{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.999936,
  "eval_steps": 500,
  "global_step": 3906,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000256,
      "grad_norm": 592.9743317031632,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 7.1027,
      "step": 1
    },
    {
      "epoch": 0.000512,
      "grad_norm": 599.1780379067313,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 7.1034,
      "step": 2
    },
    {
      "epoch": 0.000768,
      "grad_norm": 542.8746698979221,
      "learning_rate": 3e-06,
      "loss": 6.5723,
      "step": 3
    },
    {
      "epoch": 0.001024,
      "grad_norm": 260.529659383525,
      "learning_rate": 4.000000000000001e-06,
      "loss": 5.2188,
      "step": 4
    },
    {
      "epoch": 0.00128,
      "grad_norm": 113.00533898587209,
      "learning_rate": 5e-06,
      "loss": 4.0969,
      "step": 5
    },
    {
      "epoch": 0.001536,
      "grad_norm": 49.74489646576834,
      "learning_rate": 6e-06,
      "loss": 3.4141,
      "step": 6
    },
    {
      "epoch": 0.001792,
      "grad_norm": 30.811416695900597,
      "learning_rate": 7e-06,
      "loss": 3.2294,
      "step": 7
    },
    {
      "epoch": 0.002048,
      "grad_norm": 14.379119643830325,
      "learning_rate": 8.000000000000001e-06,
      "loss": 2.7187,
      "step": 8
    },
    {
      "epoch": 0.002304,
      "grad_norm": 6.924380787635255,
      "learning_rate": 9e-06,
      "loss": 2.6165,
      "step": 9
    },
    {
      "epoch": 0.00256,
      "grad_norm": 8.303192647738951,
      "learning_rate": 1e-05,
      "loss": 2.5339,
      "step": 10
    },
    {
      "epoch": 0.002816,
      "grad_norm": 6.177706017502565,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 2.4265,
      "step": 11
    },
    {
      "epoch": 0.003072,
      "grad_norm": 4.6524525901522855,
      "learning_rate": 1.2e-05,
      "loss": 2.3719,
      "step": 12
    },
    {
      "epoch": 0.003328,
      "grad_norm": 3.6718997401373885,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 2.2678,
      "step": 13
    },
    {
      "epoch": 0.003584,
      "grad_norm": 3.3357787972855113,
      "learning_rate": 1.4e-05,
      "loss": 2.4163,
      "step": 14
    },
    {
      "epoch": 0.00384,
      "grad_norm": 2.7345388229485197,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 2.2873,
      "step": 15
    },
    {
      "epoch": 0.004096,
      "grad_norm": 2.843523576780457,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 2.3126,
      "step": 16
    },
    {
      "epoch": 0.004352,
      "grad_norm": 3.023816042800059,
      "learning_rate": 1.7e-05,
      "loss": 2.1398,
      "step": 17
    },
    {
      "epoch": 0.004608,
      "grad_norm": 3.3170405372890697,
      "learning_rate": 1.8e-05,
      "loss": 2.1752,
      "step": 18
    },
    {
      "epoch": 0.004864,
      "grad_norm": 2.8559468885914283,
      "learning_rate": 1.9e-05,
      "loss": 2.2289,
      "step": 19
    },
    {
      "epoch": 0.00512,
      "grad_norm": 3.2080620203206287,
      "learning_rate": 2e-05,
      "loss": 2.2723,
      "step": 20
    },
    {
      "epoch": 0.005376,
      "grad_norm": 2.755611282975557,
      "learning_rate": 2.1000000000000002e-05,
      "loss": 2.1972,
      "step": 21
    },
    {
      "epoch": 0.005632,
      "grad_norm": 3.0930080601750194,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 2.1221,
      "step": 22
    },
    {
      "epoch": 0.005888,
      "grad_norm": 2.939046946640738,
      "learning_rate": 2.3e-05,
      "loss": 2.144,
      "step": 23
    },
    {
      "epoch": 0.006144,
      "grad_norm": 3.164263131287616,
      "learning_rate": 2.4e-05,
      "loss": 2.1498,
      "step": 24
    },
    {
      "epoch": 0.0064,
      "grad_norm": 2.526804917170411,
      "learning_rate": 2.5e-05,
      "loss": 2.2104,
      "step": 25
    },
    {
      "epoch": 0.006656,
      "grad_norm": 2.2373273892483345,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 2.1118,
      "step": 26
    },
    {
      "epoch": 0.006912,
      "grad_norm": 2.4966928094222616,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 2.0654,
      "step": 27
    },
    {
      "epoch": 0.007168,
      "grad_norm": 2.7912251583606733,
      "learning_rate": 2.8e-05,
      "loss": 2.0889,
      "step": 28
    },
    {
      "epoch": 0.007424,
      "grad_norm": 2.431054160394137,
      "learning_rate": 2.9e-05,
      "loss": 2.0099,
      "step": 29
    },
    {
      "epoch": 0.00768,
      "grad_norm": 2.902786261237885,
      "learning_rate": 3.0000000000000004e-05,
      "loss": 2.0453,
      "step": 30
    },
    {
      "epoch": 0.007936,
      "grad_norm": 2.6168819637809824,
      "learning_rate": 3.1e-05,
      "loss": 2.0467,
      "step": 31
    },
    {
      "epoch": 0.008192,
      "grad_norm": 2.429081373618289,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 2.064,
      "step": 32
    },
    {
      "epoch": 0.008448,
      "grad_norm": 2.5494920231573066,
      "learning_rate": 3.3e-05,
      "loss": 2.0269,
      "step": 33
    },
    {
      "epoch": 0.008704,
      "grad_norm": 2.057227818655378,
      "learning_rate": 3.4e-05,
      "loss": 2.0346,
      "step": 34
    },
    {
      "epoch": 0.00896,
      "grad_norm": 2.3567588259996497,
      "learning_rate": 3.5000000000000004e-05,
      "loss": 1.9448,
      "step": 35
    },
    {
      "epoch": 0.009216,
      "grad_norm": 2.3677974863708133,
      "learning_rate": 3.6e-05,
      "loss": 1.9426,
      "step": 36
    },
    {
      "epoch": 0.009472,
      "grad_norm": 2.170912625075084,
      "learning_rate": 3.7000000000000005e-05,
      "loss": 2.0256,
      "step": 37
    },
    {
      "epoch": 0.009728,
      "grad_norm": 2.471503225761493,
      "learning_rate": 3.8e-05,
      "loss": 1.9751,
      "step": 38
    },
    {
      "epoch": 0.009984,
      "grad_norm": 2.0645422510246627,
      "learning_rate": 3.9e-05,
      "loss": 1.9972,
      "step": 39
    },
    {
      "epoch": 0.01024,
      "grad_norm": 2.5945416121942295,
      "learning_rate": 4e-05,
      "loss": 1.9717,
      "step": 40
    },
    {
      "epoch": 0.010496,
      "grad_norm": 2.256990687799562,
      "learning_rate": 3.999999339647203e-05,
      "loss": 2.0233,
      "step": 41
    },
    {
      "epoch": 0.010752,
      "grad_norm": 2.856060581961514,
      "learning_rate": 3.9999973585892455e-05,
      "loss": 2.051,
      "step": 42
    },
    {
      "epoch": 0.011008,
      "grad_norm": 2.18339692829619,
      "learning_rate": 3.999994056827438e-05,
      "loss": 1.9142,
      "step": 43
    },
    {
      "epoch": 0.011264,
      "grad_norm": 2.2749191731255403,
      "learning_rate": 3.999989434363959e-05,
      "loss": 2.0122,
      "step": 44
    },
    {
      "epoch": 0.01152,
      "grad_norm": 2.3442105171322742,
      "learning_rate": 3.999983491201863e-05,
      "loss": 1.9614,
      "step": 45
    },
    {
      "epoch": 0.011776,
      "grad_norm": 3.060340586762112,
      "learning_rate": 3.999976227345072e-05,
      "loss": 1.9205,
      "step": 46
    },
    {
      "epoch": 0.012032,
      "grad_norm": 2.161941312538491,
      "learning_rate": 3.999967642798385e-05,
      "loss": 1.9214,
      "step": 47
    },
    {
      "epoch": 0.012288,
      "grad_norm": 2.170351449205052,
      "learning_rate": 3.99995773756747e-05,
      "loss": 1.9153,
      "step": 48
    },
    {
      "epoch": 0.012544,
      "grad_norm": 1.8040235993227136,
      "learning_rate": 3.9999465116588664e-05,
      "loss": 1.8511,
      "step": 49
    },
    {
      "epoch": 0.0128,
      "grad_norm": 2.1919388547398637,
      "learning_rate": 3.99993396507999e-05,
      "loss": 1.8317,
      "step": 50
    },
    {
      "epoch": 0.013056,
      "grad_norm": 1.9694273776898437,
      "learning_rate": 3.999920097839124e-05,
      "loss": 1.8839,
      "step": 51
    },
    {
      "epoch": 0.013312,
      "grad_norm": 1.8146358869514165,
      "learning_rate": 3.999904909945427e-05,
      "loss": 1.8511,
      "step": 52
    },
    {
      "epoch": 0.013568,
      "grad_norm": 1.6933734738418547,
      "learning_rate": 3.999888401408927e-05,
      "loss": 1.8425,
      "step": 53
    },
    {
      "epoch": 0.013824,
      "grad_norm": 1.8648277774057493,
      "learning_rate": 3.999870572240527e-05,
      "loss": 1.841,
      "step": 54
    },
    {
      "epoch": 0.01408,
      "grad_norm": 1.6319589161530719,
      "learning_rate": 3.999851422451999e-05,
      "loss": 1.8339,
      "step": 55
    },
    {
      "epoch": 0.014336,
      "grad_norm": 2.0773191877566846,
      "learning_rate": 3.9998309520559904e-05,
      "loss": 1.9235,
      "step": 56
    },
    {
      "epoch": 0.014592,
      "grad_norm": 1.9786939222445252,
      "learning_rate": 3.999809161066017e-05,
      "loss": 1.9388,
      "step": 57
    },
    {
      "epoch": 0.014848,
      "grad_norm": 2.112419570215387,
      "learning_rate": 3.999786049496469e-05,
      "loss": 1.815,
      "step": 58
    },
    {
      "epoch": 0.015104,
      "grad_norm": 1.681719024460609,
      "learning_rate": 3.999761617362609e-05,
      "loss": 1.8619,
      "step": 59
    },
    {
      "epoch": 0.01536,
      "grad_norm": 1.6898626214306998,
      "learning_rate": 3.99973586468057e-05,
      "loss": 1.8577,
      "step": 60
    },
    {
      "epoch": 0.015616,
      "grad_norm": 1.5779901470944477,
      "learning_rate": 3.9997087914673585e-05,
      "loss": 1.8654,
      "step": 61
    },
    {
      "epoch": 0.015872,
      "grad_norm": 1.838507428017589,
      "learning_rate": 3.999680397740852e-05,
      "loss": 1.8033,
      "step": 62
    },
    {
      "epoch": 0.016128,
      "grad_norm": 1.6109226658384828,
      "learning_rate": 3.9996506835198005e-05,
      "loss": 1.7934,
      "step": 63
    },
    {
      "epoch": 0.016384,
      "grad_norm": 1.5550114322341122,
      "learning_rate": 3.999619648823826e-05,
      "loss": 1.8388,
      "step": 64
    },
    {
      "epoch": 0.01664,
      "grad_norm": 1.5712910294227422,
      "learning_rate": 3.999587293673422e-05,
      "loss": 1.8471,
      "step": 65
    },
    {
      "epoch": 0.016896,
      "grad_norm": 1.518387212706681,
      "learning_rate": 3.9995536180899544e-05,
      "loss": 1.7755,
      "step": 66
    },
    {
      "epoch": 0.017152,
      "grad_norm": 1.6764288290558913,
      "learning_rate": 3.9995186220956616e-05,
      "loss": 1.8353,
      "step": 67
    },
    {
      "epoch": 0.017408,
      "grad_norm": 1.7902844247368257,
      "learning_rate": 3.999482305713653e-05,
      "loss": 1.8891,
      "step": 68
    },
    {
      "epoch": 0.017664,
      "grad_norm": 1.457627249670965,
      "learning_rate": 3.9994446689679085e-05,
      "loss": 1.8241,
      "step": 69
    },
    {
      "epoch": 0.01792,
      "grad_norm": 1.7805484350169922,
      "learning_rate": 3.999405711883285e-05,
      "loss": 1.7861,
      "step": 70
    },
    {
      "epoch": 0.018176,
      "grad_norm": 1.7669361023561936,
      "learning_rate": 3.999365434485504e-05,
      "loss": 1.7952,
      "step": 71
    },
    {
      "epoch": 0.018432,
      "grad_norm": 1.3036511554189805,
      "learning_rate": 3.999323836801166e-05,
      "loss": 1.8357,
      "step": 72
    },
    {
      "epoch": 0.018688,
      "grad_norm": 1.4614422147700734,
      "learning_rate": 3.999280918857739e-05,
      "loss": 1.7782,
      "step": 73
    },
    {
      "epoch": 0.018944,
      "grad_norm": 1.507240616354275,
      "learning_rate": 3.999236680683564e-05,
      "loss": 1.817,
      "step": 74
    },
    {
      "epoch": 0.0192,
      "grad_norm": 1.503660314941665,
      "learning_rate": 3.9991911223078537e-05,
      "loss": 1.7895,
      "step": 75
    },
    {
      "epoch": 0.019456,
      "grad_norm": 1.391065206748873,
      "learning_rate": 3.999144243760693e-05,
      "loss": 1.7798,
      "step": 76
    },
    {
      "epoch": 0.019712,
      "grad_norm": 1.4092222546201962,
      "learning_rate": 3.9990960450730376e-05,
      "loss": 1.7809,
      "step": 77
    },
    {
      "epoch": 0.019968,
      "grad_norm": 1.3330729002785804,
      "learning_rate": 3.999046526276716e-05,
      "loss": 1.7588,
      "step": 78
    },
    {
      "epoch": 0.020224,
      "grad_norm": 1.2940922859791728,
      "learning_rate": 3.998995687404429e-05,
      "loss": 1.7642,
      "step": 79
    },
    {
      "epoch": 0.02048,
      "grad_norm": 1.152265835668214,
      "learning_rate": 3.998943528489746e-05,
      "loss": 1.741,
      "step": 80
    },
    {
      "epoch": 0.020736,
      "grad_norm": 1.6601432070064535,
      "learning_rate": 3.998890049567113e-05,
      "loss": 1.7976,
      "step": 81
    },
    {
      "epoch": 0.020992,
      "grad_norm": 1.019135351254906,
      "learning_rate": 3.9988352506718426e-05,
      "loss": 1.7517,
      "step": 82
    },
    {
      "epoch": 0.021248,
      "grad_norm": 1.4038086263714684,
      "learning_rate": 3.998779131840123e-05,
      "loss": 1.721,
      "step": 83
    },
    {
      "epoch": 0.021504,
      "grad_norm": 1.1440209713679739,
      "learning_rate": 3.998721693109011e-05,
      "loss": 1.735,
      "step": 84
    },
    {
      "epoch": 0.02176,
      "grad_norm": 1.2241475040897805,
      "learning_rate": 3.998662934516438e-05,
      "loss": 1.7742,
      "step": 85
    },
    {
      "epoch": 0.022016,
      "grad_norm": 1.4360186297999553,
      "learning_rate": 3.998602856101205e-05,
      "loss": 1.7198,
      "step": 86
    },
    {
      "epoch": 0.022272,
      "grad_norm": 1.3780396600933922,
      "learning_rate": 3.9985414579029836e-05,
      "loss": 1.7398,
      "step": 87
    },
    {
      "epoch": 0.022528,
      "grad_norm": 1.2041825111331663,
      "learning_rate": 3.99847873996232e-05,
      "loss": 1.735,
      "step": 88
    },
    {
      "epoch": 0.022784,
      "grad_norm": 1.4226095111819237,
      "learning_rate": 3.9984147023206295e-05,
      "loss": 1.7487,
      "step": 89
    },
    {
      "epoch": 0.02304,
      "grad_norm": 1.2299632105832863,
      "learning_rate": 3.9983493450202e-05,
      "loss": 1.7461,
      "step": 90
    },
    {
      "epoch": 0.023296,
      "grad_norm": 1.0311434872404308,
      "learning_rate": 3.998282668104188e-05,
      "loss": 1.7403,
      "step": 91
    },
    {
      "epoch": 0.023552,
      "grad_norm": 1.21132035127431,
      "learning_rate": 3.998214671616627e-05,
      "loss": 1.6954,
      "step": 92
    },
    {
      "epoch": 0.023808,
      "grad_norm": 1.2658549348037387,
      "learning_rate": 3.9981453556024175e-05,
      "loss": 1.8071,
      "step": 93
    },
    {
      "epoch": 0.024064,
      "grad_norm": 1.039194378894388,
      "learning_rate": 3.998074720107332e-05,
      "loss": 1.711,
      "step": 94
    },
    {
      "epoch": 0.02432,
      "grad_norm": 1.3722879638106795,
      "learning_rate": 3.998002765178015e-05,
      "loss": 1.7698,
      "step": 95
    },
    {
      "epoch": 0.024576,
      "grad_norm": 1.1862235393233689,
      "learning_rate": 3.997929490861983e-05,
      "loss": 1.7301,
      "step": 96
    },
    {
      "epoch": 0.024832,
      "grad_norm": 1.1550824312771037,
      "learning_rate": 3.997854897207621e-05,
      "loss": 1.7179,
      "step": 97
    },
    {
      "epoch": 0.025088,
      "grad_norm": 1.3087085960813287,
      "learning_rate": 3.997778984264189e-05,
      "loss": 1.7119,
      "step": 98
    },
    {
      "epoch": 0.025344,
      "grad_norm": 1.0807007825087622,
      "learning_rate": 3.997701752081816e-05,
      "loss": 1.7394,
      "step": 99
    },
    {
      "epoch": 0.0256,
      "grad_norm": 1.3084776122166415,
      "learning_rate": 3.9976232007115015e-05,
      "loss": 1.6803,
      "step": 100
    },
    {
      "epoch": 0.025856,
      "grad_norm": 1.0923354482196317,
      "learning_rate": 3.9975433302051184e-05,
      "loss": 1.7406,
      "step": 101
    },
    {
      "epoch": 0.026112,
      "grad_norm": 1.0883912475361128,
      "learning_rate": 3.997462140615408e-05,
      "loss": 1.7512,
      "step": 102
    },
    {
      "epoch": 0.026368,
      "grad_norm": 1.438650846748812,
      "learning_rate": 3.9973796319959854e-05,
      "loss": 1.7349,
      "step": 103
    },
    {
      "epoch": 0.026624,
      "grad_norm": 1.1863717155104916,
      "learning_rate": 3.997295804401335e-05,
      "loss": 1.8028,
      "step": 104
    },
    {
      "epoch": 0.02688,
      "grad_norm": 1.1285889950460126,
      "learning_rate": 3.997210657886813e-05,
      "loss": 1.6684,
      "step": 105
    },
    {
      "epoch": 0.027136,
      "grad_norm": 1.042248894772584,
      "learning_rate": 3.9971241925086444e-05,
      "loss": 1.6906,
      "step": 106
    },
    {
      "epoch": 0.027392,
      "grad_norm": 1.1054056878331868,
      "learning_rate": 3.9970364083239284e-05,
      "loss": 1.6753,
      "step": 107
    },
    {
      "epoch": 0.027648,
      "grad_norm": 1.106329538499644,
      "learning_rate": 3.996947305390634e-05,
      "loss": 1.6673,
      "step": 108
    },
    {
      "epoch": 0.027904,
      "grad_norm": 1.1496847174342597,
      "learning_rate": 3.9968568837675986e-05,
      "loss": 1.6906,
      "step": 109
    },
    {
      "epoch": 0.02816,
      "grad_norm": 1.0523190733701966,
      "learning_rate": 3.996765143514534e-05,
      "loss": 1.7352,
      "step": 110
    },
    {
      "epoch": 0.028416,
      "grad_norm": 1.0398196895661893,
      "learning_rate": 3.996672084692021e-05,
      "loss": 1.6813,
      "step": 111
    },
    {
      "epoch": 0.028672,
      "grad_norm": 0.9599547428898851,
      "learning_rate": 3.9965777073615105e-05,
      "loss": 1.6841,
      "step": 112
    },
    {
      "epoch": 0.028928,
      "grad_norm": 1.0842378041962715,
      "learning_rate": 3.996482011585325e-05,
      "loss": 1.7104,
      "step": 113
    },
    {
      "epoch": 0.029184,
      "grad_norm": 1.2179306231138718,
      "learning_rate": 3.996384997426658e-05,
      "loss": 1.6565,
      "step": 114
    },
    {
      "epoch": 0.02944,
      "grad_norm": 0.8890508847263513,
      "learning_rate": 3.996286664949573e-05,
      "loss": 1.6614,
      "step": 115
    },
    {
      "epoch": 0.029696,
      "grad_norm": 0.8812111686417768,
      "learning_rate": 3.996187014219004e-05,
      "loss": 1.716,
      "step": 116
    },
    {
      "epoch": 0.029952,
      "grad_norm": 0.9325038134241184,
      "learning_rate": 3.9960860453007555e-05,
      "loss": 1.6901,
      "step": 117
    },
    {
      "epoch": 0.030208,
      "grad_norm": 0.9671295040562344,
      "learning_rate": 3.995983758261502e-05,
      "loss": 1.7093,
      "step": 118
    },
    {
      "epoch": 0.030464,
      "grad_norm": 0.9505147489163117,
      "learning_rate": 3.9958801531687904e-05,
      "loss": 1.6738,
      "step": 119
    },
    {
      "epoch": 0.03072,
      "grad_norm": 0.9724182748888957,
      "learning_rate": 3.995775230091035e-05,
      "loss": 1.6735,
      "step": 120
    },
    {
      "epoch": 0.030976,
      "grad_norm": 1.1693639746457531,
      "learning_rate": 3.995668989097525e-05,
      "loss": 1.7062,
      "step": 121
    },
    {
      "epoch": 0.031232,
      "grad_norm": 0.9683984161249218,
      "learning_rate": 3.995561430258413e-05,
      "loss": 1.6642,
      "step": 122
    },
    {
      "epoch": 0.031488,
      "grad_norm": 0.9383252696760273,
      "learning_rate": 3.995452553644728e-05,
      "loss": 1.6459,
      "step": 123
    },
    {
      "epoch": 0.031744,
      "grad_norm": 0.9501958624559937,
      "learning_rate": 3.995342359328367e-05,
      "loss": 1.6897,
      "step": 124
    },
    {
      "epoch": 0.032,
      "grad_norm": 0.8995019479668482,
      "learning_rate": 3.995230847382096e-05,
      "loss": 1.7132,
      "step": 125
    },
    {
      "epoch": 0.032256,
      "grad_norm": 0.8132439493043349,
      "learning_rate": 3.995118017879554e-05,
      "loss": 1.581,
      "step": 126
    },
    {
      "epoch": 0.032512,
      "grad_norm": 1.2119019710159098,
      "learning_rate": 3.995003870895247e-05,
      "loss": 1.6804,
      "step": 127
    },
    {
      "epoch": 0.032768,
      "grad_norm": 0.9120155940827004,
      "learning_rate": 3.994888406504552e-05,
      "loss": 1.7045,
      "step": 128
    },
    {
      "epoch": 0.033024,
      "grad_norm": 1.0861585776784026,
      "learning_rate": 3.994771624783718e-05,
      "loss": 1.6389,
      "step": 129
    },
    {
      "epoch": 0.03328,
      "grad_norm": 1.0207882622665796,
      "learning_rate": 3.994653525809859e-05,
      "loss": 1.7005,
      "step": 130
    },
    {
      "epoch": 0.033536,
      "grad_norm": 0.9773135378532466,
      "learning_rate": 3.9945341096609655e-05,
      "loss": 1.6992,
      "step": 131
    },
    {
      "epoch": 0.033792,
      "grad_norm": 1.0278290435463915,
      "learning_rate": 3.9944133764158925e-05,
      "loss": 1.6899,
      "step": 132
    },
    {
      "epoch": 0.034048,
      "grad_norm": 0.9079032183424462,
      "learning_rate": 3.9942913261543664e-05,
      "loss": 1.6429,
      "step": 133
    },
    {
      "epoch": 0.034304,
      "grad_norm": 0.8323298818892745,
      "learning_rate": 3.9941679589569836e-05,
      "loss": 1.6891,
      "step": 134
    },
    {
      "epoch": 0.03456,
      "grad_norm": 0.8409270975746009,
      "learning_rate": 3.99404327490521e-05,
      "loss": 1.6226,
      "step": 135
    },
    {
      "epoch": 0.034816,
      "grad_norm": 0.9149078134374603,
      "learning_rate": 3.9939172740813825e-05,
      "loss": 1.6255,
      "step": 136
    },
    {
      "epoch": 0.035072,
      "grad_norm": 1.0196329454781172,
      "learning_rate": 3.993789956568704e-05,
      "loss": 1.7171,
      "step": 137
    },
    {
      "epoch": 0.035328,
      "grad_norm": 0.7946758350260431,
      "learning_rate": 3.9936613224512495e-05,
      "loss": 1.5579,
      "step": 138
    },
    {
      "epoch": 0.035584,
      "grad_norm": 0.8874319573940996,
      "learning_rate": 3.993531371813963e-05,
      "loss": 1.6577,
      "step": 139
    },
    {
      "epoch": 0.03584,
      "grad_norm": 1.2978489551059718,
      "learning_rate": 3.9934001047426585e-05,
      "loss": 1.6809,
      "step": 140
    },
    {
      "epoch": 0.036096,
      "grad_norm": 0.8042620114066384,
      "learning_rate": 3.9932675213240175e-05,
      "loss": 1.627,
      "step": 141
    },
    {
      "epoch": 0.036352,
      "grad_norm": 0.9706716408253456,
      "learning_rate": 3.993133621645593e-05,
      "loss": 1.6764,
      "step": 142
    },
    {
      "epoch": 0.036608,
      "grad_norm": 1.16280871089314,
      "learning_rate": 3.9929984057958055e-05,
      "loss": 1.6521,
      "step": 143
    },
    {
      "epoch": 0.036864,
      "grad_norm": 0.8791842830765326,
      "learning_rate": 3.992861873863944e-05,
      "loss": 1.6497,
      "step": 144
    },
    {
      "epoch": 0.03712,
      "grad_norm": 1.2155169707095028,
      "learning_rate": 3.9927240259401694e-05,
      "loss": 1.5884,
      "step": 145
    },
    {
      "epoch": 0.037376,
      "grad_norm": 0.7562842582963687,
      "learning_rate": 3.992584862115509e-05,
      "loss": 1.5987,
      "step": 146
    },
    {
      "epoch": 0.037632,
      "grad_norm": 0.7795309651797402,
      "learning_rate": 3.992444382481862e-05,
      "loss": 1.5948,
      "step": 147
    },
    {
      "epoch": 0.037888,
      "grad_norm": 0.7771885458088762,
      "learning_rate": 3.992302587131991e-05,
      "loss": 1.6598,
      "step": 148
    },
    {
      "epoch": 0.038144,
      "grad_norm": 0.8098660490108356,
      "learning_rate": 3.9921594761595334e-05,
      "loss": 1.5991,
      "step": 149
    },
    {
      "epoch": 0.0384,
      "grad_norm": 0.7498591706711951,
      "learning_rate": 3.992015049658992e-05,
      "loss": 1.6543,
      "step": 150
    },
    {
      "epoch": 0.038656,
      "grad_norm": 0.7736378513746694,
      "learning_rate": 3.991869307725741e-05,
      "loss": 1.6648,
      "step": 151
    },
    {
      "epoch": 0.038912,
      "grad_norm": 0.742377517409635,
      "learning_rate": 3.9917222504560184e-05,
      "loss": 1.6041,
      "step": 152
    },
    {
      "epoch": 0.039168,
      "grad_norm": 0.7783701921138761,
      "learning_rate": 3.991573877946937e-05,
      "loss": 1.6443,
      "step": 153
    },
    {
      "epoch": 0.039424,
      "grad_norm": 0.7659432386168707,
      "learning_rate": 3.9914241902964725e-05,
      "loss": 1.7052,
      "step": 154
    },
    {
      "epoch": 0.03968,
      "grad_norm": 0.7518527516141134,
      "learning_rate": 3.991273187603473e-05,
      "loss": 1.6251,
      "step": 155
    },
    {
      "epoch": 0.039936,
      "grad_norm": 0.7822919113116653,
      "learning_rate": 3.9911208699676537e-05,
      "loss": 1.6287,
      "step": 156
    },
    {
      "epoch": 0.040192,
      "grad_norm": 0.8066856650406288,
      "learning_rate": 3.990967237489597e-05,
      "loss": 1.6348,
      "step": 157
    },
    {
      "epoch": 0.040448,
      "grad_norm": 0.9105078454075806,
      "learning_rate": 3.990812290270755e-05,
      "loss": 1.6133,
      "step": 158
    },
    {
      "epoch": 0.040704,
      "grad_norm": 0.8941931068770252,
      "learning_rate": 3.990656028413448e-05,
      "loss": 1.6297,
      "step": 159
    },
    {
      "epoch": 0.04096,
      "grad_norm": 0.8696780893605999,
      "learning_rate": 3.9904984520208626e-05,
      "loss": 1.6098,
      "step": 160
    },
    {
      "epoch": 0.041216,
      "grad_norm": 0.7305904830360131,
      "learning_rate": 3.9903395611970565e-05,
      "loss": 1.6131,
      "step": 161
    },
    {
      "epoch": 0.041472,
      "grad_norm": 0.7714314658140741,
      "learning_rate": 3.990179356046952e-05,
      "loss": 1.5826,
      "step": 162
    },
    {
      "epoch": 0.041728,
      "grad_norm": 0.7610912423863428,
      "learning_rate": 3.990017836676344e-05,
      "loss": 1.6436,
      "step": 163
    },
    {
      "epoch": 0.041984,
      "grad_norm": 2.4705363679369245,
      "learning_rate": 3.9898550031918886e-05,
      "loss": 1.6775,
      "step": 164
    },
    {
      "epoch": 0.04224,
      "grad_norm": 0.9467340535163156,
      "learning_rate": 3.9896908557011146e-05,
      "loss": 1.6481,
      "step": 165
    },
    {
      "epoch": 0.042496,
      "grad_norm": 0.9246624477542396,
      "learning_rate": 3.989525394312418e-05,
      "loss": 1.6508,
      "step": 166
    },
    {
      "epoch": 0.042752,
      "grad_norm": 0.9503701562561346,
      "learning_rate": 3.989358619135062e-05,
      "loss": 1.6493,
      "step": 167
    },
    {
      "epoch": 0.043008,
      "grad_norm": 0.8252572441701449,
      "learning_rate": 3.989190530279175e-05,
      "loss": 1.6417,
      "step": 168
    },
    {
      "epoch": 0.043264,
      "grad_norm": 0.848455041237527,
      "learning_rate": 3.9890211278557576e-05,
      "loss": 1.5947,
      "step": 169
    },
    {
      "epoch": 0.04352,
      "grad_norm": 0.9366609881558517,
      "learning_rate": 3.988850411976674e-05,
      "loss": 1.6538,
      "step": 170
    },
    {
      "epoch": 0.043776,
      "grad_norm": 0.8233618081995571,
      "learning_rate": 3.988678382754656e-05,
      "loss": 1.5596,
      "step": 171
    },
    {
      "epoch": 0.044032,
      "grad_norm": 0.795394583475737,
      "learning_rate": 3.988505040303304e-05,
      "loss": 1.6272,
      "step": 172
    },
    {
      "epoch": 0.044288,
      "grad_norm": 0.9468737748519184,
      "learning_rate": 3.9883303847370866e-05,
      "loss": 1.6455,
      "step": 173
    },
    {
      "epoch": 0.044544,
      "grad_norm": 0.8779097843647329,
      "learning_rate": 3.988154416171336e-05,
      "loss": 1.6205,
      "step": 174
    },
    {
      "epoch": 0.0448,
      "grad_norm": 0.7626697331087257,
      "learning_rate": 3.987977134722255e-05,
      "loss": 1.6421,
      "step": 175
    },
    {
      "epoch": 0.045056,
      "grad_norm": 0.7911748234927005,
      "learning_rate": 3.987798540506911e-05,
      "loss": 1.6395,
      "step": 176
    },
    {
      "epoch": 0.045312,
      "grad_norm": 0.8298848464014686,
      "learning_rate": 3.9876186336432404e-05,
      "loss": 1.5974,
      "step": 177
    },
    {
      "epoch": 0.045568,
      "grad_norm": 0.8004652482579814,
      "learning_rate": 3.987437414250044e-05,
      "loss": 1.6324,
      "step": 178
    },
    {
      "epoch": 0.045824,
      "grad_norm": 1.048446607740693,
      "learning_rate": 3.9872548824469913e-05,
      "loss": 1.6155,
      "step": 179
    },
    {
      "epoch": 0.04608,
      "grad_norm": 0.7739830340237577,
      "learning_rate": 3.9870710383546166e-05,
      "loss": 1.6201,
      "step": 180
    },
    {
      "epoch": 0.046336,
      "grad_norm": 0.7468121321303345,
      "learning_rate": 3.986885882094324e-05,
      "loss": 1.6282,
      "step": 181
    },
    {
      "epoch": 0.046592,
      "grad_norm": 0.7784300337192723,
      "learning_rate": 3.98669941378838e-05,
      "loss": 1.6151,
      "step": 182
    },
    {
      "epoch": 0.046848,
      "grad_norm": 0.7887247249132889,
      "learning_rate": 3.986511633559919e-05,
      "loss": 1.6322,
      "step": 183
    },
    {
      "epoch": 0.047104,
      "grad_norm": 0.7763175775630177,
      "learning_rate": 3.986322541532944e-05,
      "loss": 1.5858,
      "step": 184
    },
    {
      "epoch": 0.04736,
      "grad_norm": 0.7029643627744828,
      "learning_rate": 3.986132137832322e-05,
      "loss": 1.6501,
      "step": 185
    },
    {
      "epoch": 0.047616,
      "grad_norm": 0.7913393851592154,
      "learning_rate": 3.985940422583786e-05,
      "loss": 1.619,
      "step": 186
    },
    {
      "epoch": 0.047872,
      "grad_norm": 0.6954422281994223,
      "learning_rate": 3.985747395913936e-05,
      "loss": 1.6122,
      "step": 187
    },
    {
      "epoch": 0.048128,
      "grad_norm": 0.7370595672662306,
      "learning_rate": 3.9855530579502375e-05,
      "loss": 1.5991,
      "step": 188
    },
    {
      "epoch": 0.048384,
      "grad_norm": 0.6963060857346365,
      "learning_rate": 3.9853574088210224e-05,
      "loss": 1.6093,
      "step": 189
    },
    {
      "epoch": 0.04864,
      "grad_norm": 0.6999246871883297,
      "learning_rate": 3.985160448655488e-05,
      "loss": 1.644,
      "step": 190
    },
    {
      "epoch": 0.048896,
      "grad_norm": 0.8871350739368395,
      "learning_rate": 3.984962177583698e-05,
      "loss": 1.6516,
      "step": 191
    },
    {
      "epoch": 0.049152,
      "grad_norm": 0.665284498626241,
      "learning_rate": 3.984762595736581e-05,
      "loss": 1.5579,
      "step": 192
    },
    {
      "epoch": 0.049408,
      "grad_norm": 0.8004360626359813,
      "learning_rate": 3.98456170324593e-05,
      "loss": 1.6606,
      "step": 193
    },
    {
      "epoch": 0.049664,
      "grad_norm": 0.7551735939415242,
      "learning_rate": 3.9843595002444075e-05,
      "loss": 1.602,
      "step": 194
    },
    {
      "epoch": 0.04992,
      "grad_norm": 0.7206439722894225,
      "learning_rate": 3.984155986865537e-05,
      "loss": 1.5955,
      "step": 195
    },
    {
      "epoch": 0.050176,
      "grad_norm": 0.7753493657493119,
      "learning_rate": 3.9839511632437096e-05,
      "loss": 1.5357,
      "step": 196
    },
    {
      "epoch": 0.050432,
      "grad_norm": 0.6948348502128749,
      "learning_rate": 3.9837450295141816e-05,
      "loss": 1.5909,
      "step": 197
    },
    {
      "epoch": 0.050688,
      "grad_norm": 0.7107966792035627,
      "learning_rate": 3.983537585813073e-05,
      "loss": 1.6708,
      "step": 198
    },
    {
      "epoch": 0.050944,
      "grad_norm": 0.7395099525564277,
      "learning_rate": 3.98332883227737e-05,
      "loss": 1.5605,
      "step": 199
    },
    {
      "epoch": 0.0512,
      "grad_norm": 0.6647533169618578,
      "learning_rate": 3.9831187690449254e-05,
      "loss": 1.6156,
      "step": 200
    },
    {
      "epoch": 0.051456,
      "grad_norm": 0.7059101501205137,
      "learning_rate": 3.9829073962544525e-05,
      "loss": 1.624,
      "step": 201
    },
    {
      "epoch": 0.051712,
      "grad_norm": 0.891898916836492,
      "learning_rate": 3.9826947140455334e-05,
      "loss": 1.6255,
      "step": 202
    },
    {
      "epoch": 0.051968,
      "grad_norm": 0.6921158893254238,
      "learning_rate": 3.982480722558613e-05,
      "loss": 1.5554,
      "step": 203
    },
    {
      "epoch": 0.052224,
      "grad_norm": 0.7563556076434174,
      "learning_rate": 3.982265421935001e-05,
      "loss": 1.6158,
      "step": 204
    },
    {
      "epoch": 0.05248,
      "grad_norm": 0.7453240485860165,
      "learning_rate": 3.982048812316872e-05,
      "loss": 1.6056,
      "step": 205
    },
    {
      "epoch": 0.052736,
      "grad_norm": 0.7005042088111104,
      "learning_rate": 3.981830893847265e-05,
      "loss": 1.6017,
      "step": 206
    },
    {
      "epoch": 0.052992,
      "grad_norm": 1.3006069947635521,
      "learning_rate": 3.981611666670082e-05,
      "loss": 1.6227,
      "step": 207
    },
    {
      "epoch": 0.053248,
      "grad_norm": 0.9463170451403351,
      "learning_rate": 3.9813911309300916e-05,
      "loss": 1.6319,
      "step": 208
    },
    {
      "epoch": 0.053504,
      "grad_norm": 0.6555692155134595,
      "learning_rate": 3.981169286772925e-05,
      "loss": 1.627,
      "step": 209
    },
    {
      "epoch": 0.05376,
      "grad_norm": 0.6411962367215546,
      "learning_rate": 3.980946134345077e-05,
      "loss": 1.5803,
      "step": 210
    },
    {
      "epoch": 0.054016,
      "grad_norm": 0.7028659368248574,
      "learning_rate": 3.980721673793907e-05,
      "loss": 1.5945,
      "step": 211
    },
    {
      "epoch": 0.054272,
      "grad_norm": 0.7597298381504562,
      "learning_rate": 3.9804959052676386e-05,
      "loss": 1.6166,
      "step": 212
    },
    {
      "epoch": 0.054528,
      "grad_norm": 0.6597937276241083,
      "learning_rate": 3.980268828915359e-05,
      "loss": 1.6014,
      "step": 213
    },
    {
      "epoch": 0.054784,
      "grad_norm": 0.6527427801713436,
      "learning_rate": 3.980040444887017e-05,
      "loss": 1.6061,
      "step": 214
    },
    {
      "epoch": 0.05504,
      "grad_norm": 0.7050012981026503,
      "learning_rate": 3.979810753333428e-05,
      "loss": 1.5752,
      "step": 215
    },
    {
      "epoch": 0.055296,
      "grad_norm": 0.6641202789394471,
      "learning_rate": 3.979579754406271e-05,
      "loss": 1.5667,
      "step": 216
    },
    {
      "epoch": 0.055552,
      "grad_norm": 0.6478084200729878,
      "learning_rate": 3.979347448258083e-05,
      "loss": 1.577,
      "step": 217
    },
    {
      "epoch": 0.055808,
      "grad_norm": 0.8401856996554157,
      "learning_rate": 3.979113835042271e-05,
      "loss": 1.6513,
      "step": 218
    },
    {
      "epoch": 0.056064,
      "grad_norm": 0.6847093898858372,
      "learning_rate": 3.9788789149131003e-05,
      "loss": 1.5861,
      "step": 219
    },
    {
      "epoch": 0.05632,
      "grad_norm": 0.65843722138248,
      "learning_rate": 3.978642688025703e-05,
      "loss": 1.6002,
      "step": 220
    },
    {
      "epoch": 0.056576,
      "grad_norm": 0.6968805088274576,
      "learning_rate": 3.978405154536071e-05,
      "loss": 1.5682,
      "step": 221
    },
    {
      "epoch": 0.056832,
      "grad_norm": 0.7287039864040735,
      "learning_rate": 3.9781663146010595e-05,
      "loss": 1.5688,
      "step": 222
    },
    {
      "epoch": 0.057088,
      "grad_norm": 0.7064532206170948,
      "learning_rate": 3.977926168378389e-05,
      "loss": 1.5846,
      "step": 223
    },
    {
      "epoch": 0.057344,
      "grad_norm": 0.712986811749775,
      "learning_rate": 3.977684716026639e-05,
      "loss": 1.6461,
      "step": 224
    },
    {
      "epoch": 0.0576,
      "grad_norm": 0.646962393577212,
      "learning_rate": 3.977441957705254e-05,
      "loss": 1.5788,
      "step": 225
    },
    {
      "epoch": 0.057856,
      "grad_norm": 0.667563683766247,
      "learning_rate": 3.97719789357454e-05,
      "loss": 1.6239,
      "step": 226
    },
    {
      "epoch": 0.058112,
      "grad_norm": 0.6765771586778769,
      "learning_rate": 3.9769525237956655e-05,
      "loss": 1.5473,
      "step": 227
    },
    {
      "epoch": 0.058368,
      "grad_norm": 0.7418106909192511,
      "learning_rate": 3.976705848530662e-05,
      "loss": 1.5923,
      "step": 228
    },
    {
      "epoch": 0.058624,
      "grad_norm": 0.673324853811566,
      "learning_rate": 3.9764578679424204e-05,
      "loss": 1.603,
      "step": 229
    },
    {
      "epoch": 0.05888,
      "grad_norm": 0.6836905468874301,
      "learning_rate": 3.976208582194696e-05,
      "loss": 1.5565,
      "step": 230
    },
    {
      "epoch": 0.059136,
      "grad_norm": 0.682245161126949,
      "learning_rate": 3.975957991452107e-05,
      "loss": 1.5812,
      "step": 231
    },
    {
      "epoch": 0.059392,
      "grad_norm": 0.6865147007579201,
      "learning_rate": 3.9757060958801286e-05,
      "loss": 1.581,
      "step": 232
    },
    {
      "epoch": 0.059648,
      "grad_norm": 0.6619829854909499,
      "learning_rate": 3.9754528956451034e-05,
      "loss": 1.5698,
      "step": 233
    },
    {
      "epoch": 0.059904,
      "grad_norm": 0.6436552022457742,
      "learning_rate": 3.975198390914232e-05,
      "loss": 1.6069,
      "step": 234
    },
    {
      "epoch": 0.06016,
      "grad_norm": 0.8291322366946088,
      "learning_rate": 3.974942581855577e-05,
      "loss": 1.586,
      "step": 235
    },
    {
      "epoch": 0.060416,
      "grad_norm": 1.5559952573778277,
      "learning_rate": 3.974685468638063e-05,
      "loss": 1.5755,
      "step": 236
    },
    {
      "epoch": 0.060672,
      "grad_norm": 0.8304033995709409,
      "learning_rate": 3.9744270514314756e-05,
      "loss": 1.5765,
      "step": 237
    },
    {
      "epoch": 0.060928,
      "grad_norm": 0.7498847339030442,
      "learning_rate": 3.974167330406461e-05,
      "loss": 1.5478,
      "step": 238
    },
    {
      "epoch": 0.061184,
      "grad_norm": 0.720015328955355,
      "learning_rate": 3.973906305734526e-05,
      "loss": 1.575,
      "step": 239
    },
    {
      "epoch": 0.06144,
      "grad_norm": 1.0936969289583933,
      "learning_rate": 3.973643977588041e-05,
      "loss": 1.5564,
      "step": 240
    },
    {
      "epoch": 0.061696,
      "grad_norm": 0.7098563782536219,
      "learning_rate": 3.973380346140233e-05,
      "loss": 1.5936,
      "step": 241
    },
    {
      "epoch": 0.061952,
      "grad_norm": 0.7078867886190099,
      "learning_rate": 3.973115411565193e-05,
      "loss": 1.5702,
      "step": 242
    },
    {
      "epoch": 0.062208,
      "grad_norm": 0.7017526979035191,
      "learning_rate": 3.9728491740378705e-05,
      "loss": 1.5163,
      "step": 243
    },
    {
      "epoch": 0.062464,
      "grad_norm": 0.6909325418368992,
      "learning_rate": 3.972581633734077e-05,
      "loss": 1.5727,
      "step": 244
    },
    {
      "epoch": 0.06272,
      "grad_norm": 0.6765399870693833,
      "learning_rate": 3.972312790830483e-05,
      "loss": 1.5451,
      "step": 245
    },
    {
      "epoch": 0.062976,
      "grad_norm": 0.6690891973736529,
      "learning_rate": 3.9720426455046195e-05,
      "loss": 1.5725,
      "step": 246
    },
    {
      "epoch": 0.063232,
      "grad_norm": 0.6761236687452795,
      "learning_rate": 3.971771197934878e-05,
      "loss": 1.5935,
      "step": 247
    },
    {
      "epoch": 0.063488,
      "grad_norm": 0.6451034482316899,
      "learning_rate": 3.97149844830051e-05,
      "loss": 1.5908,
      "step": 248
    },
    {
      "epoch": 0.063744,
      "grad_norm": 0.6514685314428951,
      "learning_rate": 3.971224396781626e-05,
      "loss": 1.5662,
      "step": 249
    },
    {
      "epoch": 0.064,
      "grad_norm": 0.6542085356401529,
      "learning_rate": 3.970949043559197e-05,
      "loss": 1.5763,
      "step": 250
    },
    {
      "epoch": 0.064256,
      "grad_norm": 0.7120449728175606,
      "learning_rate": 3.970672388815052e-05,
      "loss": 1.5633,
      "step": 251
    },
    {
      "epoch": 0.064512,
      "grad_norm": 0.6360640695812156,
      "learning_rate": 3.970394432731883e-05,
      "loss": 1.5489,
      "step": 252
    },
    {
      "epoch": 0.064768,
      "grad_norm": 0.6225201283579339,
      "learning_rate": 3.9701151754932376e-05,
      "loss": 1.5615,
      "step": 253
    },
    {
      "epoch": 0.065024,
      "grad_norm": 0.6511260383517505,
      "learning_rate": 3.969834617283525e-05,
      "loss": 1.5535,
      "step": 254
    },
    {
      "epoch": 0.06528,
      "grad_norm": 0.5810087078492836,
      "learning_rate": 3.9695527582880105e-05,
      "loss": 1.6105,
      "step": 255
    },
    {
      "epoch": 0.065536,
      "grad_norm": 0.6998932628986286,
      "learning_rate": 3.969269598692823e-05,
      "loss": 1.6258,
      "step": 256
    },
    {
      "epoch": 0.065792,
      "grad_norm": 0.627737701805962,
      "learning_rate": 3.968985138684947e-05,
      "loss": 1.595,
      "step": 257
    },
    {
      "epoch": 0.066048,
      "grad_norm": 0.7087387166396533,
      "learning_rate": 3.968699378452226e-05,
      "loss": 1.5708,
      "step": 258
    },
    {
      "epoch": 0.066304,
      "grad_norm": 0.6527691127897426,
      "learning_rate": 3.968412318183362e-05,
      "loss": 1.524,
      "step": 259
    },
    {
      "epoch": 0.06656,
      "grad_norm": 0.7232076900724689,
      "learning_rate": 3.968123958067917e-05,
      "loss": 1.5344,
      "step": 260
    },
    {
      "epoch": 0.066816,
      "grad_norm": 0.6464886874713757,
      "learning_rate": 3.9678342982963105e-05,
      "loss": 1.557,
      "step": 261
    },
    {
      "epoch": 0.067072,
      "grad_norm": 0.642111277654415,
      "learning_rate": 3.96754333905982e-05,
      "loss": 1.5299,
      "step": 262
    },
    {
      "epoch": 0.067328,
      "grad_norm": 0.720672927076375,
      "learning_rate": 3.96725108055058e-05,
      "loss": 1.5723,
      "step": 263
    },
    {
      "epoch": 0.067584,
      "grad_norm": 0.6068722626184577,
      "learning_rate": 3.966957522961586e-05,
      "loss": 1.5963,
      "step": 264
    },
    {
      "epoch": 0.06784,
      "grad_norm": 0.6410058022344428,
      "learning_rate": 3.966662666486689e-05,
      "loss": 1.5364,
      "step": 265
    },
    {
      "epoch": 0.068096,
      "grad_norm": 0.6469635873979933,
      "learning_rate": 3.966366511320598e-05,
      "loss": 1.5923,
      "step": 266
    },
    {
      "epoch": 0.068352,
      "grad_norm": 0.6507160149449662,
      "learning_rate": 3.966069057658881e-05,
      "loss": 1.5576,
      "step": 267
    },
    {
      "epoch": 0.068608,
      "grad_norm": 0.7065198705386274,
      "learning_rate": 3.96577030569796e-05,
      "loss": 1.6256,
      "step": 268
    },
    {
      "epoch": 0.068864,
      "grad_norm": 0.7234247232057596,
      "learning_rate": 3.965470255635119e-05,
      "loss": 1.5431,
      "step": 269
    },
    {
      "epoch": 0.06912,
      "grad_norm": 0.6837089532034915,
      "learning_rate": 3.965168907668496e-05,
      "loss": 1.5816,
      "step": 270
    },
    {
      "epoch": 0.069376,
      "grad_norm": 0.6261301256414645,
      "learning_rate": 3.9648662619970866e-05,
      "loss": 1.59,
      "step": 271
    },
    {
      "epoch": 0.069632,
      "grad_norm": 0.6152968498369733,
      "learning_rate": 3.964562318820744e-05,
      "loss": 1.5574,
      "step": 272
    },
    {
      "epoch": 0.069888,
      "grad_norm": 0.6163553404258021,
      "learning_rate": 3.964257078340178e-05,
      "loss": 1.5142,
      "step": 273
    },
    {
      "epoch": 0.070144,
      "grad_norm": 0.5977857213091625,
      "learning_rate": 3.963950540756955e-05,
      "loss": 1.5168,
      "step": 274
    },
    {
      "epoch": 0.0704,
      "grad_norm": 0.6719016393928003,
      "learning_rate": 3.963642706273499e-05,
      "loss": 1.581,
      "step": 275
    },
    {
      "epoch": 0.070656,
      "grad_norm": 0.7004705040443666,
      "learning_rate": 3.9633335750930874e-05,
      "loss": 1.5589,
      "step": 276
    },
    {
      "epoch": 0.070912,
      "grad_norm": 0.6755307879840072,
      "learning_rate": 3.963023147419858e-05,
      "loss": 1.5933,
      "step": 277
    },
    {
      "epoch": 0.071168,
      "grad_norm": 0.6603843133572679,
      "learning_rate": 3.9627114234588e-05,
      "loss": 1.5605,
      "step": 278
    },
    {
      "epoch": 0.071424,
      "grad_norm": 0.6593151994858428,
      "learning_rate": 3.962398403415763e-05,
      "loss": 1.6025,
      "step": 279
    },
    {
      "epoch": 0.07168,
      "grad_norm": 0.6588011410619836,
      "learning_rate": 3.9620840874974505e-05,
      "loss": 1.5318,
      "step": 280
    },
    {
      "epoch": 0.071936,
      "grad_norm": 0.635351412311554,
      "learning_rate": 3.961768475911421e-05,
      "loss": 1.5637,
      "step": 281
    },
    {
      "epoch": 0.072192,
      "grad_norm": 0.6333914275501125,
      "learning_rate": 3.9614515688660906e-05,
      "loss": 1.574,
      "step": 282
    },
    {
      "epoch": 0.072448,
      "grad_norm": 0.5984690025743336,
      "learning_rate": 3.961133366570729e-05,
      "loss": 1.5605,
      "step": 283
    },
    {
      "epoch": 0.072704,
      "grad_norm": 0.6245904528965746,
      "learning_rate": 3.960813869235462e-05,
      "loss": 1.5433,
      "step": 284
    },
    {
      "epoch": 0.07296,
      "grad_norm": 0.6343112722289147,
      "learning_rate": 3.960493077071272e-05,
      "loss": 1.5768,
      "step": 285
    },
    {
      "epoch": 0.073216,
      "grad_norm": 0.6383739522234918,
      "learning_rate": 3.960170990289992e-05,
      "loss": 1.5973,
      "step": 286
    },
    {
      "epoch": 0.073472,
      "grad_norm": 0.6175087904971406,
      "learning_rate": 3.9598476091043165e-05,
      "loss": 1.5591,
      "step": 287
    },
    {
      "epoch": 0.073728,
      "grad_norm": 0.605286476330089,
      "learning_rate": 3.9595229337277884e-05,
      "loss": 1.5594,
      "step": 288
    },
    {
      "epoch": 0.073984,
      "grad_norm": 0.6150810409680053,
      "learning_rate": 3.9591969643748094e-05,
      "loss": 1.6103,
      "step": 289
    },
    {
      "epoch": 0.07424,
      "grad_norm": 0.6015092265179158,
      "learning_rate": 3.958869701260633e-05,
      "loss": 1.5677,
      "step": 290
    },
    {
      "epoch": 0.074496,
      "grad_norm": 0.5794872235125188,
      "learning_rate": 3.95854114460137e-05,
      "loss": 1.5817,
      "step": 291
    },
    {
      "epoch": 0.074752,
      "grad_norm": 0.6355098279529137,
      "learning_rate": 3.958211294613983e-05,
      "loss": 1.6012,
      "step": 292
    },
    {
      "epoch": 0.075008,
      "grad_norm": 0.6846550048372965,
      "learning_rate": 3.957880151516289e-05,
      "loss": 1.5017,
      "step": 293
    },
    {
      "epoch": 0.075264,
      "grad_norm": 0.6284211989407984,
      "learning_rate": 3.957547715526959e-05,
      "loss": 1.5916,
      "step": 294
    },
    {
      "epoch": 0.07552,
      "grad_norm": 0.6280891054495534,
      "learning_rate": 3.957213986865519e-05,
      "loss": 1.6245,
      "step": 295
    },
    {
      "epoch": 0.075776,
      "grad_norm": 0.6723872157589631,
      "learning_rate": 3.956878965752348e-05,
      "loss": 1.569,
      "step": 296
    },
    {
      "epoch": 0.076032,
      "grad_norm": 0.593574319122968,
      "learning_rate": 3.956542652408676e-05,
      "loss": 1.5321,
      "step": 297
    },
    {
      "epoch": 0.076288,
      "grad_norm": 0.629001881571098,
      "learning_rate": 3.9562050470565896e-05,
      "loss": 1.5182,
      "step": 298
    },
    {
      "epoch": 0.076544,
      "grad_norm": 0.6026761828018135,
      "learning_rate": 3.9558661499190287e-05,
      "loss": 1.5448,
      "step": 299
    },
    {
      "epoch": 0.0768,
      "grad_norm": 0.6084003125242101,
      "learning_rate": 3.955525961219783e-05,
      "loss": 1.5386,
      "step": 300
    },
    {
      "epoch": 0.077056,
      "grad_norm": 0.7230722911197127,
      "learning_rate": 3.9551844811834984e-05,
      "loss": 1.521,
      "step": 301
    },
    {
      "epoch": 0.077312,
      "grad_norm": 0.5916391928344684,
      "learning_rate": 3.954841710035672e-05,
      "loss": 1.5479,
      "step": 302
    },
    {
      "epoch": 0.077568,
      "grad_norm": 0.5655274326827907,
      "learning_rate": 3.954497648002654e-05,
      "loss": 1.5287,
      "step": 303
    },
    {
      "epoch": 0.077824,
      "grad_norm": 0.5753627416365539,
      "learning_rate": 3.9541522953116446e-05,
      "loss": 1.5556,
      "step": 304
    },
    {
      "epoch": 0.07808,
      "grad_norm": 0.5495398557131904,
      "learning_rate": 3.953805652190701e-05,
      "loss": 1.539,
      "step": 305
    },
    {
      "epoch": 0.078336,
      "grad_norm": 0.5816682760424461,
      "learning_rate": 3.9534577188687285e-05,
      "loss": 1.5823,
      "step": 306
    },
    {
      "epoch": 0.078592,
      "grad_norm": 0.5588654675495193,
      "learning_rate": 3.9531084955754865e-05,
      "loss": 1.5869,
      "step": 307
    },
    {
      "epoch": 0.078848,
      "grad_norm": 0.5739225611433424,
      "learning_rate": 3.952757982541585e-05,
      "loss": 1.5601,
      "step": 308
    },
    {
      "epoch": 0.079104,
      "grad_norm": 0.5651453272839616,
      "learning_rate": 3.952406179998487e-05,
      "loss": 1.5167,
      "step": 309
    },
    {
      "epoch": 0.07936,
      "grad_norm": 0.5759801794621782,
      "learning_rate": 3.952053088178506e-05,
      "loss": 1.5454,
      "step": 310
    },
    {
      "epoch": 0.079616,
      "grad_norm": 0.5675550904221729,
      "learning_rate": 3.951698707314808e-05,
      "loss": 1.5442,
      "step": 311
    },
    {
      "epoch": 0.079872,
      "grad_norm": 0.5798120547386769,
      "learning_rate": 3.9513430376414065e-05,
      "loss": 1.5223,
      "step": 312
    },
    {
      "epoch": 0.080128,
      "grad_norm": 0.5610484973346276,
      "learning_rate": 3.9509860793931716e-05,
      "loss": 1.5353,
      "step": 313
    },
    {
      "epoch": 0.080384,
      "grad_norm": 0.5558679322440181,
      "learning_rate": 3.9506278328058217e-05,
      "loss": 1.5443,
      "step": 314
    },
    {
      "epoch": 0.08064,
      "grad_norm": 0.5720633332565486,
      "learning_rate": 3.950268298115925e-05,
      "loss": 1.548,
      "step": 315
    },
    {
      "epoch": 0.080896,
      "grad_norm": 0.5811075172223369,
      "learning_rate": 3.949907475560901e-05,
      "loss": 1.6164,
      "step": 316
    },
    {
      "epoch": 0.081152,
      "grad_norm": 0.5749882097168282,
      "learning_rate": 3.9495453653790207e-05,
      "loss": 1.5858,
      "step": 317
    },
    {
      "epoch": 0.081408,
      "grad_norm": 0.591646361410766,
      "learning_rate": 3.949181967809404e-05,
      "loss": 1.5514,
      "step": 318
    },
    {
      "epoch": 0.081664,
      "grad_norm": 0.5754571775610835,
      "learning_rate": 3.948817283092022e-05,
      "loss": 1.5811,
      "step": 319
    },
    {
      "epoch": 0.08192,
      "grad_norm": 0.5919672737133586,
      "learning_rate": 3.948451311467695e-05,
      "loss": 1.5182,
      "step": 320
    },
    {
      "epoch": 0.082176,
      "grad_norm": 0.5434999284296101,
      "learning_rate": 3.948084053178094e-05,
      "loss": 1.5497,
      "step": 321
    },
    {
      "epoch": 0.082432,
      "grad_norm": 0.5753099865967742,
      "learning_rate": 3.947715508465738e-05,
      "loss": 1.5614,
      "step": 322
    },
    {
      "epoch": 0.082688,
      "grad_norm": 0.5889106315903021,
      "learning_rate": 3.947345677573997e-05,
      "loss": 1.5218,
      "step": 323
    },
    {
      "epoch": 0.082944,
      "grad_norm": 0.5462459473200533,
      "learning_rate": 3.946974560747089e-05,
      "loss": 1.5016,
      "step": 324
    },
    {
      "epoch": 0.0832,
      "grad_norm": 0.5698011542861564,
      "learning_rate": 3.946602158230084e-05,
      "loss": 1.5219,
      "step": 325
    },
    {
      "epoch": 0.083456,
      "grad_norm": 0.5501883442642922,
      "learning_rate": 3.946228470268898e-05,
      "loss": 1.5709,
      "step": 326
    },
    {
      "epoch": 0.083712,
      "grad_norm": 0.6146766806342341,
      "learning_rate": 3.945853497110296e-05,
      "loss": 1.5026,
      "step": 327
    },
    {
      "epoch": 0.083968,
      "grad_norm": 0.57807403453325,
      "learning_rate": 3.945477239001893e-05,
      "loss": 1.5638,
      "step": 328
    },
    {
      "epoch": 0.084224,
      "grad_norm": 0.5651786098641621,
      "learning_rate": 3.9450996961921536e-05,
      "loss": 1.5849,
      "step": 329
    },
    {
      "epoch": 0.08448,
      "grad_norm": 0.5945279558511242,
      "learning_rate": 3.944720868930388e-05,
      "loss": 1.5437,
      "step": 330
    },
    {
      "epoch": 0.084736,
      "grad_norm": 0.5805692011478812,
      "learning_rate": 3.944340757466756e-05,
      "loss": 1.5299,
      "step": 331
    },
    {
      "epoch": 0.084992,
      "grad_norm": 0.5489198981177734,
      "learning_rate": 3.9439593620522644e-05,
      "loss": 1.5006,
      "step": 332
    },
    {
      "epoch": 0.085248,
      "grad_norm": 0.5364138535319419,
      "learning_rate": 3.9435766829387706e-05,
      "loss": 1.5268,
      "step": 333
    },
    {
      "epoch": 0.085504,
      "grad_norm": 0.5350672861028023,
      "learning_rate": 3.943192720378976e-05,
      "loss": 1.5115,
      "step": 334
    },
    {
      "epoch": 0.08576,
      "grad_norm": 0.5778481008977208,
      "learning_rate": 3.942807474626433e-05,
      "loss": 1.556,
      "step": 335
    },
    {
      "epoch": 0.086016,
      "grad_norm": 0.5836054489967638,
      "learning_rate": 3.9424209459355385e-05,
      "loss": 1.5597,
      "step": 336
    },
    {
      "epoch": 0.086272,
      "grad_norm": 0.5252916782088496,
      "learning_rate": 3.942033134561538e-05,
      "loss": 1.4676,
      "step": 337
    },
    {
      "epoch": 0.086528,
      "grad_norm": 0.5523934566599342,
      "learning_rate": 3.941644040760523e-05,
      "loss": 1.5329,
      "step": 338
    },
    {
      "epoch": 0.086784,
      "grad_norm": 0.5396239268232077,
      "learning_rate": 3.941253664789435e-05,
      "loss": 1.5304,
      "step": 339
    },
    {
      "epoch": 0.08704,
      "grad_norm": 0.5922437894247253,
      "learning_rate": 3.9408620069060584e-05,
      "loss": 1.5164,
      "step": 340
    },
    {
      "epoch": 0.087296,
      "grad_norm": 0.5716311630949655,
      "learning_rate": 3.9404690673690255e-05,
      "loss": 1.5121,
      "step": 341
    },
    {
      "epoch": 0.087552,
      "grad_norm": 0.6120273215323037,
      "learning_rate": 3.940074846437815e-05,
      "loss": 1.525,
      "step": 342
    },
    {
      "epoch": 0.087808,
      "grad_norm": 0.5668241402599921,
      "learning_rate": 3.939679344372753e-05,
      "loss": 1.4974,
      "step": 343
    },
    {
      "epoch": 0.088064,
      "grad_norm": 0.6192851077538529,
      "learning_rate": 3.9392825614350085e-05,
      "loss": 1.5145,
      "step": 344
    },
    {
      "epoch": 0.08832,
      "grad_norm": 0.607306396434443,
      "learning_rate": 3.9388844978865994e-05,
      "loss": 1.5629,
      "step": 345
    },
    {
      "epoch": 0.088576,
      "grad_norm": 0.574407967776693,
      "learning_rate": 3.938485153990388e-05,
      "loss": 1.4852,
      "step": 346
    },
    {
      "epoch": 0.088832,
      "grad_norm": 0.5829761810171102,
      "learning_rate": 3.9380845300100816e-05,
      "loss": 1.5562,
      "step": 347
    },
    {
      "epoch": 0.089088,
      "grad_norm": 0.5684824398811541,
      "learning_rate": 3.937682626210234e-05,
      "loss": 1.533,
      "step": 348
    },
    {
      "epoch": 0.089344,
      "grad_norm": 0.6066649082850831,
      "learning_rate": 3.9372794428562426e-05,
      "loss": 1.5696,
      "step": 349
    },
    {
      "epoch": 0.0896,
      "grad_norm": 0.6138752672638471,
      "learning_rate": 3.9368749802143524e-05,
      "loss": 1.5335,
      "step": 350
    },
    {
      "epoch": 0.089856,
      "grad_norm": 0.5473396472418754,
      "learning_rate": 3.93646923855165e-05,
      "loss": 1.5339,
      "step": 351
    },
    {
      "epoch": 0.090112,
      "grad_norm": 0.7374851741318385,
      "learning_rate": 3.9360622181360675e-05,
      "loss": 1.5391,
      "step": 352
    },
    {
      "epoch": 0.090368,
      "grad_norm": 0.6221145074383628,
      "learning_rate": 3.9356539192363834e-05,
      "loss": 1.5424,
      "step": 353
    },
    {
      "epoch": 0.090624,
      "grad_norm": 0.6070785891521857,
      "learning_rate": 3.935244342122219e-05,
      "loss": 1.5176,
      "step": 354
    },
    {
      "epoch": 0.09088,
      "grad_norm": 0.6117861488177805,
      "learning_rate": 3.934833487064038e-05,
      "loss": 1.5632,
      "step": 355
    },
    {
      "epoch": 0.091136,
      "grad_norm": 0.5872912565281746,
      "learning_rate": 3.9344213543331524e-05,
      "loss": 1.5766,
      "step": 356
    },
    {
      "epoch": 0.091392,
      "grad_norm": 0.6171017746113925,
      "learning_rate": 3.934007944201713e-05,
      "loss": 1.5205,
      "step": 357
    },
    {
      "epoch": 0.091648,
      "grad_norm": 0.5922959018496421,
      "learning_rate": 3.933593256942717e-05,
      "loss": 1.5189,
      "step": 358
    },
    {
      "epoch": 0.091904,
      "grad_norm": 0.5645013823801353,
      "learning_rate": 3.933177292830004e-05,
      "loss": 1.498,
      "step": 359
    },
    {
      "epoch": 0.09216,
      "grad_norm": 0.5748892967721703,
      "learning_rate": 3.932760052138258e-05,
      "loss": 1.5586,
      "step": 360
    },
    {
      "epoch": 0.092416,
      "grad_norm": 0.5550513608401035,
      "learning_rate": 3.932341535143003e-05,
      "loss": 1.4637,
      "step": 361
    },
    {
      "epoch": 0.092672,
      "grad_norm": 0.5778780256127651,
      "learning_rate": 3.931921742120611e-05,
      "loss": 1.5315,
      "step": 362
    },
    {
      "epoch": 0.092928,
      "grad_norm": 0.5787100847474165,
      "learning_rate": 3.931500673348291e-05,
      "loss": 1.4683,
      "step": 363
    },
    {
      "epoch": 0.093184,
      "grad_norm": 0.6816375015592419,
      "learning_rate": 3.931078329104098e-05,
      "loss": 1.5282,
      "step": 364
    },
    {
      "epoch": 0.09344,
      "grad_norm": 0.6070308439722151,
      "learning_rate": 3.930654709666927e-05,
      "loss": 1.4843,
      "step": 365
    },
    {
      "epoch": 0.093696,
      "grad_norm": 0.611855774907062,
      "learning_rate": 3.930229815316517e-05,
      "loss": 1.507,
      "step": 366
    },
    {
      "epoch": 0.093952,
      "grad_norm": 0.5920806948004058,
      "learning_rate": 3.92980364633345e-05,
      "loss": 1.5102,
      "step": 367
    },
    {
      "epoch": 0.094208,
      "grad_norm": 0.5789481306796512,
      "learning_rate": 3.9293762029991454e-05,
      "loss": 1.4859,
      "step": 368
    },
    {
      "epoch": 0.094464,
      "grad_norm": 0.5545210379685646,
      "learning_rate": 3.9289474855958675e-05,
      "loss": 1.5334,
      "step": 369
    },
    {
      "epoch": 0.09472,
      "grad_norm": 0.5693708754070363,
      "learning_rate": 3.928517494406721e-05,
      "loss": 1.5007,
      "step": 370
    },
    {
      "epoch": 0.094976,
      "grad_norm": 0.5631181720644017,
      "learning_rate": 3.928086229715652e-05,
      "loss": 1.5516,
      "step": 371
    },
    {
      "epoch": 0.095232,
      "grad_norm": 0.5698899536539186,
      "learning_rate": 3.927653691807447e-05,
      "loss": 1.5632,
      "step": 372
    },
    {
      "epoch": 0.095488,
      "grad_norm": 0.5411018284376289,
      "learning_rate": 3.927219880967733e-05,
      "loss": 1.5356,
      "step": 373
    },
    {
      "epoch": 0.095744,
      "grad_norm": 0.5720335654312722,
      "learning_rate": 3.92678479748298e-05,
      "loss": 1.495,
      "step": 374
    },
    {
      "epoch": 0.096,
      "grad_norm": 0.5713261648400483,
      "learning_rate": 3.926348441640495e-05,
      "loss": 1.5237,
      "step": 375
    },
    {
      "epoch": 0.096256,
      "grad_norm": 0.5684117488173724,
      "learning_rate": 3.9259108137284275e-05,
      "loss": 1.5611,
      "step": 376
    },
    {
      "epoch": 0.096512,
      "grad_norm": 0.5619551640402388,
      "learning_rate": 3.9254719140357656e-05,
      "loss": 1.5373,
      "step": 377
    },
    {
      "epoch": 0.096768,
      "grad_norm": 0.5808577701688464,
      "learning_rate": 3.925031742852339e-05,
      "loss": 1.5063,
      "step": 378
    },
    {
      "epoch": 0.097024,
      "grad_norm": 0.5261948663689826,
      "learning_rate": 3.9245903004688156e-05,
      "loss": 1.5304,
      "step": 379
    },
    {
      "epoch": 0.09728,
      "grad_norm": 0.5545784152397776,
      "learning_rate": 3.9241475871767026e-05,
      "loss": 1.5237,
      "step": 380
    },
    {
      "epoch": 0.097536,
      "grad_norm": 0.5415429526824642,
      "learning_rate": 3.923703603268348e-05,
      "loss": 1.5362,
      "step": 381
    },
    {
      "epoch": 0.097792,
      "grad_norm": 0.5448497604090797,
      "learning_rate": 3.9232583490369363e-05,
      "loss": 1.4965,
      "step": 382
    },
    {
      "epoch": 0.098048,
      "grad_norm": 0.5058967944470406,
      "learning_rate": 3.922811824776494e-05,
      "loss": 1.5107,
      "step": 383
    },
    {
      "epoch": 0.098304,
      "grad_norm": 0.5658118095971683,
      "learning_rate": 3.922364030781884e-05,
      "loss": 1.4619,
      "step": 384
    },
    {
      "epoch": 0.09856,
      "grad_norm": 0.5407229307860603,
      "learning_rate": 3.921914967348807e-05,
      "loss": 1.5492,
      "step": 385
    },
    {
      "epoch": 0.098816,
      "grad_norm": 0.6391689257858224,
      "learning_rate": 3.9214646347738055e-05,
      "loss": 1.5979,
      "step": 386
    },
    {
      "epoch": 0.099072,
      "grad_norm": 0.5354689680191062,
      "learning_rate": 3.921013033354256e-05,
      "loss": 1.4744,
      "step": 387
    },
    {
      "epoch": 0.099328,
      "grad_norm": 0.5449863102426697,
      "learning_rate": 3.920560163388377e-05,
      "loss": 1.4878,
      "step": 388
    },
    {
      "epoch": 0.099584,
      "grad_norm": 0.5234049424534529,
      "learning_rate": 3.92010602517522e-05,
      "loss": 1.4809,
      "step": 389
    },
    {
      "epoch": 0.09984,
      "grad_norm": 0.5577205865798737,
      "learning_rate": 3.919650619014678e-05,
      "loss": 1.5365,
      "step": 390
    },
    {
      "epoch": 0.100096,
      "grad_norm": 0.5620202759361268,
      "learning_rate": 3.9191939452074786e-05,
      "loss": 1.5147,
      "step": 391
    },
    {
      "epoch": 0.100352,
      "grad_norm": 0.5425613914169927,
      "learning_rate": 3.9187360040551886e-05,
      "loss": 1.5106,
      "step": 392
    },
    {
      "epoch": 0.100608,
      "grad_norm": 0.5695548764067173,
      "learning_rate": 3.9182767958602104e-05,
      "loss": 1.4999,
      "step": 393
    },
    {
      "epoch": 0.100864,
      "grad_norm": 0.5402815061869861,
      "learning_rate": 3.917816320925784e-05,
      "loss": 1.5021,
      "step": 394
    },
    {
      "epoch": 0.10112,
      "grad_norm": 0.5897767546820807,
      "learning_rate": 3.917354579555984e-05,
      "loss": 1.4991,
      "step": 395
    },
    {
      "epoch": 0.101376,
      "grad_norm": 0.5548953840790886,
      "learning_rate": 3.916891572055724e-05,
      "loss": 1.5445,
      "step": 396
    },
    {
      "epoch": 0.101632,
      "grad_norm": 0.5869716128855055,
      "learning_rate": 3.916427298730751e-05,
      "loss": 1.5418,
      "step": 397
    },
    {
      "epoch": 0.101888,
      "grad_norm": 0.5846028031929945,
      "learning_rate": 3.9159617598876495e-05,
      "loss": 1.4997,
      "step": 398
    },
    {
      "epoch": 0.102144,
      "grad_norm": 0.5418112164138075,
      "learning_rate": 3.9154949558338404e-05,
      "loss": 1.4576,
      "step": 399
    },
    {
      "epoch": 0.1024,
      "grad_norm": 0.5778710196897984,
      "learning_rate": 3.915026886877578e-05,
      "loss": 1.4819,
      "step": 400
    },
    {
      "epoch": 0.102656,
      "grad_norm": 0.5655149496241458,
      "learning_rate": 3.914557553327954e-05,
      "loss": 1.4622,
      "step": 401
    },
    {
      "epoch": 0.102912,
      "grad_norm": 0.5647033525492865,
      "learning_rate": 3.914086955494893e-05,
      "loss": 1.5045,
      "step": 402
    },
    {
      "epoch": 0.103168,
      "grad_norm": 0.7016332090993277,
      "learning_rate": 3.913615093689155e-05,
      "loss": 1.5452,
      "step": 403
    },
    {
      "epoch": 0.103424,
      "grad_norm": 0.6138038403589344,
      "learning_rate": 3.9131419682223376e-05,
      "loss": 1.5164,
      "step": 404
    },
    {
      "epoch": 0.10368,
      "grad_norm": 0.5950084337866036,
      "learning_rate": 3.9126675794068686e-05,
      "loss": 1.5413,
      "step": 405
    },
    {
      "epoch": 0.103936,
      "grad_norm": 0.5549327135312124,
      "learning_rate": 3.912191927556013e-05,
      "loss": 1.5168,
      "step": 406
    },
    {
      "epoch": 0.104192,
      "grad_norm": 0.6062882109514001,
      "learning_rate": 3.911715012983868e-05,
      "loss": 1.5974,
      "step": 407
    },
    {
      "epoch": 0.104448,
      "grad_norm": 0.5412167037175736,
      "learning_rate": 3.911236836005366e-05,
      "loss": 1.506,
      "step": 408
    },
    {
      "epoch": 0.104704,
      "grad_norm": 0.6335027417269128,
      "learning_rate": 3.910757396936273e-05,
      "loss": 1.5209,
      "step": 409
    },
    {
      "epoch": 0.10496,
      "grad_norm": 0.5517968775497638,
      "learning_rate": 3.910276696093186e-05,
      "loss": 1.4858,
      "step": 410
    },
    {
      "epoch": 0.105216,
      "grad_norm": 0.6864776560063669,
      "learning_rate": 3.90979473379354e-05,
      "loss": 1.54,
      "step": 411
    },
    {
      "epoch": 0.105472,
      "grad_norm": 0.5512986659989538,
      "learning_rate": 3.909311510355598e-05,
      "loss": 1.4793,
      "step": 412
    },
    {
      "epoch": 0.105728,
      "grad_norm": 0.5330731696329115,
      "learning_rate": 3.908827026098458e-05,
      "loss": 1.4989,
      "step": 413
    },
    {
      "epoch": 0.105984,
      "grad_norm": 0.569680860290677,
      "learning_rate": 3.908341281342052e-05,
      "loss": 1.5122,
      "step": 414
    },
    {
      "epoch": 0.10624,
      "grad_norm": 0.5655454222855373,
      "learning_rate": 3.907854276407141e-05,
      "loss": 1.4959,
      "step": 415
    },
    {
      "epoch": 0.106496,
      "grad_norm": 0.5408547908496869,
      "learning_rate": 3.907366011615322e-05,
      "loss": 1.5024,
      "step": 416
    },
    {
      "epoch": 0.106752,
      "grad_norm": 0.5775268900256835,
      "learning_rate": 3.90687648728902e-05,
      "loss": 1.5268,
      "step": 417
    },
    {
      "epoch": 0.107008,
      "grad_norm": 0.5694947000413031,
      "learning_rate": 3.906385703751496e-05,
      "loss": 1.5009,
      "step": 418
    },
    {
      "epoch": 0.107264,
      "grad_norm": 0.5427655898991233,
      "learning_rate": 3.905893661326839e-05,
      "loss": 1.5121,
      "step": 419
    },
    {
      "epoch": 0.10752,
      "grad_norm": 0.5642574792361763,
      "learning_rate": 3.9054003603399706e-05,
      "loss": 1.4864,
      "step": 420
    },
    {
      "epoch": 0.107776,
      "grad_norm": 0.5416495096619954,
      "learning_rate": 3.904905801116644e-05,
      "loss": 1.5056,
      "step": 421
    },
    {
      "epoch": 0.108032,
      "grad_norm": 0.5776572675505547,
      "learning_rate": 3.904409983983442e-05,
      "loss": 1.5013,
      "step": 422
    },
    {
      "epoch": 0.108288,
      "grad_norm": 0.5234234814164501,
      "learning_rate": 3.903912909267779e-05,
      "loss": 1.5228,
      "step": 423
    },
    {
      "epoch": 0.108544,
      "grad_norm": 0.5601952743601876,
      "learning_rate": 3.9034145772979e-05,
      "loss": 1.4789,
      "step": 424
    },
    {
      "epoch": 0.1088,
      "grad_norm": 0.5202010532139367,
      "learning_rate": 3.90291498840288e-05,
      "loss": 1.5558,
      "step": 425
    },
    {
      "epoch": 0.109056,
      "grad_norm": 0.5464202938890733,
      "learning_rate": 3.9024141429126234e-05,
      "loss": 1.4734,
      "step": 426
    },
    {
      "epoch": 0.109312,
      "grad_norm": 0.5168379327374141,
      "learning_rate": 3.901912041157866e-05,
      "loss": 1.5119,
      "step": 427
    },
    {
      "epoch": 0.109568,
      "grad_norm": 0.587502511308583,
      "learning_rate": 3.9014086834701704e-05,
      "loss": 1.5305,
      "step": 428
    },
    {
      "epoch": 0.109824,
      "grad_norm": 0.5231067644776921,
      "learning_rate": 3.900904070181932e-05,
      "loss": 1.5085,
      "step": 429
    },
    {
      "epoch": 0.11008,
      "grad_norm": 0.5542602007102981,
      "learning_rate": 3.900398201626372e-05,
      "loss": 1.4959,
      "step": 430
    },
    {
      "epoch": 0.110336,
      "grad_norm": 0.5379469729738892,
      "learning_rate": 3.8998910781375443e-05,
      "loss": 1.4824,
      "step": 431
    },
    {
      "epoch": 0.110592,
      "grad_norm": 0.5207824755557448,
      "learning_rate": 3.8993827000503266e-05,
      "loss": 1.4901,
      "step": 432
    },
    {
      "epoch": 0.110848,
      "grad_norm": 0.5512719568744536,
      "learning_rate": 3.8988730677004294e-05,
      "loss": 1.515,
      "step": 433
    },
    {
      "epoch": 0.111104,
      "grad_norm": 0.5121929398097549,
      "learning_rate": 3.8983621814243903e-05,
      "loss": 1.4651,
      "step": 434
    },
    {
      "epoch": 0.11136,
      "grad_norm": 0.5634050326224734,
      "learning_rate": 3.897850041559573e-05,
      "loss": 1.493,
      "step": 435
    },
    {
      "epoch": 0.111616,
      "grad_norm": 0.5670568845406377,
      "learning_rate": 3.897336648444172e-05,
      "loss": 1.4999,
      "step": 436
    },
    {
      "epoch": 0.111872,
      "grad_norm": 0.5111885029033698,
      "learning_rate": 3.896822002417206e-05,
      "loss": 1.456,
      "step": 437
    },
    {
      "epoch": 0.112128,
      "grad_norm": 0.5590287667908839,
      "learning_rate": 3.896306103818525e-05,
      "loss": 1.4698,
      "step": 438
    },
    {
      "epoch": 0.112384,
      "grad_norm": 0.5665805787207899,
      "learning_rate": 3.895788952988803e-05,
      "loss": 1.4595,
      "step": 439
    },
    {
      "epoch": 0.11264,
      "grad_norm": 0.5219584430832925,
      "learning_rate": 3.8952705502695424e-05,
      "loss": 1.5176,
      "step": 440
    },
    {
      "epoch": 0.112896,
      "grad_norm": 0.6025385181768772,
      "learning_rate": 3.8947508960030715e-05,
      "loss": 1.5346,
      "step": 441
    },
    {
      "epoch": 0.113152,
      "grad_norm": 0.5060068922884791,
      "learning_rate": 3.894229990532545e-05,
      "loss": 1.5245,
      "step": 442
    },
    {
      "epoch": 0.113408,
      "grad_norm": 0.5843206639819737,
      "learning_rate": 3.893707834201945e-05,
      "loss": 1.538,
      "step": 443
    },
    {
      "epoch": 0.113664,
      "grad_norm": 0.5244030530306381,
      "learning_rate": 3.8931844273560796e-05,
      "loss": 1.5176,
      "step": 444
    },
    {
      "epoch": 0.11392,
      "grad_norm": 0.559066700032701,
      "learning_rate": 3.8926597703405805e-05,
      "loss": 1.5139,
      "step": 445
    },
    {
      "epoch": 0.114176,
      "grad_norm": 0.5672067440229976,
      "learning_rate": 3.892133863501907e-05,
      "loss": 1.4721,
      "step": 446
    },
    {
      "epoch": 0.114432,
      "grad_norm": 0.5614283167360169,
      "learning_rate": 3.891606707187344e-05,
      "loss": 1.5847,
      "step": 447
    },
    {
      "epoch": 0.114688,
      "grad_norm": 0.5397568887526905,
      "learning_rate": 3.891078301744999e-05,
      "loss": 1.4329,
      "step": 448
    },
    {
      "epoch": 0.114944,
      "grad_norm": 0.5588517674522153,
      "learning_rate": 3.890548647523807e-05,
      "loss": 1.4528,
      "step": 449
    },
    {
      "epoch": 0.1152,
      "grad_norm": 0.529750087303853,
      "learning_rate": 3.890017744873527e-05,
      "loss": 1.5092,
      "step": 450
    },
    {
      "epoch": 0.115456,
      "grad_norm": 0.5759067484700479,
      "learning_rate": 3.8894855941447417e-05,
      "loss": 1.5124,
      "step": 451
    },
    {
      "epoch": 0.115712,
      "grad_norm": 0.5231313240554187,
      "learning_rate": 3.888952195688858e-05,
      "loss": 1.4831,
      "step": 452
    },
    {
      "epoch": 0.115968,
      "grad_norm": 0.5129406310715484,
      "learning_rate": 3.888417549858106e-05,
      "loss": 1.4628,
      "step": 453
    },
    {
      "epoch": 0.116224,
      "grad_norm": 0.4968070937138596,
      "learning_rate": 3.887881657005544e-05,
      "loss": 1.4634,
      "step": 454
    },
    {
      "epoch": 0.11648,
      "grad_norm": 0.5019519687196918,
      "learning_rate": 3.8873445174850466e-05,
      "loss": 1.4818,
      "step": 455
    },
    {
      "epoch": 0.116736,
      "grad_norm": 0.5636711660602768,
      "learning_rate": 3.8868061316513175e-05,
      "loss": 1.5089,
      "step": 456
    },
    {
      "epoch": 0.116992,
      "grad_norm": 0.5625467989322707,
      "learning_rate": 3.886266499859881e-05,
      "loss": 1.5297,
      "step": 457
    },
    {
      "epoch": 0.117248,
      "grad_norm": 0.4986056452907962,
      "learning_rate": 3.8857256224670845e-05,
      "loss": 1.4898,
      "step": 458
    },
    {
      "epoch": 0.117504,
      "grad_norm": 0.5566384911612768,
      "learning_rate": 3.885183499830097e-05,
      "loss": 1.4806,
      "step": 459
    },
    {
      "epoch": 0.11776,
      "grad_norm": 0.5060479806690903,
      "learning_rate": 3.884640132306911e-05,
      "loss": 1.5382,
      "step": 460
    },
    {
      "epoch": 0.118016,
      "grad_norm": 0.6047495315011246,
      "learning_rate": 3.8840955202563416e-05,
      "loss": 1.5075,
      "step": 461
    },
    {
      "epoch": 0.118272,
      "grad_norm": 0.5481938397051778,
      "learning_rate": 3.8835496640380246e-05,
      "loss": 1.4574,
      "step": 462
    },
    {
      "epoch": 0.118528,
      "grad_norm": 0.5811921477566268,
      "learning_rate": 3.883002564012417e-05,
      "loss": 1.4763,
      "step": 463
    },
    {
      "epoch": 0.118784,
      "grad_norm": 0.5117611751208486,
      "learning_rate": 3.8824542205407986e-05,
      "loss": 1.5086,
      "step": 464
    },
    {
      "epoch": 0.11904,
      "grad_norm": 0.5287742632770791,
      "learning_rate": 3.881904633985269e-05,
      "loss": 1.526,
      "step": 465
    },
    {
      "epoch": 0.119296,
      "grad_norm": 0.5141498015718505,
      "learning_rate": 3.88135380470875e-05,
      "loss": 1.4689,
      "step": 466
    },
    {
      "epoch": 0.119552,
      "grad_norm": 0.5650669180391636,
      "learning_rate": 3.880801733074982e-05,
      "loss": 1.4704,
      "step": 467
    },
    {
      "epoch": 0.119808,
      "grad_norm": 0.519768278528569,
      "learning_rate": 3.8802484194485286e-05,
      "loss": 1.497,
      "step": 468
    },
    {
      "epoch": 0.120064,
      "grad_norm": 0.516148321621722,
      "learning_rate": 3.87969386419477e-05,
      "loss": 1.4748,
      "step": 469
    },
    {
      "epoch": 0.12032,
      "grad_norm": 0.5741498108887203,
      "learning_rate": 3.87913806767991e-05,
      "loss": 1.5364,
      "step": 470
    },
    {
      "epoch": 0.120576,
      "grad_norm": 0.5370193113419436,
      "learning_rate": 3.8785810302709705e-05,
      "loss": 1.4869,
      "step": 471
    },
    {
      "epoch": 0.120832,
      "grad_norm": 0.5141958953741277,
      "learning_rate": 3.878022752335792e-05,
      "loss": 1.483,
      "step": 472
    },
    {
      "epoch": 0.121088,
      "grad_norm": 0.5227953184172967,
      "learning_rate": 3.8774632342430335e-05,
      "loss": 1.5261,
      "step": 473
    },
    {
      "epoch": 0.121344,
      "grad_norm": 0.5680322108237085,
      "learning_rate": 3.876902476362178e-05,
      "loss": 1.4864,
      "step": 474
    },
    {
      "epoch": 0.1216,
      "grad_norm": 0.5148615981419378,
      "learning_rate": 3.8763404790635196e-05,
      "loss": 1.5101,
      "step": 475
    },
    {
      "epoch": 0.121856,
      "grad_norm": 0.5278168427572507,
      "learning_rate": 3.875777242718177e-05,
      "loss": 1.4869,
      "step": 476
    },
    {
      "epoch": 0.122112,
      "grad_norm": 0.5262349561788399,
      "learning_rate": 3.875212767698084e-05,
      "loss": 1.5193,
      "step": 477
    },
    {
      "epoch": 0.122368,
      "grad_norm": 0.5088167825577808,
      "learning_rate": 3.874647054375995e-05,
      "loss": 1.5216,
      "step": 478
    },
    {
      "epoch": 0.122624,
      "grad_norm": 0.515389026746513,
      "learning_rate": 3.874080103125479e-05,
      "loss": 1.5152,
      "step": 479
    },
    {
      "epoch": 0.12288,
      "grad_norm": 0.5424517329693198,
      "learning_rate": 3.873511914320922e-05,
      "loss": 1.4984,
      "step": 480
    },
    {
      "epoch": 0.123136,
      "grad_norm": 0.5246491931065252,
      "learning_rate": 3.8729424883375324e-05,
      "loss": 1.464,
      "step": 481
    },
    {
      "epoch": 0.123392,
      "grad_norm": 0.4988730498367362,
      "learning_rate": 3.872371825551331e-05,
      "loss": 1.4557,
      "step": 482
    },
    {
      "epoch": 0.123648,
      "grad_norm": 0.5029333644406483,
      "learning_rate": 3.871799926339156e-05,
      "loss": 1.4854,
      "step": 483
    },
    {
      "epoch": 0.123904,
      "grad_norm": 0.9383990698675945,
      "learning_rate": 3.8712267910786625e-05,
      "loss": 1.5167,
      "step": 484
    },
    {
      "epoch": 0.12416,
      "grad_norm": 0.5139016651407603,
      "learning_rate": 3.870652420148323e-05,
      "loss": 1.5332,
      "step": 485
    },
    {
      "epoch": 0.124416,
      "grad_norm": 0.5959931001212663,
      "learning_rate": 3.8700768139274246e-05,
      "loss": 1.4974,
      "step": 486
    },
    {
      "epoch": 0.124672,
      "grad_norm": 0.5187408664632372,
      "learning_rate": 3.86949997279607e-05,
      "loss": 1.5043,
      "step": 487
    },
    {
      "epoch": 0.124928,
      "grad_norm": 0.5443000788345685,
      "learning_rate": 3.868921897135178e-05,
      "loss": 1.5182,
      "step": 488
    },
    {
      "epoch": 0.125184,
      "grad_norm": 0.5072292115459439,
      "learning_rate": 3.868342587326483e-05,
      "loss": 1.4451,
      "step": 489
    },
    {
      "epoch": 0.12544,
      "grad_norm": 0.5518760340694461,
      "learning_rate": 3.867762043752534e-05,
      "loss": 1.5107,
      "step": 490
    },
    {
      "epoch": 0.125696,
      "grad_norm": 0.5051192418000028,
      "learning_rate": 3.867180266796693e-05,
      "loss": 1.5035,
      "step": 491
    },
    {
      "epoch": 0.125952,
      "grad_norm": 0.52273969926132,
      "learning_rate": 3.86659725684314e-05,
      "loss": 1.4893,
      "step": 492
    },
    {
      "epoch": 0.126208,
      "grad_norm": 0.5137312599365389,
      "learning_rate": 3.866013014276866e-05,
      "loss": 1.5142,
      "step": 493
    },
    {
      "epoch": 0.126464,
      "grad_norm": 0.5296237991513308,
      "learning_rate": 3.865427539483678e-05,
      "loss": 1.4812,
      "step": 494
    },
    {
      "epoch": 0.12672,
      "grad_norm": 0.5387163912432142,
      "learning_rate": 3.864840832850195e-05,
      "loss": 1.4925,
      "step": 495
    },
    {
      "epoch": 0.126976,
      "grad_norm": 0.5684312270648573,
      "learning_rate": 3.864252894763851e-05,
      "loss": 1.5178,
      "step": 496
    },
    {
      "epoch": 0.127232,
      "grad_norm": 0.5322176602251488,
      "learning_rate": 3.863663725612892e-05,
      "loss": 1.4826,
      "step": 497
    },
    {
      "epoch": 0.127488,
      "grad_norm": 0.5825864174252922,
      "learning_rate": 3.863073325786379e-05,
      "loss": 1.4694,
      "step": 498
    },
    {
      "epoch": 0.127744,
      "grad_norm": 0.5276903306974429,
      "learning_rate": 3.862481695674183e-05,
      "loss": 1.488,
      "step": 499
    },
    {
      "epoch": 0.128,
      "grad_norm": 0.5260002029782144,
      "learning_rate": 3.861888835666988e-05,
      "loss": 1.4839,
      "step": 500
    },
    {
      "epoch": 0.128256,
      "grad_norm": 0.5300556397109144,
      "learning_rate": 3.8612947461562916e-05,
      "loss": 1.4207,
      "step": 501
    },
    {
      "epoch": 0.128512,
      "grad_norm": 0.5314092141445704,
      "learning_rate": 3.8606994275344026e-05,
      "loss": 1.5032,
      "step": 502
    },
    {
      "epoch": 0.128768,
      "grad_norm": 0.5330073607818647,
      "learning_rate": 3.860102880194441e-05,
      "loss": 1.5406,
      "step": 503
    },
    {
      "epoch": 0.129024,
      "grad_norm": 0.49913744413529537,
      "learning_rate": 3.8595051045303386e-05,
      "loss": 1.4283,
      "step": 504
    },
    {
      "epoch": 0.12928,
      "grad_norm": 0.5087414500773146,
      "learning_rate": 3.858906100936838e-05,
      "loss": 1.4167,
      "step": 505
    },
    {
      "epoch": 0.129536,
      "grad_norm": 0.5152256616225279,
      "learning_rate": 3.858305869809493e-05,
      "loss": 1.4878,
      "step": 506
    },
    {
      "epoch": 0.129792,
      "grad_norm": 0.5698963239266701,
      "learning_rate": 3.857704411544668e-05,
      "loss": 1.486,
      "step": 507
    },
    {
      "epoch": 0.130048,
      "grad_norm": 0.5311600179370706,
      "learning_rate": 3.8571017265395386e-05,
      "loss": 1.4734,
      "step": 508
    },
    {
      "epoch": 0.130304,
      "grad_norm": 0.5320096482787684,
      "learning_rate": 3.8564978151920876e-05,
      "loss": 1.4958,
      "step": 509
    },
    {
      "epoch": 0.13056,
      "grad_norm": 0.5483941145995145,
      "learning_rate": 3.8558926779011105e-05,
      "loss": 1.4803,
      "step": 510
    },
    {
      "epoch": 0.130816,
      "grad_norm": 0.5204917336788906,
      "learning_rate": 3.855286315066212e-05,
      "loss": 1.5323,
      "step": 511
    },
    {
      "epoch": 0.131072,
      "grad_norm": 0.5099999328110443,
      "learning_rate": 3.8546787270878046e-05,
      "loss": 1.4984,
      "step": 512
    },
    {
      "epoch": 0.131328,
      "grad_norm": 0.5685714710884843,
      "learning_rate": 3.854069914367111e-05,
      "loss": 1.489,
      "step": 513
    },
    {
      "epoch": 0.131584,
      "grad_norm": 0.5191929125191872,
      "learning_rate": 3.8534598773061624e-05,
      "loss": 1.4582,
      "step": 514
    },
    {
      "epoch": 0.13184,
      "grad_norm": 0.49900355580993516,
      "learning_rate": 3.852848616307799e-05,
      "loss": 1.506,
      "step": 515
    },
    {
      "epoch": 0.132096,
      "grad_norm": 0.5523320180053961,
      "learning_rate": 3.8522361317756674e-05,
      "loss": 1.529,
      "step": 516
    },
    {
      "epoch": 0.132352,
      "grad_norm": 0.5312634485733235,
      "learning_rate": 3.851622424114225e-05,
      "loss": 1.4752,
      "step": 517
    },
    {
      "epoch": 0.132608,
      "grad_norm": 0.5289885881469029,
      "learning_rate": 3.851007493728735e-05,
      "loss": 1.5559,
      "step": 518
    },
    {
      "epoch": 0.132864,
      "grad_norm": 0.521808855036893,
      "learning_rate": 3.8503913410252674e-05,
      "loss": 1.4967,
      "step": 519
    },
    {
      "epoch": 0.13312,
      "grad_norm": 0.5293852897838439,
      "learning_rate": 3.849773966410701e-05,
      "loss": 1.5073,
      "step": 520
    },
    {
      "epoch": 0.133376,
      "grad_norm": 0.6082067323529153,
      "learning_rate": 3.849155370292721e-05,
      "loss": 1.4817,
      "step": 521
    },
    {
      "epoch": 0.133632,
      "grad_norm": 0.5673519866765149,
      "learning_rate": 3.848535553079819e-05,
      "loss": 1.467,
      "step": 522
    },
    {
      "epoch": 0.133888,
      "grad_norm": 0.5226145357068925,
      "learning_rate": 3.847914515181293e-05,
      "loss": 1.4625,
      "step": 523
    },
    {
      "epoch": 0.134144,
      "grad_norm": 0.5276843816043325,
      "learning_rate": 3.847292257007247e-05,
      "loss": 1.4948,
      "step": 524
    },
    {
      "epoch": 0.1344,
      "grad_norm": 0.5370620172020053,
      "learning_rate": 3.84666877896859e-05,
      "loss": 1.4742,
      "step": 525
    },
    {
      "epoch": 0.134656,
      "grad_norm": 0.5453507354738568,
      "learning_rate": 3.846044081477041e-05,
      "loss": 1.4685,
      "step": 526
    },
    {
      "epoch": 0.134912,
      "grad_norm": 0.5438162366383884,
      "learning_rate": 3.845418164945116e-05,
      "loss": 1.4708,
      "step": 527
    },
    {
      "epoch": 0.135168,
      "grad_norm": 0.5234516812112022,
      "learning_rate": 3.844791029786144e-05,
      "loss": 1.4996,
      "step": 528
    },
    {
      "epoch": 0.135424,
      "grad_norm": 0.48639610207845557,
      "learning_rate": 3.8441626764142546e-05,
      "loss": 1.4946,
      "step": 529
    },
    {
      "epoch": 0.13568,
      "grad_norm": 0.496595221426787,
      "learning_rate": 3.843533105244382e-05,
      "loss": 1.4811,
      "step": 530
    },
    {
      "epoch": 0.135936,
      "grad_norm": 0.512180114016878,
      "learning_rate": 3.842902316692266e-05,
      "loss": 1.4363,
      "step": 531
    },
    {
      "epoch": 0.136192,
      "grad_norm": 0.5141028053289458,
      "learning_rate": 3.842270311174449e-05,
      "loss": 1.4342,
      "step": 532
    },
    {
      "epoch": 0.136448,
      "grad_norm": 0.4910910351768579,
      "learning_rate": 3.8416370891082785e-05,
      "loss": 1.4621,
      "step": 533
    },
    {
      "epoch": 0.136704,
      "grad_norm": 0.5641400246998611,
      "learning_rate": 3.841002650911904e-05,
      "loss": 1.5031,
      "step": 534
    },
    {
      "epoch": 0.13696,
      "grad_norm": 0.5355794269147915,
      "learning_rate": 3.840366997004278e-05,
      "loss": 1.4621,
      "step": 535
    },
    {
      "epoch": 0.137216,
      "grad_norm": 0.5183559275271968,
      "learning_rate": 3.839730127805157e-05,
      "loss": 1.5286,
      "step": 536
    },
    {
      "epoch": 0.137472,
      "grad_norm": 0.5517484333135494,
      "learning_rate": 3.8390920437351e-05,
      "loss": 1.4496,
      "step": 537
    },
    {
      "epoch": 0.137728,
      "grad_norm": 0.5250823206171249,
      "learning_rate": 3.8384527452154656e-05,
      "loss": 1.4718,
      "step": 538
    },
    {
      "epoch": 0.137984,
      "grad_norm": 0.527386533733061,
      "learning_rate": 3.837812232668419e-05,
      "loss": 1.4355,
      "step": 539
    },
    {
      "epoch": 0.13824,
      "grad_norm": 0.5301297939746005,
      "learning_rate": 3.837170506516922e-05,
      "loss": 1.5279,
      "step": 540
    },
    {
      "epoch": 0.138496,
      "grad_norm": 0.4965953653101356,
      "learning_rate": 3.836527567184741e-05,
      "loss": 1.4578,
      "step": 541
    },
    {
      "epoch": 0.138752,
      "grad_norm": 0.4987058389995067,
      "learning_rate": 3.8358834150964435e-05,
      "loss": 1.4939,
      "step": 542
    },
    {
      "epoch": 0.139008,
      "grad_norm": 0.5199682402091151,
      "learning_rate": 3.835238050677397e-05,
      "loss": 1.5284,
      "step": 543
    },
    {
      "epoch": 0.139264,
      "grad_norm": 0.5221150016537505,
      "learning_rate": 3.83459147435377e-05,
      "loss": 1.4753,
      "step": 544
    },
    {
      "epoch": 0.13952,
      "grad_norm": 0.49808617554982365,
      "learning_rate": 3.833943686552529e-05,
      "loss": 1.4573,
      "step": 545
    },
    {
      "epoch": 0.139776,
      "grad_norm": 0.5187924396388296,
      "learning_rate": 3.8332946877014444e-05,
      "loss": 1.5393,
      "step": 546
    },
    {
      "epoch": 0.140032,
      "grad_norm": 0.518681013396759,
      "learning_rate": 3.8326444782290844e-05,
      "loss": 1.4876,
      "step": 547
    },
    {
      "epoch": 0.140288,
      "grad_norm": 0.5246208067375517,
      "learning_rate": 3.8319930585648155e-05,
      "loss": 1.5283,
      "step": 548
    },
    {
      "epoch": 0.140544,
      "grad_norm": 0.5144409469740104,
      "learning_rate": 3.831340429138805e-05,
      "loss": 1.4892,
      "step": 549
    },
    {
      "epoch": 0.1408,
      "grad_norm": 0.516972607434327,
      "learning_rate": 3.830686590382019e-05,
      "loss": 1.5056,
      "step": 550
    },
    {
      "epoch": 0.141056,
      "grad_norm": 0.5068358833896857,
      "learning_rate": 3.8300315427262213e-05,
      "loss": 1.4458,
      "step": 551
    },
    {
      "epoch": 0.141312,
      "grad_norm": 0.511934063044909,
      "learning_rate": 3.829375286603975e-05,
      "loss": 1.4648,
      "step": 552
    },
    {
      "epoch": 0.141568,
      "grad_norm": 0.5005688295598651,
      "learning_rate": 3.82871782244864e-05,
      "loss": 1.4702,
      "step": 553
    },
    {
      "epoch": 0.141824,
      "grad_norm": 0.5184019607769594,
      "learning_rate": 3.8280591506943747e-05,
      "loss": 1.4841,
      "step": 554
    },
    {
      "epoch": 0.14208,
      "grad_norm": 0.4899580643761107,
      "learning_rate": 3.827399271776135e-05,
      "loss": 1.4575,
      "step": 555
    },
    {
      "epoch": 0.142336,
      "grad_norm": 0.5289866400452454,
      "learning_rate": 3.826738186129674e-05,
      "loss": 1.4765,
      "step": 556
    },
    {
      "epoch": 0.142592,
      "grad_norm": 0.5200607580905917,
      "learning_rate": 3.826075894191541e-05,
      "loss": 1.5443,
      "step": 557
    },
    {
      "epoch": 0.142848,
      "grad_norm": 0.47575269376247736,
      "learning_rate": 3.825412396399083e-05,
      "loss": 1.4438,
      "step": 558
    },
    {
      "epoch": 0.143104,
      "grad_norm": 0.4935999190617765,
      "learning_rate": 3.824747693190442e-05,
      "loss": 1.4821,
      "step": 559
    },
    {
      "epoch": 0.14336,
      "grad_norm": 0.5003064248614711,
      "learning_rate": 3.824081785004557e-05,
      "loss": 1.48,
      "step": 560
    },
    {
      "epoch": 0.143616,
      "grad_norm": 0.5051215704107598,
      "learning_rate": 3.8234146722811615e-05,
      "loss": 1.4567,
      "step": 561
    },
    {
      "epoch": 0.143872,
      "grad_norm": 0.4982539484872333,
      "learning_rate": 3.822746355460786e-05,
      "loss": 1.5261,
      "step": 562
    },
    {
      "epoch": 0.144128,
      "grad_norm": 0.45396684160573436,
      "learning_rate": 3.8220768349847556e-05,
      "loss": 1.4624,
      "step": 563
    },
    {
      "epoch": 0.144384,
      "grad_norm": 0.4772507381881317,
      "learning_rate": 3.82140611129519e-05,
      "loss": 1.4324,
      "step": 564
    },
    {
      "epoch": 0.14464,
      "grad_norm": 0.4800115566588089,
      "learning_rate": 3.8207341848350024e-05,
      "loss": 1.4993,
      "step": 565
    },
    {
      "epoch": 0.144896,
      "grad_norm": 0.46880468290110133,
      "learning_rate": 3.820061056047903e-05,
      "loss": 1.4359,
      "step": 566
    },
    {
      "epoch": 0.145152,
      "grad_norm": 0.5906299935246266,
      "learning_rate": 3.819386725378392e-05,
      "loss": 1.4459,
      "step": 567
    },
    {
      "epoch": 0.145408,
      "grad_norm": 0.514687473031849,
      "learning_rate": 3.8187111932717685e-05,
      "loss": 1.482,
      "step": 568
    },
    {
      "epoch": 0.145664,
      "grad_norm": 0.48950218466943946,
      "learning_rate": 3.8180344601741196e-05,
      "loss": 1.4797,
      "step": 569
    },
    {
      "epoch": 0.14592,
      "grad_norm": 0.5774990145961018,
      "learning_rate": 3.817356526532329e-05,
      "loss": 1.4983,
      "step": 570
    },
    {
      "epoch": 0.146176,
      "grad_norm": 0.4805298739520118,
      "learning_rate": 3.8166773927940716e-05,
      "loss": 1.4527,
      "step": 571
    },
    {
      "epoch": 0.146432,
      "grad_norm": 0.5426702797305496,
      "learning_rate": 3.815997059407815e-05,
      "loss": 1.4959,
      "step": 572
    },
    {
      "epoch": 0.146688,
      "grad_norm": 0.5192554251827838,
      "learning_rate": 3.8153155268228205e-05,
      "loss": 1.456,
      "step": 573
    },
    {
      "epoch": 0.146944,
      "grad_norm": 0.4827986566128428,
      "learning_rate": 3.814632795489139e-05,
      "loss": 1.4796,
      "step": 574
    },
    {
      "epoch": 0.1472,
      "grad_norm": 0.5661508656846217,
      "learning_rate": 3.813948865857614e-05,
      "loss": 1.4821,
      "step": 575
    },
    {
      "epoch": 0.147456,
      "grad_norm": 0.47518334928370554,
      "learning_rate": 3.813263738379882e-05,
      "loss": 1.4686,
      "step": 576
    },
    {
      "epoch": 0.147712,
      "grad_norm": 0.5053523822492876,
      "learning_rate": 3.812577413508366e-05,
      "loss": 1.475,
      "step": 577
    },
    {
      "epoch": 0.147968,
      "grad_norm": 0.4921007803106813,
      "learning_rate": 3.811889891696285e-05,
      "loss": 1.5357,
      "step": 578
    },
    {
      "epoch": 0.148224,
      "grad_norm": 0.47279084077275696,
      "learning_rate": 3.8112011733976445e-05,
      "loss": 1.4627,
      "step": 579
    },
    {
      "epoch": 0.14848,
      "grad_norm": 0.5151988977538712,
      "learning_rate": 3.810511259067242e-05,
      "loss": 1.4658,
      "step": 580
    },
    {
      "epoch": 0.148736,
      "grad_norm": 0.5937060221422091,
      "learning_rate": 3.8098201491606643e-05,
      "loss": 1.4658,
      "step": 581
    },
    {
      "epoch": 0.148992,
      "grad_norm": 0.5033060433188645,
      "learning_rate": 3.8091278441342884e-05,
      "loss": 1.5192,
      "step": 582
    },
    {
      "epoch": 0.149248,
      "grad_norm": 0.5537971605094045,
      "learning_rate": 3.808434344445279e-05,
      "loss": 1.5079,
      "step": 583
    },
    {
      "epoch": 0.149504,
      "grad_norm": 0.5461559473122115,
      "learning_rate": 3.80773965055159e-05,
      "loss": 1.4624,
      "step": 584
    },
    {
      "epoch": 0.14976,
      "grad_norm": 0.4971135641407436,
      "learning_rate": 3.807043762911967e-05,
      "loss": 1.4506,
      "step": 585
    },
    {
      "epoch": 0.150016,
      "grad_norm": 0.5801081312766178,
      "learning_rate": 3.8063466819859384e-05,
      "loss": 1.5032,
      "step": 586
    },
    {
      "epoch": 0.150272,
      "grad_norm": 0.5342732781667683,
      "learning_rate": 3.8056484082338264e-05,
      "loss": 1.5277,
      "step": 587
    },
    {
      "epoch": 0.150528,
      "grad_norm": 0.5233499402357944,
      "learning_rate": 3.804948942116735e-05,
      "loss": 1.4667,
      "step": 588
    },
    {
      "epoch": 0.150784,
      "grad_norm": 0.4809954602085254,
      "learning_rate": 3.8042482840965614e-05,
      "loss": 1.5182,
      "step": 589
    },
    {
      "epoch": 0.15104,
      "grad_norm": 0.5378209875866703,
      "learning_rate": 3.803546434635984e-05,
      "loss": 1.4919,
      "step": 590
    },
    {
      "epoch": 0.151296,
      "grad_norm": 0.5115462527699177,
      "learning_rate": 3.8028433941984745e-05,
      "loss": 1.4873,
      "step": 591
    },
    {
      "epoch": 0.151552,
      "grad_norm": 0.48884896510217246,
      "learning_rate": 3.8021391632482846e-05,
      "loss": 1.4909,
      "step": 592
    },
    {
      "epoch": 0.151808,
      "grad_norm": 0.5063767104197124,
      "learning_rate": 3.8014337422504575e-05,
      "loss": 1.4874,
      "step": 593
    },
    {
      "epoch": 0.152064,
      "grad_norm": 0.488906627315114,
      "learning_rate": 3.8007271316708186e-05,
      "loss": 1.4503,
      "step": 594
    },
    {
      "epoch": 0.15232,
      "grad_norm": 0.48264269817230804,
      "learning_rate": 3.8000193319759806e-05,
      "loss": 1.4654,
      "step": 595
    },
    {
      "epoch": 0.152576,
      "grad_norm": 0.45909432938574485,
      "learning_rate": 3.799310343633341e-05,
      "loss": 1.4605,
      "step": 596
    },
    {
      "epoch": 0.152832,
      "grad_norm": 0.4869282154270548,
      "learning_rate": 3.798600167111082e-05,
      "loss": 1.4774,
      "step": 597
    },
    {
      "epoch": 0.153088,
      "grad_norm": 0.4678186464166902,
      "learning_rate": 3.797888802878171e-05,
      "loss": 1.4291,
      "step": 598
    },
    {
      "epoch": 0.153344,
      "grad_norm": 0.4898434772262499,
      "learning_rate": 3.7971762514043585e-05,
      "loss": 1.4803,
      "step": 599
    },
    {
      "epoch": 0.1536,
      "grad_norm": 0.45885928624571815,
      "learning_rate": 3.7964625131601816e-05,
      "loss": 1.4063,
      "step": 600
    },
    {
      "epoch": 0.153856,
      "grad_norm": 0.4638369735793162,
      "learning_rate": 3.795747588616957e-05,
      "loss": 1.4827,
      "step": 601
    },
    {
      "epoch": 0.154112,
      "grad_norm": 0.48705182866429164,
      "learning_rate": 3.7950314782467895e-05,
      "loss": 1.4864,
      "step": 602
    },
    {
      "epoch": 0.154368,
      "grad_norm": 0.44948040749515983,
      "learning_rate": 3.794314182522563e-05,
      "loss": 1.4497,
      "step": 603
    },
    {
      "epoch": 0.154624,
      "grad_norm": 0.46998441867284246,
      "learning_rate": 3.793595701917946e-05,
      "loss": 1.4646,
      "step": 604
    },
    {
      "epoch": 0.15488,
      "grad_norm": 0.47487513935304704,
      "learning_rate": 3.79287603690739e-05,
      "loss": 1.4278,
      "step": 605
    },
    {
      "epoch": 0.155136,
      "grad_norm": 0.5071892523988517,
      "learning_rate": 3.792155187966127e-05,
      "loss": 1.5223,
      "step": 606
    },
    {
      "epoch": 0.155392,
      "grad_norm": 0.47046687460328607,
      "learning_rate": 3.791433155570172e-05,
      "loss": 1.4627,
      "step": 607
    },
    {
      "epoch": 0.155648,
      "grad_norm": 0.46365220599394413,
      "learning_rate": 3.790709940196321e-05,
      "loss": 1.4587,
      "step": 608
    },
    {
      "epoch": 0.155904,
      "grad_norm": 0.4618125397640924,
      "learning_rate": 3.7899855423221506e-05,
      "loss": 1.4614,
      "step": 609
    },
    {
      "epoch": 0.15616,
      "grad_norm": 0.46405342109371617,
      "learning_rate": 3.78925996242602e-05,
      "loss": 1.4658,
      "step": 610
    },
    {
      "epoch": 0.156416,
      "grad_norm": 0.4964360980835685,
      "learning_rate": 3.7885332009870674e-05,
      "loss": 1.4925,
      "step": 611
    },
    {
      "epoch": 0.156672,
      "grad_norm": 0.4729897541533447,
      "learning_rate": 3.787805258485212e-05,
      "loss": 1.502,
      "step": 612
    },
    {
      "epoch": 0.156928,
      "grad_norm": 0.4984420734854259,
      "learning_rate": 3.787076135401153e-05,
      "loss": 1.4908,
      "step": 613
    },
    {
      "epoch": 0.157184,
      "grad_norm": 0.4725106010755061,
      "learning_rate": 3.786345832216367e-05,
      "loss": 1.4515,
      "step": 614
    },
    {
      "epoch": 0.15744,
      "grad_norm": 0.503086806244627,
      "learning_rate": 3.7856143494131147e-05,
      "loss": 1.4209,
      "step": 615
    },
    {
      "epoch": 0.157696,
      "grad_norm": 0.4759086188205728,
      "learning_rate": 3.78488168747443e-05,
      "loss": 1.4461,
      "step": 616
    },
    {
      "epoch": 0.157952,
      "grad_norm": 0.4829840820219489,
      "learning_rate": 3.78414784688413e-05,
      "loss": 1.469,
      "step": 617
    },
    {
      "epoch": 0.158208,
      "grad_norm": 0.4729769255462357,
      "learning_rate": 3.7834128281268086e-05,
      "loss": 1.4876,
      "step": 618
    },
    {
      "epoch": 0.158464,
      "grad_norm": 0.49318652325924256,
      "learning_rate": 3.782676631687836e-05,
      "loss": 1.4559,
      "step": 619
    },
    {
      "epoch": 0.15872,
      "grad_norm": 0.48267335451024324,
      "learning_rate": 3.7819392580533625e-05,
      "loss": 1.4416,
      "step": 620
    },
    {
      "epoch": 0.158976,
      "grad_norm": 0.4940922132924232,
      "learning_rate": 3.781200707710315e-05,
      "loss": 1.4512,
      "step": 621
    },
    {
      "epoch": 0.159232,
      "grad_norm": 0.5243821942738289,
      "learning_rate": 3.7804609811463965e-05,
      "loss": 1.4621,
      "step": 622
    },
    {
      "epoch": 0.159488,
      "grad_norm": 0.49669911080722495,
      "learning_rate": 3.779720078850089e-05,
      "loss": 1.4747,
      "step": 623
    },
    {
      "epoch": 0.159744,
      "grad_norm": 0.5294685222181122,
      "learning_rate": 3.7789780013106473e-05,
      "loss": 1.4648,
      "step": 624
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4945110689001128,
      "learning_rate": 3.778234749018106e-05,
      "loss": 1.4973,
      "step": 625
    },
    {
      "epoch": 0.160256,
      "grad_norm": 0.5007043208705445,
      "learning_rate": 3.777490322463273e-05,
      "loss": 1.4288,
      "step": 626
    },
    {
      "epoch": 0.160512,
      "grad_norm": 0.48830457806052846,
      "learning_rate": 3.776744722137733e-05,
      "loss": 1.461,
      "step": 627
    },
    {
      "epoch": 0.160768,
      "grad_norm": 0.48616194667411944,
      "learning_rate": 3.7759979485338454e-05,
      "loss": 1.4579,
      "step": 628
    },
    {
      "epoch": 0.161024,
      "grad_norm": 0.4982998450825503,
      "learning_rate": 3.775250002144744e-05,
      "loss": 1.4687,
      "step": 629
    },
    {
      "epoch": 0.16128,
      "grad_norm": 0.5711337803292439,
      "learning_rate": 3.774500883464336e-05,
      "loss": 1.4386,
      "step": 630
    },
    {
      "epoch": 0.161536,
      "grad_norm": 0.5159006545962662,
      "learning_rate": 3.7737505929873064e-05,
      "loss": 1.4705,
      "step": 631
    },
    {
      "epoch": 0.161792,
      "grad_norm": 0.493604946037014,
      "learning_rate": 3.77299913120911e-05,
      "loss": 1.5111,
      "step": 632
    },
    {
      "epoch": 0.162048,
      "grad_norm": 0.5143390979430464,
      "learning_rate": 3.772246498625977e-05,
      "loss": 1.4665,
      "step": 633
    },
    {
      "epoch": 0.162304,
      "grad_norm": 0.5198568541339368,
      "learning_rate": 3.771492695734911e-05,
      "loss": 1.4839,
      "step": 634
    },
    {
      "epoch": 0.16256,
      "grad_norm": 0.5330600104577458,
      "learning_rate": 3.770737723033687e-05,
      "loss": 1.4897,
      "step": 635
    },
    {
      "epoch": 0.162816,
      "grad_norm": 0.5350729225915506,
      "learning_rate": 3.769981581020854e-05,
      "loss": 1.499,
      "step": 636
    },
    {
      "epoch": 0.163072,
      "grad_norm": 0.5133390460402681,
      "learning_rate": 3.769224270195731e-05,
      "loss": 1.4116,
      "step": 637
    },
    {
      "epoch": 0.163328,
      "grad_norm": 0.5493428046404916,
      "learning_rate": 3.7684657910584126e-05,
      "loss": 1.4792,
      "step": 638
    },
    {
      "epoch": 0.163584,
      "grad_norm": 0.5007097732837082,
      "learning_rate": 3.767706144109761e-05,
      "loss": 1.4847,
      "step": 639
    },
    {
      "epoch": 0.16384,
      "grad_norm": 0.544321180685805,
      "learning_rate": 3.766945329851412e-05,
      "loss": 1.4654,
      "step": 640
    },
    {
      "epoch": 0.164096,
      "grad_norm": 0.5138555751833251,
      "learning_rate": 3.766183348785772e-05,
      "loss": 1.4376,
      "step": 641
    },
    {
      "epoch": 0.164352,
      "grad_norm": 0.5316539176387364,
      "learning_rate": 3.7654202014160156e-05,
      "loss": 1.5147,
      "step": 642
    },
    {
      "epoch": 0.164608,
      "grad_norm": 0.7945075408055546,
      "learning_rate": 3.7646558882460905e-05,
      "loss": 1.4844,
      "step": 643
    },
    {
      "epoch": 0.164864,
      "grad_norm": 0.4873574610896713,
      "learning_rate": 3.7638904097807126e-05,
      "loss": 1.496,
      "step": 644
    },
    {
      "epoch": 0.16512,
      "grad_norm": 0.495015641625426,
      "learning_rate": 3.7631237665253674e-05,
      "loss": 1.4455,
      "step": 645
    },
    {
      "epoch": 0.165376,
      "grad_norm": 0.4838969037064851,
      "learning_rate": 3.762355958986312e-05,
      "loss": 1.4342,
      "step": 646
    },
    {
      "epoch": 0.165632,
      "grad_norm": 0.4970867423330155,
      "learning_rate": 3.7615869876705665e-05,
      "loss": 1.5032,
      "step": 647
    },
    {
      "epoch": 0.165888,
      "grad_norm": 0.49865463928292414,
      "learning_rate": 3.760816853085927e-05,
      "loss": 1.4649,
      "step": 648
    },
    {
      "epoch": 0.166144,
      "grad_norm": 0.5398464692283329,
      "learning_rate": 3.760045555740951e-05,
      "loss": 1.463,
      "step": 649
    },
    {
      "epoch": 0.1664,
      "grad_norm": 0.46044969075030867,
      "learning_rate": 3.7592730961449696e-05,
      "loss": 1.4439,
      "step": 650
    },
    {
      "epoch": 0.166656,
      "grad_norm": 0.4987451269097154,
      "learning_rate": 3.7584994748080766e-05,
      "loss": 1.495,
      "step": 651
    },
    {
      "epoch": 0.166912,
      "grad_norm": 0.45642589375875126,
      "learning_rate": 3.757724692241136e-05,
      "loss": 1.4331,
      "step": 652
    },
    {
      "epoch": 0.167168,
      "grad_norm": 0.4826557350612097,
      "learning_rate": 3.7569487489557774e-05,
      "loss": 1.4376,
      "step": 653
    },
    {
      "epoch": 0.167424,
      "grad_norm": 0.5181139554632707,
      "learning_rate": 3.756171645464397e-05,
      "loss": 1.4445,
      "step": 654
    },
    {
      "epoch": 0.16768,
      "grad_norm": 0.48851146823809705,
      "learning_rate": 3.755393382280158e-05,
      "loss": 1.4673,
      "step": 655
    },
    {
      "epoch": 0.167936,
      "grad_norm": 0.4690354425745252,
      "learning_rate": 3.754613959916988e-05,
      "loss": 1.4609,
      "step": 656
    },
    {
      "epoch": 0.168192,
      "grad_norm": 0.4724438289893957,
      "learning_rate": 3.75383337888958e-05,
      "loss": 1.4474,
      "step": 657
    },
    {
      "epoch": 0.168448,
      "grad_norm": 0.44555793362281726,
      "learning_rate": 3.7530516397133945e-05,
      "loss": 1.4201,
      "step": 658
    },
    {
      "epoch": 0.168704,
      "grad_norm": 0.6212787372444118,
      "learning_rate": 3.752268742904654e-05,
      "loss": 1.4805,
      "step": 659
    },
    {
      "epoch": 0.16896,
      "grad_norm": 0.46121297168423186,
      "learning_rate": 3.7514846889803464e-05,
      "loss": 1.4796,
      "step": 660
    },
    {
      "epoch": 0.169216,
      "grad_norm": 0.47184569588969577,
      "learning_rate": 3.750699478458225e-05,
      "loss": 1.4505,
      "step": 661
    },
    {
      "epoch": 0.169472,
      "grad_norm": 0.4656954533401971,
      "learning_rate": 3.749913111856805e-05,
      "loss": 1.4444,
      "step": 662
    },
    {
      "epoch": 0.169728,
      "grad_norm": 0.5259675670817636,
      "learning_rate": 3.749125589695366e-05,
      "loss": 1.4367,
      "step": 663
    },
    {
      "epoch": 0.169984,
      "grad_norm": 0.4873328702186002,
      "learning_rate": 3.7483369124939504e-05,
      "loss": 1.4749,
      "step": 664
    },
    {
      "epoch": 0.17024,
      "grad_norm": 0.4655814623317464,
      "learning_rate": 3.7475470807733636e-05,
      "loss": 1.4381,
      "step": 665
    },
    {
      "epoch": 0.170496,
      "grad_norm": 0.4931827109254648,
      "learning_rate": 3.746756095055173e-05,
      "loss": 1.5149,
      "step": 666
    },
    {
      "epoch": 0.170752,
      "grad_norm": 0.4911179567194494,
      "learning_rate": 3.745963955861707e-05,
      "loss": 1.4626,
      "step": 667
    },
    {
      "epoch": 0.171008,
      "grad_norm": 0.45248131791078317,
      "learning_rate": 3.7451706637160594e-05,
      "loss": 1.4393,
      "step": 668
    },
    {
      "epoch": 0.171264,
      "grad_norm": 0.49779655051956917,
      "learning_rate": 3.744376219142082e-05,
      "loss": 1.4999,
      "step": 669
    },
    {
      "epoch": 0.17152,
      "grad_norm": 0.7577418007010718,
      "learning_rate": 3.743580622664387e-05,
      "loss": 1.4981,
      "step": 670
    },
    {
      "epoch": 0.171776,
      "grad_norm": 0.6362415017807813,
      "learning_rate": 3.7427838748083505e-05,
      "loss": 1.4147,
      "step": 671
    },
    {
      "epoch": 0.172032,
      "grad_norm": 0.5072735060669674,
      "learning_rate": 3.7419859761001076e-05,
      "loss": 1.4226,
      "step": 672
    },
    {
      "epoch": 0.172288,
      "grad_norm": 0.48307694022744335,
      "learning_rate": 3.7411869270665504e-05,
      "loss": 1.4619,
      "step": 673
    },
    {
      "epoch": 0.172544,
      "grad_norm": 0.4859396496396663,
      "learning_rate": 3.740386728235336e-05,
      "loss": 1.4478,
      "step": 674
    },
    {
      "epoch": 0.1728,
      "grad_norm": 0.518520163527151,
      "learning_rate": 3.739585380134875e-05,
      "loss": 1.4659,
      "step": 675
    },
    {
      "epoch": 0.173056,
      "grad_norm": 0.4793495271657428,
      "learning_rate": 3.7387828832943426e-05,
      "loss": 1.4412,
      "step": 676
    },
    {
      "epoch": 0.173312,
      "grad_norm": 0.5723031771791277,
      "learning_rate": 3.7379792382436686e-05,
      "loss": 1.4516,
      "step": 677
    },
    {
      "epoch": 0.173568,
      "grad_norm": 0.7346066886288481,
      "learning_rate": 3.737174445513542e-05,
      "loss": 1.4862,
      "step": 678
    },
    {
      "epoch": 0.173824,
      "grad_norm": 0.5081885555916911,
      "learning_rate": 3.7363685056354107e-05,
      "loss": 1.4746,
      "step": 679
    },
    {
      "epoch": 0.17408,
      "grad_norm": 0.47154578283033843,
      "learning_rate": 3.735561419141479e-05,
      "loss": 1.4833,
      "step": 680
    },
    {
      "epoch": 0.174336,
      "grad_norm": 0.4821856955099171,
      "learning_rate": 3.7347531865647085e-05,
      "loss": 1.492,
      "step": 681
    },
    {
      "epoch": 0.174592,
      "grad_norm": 0.4808741372679512,
      "learning_rate": 3.7339438084388186e-05,
      "loss": 1.4682,
      "step": 682
    },
    {
      "epoch": 0.174848,
      "grad_norm": 0.4689926710044866,
      "learning_rate": 3.733133285298283e-05,
      "loss": 1.412,
      "step": 683
    },
    {
      "epoch": 0.175104,
      "grad_norm": 0.4739136666298373,
      "learning_rate": 3.7323216176783344e-05,
      "loss": 1.4777,
      "step": 684
    },
    {
      "epoch": 0.17536,
      "grad_norm": 0.48599165640258113,
      "learning_rate": 3.731508806114959e-05,
      "loss": 1.4491,
      "step": 685
    },
    {
      "epoch": 0.175616,
      "grad_norm": 0.48200465340790805,
      "learning_rate": 3.7306948511448995e-05,
      "loss": 1.4002,
      "step": 686
    },
    {
      "epoch": 0.175872,
      "grad_norm": 0.478024669313647,
      "learning_rate": 3.729879753305653e-05,
      "loss": 1.4036,
      "step": 687
    },
    {
      "epoch": 0.176128,
      "grad_norm": 0.4702945080370846,
      "learning_rate": 3.729063513135472e-05,
      "loss": 1.4725,
      "step": 688
    },
    {
      "epoch": 0.176384,
      "grad_norm": 0.48183401358468375,
      "learning_rate": 3.7282461311733625e-05,
      "loss": 1.442,
      "step": 689
    },
    {
      "epoch": 0.17664,
      "grad_norm": 0.48111172202528113,
      "learning_rate": 3.727427607959085e-05,
      "loss": 1.4793,
      "step": 690
    },
    {
      "epoch": 0.176896,
      "grad_norm": 0.4916252800067881,
      "learning_rate": 3.726607944033155e-05,
      "loss": 1.4696,
      "step": 691
    },
    {
      "epoch": 0.177152,
      "grad_norm": 0.508033560450045,
      "learning_rate": 3.725787139936837e-05,
      "loss": 1.476,
      "step": 692
    },
    {
      "epoch": 0.177408,
      "grad_norm": 0.4771154913664944,
      "learning_rate": 3.7249651962121546e-05,
      "loss": 1.4514,
      "step": 693
    },
    {
      "epoch": 0.177664,
      "grad_norm": 0.5409437490557761,
      "learning_rate": 3.724142113401878e-05,
      "loss": 1.4154,
      "step": 694
    },
    {
      "epoch": 0.17792,
      "grad_norm": 0.48600439887575736,
      "learning_rate": 3.7233178920495337e-05,
      "loss": 1.4357,
      "step": 695
    },
    {
      "epoch": 0.178176,
      "grad_norm": 0.4738338468476675,
      "learning_rate": 3.7224925326993984e-05,
      "loss": 1.4332,
      "step": 696
    },
    {
      "epoch": 0.178432,
      "grad_norm": 0.5001509463708051,
      "learning_rate": 3.7216660358965004e-05,
      "loss": 1.4811,
      "step": 697
    },
    {
      "epoch": 0.178688,
      "grad_norm": 0.48243325721776514,
      "learning_rate": 3.720838402186619e-05,
      "loss": 1.506,
      "step": 698
    },
    {
      "epoch": 0.178944,
      "grad_norm": 0.5061747044809768,
      "learning_rate": 3.720009632116284e-05,
      "loss": 1.4843,
      "step": 699
    },
    {
      "epoch": 0.1792,
      "grad_norm": 0.49368302377920203,
      "learning_rate": 3.719179726232776e-05,
      "loss": 1.4761,
      "step": 700
    },
    {
      "epoch": 0.179456,
      "grad_norm": 0.4880094254285324,
      "learning_rate": 3.718348685084127e-05,
      "loss": 1.4306,
      "step": 701
    },
    {
      "epoch": 0.179712,
      "grad_norm": 0.5616887770597707,
      "learning_rate": 3.717516509219116e-05,
      "loss": 1.4235,
      "step": 702
    },
    {
      "epoch": 0.179968,
      "grad_norm": 0.4605491705502186,
      "learning_rate": 3.716683199187273e-05,
      "loss": 1.4769,
      "step": 703
    },
    {
      "epoch": 0.180224,
      "grad_norm": 0.4806293115270143,
      "learning_rate": 3.7158487555388774e-05,
      "loss": 1.4758,
      "step": 704
    },
    {
      "epoch": 0.18048,
      "grad_norm": 0.4491839107398391,
      "learning_rate": 3.715013178824955e-05,
      "loss": 1.4049,
      "step": 705
    },
    {
      "epoch": 0.180736,
      "grad_norm": 0.4834227028108713,
      "learning_rate": 3.7141764695972826e-05,
      "loss": 1.4269,
      "step": 706
    },
    {
      "epoch": 0.180992,
      "grad_norm": 0.4689425321819967,
      "learning_rate": 3.7133386284083827e-05,
      "loss": 1.4878,
      "step": 707
    },
    {
      "epoch": 0.181248,
      "grad_norm": 0.47414622085498964,
      "learning_rate": 3.7124996558115264e-05,
      "loss": 1.3898,
      "step": 708
    },
    {
      "epoch": 0.181504,
      "grad_norm": 0.4829954020026905,
      "learning_rate": 3.711659552360731e-05,
      "loss": 1.4791,
      "step": 709
    },
    {
      "epoch": 0.18176,
      "grad_norm": 0.5978676606850087,
      "learning_rate": 3.710818318610762e-05,
      "loss": 1.4729,
      "step": 710
    },
    {
      "epoch": 0.182016,
      "grad_norm": 0.4939727208532775,
      "learning_rate": 3.70997595511713e-05,
      "loss": 1.4198,
      "step": 711
    },
    {
      "epoch": 0.182272,
      "grad_norm": 0.49515408707729713,
      "learning_rate": 3.7091324624360915e-05,
      "loss": 1.4738,
      "step": 712
    },
    {
      "epoch": 0.182528,
      "grad_norm": 0.47417440850950404,
      "learning_rate": 3.708287841124651e-05,
      "loss": 1.4235,
      "step": 713
    },
    {
      "epoch": 0.182784,
      "grad_norm": 0.5255752273278199,
      "learning_rate": 3.707442091740555e-05,
      "loss": 1.4892,
      "step": 714
    },
    {
      "epoch": 0.18304,
      "grad_norm": 0.5000673101521823,
      "learning_rate": 3.7065952148422964e-05,
      "loss": 1.4919,
      "step": 715
    },
    {
      "epoch": 0.183296,
      "grad_norm": 0.5019433803860871,
      "learning_rate": 3.705747210989114e-05,
      "loss": 1.4335,
      "step": 716
    },
    {
      "epoch": 0.183552,
      "grad_norm": 0.5053894054722073,
      "learning_rate": 3.704898080740989e-05,
      "loss": 1.4196,
      "step": 717
    },
    {
      "epoch": 0.183808,
      "grad_norm": 0.48255015197509354,
      "learning_rate": 3.7040478246586455e-05,
      "loss": 1.4798,
      "step": 718
    },
    {
      "epoch": 0.184064,
      "grad_norm": 0.49431846126325724,
      "learning_rate": 3.703196443303554e-05,
      "loss": 1.4619,
      "step": 719
    },
    {
      "epoch": 0.18432,
      "grad_norm": 0.4650252165379995,
      "learning_rate": 3.702343937237927e-05,
      "loss": 1.4242,
      "step": 720
    },
    {
      "epoch": 0.184576,
      "grad_norm": 0.49721326666157056,
      "learning_rate": 3.701490307024718e-05,
      "loss": 1.4753,
      "step": 721
    },
    {
      "epoch": 0.184832,
      "grad_norm": 0.477050808015703,
      "learning_rate": 3.700635553227624e-05,
      "loss": 1.4541,
      "step": 722
    },
    {
      "epoch": 0.185088,
      "grad_norm": 0.4837377632149249,
      "learning_rate": 3.699779676411085e-05,
      "loss": 1.4864,
      "step": 723
    },
    {
      "epoch": 0.185344,
      "grad_norm": 0.4791408518794529,
      "learning_rate": 3.698922677140282e-05,
      "loss": 1.4349,
      "step": 724
    },
    {
      "epoch": 0.1856,
      "grad_norm": 0.4766021409557764,
      "learning_rate": 3.698064555981136e-05,
      "loss": 1.4612,
      "step": 725
    },
    {
      "epoch": 0.185856,
      "grad_norm": 0.49333610372377656,
      "learning_rate": 3.6972053135003095e-05,
      "loss": 1.4764,
      "step": 726
    },
    {
      "epoch": 0.186112,
      "grad_norm": 0.46686646062848547,
      "learning_rate": 3.6963449502652065e-05,
      "loss": 1.4979,
      "step": 727
    },
    {
      "epoch": 0.186368,
      "grad_norm": 0.5033261014249186,
      "learning_rate": 3.6954834668439696e-05,
      "loss": 1.4581,
      "step": 728
    },
    {
      "epoch": 0.186624,
      "grad_norm": 0.46343053384391353,
      "learning_rate": 3.694620863805481e-05,
      "loss": 1.4686,
      "step": 729
    },
    {
      "epoch": 0.18688,
      "grad_norm": 0.47011260118187753,
      "learning_rate": 3.6937571417193655e-05,
      "loss": 1.4574,
      "step": 730
    },
    {
      "epoch": 0.187136,
      "grad_norm": 0.47261326716617896,
      "learning_rate": 3.6928923011559826e-05,
      "loss": 1.4756,
      "step": 731
    },
    {
      "epoch": 0.187392,
      "grad_norm": 0.49385632479513597,
      "learning_rate": 3.6920263426864326e-05,
      "loss": 1.4633,
      "step": 732
    },
    {
      "epoch": 0.187648,
      "grad_norm": 0.47960057867052397,
      "learning_rate": 3.691159266882553e-05,
      "loss": 1.4542,
      "step": 733
    },
    {
      "epoch": 0.187904,
      "grad_norm": 0.533628733946529,
      "learning_rate": 3.690291074316921e-05,
      "loss": 1.398,
      "step": 734
    },
    {
      "epoch": 0.18816,
      "grad_norm": 0.4724189989723257,
      "learning_rate": 3.689421765562848e-05,
      "loss": 1.4341,
      "step": 735
    },
    {
      "epoch": 0.188416,
      "grad_norm": 0.46042216846702017,
      "learning_rate": 3.688551341194387e-05,
      "loss": 1.4933,
      "step": 736
    },
    {
      "epoch": 0.188672,
      "grad_norm": 0.5143585371098637,
      "learning_rate": 3.6876798017863234e-05,
      "loss": 1.4663,
      "step": 737
    },
    {
      "epoch": 0.188928,
      "grad_norm": 0.470956616666743,
      "learning_rate": 3.686807147914181e-05,
      "loss": 1.4381,
      "step": 738
    },
    {
      "epoch": 0.189184,
      "grad_norm": 0.47002703687021247,
      "learning_rate": 3.68593338015422e-05,
      "loss": 1.4263,
      "step": 739
    },
    {
      "epoch": 0.18944,
      "grad_norm": 0.48844803001338105,
      "learning_rate": 3.685058499083434e-05,
      "loss": 1.4685,
      "step": 740
    },
    {
      "epoch": 0.189696,
      "grad_norm": 0.46472404799056205,
      "learning_rate": 3.684182505279555e-05,
      "loss": 1.4374,
      "step": 741
    },
    {
      "epoch": 0.189952,
      "grad_norm": 0.4905148059461764,
      "learning_rate": 3.6833053993210454e-05,
      "loss": 1.4932,
      "step": 742
    },
    {
      "epoch": 0.190208,
      "grad_norm": 0.4830046226099318,
      "learning_rate": 3.682427181787107e-05,
      "loss": 1.44,
      "step": 743
    },
    {
      "epoch": 0.190464,
      "grad_norm": 0.46966230620668026,
      "learning_rate": 3.681547853257673e-05,
      "loss": 1.4665,
      "step": 744
    },
    {
      "epoch": 0.19072,
      "grad_norm": 0.47445681840910275,
      "learning_rate": 3.6806674143134084e-05,
      "loss": 1.4422,
      "step": 745
    },
    {
      "epoch": 0.190976,
      "grad_norm": 0.4592002328987519,
      "learning_rate": 3.679785865535716e-05,
      "loss": 1.4172,
      "step": 746
    },
    {
      "epoch": 0.191232,
      "grad_norm": 0.49365903145792833,
      "learning_rate": 3.6789032075067276e-05,
      "loss": 1.473,
      "step": 747
    },
    {
      "epoch": 0.191488,
      "grad_norm": 0.4763355892831194,
      "learning_rate": 3.678019440809308e-05,
      "loss": 1.4549,
      "step": 748
    },
    {
      "epoch": 0.191744,
      "grad_norm": 0.45940345523933357,
      "learning_rate": 3.677134566027057e-05,
      "loss": 1.4833,
      "step": 749
    },
    {
      "epoch": 0.192,
      "grad_norm": 0.4830970937752208,
      "learning_rate": 3.676248583744304e-05,
      "loss": 1.4435,
      "step": 750
    },
    {
      "epoch": 0.192256,
      "grad_norm": 0.4512327338076334,
      "learning_rate": 3.675361494546108e-05,
      "loss": 1.4113,
      "step": 751
    },
    {
      "epoch": 0.192512,
      "grad_norm": 0.44796981375520895,
      "learning_rate": 3.674473299018263e-05,
      "loss": 1.435,
      "step": 752
    },
    {
      "epoch": 0.192768,
      "grad_norm": 0.44441736502781654,
      "learning_rate": 3.67358399774729e-05,
      "loss": 1.379,
      "step": 753
    },
    {
      "epoch": 0.193024,
      "grad_norm": 0.45911451601843406,
      "learning_rate": 3.672693591320441e-05,
      "loss": 1.4421,
      "step": 754
    },
    {
      "epoch": 0.19328,
      "grad_norm": 0.4610800402343418,
      "learning_rate": 3.671802080325701e-05,
      "loss": 1.4609,
      "step": 755
    },
    {
      "epoch": 0.193536,
      "grad_norm": 0.46315170781973786,
      "learning_rate": 3.6709094653517795e-05,
      "loss": 1.435,
      "step": 756
    },
    {
      "epoch": 0.193792,
      "grad_norm": 0.46377093430214594,
      "learning_rate": 3.670015746988118e-05,
      "loss": 1.4343,
      "step": 757
    },
    {
      "epoch": 0.194048,
      "grad_norm": 0.4510685806409105,
      "learning_rate": 3.6691209258248856e-05,
      "loss": 1.4727,
      "step": 758
    },
    {
      "epoch": 0.194304,
      "grad_norm": 0.45491905747998546,
      "learning_rate": 3.6682250024529806e-05,
      "loss": 1.4532,
      "step": 759
    },
    {
      "epoch": 0.19456,
      "grad_norm": 0.45986555720298133,
      "learning_rate": 3.667327977464028e-05,
      "loss": 1.4567,
      "step": 760
    },
    {
      "epoch": 0.194816,
      "grad_norm": 0.4781624059016721,
      "learning_rate": 3.666429851450381e-05,
      "loss": 1.4763,
      "step": 761
    },
    {
      "epoch": 0.195072,
      "grad_norm": 0.46519209759210833,
      "learning_rate": 3.66553062500512e-05,
      "loss": 1.4453,
      "step": 762
    },
    {
      "epoch": 0.195328,
      "grad_norm": 0.45677026801590037,
      "learning_rate": 3.66463029872205e-05,
      "loss": 1.4327,
      "step": 763
    },
    {
      "epoch": 0.195584,
      "grad_norm": 0.47067501655346805,
      "learning_rate": 3.663728873195706e-05,
      "loss": 1.4398,
      "step": 764
    },
    {
      "epoch": 0.19584,
      "grad_norm": 0.45734875871697284,
      "learning_rate": 3.662826349021346e-05,
      "loss": 1.4768,
      "step": 765
    },
    {
      "epoch": 0.196096,
      "grad_norm": 0.46940587574844156,
      "learning_rate": 3.6619227267949546e-05,
      "loss": 1.4841,
      "step": 766
    },
    {
      "epoch": 0.196352,
      "grad_norm": 0.5151508689949047,
      "learning_rate": 3.6610180071132406e-05,
      "loss": 1.4208,
      "step": 767
    },
    {
      "epoch": 0.196608,
      "grad_norm": 0.45931076078920907,
      "learning_rate": 3.660112190573639e-05,
      "loss": 1.4526,
      "step": 768
    },
    {
      "epoch": 0.196864,
      "grad_norm": 0.5532466662592809,
      "learning_rate": 3.659205277774308e-05,
      "loss": 1.4271,
      "step": 769
    },
    {
      "epoch": 0.19712,
      "grad_norm": 0.4545172639445102,
      "learning_rate": 3.65829726931413e-05,
      "loss": 1.426,
      "step": 770
    },
    {
      "epoch": 0.197376,
      "grad_norm": 0.4508756295326525,
      "learning_rate": 3.657388165792711e-05,
      "loss": 1.4362,
      "step": 771
    },
    {
      "epoch": 0.197632,
      "grad_norm": 0.4588418128081962,
      "learning_rate": 3.6564779678103795e-05,
      "loss": 1.4734,
      "step": 772
    },
    {
      "epoch": 0.197888,
      "grad_norm": 0.46396276381331936,
      "learning_rate": 3.6555666759681875e-05,
      "loss": 1.4817,
      "step": 773
    },
    {
      "epoch": 0.198144,
      "grad_norm": 0.4418862757637594,
      "learning_rate": 3.6546542908679105e-05,
      "loss": 1.4256,
      "step": 774
    },
    {
      "epoch": 0.1984,
      "grad_norm": 0.44269696032275824,
      "learning_rate": 3.6537408131120425e-05,
      "loss": 1.37,
      "step": 775
    },
    {
      "epoch": 0.198656,
      "grad_norm": 0.47068728215224725,
      "learning_rate": 3.652826243303802e-05,
      "loss": 1.4272,
      "step": 776
    },
    {
      "epoch": 0.198912,
      "grad_norm": 0.447381755116166,
      "learning_rate": 3.651910582047128e-05,
      "loss": 1.4493,
      "step": 777
    },
    {
      "epoch": 0.199168,
      "grad_norm": 0.45861520130904376,
      "learning_rate": 3.6509938299466796e-05,
      "loss": 1.4452,
      "step": 778
    },
    {
      "epoch": 0.199424,
      "grad_norm": 0.44731709964897715,
      "learning_rate": 3.650075987607837e-05,
      "loss": 1.4562,
      "step": 779
    },
    {
      "epoch": 0.19968,
      "grad_norm": 0.44168344296001966,
      "learning_rate": 3.649157055636699e-05,
      "loss": 1.432,
      "step": 780
    },
    {
      "epoch": 0.199936,
      "grad_norm": 0.45551124959587086,
      "learning_rate": 3.6482370346400866e-05,
      "loss": 1.4499,
      "step": 781
    },
    {
      "epoch": 0.200192,
      "grad_norm": 0.4680595228925858,
      "learning_rate": 3.6473159252255366e-05,
      "loss": 1.5188,
      "step": 782
    },
    {
      "epoch": 0.200448,
      "grad_norm": 0.45692743235566396,
      "learning_rate": 3.646393728001306e-05,
      "loss": 1.4702,
      "step": 783
    },
    {
      "epoch": 0.200704,
      "grad_norm": 0.44747454106392787,
      "learning_rate": 3.645470443576372e-05,
      "loss": 1.441,
      "step": 784
    },
    {
      "epoch": 0.20096,
      "grad_norm": 0.44838152911562196,
      "learning_rate": 3.6445460725604275e-05,
      "loss": 1.4515,
      "step": 785
    },
    {
      "epoch": 0.201216,
      "grad_norm": 0.44406753895378415,
      "learning_rate": 3.6436206155638826e-05,
      "loss": 1.4203,
      "step": 786
    },
    {
      "epoch": 0.201472,
      "grad_norm": 0.4701345650128569,
      "learning_rate": 3.642694073197866e-05,
      "loss": 1.4326,
      "step": 787
    },
    {
      "epoch": 0.201728,
      "grad_norm": 0.4564607821441837,
      "learning_rate": 3.641766446074223e-05,
      "loss": 1.4644,
      "step": 788
    },
    {
      "epoch": 0.201984,
      "grad_norm": 0.5001405194965799,
      "learning_rate": 3.640837734805513e-05,
      "loss": 1.3867,
      "step": 789
    },
    {
      "epoch": 0.20224,
      "grad_norm": 0.45280261818981854,
      "learning_rate": 3.6399079400050155e-05,
      "loss": 1.4541,
      "step": 790
    },
    {
      "epoch": 0.202496,
      "grad_norm": 0.4575027220844858,
      "learning_rate": 3.6389770622867216e-05,
      "loss": 1.4884,
      "step": 791
    },
    {
      "epoch": 0.202752,
      "grad_norm": 0.4559779231466351,
      "learning_rate": 3.6380451022653396e-05,
      "loss": 1.4331,
      "step": 792
    },
    {
      "epoch": 0.203008,
      "grad_norm": 0.452317123948773,
      "learning_rate": 3.637112060556291e-05,
      "loss": 1.4153,
      "step": 793
    },
    {
      "epoch": 0.203264,
      "grad_norm": 0.4646911337464457,
      "learning_rate": 3.6361779377757135e-05,
      "loss": 1.4635,
      "step": 794
    },
    {
      "epoch": 0.20352,
      "grad_norm": 0.4635149832446141,
      "learning_rate": 3.635242734540458e-05,
      "loss": 1.4731,
      "step": 795
    },
    {
      "epoch": 0.203776,
      "grad_norm": 0.45884534851476744,
      "learning_rate": 3.634306451468087e-05,
      "loss": 1.4468,
      "step": 796
    },
    {
      "epoch": 0.204032,
      "grad_norm": 0.4544762431441358,
      "learning_rate": 3.633369089176879e-05,
      "loss": 1.414,
      "step": 797
    },
    {
      "epoch": 0.204288,
      "grad_norm": 0.474944708134245,
      "learning_rate": 3.632430648285824e-05,
      "loss": 1.4567,
      "step": 798
    },
    {
      "epoch": 0.204544,
      "grad_norm": 0.44940316172856604,
      "learning_rate": 3.631491129414622e-05,
      "loss": 1.4213,
      "step": 799
    },
    {
      "epoch": 0.2048,
      "grad_norm": 0.4548769371713411,
      "learning_rate": 3.630550533183689e-05,
      "loss": 1.4889,
      "step": 800
    },
    {
      "epoch": 0.205056,
      "grad_norm": 0.45162231427773136,
      "learning_rate": 3.62960886021415e-05,
      "loss": 1.4544,
      "step": 801
    },
    {
      "epoch": 0.205312,
      "grad_norm": 0.45503051541640666,
      "learning_rate": 3.628666111127841e-05,
      "loss": 1.4252,
      "step": 802
    },
    {
      "epoch": 0.205568,
      "grad_norm": 0.46646098174781164,
      "learning_rate": 3.627722286547309e-05,
      "loss": 1.4586,
      "step": 803
    },
    {
      "epoch": 0.205824,
      "grad_norm": 0.5207617344446815,
      "learning_rate": 3.6267773870958116e-05,
      "loss": 1.4168,
      "step": 804
    },
    {
      "epoch": 0.20608,
      "grad_norm": 0.4348409570811274,
      "learning_rate": 3.625831413397315e-05,
      "loss": 1.4237,
      "step": 805
    },
    {
      "epoch": 0.206336,
      "grad_norm": 0.5123807862037675,
      "learning_rate": 3.624884366076496e-05,
      "loss": 1.448,
      "step": 806
    },
    {
      "epoch": 0.206592,
      "grad_norm": 0.4471707287448433,
      "learning_rate": 3.6239362457587405e-05,
      "loss": 1.4571,
      "step": 807
    },
    {
      "epoch": 0.206848,
      "grad_norm": 0.4767021845627687,
      "learning_rate": 3.622987053070143e-05,
      "loss": 1.4618,
      "step": 808
    },
    {
      "epoch": 0.207104,
      "grad_norm": 0.4736383286213043,
      "learning_rate": 3.6220367886375025e-05,
      "loss": 1.4976,
      "step": 809
    },
    {
      "epoch": 0.20736,
      "grad_norm": 0.5230317497824801,
      "learning_rate": 3.621085453088332e-05,
      "loss": 1.4596,
      "step": 810
    },
    {
      "epoch": 0.207616,
      "grad_norm": 0.459736186899135,
      "learning_rate": 3.620133047050846e-05,
      "loss": 1.4626,
      "step": 811
    },
    {
      "epoch": 0.207872,
      "grad_norm": 0.4504462983907288,
      "learning_rate": 3.619179571153971e-05,
      "loss": 1.4321,
      "step": 812
    },
    {
      "epoch": 0.208128,
      "grad_norm": 0.4660932448028296,
      "learning_rate": 3.618225026027336e-05,
      "loss": 1.439,
      "step": 813
    },
    {
      "epoch": 0.208384,
      "grad_norm": 0.44896562896314646,
      "learning_rate": 3.617269412301278e-05,
      "loss": 1.4705,
      "step": 814
    },
    {
      "epoch": 0.20864,
      "grad_norm": 0.45092059364560955,
      "learning_rate": 3.616312730606839e-05,
      "loss": 1.448,
      "step": 815
    },
    {
      "epoch": 0.208896,
      "grad_norm": 0.4619945074066421,
      "learning_rate": 3.6153549815757665e-05,
      "loss": 1.4701,
      "step": 816
    },
    {
      "epoch": 0.209152,
      "grad_norm": 0.4638576733714117,
      "learning_rate": 3.6143961658405125e-05,
      "loss": 1.4519,
      "step": 817
    },
    {
      "epoch": 0.209408,
      "grad_norm": 0.48783427361540643,
      "learning_rate": 3.6134362840342336e-05,
      "loss": 1.387,
      "step": 818
    },
    {
      "epoch": 0.209664,
      "grad_norm": 0.47238950806209135,
      "learning_rate": 3.612475336790792e-05,
      "loss": 1.4555,
      "step": 819
    },
    {
      "epoch": 0.20992,
      "grad_norm": 0.4461048762613441,
      "learning_rate": 3.6115133247447495e-05,
      "loss": 1.4379,
      "step": 820
    },
    {
      "epoch": 0.210176,
      "grad_norm": 0.4497081123664542,
      "learning_rate": 3.610550248531375e-05,
      "loss": 1.417,
      "step": 821
    },
    {
      "epoch": 0.210432,
      "grad_norm": 0.4478174355250229,
      "learning_rate": 3.609586108786638e-05,
      "loss": 1.3755,
      "step": 822
    },
    {
      "epoch": 0.210688,
      "grad_norm": 0.6191801428730218,
      "learning_rate": 3.608620906147211e-05,
      "loss": 1.46,
      "step": 823
    },
    {
      "epoch": 0.210944,
      "grad_norm": 0.43865434380083596,
      "learning_rate": 3.6076546412504685e-05,
      "loss": 1.3665,
      "step": 824
    },
    {
      "epoch": 0.2112,
      "grad_norm": 0.46102716217578815,
      "learning_rate": 3.6066873147344855e-05,
      "loss": 1.4565,
      "step": 825
    },
    {
      "epoch": 0.211456,
      "grad_norm": 0.46596669761121484,
      "learning_rate": 3.60571892723804e-05,
      "loss": 1.4264,
      "step": 826
    },
    {
      "epoch": 0.211712,
      "grad_norm": 0.47108078258071157,
      "learning_rate": 3.6047494794006076e-05,
      "loss": 1.4807,
      "step": 827
    },
    {
      "epoch": 0.211968,
      "grad_norm": 0.447382530612611,
      "learning_rate": 3.603778971862368e-05,
      "loss": 1.4189,
      "step": 828
    },
    {
      "epoch": 0.212224,
      "grad_norm": 0.45889576853147074,
      "learning_rate": 3.602807405264197e-05,
      "loss": 1.4696,
      "step": 829
    },
    {
      "epoch": 0.21248,
      "grad_norm": 0.4470715047470063,
      "learning_rate": 3.6018347802476726e-05,
      "loss": 1.4355,
      "step": 830
    },
    {
      "epoch": 0.212736,
      "grad_norm": 0.44725003576501904,
      "learning_rate": 3.6008610974550695e-05,
      "loss": 1.4349,
      "step": 831
    },
    {
      "epoch": 0.212992,
      "grad_norm": 0.44511956010906895,
      "learning_rate": 3.5998863575293615e-05,
      "loss": 1.4006,
      "step": 832
    },
    {
      "epoch": 0.213248,
      "grad_norm": 0.4725224443802951,
      "learning_rate": 3.598910561114221e-05,
      "loss": 1.4823,
      "step": 833
    },
    {
      "epoch": 0.213504,
      "grad_norm": 0.4468391985029001,
      "learning_rate": 3.5979337088540196e-05,
      "loss": 1.4267,
      "step": 834
    },
    {
      "epoch": 0.21376,
      "grad_norm": 0.45018489097674963,
      "learning_rate": 3.596955801393822e-05,
      "loss": 1.4197,
      "step": 835
    },
    {
      "epoch": 0.214016,
      "grad_norm": 0.4473563636509467,
      "learning_rate": 3.595976839379394e-05,
      "loss": 1.441,
      "step": 836
    },
    {
      "epoch": 0.214272,
      "grad_norm": 0.4686279590465284,
      "learning_rate": 3.594996823457195e-05,
      "loss": 1.4295,
      "step": 837
    },
    {
      "epoch": 0.214528,
      "grad_norm": 0.4695134403853054,
      "learning_rate": 3.594015754274381e-05,
      "loss": 1.3955,
      "step": 838
    },
    {
      "epoch": 0.214784,
      "grad_norm": 0.4614248889234679,
      "learning_rate": 3.5930336324788044e-05,
      "loss": 1.4499,
      "step": 839
    },
    {
      "epoch": 0.21504,
      "grad_norm": 0.4731496378151073,
      "learning_rate": 3.592050458719012e-05,
      "loss": 1.4287,
      "step": 840
    },
    {
      "epoch": 0.215296,
      "grad_norm": 0.6173954696775312,
      "learning_rate": 3.591066233644245e-05,
      "loss": 1.4411,
      "step": 841
    },
    {
      "epoch": 0.215552,
      "grad_norm": 0.46537449849811974,
      "learning_rate": 3.590080957904439e-05,
      "loss": 1.4431,
      "step": 842
    },
    {
      "epoch": 0.215808,
      "grad_norm": 0.4863362936600211,
      "learning_rate": 3.589094632150225e-05,
      "loss": 1.4503,
      "step": 843
    },
    {
      "epoch": 0.216064,
      "grad_norm": 0.4540882864080526,
      "learning_rate": 3.5881072570329246e-05,
      "loss": 1.4193,
      "step": 844
    },
    {
      "epoch": 0.21632,
      "grad_norm": 0.47353306881464596,
      "learning_rate": 3.587118833204554e-05,
      "loss": 1.4313,
      "step": 845
    },
    {
      "epoch": 0.216576,
      "grad_norm": 0.43443280437311405,
      "learning_rate": 3.586129361317821e-05,
      "loss": 1.3893,
      "step": 846
    },
    {
      "epoch": 0.216832,
      "grad_norm": 0.46863556095653064,
      "learning_rate": 3.585138842026128e-05,
      "loss": 1.4558,
      "step": 847
    },
    {
      "epoch": 0.217088,
      "grad_norm": 0.46147326841517294,
      "learning_rate": 3.584147275983565e-05,
      "loss": 1.4208,
      "step": 848
    },
    {
      "epoch": 0.217344,
      "grad_norm": 0.4419767536875721,
      "learning_rate": 3.5831546638449176e-05,
      "loss": 1.4774,
      "step": 849
    },
    {
      "epoch": 0.2176,
      "grad_norm": 0.46709183364111473,
      "learning_rate": 3.5821610062656576e-05,
      "loss": 1.4332,
      "step": 850
    },
    {
      "epoch": 0.217856,
      "grad_norm": 0.43366345622333197,
      "learning_rate": 3.5811663039019515e-05,
      "loss": 1.4589,
      "step": 851
    },
    {
      "epoch": 0.218112,
      "grad_norm": 0.8690166031861539,
      "learning_rate": 3.580170557410653e-05,
      "loss": 1.4928,
      "step": 852
    },
    {
      "epoch": 0.218368,
      "grad_norm": 0.48403149174316906,
      "learning_rate": 3.579173767449306e-05,
      "loss": 1.4997,
      "step": 853
    },
    {
      "epoch": 0.218624,
      "grad_norm": 0.4682968680756923,
      "learning_rate": 3.5781759346761436e-05,
      "loss": 1.4918,
      "step": 854
    },
    {
      "epoch": 0.21888,
      "grad_norm": 0.4457127520054066,
      "learning_rate": 3.577177059750088e-05,
      "loss": 1.4432,
      "step": 855
    },
    {
      "epoch": 0.219136,
      "grad_norm": 0.47481852528620266,
      "learning_rate": 3.5761771433307484e-05,
      "loss": 1.4319,
      "step": 856
    },
    {
      "epoch": 0.219392,
      "grad_norm": 0.44455023741613414,
      "learning_rate": 3.575176186078423e-05,
      "loss": 1.4044,
      "step": 857
    },
    {
      "epoch": 0.219648,
      "grad_norm": 0.4384367619677731,
      "learning_rate": 3.574174188654095e-05,
      "loss": 1.3796,
      "step": 858
    },
    {
      "epoch": 0.219904,
      "grad_norm": 0.4519523412451532,
      "learning_rate": 3.573171151719439e-05,
      "loss": 1.4814,
      "step": 859
    },
    {
      "epoch": 0.22016,
      "grad_norm": 0.4316309401500639,
      "learning_rate": 3.572167075936812e-05,
      "loss": 1.421,
      "step": 860
    },
    {
      "epoch": 0.220416,
      "grad_norm": 0.4520666384599931,
      "learning_rate": 3.571161961969257e-05,
      "loss": 1.4671,
      "step": 861
    },
    {
      "epoch": 0.220672,
      "grad_norm": 0.42921942682765796,
      "learning_rate": 3.5701558104805056e-05,
      "loss": 1.4075,
      "step": 862
    },
    {
      "epoch": 0.220928,
      "grad_norm": 0.4596437013341911,
      "learning_rate": 3.5691486221349716e-05,
      "loss": 1.4885,
      "step": 863
    },
    {
      "epoch": 0.221184,
      "grad_norm": 0.4410163086601325,
      "learning_rate": 3.568140397597755e-05,
      "loss": 1.4476,
      "step": 864
    },
    {
      "epoch": 0.22144,
      "grad_norm": 0.448878647999851,
      "learning_rate": 3.5671311375346396e-05,
      "loss": 1.4394,
      "step": 865
    },
    {
      "epoch": 0.221696,
      "grad_norm": 0.4407542992747404,
      "learning_rate": 3.5661208426120936e-05,
      "loss": 1.4153,
      "step": 866
    },
    {
      "epoch": 0.221952,
      "grad_norm": 0.4471569353757589,
      "learning_rate": 3.5651095134972675e-05,
      "loss": 1.4128,
      "step": 867
    },
    {
      "epoch": 0.222208,
      "grad_norm": 0.46010435781203535,
      "learning_rate": 3.564097150857996e-05,
      "loss": 1.4565,
      "step": 868
    },
    {
      "epoch": 0.222464,
      "grad_norm": 0.4635963573709651,
      "learning_rate": 3.563083755362795e-05,
      "loss": 1.4753,
      "step": 869
    },
    {
      "epoch": 0.22272,
      "grad_norm": 0.45296636411408797,
      "learning_rate": 3.5620693276808636e-05,
      "loss": 1.43,
      "step": 870
    },
    {
      "epoch": 0.222976,
      "grad_norm": 0.48554147437404693,
      "learning_rate": 3.56105386848208e-05,
      "loss": 1.4635,
      "step": 871
    },
    {
      "epoch": 0.223232,
      "grad_norm": 0.4412503509821529,
      "learning_rate": 3.560037378437009e-05,
      "loss": 1.4546,
      "step": 872
    },
    {
      "epoch": 0.223488,
      "grad_norm": 0.4616564937072791,
      "learning_rate": 3.55901985821689e-05,
      "loss": 1.4781,
      "step": 873
    },
    {
      "epoch": 0.223744,
      "grad_norm": 0.4410249479054077,
      "learning_rate": 3.558001308493646e-05,
      "loss": 1.4383,
      "step": 874
    },
    {
      "epoch": 0.224,
      "grad_norm": 0.45713132279155816,
      "learning_rate": 3.556981729939879e-05,
      "loss": 1.4322,
      "step": 875
    },
    {
      "epoch": 0.224256,
      "grad_norm": 0.44717743337287874,
      "learning_rate": 3.555961123228871e-05,
      "loss": 1.4378,
      "step": 876
    },
    {
      "epoch": 0.224512,
      "grad_norm": 0.45932042445201127,
      "learning_rate": 3.554939489034582e-05,
      "loss": 1.4151,
      "step": 877
    },
    {
      "epoch": 0.224768,
      "grad_norm": 0.44338562280806193,
      "learning_rate": 3.5539168280316514e-05,
      "loss": 1.425,
      "step": 878
    },
    {
      "epoch": 0.225024,
      "grad_norm": 0.4652361071614257,
      "learning_rate": 3.552893140895396e-05,
      "loss": 1.4614,
      "step": 879
    },
    {
      "epoch": 0.22528,
      "grad_norm": 0.4927838123745844,
      "learning_rate": 3.551868428301811e-05,
      "loss": 1.5102,
      "step": 880
    },
    {
      "epoch": 0.225536,
      "grad_norm": 0.4658091994670612,
      "learning_rate": 3.550842690927567e-05,
      "loss": 1.4125,
      "step": 881
    },
    {
      "epoch": 0.225792,
      "grad_norm": 0.4562348811538113,
      "learning_rate": 3.549815929450015e-05,
      "loss": 1.4112,
      "step": 882
    },
    {
      "epoch": 0.226048,
      "grad_norm": 0.45055802924084537,
      "learning_rate": 3.548788144547176e-05,
      "loss": 1.4397,
      "step": 883
    },
    {
      "epoch": 0.226304,
      "grad_norm": 0.44549937773237536,
      "learning_rate": 3.547759336897755e-05,
      "loss": 1.4188,
      "step": 884
    },
    {
      "epoch": 0.22656,
      "grad_norm": 0.4518598072026557,
      "learning_rate": 3.546729507181124e-05,
      "loss": 1.4139,
      "step": 885
    },
    {
      "epoch": 0.226816,
      "grad_norm": 0.46965867493303237,
      "learning_rate": 3.545698656077336e-05,
      "loss": 1.4642,
      "step": 886
    },
    {
      "epoch": 0.227072,
      "grad_norm": 0.447646521939099,
      "learning_rate": 3.5446667842671165e-05,
      "loss": 1.4297,
      "step": 887
    },
    {
      "epoch": 0.227328,
      "grad_norm": 0.4653145025348672,
      "learning_rate": 3.5436338924318655e-05,
      "loss": 1.4619,
      "step": 888
    },
    {
      "epoch": 0.227584,
      "grad_norm": 0.44904405354389765,
      "learning_rate": 3.542599981253654e-05,
      "loss": 1.4244,
      "step": 889
    },
    {
      "epoch": 0.22784,
      "grad_norm": 0.4502027699866497,
      "learning_rate": 3.541565051415229e-05,
      "loss": 1.372,
      "step": 890
    },
    {
      "epoch": 0.228096,
      "grad_norm": 0.4573099145423407,
      "learning_rate": 3.5405291036000106e-05,
      "loss": 1.4387,
      "step": 891
    },
    {
      "epoch": 0.228352,
      "grad_norm": 0.45492528684437367,
      "learning_rate": 3.539492138492088e-05,
      "loss": 1.3732,
      "step": 892
    },
    {
      "epoch": 0.228608,
      "grad_norm": 0.44881247790223366,
      "learning_rate": 3.538454156776225e-05,
      "loss": 1.4001,
      "step": 893
    },
    {
      "epoch": 0.228864,
      "grad_norm": 0.46080139940224124,
      "learning_rate": 3.537415159137856e-05,
      "loss": 1.4577,
      "step": 894
    },
    {
      "epoch": 0.22912,
      "grad_norm": 0.4633337052245069,
      "learning_rate": 3.536375146263085e-05,
      "loss": 1.4956,
      "step": 895
    },
    {
      "epoch": 0.229376,
      "grad_norm": 0.45084353848061176,
      "learning_rate": 3.535334118838688e-05,
      "loss": 1.4344,
      "step": 896
    },
    {
      "epoch": 0.229632,
      "grad_norm": 0.47855408463606197,
      "learning_rate": 3.5342920775521105e-05,
      "loss": 1.3802,
      "step": 897
    },
    {
      "epoch": 0.229888,
      "grad_norm": 0.42919002715914023,
      "learning_rate": 3.5332490230914666e-05,
      "loss": 1.4556,
      "step": 898
    },
    {
      "epoch": 0.230144,
      "grad_norm": 0.45692038197991075,
      "learning_rate": 3.5322049561455415e-05,
      "loss": 1.4184,
      "step": 899
    },
    {
      "epoch": 0.2304,
      "grad_norm": 0.4240597464637973,
      "learning_rate": 3.531159877403787e-05,
      "loss": 1.4079,
      "step": 900
    },
    {
      "epoch": 0.230656,
      "grad_norm": 0.47541987226950216,
      "learning_rate": 3.5301137875563234e-05,
      "loss": 1.4152,
      "step": 901
    },
    {
      "epoch": 0.230912,
      "grad_norm": 0.4360433996520917,
      "learning_rate": 3.52906668729394e-05,
      "loss": 1.4499,
      "step": 902
    },
    {
      "epoch": 0.231168,
      "grad_norm": 0.4601324602891305,
      "learning_rate": 3.528018577308092e-05,
      "loss": 1.4248,
      "step": 903
    },
    {
      "epoch": 0.231424,
      "grad_norm": 0.44550797836289335,
      "learning_rate": 3.526969458290901e-05,
      "loss": 1.3967,
      "step": 904
    },
    {
      "epoch": 0.23168,
      "grad_norm": 0.46466046515146925,
      "learning_rate": 3.525919330935157e-05,
      "loss": 1.4707,
      "step": 905
    },
    {
      "epoch": 0.231936,
      "grad_norm": 0.4739550055141789,
      "learning_rate": 3.5248681959343134e-05,
      "loss": 1.4413,
      "step": 906
    },
    {
      "epoch": 0.232192,
      "grad_norm": 0.46647827174953926,
      "learning_rate": 3.523816053982491e-05,
      "loss": 1.4281,
      "step": 907
    },
    {
      "epoch": 0.232448,
      "grad_norm": 0.46069258024961396,
      "learning_rate": 3.522762905774473e-05,
      "loss": 1.4925,
      "step": 908
    },
    {
      "epoch": 0.232704,
      "grad_norm": 0.4578784012913905,
      "learning_rate": 3.5217087520057104e-05,
      "loss": 1.4534,
      "step": 909
    },
    {
      "epoch": 0.23296,
      "grad_norm": 0.4454753001477312,
      "learning_rate": 3.5206535933723176e-05,
      "loss": 1.4192,
      "step": 910
    },
    {
      "epoch": 0.233216,
      "grad_norm": 0.45421816655951225,
      "learning_rate": 3.5195974305710686e-05,
      "loss": 1.4427,
      "step": 911
    },
    {
      "epoch": 0.233472,
      "grad_norm": 0.4490127589491534,
      "learning_rate": 3.518540264299406e-05,
      "loss": 1.4016,
      "step": 912
    },
    {
      "epoch": 0.233728,
      "grad_norm": 0.4366203378136273,
      "learning_rate": 3.5174820952554315e-05,
      "loss": 1.3855,
      "step": 913
    },
    {
      "epoch": 0.233984,
      "grad_norm": 0.44468041682079235,
      "learning_rate": 3.51642292413791e-05,
      "loss": 1.4152,
      "step": 914
    },
    {
      "epoch": 0.23424,
      "grad_norm": 0.43867625176718084,
      "learning_rate": 3.5153627516462675e-05,
      "loss": 1.482,
      "step": 915
    },
    {
      "epoch": 0.234496,
      "grad_norm": 0.46086789768893743,
      "learning_rate": 3.5143015784805936e-05,
      "loss": 1.423,
      "step": 916
    },
    {
      "epoch": 0.234752,
      "grad_norm": 0.43066697230713963,
      "learning_rate": 3.513239405341635e-05,
      "loss": 1.4294,
      "step": 917
    },
    {
      "epoch": 0.235008,
      "grad_norm": 0.43799191965319134,
      "learning_rate": 3.512176232930802e-05,
      "loss": 1.4121,
      "step": 918
    },
    {
      "epoch": 0.235264,
      "grad_norm": 0.4842481596239167,
      "learning_rate": 3.5111120619501634e-05,
      "loss": 1.493,
      "step": 919
    },
    {
      "epoch": 0.23552,
      "grad_norm": 0.4279747857679062,
      "learning_rate": 3.510046893102446e-05,
      "loss": 1.4338,
      "step": 920
    },
    {
      "epoch": 0.235776,
      "grad_norm": 0.45756852139816373,
      "learning_rate": 3.508980727091039e-05,
      "loss": 1.4333,
      "step": 921
    },
    {
      "epoch": 0.236032,
      "grad_norm": 0.4486105560370268,
      "learning_rate": 3.5079135646199874e-05,
      "loss": 1.4665,
      "step": 922
    },
    {
      "epoch": 0.236288,
      "grad_norm": 0.43191304608258113,
      "learning_rate": 3.506845406393995e-05,
      "loss": 1.4383,
      "step": 923
    },
    {
      "epoch": 0.236544,
      "grad_norm": 0.4402610923494255,
      "learning_rate": 3.505776253118422e-05,
      "loss": 1.4572,
      "step": 924
    },
    {
      "epoch": 0.2368,
      "grad_norm": 0.4470400059832819,
      "learning_rate": 3.504706105499288e-05,
      "loss": 1.4452,
      "step": 925
    },
    {
      "epoch": 0.237056,
      "grad_norm": 0.4317959250025687,
      "learning_rate": 3.503634964243268e-05,
      "loss": 1.4008,
      "step": 926
    },
    {
      "epoch": 0.237312,
      "grad_norm": 0.4490832662406321,
      "learning_rate": 3.502562830057692e-05,
      "loss": 1.4472,
      "step": 927
    },
    {
      "epoch": 0.237568,
      "grad_norm": 0.4457825493520661,
      "learning_rate": 3.501489703650548e-05,
      "loss": 1.4335,
      "step": 928
    },
    {
      "epoch": 0.237824,
      "grad_norm": 0.4351634660935205,
      "learning_rate": 3.500415585730477e-05,
      "loss": 1.4003,
      "step": 929
    },
    {
      "epoch": 0.23808,
      "grad_norm": 0.4885225929546436,
      "learning_rate": 3.499340477006777e-05,
      "loss": 1.4734,
      "step": 930
    },
    {
      "epoch": 0.238336,
      "grad_norm": 0.4556235308025768,
      "learning_rate": 3.498264378189398e-05,
      "loss": 1.4339,
      "step": 931
    },
    {
      "epoch": 0.238592,
      "grad_norm": 0.4419259292218684,
      "learning_rate": 3.497187289988946e-05,
      "loss": 1.4713,
      "step": 932
    },
    {
      "epoch": 0.238848,
      "grad_norm": 0.44043595385150985,
      "learning_rate": 3.4961092131166783e-05,
      "loss": 1.3921,
      "step": 933
    },
    {
      "epoch": 0.239104,
      "grad_norm": 0.44455364945708115,
      "learning_rate": 3.495030148284506e-05,
      "loss": 1.4704,
      "step": 934
    },
    {
      "epoch": 0.23936,
      "grad_norm": 0.4420177524636941,
      "learning_rate": 3.4939500962049925e-05,
      "loss": 1.4188,
      "step": 935
    },
    {
      "epoch": 0.239616,
      "grad_norm": 0.42233265302722817,
      "learning_rate": 3.492869057591353e-05,
      "loss": 1.4258,
      "step": 936
    },
    {
      "epoch": 0.239872,
      "grad_norm": 0.4532793942780002,
      "learning_rate": 3.4917870331574555e-05,
      "loss": 1.4231,
      "step": 937
    },
    {
      "epoch": 0.240128,
      "grad_norm": 0.4349911554670856,
      "learning_rate": 3.490704023617817e-05,
      "loss": 1.4043,
      "step": 938
    },
    {
      "epoch": 0.240384,
      "grad_norm": 0.455098266486817,
      "learning_rate": 3.489620029687606e-05,
      "loss": 1.4187,
      "step": 939
    },
    {
      "epoch": 0.24064,
      "grad_norm": 0.4227489801709916,
      "learning_rate": 3.488535052082641e-05,
      "loss": 1.4038,
      "step": 940
    },
    {
      "epoch": 0.240896,
      "grad_norm": 0.4465570310466254,
      "learning_rate": 3.4874490915193904e-05,
      "loss": 1.3604,
      "step": 941
    },
    {
      "epoch": 0.241152,
      "grad_norm": 0.4278854623964549,
      "learning_rate": 3.48636214871497e-05,
      "loss": 1.4159,
      "step": 942
    },
    {
      "epoch": 0.241408,
      "grad_norm": 0.4314271056468017,
      "learning_rate": 3.485274224387147e-05,
      "loss": 1.3796,
      "step": 943
    },
    {
      "epoch": 0.241664,
      "grad_norm": 0.45032212849186587,
      "learning_rate": 3.484185319254334e-05,
      "loss": 1.4706,
      "step": 944
    },
    {
      "epoch": 0.24192,
      "grad_norm": 0.46227312038582874,
      "learning_rate": 3.483095434035594e-05,
      "loss": 1.4106,
      "step": 945
    },
    {
      "epoch": 0.242176,
      "grad_norm": 0.44790477242324117,
      "learning_rate": 3.4820045694506354e-05,
      "loss": 1.4689,
      "step": 946
    },
    {
      "epoch": 0.242432,
      "grad_norm": 0.44688685866097144,
      "learning_rate": 3.480912726219812e-05,
      "loss": 1.4331,
      "step": 947
    },
    {
      "epoch": 0.242688,
      "grad_norm": 0.45058692640925385,
      "learning_rate": 3.479819905064127e-05,
      "loss": 1.4484,
      "step": 948
    },
    {
      "epoch": 0.242944,
      "grad_norm": 0.42111017883033847,
      "learning_rate": 3.478726106705229e-05,
      "loss": 1.4302,
      "step": 949
    },
    {
      "epoch": 0.2432,
      "grad_norm": 0.43284840963226584,
      "learning_rate": 3.477631331865408e-05,
      "loss": 1.447,
      "step": 950
    },
    {
      "epoch": 0.243456,
      "grad_norm": 0.4282124874745718,
      "learning_rate": 3.476535581267604e-05,
      "loss": 1.4419,
      "step": 951
    },
    {
      "epoch": 0.243712,
      "grad_norm": 0.5000892409442088,
      "learning_rate": 3.4754388556353975e-05,
      "loss": 1.4384,
      "step": 952
    },
    {
      "epoch": 0.243968,
      "grad_norm": 0.437808391811369,
      "learning_rate": 3.474341155693015e-05,
      "loss": 1.4044,
      "step": 953
    },
    {
      "epoch": 0.244224,
      "grad_norm": 0.4261688588576915,
      "learning_rate": 3.4732424821653264e-05,
      "loss": 1.4221,
      "step": 954
    },
    {
      "epoch": 0.24448,
      "grad_norm": 0.45446959065478565,
      "learning_rate": 3.472142835777843e-05,
      "loss": 1.4587,
      "step": 955
    },
    {
      "epoch": 0.244736,
      "grad_norm": 0.4336286614191386,
      "learning_rate": 3.471042217256719e-05,
      "loss": 1.3963,
      "step": 956
    },
    {
      "epoch": 0.244992,
      "grad_norm": 0.4510498464342703,
      "learning_rate": 3.469940627328751e-05,
      "loss": 1.432,
      "step": 957
    },
    {
      "epoch": 0.245248,
      "grad_norm": 0.4336455219214784,
      "learning_rate": 3.468838066721378e-05,
      "loss": 1.4191,
      "step": 958
    },
    {
      "epoch": 0.245504,
      "grad_norm": 0.48064724030236716,
      "learning_rate": 3.4677345361626775e-05,
      "loss": 1.452,
      "step": 959
    },
    {
      "epoch": 0.24576,
      "grad_norm": 0.4274295696997247,
      "learning_rate": 3.4666300363813706e-05,
      "loss": 1.4437,
      "step": 960
    },
    {
      "epoch": 0.246016,
      "grad_norm": 0.578011845147695,
      "learning_rate": 3.465524568106816e-05,
      "loss": 1.4364,
      "step": 961
    },
    {
      "epoch": 0.246272,
      "grad_norm": 0.453456942161563,
      "learning_rate": 3.4644181320690134e-05,
      "loss": 1.4038,
      "step": 962
    },
    {
      "epoch": 0.246528,
      "grad_norm": 0.4983859308605308,
      "learning_rate": 3.4633107289985996e-05,
      "loss": 1.4394,
      "step": 963
    },
    {
      "epoch": 0.246784,
      "grad_norm": 0.473652556116915,
      "learning_rate": 3.462202359626852e-05,
      "loss": 1.4687,
      "step": 964
    },
    {
      "epoch": 0.24704,
      "grad_norm": 0.4392706175781353,
      "learning_rate": 3.461093024685686e-05,
      "loss": 1.4501,
      "step": 965
    },
    {
      "epoch": 0.247296,
      "grad_norm": 0.4493452178042283,
      "learning_rate": 3.459982724907654e-05,
      "loss": 1.428,
      "step": 966
    },
    {
      "epoch": 0.247552,
      "grad_norm": 0.434453604464245,
      "learning_rate": 3.4588714610259444e-05,
      "loss": 1.4207,
      "step": 967
    },
    {
      "epoch": 0.247808,
      "grad_norm": 0.45005027420718363,
      "learning_rate": 3.4577592337743834e-05,
      "loss": 1.3774,
      "step": 968
    },
    {
      "epoch": 0.248064,
      "grad_norm": 0.44151944570975715,
      "learning_rate": 3.456646043887435e-05,
      "loss": 1.4284,
      "step": 969
    },
    {
      "epoch": 0.24832,
      "grad_norm": 0.4537005100990312,
      "learning_rate": 3.455531892100196e-05,
      "loss": 1.4298,
      "step": 970
    },
    {
      "epoch": 0.248576,
      "grad_norm": 0.447138459673921,
      "learning_rate": 3.4544167791484e-05,
      "loss": 1.4121,
      "step": 971
    },
    {
      "epoch": 0.248832,
      "grad_norm": 0.4546885780839055,
      "learning_rate": 3.453300705768416e-05,
      "loss": 1.4408,
      "step": 972
    },
    {
      "epoch": 0.249088,
      "grad_norm": 0.4589454927084073,
      "learning_rate": 3.452183672697244e-05,
      "loss": 1.4117,
      "step": 973
    },
    {
      "epoch": 0.249344,
      "grad_norm": 0.4544498078183396,
      "learning_rate": 3.451065680672522e-05,
      "loss": 1.4374,
      "step": 974
    },
    {
      "epoch": 0.2496,
      "grad_norm": 0.4465668176926965,
      "learning_rate": 3.449946730432517e-05,
      "loss": 1.416,
      "step": 975
    },
    {
      "epoch": 0.249856,
      "grad_norm": 0.42919312920462227,
      "learning_rate": 3.448826822716133e-05,
      "loss": 1.4194,
      "step": 976
    },
    {
      "epoch": 0.250112,
      "grad_norm": 0.44610945903061555,
      "learning_rate": 3.447705958262903e-05,
      "loss": 1.4256,
      "step": 977
    },
    {
      "epoch": 0.250368,
      "grad_norm": 0.45220100030175053,
      "learning_rate": 3.446584137812993e-05,
      "loss": 1.4311,
      "step": 978
    },
    {
      "epoch": 0.250624,
      "grad_norm": 0.43055243003794796,
      "learning_rate": 3.445461362107201e-05,
      "loss": 1.4435,
      "step": 979
    },
    {
      "epoch": 0.25088,
      "grad_norm": 0.44172772018344914,
      "learning_rate": 3.444337631886955e-05,
      "loss": 1.4692,
      "step": 980
    },
    {
      "epoch": 0.251136,
      "grad_norm": 0.44030475318337275,
      "learning_rate": 3.443212947894312e-05,
      "loss": 1.4482,
      "step": 981
    },
    {
      "epoch": 0.251392,
      "grad_norm": 0.415691608094687,
      "learning_rate": 3.442087310871963e-05,
      "loss": 1.3929,
      "step": 982
    },
    {
      "epoch": 0.251648,
      "grad_norm": 0.44304165148469893,
      "learning_rate": 3.440960721563222e-05,
      "loss": 1.4521,
      "step": 983
    },
    {
      "epoch": 0.251904,
      "grad_norm": 0.4172736072418633,
      "learning_rate": 3.439833180712038e-05,
      "loss": 1.4027,
      "step": 984
    },
    {
      "epoch": 0.25216,
      "grad_norm": 0.45109771921711805,
      "learning_rate": 3.438704689062985e-05,
      "loss": 1.468,
      "step": 985
    },
    {
      "epoch": 0.252416,
      "grad_norm": 0.41347473200382073,
      "learning_rate": 3.437575247361266e-05,
      "loss": 1.4347,
      "step": 986
    },
    {
      "epoch": 0.252672,
      "grad_norm": 0.43592305242084495,
      "learning_rate": 3.43644485635271e-05,
      "loss": 1.3982,
      "step": 987
    },
    {
      "epoch": 0.252928,
      "grad_norm": 0.4278707954991748,
      "learning_rate": 3.4353135167837746e-05,
      "loss": 1.4202,
      "step": 988
    },
    {
      "epoch": 0.253184,
      "grad_norm": 0.43683805167400386,
      "learning_rate": 3.434181229401543e-05,
      "loss": 1.3886,
      "step": 989
    },
    {
      "epoch": 0.25344,
      "grad_norm": 0.4230244794105691,
      "learning_rate": 3.4330479949537244e-05,
      "loss": 1.4486,
      "step": 990
    },
    {
      "epoch": 0.253696,
      "grad_norm": 0.46454049363597516,
      "learning_rate": 3.431913814188653e-05,
      "loss": 1.4418,
      "step": 991
    },
    {
      "epoch": 0.253952,
      "grad_norm": 0.43683065193530624,
      "learning_rate": 3.430778687855288e-05,
      "loss": 1.4614,
      "step": 992
    },
    {
      "epoch": 0.254208,
      "grad_norm": 0.4280593917776047,
      "learning_rate": 3.429642616703214e-05,
      "loss": 1.426,
      "step": 993
    },
    {
      "epoch": 0.254464,
      "grad_norm": 0.44574625900574566,
      "learning_rate": 3.4285056014826384e-05,
      "loss": 1.444,
      "step": 994
    },
    {
      "epoch": 0.25472,
      "grad_norm": 0.47125727307256127,
      "learning_rate": 3.427367642944393e-05,
      "loss": 1.4313,
      "step": 995
    },
    {
      "epoch": 0.254976,
      "grad_norm": 0.4466182459406985,
      "learning_rate": 3.42622874183993e-05,
      "loss": 1.4231,
      "step": 996
    },
    {
      "epoch": 0.255232,
      "grad_norm": 0.43873867769444513,
      "learning_rate": 3.4250888989213284e-05,
      "loss": 1.4379,
      "step": 997
    },
    {
      "epoch": 0.255488,
      "grad_norm": 0.44269828190317606,
      "learning_rate": 3.4239481149412856e-05,
      "loss": 1.3736,
      "step": 998
    },
    {
      "epoch": 0.255744,
      "grad_norm": 0.4350156597314667,
      "learning_rate": 3.42280639065312e-05,
      "loss": 1.4446,
      "step": 999
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.42429994440064317,
      "learning_rate": 3.4216637268107756e-05,
      "loss": 1.3959,
      "step": 1000
    },
    {
      "epoch": 0.256256,
      "grad_norm": 0.5709669353474612,
      "learning_rate": 3.420520124168811e-05,
      "loss": 1.433,
      "step": 1001
    },
    {
      "epoch": 0.256512,
      "grad_norm": 0.42665045231310056,
      "learning_rate": 3.419375583482409e-05,
      "loss": 1.4015,
      "step": 1002
    },
    {
      "epoch": 0.256768,
      "grad_norm": 0.4294521621964963,
      "learning_rate": 3.4182301055073685e-05,
      "loss": 1.4472,
      "step": 1003
    },
    {
      "epoch": 0.257024,
      "grad_norm": 0.45581245450102015,
      "learning_rate": 3.4170836910001106e-05,
      "loss": 1.4398,
      "step": 1004
    },
    {
      "epoch": 0.25728,
      "grad_norm": 0.4271422787799621,
      "learning_rate": 3.415936340717673e-05,
      "loss": 1.4221,
      "step": 1005
    },
    {
      "epoch": 0.257536,
      "grad_norm": 0.4369725844285113,
      "learning_rate": 3.414788055417711e-05,
      "loss": 1.3866,
      "step": 1006
    },
    {
      "epoch": 0.257792,
      "grad_norm": 0.4426497566352074,
      "learning_rate": 3.413638835858499e-05,
      "loss": 1.4413,
      "step": 1007
    },
    {
      "epoch": 0.258048,
      "grad_norm": 0.4259937511799618,
      "learning_rate": 3.4124886827989266e-05,
      "loss": 1.4084,
      "step": 1008
    },
    {
      "epoch": 0.258304,
      "grad_norm": 0.4322674807350434,
      "learning_rate": 3.4113375969985014e-05,
      "loss": 1.4824,
      "step": 1009
    },
    {
      "epoch": 0.25856,
      "grad_norm": 0.42823381043090136,
      "learning_rate": 3.410185579217346e-05,
      "loss": 1.446,
      "step": 1010
    },
    {
      "epoch": 0.258816,
      "grad_norm": 0.43528220386194627,
      "learning_rate": 3.409032630216197e-05,
      "loss": 1.427,
      "step": 1011
    },
    {
      "epoch": 0.259072,
      "grad_norm": 0.452054793422249,
      "learning_rate": 3.40787875075641e-05,
      "loss": 1.4119,
      "step": 1012
    },
    {
      "epoch": 0.259328,
      "grad_norm": 0.4250279989064599,
      "learning_rate": 3.4067239415999495e-05,
      "loss": 1.3977,
      "step": 1013
    },
    {
      "epoch": 0.259584,
      "grad_norm": 0.44126424842155326,
      "learning_rate": 3.4055682035093995e-05,
      "loss": 1.4529,
      "step": 1014
    },
    {
      "epoch": 0.25984,
      "grad_norm": 0.4256684350588182,
      "learning_rate": 3.404411537247954e-05,
      "loss": 1.4067,
      "step": 1015
    },
    {
      "epoch": 0.260096,
      "grad_norm": 0.4335721487876929,
      "learning_rate": 3.403253943579421e-05,
      "loss": 1.4464,
      "step": 1016
    },
    {
      "epoch": 0.260352,
      "grad_norm": 0.4303143452203573,
      "learning_rate": 3.4020954232682206e-05,
      "loss": 1.4215,
      "step": 1017
    },
    {
      "epoch": 0.260608,
      "grad_norm": 1.0123183357848653,
      "learning_rate": 3.400935977079385e-05,
      "loss": 1.4353,
      "step": 1018
    },
    {
      "epoch": 0.260864,
      "grad_norm": 0.6482977388833904,
      "learning_rate": 3.399775605778557e-05,
      "loss": 1.438,
      "step": 1019
    },
    {
      "epoch": 0.26112,
      "grad_norm": 0.44967285191479667,
      "learning_rate": 3.398614310131993e-05,
      "loss": 1.4229,
      "step": 1020
    },
    {
      "epoch": 0.261376,
      "grad_norm": 0.6057682988476251,
      "learning_rate": 3.397452090906555e-05,
      "loss": 1.4651,
      "step": 1021
    },
    {
      "epoch": 0.261632,
      "grad_norm": 0.43317326170891146,
      "learning_rate": 3.39628894886972e-05,
      "loss": 1.4256,
      "step": 1022
    },
    {
      "epoch": 0.261888,
      "grad_norm": 0.45491065500195377,
      "learning_rate": 3.395124884789571e-05,
      "loss": 1.4433,
      "step": 1023
    },
    {
      "epoch": 0.262144,
      "grad_norm": 0.4437203654837756,
      "learning_rate": 3.3939598994348013e-05,
      "loss": 1.4422,
      "step": 1024
    },
    {
      "epoch": 0.2624,
      "grad_norm": 0.4191056580718858,
      "learning_rate": 3.392793993574712e-05,
      "loss": 1.4045,
      "step": 1025
    },
    {
      "epoch": 0.262656,
      "grad_norm": 0.4398937026570217,
      "learning_rate": 3.3916271679792126e-05,
      "loss": 1.4321,
      "step": 1026
    },
    {
      "epoch": 0.262912,
      "grad_norm": 0.42128521009592906,
      "learning_rate": 3.39045942341882e-05,
      "loss": 1.4127,
      "step": 1027
    },
    {
      "epoch": 0.263168,
      "grad_norm": 0.4120713687464483,
      "learning_rate": 3.389290760664657e-05,
      "loss": 1.3566,
      "step": 1028
    },
    {
      "epoch": 0.263424,
      "grad_norm": 0.44944326334583873,
      "learning_rate": 3.388121180488453e-05,
      "loss": 1.4551,
      "step": 1029
    },
    {
      "epoch": 0.26368,
      "grad_norm": 0.4231223193237265,
      "learning_rate": 3.386950683662544e-05,
      "loss": 1.3996,
      "step": 1030
    },
    {
      "epoch": 0.263936,
      "grad_norm": 0.4179059780817227,
      "learning_rate": 3.385779270959872e-05,
      "loss": 1.393,
      "step": 1031
    },
    {
      "epoch": 0.264192,
      "grad_norm": 0.44701583616252405,
      "learning_rate": 3.38460694315398e-05,
      "loss": 1.4247,
      "step": 1032
    },
    {
      "epoch": 0.264448,
      "grad_norm": 0.4467420507604214,
      "learning_rate": 3.38343370101902e-05,
      "loss": 1.4173,
      "step": 1033
    },
    {
      "epoch": 0.264704,
      "grad_norm": 0.4203918013562224,
      "learning_rate": 3.382259545329746e-05,
      "loss": 1.4375,
      "step": 1034
    },
    {
      "epoch": 0.26496,
      "grad_norm": 0.44194798806286295,
      "learning_rate": 3.381084476861513e-05,
      "loss": 1.401,
      "step": 1035
    },
    {
      "epoch": 0.265216,
      "grad_norm": 0.4270039794321105,
      "learning_rate": 3.379908496390283e-05,
      "loss": 1.4238,
      "step": 1036
    },
    {
      "epoch": 0.265472,
      "grad_norm": 0.4519111170939524,
      "learning_rate": 3.3787316046926165e-05,
      "loss": 1.4104,
      "step": 1037
    },
    {
      "epoch": 0.265728,
      "grad_norm": 0.43028522250944884,
      "learning_rate": 3.377553802545677e-05,
      "loss": 1.372,
      "step": 1038
    },
    {
      "epoch": 0.265984,
      "grad_norm": 0.44533641862592893,
      "learning_rate": 3.376375090727231e-05,
      "loss": 1.4304,
      "step": 1039
    },
    {
      "epoch": 0.26624,
      "grad_norm": 0.4299567222911219,
      "learning_rate": 3.375195470015643e-05,
      "loss": 1.405,
      "step": 1040
    },
    {
      "epoch": 0.266496,
      "grad_norm": 0.4279312397017779,
      "learning_rate": 3.3740149411898786e-05,
      "loss": 1.3848,
      "step": 1041
    },
    {
      "epoch": 0.266752,
      "grad_norm": 0.454111960456516,
      "learning_rate": 3.3728335050295044e-05,
      "loss": 1.4131,
      "step": 1042
    },
    {
      "epoch": 0.267008,
      "grad_norm": 0.42596979865495277,
      "learning_rate": 3.3716511623146843e-05,
      "loss": 1.3527,
      "step": 1043
    },
    {
      "epoch": 0.267264,
      "grad_norm": 0.4728077247376842,
      "learning_rate": 3.370467913826181e-05,
      "loss": 1.4494,
      "step": 1044
    },
    {
      "epoch": 0.26752,
      "grad_norm": 0.41880870179750784,
      "learning_rate": 3.369283760345358e-05,
      "loss": 1.3987,
      "step": 1045
    },
    {
      "epoch": 0.267776,
      "grad_norm": 0.46394536859563795,
      "learning_rate": 3.3680987026541726e-05,
      "loss": 1.41,
      "step": 1046
    },
    {
      "epoch": 0.268032,
      "grad_norm": 0.43104738083304367,
      "learning_rate": 3.3669127415351814e-05,
      "loss": 1.409,
      "step": 1047
    },
    {
      "epoch": 0.268288,
      "grad_norm": 0.5752830793020016,
      "learning_rate": 3.365725877771537e-05,
      "loss": 1.4452,
      "step": 1048
    },
    {
      "epoch": 0.268544,
      "grad_norm": 0.4307695707750591,
      "learning_rate": 3.364538112146989e-05,
      "loss": 1.414,
      "step": 1049
    },
    {
      "epoch": 0.2688,
      "grad_norm": 0.4408222347775491,
      "learning_rate": 3.3633494454458797e-05,
      "loss": 1.3881,
      "step": 1050
    },
    {
      "epoch": 0.269056,
      "grad_norm": 0.4472505855386507,
      "learning_rate": 3.3621598784531516e-05,
      "loss": 1.4303,
      "step": 1051
    },
    {
      "epoch": 0.269312,
      "grad_norm": 0.44575366365267904,
      "learning_rate": 3.360969411954336e-05,
      "loss": 1.458,
      "step": 1052
    },
    {
      "epoch": 0.269568,
      "grad_norm": 0.4488634034635231,
      "learning_rate": 3.3597780467355617e-05,
      "loss": 1.4352,
      "step": 1053
    },
    {
      "epoch": 0.269824,
      "grad_norm": 0.4518954947339265,
      "learning_rate": 3.3585857835835506e-05,
      "loss": 1.432,
      "step": 1054
    },
    {
      "epoch": 0.27008,
      "grad_norm": 0.4628353190928325,
      "learning_rate": 3.357392623285616e-05,
      "loss": 1.4005,
      "step": 1055
    },
    {
      "epoch": 0.270336,
      "grad_norm": 0.4222741642550358,
      "learning_rate": 3.356198566629666e-05,
      "loss": 1.4066,
      "step": 1056
    },
    {
      "epoch": 0.270592,
      "grad_norm": 0.45883862879413384,
      "learning_rate": 3.355003614404198e-05,
      "loss": 1.4518,
      "step": 1057
    },
    {
      "epoch": 0.270848,
      "grad_norm": 0.4324613345191155,
      "learning_rate": 3.353807767398302e-05,
      "loss": 1.3763,
      "step": 1058
    },
    {
      "epoch": 0.271104,
      "grad_norm": 0.4625868763703739,
      "learning_rate": 3.35261102640166e-05,
      "loss": 1.441,
      "step": 1059
    },
    {
      "epoch": 0.27136,
      "grad_norm": 0.438932783503285,
      "learning_rate": 3.351413392204542e-05,
      "loss": 1.3859,
      "step": 1060
    },
    {
      "epoch": 0.271616,
      "grad_norm": 0.4411975836219631,
      "learning_rate": 3.350214865597809e-05,
      "loss": 1.4521,
      "step": 1061
    },
    {
      "epoch": 0.271872,
      "grad_norm": 0.4505830947630525,
      "learning_rate": 3.349015447372913e-05,
      "loss": 1.4685,
      "step": 1062
    },
    {
      "epoch": 0.272128,
      "grad_norm": 0.42621480759800845,
      "learning_rate": 3.347815138321893e-05,
      "loss": 1.4029,
      "step": 1063
    },
    {
      "epoch": 0.272384,
      "grad_norm": 0.4273730439682329,
      "learning_rate": 3.3466139392373737e-05,
      "loss": 1.3969,
      "step": 1064
    },
    {
      "epoch": 0.27264,
      "grad_norm": 0.43379605570568897,
      "learning_rate": 3.345411850912574e-05,
      "loss": 1.4322,
      "step": 1065
    },
    {
      "epoch": 0.272896,
      "grad_norm": 0.45587258642895095,
      "learning_rate": 3.3442088741412936e-05,
      "loss": 1.4512,
      "step": 1066
    },
    {
      "epoch": 0.273152,
      "grad_norm": 0.43330064555663333,
      "learning_rate": 3.3430050097179236e-05,
      "loss": 1.373,
      "step": 1067
    },
    {
      "epoch": 0.273408,
      "grad_norm": 0.4539309118279507,
      "learning_rate": 3.341800258437437e-05,
      "loss": 1.3822,
      "step": 1068
    },
    {
      "epoch": 0.273664,
      "grad_norm": 0.42680792034258064,
      "learning_rate": 3.340594621095397e-05,
      "loss": 1.3836,
      "step": 1069
    },
    {
      "epoch": 0.27392,
      "grad_norm": 0.43629943792231574,
      "learning_rate": 3.3393880984879486e-05,
      "loss": 1.3919,
      "step": 1070
    },
    {
      "epoch": 0.274176,
      "grad_norm": 0.4318062252229854,
      "learning_rate": 3.338180691411821e-05,
      "loss": 1.4112,
      "step": 1071
    },
    {
      "epoch": 0.274432,
      "grad_norm": 0.48812451105607574,
      "learning_rate": 3.3369724006643306e-05,
      "loss": 1.3769,
      "step": 1072
    },
    {
      "epoch": 0.274688,
      "grad_norm": 0.4430550721916868,
      "learning_rate": 3.335763227043375e-05,
      "loss": 1.428,
      "step": 1073
    },
    {
      "epoch": 0.274944,
      "grad_norm": 0.4927411254144498,
      "learning_rate": 3.334553171347436e-05,
      "loss": 1.4132,
      "step": 1074
    },
    {
      "epoch": 0.2752,
      "grad_norm": 0.5198335599501936,
      "learning_rate": 3.3333422343755754e-05,
      "loss": 1.4295,
      "step": 1075
    },
    {
      "epoch": 0.275456,
      "grad_norm": 0.4181677523654909,
      "learning_rate": 3.332130416927442e-05,
      "loss": 1.4117,
      "step": 1076
    },
    {
      "epoch": 0.275712,
      "grad_norm": 0.4278501225422261,
      "learning_rate": 3.3309177198032596e-05,
      "loss": 1.399,
      "step": 1077
    },
    {
      "epoch": 0.275968,
      "grad_norm": 0.4381751446210983,
      "learning_rate": 3.3297041438038374e-05,
      "loss": 1.3807,
      "step": 1078
    },
    {
      "epoch": 0.276224,
      "grad_norm": 0.4259930691909271,
      "learning_rate": 3.3284896897305646e-05,
      "loss": 1.3801,
      "step": 1079
    },
    {
      "epoch": 0.27648,
      "grad_norm": 0.43998456590072904,
      "learning_rate": 3.327274358385408e-05,
      "loss": 1.4196,
      "step": 1080
    },
    {
      "epoch": 0.276736,
      "grad_norm": 0.43467970031842434,
      "learning_rate": 3.3260581505709155e-05,
      "loss": 1.4248,
      "step": 1081
    },
    {
      "epoch": 0.276992,
      "grad_norm": 0.42766015316242284,
      "learning_rate": 3.3248410670902135e-05,
      "loss": 1.4662,
      "step": 1082
    },
    {
      "epoch": 0.277248,
      "grad_norm": 0.42400473999462046,
      "learning_rate": 3.323623108747007e-05,
      "loss": 1.3948,
      "step": 1083
    },
    {
      "epoch": 0.277504,
      "grad_norm": 0.4089140295603216,
      "learning_rate": 3.322404276345577e-05,
      "loss": 1.382,
      "step": 1084
    },
    {
      "epoch": 0.27776,
      "grad_norm": 0.48292140338198125,
      "learning_rate": 3.321184570690784e-05,
      "loss": 1.4351,
      "step": 1085
    },
    {
      "epoch": 0.278016,
      "grad_norm": 0.4441478255860841,
      "learning_rate": 3.3199639925880635e-05,
      "loss": 1.4128,
      "step": 1086
    },
    {
      "epoch": 0.278272,
      "grad_norm": 0.43220173522750355,
      "learning_rate": 3.3187425428434274e-05,
      "loss": 1.4176,
      "step": 1087
    },
    {
      "epoch": 0.278528,
      "grad_norm": 0.4710304147933427,
      "learning_rate": 3.317520222263464e-05,
      "loss": 1.3941,
      "step": 1088
    },
    {
      "epoch": 0.278784,
      "grad_norm": 0.5787973525389272,
      "learning_rate": 3.316297031655336e-05,
      "loss": 1.4354,
      "step": 1089
    },
    {
      "epoch": 0.27904,
      "grad_norm": 0.45326684848791515,
      "learning_rate": 3.31507297182678e-05,
      "loss": 1.4445,
      "step": 1090
    },
    {
      "epoch": 0.279296,
      "grad_norm": 0.42260688416974257,
      "learning_rate": 3.313848043586109e-05,
      "loss": 1.4049,
      "step": 1091
    },
    {
      "epoch": 0.279552,
      "grad_norm": 0.43888705009446605,
      "learning_rate": 3.312622247742206e-05,
      "loss": 1.4002,
      "step": 1092
    },
    {
      "epoch": 0.279808,
      "grad_norm": 0.46982478052778404,
      "learning_rate": 3.3113955851045294e-05,
      "loss": 1.4006,
      "step": 1093
    },
    {
      "epoch": 0.280064,
      "grad_norm": 0.46131938755057716,
      "learning_rate": 3.31016805648311e-05,
      "loss": 1.3973,
      "step": 1094
    },
    {
      "epoch": 0.28032,
      "grad_norm": 0.42695791819364337,
      "learning_rate": 3.3089396626885484e-05,
      "loss": 1.4096,
      "step": 1095
    },
    {
      "epoch": 0.280576,
      "grad_norm": 0.48540496424955604,
      "learning_rate": 3.307710404532019e-05,
      "loss": 1.4218,
      "step": 1096
    },
    {
      "epoch": 0.280832,
      "grad_norm": 0.4181815020740165,
      "learning_rate": 3.3064802828252656e-05,
      "loss": 1.4089,
      "step": 1097
    },
    {
      "epoch": 0.281088,
      "grad_norm": 0.4624292426744591,
      "learning_rate": 3.3052492983806014e-05,
      "loss": 1.4146,
      "step": 1098
    },
    {
      "epoch": 0.281344,
      "grad_norm": 0.41945629540394497,
      "learning_rate": 3.304017452010913e-05,
      "loss": 1.3996,
      "step": 1099
    },
    {
      "epoch": 0.2816,
      "grad_norm": 0.487680540768905,
      "learning_rate": 3.302784744529651e-05,
      "loss": 1.4213,
      "step": 1100
    },
    {
      "epoch": 0.281856,
      "grad_norm": 0.4326681860030608,
      "learning_rate": 3.3015511767508384e-05,
      "loss": 1.4283,
      "step": 1101
    },
    {
      "epoch": 0.282112,
      "grad_norm": 0.4728747105607696,
      "learning_rate": 3.300316749489065e-05,
      "loss": 1.4193,
      "step": 1102
    },
    {
      "epoch": 0.282368,
      "grad_norm": 0.4395615466395717,
      "learning_rate": 3.2990814635594876e-05,
      "loss": 1.3998,
      "step": 1103
    },
    {
      "epoch": 0.282624,
      "grad_norm": 0.4577230782734413,
      "learning_rate": 3.2978453197778325e-05,
      "loss": 1.4163,
      "step": 1104
    },
    {
      "epoch": 0.28288,
      "grad_norm": 0.43701381588125393,
      "learning_rate": 3.296608318960388e-05,
      "loss": 1.4272,
      "step": 1105
    },
    {
      "epoch": 0.283136,
      "grad_norm": 0.42056232786760994,
      "learning_rate": 3.295370461924014e-05,
      "loss": 1.3768,
      "step": 1106
    },
    {
      "epoch": 0.283392,
      "grad_norm": 0.43698364999106304,
      "learning_rate": 3.2941317494861316e-05,
      "loss": 1.4219,
      "step": 1107
    },
    {
      "epoch": 0.283648,
      "grad_norm": 0.42205577557787666,
      "learning_rate": 3.2928921824647276e-05,
      "loss": 1.4239,
      "step": 1108
    },
    {
      "epoch": 0.283904,
      "grad_norm": 0.4541527606081171,
      "learning_rate": 3.2916517616783536e-05,
      "loss": 1.4737,
      "step": 1109
    },
    {
      "epoch": 0.28416,
      "grad_norm": 0.433999959387067,
      "learning_rate": 3.290410487946125e-05,
      "loss": 1.4476,
      "step": 1110
    },
    {
      "epoch": 0.284416,
      "grad_norm": 0.4310777615335861,
      "learning_rate": 3.289168362087721e-05,
      "loss": 1.4207,
      "step": 1111
    },
    {
      "epoch": 0.284672,
      "grad_norm": 0.42839054977842494,
      "learning_rate": 3.287925384923383e-05,
      "loss": 1.4163,
      "step": 1112
    },
    {
      "epoch": 0.284928,
      "grad_norm": 0.4420005420447975,
      "learning_rate": 3.286681557273913e-05,
      "loss": 1.4234,
      "step": 1113
    },
    {
      "epoch": 0.285184,
      "grad_norm": 0.42448549675239444,
      "learning_rate": 3.285436879960677e-05,
      "loss": 1.4396,
      "step": 1114
    },
    {
      "epoch": 0.28544,
      "grad_norm": 0.42545444935575516,
      "learning_rate": 3.284191353805602e-05,
      "loss": 1.3966,
      "step": 1115
    },
    {
      "epoch": 0.285696,
      "grad_norm": 0.43663761886815217,
      "learning_rate": 3.282944979631173e-05,
      "loss": 1.4307,
      "step": 1116
    },
    {
      "epoch": 0.285952,
      "grad_norm": 0.4192930986251806,
      "learning_rate": 3.281697758260437e-05,
      "loss": 1.4183,
      "step": 1117
    },
    {
      "epoch": 0.286208,
      "grad_norm": 0.4228677427831289,
      "learning_rate": 3.280449690517002e-05,
      "loss": 1.4396,
      "step": 1118
    },
    {
      "epoch": 0.286464,
      "grad_norm": 0.42687309118446376,
      "learning_rate": 3.2792007772250305e-05,
      "loss": 1.3996,
      "step": 1119
    },
    {
      "epoch": 0.28672,
      "grad_norm": 0.4213739266791507,
      "learning_rate": 3.277951019209248e-05,
      "loss": 1.3843,
      "step": 1120
    },
    {
      "epoch": 0.286976,
      "grad_norm": 0.4124062472240904,
      "learning_rate": 3.2767004172949346e-05,
      "loss": 1.3629,
      "step": 1121
    },
    {
      "epoch": 0.287232,
      "grad_norm": 0.4081569520840997,
      "learning_rate": 3.275448972307928e-05,
      "loss": 1.3957,
      "step": 1122
    },
    {
      "epoch": 0.287488,
      "grad_norm": 0.4134585607339923,
      "learning_rate": 3.2741966850746244e-05,
      "loss": 1.4243,
      "step": 1123
    },
    {
      "epoch": 0.287744,
      "grad_norm": 0.4133468632321385,
      "learning_rate": 3.272943556421976e-05,
      "loss": 1.3759,
      "step": 1124
    },
    {
      "epoch": 0.288,
      "grad_norm": 0.4354601600875955,
      "learning_rate": 3.2716895871774886e-05,
      "loss": 1.4441,
      "step": 1125
    },
    {
      "epoch": 0.288256,
      "grad_norm": 0.509393927859276,
      "learning_rate": 3.270434778169224e-05,
      "loss": 1.4118,
      "step": 1126
    },
    {
      "epoch": 0.288512,
      "grad_norm": 0.42639485732992566,
      "learning_rate": 3.2691791302258e-05,
      "loss": 1.4334,
      "step": 1127
    },
    {
      "epoch": 0.288768,
      "grad_norm": 0.44073563651309283,
      "learning_rate": 3.267922644176386e-05,
      "loss": 1.4655,
      "step": 1128
    },
    {
      "epoch": 0.289024,
      "grad_norm": 0.4154603477143575,
      "learning_rate": 3.2666653208507075e-05,
      "loss": 1.3896,
      "step": 1129
    },
    {
      "epoch": 0.28928,
      "grad_norm": 0.4302593491785022,
      "learning_rate": 3.265407161079041e-05,
      "loss": 1.403,
      "step": 1130
    },
    {
      "epoch": 0.289536,
      "grad_norm": 0.4317919009448384,
      "learning_rate": 3.264148165692214e-05,
      "loss": 1.3681,
      "step": 1131
    },
    {
      "epoch": 0.289792,
      "grad_norm": 0.4408316085454837,
      "learning_rate": 3.26288833552161e-05,
      "loss": 1.3998,
      "step": 1132
    },
    {
      "epoch": 0.290048,
      "grad_norm": 0.4480675731399566,
      "learning_rate": 3.2616276713991606e-05,
      "loss": 1.4056,
      "step": 1133
    },
    {
      "epoch": 0.290304,
      "grad_norm": 0.45969072756888313,
      "learning_rate": 3.2603661741573484e-05,
      "loss": 1.4175,
      "step": 1134
    },
    {
      "epoch": 0.29056,
      "grad_norm": 0.5901718115802729,
      "learning_rate": 3.259103844629208e-05,
      "loss": 1.3782,
      "step": 1135
    },
    {
      "epoch": 0.290816,
      "grad_norm": 0.6404182420252112,
      "learning_rate": 3.25784068364832e-05,
      "loss": 1.4055,
      "step": 1136
    },
    {
      "epoch": 0.291072,
      "grad_norm": 0.4544265445156285,
      "learning_rate": 3.2565766920488186e-05,
      "loss": 1.4107,
      "step": 1137
    },
    {
      "epoch": 0.291328,
      "grad_norm": 0.4281151328847092,
      "learning_rate": 3.255311870665383e-05,
      "loss": 1.3876,
      "step": 1138
    },
    {
      "epoch": 0.291584,
      "grad_norm": 0.4760937046716286,
      "learning_rate": 3.2540462203332414e-05,
      "loss": 1.426,
      "step": 1139
    },
    {
      "epoch": 0.29184,
      "grad_norm": 0.45430832309023095,
      "learning_rate": 3.25277974188817e-05,
      "loss": 1.4436,
      "step": 1140
    },
    {
      "epoch": 0.292096,
      "grad_norm": 0.44345885986922606,
      "learning_rate": 3.251512436166492e-05,
      "loss": 1.3907,
      "step": 1141
    },
    {
      "epoch": 0.292352,
      "grad_norm": 0.42117929236071266,
      "learning_rate": 3.250244304005074e-05,
      "loss": 1.3982,
      "step": 1142
    },
    {
      "epoch": 0.292608,
      "grad_norm": 0.44869795415933883,
      "learning_rate": 3.248975346241333e-05,
      "loss": 1.4127,
      "step": 1143
    },
    {
      "epoch": 0.292864,
      "grad_norm": 0.42775831456898844,
      "learning_rate": 3.247705563713228e-05,
      "loss": 1.4351,
      "step": 1144
    },
    {
      "epoch": 0.29312,
      "grad_norm": 0.44403054624269184,
      "learning_rate": 3.246434957259263e-05,
      "loss": 1.4263,
      "step": 1145
    },
    {
      "epoch": 0.293376,
      "grad_norm": 0.4131576238556579,
      "learning_rate": 3.2451635277184866e-05,
      "loss": 1.4178,
      "step": 1146
    },
    {
      "epoch": 0.293632,
      "grad_norm": 0.42608550129408124,
      "learning_rate": 3.2438912759304916e-05,
      "loss": 1.398,
      "step": 1147
    },
    {
      "epoch": 0.293888,
      "grad_norm": 0.42057962527261933,
      "learning_rate": 3.242618202735413e-05,
      "loss": 1.4502,
      "step": 1148
    },
    {
      "epoch": 0.294144,
      "grad_norm": 0.4364781298778433,
      "learning_rate": 3.241344308973927e-05,
      "loss": 1.4394,
      "step": 1149
    },
    {
      "epoch": 0.2944,
      "grad_norm": 0.41859349497917914,
      "learning_rate": 3.240069595487254e-05,
      "loss": 1.4237,
      "step": 1150
    },
    {
      "epoch": 0.294656,
      "grad_norm": 0.43178868912811974,
      "learning_rate": 3.2387940631171545e-05,
      "loss": 1.4788,
      "step": 1151
    },
    {
      "epoch": 0.294912,
      "grad_norm": 0.4375459849608849,
      "learning_rate": 3.23751771270593e-05,
      "loss": 1.4067,
      "step": 1152
    },
    {
      "epoch": 0.295168,
      "grad_norm": 0.4250862980316044,
      "learning_rate": 3.236240545096421e-05,
      "loss": 1.4313,
      "step": 1153
    },
    {
      "epoch": 0.295424,
      "grad_norm": 0.4315547237652792,
      "learning_rate": 3.2349625611320105e-05,
      "loss": 1.4184,
      "step": 1154
    },
    {
      "epoch": 0.29568,
      "grad_norm": 0.40328560367450433,
      "learning_rate": 3.233683761656617e-05,
      "loss": 1.3819,
      "step": 1155
    },
    {
      "epoch": 0.295936,
      "grad_norm": 0.4345655021400087,
      "learning_rate": 3.2324041475147004e-05,
      "loss": 1.4123,
      "step": 1156
    },
    {
      "epoch": 0.296192,
      "grad_norm": 0.40972345051257697,
      "learning_rate": 3.231123719551257e-05,
      "loss": 1.4267,
      "step": 1157
    },
    {
      "epoch": 0.296448,
      "grad_norm": 0.41479548372189257,
      "learning_rate": 3.2298424786118215e-05,
      "loss": 1.4014,
      "step": 1158
    },
    {
      "epoch": 0.296704,
      "grad_norm": 0.401898097166026,
      "learning_rate": 3.228560425542465e-05,
      "loss": 1.3984,
      "step": 1159
    },
    {
      "epoch": 0.29696,
      "grad_norm": 0.39568322630725306,
      "learning_rate": 3.227277561189793e-05,
      "loss": 1.3999,
      "step": 1160
    },
    {
      "epoch": 0.297216,
      "grad_norm": 0.43401559163454545,
      "learning_rate": 3.225993886400951e-05,
      "loss": 1.4252,
      "step": 1161
    },
    {
      "epoch": 0.297472,
      "grad_norm": 0.40806443584462176,
      "learning_rate": 3.224709402023616e-05,
      "loss": 1.406,
      "step": 1162
    },
    {
      "epoch": 0.297728,
      "grad_norm": 0.4339995130953313,
      "learning_rate": 3.223424108906001e-05,
      "loss": 1.398,
      "step": 1163
    },
    {
      "epoch": 0.297984,
      "grad_norm": 0.43209963228877146,
      "learning_rate": 3.222138007896853e-05,
      "loss": 1.4249,
      "step": 1164
    },
    {
      "epoch": 0.29824,
      "grad_norm": 0.432238127372735,
      "learning_rate": 3.220851099845453e-05,
      "loss": 1.3565,
      "step": 1165
    },
    {
      "epoch": 0.298496,
      "grad_norm": 0.42107503053955303,
      "learning_rate": 3.219563385601613e-05,
      "loss": 1.426,
      "step": 1166
    },
    {
      "epoch": 0.298752,
      "grad_norm": 0.43054109840331417,
      "learning_rate": 3.2182748660156795e-05,
      "loss": 1.4325,
      "step": 1167
    },
    {
      "epoch": 0.299008,
      "grad_norm": 0.4121611963396635,
      "learning_rate": 3.21698554193853e-05,
      "loss": 1.3735,
      "step": 1168
    },
    {
      "epoch": 0.299264,
      "grad_norm": 0.4243039112207908,
      "learning_rate": 3.215695414221573e-05,
      "loss": 1.3839,
      "step": 1169
    },
    {
      "epoch": 0.29952,
      "grad_norm": 0.4194864084514799,
      "learning_rate": 3.214404483716749e-05,
      "loss": 1.4287,
      "step": 1170
    },
    {
      "epoch": 0.299776,
      "grad_norm": 0.442795350924567,
      "learning_rate": 3.2131127512765254e-05,
      "loss": 1.4376,
      "step": 1171
    },
    {
      "epoch": 0.300032,
      "grad_norm": 0.7291214163370128,
      "learning_rate": 3.211820217753903e-05,
      "loss": 1.3666,
      "step": 1172
    },
    {
      "epoch": 0.300288,
      "grad_norm": 0.43035110944910393,
      "learning_rate": 3.210526884002411e-05,
      "loss": 1.4319,
      "step": 1173
    },
    {
      "epoch": 0.300544,
      "grad_norm": 0.43007259932303177,
      "learning_rate": 3.209232750876103e-05,
      "loss": 1.4455,
      "step": 1174
    },
    {
      "epoch": 0.3008,
      "grad_norm": 0.4366584717399654,
      "learning_rate": 3.207937819229565e-05,
      "loss": 1.4037,
      "step": 1175
    },
    {
      "epoch": 0.301056,
      "grad_norm": 0.4242114268120552,
      "learning_rate": 3.206642089917909e-05,
      "loss": 1.4362,
      "step": 1176
    },
    {
      "epoch": 0.301312,
      "grad_norm": 0.4592443154762359,
      "learning_rate": 3.205345563796774e-05,
      "loss": 1.4522,
      "step": 1177
    },
    {
      "epoch": 0.301568,
      "grad_norm": 0.41578437846859323,
      "learning_rate": 3.2040482417223226e-05,
      "loss": 1.4031,
      "step": 1178
    },
    {
      "epoch": 0.301824,
      "grad_norm": 0.4308914903807655,
      "learning_rate": 3.2027501245512467e-05,
      "loss": 1.4179,
      "step": 1179
    },
    {
      "epoch": 0.30208,
      "grad_norm": 0.4226380104191961,
      "learning_rate": 3.201451213140762e-05,
      "loss": 1.3722,
      "step": 1180
    },
    {
      "epoch": 0.302336,
      "grad_norm": 0.44056503127289953,
      "learning_rate": 3.2001515083486064e-05,
      "loss": 1.4113,
      "step": 1181
    },
    {
      "epoch": 0.302592,
      "grad_norm": 0.44529615595672334,
      "learning_rate": 3.1988510110330454e-05,
      "loss": 1.4006,
      "step": 1182
    },
    {
      "epoch": 0.302848,
      "grad_norm": 0.40886495600605616,
      "learning_rate": 3.1975497220528644e-05,
      "loss": 1.3829,
      "step": 1183
    },
    {
      "epoch": 0.303104,
      "grad_norm": 0.45877194127997917,
      "learning_rate": 3.196247642267375e-05,
      "loss": 1.4025,
      "step": 1184
    },
    {
      "epoch": 0.30336,
      "grad_norm": 0.4228189306259059,
      "learning_rate": 3.194944772536408e-05,
      "loss": 1.4354,
      "step": 1185
    },
    {
      "epoch": 0.303616,
      "grad_norm": 0.432010486512588,
      "learning_rate": 3.1936411137203174e-05,
      "loss": 1.4126,
      "step": 1186
    },
    {
      "epoch": 0.303872,
      "grad_norm": 0.41472261705120805,
      "learning_rate": 3.192336666679978e-05,
      "loss": 1.4233,
      "step": 1187
    },
    {
      "epoch": 0.304128,
      "grad_norm": 0.4077540182057449,
      "learning_rate": 3.191031432276785e-05,
      "loss": 1.3485,
      "step": 1188
    },
    {
      "epoch": 0.304384,
      "grad_norm": 0.4191130733988432,
      "learning_rate": 3.189725411372654e-05,
      "loss": 1.3602,
      "step": 1189
    },
    {
      "epoch": 0.30464,
      "grad_norm": 0.4181421316061501,
      "learning_rate": 3.188418604830019e-05,
      "loss": 1.3882,
      "step": 1190
    },
    {
      "epoch": 0.304896,
      "grad_norm": 0.42148518526798495,
      "learning_rate": 3.187111013511833e-05,
      "loss": 1.4192,
      "step": 1191
    },
    {
      "epoch": 0.305152,
      "grad_norm": 0.4246014981854268,
      "learning_rate": 3.185802638281569e-05,
      "loss": 1.4109,
      "step": 1192
    },
    {
      "epoch": 0.305408,
      "grad_norm": 0.41941954042274,
      "learning_rate": 3.184493480003214e-05,
      "loss": 1.426,
      "step": 1193
    },
    {
      "epoch": 0.305664,
      "grad_norm": 0.41303926450149886,
      "learning_rate": 3.183183539541277e-05,
      "loss": 1.4235,
      "step": 1194
    },
    {
      "epoch": 0.30592,
      "grad_norm": 0.41251837929526514,
      "learning_rate": 3.18187281776078e-05,
      "loss": 1.3894,
      "step": 1195
    },
    {
      "epoch": 0.306176,
      "grad_norm": 0.4128128018691263,
      "learning_rate": 3.18056131552726e-05,
      "loss": 1.3868,
      "step": 1196
    },
    {
      "epoch": 0.306432,
      "grad_norm": 0.4171460384979787,
      "learning_rate": 3.1792490337067725e-05,
      "loss": 1.3649,
      "step": 1197
    },
    {
      "epoch": 0.306688,
      "grad_norm": 0.42102456348689277,
      "learning_rate": 3.177935973165887e-05,
      "loss": 1.4214,
      "step": 1198
    },
    {
      "epoch": 0.306944,
      "grad_norm": 0.4340899247810748,
      "learning_rate": 3.176622134771687e-05,
      "loss": 1.3913,
      "step": 1199
    },
    {
      "epoch": 0.3072,
      "grad_norm": 0.4088220280950938,
      "learning_rate": 3.1753075193917674e-05,
      "loss": 1.3499,
      "step": 1200
    },
    {
      "epoch": 0.307456,
      "grad_norm": 0.4108606099279782,
      "learning_rate": 3.17399212789424e-05,
      "loss": 1.4277,
      "step": 1201
    },
    {
      "epoch": 0.307712,
      "grad_norm": 0.4307523834209843,
      "learning_rate": 3.1726759611477256e-05,
      "loss": 1.4361,
      "step": 1202
    },
    {
      "epoch": 0.307968,
      "grad_norm": 0.41750151385651174,
      "learning_rate": 3.17135902002136e-05,
      "loss": 1.4131,
      "step": 1203
    },
    {
      "epoch": 0.308224,
      "grad_norm": 0.4107146101173538,
      "learning_rate": 3.1700413053847885e-05,
      "loss": 1.378,
      "step": 1204
    },
    {
      "epoch": 0.30848,
      "grad_norm": 0.42862077911334096,
      "learning_rate": 3.168722818108168e-05,
      "loss": 1.4468,
      "step": 1205
    },
    {
      "epoch": 0.308736,
      "grad_norm": 0.4131203878404391,
      "learning_rate": 3.167403559062164e-05,
      "loss": 1.3732,
      "step": 1206
    },
    {
      "epoch": 0.308992,
      "grad_norm": 0.41794424909767036,
      "learning_rate": 3.1660835291179545e-05,
      "loss": 1.3917,
      "step": 1207
    },
    {
      "epoch": 0.309248,
      "grad_norm": 0.40816798427888107,
      "learning_rate": 3.164762729147224e-05,
      "loss": 1.4394,
      "step": 1208
    },
    {
      "epoch": 0.309504,
      "grad_norm": 0.43607272276646936,
      "learning_rate": 3.163441160022167e-05,
      "loss": 1.4557,
      "step": 1209
    },
    {
      "epoch": 0.30976,
      "grad_norm": 0.41024301357443327,
      "learning_rate": 3.162118822615485e-05,
      "loss": 1.3978,
      "step": 1210
    },
    {
      "epoch": 0.310016,
      "grad_norm": 0.4105175082791347,
      "learning_rate": 3.1607957178003874e-05,
      "loss": 1.3983,
      "step": 1211
    },
    {
      "epoch": 0.310272,
      "grad_norm": 0.4229197688012235,
      "learning_rate": 3.1594718464505896e-05,
      "loss": 1.4291,
      "step": 1212
    },
    {
      "epoch": 0.310528,
      "grad_norm": 0.4082222047043843,
      "learning_rate": 3.158147209440315e-05,
      "loss": 1.4169,
      "step": 1213
    },
    {
      "epoch": 0.310784,
      "grad_norm": 0.7096120274727995,
      "learning_rate": 3.1568218076442904e-05,
      "loss": 1.4024,
      "step": 1214
    },
    {
      "epoch": 0.31104,
      "grad_norm": 0.4159323049778268,
      "learning_rate": 3.1554956419377476e-05,
      "loss": 1.3848,
      "step": 1215
    },
    {
      "epoch": 0.311296,
      "grad_norm": 0.4193137849966752,
      "learning_rate": 3.154168713196426e-05,
      "loss": 1.3952,
      "step": 1216
    },
    {
      "epoch": 0.311552,
      "grad_norm": 0.41871491169588926,
      "learning_rate": 3.1528410222965664e-05,
      "loss": 1.4513,
      "step": 1217
    },
    {
      "epoch": 0.311808,
      "grad_norm": 0.4099675434317884,
      "learning_rate": 3.151512570114911e-05,
      "loss": 1.4055,
      "step": 1218
    },
    {
      "epoch": 0.312064,
      "grad_norm": 0.41434091244194343,
      "learning_rate": 3.150183357528709e-05,
      "loss": 1.3911,
      "step": 1219
    },
    {
      "epoch": 0.31232,
      "grad_norm": 0.4253088197848956,
      "learning_rate": 3.1488533854157096e-05,
      "loss": 1.4421,
      "step": 1220
    },
    {
      "epoch": 0.312576,
      "grad_norm": 0.40912407238603443,
      "learning_rate": 3.1475226546541624e-05,
      "loss": 1.4059,
      "step": 1221
    },
    {
      "epoch": 0.312832,
      "grad_norm": 0.41612533190360373,
      "learning_rate": 3.14619116612282e-05,
      "loss": 1.3935,
      "step": 1222
    },
    {
      "epoch": 0.313088,
      "grad_norm": 0.4035512041461475,
      "learning_rate": 3.144858920700935e-05,
      "loss": 1.367,
      "step": 1223
    },
    {
      "epoch": 0.313344,
      "grad_norm": 0.4088555183597548,
      "learning_rate": 3.1435259192682575e-05,
      "loss": 1.3806,
      "step": 1224
    },
    {
      "epoch": 0.3136,
      "grad_norm": 0.4038294794185272,
      "learning_rate": 3.1421921627050416e-05,
      "loss": 1.3536,
      "step": 1225
    },
    {
      "epoch": 0.313856,
      "grad_norm": 0.40885215306337575,
      "learning_rate": 3.1408576518920346e-05,
      "loss": 1.3924,
      "step": 1226
    },
    {
      "epoch": 0.314112,
      "grad_norm": 0.40657851070025025,
      "learning_rate": 3.139522387710485e-05,
      "loss": 1.3767,
      "step": 1227
    },
    {
      "epoch": 0.314368,
      "grad_norm": 0.4902339272017824,
      "learning_rate": 3.13818637104214e-05,
      "loss": 1.4227,
      "step": 1228
    },
    {
      "epoch": 0.314624,
      "grad_norm": 0.422462749919461,
      "learning_rate": 3.13684960276924e-05,
      "loss": 1.3923,
      "step": 1229
    },
    {
      "epoch": 0.31488,
      "grad_norm": 0.418395953053438,
      "learning_rate": 3.135512083774524e-05,
      "loss": 1.4874,
      "step": 1230
    },
    {
      "epoch": 0.315136,
      "grad_norm": 0.4251976016544273,
      "learning_rate": 3.134173814941227e-05,
      "loss": 1.434,
      "step": 1231
    },
    {
      "epoch": 0.315392,
      "grad_norm": 0.4212149400175426,
      "learning_rate": 3.132834797153079e-05,
      "loss": 1.4046,
      "step": 1232
    },
    {
      "epoch": 0.315648,
      "grad_norm": 0.43017280246410944,
      "learning_rate": 3.131495031294303e-05,
      "loss": 1.4447,
      "step": 1233
    },
    {
      "epoch": 0.315904,
      "grad_norm": 0.41643787912906144,
      "learning_rate": 3.130154518249617e-05,
      "loss": 1.4225,
      "step": 1234
    },
    {
      "epoch": 0.31616,
      "grad_norm": 0.4177713148189969,
      "learning_rate": 3.1288132589042335e-05,
      "loss": 1.4147,
      "step": 1235
    },
    {
      "epoch": 0.316416,
      "grad_norm": 0.4246517513297747,
      "learning_rate": 3.1274712541438564e-05,
      "loss": 1.4012,
      "step": 1236
    },
    {
      "epoch": 0.316672,
      "grad_norm": 0.41914662504676936,
      "learning_rate": 3.126128504854682e-05,
      "loss": 1.4041,
      "step": 1237
    },
    {
      "epoch": 0.316928,
      "grad_norm": 0.42552289495361123,
      "learning_rate": 3.1247850119233995e-05,
      "loss": 1.4056,
      "step": 1238
    },
    {
      "epoch": 0.317184,
      "grad_norm": 0.4235625622010515,
      "learning_rate": 3.1234407762371875e-05,
      "loss": 1.41,
      "step": 1239
    },
    {
      "epoch": 0.31744,
      "grad_norm": 0.42041864931760375,
      "learning_rate": 3.1220957986837165e-05,
      "loss": 1.4256,
      "step": 1240
    },
    {
      "epoch": 0.317696,
      "grad_norm": 0.46716382347814195,
      "learning_rate": 3.120750080151145e-05,
      "loss": 1.3845,
      "step": 1241
    },
    {
      "epoch": 0.317952,
      "grad_norm": 0.4288107361680559,
      "learning_rate": 3.119403621528122e-05,
      "loss": 1.3805,
      "step": 1242
    },
    {
      "epoch": 0.318208,
      "grad_norm": 0.42585324552708786,
      "learning_rate": 3.118056423703787e-05,
      "loss": 1.4314,
      "step": 1243
    },
    {
      "epoch": 0.318464,
      "grad_norm": 0.43413380000721746,
      "learning_rate": 3.116708487567764e-05,
      "loss": 1.4102,
      "step": 1244
    },
    {
      "epoch": 0.31872,
      "grad_norm": 0.4383729186698269,
      "learning_rate": 3.1153598140101676e-05,
      "loss": 1.3789,
      "step": 1245
    },
    {
      "epoch": 0.318976,
      "grad_norm": 0.4439772022027527,
      "learning_rate": 3.114010403921597e-05,
      "loss": 1.4489,
      "step": 1246
    },
    {
      "epoch": 0.319232,
      "grad_norm": 0.4661506932589023,
      "learning_rate": 3.11266025819314e-05,
      "loss": 1.4098,
      "step": 1247
    },
    {
      "epoch": 0.319488,
      "grad_norm": 0.4468540644831854,
      "learning_rate": 3.1113093777163686e-05,
      "loss": 1.4293,
      "step": 1248
    },
    {
      "epoch": 0.319744,
      "grad_norm": 0.40812565348578894,
      "learning_rate": 3.1099577633833406e-05,
      "loss": 1.364,
      "step": 1249
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4387320152984219,
      "learning_rate": 3.1086054160865985e-05,
      "loss": 1.4582,
      "step": 1250
    },
    {
      "epoch": 0.320256,
      "grad_norm": 0.42933384170518335,
      "learning_rate": 3.107252336719168e-05,
      "loss": 1.4147,
      "step": 1251
    },
    {
      "epoch": 0.320512,
      "grad_norm": 0.4191023901502762,
      "learning_rate": 3.1058985261745596e-05,
      "loss": 1.4306,
      "step": 1252
    },
    {
      "epoch": 0.320768,
      "grad_norm": 0.4354160868992651,
      "learning_rate": 3.104543985346765e-05,
      "loss": 1.4376,
      "step": 1253
    },
    {
      "epoch": 0.321024,
      "grad_norm": 0.45779761194874546,
      "learning_rate": 3.10318871513026e-05,
      "loss": 1.4405,
      "step": 1254
    },
    {
      "epoch": 0.32128,
      "grad_norm": 0.40230870893151843,
      "learning_rate": 3.101832716420001e-05,
      "loss": 1.4103,
      "step": 1255
    },
    {
      "epoch": 0.321536,
      "grad_norm": 0.422176785304716,
      "learning_rate": 3.1004759901114245e-05,
      "loss": 1.4084,
      "step": 1256
    },
    {
      "epoch": 0.321792,
      "grad_norm": 0.416659527570115,
      "learning_rate": 3.09911853710045e-05,
      "loss": 1.3816,
      "step": 1257
    },
    {
      "epoch": 0.322048,
      "grad_norm": 0.4679334693268822,
      "learning_rate": 3.097760358283473e-05,
      "loss": 1.3757,
      "step": 1258
    },
    {
      "epoch": 0.322304,
      "grad_norm": 0.4339406676317571,
      "learning_rate": 3.096401454557374e-05,
      "loss": 1.4663,
      "step": 1259
    },
    {
      "epoch": 0.32256,
      "grad_norm": 0.42229257580469615,
      "learning_rate": 3.095041826819506e-05,
      "loss": 1.4102,
      "step": 1260
    },
    {
      "epoch": 0.322816,
      "grad_norm": 0.4250543831838138,
      "learning_rate": 3.093681475967705e-05,
      "loss": 1.3755,
      "step": 1261
    },
    {
      "epoch": 0.323072,
      "grad_norm": 0.43920831376044683,
      "learning_rate": 3.0923204029002815e-05,
      "loss": 1.4312,
      "step": 1262
    },
    {
      "epoch": 0.323328,
      "grad_norm": 0.41301480771951404,
      "learning_rate": 3.090958608516023e-05,
      "loss": 1.3619,
      "step": 1263
    },
    {
      "epoch": 0.323584,
      "grad_norm": 0.4312385959997988,
      "learning_rate": 3.089596093714196e-05,
      "loss": 1.4412,
      "step": 1264
    },
    {
      "epoch": 0.32384,
      "grad_norm": 0.39918549346762466,
      "learning_rate": 3.088232859394541e-05,
      "loss": 1.3637,
      "step": 1265
    },
    {
      "epoch": 0.324096,
      "grad_norm": 0.41279532150483866,
      "learning_rate": 3.086868906457271e-05,
      "loss": 1.4028,
      "step": 1266
    },
    {
      "epoch": 0.324352,
      "grad_norm": 0.4108708369390747,
      "learning_rate": 3.0855042358030795e-05,
      "loss": 1.3657,
      "step": 1267
    },
    {
      "epoch": 0.324608,
      "grad_norm": 0.3986934058253757,
      "learning_rate": 3.0841388483331284e-05,
      "loss": 1.3671,
      "step": 1268
    },
    {
      "epoch": 0.324864,
      "grad_norm": 0.41476267719154275,
      "learning_rate": 3.082772744949056e-05,
      "loss": 1.3774,
      "step": 1269
    },
    {
      "epoch": 0.32512,
      "grad_norm": 0.3987211561793892,
      "learning_rate": 3.0814059265529726e-05,
      "loss": 1.3822,
      "step": 1270
    },
    {
      "epoch": 0.325376,
      "grad_norm": 0.431107851684075,
      "learning_rate": 3.08003839404746e-05,
      "loss": 1.4003,
      "step": 1271
    },
    {
      "epoch": 0.325632,
      "grad_norm": 0.40383602951971925,
      "learning_rate": 3.0786701483355716e-05,
      "loss": 1.4432,
      "step": 1272
    },
    {
      "epoch": 0.325888,
      "grad_norm": 0.4234532770760919,
      "learning_rate": 3.077301190320834e-05,
      "loss": 1.4062,
      "step": 1273
    },
    {
      "epoch": 0.326144,
      "grad_norm": 0.4253562186601696,
      "learning_rate": 3.075931520907241e-05,
      "loss": 1.4306,
      "step": 1274
    },
    {
      "epoch": 0.3264,
      "grad_norm": 0.43180697461774165,
      "learning_rate": 3.074561140999258e-05,
      "loss": 1.4227,
      "step": 1275
    },
    {
      "epoch": 0.326656,
      "grad_norm": 0.43759673557827045,
      "learning_rate": 3.073190051501819e-05,
      "loss": 1.4058,
      "step": 1276
    },
    {
      "epoch": 0.326912,
      "grad_norm": 0.3924255525754246,
      "learning_rate": 3.071818253320328e-05,
      "loss": 1.4032,
      "step": 1277
    },
    {
      "epoch": 0.327168,
      "grad_norm": 0.4422430576562213,
      "learning_rate": 3.070445747360654e-05,
      "loss": 1.4103,
      "step": 1278
    },
    {
      "epoch": 0.327424,
      "grad_norm": 0.42782128255942475,
      "learning_rate": 3.0690725345291364e-05,
      "loss": 1.459,
      "step": 1279
    },
    {
      "epoch": 0.32768,
      "grad_norm": 0.4180700988964281,
      "learning_rate": 3.06769861573258e-05,
      "loss": 1.4052,
      "step": 1280
    },
    {
      "epoch": 0.327936,
      "grad_norm": 0.4121329345059298,
      "learning_rate": 3.0663239918782555e-05,
      "loss": 1.4315,
      "step": 1281
    },
    {
      "epoch": 0.328192,
      "grad_norm": 0.39184343428686336,
      "learning_rate": 3.0649486638738994e-05,
      "loss": 1.3906,
      "step": 1282
    },
    {
      "epoch": 0.328448,
      "grad_norm": 0.41153734696899574,
      "learning_rate": 3.0635726326277145e-05,
      "loss": 1.4242,
      "step": 1283
    },
    {
      "epoch": 0.328704,
      "grad_norm": 0.42122867421333227,
      "learning_rate": 3.062195899048365e-05,
      "loss": 1.4153,
      "step": 1284
    },
    {
      "epoch": 0.32896,
      "grad_norm": 0.4173491850861933,
      "learning_rate": 3.060818464044984e-05,
      "loss": 1.3687,
      "step": 1285
    },
    {
      "epoch": 0.329216,
      "grad_norm": 0.4077471444427495,
      "learning_rate": 3.0594403285271604e-05,
      "loss": 1.4451,
      "step": 1286
    },
    {
      "epoch": 0.329472,
      "grad_norm": 0.41324433942453037,
      "learning_rate": 3.0580614934049524e-05,
      "loss": 1.4418,
      "step": 1287
    },
    {
      "epoch": 0.329728,
      "grad_norm": 0.399379532041928,
      "learning_rate": 3.0566819595888775e-05,
      "loss": 1.3672,
      "step": 1288
    },
    {
      "epoch": 0.329984,
      "grad_norm": 0.41275025901887585,
      "learning_rate": 3.055301727989914e-05,
      "loss": 1.3802,
      "step": 1289
    },
    {
      "epoch": 0.33024,
      "grad_norm": 0.41859219060050396,
      "learning_rate": 3.053920799519502e-05,
      "loss": 1.4245,
      "step": 1290
    },
    {
      "epoch": 0.330496,
      "grad_norm": 0.4076327428619652,
      "learning_rate": 3.0525391750895426e-05,
      "loss": 1.4043,
      "step": 1291
    },
    {
      "epoch": 0.330752,
      "grad_norm": 0.4001922313684657,
      "learning_rate": 3.0511568556123934e-05,
      "loss": 1.4273,
      "step": 1292
    },
    {
      "epoch": 0.331008,
      "grad_norm": 0.42313948842744575,
      "learning_rate": 3.049773842000874e-05,
      "loss": 1.4695,
      "step": 1293
    },
    {
      "epoch": 0.331264,
      "grad_norm": 0.40838141258835375,
      "learning_rate": 3.0483901351682612e-05,
      "loss": 1.3931,
      "step": 1294
    },
    {
      "epoch": 0.33152,
      "grad_norm": 0.41358465540726363,
      "learning_rate": 3.0470057360282896e-05,
      "loss": 1.4059,
      "step": 1295
    },
    {
      "epoch": 0.331776,
      "grad_norm": 2.507835536111388,
      "learning_rate": 3.0456206454951514e-05,
      "loss": 1.4334,
      "step": 1296
    },
    {
      "epoch": 0.332032,
      "grad_norm": 0.46880057278022064,
      "learning_rate": 3.0442348644834944e-05,
      "loss": 1.4683,
      "step": 1297
    },
    {
      "epoch": 0.332288,
      "grad_norm": 0.4173886689216587,
      "learning_rate": 3.042848393908423e-05,
      "loss": 1.3984,
      "step": 1298
    },
    {
      "epoch": 0.332544,
      "grad_norm": 0.42698107439204497,
      "learning_rate": 3.0414612346854972e-05,
      "loss": 1.4095,
      "step": 1299
    },
    {
      "epoch": 0.3328,
      "grad_norm": 1.1165984828329885,
      "learning_rate": 3.0400733877307316e-05,
      "loss": 1.4433,
      "step": 1300
    },
    {
      "epoch": 0.333056,
      "grad_norm": 0.4452702291973284,
      "learning_rate": 3.0386848539605947e-05,
      "loss": 1.3945,
      "step": 1301
    },
    {
      "epoch": 0.333312,
      "grad_norm": 0.48079409045222515,
      "learning_rate": 3.0372956342920088e-05,
      "loss": 1.4524,
      "step": 1302
    },
    {
      "epoch": 0.333568,
      "grad_norm": 0.4379537970147366,
      "learning_rate": 3.0359057296423487e-05,
      "loss": 1.3809,
      "step": 1303
    },
    {
      "epoch": 0.333824,
      "grad_norm": 0.4202475726051076,
      "learning_rate": 3.034515140929442e-05,
      "loss": 1.3242,
      "step": 1304
    },
    {
      "epoch": 0.33408,
      "grad_norm": 0.44835869342948403,
      "learning_rate": 3.0331238690715676e-05,
      "loss": 1.3987,
      "step": 1305
    },
    {
      "epoch": 0.334336,
      "grad_norm": 0.4319324120842725,
      "learning_rate": 3.0317319149874556e-05,
      "loss": 1.3937,
      "step": 1306
    },
    {
      "epoch": 0.334592,
      "grad_norm": 0.41385006731848517,
      "learning_rate": 3.0303392795962878e-05,
      "loss": 1.4053,
      "step": 1307
    },
    {
      "epoch": 0.334848,
      "grad_norm": 0.4336406684374519,
      "learning_rate": 3.028945963817694e-05,
      "loss": 1.3845,
      "step": 1308
    },
    {
      "epoch": 0.335104,
      "grad_norm": 0.4290693139712829,
      "learning_rate": 3.0275519685717546e-05,
      "loss": 1.4014,
      "step": 1309
    },
    {
      "epoch": 0.33536,
      "grad_norm": 0.42024043372411046,
      "learning_rate": 3.0261572947789975e-05,
      "loss": 1.3781,
      "step": 1310
    },
    {
      "epoch": 0.335616,
      "grad_norm": 0.4339449661013725,
      "learning_rate": 3.0247619433604e-05,
      "loss": 1.4144,
      "step": 1311
    },
    {
      "epoch": 0.335872,
      "grad_norm": 0.4309262942451594,
      "learning_rate": 3.0233659152373868e-05,
      "loss": 1.4027,
      "step": 1312
    },
    {
      "epoch": 0.336128,
      "grad_norm": 0.41669504307845745,
      "learning_rate": 3.0219692113318283e-05,
      "loss": 1.4412,
      "step": 1313
    },
    {
      "epoch": 0.336384,
      "grad_norm": 0.4403791338547325,
      "learning_rate": 3.0205718325660414e-05,
      "loss": 1.4148,
      "step": 1314
    },
    {
      "epoch": 0.33664,
      "grad_norm": 0.42834913537761277,
      "learning_rate": 3.01917377986279e-05,
      "loss": 1.4019,
      "step": 1315
    },
    {
      "epoch": 0.336896,
      "grad_norm": 0.41394411385160595,
      "learning_rate": 3.0177750541452826e-05,
      "loss": 1.3879,
      "step": 1316
    },
    {
      "epoch": 0.337152,
      "grad_norm": 0.42386207164349365,
      "learning_rate": 3.0163756563371697e-05,
      "loss": 1.3705,
      "step": 1317
    },
    {
      "epoch": 0.337408,
      "grad_norm": 0.4228294359643049,
      "learning_rate": 3.0149755873625486e-05,
      "loss": 1.4404,
      "step": 1318
    },
    {
      "epoch": 0.337664,
      "grad_norm": 0.41303349259162797,
      "learning_rate": 3.01357484814596e-05,
      "loss": 1.4013,
      "step": 1319
    },
    {
      "epoch": 0.33792,
      "grad_norm": 0.4312530347788422,
      "learning_rate": 3.0121734396123848e-05,
      "loss": 1.3926,
      "step": 1320
    },
    {
      "epoch": 0.338176,
      "grad_norm": 0.421555599824878,
      "learning_rate": 3.010771362687246e-05,
      "loss": 1.4442,
      "step": 1321
    },
    {
      "epoch": 0.338432,
      "grad_norm": 0.4380217840607509,
      "learning_rate": 3.0093686182964102e-05,
      "loss": 1.4205,
      "step": 1322
    },
    {
      "epoch": 0.338688,
      "grad_norm": 0.4214516089879668,
      "learning_rate": 3.0079652073661846e-05,
      "loss": 1.3799,
      "step": 1323
    },
    {
      "epoch": 0.338944,
      "grad_norm": 0.42724323935372926,
      "learning_rate": 3.0065611308233134e-05,
      "loss": 1.475,
      "step": 1324
    },
    {
      "epoch": 0.3392,
      "grad_norm": 0.40168310253274403,
      "learning_rate": 3.0051563895949836e-05,
      "loss": 1.3881,
      "step": 1325
    },
    {
      "epoch": 0.339456,
      "grad_norm": 0.42110570812271714,
      "learning_rate": 3.0037509846088197e-05,
      "loss": 1.4598,
      "step": 1326
    },
    {
      "epoch": 0.339712,
      "grad_norm": 0.4073135562375942,
      "learning_rate": 3.0023449167928854e-05,
      "loss": 1.3696,
      "step": 1327
    },
    {
      "epoch": 0.339968,
      "grad_norm": 0.41175537780242827,
      "learning_rate": 3.0009381870756808e-05,
      "loss": 1.3955,
      "step": 1328
    },
    {
      "epoch": 0.340224,
      "grad_norm": 0.4025994893286787,
      "learning_rate": 2.9995307963861443e-05,
      "loss": 1.4023,
      "step": 1329
    },
    {
      "epoch": 0.34048,
      "grad_norm": 0.3999826325095339,
      "learning_rate": 2.9981227456536502e-05,
      "loss": 1.3582,
      "step": 1330
    },
    {
      "epoch": 0.340736,
      "grad_norm": 0.39653252481757734,
      "learning_rate": 2.996714035808008e-05,
      "loss": 1.3384,
      "step": 1331
    },
    {
      "epoch": 0.340992,
      "grad_norm": 0.4017121823864137,
      "learning_rate": 2.9953046677794648e-05,
      "loss": 1.3892,
      "step": 1332
    },
    {
      "epoch": 0.341248,
      "grad_norm": 0.3952039600621715,
      "learning_rate": 2.993894642498699e-05,
      "loss": 1.3885,
      "step": 1333
    },
    {
      "epoch": 0.341504,
      "grad_norm": 0.4309079034094291,
      "learning_rate": 2.9924839608968254e-05,
      "loss": 1.3966,
      "step": 1334
    },
    {
      "epoch": 0.34176,
      "grad_norm": 0.3912068992899185,
      "learning_rate": 2.991072623905392e-05,
      "loss": 1.3983,
      "step": 1335
    },
    {
      "epoch": 0.342016,
      "grad_norm": 0.4096085625722772,
      "learning_rate": 2.989660632456379e-05,
      "loss": 1.4132,
      "step": 1336
    },
    {
      "epoch": 0.342272,
      "grad_norm": 0.39100270613168364,
      "learning_rate": 2.988247987482198e-05,
      "loss": 1.3827,
      "step": 1337
    },
    {
      "epoch": 0.342528,
      "grad_norm": 0.40875733850657575,
      "learning_rate": 2.9868346899156938e-05,
      "loss": 1.3821,
      "step": 1338
    },
    {
      "epoch": 0.342784,
      "grad_norm": 0.3983344888294015,
      "learning_rate": 2.9854207406901413e-05,
      "loss": 1.4158,
      "step": 1339
    },
    {
      "epoch": 0.34304,
      "grad_norm": 0.4133525821701678,
      "learning_rate": 2.984006140739246e-05,
      "loss": 1.3897,
      "step": 1340
    },
    {
      "epoch": 0.343296,
      "grad_norm": 0.41878414870417735,
      "learning_rate": 2.982590890997143e-05,
      "loss": 1.4366,
      "step": 1341
    },
    {
      "epoch": 0.343552,
      "grad_norm": 0.41036239739906116,
      "learning_rate": 2.9811749923983958e-05,
      "loss": 1.4035,
      "step": 1342
    },
    {
      "epoch": 0.343808,
      "grad_norm": 0.4099744812173714,
      "learning_rate": 2.9797584458779973e-05,
      "loss": 1.418,
      "step": 1343
    },
    {
      "epoch": 0.344064,
      "grad_norm": 0.39679300109156596,
      "learning_rate": 2.9783412523713688e-05,
      "loss": 1.366,
      "step": 1344
    },
    {
      "epoch": 0.34432,
      "grad_norm": 0.407202682154567,
      "learning_rate": 2.9769234128143565e-05,
      "loss": 1.428,
      "step": 1345
    },
    {
      "epoch": 0.344576,
      "grad_norm": 0.4653510458646779,
      "learning_rate": 2.9755049281432354e-05,
      "loss": 1.3757,
      "step": 1346
    },
    {
      "epoch": 0.344832,
      "grad_norm": 0.4268516829619504,
      "learning_rate": 2.974085799294707e-05,
      "loss": 1.4476,
      "step": 1347
    },
    {
      "epoch": 0.345088,
      "grad_norm": 0.39093214171547186,
      "learning_rate": 2.9726660272058948e-05,
      "loss": 1.3657,
      "step": 1348
    },
    {
      "epoch": 0.345344,
      "grad_norm": 0.4063958212976025,
      "learning_rate": 2.97124561281435e-05,
      "loss": 1.3682,
      "step": 1349
    },
    {
      "epoch": 0.3456,
      "grad_norm": 0.4204029768691132,
      "learning_rate": 2.9698245570580488e-05,
      "loss": 1.3964,
      "step": 1350
    },
    {
      "epoch": 0.345856,
      "grad_norm": 0.40915344851946944,
      "learning_rate": 2.9684028608753874e-05,
      "loss": 1.4048,
      "step": 1351
    },
    {
      "epoch": 0.346112,
      "grad_norm": 0.41143302070026544,
      "learning_rate": 2.9669805252051884e-05,
      "loss": 1.3801,
      "step": 1352
    },
    {
      "epoch": 0.346368,
      "grad_norm": 0.40408748499763036,
      "learning_rate": 2.9655575509866935e-05,
      "loss": 1.3964,
      "step": 1353
    },
    {
      "epoch": 0.346624,
      "grad_norm": 0.4070544059317756,
      "learning_rate": 2.9641339391595685e-05,
      "loss": 1.393,
      "step": 1354
    },
    {
      "epoch": 0.34688,
      "grad_norm": 0.4117257452047398,
      "learning_rate": 2.9627096906639e-05,
      "loss": 1.4146,
      "step": 1355
    },
    {
      "epoch": 0.347136,
      "grad_norm": 0.41523801930988,
      "learning_rate": 2.9612848064401934e-05,
      "loss": 1.3954,
      "step": 1356
    },
    {
      "epoch": 0.347392,
      "grad_norm": 0.40424233120719416,
      "learning_rate": 2.959859287429376e-05,
      "loss": 1.3973,
      "step": 1357
    },
    {
      "epoch": 0.347648,
      "grad_norm": 0.3952105902502218,
      "learning_rate": 2.9584331345727924e-05,
      "loss": 1.3594,
      "step": 1358
    },
    {
      "epoch": 0.347904,
      "grad_norm": 0.4002731102703411,
      "learning_rate": 2.9570063488122077e-05,
      "loss": 1.3644,
      "step": 1359
    },
    {
      "epoch": 0.34816,
      "grad_norm": 0.4062748863017373,
      "learning_rate": 2.9555789310898025e-05,
      "loss": 1.3972,
      "step": 1360
    },
    {
      "epoch": 0.348416,
      "grad_norm": 0.4006636196780595,
      "learning_rate": 2.954150882348177e-05,
      "loss": 1.3761,
      "step": 1361
    },
    {
      "epoch": 0.348672,
      "grad_norm": 0.42746420943661606,
      "learning_rate": 2.952722203530347e-05,
      "loss": 1.4657,
      "step": 1362
    },
    {
      "epoch": 0.348928,
      "grad_norm": 0.40353070498438187,
      "learning_rate": 2.9512928955797443e-05,
      "loss": 1.3721,
      "step": 1363
    },
    {
      "epoch": 0.349184,
      "grad_norm": 0.40839741594261064,
      "learning_rate": 2.9498629594402167e-05,
      "loss": 1.3842,
      "step": 1364
    },
    {
      "epoch": 0.34944,
      "grad_norm": 0.4088648815029996,
      "learning_rate": 2.9484323960560273e-05,
      "loss": 1.407,
      "step": 1365
    },
    {
      "epoch": 0.349696,
      "grad_norm": 0.4194551301113532,
      "learning_rate": 2.9470012063718508e-05,
      "loss": 1.4074,
      "step": 1366
    },
    {
      "epoch": 0.349952,
      "grad_norm": 0.41113900413829035,
      "learning_rate": 2.9455693913327788e-05,
      "loss": 1.3886,
      "step": 1367
    },
    {
      "epoch": 0.350208,
      "grad_norm": 0.4075661853838565,
      "learning_rate": 2.9441369518843138e-05,
      "loss": 1.421,
      "step": 1368
    },
    {
      "epoch": 0.350464,
      "grad_norm": 0.4326718183685464,
      "learning_rate": 2.9427038889723708e-05,
      "loss": 1.4226,
      "step": 1369
    },
    {
      "epoch": 0.35072,
      "grad_norm": 0.3978549833374469,
      "learning_rate": 2.941270203543278e-05,
      "loss": 1.386,
      "step": 1370
    },
    {
      "epoch": 0.350976,
      "grad_norm": 0.41513633886157214,
      "learning_rate": 2.9398358965437726e-05,
      "loss": 1.3876,
      "step": 1371
    },
    {
      "epoch": 0.351232,
      "grad_norm": 0.4072081871458697,
      "learning_rate": 2.9384009689210038e-05,
      "loss": 1.4214,
      "step": 1372
    },
    {
      "epoch": 0.351488,
      "grad_norm": 0.4102887266971176,
      "learning_rate": 2.9369654216225294e-05,
      "loss": 1.3733,
      "step": 1373
    },
    {
      "epoch": 0.351744,
      "grad_norm": 0.4247078164587722,
      "learning_rate": 2.9355292555963185e-05,
      "loss": 1.4051,
      "step": 1374
    },
    {
      "epoch": 0.352,
      "grad_norm": 0.3979339879640671,
      "learning_rate": 2.9340924717907455e-05,
      "loss": 1.3585,
      "step": 1375
    },
    {
      "epoch": 0.352256,
      "grad_norm": 0.4091242783323288,
      "learning_rate": 2.932655071154596e-05,
      "loss": 1.3668,
      "step": 1376
    },
    {
      "epoch": 0.352512,
      "grad_norm": 0.401848634888522,
      "learning_rate": 2.9312170546370605e-05,
      "loss": 1.3845,
      "step": 1377
    },
    {
      "epoch": 0.352768,
      "grad_norm": 0.39291242459777115,
      "learning_rate": 2.9297784231877382e-05,
      "loss": 1.3894,
      "step": 1378
    },
    {
      "epoch": 0.353024,
      "grad_norm": 0.4152876137595836,
      "learning_rate": 2.9283391777566325e-05,
      "loss": 1.3088,
      "step": 1379
    },
    {
      "epoch": 0.35328,
      "grad_norm": 0.39267878980640647,
      "learning_rate": 2.9268993192941547e-05,
      "loss": 1.4052,
      "step": 1380
    },
    {
      "epoch": 0.353536,
      "grad_norm": 0.4243710580844867,
      "learning_rate": 2.9254588487511175e-05,
      "loss": 1.4289,
      "step": 1381
    },
    {
      "epoch": 0.353792,
      "grad_norm": 0.4187107223001199,
      "learning_rate": 2.9240177670787406e-05,
      "loss": 1.4168,
      "step": 1382
    },
    {
      "epoch": 0.354048,
      "grad_norm": 0.3930573690233906,
      "learning_rate": 2.9225760752286452e-05,
      "loss": 1.3699,
      "step": 1383
    },
    {
      "epoch": 0.354304,
      "grad_norm": 0.41248128489925057,
      "learning_rate": 2.9211337741528587e-05,
      "loss": 1.4088,
      "step": 1384
    },
    {
      "epoch": 0.35456,
      "grad_norm": 0.40385969250358,
      "learning_rate": 2.9196908648038067e-05,
      "loss": 1.392,
      "step": 1385
    },
    {
      "epoch": 0.354816,
      "grad_norm": 0.41201070774800636,
      "learning_rate": 2.91824734813432e-05,
      "loss": 1.3823,
      "step": 1386
    },
    {
      "epoch": 0.355072,
      "grad_norm": 0.4111156865828313,
      "learning_rate": 2.9168032250976276e-05,
      "loss": 1.363,
      "step": 1387
    },
    {
      "epoch": 0.355328,
      "grad_norm": 0.4090791112503471,
      "learning_rate": 2.9153584966473603e-05,
      "loss": 1.4184,
      "step": 1388
    },
    {
      "epoch": 0.355584,
      "grad_norm": 0.39436573608160275,
      "learning_rate": 2.913913163737549e-05,
      "loss": 1.3679,
      "step": 1389
    },
    {
      "epoch": 0.35584,
      "grad_norm": 0.42828945672571866,
      "learning_rate": 2.9124672273226235e-05,
      "loss": 1.386,
      "step": 1390
    },
    {
      "epoch": 0.356096,
      "grad_norm": 0.4169046921649057,
      "learning_rate": 2.9110206883574116e-05,
      "loss": 1.4224,
      "step": 1391
    },
    {
      "epoch": 0.356352,
      "grad_norm": 0.4059191438661423,
      "learning_rate": 2.9095735477971394e-05,
      "loss": 1.392,
      "step": 1392
    },
    {
      "epoch": 0.356608,
      "grad_norm": 0.4010680595741339,
      "learning_rate": 2.908125806597431e-05,
      "loss": 1.4451,
      "step": 1393
    },
    {
      "epoch": 0.356864,
      "grad_norm": 0.4197912470550911,
      "learning_rate": 2.9066774657143044e-05,
      "loss": 1.3578,
      "step": 1394
    },
    {
      "epoch": 0.35712,
      "grad_norm": 0.40507250291805036,
      "learning_rate": 2.905228526104177e-05,
      "loss": 1.3947,
      "step": 1395
    },
    {
      "epoch": 0.357376,
      "grad_norm": 0.40100006589374415,
      "learning_rate": 2.9037789887238604e-05,
      "loss": 1.3946,
      "step": 1396
    },
    {
      "epoch": 0.357632,
      "grad_norm": 0.41565844507070737,
      "learning_rate": 2.9023288545305603e-05,
      "loss": 1.3639,
      "step": 1397
    },
    {
      "epoch": 0.357888,
      "grad_norm": 0.40506390419502336,
      "learning_rate": 2.9008781244818757e-05,
      "loss": 1.3902,
      "step": 1398
    },
    {
      "epoch": 0.358144,
      "grad_norm": 0.41078857891945475,
      "learning_rate": 2.8994267995358023e-05,
      "loss": 1.3904,
      "step": 1399
    },
    {
      "epoch": 0.3584,
      "grad_norm": 0.41349718420951437,
      "learning_rate": 2.8979748806507246e-05,
      "loss": 1.3436,
      "step": 1400
    },
    {
      "epoch": 0.358656,
      "grad_norm": 0.40497883809613155,
      "learning_rate": 2.8965223687854228e-05,
      "loss": 1.4263,
      "step": 1401
    },
    {
      "epoch": 0.358912,
      "grad_norm": 0.4186045316020322,
      "learning_rate": 2.895069264899067e-05,
      "loss": 1.4104,
      "step": 1402
    },
    {
      "epoch": 0.359168,
      "grad_norm": 0.4050900563293329,
      "learning_rate": 2.8936155699512177e-05,
      "loss": 1.3894,
      "step": 1403
    },
    {
      "epoch": 0.359424,
      "grad_norm": 0.4217906921509169,
      "learning_rate": 2.8921612849018264e-05,
      "loss": 1.4188,
      "step": 1404
    },
    {
      "epoch": 0.35968,
      "grad_norm": 0.406031775708933,
      "learning_rate": 2.8907064107112352e-05,
      "loss": 1.4038,
      "step": 1405
    },
    {
      "epoch": 0.359936,
      "grad_norm": 0.42501135712568533,
      "learning_rate": 2.8892509483401733e-05,
      "loss": 1.3864,
      "step": 1406
    },
    {
      "epoch": 0.360192,
      "grad_norm": 0.4004574305482252,
      "learning_rate": 2.88779489874976e-05,
      "loss": 1.3794,
      "step": 1407
    },
    {
      "epoch": 0.360448,
      "grad_norm": 0.4322073810561032,
      "learning_rate": 2.886338262901502e-05,
      "loss": 1.4033,
      "step": 1408
    },
    {
      "epoch": 0.360704,
      "grad_norm": 0.407532779928784,
      "learning_rate": 2.8848810417572926e-05,
      "loss": 1.3653,
      "step": 1409
    },
    {
      "epoch": 0.36096,
      "grad_norm": 0.4160188733666925,
      "learning_rate": 2.883423236279411e-05,
      "loss": 1.3987,
      "step": 1410
    },
    {
      "epoch": 0.361216,
      "grad_norm": 0.3946384821257591,
      "learning_rate": 2.8819648474305242e-05,
      "loss": 1.3528,
      "step": 1411
    },
    {
      "epoch": 0.361472,
      "grad_norm": 0.40946563151926035,
      "learning_rate": 2.8805058761736826e-05,
      "loss": 1.4338,
      "step": 1412
    },
    {
      "epoch": 0.361728,
      "grad_norm": 0.4140866205862416,
      "learning_rate": 2.8790463234723228e-05,
      "loss": 1.3277,
      "step": 1413
    },
    {
      "epoch": 0.361984,
      "grad_norm": 0.41329119127896946,
      "learning_rate": 2.8775861902902637e-05,
      "loss": 1.4364,
      "step": 1414
    },
    {
      "epoch": 0.36224,
      "grad_norm": 0.4148301414120206,
      "learning_rate": 2.876125477591709e-05,
      "loss": 1.3876,
      "step": 1415
    },
    {
      "epoch": 0.362496,
      "grad_norm": 0.42922962124142383,
      "learning_rate": 2.8746641863412442e-05,
      "loss": 1.4215,
      "step": 1416
    },
    {
      "epoch": 0.362752,
      "grad_norm": 0.3996671823822803,
      "learning_rate": 2.8732023175038367e-05,
      "loss": 1.4017,
      "step": 1417
    },
    {
      "epoch": 0.363008,
      "grad_norm": 0.4243935379713655,
      "learning_rate": 2.8717398720448354e-05,
      "loss": 1.4023,
      "step": 1418
    },
    {
      "epoch": 0.363264,
      "grad_norm": 0.40366481180258595,
      "learning_rate": 2.8702768509299713e-05,
      "loss": 1.3576,
      "step": 1419
    },
    {
      "epoch": 0.36352,
      "grad_norm": 0.40237319153594614,
      "learning_rate": 2.8688132551253546e-05,
      "loss": 1.3771,
      "step": 1420
    },
    {
      "epoch": 0.363776,
      "grad_norm": 0.3979352056097781,
      "learning_rate": 2.8673490855974734e-05,
      "loss": 1.3749,
      "step": 1421
    },
    {
      "epoch": 0.364032,
      "grad_norm": 0.411800551565504,
      "learning_rate": 2.865884343313197e-05,
      "loss": 1.3891,
      "step": 1422
    },
    {
      "epoch": 0.364288,
      "grad_norm": 0.4357601236293486,
      "learning_rate": 2.8644190292397724e-05,
      "loss": 1.4419,
      "step": 1423
    },
    {
      "epoch": 0.364544,
      "grad_norm": 0.41655976059382555,
      "learning_rate": 2.8629531443448235e-05,
      "loss": 1.3506,
      "step": 1424
    },
    {
      "epoch": 0.3648,
      "grad_norm": 0.41080576554614484,
      "learning_rate": 2.8614866895963516e-05,
      "loss": 1.3921,
      "step": 1425
    },
    {
      "epoch": 0.365056,
      "grad_norm": 0.41253353404570114,
      "learning_rate": 2.8600196659627338e-05,
      "loss": 1.3865,
      "step": 1426
    },
    {
      "epoch": 0.365312,
      "grad_norm": 0.3986542374866559,
      "learning_rate": 2.8585520744127237e-05,
      "loss": 1.3525,
      "step": 1427
    },
    {
      "epoch": 0.365568,
      "grad_norm": 0.4096729370444157,
      "learning_rate": 2.857083915915449e-05,
      "loss": 1.3653,
      "step": 1428
    },
    {
      "epoch": 0.365824,
      "grad_norm": 0.39774045543567765,
      "learning_rate": 2.8556151914404134e-05,
      "loss": 1.4021,
      "step": 1429
    },
    {
      "epoch": 0.36608,
      "grad_norm": 0.4049965435487163,
      "learning_rate": 2.8541459019574917e-05,
      "loss": 1.3832,
      "step": 1430
    },
    {
      "epoch": 0.366336,
      "grad_norm": 0.42629838563661254,
      "learning_rate": 2.8526760484369343e-05,
      "loss": 1.4137,
      "step": 1431
    },
    {
      "epoch": 0.366592,
      "grad_norm": 0.4116529028138325,
      "learning_rate": 2.851205631849363e-05,
      "loss": 1.4292,
      "step": 1432
    },
    {
      "epoch": 0.366848,
      "grad_norm": 0.4045372916833185,
      "learning_rate": 2.8497346531657707e-05,
      "loss": 1.3847,
      "step": 1433
    },
    {
      "epoch": 0.367104,
      "grad_norm": 0.38842958290686924,
      "learning_rate": 2.8482631133575234e-05,
      "loss": 1.3454,
      "step": 1434
    },
    {
      "epoch": 0.36736,
      "grad_norm": 0.41313710411587384,
      "learning_rate": 2.8467910133963563e-05,
      "loss": 1.4246,
      "step": 1435
    },
    {
      "epoch": 0.367616,
      "grad_norm": 0.415673512083582,
      "learning_rate": 2.8453183542543743e-05,
      "loss": 1.394,
      "step": 1436
    },
    {
      "epoch": 0.367872,
      "grad_norm": 0.40776592561021396,
      "learning_rate": 2.8438451369040517e-05,
      "loss": 1.3712,
      "step": 1437
    },
    {
      "epoch": 0.368128,
      "grad_norm": 0.40050035240394743,
      "learning_rate": 2.8423713623182335e-05,
      "loss": 1.4022,
      "step": 1438
    },
    {
      "epoch": 0.368384,
      "grad_norm": 0.413546865878269,
      "learning_rate": 2.840897031470128e-05,
      "loss": 1.3992,
      "step": 1439
    },
    {
      "epoch": 0.36864,
      "grad_norm": 0.3977846604527369,
      "learning_rate": 2.8394221453333156e-05,
      "loss": 1.4011,
      "step": 1440
    },
    {
      "epoch": 0.368896,
      "grad_norm": 0.4465853157576206,
      "learning_rate": 2.8379467048817417e-05,
      "loss": 1.39,
      "step": 1441
    },
    {
      "epoch": 0.369152,
      "grad_norm": 0.4117433738404153,
      "learning_rate": 2.8364707110897164e-05,
      "loss": 1.3279,
      "step": 1442
    },
    {
      "epoch": 0.369408,
      "grad_norm": 0.4099006976335367,
      "learning_rate": 2.8349941649319177e-05,
      "loss": 1.4148,
      "step": 1443
    },
    {
      "epoch": 0.369664,
      "grad_norm": 0.4033069633736659,
      "learning_rate": 2.8335170673833853e-05,
      "loss": 1.4042,
      "step": 1444
    },
    {
      "epoch": 0.36992,
      "grad_norm": 0.42122434363078604,
      "learning_rate": 2.832039419419526e-05,
      "loss": 1.3894,
      "step": 1445
    },
    {
      "epoch": 0.370176,
      "grad_norm": 0.403945412759211,
      "learning_rate": 2.8305612220161083e-05,
      "loss": 1.4065,
      "step": 1446
    },
    {
      "epoch": 0.370432,
      "grad_norm": 0.41425040899401516,
      "learning_rate": 2.8290824761492638e-05,
      "loss": 1.4355,
      "step": 1447
    },
    {
      "epoch": 0.370688,
      "grad_norm": 0.3982234942522407,
      "learning_rate": 2.827603182795487e-05,
      "loss": 1.3782,
      "step": 1448
    },
    {
      "epoch": 0.370944,
      "grad_norm": 0.4202672535883716,
      "learning_rate": 2.8261233429316332e-05,
      "loss": 1.4379,
      "step": 1449
    },
    {
      "epoch": 0.3712,
      "grad_norm": 0.40147854079857526,
      "learning_rate": 2.824642957534918e-05,
      "loss": 1.4362,
      "step": 1450
    },
    {
      "epoch": 0.371456,
      "grad_norm": 0.39897217803227375,
      "learning_rate": 2.8231620275829195e-05,
      "loss": 1.407,
      "step": 1451
    },
    {
      "epoch": 0.371712,
      "grad_norm": 0.40318497864964664,
      "learning_rate": 2.8216805540535726e-05,
      "loss": 1.4066,
      "step": 1452
    },
    {
      "epoch": 0.371968,
      "grad_norm": 0.4103414428318573,
      "learning_rate": 2.820198537925174e-05,
      "loss": 1.3865,
      "step": 1453
    },
    {
      "epoch": 0.372224,
      "grad_norm": 0.4182121543077132,
      "learning_rate": 2.818715980176376e-05,
      "loss": 1.3869,
      "step": 1454
    },
    {
      "epoch": 0.37248,
      "grad_norm": 0.40288453253981876,
      "learning_rate": 2.8172328817861893e-05,
      "loss": 1.3704,
      "step": 1455
    },
    {
      "epoch": 0.372736,
      "grad_norm": 0.4069207416474802,
      "learning_rate": 2.815749243733983e-05,
      "loss": 1.4013,
      "step": 1456
    },
    {
      "epoch": 0.372992,
      "grad_norm": 0.39384935194889376,
      "learning_rate": 2.8142650669994815e-05,
      "loss": 1.3887,
      "step": 1457
    },
    {
      "epoch": 0.373248,
      "grad_norm": 0.41284920347155823,
      "learning_rate": 2.812780352562765e-05,
      "loss": 1.4167,
      "step": 1458
    },
    {
      "epoch": 0.373504,
      "grad_norm": 0.39494067135759336,
      "learning_rate": 2.811295101404269e-05,
      "loss": 1.4529,
      "step": 1459
    },
    {
      "epoch": 0.37376,
      "grad_norm": 0.39146528650555706,
      "learning_rate": 2.8098093145047835e-05,
      "loss": 1.3363,
      "step": 1460
    },
    {
      "epoch": 0.374016,
      "grad_norm": 0.3938517203877017,
      "learning_rate": 2.808322992845451e-05,
      "loss": 1.4158,
      "step": 1461
    },
    {
      "epoch": 0.374272,
      "grad_norm": 0.38717281782784313,
      "learning_rate": 2.8068361374077685e-05,
      "loss": 1.3325,
      "step": 1462
    },
    {
      "epoch": 0.374528,
      "grad_norm": 0.40039877511324534,
      "learning_rate": 2.805348749173586e-05,
      "loss": 1.4214,
      "step": 1463
    },
    {
      "epoch": 0.374784,
      "grad_norm": 0.4040339933397001,
      "learning_rate": 2.8038608291251034e-05,
      "loss": 1.4046,
      "step": 1464
    },
    {
      "epoch": 0.37504,
      "grad_norm": 0.4012907304983749,
      "learning_rate": 2.802372378244874e-05,
      "loss": 1.3418,
      "step": 1465
    },
    {
      "epoch": 0.375296,
      "grad_norm": 0.40684512393503225,
      "learning_rate": 2.8008833975158e-05,
      "loss": 1.4432,
      "step": 1466
    },
    {
      "epoch": 0.375552,
      "grad_norm": 0.4015132924502184,
      "learning_rate": 2.7993938879211326e-05,
      "loss": 1.3388,
      "step": 1467
    },
    {
      "epoch": 0.375808,
      "grad_norm": 0.39201937175882234,
      "learning_rate": 2.7979038504444758e-05,
      "loss": 1.3599,
      "step": 1468
    },
    {
      "epoch": 0.376064,
      "grad_norm": 0.3954785930528305,
      "learning_rate": 2.796413286069779e-05,
      "loss": 1.361,
      "step": 1469
    },
    {
      "epoch": 0.37632,
      "grad_norm": 0.3991007536549761,
      "learning_rate": 2.794922195781341e-05,
      "loss": 1.3837,
      "step": 1470
    },
    {
      "epoch": 0.376576,
      "grad_norm": 0.3962004416655387,
      "learning_rate": 2.793430580563807e-05,
      "loss": 1.3832,
      "step": 1471
    },
    {
      "epoch": 0.376832,
      "grad_norm": 0.4151718797988861,
      "learning_rate": 2.7919384414021694e-05,
      "loss": 1.3845,
      "step": 1472
    },
    {
      "epoch": 0.377088,
      "grad_norm": 0.4070998709006792,
      "learning_rate": 2.7904457792817657e-05,
      "loss": 1.4441,
      "step": 1473
    },
    {
      "epoch": 0.377344,
      "grad_norm": 0.40227154642403934,
      "learning_rate": 2.788952595188281e-05,
      "loss": 1.3828,
      "step": 1474
    },
    {
      "epoch": 0.3776,
      "grad_norm": 0.41053363456182507,
      "learning_rate": 2.7874588901077423e-05,
      "loss": 1.401,
      "step": 1475
    },
    {
      "epoch": 0.377856,
      "grad_norm": 0.4156256717670139,
      "learning_rate": 2.7859646650265234e-05,
      "loss": 1.4609,
      "step": 1476
    },
    {
      "epoch": 0.378112,
      "grad_norm": 0.42156722753830067,
      "learning_rate": 2.7844699209313386e-05,
      "loss": 1.3666,
      "step": 1477
    },
    {
      "epoch": 0.378368,
      "grad_norm": 0.39362018482759137,
      "learning_rate": 2.782974658809247e-05,
      "loss": 1.3902,
      "step": 1478
    },
    {
      "epoch": 0.378624,
      "grad_norm": 0.42342397930782977,
      "learning_rate": 2.7814788796476485e-05,
      "loss": 1.3568,
      "step": 1479
    },
    {
      "epoch": 0.37888,
      "grad_norm": 0.410550145888425,
      "learning_rate": 2.7799825844342855e-05,
      "loss": 1.3763,
      "step": 1480
    },
    {
      "epoch": 0.379136,
      "grad_norm": 0.42195017934877654,
      "learning_rate": 2.7784857741572417e-05,
      "loss": 1.4173,
      "step": 1481
    },
    {
      "epoch": 0.379392,
      "grad_norm": 0.41889711774759975,
      "learning_rate": 2.7769884498049386e-05,
      "loss": 1.4141,
      "step": 1482
    },
    {
      "epoch": 0.379648,
      "grad_norm": 0.40309358374814935,
      "learning_rate": 2.7754906123661387e-05,
      "loss": 1.3913,
      "step": 1483
    },
    {
      "epoch": 0.379904,
      "grad_norm": 0.4332191496375678,
      "learning_rate": 2.7739922628299437e-05,
      "loss": 1.3699,
      "step": 1484
    },
    {
      "epoch": 0.38016,
      "grad_norm": 0.3978072067686622,
      "learning_rate": 2.7724934021857918e-05,
      "loss": 1.4148,
      "step": 1485
    },
    {
      "epoch": 0.380416,
      "grad_norm": 0.4073415193597892,
      "learning_rate": 2.770994031423461e-05,
      "loss": 1.4123,
      "step": 1486
    },
    {
      "epoch": 0.380672,
      "grad_norm": 0.39202950987697,
      "learning_rate": 2.7694941515330647e-05,
      "loss": 1.3344,
      "step": 1487
    },
    {
      "epoch": 0.380928,
      "grad_norm": 0.39459818264273605,
      "learning_rate": 2.7679937635050528e-05,
      "loss": 1.3952,
      "step": 1488
    },
    {
      "epoch": 0.381184,
      "grad_norm": 0.41533187245370556,
      "learning_rate": 2.7664928683302105e-05,
      "loss": 1.4183,
      "step": 1489
    },
    {
      "epoch": 0.38144,
      "grad_norm": 0.38598544926438805,
      "learning_rate": 2.7649914669996586e-05,
      "loss": 1.3933,
      "step": 1490
    },
    {
      "epoch": 0.381696,
      "grad_norm": 0.3995132263263834,
      "learning_rate": 2.7634895605048507e-05,
      "loss": 1.3949,
      "step": 1491
    },
    {
      "epoch": 0.381952,
      "grad_norm": 0.38762983789706557,
      "learning_rate": 2.761987149837576e-05,
      "loss": 1.3958,
      "step": 1492
    },
    {
      "epoch": 0.382208,
      "grad_norm": 0.41064617496478406,
      "learning_rate": 2.7604842359899548e-05,
      "loss": 1.4421,
      "step": 1493
    },
    {
      "epoch": 0.382464,
      "grad_norm": 0.3878237625353223,
      "learning_rate": 2.758980819954441e-05,
      "loss": 1.3611,
      "step": 1494
    },
    {
      "epoch": 0.38272,
      "grad_norm": 0.4050652158905262,
      "learning_rate": 2.7574769027238192e-05,
      "loss": 1.4051,
      "step": 1495
    },
    {
      "epoch": 0.382976,
      "grad_norm": 0.40194902335667826,
      "learning_rate": 2.755972485291206e-05,
      "loss": 1.3822,
      "step": 1496
    },
    {
      "epoch": 0.383232,
      "grad_norm": 0.3942544952669704,
      "learning_rate": 2.7544675686500463e-05,
      "loss": 1.4076,
      "step": 1497
    },
    {
      "epoch": 0.383488,
      "grad_norm": 0.41318429138575724,
      "learning_rate": 2.7529621537941178e-05,
      "loss": 1.388,
      "step": 1498
    },
    {
      "epoch": 0.383744,
      "grad_norm": 0.39386145940840683,
      "learning_rate": 2.7514562417175245e-05,
      "loss": 1.3904,
      "step": 1499
    },
    {
      "epoch": 0.384,
      "grad_norm": 0.3990637429649315,
      "learning_rate": 2.749949833414699e-05,
      "loss": 1.3692,
      "step": 1500
    },
    {
      "epoch": 0.384256,
      "grad_norm": 0.401501946969665,
      "learning_rate": 2.7484429298804035e-05,
      "loss": 1.3961,
      "step": 1501
    },
    {
      "epoch": 0.384512,
      "grad_norm": 0.4128300084646836,
      "learning_rate": 2.746935532109725e-05,
      "loss": 1.3988,
      "step": 1502
    },
    {
      "epoch": 0.384768,
      "grad_norm": 0.39017572698396386,
      "learning_rate": 2.7454276410980782e-05,
      "loss": 1.3943,
      "step": 1503
    },
    {
      "epoch": 0.385024,
      "grad_norm": 0.40030785606713126,
      "learning_rate": 2.7439192578412038e-05,
      "loss": 1.354,
      "step": 1504
    },
    {
      "epoch": 0.38528,
      "grad_norm": 0.3982596970351547,
      "learning_rate": 2.742410383335166e-05,
      "loss": 1.3905,
      "step": 1505
    },
    {
      "epoch": 0.385536,
      "grad_norm": 0.40237759112183574,
      "learning_rate": 2.740901018576354e-05,
      "loss": 1.3959,
      "step": 1506
    },
    {
      "epoch": 0.385792,
      "grad_norm": 0.39582368540774776,
      "learning_rate": 2.7393911645614823e-05,
      "loss": 1.3791,
      "step": 1507
    },
    {
      "epoch": 0.386048,
      "grad_norm": 0.41353091741160364,
      "learning_rate": 2.7378808222875857e-05,
      "loss": 1.4195,
      "step": 1508
    },
    {
      "epoch": 0.386304,
      "grad_norm": 0.40337474613787083,
      "learning_rate": 2.7363699927520245e-05,
      "loss": 1.3876,
      "step": 1509
    },
    {
      "epoch": 0.38656,
      "grad_norm": 0.4262651794610066,
      "learning_rate": 2.7348586769524787e-05,
      "loss": 1.4114,
      "step": 1510
    },
    {
      "epoch": 0.386816,
      "grad_norm": 0.3973129693246807,
      "learning_rate": 2.733346875886949e-05,
      "loss": 1.4232,
      "step": 1511
    },
    {
      "epoch": 0.387072,
      "grad_norm": 0.4091811894812489,
      "learning_rate": 2.7318345905537586e-05,
      "loss": 1.3488,
      "step": 1512
    },
    {
      "epoch": 0.387328,
      "grad_norm": 0.42086499572736796,
      "learning_rate": 2.730321821951549e-05,
      "loss": 1.4452,
      "step": 1513
    },
    {
      "epoch": 0.387584,
      "grad_norm": 0.40117605061590655,
      "learning_rate": 2.7288085710792803e-05,
      "loss": 1.3529,
      "step": 1514
    },
    {
      "epoch": 0.38784,
      "grad_norm": 0.421129178260488,
      "learning_rate": 2.7272948389362336e-05,
      "loss": 1.3757,
      "step": 1515
    },
    {
      "epoch": 0.388096,
      "grad_norm": 0.39138122388578267,
      "learning_rate": 2.7257806265220054e-05,
      "loss": 1.3908,
      "step": 1516
    },
    {
      "epoch": 0.388352,
      "grad_norm": 0.4164644408626294,
      "learning_rate": 2.7242659348365095e-05,
      "loss": 1.3961,
      "step": 1517
    },
    {
      "epoch": 0.388608,
      "grad_norm": 0.5209449487008662,
      "learning_rate": 2.722750764879978e-05,
      "loss": 1.3795,
      "step": 1518
    },
    {
      "epoch": 0.388864,
      "grad_norm": 0.3911144110290774,
      "learning_rate": 2.721235117652956e-05,
      "loss": 1.3808,
      "step": 1519
    },
    {
      "epoch": 0.38912,
      "grad_norm": 0.39729576106363,
      "learning_rate": 2.7197189941563062e-05,
      "loss": 1.3983,
      "step": 1520
    },
    {
      "epoch": 0.389376,
      "grad_norm": 0.41272447874451607,
      "learning_rate": 2.7182023953912064e-05,
      "loss": 1.3985,
      "step": 1521
    },
    {
      "epoch": 0.389632,
      "grad_norm": 0.3865516565294949,
      "learning_rate": 2.7166853223591443e-05,
      "loss": 1.3647,
      "step": 1522
    },
    {
      "epoch": 0.389888,
      "grad_norm": 0.4021935699821488,
      "learning_rate": 2.715167776061925e-05,
      "loss": 1.3027,
      "step": 1523
    },
    {
      "epoch": 0.390144,
      "grad_norm": 0.39466405197335036,
      "learning_rate": 2.713649757501664e-05,
      "loss": 1.4065,
      "step": 1524
    },
    {
      "epoch": 0.3904,
      "grad_norm": 0.41187275749423646,
      "learning_rate": 2.712131267680789e-05,
      "loss": 1.4002,
      "step": 1525
    },
    {
      "epoch": 0.390656,
      "grad_norm": 0.3885017632820429,
      "learning_rate": 2.7106123076020395e-05,
      "loss": 1.4001,
      "step": 1526
    },
    {
      "epoch": 0.390912,
      "grad_norm": 0.40901946692529934,
      "learning_rate": 2.7090928782684645e-05,
      "loss": 1.3776,
      "step": 1527
    },
    {
      "epoch": 0.391168,
      "grad_norm": 0.3773922895454674,
      "learning_rate": 2.707572980683423e-05,
      "loss": 1.3806,
      "step": 1528
    },
    {
      "epoch": 0.391424,
      "grad_norm": 0.41086430448126915,
      "learning_rate": 2.7060526158505842e-05,
      "loss": 1.3759,
      "step": 1529
    },
    {
      "epoch": 0.39168,
      "grad_norm": 0.40622266435226445,
      "learning_rate": 2.7045317847739257e-05,
      "loss": 1.388,
      "step": 1530
    },
    {
      "epoch": 0.391936,
      "grad_norm": 0.39405916912369365,
      "learning_rate": 2.7030104884577318e-05,
      "loss": 1.3558,
      "step": 1531
    },
    {
      "epoch": 0.392192,
      "grad_norm": 0.39198447399118386,
      "learning_rate": 2.7014887279065957e-05,
      "loss": 1.3405,
      "step": 1532
    },
    {
      "epoch": 0.392448,
      "grad_norm": 0.4012359202522085,
      "learning_rate": 2.6999665041254153e-05,
      "loss": 1.3942,
      "step": 1533
    },
    {
      "epoch": 0.392704,
      "grad_norm": 0.42075057648329756,
      "learning_rate": 2.6984438181193953e-05,
      "loss": 1.3786,
      "step": 1534
    },
    {
      "epoch": 0.39296,
      "grad_norm": 0.3891041261713129,
      "learning_rate": 2.696920670894046e-05,
      "loss": 1.3364,
      "step": 1535
    },
    {
      "epoch": 0.393216,
      "grad_norm": 0.42206820076883245,
      "learning_rate": 2.6953970634551826e-05,
      "loss": 1.405,
      "step": 1536
    },
    {
      "epoch": 0.393472,
      "grad_norm": 0.3998697342408393,
      "learning_rate": 2.693872996808922e-05,
      "loss": 1.3948,
      "step": 1537
    },
    {
      "epoch": 0.393728,
      "grad_norm": 0.4276627844704088,
      "learning_rate": 2.692348471961687e-05,
      "loss": 1.4086,
      "step": 1538
    },
    {
      "epoch": 0.393984,
      "grad_norm": 0.4126012008033037,
      "learning_rate": 2.6908234899202022e-05,
      "loss": 1.4199,
      "step": 1539
    },
    {
      "epoch": 0.39424,
      "grad_norm": 0.4138237646983024,
      "learning_rate": 2.6892980516914928e-05,
      "loss": 1.4126,
      "step": 1540
    },
    {
      "epoch": 0.394496,
      "grad_norm": 0.39757716447399477,
      "learning_rate": 2.6877721582828866e-05,
      "loss": 1.3923,
      "step": 1541
    },
    {
      "epoch": 0.394752,
      "grad_norm": 0.3920675739350379,
      "learning_rate": 2.686245810702011e-05,
      "loss": 1.3315,
      "step": 1542
    },
    {
      "epoch": 0.395008,
      "grad_norm": 0.4025739810672015,
      "learning_rate": 2.684719009956795e-05,
      "loss": 1.4146,
      "step": 1543
    },
    {
      "epoch": 0.395264,
      "grad_norm": 0.39771850671881304,
      "learning_rate": 2.683191757055465e-05,
      "loss": 1.3387,
      "step": 1544
    },
    {
      "epoch": 0.39552,
      "grad_norm": 0.3990656693840695,
      "learning_rate": 2.6816640530065473e-05,
      "loss": 1.3829,
      "step": 1545
    },
    {
      "epoch": 0.395776,
      "grad_norm": 0.40071890774475416,
      "learning_rate": 2.680135898818865e-05,
      "loss": 1.3765,
      "step": 1546
    },
    {
      "epoch": 0.396032,
      "grad_norm": 0.39735506237863144,
      "learning_rate": 2.6786072955015385e-05,
      "loss": 1.3892,
      "step": 1547
    },
    {
      "epoch": 0.396288,
      "grad_norm": 0.4175223299259817,
      "learning_rate": 2.6770782440639863e-05,
      "loss": 1.3874,
      "step": 1548
    },
    {
      "epoch": 0.396544,
      "grad_norm": 0.433375401605186,
      "learning_rate": 2.675548745515922e-05,
      "loss": 1.3757,
      "step": 1549
    },
    {
      "epoch": 0.3968,
      "grad_norm": 0.3930721801250507,
      "learning_rate": 2.674018800867353e-05,
      "loss": 1.3744,
      "step": 1550
    },
    {
      "epoch": 0.397056,
      "grad_norm": 0.3954251438693906,
      "learning_rate": 2.6724884111285833e-05,
      "loss": 1.3824,
      "step": 1551
    },
    {
      "epoch": 0.397312,
      "grad_norm": 0.40441968467011125,
      "learning_rate": 2.67095757731021e-05,
      "loss": 1.3769,
      "step": 1552
    },
    {
      "epoch": 0.397568,
      "grad_norm": 0.4183542898345257,
      "learning_rate": 2.6694263004231233e-05,
      "loss": 1.3533,
      "step": 1553
    },
    {
      "epoch": 0.397824,
      "grad_norm": 0.40230419987606225,
      "learning_rate": 2.6678945814785064e-05,
      "loss": 1.4029,
      "step": 1554
    },
    {
      "epoch": 0.39808,
      "grad_norm": 0.38673022424732684,
      "learning_rate": 2.6663624214878346e-05,
      "loss": 1.3579,
      "step": 1555
    },
    {
      "epoch": 0.398336,
      "grad_norm": 0.4023746571208338,
      "learning_rate": 2.664829821462872e-05,
      "loss": 1.3966,
      "step": 1556
    },
    {
      "epoch": 0.398592,
      "grad_norm": 0.4124529070247121,
      "learning_rate": 2.6632967824156787e-05,
      "loss": 1.3776,
      "step": 1557
    },
    {
      "epoch": 0.398848,
      "grad_norm": 0.397544880264132,
      "learning_rate": 2.6617633053585982e-05,
      "loss": 1.3969,
      "step": 1558
    },
    {
      "epoch": 0.399104,
      "grad_norm": 0.38090972098310716,
      "learning_rate": 2.660229391304268e-05,
      "loss": 1.3499,
      "step": 1559
    },
    {
      "epoch": 0.39936,
      "grad_norm": 0.5782454723460566,
      "learning_rate": 2.6586950412656123e-05,
      "loss": 1.3468,
      "step": 1560
    },
    {
      "epoch": 0.399616,
      "grad_norm": 0.3993088875489005,
      "learning_rate": 2.6571602562558437e-05,
      "loss": 1.371,
      "step": 1561
    },
    {
      "epoch": 0.399872,
      "grad_norm": 0.40302494861911015,
      "learning_rate": 2.6556250372884607e-05,
      "loss": 1.3778,
      "step": 1562
    },
    {
      "epoch": 0.400128,
      "grad_norm": 0.4114994326643323,
      "learning_rate": 2.6540893853772503e-05,
      "loss": 1.3992,
      "step": 1563
    },
    {
      "epoch": 0.400384,
      "grad_norm": 0.40272367426831235,
      "learning_rate": 2.6525533015362842e-05,
      "loss": 1.4196,
      "step": 1564
    },
    {
      "epoch": 0.40064,
      "grad_norm": 0.427405695044878,
      "learning_rate": 2.65101678677992e-05,
      "loss": 1.4068,
      "step": 1565
    },
    {
      "epoch": 0.400896,
      "grad_norm": 0.3913728419721429,
      "learning_rate": 2.6494798421227998e-05,
      "loss": 1.3798,
      "step": 1566
    },
    {
      "epoch": 0.401152,
      "grad_norm": 0.4171393815183837,
      "learning_rate": 2.6479424685798486e-05,
      "loss": 1.3898,
      "step": 1567
    },
    {
      "epoch": 0.401408,
      "grad_norm": 0.42903904918849795,
      "learning_rate": 2.646404667166275e-05,
      "loss": 1.4504,
      "step": 1568
    },
    {
      "epoch": 0.401664,
      "grad_norm": 0.40441914554955816,
      "learning_rate": 2.6448664388975715e-05,
      "loss": 1.363,
      "step": 1569
    },
    {
      "epoch": 0.40192,
      "grad_norm": 0.4021049132089498,
      "learning_rate": 2.6433277847895112e-05,
      "loss": 1.3656,
      "step": 1570
    },
    {
      "epoch": 0.402176,
      "grad_norm": 0.4126885458524139,
      "learning_rate": 2.6417887058581477e-05,
      "loss": 1.4309,
      "step": 1571
    },
    {
      "epoch": 0.402432,
      "grad_norm": 0.40419344696231346,
      "learning_rate": 2.640249203119817e-05,
      "loss": 1.3955,
      "step": 1572
    },
    {
      "epoch": 0.402688,
      "grad_norm": 0.4298677193961361,
      "learning_rate": 2.6387092775911346e-05,
      "loss": 1.39,
      "step": 1573
    },
    {
      "epoch": 0.402944,
      "grad_norm": 0.39841193163686267,
      "learning_rate": 2.6371689302889932e-05,
      "loss": 1.3903,
      "step": 1574
    },
    {
      "epoch": 0.4032,
      "grad_norm": 0.43115914304598324,
      "learning_rate": 2.635628162230566e-05,
      "loss": 1.3853,
      "step": 1575
    },
    {
      "epoch": 0.403456,
      "grad_norm": 0.3949831876606797,
      "learning_rate": 2.634086974433304e-05,
      "loss": 1.3685,
      "step": 1576
    },
    {
      "epoch": 0.403712,
      "grad_norm": 0.5145054740614577,
      "learning_rate": 2.6325453679149345e-05,
      "loss": 1.3827,
      "step": 1577
    },
    {
      "epoch": 0.403968,
      "grad_norm": 0.40356041639044127,
      "learning_rate": 2.631003343693461e-05,
      "loss": 1.3764,
      "step": 1578
    },
    {
      "epoch": 0.404224,
      "grad_norm": 0.40962786395355416,
      "learning_rate": 2.629460902787165e-05,
      "loss": 1.3928,
      "step": 1579
    },
    {
      "epoch": 0.40448,
      "grad_norm": 0.3951557965433583,
      "learning_rate": 2.6279180462146007e-05,
      "loss": 1.3821,
      "step": 1580
    },
    {
      "epoch": 0.404736,
      "grad_norm": 0.42080773079936673,
      "learning_rate": 2.626374774994597e-05,
      "loss": 1.3775,
      "step": 1581
    },
    {
      "epoch": 0.404992,
      "grad_norm": 0.4019048141081693,
      "learning_rate": 2.624831090146259e-05,
      "loss": 1.3775,
      "step": 1582
    },
    {
      "epoch": 0.405248,
      "grad_norm": 0.42041255372460984,
      "learning_rate": 2.623286992688963e-05,
      "loss": 1.3865,
      "step": 1583
    },
    {
      "epoch": 0.405504,
      "grad_norm": 0.4171818585494644,
      "learning_rate": 2.6217424836423568e-05,
      "loss": 1.3765,
      "step": 1584
    },
    {
      "epoch": 0.40576,
      "grad_norm": 0.4206426386265979,
      "learning_rate": 2.620197564026363e-05,
      "loss": 1.3837,
      "step": 1585
    },
    {
      "epoch": 0.406016,
      "grad_norm": 0.40643713965187,
      "learning_rate": 2.618652234861172e-05,
      "loss": 1.3902,
      "step": 1586
    },
    {
      "epoch": 0.406272,
      "grad_norm": 0.4103606704571333,
      "learning_rate": 2.617106497167247e-05,
      "loss": 1.3846,
      "step": 1587
    },
    {
      "epoch": 0.406528,
      "grad_norm": 0.40351380650539226,
      "learning_rate": 2.6155603519653204e-05,
      "loss": 1.4024,
      "step": 1588
    },
    {
      "epoch": 0.406784,
      "grad_norm": 0.3935193501074164,
      "learning_rate": 2.6140138002763934e-05,
      "loss": 1.3798,
      "step": 1589
    },
    {
      "epoch": 0.40704,
      "grad_norm": 0.39594658998954835,
      "learning_rate": 2.6124668431217348e-05,
      "loss": 1.3504,
      "step": 1590
    },
    {
      "epoch": 0.407296,
      "grad_norm": 0.4263075340248201,
      "learning_rate": 2.610919481522884e-05,
      "loss": 1.4077,
      "step": 1591
    },
    {
      "epoch": 0.407552,
      "grad_norm": 0.4054166615334729,
      "learning_rate": 2.6093717165016435e-05,
      "loss": 1.3901,
      "step": 1592
    },
    {
      "epoch": 0.407808,
      "grad_norm": 0.40408189295138713,
      "learning_rate": 2.6078235490800854e-05,
      "loss": 1.3998,
      "step": 1593
    },
    {
      "epoch": 0.408064,
      "grad_norm": 0.39219033299624,
      "learning_rate": 2.606274980280547e-05,
      "loss": 1.3429,
      "step": 1594
    },
    {
      "epoch": 0.40832,
      "grad_norm": 0.4198503629754004,
      "learning_rate": 2.6047260111256285e-05,
      "loss": 1.3918,
      "step": 1595
    },
    {
      "epoch": 0.408576,
      "grad_norm": 0.3992637269410784,
      "learning_rate": 2.6031766426381966e-05,
      "loss": 1.3521,
      "step": 1596
    },
    {
      "epoch": 0.408832,
      "grad_norm": 0.420595935974799,
      "learning_rate": 2.601626875841382e-05,
      "loss": 1.3826,
      "step": 1597
    },
    {
      "epoch": 0.409088,
      "grad_norm": 0.4073796415180875,
      "learning_rate": 2.6000767117585757e-05,
      "loss": 1.3647,
      "step": 1598
    },
    {
      "epoch": 0.409344,
      "grad_norm": 0.41121208429459427,
      "learning_rate": 2.598526151413435e-05,
      "loss": 1.3888,
      "step": 1599
    },
    {
      "epoch": 0.4096,
      "grad_norm": 0.41195076366901184,
      "learning_rate": 2.5969751958298757e-05,
      "loss": 1.3756,
      "step": 1600
    },
    {
      "epoch": 0.409856,
      "grad_norm": 0.3948716051630853,
      "learning_rate": 2.595423846032075e-05,
      "loss": 1.3564,
      "step": 1601
    },
    {
      "epoch": 0.410112,
      "grad_norm": 0.4952150340783802,
      "learning_rate": 2.593872103044472e-05,
      "loss": 1.3446,
      "step": 1602
    },
    {
      "epoch": 0.410368,
      "grad_norm": 0.41606517793409226,
      "learning_rate": 2.592319967891765e-05,
      "loss": 1.4168,
      "step": 1603
    },
    {
      "epoch": 0.410624,
      "grad_norm": 0.3977857583909906,
      "learning_rate": 2.5907674415989093e-05,
      "loss": 1.4139,
      "step": 1604
    },
    {
      "epoch": 0.41088,
      "grad_norm": 0.4139785295481217,
      "learning_rate": 2.5892145251911212e-05,
      "loss": 1.3831,
      "step": 1605
    },
    {
      "epoch": 0.411136,
      "grad_norm": 0.4050706492706079,
      "learning_rate": 2.5876612196938733e-05,
      "loss": 1.3645,
      "step": 1606
    },
    {
      "epoch": 0.411392,
      "grad_norm": 0.3972742936957601,
      "learning_rate": 2.586107526132894e-05,
      "loss": 1.3982,
      "step": 1607
    },
    {
      "epoch": 0.411648,
      "grad_norm": 0.3963830822595578,
      "learning_rate": 2.584553445534171e-05,
      "loss": 1.3901,
      "step": 1608
    },
    {
      "epoch": 0.411904,
      "grad_norm": 0.39906050797707837,
      "learning_rate": 2.5829989789239444e-05,
      "loss": 1.4044,
      "step": 1609
    },
    {
      "epoch": 0.41216,
      "grad_norm": 0.4041355026275094,
      "learning_rate": 2.5814441273287107e-05,
      "loss": 1.3526,
      "step": 1610
    },
    {
      "epoch": 0.412416,
      "grad_norm": 0.3901220716060119,
      "learning_rate": 2.5798888917752214e-05,
      "loss": 1.3944,
      "step": 1611
    },
    {
      "epoch": 0.412672,
      "grad_norm": 0.3951631264314115,
      "learning_rate": 2.5783332732904804e-05,
      "loss": 1.374,
      "step": 1612
    },
    {
      "epoch": 0.412928,
      "grad_norm": 0.3969236325227861,
      "learning_rate": 2.576777272901744e-05,
      "loss": 1.3745,
      "step": 1613
    },
    {
      "epoch": 0.413184,
      "grad_norm": 0.399605500782134,
      "learning_rate": 2.5752208916365212e-05,
      "loss": 1.3849,
      "step": 1614
    },
    {
      "epoch": 0.41344,
      "grad_norm": 0.3933122149760464,
      "learning_rate": 2.5736641305225743e-05,
      "loss": 1.3861,
      "step": 1615
    },
    {
      "epoch": 0.413696,
      "grad_norm": 0.40477851704107515,
      "learning_rate": 2.5721069905879133e-05,
      "loss": 1.4244,
      "step": 1616
    },
    {
      "epoch": 0.413952,
      "grad_norm": 0.3954583940407715,
      "learning_rate": 2.570549472860801e-05,
      "loss": 1.3637,
      "step": 1617
    },
    {
      "epoch": 0.414208,
      "grad_norm": 0.3959632146371025,
      "learning_rate": 2.5689915783697482e-05,
      "loss": 1.3453,
      "step": 1618
    },
    {
      "epoch": 0.414464,
      "grad_norm": 0.3991134814273877,
      "learning_rate": 2.5674333081435143e-05,
      "loss": 1.37,
      "step": 1619
    },
    {
      "epoch": 0.41472,
      "grad_norm": 0.4051899206058478,
      "learning_rate": 2.5658746632111074e-05,
      "loss": 1.3795,
      "step": 1620
    },
    {
      "epoch": 0.414976,
      "grad_norm": 0.4185562541025897,
      "learning_rate": 2.5643156446017848e-05,
      "loss": 1.3535,
      "step": 1621
    },
    {
      "epoch": 0.415232,
      "grad_norm": 0.38986333830470227,
      "learning_rate": 2.5627562533450472e-05,
      "loss": 1.3604,
      "step": 1622
    },
    {
      "epoch": 0.415488,
      "grad_norm": 0.42337006686198586,
      "learning_rate": 2.5611964904706428e-05,
      "loss": 1.3832,
      "step": 1623
    },
    {
      "epoch": 0.415744,
      "grad_norm": 0.405324673797134,
      "learning_rate": 2.5596363570085662e-05,
      "loss": 1.4134,
      "step": 1624
    },
    {
      "epoch": 0.416,
      "grad_norm": 0.39206757474833637,
      "learning_rate": 2.5580758539890553e-05,
      "loss": 1.3489,
      "step": 1625
    },
    {
      "epoch": 0.416256,
      "grad_norm": 0.42518732036158596,
      "learning_rate": 2.5565149824425934e-05,
      "loss": 1.3886,
      "step": 1626
    },
    {
      "epoch": 0.416512,
      "grad_norm": 0.4115238669294899,
      "learning_rate": 2.5549537433999056e-05,
      "loss": 1.3808,
      "step": 1627
    },
    {
      "epoch": 0.416768,
      "grad_norm": 0.40265154562167854,
      "learning_rate": 2.5533921378919613e-05,
      "loss": 1.3681,
      "step": 1628
    },
    {
      "epoch": 0.417024,
      "grad_norm": 0.39696003699747384,
      "learning_rate": 2.55183016694997e-05,
      "loss": 1.384,
      "step": 1629
    },
    {
      "epoch": 0.41728,
      "grad_norm": 0.4235769321836882,
      "learning_rate": 2.5502678316053843e-05,
      "loss": 1.3844,
      "step": 1630
    },
    {
      "epoch": 0.417536,
      "grad_norm": 0.41409911866652294,
      "learning_rate": 2.548705132889896e-05,
      "loss": 1.3767,
      "step": 1631
    },
    {
      "epoch": 0.417792,
      "grad_norm": 0.48422399562662255,
      "learning_rate": 2.5471420718354388e-05,
      "loss": 1.38,
      "step": 1632
    },
    {
      "epoch": 0.418048,
      "grad_norm": 0.42036501796305853,
      "learning_rate": 2.5455786494741837e-05,
      "loss": 1.3975,
      "step": 1633
    },
    {
      "epoch": 0.418304,
      "grad_norm": 0.4195381697731602,
      "learning_rate": 2.544014866838541e-05,
      "loss": 1.4076,
      "step": 1634
    },
    {
      "epoch": 0.41856,
      "grad_norm": 0.4548925116959188,
      "learning_rate": 2.542450724961159e-05,
      "loss": 1.3902,
      "step": 1635
    },
    {
      "epoch": 0.418816,
      "grad_norm": 0.3940400315054187,
      "learning_rate": 2.540886224874923e-05,
      "loss": 1.3779,
      "step": 1636
    },
    {
      "epoch": 0.419072,
      "grad_norm": 0.43945606726245035,
      "learning_rate": 2.5393213676129554e-05,
      "loss": 1.3826,
      "step": 1637
    },
    {
      "epoch": 0.419328,
      "grad_norm": 0.4163662354538395,
      "learning_rate": 2.5377561542086137e-05,
      "loss": 1.4204,
      "step": 1638
    },
    {
      "epoch": 0.419584,
      "grad_norm": 0.4193152758276446,
      "learning_rate": 2.5361905856954914e-05,
      "loss": 1.4037,
      "step": 1639
    },
    {
      "epoch": 0.41984,
      "grad_norm": 0.44274764480832757,
      "learning_rate": 2.534624663107416e-05,
      "loss": 1.3718,
      "step": 1640
    },
    {
      "epoch": 0.420096,
      "grad_norm": 0.3876028697996953,
      "learning_rate": 2.5330583874784482e-05,
      "loss": 1.3503,
      "step": 1641
    },
    {
      "epoch": 0.420352,
      "grad_norm": 0.41644118695416305,
      "learning_rate": 2.5314917598428825e-05,
      "loss": 1.4129,
      "step": 1642
    },
    {
      "epoch": 0.420608,
      "grad_norm": 0.39810268710247143,
      "learning_rate": 2.5299247812352473e-05,
      "loss": 1.3932,
      "step": 1643
    },
    {
      "epoch": 0.420864,
      "grad_norm": 0.40326367954337183,
      "learning_rate": 2.5283574526902993e-05,
      "loss": 1.3481,
      "step": 1644
    },
    {
      "epoch": 0.42112,
      "grad_norm": 0.39892787672129604,
      "learning_rate": 2.5267897752430298e-05,
      "loss": 1.349,
      "step": 1645
    },
    {
      "epoch": 0.421376,
      "grad_norm": 0.396709095168283,
      "learning_rate": 2.5252217499286588e-05,
      "loss": 1.3434,
      "step": 1646
    },
    {
      "epoch": 0.421632,
      "grad_norm": 0.41170280773964757,
      "learning_rate": 2.5236533777826352e-05,
      "loss": 1.4296,
      "step": 1647
    },
    {
      "epoch": 0.421888,
      "grad_norm": 0.3943068969914347,
      "learning_rate": 2.522084659840638e-05,
      "loss": 1.4062,
      "step": 1648
    },
    {
      "epoch": 0.422144,
      "grad_norm": 0.41427815670803986,
      "learning_rate": 2.5205155971385767e-05,
      "loss": 1.3614,
      "step": 1649
    },
    {
      "epoch": 0.4224,
      "grad_norm": 0.47798552404873934,
      "learning_rate": 2.518946190712584e-05,
      "loss": 1.3762,
      "step": 1650
    },
    {
      "epoch": 0.422656,
      "grad_norm": 0.4035673682359989,
      "learning_rate": 2.517376441599022e-05,
      "loss": 1.3922,
      "step": 1651
    },
    {
      "epoch": 0.422912,
      "grad_norm": 0.4253600897756832,
      "learning_rate": 2.5158063508344795e-05,
      "loss": 1.3904,
      "step": 1652
    },
    {
      "epoch": 0.423168,
      "grad_norm": 0.4501247555429117,
      "learning_rate": 2.51423591945577e-05,
      "loss": 1.362,
      "step": 1653
    },
    {
      "epoch": 0.423424,
      "grad_norm": 0.40857521892542986,
      "learning_rate": 2.5126651484999326e-05,
      "loss": 1.3682,
      "step": 1654
    },
    {
      "epoch": 0.42368,
      "grad_norm": 0.3973805906450857,
      "learning_rate": 2.5110940390042307e-05,
      "loss": 1.3425,
      "step": 1655
    },
    {
      "epoch": 0.423936,
      "grad_norm": 0.39188616018937367,
      "learning_rate": 2.5095225920061497e-05,
      "loss": 1.3785,
      "step": 1656
    },
    {
      "epoch": 0.424192,
      "grad_norm": 0.4091780966827691,
      "learning_rate": 2.5079508085433997e-05,
      "loss": 1.3701,
      "step": 1657
    },
    {
      "epoch": 0.424448,
      "grad_norm": 0.3996626694571495,
      "learning_rate": 2.5063786896539127e-05,
      "loss": 1.404,
      "step": 1658
    },
    {
      "epoch": 0.424704,
      "grad_norm": 0.40939621792263153,
      "learning_rate": 2.5048062363758402e-05,
      "loss": 1.3714,
      "step": 1659
    },
    {
      "epoch": 0.42496,
      "grad_norm": 0.3971689720247293,
      "learning_rate": 2.503233449747558e-05,
      "loss": 1.3984,
      "step": 1660
    },
    {
      "epoch": 0.425216,
      "grad_norm": 0.3821679446009892,
      "learning_rate": 2.5016603308076595e-05,
      "loss": 1.3376,
      "step": 1661
    },
    {
      "epoch": 0.425472,
      "grad_norm": 0.4043898043823833,
      "learning_rate": 2.500086880594958e-05,
      "loss": 1.3666,
      "step": 1662
    },
    {
      "epoch": 0.425728,
      "grad_norm": 0.39778875096692295,
      "learning_rate": 2.4985131001484856e-05,
      "loss": 1.3714,
      "step": 1663
    },
    {
      "epoch": 0.425984,
      "grad_norm": 0.38665387374871996,
      "learning_rate": 2.4969389905074933e-05,
      "loss": 1.3628,
      "step": 1664
    },
    {
      "epoch": 0.42624,
      "grad_norm": 0.40061361960772834,
      "learning_rate": 2.495364552711448e-05,
      "loss": 1.4037,
      "step": 1665
    },
    {
      "epoch": 0.426496,
      "grad_norm": 0.4097538578443415,
      "learning_rate": 2.4937897878000342e-05,
      "loss": 1.3997,
      "step": 1666
    },
    {
      "epoch": 0.426752,
      "grad_norm": 0.4194364300436425,
      "learning_rate": 2.4922146968131532e-05,
      "loss": 1.4175,
      "step": 1667
    },
    {
      "epoch": 0.427008,
      "grad_norm": 0.3904163363390266,
      "learning_rate": 2.4906392807909197e-05,
      "loss": 1.3648,
      "step": 1668
    },
    {
      "epoch": 0.427264,
      "grad_norm": 0.38829791877734754,
      "learning_rate": 2.489063540773665e-05,
      "loss": 1.378,
      "step": 1669
    },
    {
      "epoch": 0.42752,
      "grad_norm": 0.39804854115858945,
      "learning_rate": 2.4874874778019325e-05,
      "loss": 1.3753,
      "step": 1670
    },
    {
      "epoch": 0.427776,
      "grad_norm": 0.40394219971487694,
      "learning_rate": 2.4859110929164804e-05,
      "loss": 1.3904,
      "step": 1671
    },
    {
      "epoch": 0.428032,
      "grad_norm": 0.40593103276439263,
      "learning_rate": 2.4843343871582782e-05,
      "loss": 1.442,
      "step": 1672
    },
    {
      "epoch": 0.428288,
      "grad_norm": 0.40883603705905375,
      "learning_rate": 2.4827573615685094e-05,
      "loss": 1.3966,
      "step": 1673
    },
    {
      "epoch": 0.428544,
      "grad_norm": 0.39298387029552795,
      "learning_rate": 2.4811800171885657e-05,
      "loss": 1.3697,
      "step": 1674
    },
    {
      "epoch": 0.4288,
      "grad_norm": 0.3975635468666579,
      "learning_rate": 2.4796023550600514e-05,
      "loss": 1.3686,
      "step": 1675
    },
    {
      "epoch": 0.429056,
      "grad_norm": 0.4039602951835481,
      "learning_rate": 2.47802437622478e-05,
      "loss": 1.3925,
      "step": 1676
    },
    {
      "epoch": 0.429312,
      "grad_norm": 0.3937224765480824,
      "learning_rate": 2.476446081724775e-05,
      "loss": 1.3974,
      "step": 1677
    },
    {
      "epoch": 0.429568,
      "grad_norm": 0.3962916721102795,
      "learning_rate": 2.4748674726022664e-05,
      "loss": 1.3221,
      "step": 1678
    },
    {
      "epoch": 0.429824,
      "grad_norm": 0.43345760168637193,
      "learning_rate": 2.4732885498996946e-05,
      "loss": 1.3095,
      "step": 1679
    },
    {
      "epoch": 0.43008,
      "grad_norm": 0.5311447891485607,
      "learning_rate": 2.471709314659704e-05,
      "loss": 1.388,
      "step": 1680
    },
    {
      "epoch": 0.430336,
      "grad_norm": 0.40774384231581445,
      "learning_rate": 2.4701297679251477e-05,
      "loss": 1.415,
      "step": 1681
    },
    {
      "epoch": 0.430592,
      "grad_norm": 0.3959012487119448,
      "learning_rate": 2.4685499107390845e-05,
      "loss": 1.3379,
      "step": 1682
    },
    {
      "epoch": 0.430848,
      "grad_norm": 0.39720729155074513,
      "learning_rate": 2.4669697441447765e-05,
      "loss": 1.3943,
      "step": 1683
    },
    {
      "epoch": 0.431104,
      "grad_norm": 0.4044871575064922,
      "learning_rate": 2.4653892691856917e-05,
      "loss": 1.3809,
      "step": 1684
    },
    {
      "epoch": 0.43136,
      "grad_norm": 0.4128011927443433,
      "learning_rate": 2.463808486905501e-05,
      "loss": 1.3692,
      "step": 1685
    },
    {
      "epoch": 0.431616,
      "grad_norm": 0.3945382851301297,
      "learning_rate": 2.4622273983480786e-05,
      "loss": 1.4103,
      "step": 1686
    },
    {
      "epoch": 0.431872,
      "grad_norm": 0.40348049733922964,
      "learning_rate": 2.4606460045575e-05,
      "loss": 1.4062,
      "step": 1687
    },
    {
      "epoch": 0.432128,
      "grad_norm": 0.39468307064477304,
      "learning_rate": 2.4590643065780442e-05,
      "loss": 1.3561,
      "step": 1688
    },
    {
      "epoch": 0.432384,
      "grad_norm": 0.38637036351251514,
      "learning_rate": 2.4574823054541888e-05,
      "loss": 1.376,
      "step": 1689
    },
    {
      "epoch": 0.43264,
      "grad_norm": 0.5706557146219178,
      "learning_rate": 2.4559000022306138e-05,
      "loss": 1.3618,
      "step": 1690
    },
    {
      "epoch": 0.432896,
      "grad_norm": 0.4014697464370375,
      "learning_rate": 2.4543173979521963e-05,
      "loss": 1.3738,
      "step": 1691
    },
    {
      "epoch": 0.433152,
      "grad_norm": 0.4234522871570587,
      "learning_rate": 2.4527344936640144e-05,
      "loss": 1.3997,
      "step": 1692
    },
    {
      "epoch": 0.433408,
      "grad_norm": 0.40070027803367486,
      "learning_rate": 2.451151290411342e-05,
      "loss": 1.3701,
      "step": 1693
    },
    {
      "epoch": 0.433664,
      "grad_norm": 0.4223395250073125,
      "learning_rate": 2.4495677892396536e-05,
      "loss": 1.3876,
      "step": 1694
    },
    {
      "epoch": 0.43392,
      "grad_norm": 0.39766966616935306,
      "learning_rate": 2.447983991194618e-05,
      "loss": 1.3846,
      "step": 1695
    },
    {
      "epoch": 0.434176,
      "grad_norm": 0.395694837718773,
      "learning_rate": 2.4463998973221008e-05,
      "loss": 1.3401,
      "step": 1696
    },
    {
      "epoch": 0.434432,
      "grad_norm": 0.43721143989623745,
      "learning_rate": 2.444815508668162e-05,
      "loss": 1.3799,
      "step": 1697
    },
    {
      "epoch": 0.434688,
      "grad_norm": 0.3977996446282918,
      "learning_rate": 2.4432308262790578e-05,
      "loss": 1.3811,
      "step": 1698
    },
    {
      "epoch": 0.434944,
      "grad_norm": 0.3898714921064679,
      "learning_rate": 2.441645851201237e-05,
      "loss": 1.3798,
      "step": 1699
    },
    {
      "epoch": 0.4352,
      "grad_norm": 0.4120512566566544,
      "learning_rate": 2.4400605844813435e-05,
      "loss": 1.4282,
      "step": 1700
    },
    {
      "epoch": 0.435456,
      "grad_norm": 0.3855122018243483,
      "learning_rate": 2.4384750271662115e-05,
      "loss": 1.358,
      "step": 1701
    },
    {
      "epoch": 0.435712,
      "grad_norm": 0.40242886120197624,
      "learning_rate": 2.4368891803028686e-05,
      "loss": 1.3855,
      "step": 1702
    },
    {
      "epoch": 0.435968,
      "grad_norm": 0.40703300960909283,
      "learning_rate": 2.435303044938533e-05,
      "loss": 1.402,
      "step": 1703
    },
    {
      "epoch": 0.436224,
      "grad_norm": 0.4092185102587738,
      "learning_rate": 2.433716622120615e-05,
      "loss": 1.3923,
      "step": 1704
    },
    {
      "epoch": 0.43648,
      "grad_norm": 0.3966975647335042,
      "learning_rate": 2.432129912896711e-05,
      "loss": 1.3442,
      "step": 1705
    },
    {
      "epoch": 0.436736,
      "grad_norm": 0.39348873043614585,
      "learning_rate": 2.4305429183146103e-05,
      "loss": 1.3597,
      "step": 1706
    },
    {
      "epoch": 0.436992,
      "grad_norm": 0.39246132168166764,
      "learning_rate": 2.4289556394222898e-05,
      "loss": 1.3902,
      "step": 1707
    },
    {
      "epoch": 0.437248,
      "grad_norm": 0.41949718832606925,
      "learning_rate": 2.4273680772679117e-05,
      "loss": 1.3581,
      "step": 1708
    },
    {
      "epoch": 0.437504,
      "grad_norm": 0.3847202589782453,
      "learning_rate": 2.4257802328998282e-05,
      "loss": 1.3445,
      "step": 1709
    },
    {
      "epoch": 0.43776,
      "grad_norm": 0.3980203822033651,
      "learning_rate": 2.4241921073665774e-05,
      "loss": 1.4048,
      "step": 1710
    },
    {
      "epoch": 0.438016,
      "grad_norm": 0.3998175690546011,
      "learning_rate": 2.422603701716881e-05,
      "loss": 1.3807,
      "step": 1711
    },
    {
      "epoch": 0.438272,
      "grad_norm": 0.4005580201490593,
      "learning_rate": 2.4210150169996488e-05,
      "loss": 1.3966,
      "step": 1712
    },
    {
      "epoch": 0.438528,
      "grad_norm": 0.3892653316982133,
      "learning_rate": 2.4194260542639718e-05,
      "loss": 1.4274,
      "step": 1713
    },
    {
      "epoch": 0.438784,
      "grad_norm": 0.4030769465151541,
      "learning_rate": 2.4178368145591263e-05,
      "loss": 1.4513,
      "step": 1714
    },
    {
      "epoch": 0.43904,
      "grad_norm": 0.3877759525628278,
      "learning_rate": 2.4162472989345712e-05,
      "loss": 1.3453,
      "step": 1715
    },
    {
      "epoch": 0.439296,
      "grad_norm": 0.3990219360361329,
      "learning_rate": 2.4146575084399486e-05,
      "loss": 1.389,
      "step": 1716
    },
    {
      "epoch": 0.439552,
      "grad_norm": 0.3918722994154548,
      "learning_rate": 2.4130674441250794e-05,
      "loss": 1.3979,
      "step": 1717
    },
    {
      "epoch": 0.439808,
      "grad_norm": 0.3959758657045125,
      "learning_rate": 2.4114771070399682e-05,
      "loss": 1.3684,
      "step": 1718
    },
    {
      "epoch": 0.440064,
      "grad_norm": 0.40699097011603336,
      "learning_rate": 2.4098864982347987e-05,
      "loss": 1.421,
      "step": 1719
    },
    {
      "epoch": 0.44032,
      "grad_norm": 0.405875905668041,
      "learning_rate": 2.408295618759932e-05,
      "loss": 1.4194,
      "step": 1720
    },
    {
      "epoch": 0.440576,
      "grad_norm": 0.3937131816620548,
      "learning_rate": 2.4067044696659123e-05,
      "loss": 1.3582,
      "step": 1721
    },
    {
      "epoch": 0.440832,
      "grad_norm": 0.40362599806045574,
      "learning_rate": 2.4051130520034583e-05,
      "loss": 1.4082,
      "step": 1722
    },
    {
      "epoch": 0.441088,
      "grad_norm": 0.40096338058688386,
      "learning_rate": 2.4035213668234665e-05,
      "loss": 1.3576,
      "step": 1723
    },
    {
      "epoch": 0.441344,
      "grad_norm": 0.392911513846458,
      "learning_rate": 2.4019294151770112e-05,
      "loss": 1.397,
      "step": 1724
    },
    {
      "epoch": 0.4416,
      "grad_norm": 0.41780738843330906,
      "learning_rate": 2.400337198115343e-05,
      "loss": 1.3646,
      "step": 1725
    },
    {
      "epoch": 0.441856,
      "grad_norm": 0.3866055700530504,
      "learning_rate": 2.3987447166898852e-05,
      "loss": 1.3027,
      "step": 1726
    },
    {
      "epoch": 0.442112,
      "grad_norm": 0.40076950230821623,
      "learning_rate": 2.397151971952238e-05,
      "loss": 1.382,
      "step": 1727
    },
    {
      "epoch": 0.442368,
      "grad_norm": 0.4365448033887497,
      "learning_rate": 2.395558964954176e-05,
      "loss": 1.411,
      "step": 1728
    },
    {
      "epoch": 0.442624,
      "grad_norm": 0.4848205715225887,
      "learning_rate": 2.3939656967476438e-05,
      "loss": 1.3677,
      "step": 1729
    },
    {
      "epoch": 0.44288,
      "grad_norm": 0.40848934176686613,
      "learning_rate": 2.3923721683847622e-05,
      "loss": 1.3604,
      "step": 1730
    },
    {
      "epoch": 0.443136,
      "grad_norm": 0.4086949613269775,
      "learning_rate": 2.3907783809178217e-05,
      "loss": 1.4061,
      "step": 1731
    },
    {
      "epoch": 0.443392,
      "grad_norm": 0.40735356106168075,
      "learning_rate": 2.3891843353992834e-05,
      "loss": 1.3478,
      "step": 1732
    },
    {
      "epoch": 0.443648,
      "grad_norm": 0.39930100878519625,
      "learning_rate": 2.3875900328817803e-05,
      "loss": 1.3767,
      "step": 1733
    },
    {
      "epoch": 0.443904,
      "grad_norm": 0.40953445784506254,
      "learning_rate": 2.3859954744181155e-05,
      "loss": 1.4088,
      "step": 1734
    },
    {
      "epoch": 0.44416,
      "grad_norm": 0.3996727107528165,
      "learning_rate": 2.384400661061259e-05,
      "loss": 1.4105,
      "step": 1735
    },
    {
      "epoch": 0.444416,
      "grad_norm": 0.3904639117082841,
      "learning_rate": 2.38280559386435e-05,
      "loss": 1.3837,
      "step": 1736
    },
    {
      "epoch": 0.444672,
      "grad_norm": 0.39470341174968043,
      "learning_rate": 2.3812102738806972e-05,
      "loss": 1.3887,
      "step": 1737
    },
    {
      "epoch": 0.444928,
      "grad_norm": 0.418702806625604,
      "learning_rate": 2.3796147021637727e-05,
      "loss": 1.3768,
      "step": 1738
    },
    {
      "epoch": 0.445184,
      "grad_norm": 0.4125066196253555,
      "learning_rate": 2.3780188797672174e-05,
      "loss": 1.3837,
      "step": 1739
    },
    {
      "epoch": 0.44544,
      "grad_norm": 0.3722167887884071,
      "learning_rate": 2.376422807744838e-05,
      "loss": 1.3211,
      "step": 1740
    },
    {
      "epoch": 0.445696,
      "grad_norm": 0.3858516208747193,
      "learning_rate": 2.3748264871506046e-05,
      "loss": 1.3819,
      "step": 1741
    },
    {
      "epoch": 0.445952,
      "grad_norm": 0.3950251437618387,
      "learning_rate": 2.373229919038651e-05,
      "loss": 1.3866,
      "step": 1742
    },
    {
      "epoch": 0.446208,
      "grad_norm": 0.3939520350075915,
      "learning_rate": 2.371633104463277e-05,
      "loss": 1.3869,
      "step": 1743
    },
    {
      "epoch": 0.446464,
      "grad_norm": 0.38946413317359496,
      "learning_rate": 2.370036044478942e-05,
      "loss": 1.3628,
      "step": 1744
    },
    {
      "epoch": 0.44672,
      "grad_norm": 0.3956520200769147,
      "learning_rate": 2.36843874014027e-05,
      "loss": 1.3705,
      "step": 1745
    },
    {
      "epoch": 0.446976,
      "grad_norm": 0.40353459534633557,
      "learning_rate": 2.3668411925020455e-05,
      "loss": 1.4091,
      "step": 1746
    },
    {
      "epoch": 0.447232,
      "grad_norm": 0.40007440247880177,
      "learning_rate": 2.3652434026192133e-05,
      "loss": 1.3655,
      "step": 1747
    },
    {
      "epoch": 0.447488,
      "grad_norm": 0.40150312507951885,
      "learning_rate": 2.363645371546878e-05,
      "loss": 1.4011,
      "step": 1748
    },
    {
      "epoch": 0.447744,
      "grad_norm": 0.39333764690121004,
      "learning_rate": 2.3620471003403042e-05,
      "loss": 1.3843,
      "step": 1749
    },
    {
      "epoch": 0.448,
      "grad_norm": 0.3926520464218505,
      "learning_rate": 2.360448590054915e-05,
      "loss": 1.3105,
      "step": 1750
    },
    {
      "epoch": 0.448256,
      "grad_norm": 0.4169859328134903,
      "learning_rate": 2.358849841746291e-05,
      "loss": 1.3721,
      "step": 1751
    },
    {
      "epoch": 0.448512,
      "grad_norm": 0.39648294512606946,
      "learning_rate": 2.35725085647017e-05,
      "loss": 1.4077,
      "step": 1752
    },
    {
      "epoch": 0.448768,
      "grad_norm": 0.3955064408061962,
      "learning_rate": 2.3556516352824463e-05,
      "loss": 1.4041,
      "step": 1753
    },
    {
      "epoch": 0.449024,
      "grad_norm": 0.38764943218850895,
      "learning_rate": 2.3540521792391702e-05,
      "loss": 1.3843,
      "step": 1754
    },
    {
      "epoch": 0.44928,
      "grad_norm": 0.382829642481448,
      "learning_rate": 2.352452489396547e-05,
      "loss": 1.3633,
      "step": 1755
    },
    {
      "epoch": 0.449536,
      "grad_norm": 0.38171491764487375,
      "learning_rate": 2.3508525668109372e-05,
      "loss": 1.3413,
      "step": 1756
    },
    {
      "epoch": 0.449792,
      "grad_norm": 0.407732971436391,
      "learning_rate": 2.3492524125388528e-05,
      "loss": 1.3797,
      "step": 1757
    },
    {
      "epoch": 0.450048,
      "grad_norm": 0.3979240541405608,
      "learning_rate": 2.3476520276369604e-05,
      "loss": 1.395,
      "step": 1758
    },
    {
      "epoch": 0.450304,
      "grad_norm": 0.41229659307524663,
      "learning_rate": 2.3460514131620794e-05,
      "loss": 1.3687,
      "step": 1759
    },
    {
      "epoch": 0.45056,
      "grad_norm": 0.3963502848257384,
      "learning_rate": 2.3444505701711797e-05,
      "loss": 1.3633,
      "step": 1760
    },
    {
      "epoch": 0.450816,
      "grad_norm": 0.4521114295100032,
      "learning_rate": 2.3428494997213815e-05,
      "loss": 1.4106,
      "step": 1761
    },
    {
      "epoch": 0.451072,
      "grad_norm": 0.3973039631210743,
      "learning_rate": 2.3412482028699578e-05,
      "loss": 1.37,
      "step": 1762
    },
    {
      "epoch": 0.451328,
      "grad_norm": 0.4096894893328068,
      "learning_rate": 2.339646680674329e-05,
      "loss": 1.393,
      "step": 1763
    },
    {
      "epoch": 0.451584,
      "grad_norm": 0.38696194505378206,
      "learning_rate": 2.3380449341920636e-05,
      "loss": 1.3641,
      "step": 1764
    },
    {
      "epoch": 0.45184,
      "grad_norm": 0.4168961251278173,
      "learning_rate": 2.33644296448088e-05,
      "loss": 1.408,
      "step": 1765
    },
    {
      "epoch": 0.452096,
      "grad_norm": 0.5591728900837221,
      "learning_rate": 2.334840772598644e-05,
      "loss": 1.4023,
      "step": 1766
    },
    {
      "epoch": 0.452352,
      "grad_norm": 0.3950170126233293,
      "learning_rate": 2.3332383596033667e-05,
      "loss": 1.3824,
      "step": 1767
    },
    {
      "epoch": 0.452608,
      "grad_norm": 0.41596636317532765,
      "learning_rate": 2.331635726553207e-05,
      "loss": 1.3809,
      "step": 1768
    },
    {
      "epoch": 0.452864,
      "grad_norm": 0.3843489452862653,
      "learning_rate": 2.3300328745064676e-05,
      "loss": 1.378,
      "step": 1769
    },
    {
      "epoch": 0.45312,
      "grad_norm": 0.41390966007592817,
      "learning_rate": 2.328429804521595e-05,
      "loss": 1.4255,
      "step": 1770
    },
    {
      "epoch": 0.453376,
      "grad_norm": 0.39683044682829405,
      "learning_rate": 2.326826517657183e-05,
      "loss": 1.3837,
      "step": 1771
    },
    {
      "epoch": 0.453632,
      "grad_norm": 0.3847065092201651,
      "learning_rate": 2.325223014971966e-05,
      "loss": 1.3953,
      "step": 1772
    },
    {
      "epoch": 0.453888,
      "grad_norm": 0.3842903954199142,
      "learning_rate": 2.3236192975248204e-05,
      "loss": 1.3593,
      "step": 1773
    },
    {
      "epoch": 0.454144,
      "grad_norm": 0.40201925252571474,
      "learning_rate": 2.3220153663747675e-05,
      "loss": 1.3864,
      "step": 1774
    },
    {
      "epoch": 0.4544,
      "grad_norm": 0.3777445785109359,
      "learning_rate": 2.3204112225809653e-05,
      "loss": 1.3559,
      "step": 1775
    },
    {
      "epoch": 0.454656,
      "grad_norm": 0.4033279428350374,
      "learning_rate": 2.3188068672027163e-05,
      "loss": 1.3656,
      "step": 1776
    },
    {
      "epoch": 0.454912,
      "grad_norm": 0.42131659789269554,
      "learning_rate": 2.3172023012994608e-05,
      "loss": 1.3833,
      "step": 1777
    },
    {
      "epoch": 0.455168,
      "grad_norm": 0.38228953431753593,
      "learning_rate": 2.3155975259307778e-05,
      "loss": 1.3408,
      "step": 1778
    },
    {
      "epoch": 0.455424,
      "grad_norm": 0.39570407255298823,
      "learning_rate": 2.3139925421563863e-05,
      "loss": 1.3417,
      "step": 1779
    },
    {
      "epoch": 0.45568,
      "grad_norm": 0.38359210009177375,
      "learning_rate": 2.312387351036141e-05,
      "loss": 1.3441,
      "step": 1780
    },
    {
      "epoch": 0.455936,
      "grad_norm": 0.38266597427093224,
      "learning_rate": 2.310781953630034e-05,
      "loss": 1.3752,
      "step": 1781
    },
    {
      "epoch": 0.456192,
      "grad_norm": 0.3955574821178628,
      "learning_rate": 2.3091763509981946e-05,
      "loss": 1.3558,
      "step": 1782
    },
    {
      "epoch": 0.456448,
      "grad_norm": 0.3831525546439454,
      "learning_rate": 2.3075705442008874e-05,
      "loss": 1.3719,
      "step": 1783
    },
    {
      "epoch": 0.456704,
      "grad_norm": 0.3869376162023207,
      "learning_rate": 2.30596453429851e-05,
      "loss": 1.3435,
      "step": 1784
    },
    {
      "epoch": 0.45696,
      "grad_norm": 0.4093846303352642,
      "learning_rate": 2.304358322351597e-05,
      "loss": 1.4077,
      "step": 1785
    },
    {
      "epoch": 0.457216,
      "grad_norm": 0.39645698038591165,
      "learning_rate": 2.3027519094208148e-05,
      "loss": 1.3475,
      "step": 1786
    },
    {
      "epoch": 0.457472,
      "grad_norm": 0.41232689714750037,
      "learning_rate": 2.301145296566961e-05,
      "loss": 1.4272,
      "step": 1787
    },
    {
      "epoch": 0.457728,
      "grad_norm": 0.39099450538922387,
      "learning_rate": 2.2995384848509684e-05,
      "loss": 1.3644,
      "step": 1788
    },
    {
      "epoch": 0.457984,
      "grad_norm": 0.38596895877610593,
      "learning_rate": 2.2979314753339e-05,
      "loss": 1.3725,
      "step": 1789
    },
    {
      "epoch": 0.45824,
      "grad_norm": 0.39215342491795824,
      "learning_rate": 2.296324269076948e-05,
      "loss": 1.3867,
      "step": 1790
    },
    {
      "epoch": 0.458496,
      "grad_norm": 0.4038879757570262,
      "learning_rate": 2.294716867141436e-05,
      "loss": 1.3811,
      "step": 1791
    },
    {
      "epoch": 0.458752,
      "grad_norm": 0.4037059234135371,
      "learning_rate": 2.2931092705888167e-05,
      "loss": 1.4255,
      "step": 1792
    },
    {
      "epoch": 0.459008,
      "grad_norm": 0.38314522450078486,
      "learning_rate": 2.2915014804806693e-05,
      "loss": 1.34,
      "step": 1793
    },
    {
      "epoch": 0.459264,
      "grad_norm": 0.38970609420407243,
      "learning_rate": 2.289893497878705e-05,
      "loss": 1.3865,
      "step": 1794
    },
    {
      "epoch": 0.45952,
      "grad_norm": 0.38437592094893025,
      "learning_rate": 2.2882853238447576e-05,
      "loss": 1.3564,
      "step": 1795
    },
    {
      "epoch": 0.459776,
      "grad_norm": 0.39797191113556035,
      "learning_rate": 2.28667695944079e-05,
      "loss": 1.3872,
      "step": 1796
    },
    {
      "epoch": 0.460032,
      "grad_norm": 0.38772174065389203,
      "learning_rate": 2.285068405728891e-05,
      "loss": 1.3826,
      "step": 1797
    },
    {
      "epoch": 0.460288,
      "grad_norm": 0.3862781126275468,
      "learning_rate": 2.2834596637712724e-05,
      "loss": 1.3999,
      "step": 1798
    },
    {
      "epoch": 0.460544,
      "grad_norm": 0.3932140463601221,
      "learning_rate": 2.281850734630272e-05,
      "loss": 1.3887,
      "step": 1799
    },
    {
      "epoch": 0.4608,
      "grad_norm": 0.39389600096501287,
      "learning_rate": 2.28024161936835e-05,
      "loss": 1.3663,
      "step": 1800
    },
    {
      "epoch": 0.461056,
      "grad_norm": 0.3723341401885224,
      "learning_rate": 2.2786323190480907e-05,
      "loss": 1.3218,
      "step": 1801
    },
    {
      "epoch": 0.461312,
      "grad_norm": 0.38596343457932986,
      "learning_rate": 2.2770228347322005e-05,
      "loss": 1.3694,
      "step": 1802
    },
    {
      "epoch": 0.461568,
      "grad_norm": 0.3882757892178642,
      "learning_rate": 2.2754131674835057e-05,
      "loss": 1.3591,
      "step": 1803
    },
    {
      "epoch": 0.461824,
      "grad_norm": 0.4018501524739478,
      "learning_rate": 2.273803318364956e-05,
      "loss": 1.4035,
      "step": 1804
    },
    {
      "epoch": 0.46208,
      "grad_norm": 0.3821632736044547,
      "learning_rate": 2.272193288439618e-05,
      "loss": 1.3387,
      "step": 1805
    },
    {
      "epoch": 0.462336,
      "grad_norm": 0.38408088795486967,
      "learning_rate": 2.270583078770681e-05,
      "loss": 1.3106,
      "step": 1806
    },
    {
      "epoch": 0.462592,
      "grad_norm": 0.39667015718044396,
      "learning_rate": 2.2689726904214514e-05,
      "loss": 1.3806,
      "step": 1807
    },
    {
      "epoch": 0.462848,
      "grad_norm": 0.39655850443471397,
      "learning_rate": 2.2673621244553526e-05,
      "loss": 1.3552,
      "step": 1808
    },
    {
      "epoch": 0.463104,
      "grad_norm": 0.3919611109646091,
      "learning_rate": 2.2657513819359273e-05,
      "loss": 1.3692,
      "step": 1809
    },
    {
      "epoch": 0.46336,
      "grad_norm": 0.3982351083472095,
      "learning_rate": 2.2641404639268333e-05,
      "loss": 1.397,
      "step": 1810
    },
    {
      "epoch": 0.463616,
      "grad_norm": 0.40006889799958345,
      "learning_rate": 2.262529371491845e-05,
      "loss": 1.3409,
      "step": 1811
    },
    {
      "epoch": 0.463872,
      "grad_norm": 0.39670665142998657,
      "learning_rate": 2.2609181056948514e-05,
      "loss": 1.3821,
      "step": 1812
    },
    {
      "epoch": 0.464128,
      "grad_norm": 0.38493550905875334,
      "learning_rate": 2.2593066675998573e-05,
      "loss": 1.3788,
      "step": 1813
    },
    {
      "epoch": 0.464384,
      "grad_norm": 0.3987303359163086,
      "learning_rate": 2.2576950582709798e-05,
      "loss": 1.3963,
      "step": 1814
    },
    {
      "epoch": 0.46464,
      "grad_norm": 0.39799467532596533,
      "learning_rate": 2.256083278772449e-05,
      "loss": 1.4016,
      "step": 1815
    },
    {
      "epoch": 0.464896,
      "grad_norm": 0.37947102091190804,
      "learning_rate": 2.2544713301686095e-05,
      "loss": 1.3973,
      "step": 1816
    },
    {
      "epoch": 0.465152,
      "grad_norm": 0.4059182442431278,
      "learning_rate": 2.252859213523915e-05,
      "loss": 1.3817,
      "step": 1817
    },
    {
      "epoch": 0.465408,
      "grad_norm": 0.39267172911681414,
      "learning_rate": 2.2512469299029305e-05,
      "loss": 1.3936,
      "step": 1818
    },
    {
      "epoch": 0.465664,
      "grad_norm": 0.39756579908049916,
      "learning_rate": 2.249634480370334e-05,
      "loss": 1.3946,
      "step": 1819
    },
    {
      "epoch": 0.46592,
      "grad_norm": 0.4008096913828974,
      "learning_rate": 2.24802186599091e-05,
      "loss": 1.3553,
      "step": 1820
    },
    {
      "epoch": 0.466176,
      "grad_norm": 0.3934787317972552,
      "learning_rate": 2.246409087829552e-05,
      "loss": 1.3999,
      "step": 1821
    },
    {
      "epoch": 0.466432,
      "grad_norm": 0.3868372313340382,
      "learning_rate": 2.244796146951264e-05,
      "loss": 1.3506,
      "step": 1822
    },
    {
      "epoch": 0.466688,
      "grad_norm": 0.39393581463126714,
      "learning_rate": 2.243183044421156e-05,
      "loss": 1.3772,
      "step": 1823
    },
    {
      "epoch": 0.466944,
      "grad_norm": 0.38846836998563544,
      "learning_rate": 2.2415697813044437e-05,
      "loss": 1.394,
      "step": 1824
    },
    {
      "epoch": 0.4672,
      "grad_norm": 0.39569930690283633,
      "learning_rate": 2.2399563586664506e-05,
      "loss": 1.3555,
      "step": 1825
    },
    {
      "epoch": 0.467456,
      "grad_norm": 0.39772708408470236,
      "learning_rate": 2.2383427775726048e-05,
      "loss": 1.3721,
      "step": 1826
    },
    {
      "epoch": 0.467712,
      "grad_norm": 0.3984931759651297,
      "learning_rate": 2.2367290390884388e-05,
      "loss": 1.3459,
      "step": 1827
    },
    {
      "epoch": 0.467968,
      "grad_norm": 0.4112939598437003,
      "learning_rate": 2.23511514427959e-05,
      "loss": 1.3884,
      "step": 1828
    },
    {
      "epoch": 0.468224,
      "grad_norm": 0.3924848225380971,
      "learning_rate": 2.2335010942117975e-05,
      "loss": 1.3986,
      "step": 1829
    },
    {
      "epoch": 0.46848,
      "grad_norm": 0.39693183356509976,
      "learning_rate": 2.2318868899509044e-05,
      "loss": 1.382,
      "step": 1830
    },
    {
      "epoch": 0.468736,
      "grad_norm": 0.3989722636859817,
      "learning_rate": 2.2302725325628546e-05,
      "loss": 1.4032,
      "step": 1831
    },
    {
      "epoch": 0.468992,
      "grad_norm": 0.3917379726246244,
      "learning_rate": 2.2286580231136944e-05,
      "loss": 1.3506,
      "step": 1832
    },
    {
      "epoch": 0.469248,
      "grad_norm": 0.39012441905950607,
      "learning_rate": 2.227043362669568e-05,
      "loss": 1.3476,
      "step": 1833
    },
    {
      "epoch": 0.469504,
      "grad_norm": 0.39173346519766833,
      "learning_rate": 2.2254285522967222e-05,
      "loss": 1.363,
      "step": 1834
    },
    {
      "epoch": 0.46976,
      "grad_norm": 0.3876010772993033,
      "learning_rate": 2.2238135930615013e-05,
      "loss": 1.3761,
      "step": 1835
    },
    {
      "epoch": 0.470016,
      "grad_norm": 0.3894214133316552,
      "learning_rate": 2.222198486030348e-05,
      "loss": 1.3248,
      "step": 1836
    },
    {
      "epoch": 0.470272,
      "grad_norm": 0.37941164908231445,
      "learning_rate": 2.2205832322698032e-05,
      "loss": 1.3613,
      "step": 1837
    },
    {
      "epoch": 0.470528,
      "grad_norm": 0.38428399869063484,
      "learning_rate": 2.2189678328465037e-05,
      "loss": 1.3583,
      "step": 1838
    },
    {
      "epoch": 0.470784,
      "grad_norm": 0.3884299629309424,
      "learning_rate": 2.2173522888271828e-05,
      "loss": 1.3427,
      "step": 1839
    },
    {
      "epoch": 0.47104,
      "grad_norm": 0.3967330061801342,
      "learning_rate": 2.2157366012786697e-05,
      "loss": 1.4158,
      "step": 1840
    },
    {
      "epoch": 0.471296,
      "grad_norm": 0.39378558378341266,
      "learning_rate": 2.2141207712678892e-05,
      "loss": 1.3519,
      "step": 1841
    },
    {
      "epoch": 0.471552,
      "grad_norm": 0.4060692858362815,
      "learning_rate": 2.2125047998618585e-05,
      "loss": 1.417,
      "step": 1842
    },
    {
      "epoch": 0.471808,
      "grad_norm": 0.3937915115932798,
      "learning_rate": 2.2108886881276876e-05,
      "loss": 1.4381,
      "step": 1843
    },
    {
      "epoch": 0.472064,
      "grad_norm": 0.40559843964990816,
      "learning_rate": 2.2092724371325822e-05,
      "loss": 1.3945,
      "step": 1844
    },
    {
      "epoch": 0.47232,
      "grad_norm": 0.39586516874472877,
      "learning_rate": 2.207656047943837e-05,
      "loss": 1.3791,
      "step": 1845
    },
    {
      "epoch": 0.472576,
      "grad_norm": 0.4038597628746731,
      "learning_rate": 2.20603952162884e-05,
      "loss": 1.4425,
      "step": 1846
    },
    {
      "epoch": 0.472832,
      "grad_norm": 0.3825390250815511,
      "learning_rate": 2.2044228592550692e-05,
      "loss": 1.3741,
      "step": 1847
    },
    {
      "epoch": 0.473088,
      "grad_norm": 0.3879539812390648,
      "learning_rate": 2.2028060618900904e-05,
      "loss": 1.3422,
      "step": 1848
    },
    {
      "epoch": 0.473344,
      "grad_norm": 0.40371587380346063,
      "learning_rate": 2.2011891306015613e-05,
      "loss": 1.3301,
      "step": 1849
    },
    {
      "epoch": 0.4736,
      "grad_norm": 0.39304235831918916,
      "learning_rate": 2.1995720664572274e-05,
      "loss": 1.4056,
      "step": 1850
    },
    {
      "epoch": 0.473856,
      "grad_norm": 0.3815801828760049,
      "learning_rate": 2.1979548705249207e-05,
      "loss": 1.337,
      "step": 1851
    },
    {
      "epoch": 0.474112,
      "grad_norm": 0.40892262823207376,
      "learning_rate": 2.1963375438725616e-05,
      "loss": 1.4087,
      "step": 1852
    },
    {
      "epoch": 0.474368,
      "grad_norm": 0.3759524716635272,
      "learning_rate": 2.1947200875681562e-05,
      "loss": 1.3276,
      "step": 1853
    },
    {
      "epoch": 0.474624,
      "grad_norm": 0.38996220056608855,
      "learning_rate": 2.193102502679796e-05,
      "loss": 1.3708,
      "step": 1854
    },
    {
      "epoch": 0.47488,
      "grad_norm": 0.38408014772660437,
      "learning_rate": 2.1914847902756576e-05,
      "loss": 1.3561,
      "step": 1855
    },
    {
      "epoch": 0.475136,
      "grad_norm": 0.490147038901984,
      "learning_rate": 2.1898669514240027e-05,
      "loss": 1.4004,
      "step": 1856
    },
    {
      "epoch": 0.475392,
      "grad_norm": 0.405561535558326,
      "learning_rate": 2.188248987193175e-05,
      "loss": 1.348,
      "step": 1857
    },
    {
      "epoch": 0.475648,
      "grad_norm": 0.3919274110593133,
      "learning_rate": 2.1866308986516024e-05,
      "loss": 1.4026,
      "step": 1858
    },
    {
      "epoch": 0.475904,
      "grad_norm": 0.3815444236706001,
      "learning_rate": 2.185012686867794e-05,
      "loss": 1.3524,
      "step": 1859
    },
    {
      "epoch": 0.47616,
      "grad_norm": 0.3751516282870969,
      "learning_rate": 2.183394352910339e-05,
      "loss": 1.3222,
      "step": 1860
    },
    {
      "epoch": 0.476416,
      "grad_norm": 0.43472332841938144,
      "learning_rate": 2.181775897847911e-05,
      "loss": 1.3663,
      "step": 1861
    },
    {
      "epoch": 0.476672,
      "grad_norm": 0.3949922058601869,
      "learning_rate": 2.18015732274926e-05,
      "loss": 1.3772,
      "step": 1862
    },
    {
      "epoch": 0.476928,
      "grad_norm": 0.39306434480459695,
      "learning_rate": 2.1785386286832168e-05,
      "loss": 1.3503,
      "step": 1863
    },
    {
      "epoch": 0.477184,
      "grad_norm": 0.38099835002121996,
      "learning_rate": 2.176919816718691e-05,
      "loss": 1.3254,
      "step": 1864
    },
    {
      "epoch": 0.47744,
      "grad_norm": 0.39097982013339694,
      "learning_rate": 2.175300887924669e-05,
      "loss": 1.374,
      "step": 1865
    },
    {
      "epoch": 0.477696,
      "grad_norm": 0.4237109011944741,
      "learning_rate": 2.1736818433702154e-05,
      "loss": 1.3544,
      "step": 1866
    },
    {
      "epoch": 0.477952,
      "grad_norm": 0.3866707953519438,
      "learning_rate": 2.17206268412447e-05,
      "loss": 1.3532,
      "step": 1867
    },
    {
      "epoch": 0.478208,
      "grad_norm": 0.3991120349847027,
      "learning_rate": 2.170443411256651e-05,
      "loss": 1.3726,
      "step": 1868
    },
    {
      "epoch": 0.478464,
      "grad_norm": 0.394147724593643,
      "learning_rate": 2.168824025836047e-05,
      "loss": 1.341,
      "step": 1869
    },
    {
      "epoch": 0.47872,
      "grad_norm": 0.38688708760088025,
      "learning_rate": 2.1672045289320266e-05,
      "loss": 1.367,
      "step": 1870
    },
    {
      "epoch": 0.478976,
      "grad_norm": 0.3883878610835278,
      "learning_rate": 2.1655849216140275e-05,
      "loss": 1.3279,
      "step": 1871
    },
    {
      "epoch": 0.479232,
      "grad_norm": 0.38419541851279776,
      "learning_rate": 2.1639652049515612e-05,
      "loss": 1.3552,
      "step": 1872
    },
    {
      "epoch": 0.479488,
      "grad_norm": 0.38071993014633887,
      "learning_rate": 2.1623453800142137e-05,
      "loss": 1.3455,
      "step": 1873
    },
    {
      "epoch": 0.479744,
      "grad_norm": 0.3883589661871264,
      "learning_rate": 2.1607254478716408e-05,
      "loss": 1.3495,
      "step": 1874
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38360991489015156,
      "learning_rate": 2.1591054095935685e-05,
      "loss": 1.3665,
      "step": 1875
    },
    {
      "epoch": 0.480256,
      "grad_norm": 0.38566315185836325,
      "learning_rate": 2.1574852662497934e-05,
      "loss": 1.3692,
      "step": 1876
    },
    {
      "epoch": 0.480512,
      "grad_norm": 0.39595454956347353,
      "learning_rate": 2.1558650189101827e-05,
      "loss": 1.3769,
      "step": 1877
    },
    {
      "epoch": 0.480768,
      "grad_norm": 0.3771882913414376,
      "learning_rate": 2.1542446686446703e-05,
      "loss": 1.3559,
      "step": 1878
    },
    {
      "epoch": 0.481024,
      "grad_norm": 0.3848277571542053,
      "learning_rate": 2.1526242165232593e-05,
      "loss": 1.3625,
      "step": 1879
    },
    {
      "epoch": 0.48128,
      "grad_norm": 0.3875537295192209,
      "learning_rate": 2.1510036636160203e-05,
      "loss": 1.3586,
      "step": 1880
    },
    {
      "epoch": 0.481536,
      "grad_norm": 0.3840800721837618,
      "learning_rate": 2.14938301099309e-05,
      "loss": 1.3681,
      "step": 1881
    },
    {
      "epoch": 0.481792,
      "grad_norm": 0.3829801938867898,
      "learning_rate": 2.1477622597246698e-05,
      "loss": 1.3809,
      "step": 1882
    },
    {
      "epoch": 0.482048,
      "grad_norm": 0.3814671971062877,
      "learning_rate": 2.1461414108810275e-05,
      "loss": 1.3692,
      "step": 1883
    },
    {
      "epoch": 0.482304,
      "grad_norm": 0.3904328847746494,
      "learning_rate": 2.1445204655324966e-05,
      "loss": 1.3847,
      "step": 1884
    },
    {
      "epoch": 0.48256,
      "grad_norm": 0.38612942551642143,
      "learning_rate": 2.1428994247494717e-05,
      "loss": 1.3132,
      "step": 1885
    },
    {
      "epoch": 0.482816,
      "grad_norm": 0.3967469284918172,
      "learning_rate": 2.1412782896024123e-05,
      "loss": 1.4053,
      "step": 1886
    },
    {
      "epoch": 0.483072,
      "grad_norm": 0.3832059171805036,
      "learning_rate": 2.139657061161839e-05,
      "loss": 1.412,
      "step": 1887
    },
    {
      "epoch": 0.483328,
      "grad_norm": 0.3838962416148976,
      "learning_rate": 2.1380357404983337e-05,
      "loss": 1.3454,
      "step": 1888
    },
    {
      "epoch": 0.483584,
      "grad_norm": 0.3827608827431334,
      "learning_rate": 2.1364143286825416e-05,
      "loss": 1.3567,
      "step": 1889
    },
    {
      "epoch": 0.48384,
      "grad_norm": 0.37557705371612216,
      "learning_rate": 2.134792826785166e-05,
      "loss": 1.3464,
      "step": 1890
    },
    {
      "epoch": 0.484096,
      "grad_norm": 0.39159301675637603,
      "learning_rate": 2.13317123587697e-05,
      "loss": 1.3741,
      "step": 1891
    },
    {
      "epoch": 0.484352,
      "grad_norm": 0.46588329434799575,
      "learning_rate": 2.1315495570287763e-05,
      "loss": 1.3658,
      "step": 1892
    },
    {
      "epoch": 0.484608,
      "grad_norm": 0.3835301814445272,
      "learning_rate": 2.1299277913114644e-05,
      "loss": 1.3823,
      "step": 1893
    },
    {
      "epoch": 0.484864,
      "grad_norm": 0.3804298421496852,
      "learning_rate": 2.1283059397959715e-05,
      "loss": 1.346,
      "step": 1894
    },
    {
      "epoch": 0.48512,
      "grad_norm": 0.39296518406531694,
      "learning_rate": 2.1266840035532924e-05,
      "loss": 1.3401,
      "step": 1895
    },
    {
      "epoch": 0.485376,
      "grad_norm": 0.3863026882979723,
      "learning_rate": 2.1250619836544777e-05,
      "loss": 1.3979,
      "step": 1896
    },
    {
      "epoch": 0.485632,
      "grad_norm": 0.39231799033018405,
      "learning_rate": 2.1234398811706316e-05,
      "loss": 1.3448,
      "step": 1897
    },
    {
      "epoch": 0.485888,
      "grad_norm": 0.3927200286700311,
      "learning_rate": 2.121817697172915e-05,
      "loss": 1.3527,
      "step": 1898
    },
    {
      "epoch": 0.486144,
      "grad_norm": 0.37090478976427615,
      "learning_rate": 2.120195432732542e-05,
      "loss": 1.3447,
      "step": 1899
    },
    {
      "epoch": 0.4864,
      "grad_norm": 0.3860129441087777,
      "learning_rate": 2.1185730889207776e-05,
      "loss": 1.3671,
      "step": 1900
    },
    {
      "epoch": 0.486656,
      "grad_norm": 0.3859971159967315,
      "learning_rate": 2.116950666808942e-05,
      "loss": 1.3707,
      "step": 1901
    },
    {
      "epoch": 0.486912,
      "grad_norm": 0.380245783494106,
      "learning_rate": 2.115328167468407e-05,
      "loss": 1.3536,
      "step": 1902
    },
    {
      "epoch": 0.487168,
      "grad_norm": 0.3705195318013785,
      "learning_rate": 2.1137055919705943e-05,
      "loss": 1.3556,
      "step": 1903
    },
    {
      "epoch": 0.487424,
      "grad_norm": 0.3848309242438573,
      "learning_rate": 2.1120829413869753e-05,
      "loss": 1.3176,
      "step": 1904
    },
    {
      "epoch": 0.48768,
      "grad_norm": 0.4671167042750147,
      "learning_rate": 2.110460216789073e-05,
      "loss": 1.3928,
      "step": 1905
    },
    {
      "epoch": 0.487936,
      "grad_norm": 0.4164478081380505,
      "learning_rate": 2.1088374192484563e-05,
      "loss": 1.3438,
      "step": 1906
    },
    {
      "epoch": 0.488192,
      "grad_norm": 0.43360540042775003,
      "learning_rate": 2.107214549836746e-05,
      "loss": 1.3727,
      "step": 1907
    },
    {
      "epoch": 0.488448,
      "grad_norm": 0.4014460329954866,
      "learning_rate": 2.105591609625608e-05,
      "loss": 1.3998,
      "step": 1908
    },
    {
      "epoch": 0.488704,
      "grad_norm": 0.39470932194340597,
      "learning_rate": 2.1039685996867553e-05,
      "loss": 1.3545,
      "step": 1909
    },
    {
      "epoch": 0.48896,
      "grad_norm": 0.3779309817406106,
      "learning_rate": 2.1023455210919464e-05,
      "loss": 1.3421,
      "step": 1910
    },
    {
      "epoch": 0.489216,
      "grad_norm": 0.42963191579012516,
      "learning_rate": 2.1007223749129868e-05,
      "loss": 1.39,
      "step": 1911
    },
    {
      "epoch": 0.489472,
      "grad_norm": 0.39017051602695113,
      "learning_rate": 2.0990991622217245e-05,
      "loss": 1.3524,
      "step": 1912
    },
    {
      "epoch": 0.489728,
      "grad_norm": 0.397072377704159,
      "learning_rate": 2.097475884090053e-05,
      "loss": 1.3887,
      "step": 1913
    },
    {
      "epoch": 0.489984,
      "grad_norm": 0.39216348539631113,
      "learning_rate": 2.0958525415899097e-05,
      "loss": 1.3522,
      "step": 1914
    },
    {
      "epoch": 0.49024,
      "grad_norm": 0.3912211902857766,
      "learning_rate": 2.094229135793272e-05,
      "loss": 1.3368,
      "step": 1915
    },
    {
      "epoch": 0.490496,
      "grad_norm": 0.3763842134676675,
      "learning_rate": 2.092605667772161e-05,
      "loss": 1.3398,
      "step": 1916
    },
    {
      "epoch": 0.490752,
      "grad_norm": 0.40522392708546073,
      "learning_rate": 2.090982138598638e-05,
      "loss": 1.4091,
      "step": 1917
    },
    {
      "epoch": 0.491008,
      "grad_norm": 0.3877709008886343,
      "learning_rate": 2.089358549344805e-05,
      "loss": 1.3868,
      "step": 1918
    },
    {
      "epoch": 0.491264,
      "grad_norm": 0.3823586201376526,
      "learning_rate": 2.0877349010828044e-05,
      "loss": 1.3416,
      "step": 1919
    },
    {
      "epoch": 0.49152,
      "grad_norm": 0.3937433305394222,
      "learning_rate": 2.0861111948848164e-05,
      "loss": 1.4201,
      "step": 1920
    },
    {
      "epoch": 0.491776,
      "grad_norm": 0.38459863882126133,
      "learning_rate": 2.08448743182306e-05,
      "loss": 1.3307,
      "step": 1921
    },
    {
      "epoch": 0.492032,
      "grad_norm": 0.38562441496805383,
      "learning_rate": 2.0828636129697913e-05,
      "loss": 1.3439,
      "step": 1922
    },
    {
      "epoch": 0.492288,
      "grad_norm": 0.4057156401896175,
      "learning_rate": 2.081239739397304e-05,
      "loss": 1.4366,
      "step": 1923
    },
    {
      "epoch": 0.492544,
      "grad_norm": 0.3756407061799462,
      "learning_rate": 2.0796158121779276e-05,
      "loss": 1.3363,
      "step": 1924
    },
    {
      "epoch": 0.4928,
      "grad_norm": 0.3913577138131945,
      "learning_rate": 2.0779918323840272e-05,
      "loss": 1.3487,
      "step": 1925
    },
    {
      "epoch": 0.493056,
      "grad_norm": 0.39703055525485664,
      "learning_rate": 2.0763678010880022e-05,
      "loss": 1.3578,
      "step": 1926
    },
    {
      "epoch": 0.493312,
      "grad_norm": 0.3855821699483612,
      "learning_rate": 2.0747437193622853e-05,
      "loss": 1.3834,
      "step": 1927
    },
    {
      "epoch": 0.493568,
      "grad_norm": 0.3976740420628351,
      "learning_rate": 2.0731195882793444e-05,
      "loss": 1.392,
      "step": 1928
    },
    {
      "epoch": 0.493824,
      "grad_norm": 0.38616943432626455,
      "learning_rate": 2.071495408911679e-05,
      "loss": 1.3432,
      "step": 1929
    },
    {
      "epoch": 0.49408,
      "grad_norm": 0.39878652095482303,
      "learning_rate": 2.0698711823318206e-05,
      "loss": 1.4333,
      "step": 1930
    },
    {
      "epoch": 0.494336,
      "grad_norm": 0.37809552547894043,
      "learning_rate": 2.068246909612331e-05,
      "loss": 1.3208,
      "step": 1931
    },
    {
      "epoch": 0.494592,
      "grad_norm": 0.40437408145789877,
      "learning_rate": 2.0666225918258044e-05,
      "loss": 1.3466,
      "step": 1932
    },
    {
      "epoch": 0.494848,
      "grad_norm": 0.3825355317614924,
      "learning_rate": 2.0649982300448622e-05,
      "loss": 1.3353,
      "step": 1933
    },
    {
      "epoch": 0.495104,
      "grad_norm": 0.3814665792747797,
      "learning_rate": 2.0633738253421568e-05,
      "loss": 1.4063,
      "step": 1934
    },
    {
      "epoch": 0.49536,
      "grad_norm": 0.4169007755233025,
      "learning_rate": 2.0617493787903693e-05,
      "loss": 1.3536,
      "step": 1935
    },
    {
      "epoch": 0.495616,
      "grad_norm": 0.38418396725842263,
      "learning_rate": 2.060124891462206e-05,
      "loss": 1.3588,
      "step": 1936
    },
    {
      "epoch": 0.495872,
      "grad_norm": 0.37767525766944776,
      "learning_rate": 2.058500364430403e-05,
      "loss": 1.3787,
      "step": 1937
    },
    {
      "epoch": 0.496128,
      "grad_norm": 0.3934807105107317,
      "learning_rate": 2.056875798767721e-05,
      "loss": 1.3801,
      "step": 1938
    },
    {
      "epoch": 0.496384,
      "grad_norm": 0.3951314838809165,
      "learning_rate": 2.055251195546945e-05,
      "loss": 1.3386,
      "step": 1939
    },
    {
      "epoch": 0.49664,
      "grad_norm": 0.38609768348745493,
      "learning_rate": 2.0536265558408882e-05,
      "loss": 1.3933,
      "step": 1940
    },
    {
      "epoch": 0.496896,
      "grad_norm": 0.392244717604293,
      "learning_rate": 2.052001880722385e-05,
      "loss": 1.4091,
      "step": 1941
    },
    {
      "epoch": 0.497152,
      "grad_norm": 0.37865861275173585,
      "learning_rate": 2.0503771712642943e-05,
      "loss": 1.3607,
      "step": 1942
    },
    {
      "epoch": 0.497408,
      "grad_norm": 0.3963807460907485,
      "learning_rate": 2.048752428539498e-05,
      "loss": 1.3971,
      "step": 1943
    },
    {
      "epoch": 0.497664,
      "grad_norm": 0.3750881740326447,
      "learning_rate": 2.0471276536208985e-05,
      "loss": 1.3329,
      "step": 1944
    },
    {
      "epoch": 0.49792,
      "grad_norm": 0.38133354834594463,
      "learning_rate": 2.045502847581422e-05,
      "loss": 1.3677,
      "step": 1945
    },
    {
      "epoch": 0.498176,
      "grad_norm": 0.3781436719748803,
      "learning_rate": 2.043878011494012e-05,
      "loss": 1.3767,
      "step": 1946
    },
    {
      "epoch": 0.498432,
      "grad_norm": 0.3823177173813485,
      "learning_rate": 2.0422531464316348e-05,
      "loss": 1.371,
      "step": 1947
    },
    {
      "epoch": 0.498688,
      "grad_norm": 0.3853247153477152,
      "learning_rate": 2.040628253467274e-05,
      "loss": 1.3701,
      "step": 1948
    },
    {
      "epoch": 0.498944,
      "grad_norm": 0.39146602670334707,
      "learning_rate": 2.0390033336739324e-05,
      "loss": 1.3573,
      "step": 1949
    },
    {
      "epoch": 0.4992,
      "grad_norm": 0.3809707236503774,
      "learning_rate": 2.0373783881246304e-05,
      "loss": 1.3735,
      "step": 1950
    },
    {
      "epoch": 0.499456,
      "grad_norm": 0.38373386962231826,
      "learning_rate": 2.035753417892405e-05,
      "loss": 1.3625,
      "step": 1951
    },
    {
      "epoch": 0.499712,
      "grad_norm": 0.43752232820186826,
      "learning_rate": 2.03412842405031e-05,
      "loss": 1.3675,
      "step": 1952
    },
    {
      "epoch": 0.499968,
      "grad_norm": 0.43061717408471223,
      "learning_rate": 2.0325034076714154e-05,
      "loss": 1.3578,
      "step": 1953
    },
    {
      "epoch": 0.500224,
      "grad_norm": 0.38957311795657534,
      "learning_rate": 2.030878369828804e-05,
      "loss": 1.3417,
      "step": 1954
    },
    {
      "epoch": 0.50048,
      "grad_norm": 0.38734693860595065,
      "learning_rate": 2.0292533115955747e-05,
      "loss": 1.3483,
      "step": 1955
    },
    {
      "epoch": 0.500736,
      "grad_norm": 0.4001206589592376,
      "learning_rate": 2.0276282340448388e-05,
      "loss": 1.3746,
      "step": 1956
    },
    {
      "epoch": 0.500992,
      "grad_norm": 0.3932650495410377,
      "learning_rate": 2.0260031382497223e-05,
      "loss": 1.3732,
      "step": 1957
    },
    {
      "epoch": 0.501248,
      "grad_norm": 0.40347394899495653,
      "learning_rate": 2.0243780252833595e-05,
      "loss": 1.3855,
      "step": 1958
    },
    {
      "epoch": 0.501504,
      "grad_norm": 0.39723513653260434,
      "learning_rate": 2.0227528962189007e-05,
      "loss": 1.3594,
      "step": 1959
    },
    {
      "epoch": 0.50176,
      "grad_norm": 0.3755346349346534,
      "learning_rate": 2.0211277521295028e-05,
      "loss": 1.3553,
      "step": 1960
    },
    {
      "epoch": 0.502016,
      "grad_norm": 0.3845739316487716,
      "learning_rate": 2.0195025940883345e-05,
      "loss": 1.3613,
      "step": 1961
    },
    {
      "epoch": 0.502272,
      "grad_norm": 0.3997167923209803,
      "learning_rate": 2.0178774231685737e-05,
      "loss": 1.4203,
      "step": 1962
    },
    {
      "epoch": 0.502528,
      "grad_norm": 0.39110812057016336,
      "learning_rate": 2.0162522404434064e-05,
      "loss": 1.3929,
      "step": 1963
    },
    {
      "epoch": 0.502784,
      "grad_norm": 0.3826156000320015,
      "learning_rate": 2.0146270469860267e-05,
      "loss": 1.3769,
      "step": 1964
    },
    {
      "epoch": 0.50304,
      "grad_norm": 0.39525167791832627,
      "learning_rate": 2.0130018438696356e-05,
      "loss": 1.3841,
      "step": 1965
    },
    {
      "epoch": 0.503296,
      "grad_norm": 0.46290825246172496,
      "learning_rate": 2.011376632167441e-05,
      "loss": 1.3184,
      "step": 1966
    },
    {
      "epoch": 0.503552,
      "grad_norm": 0.3902020720325329,
      "learning_rate": 2.009751412952655e-05,
      "loss": 1.3799,
      "step": 1967
    },
    {
      "epoch": 0.503808,
      "grad_norm": 0.37705753303203643,
      "learning_rate": 2.008126187298496e-05,
      "loss": 1.3499,
      "step": 1968
    },
    {
      "epoch": 0.504064,
      "grad_norm": 0.37233911352395976,
      "learning_rate": 2.006500956278187e-05,
      "loss": 1.2956,
      "step": 1969
    },
    {
      "epoch": 0.50432,
      "grad_norm": 0.39048293745012347,
      "learning_rate": 2.004875720964953e-05,
      "loss": 1.3707,
      "step": 1970
    },
    {
      "epoch": 0.504576,
      "grad_norm": 0.3816016600125312,
      "learning_rate": 2.0032504824320232e-05,
      "loss": 1.3477,
      "step": 1971
    },
    {
      "epoch": 0.504832,
      "grad_norm": 0.39998943464436226,
      "learning_rate": 2.0016252417526286e-05,
      "loss": 1.4148,
      "step": 1972
    },
    {
      "epoch": 0.505088,
      "grad_norm": 0.36275553689547413,
      "learning_rate": 2e-05,
      "loss": 1.3116,
      "step": 1973
    },
    {
      "epoch": 0.505344,
      "grad_norm": 0.3871694522161321,
      "learning_rate": 1.998374758247372e-05,
      "loss": 1.3867,
      "step": 1974
    },
    {
      "epoch": 0.5056,
      "grad_norm": 0.39756330652623084,
      "learning_rate": 1.9967495175679767e-05,
      "loss": 1.3371,
      "step": 1975
    },
    {
      "epoch": 0.505856,
      "grad_norm": 0.3811377904000282,
      "learning_rate": 1.9951242790350473e-05,
      "loss": 1.3702,
      "step": 1976
    },
    {
      "epoch": 0.506112,
      "grad_norm": 0.39480916686406853,
      "learning_rate": 1.993499043721814e-05,
      "loss": 1.4002,
      "step": 1977
    },
    {
      "epoch": 0.506368,
      "grad_norm": 0.3949932713288373,
      "learning_rate": 1.9918738127015044e-05,
      "loss": 1.4003,
      "step": 1978
    },
    {
      "epoch": 0.506624,
      "grad_norm": 0.40511077698438464,
      "learning_rate": 1.9902485870473456e-05,
      "loss": 1.3934,
      "step": 1979
    },
    {
      "epoch": 0.50688,
      "grad_norm": 0.37228094962321134,
      "learning_rate": 1.9886233678325593e-05,
      "loss": 1.323,
      "step": 1980
    },
    {
      "epoch": 0.507136,
      "grad_norm": 0.39072542349002837,
      "learning_rate": 1.9869981561303648e-05,
      "loss": 1.3769,
      "step": 1981
    },
    {
      "epoch": 0.507392,
      "grad_norm": 0.3874456384980606,
      "learning_rate": 1.9853729530139733e-05,
      "loss": 1.3545,
      "step": 1982
    },
    {
      "epoch": 0.507648,
      "grad_norm": 0.38616909369123936,
      "learning_rate": 1.9837477595565943e-05,
      "loss": 1.3723,
      "step": 1983
    },
    {
      "epoch": 0.507904,
      "grad_norm": 0.380014806887977,
      "learning_rate": 1.9821225768314273e-05,
      "loss": 1.3552,
      "step": 1984
    },
    {
      "epoch": 0.50816,
      "grad_norm": 0.3918810517682189,
      "learning_rate": 1.9804974059116662e-05,
      "loss": 1.4037,
      "step": 1985
    },
    {
      "epoch": 0.508416,
      "grad_norm": 0.38827740239059316,
      "learning_rate": 1.9788722478704976e-05,
      "loss": 1.3649,
      "step": 1986
    },
    {
      "epoch": 0.508672,
      "grad_norm": 0.6764152648389595,
      "learning_rate": 1.9772471037811e-05,
      "loss": 1.3329,
      "step": 1987
    },
    {
      "epoch": 0.508928,
      "grad_norm": 0.4038041178815253,
      "learning_rate": 1.9756219747166402e-05,
      "loss": 1.3828,
      "step": 1988
    },
    {
      "epoch": 0.509184,
      "grad_norm": 0.3939024982146074,
      "learning_rate": 1.973996861750279e-05,
      "loss": 1.4168,
      "step": 1989
    },
    {
      "epoch": 0.50944,
      "grad_norm": 0.3799535259187517,
      "learning_rate": 1.9723717659551615e-05,
      "loss": 1.3796,
      "step": 1990
    },
    {
      "epoch": 0.509696,
      "grad_norm": 0.38191301074008077,
      "learning_rate": 1.970746688404426e-05,
      "loss": 1.3685,
      "step": 1991
    },
    {
      "epoch": 0.509952,
      "grad_norm": 0.3786011550570985,
      "learning_rate": 1.9691216301711966e-05,
      "loss": 1.3765,
      "step": 1992
    },
    {
      "epoch": 0.510208,
      "grad_norm": 0.3825478153169017,
      "learning_rate": 1.9674965923285853e-05,
      "loss": 1.3655,
      "step": 1993
    },
    {
      "epoch": 0.510464,
      "grad_norm": 0.38829089445025744,
      "learning_rate": 1.96587157594969e-05,
      "loss": 1.3714,
      "step": 1994
    },
    {
      "epoch": 0.51072,
      "grad_norm": 0.3887355422106671,
      "learning_rate": 1.964246582107596e-05,
      "loss": 1.3364,
      "step": 1995
    },
    {
      "epoch": 0.510976,
      "grad_norm": 0.3739339344603865,
      "learning_rate": 1.9626216118753706e-05,
      "loss": 1.3001,
      "step": 1996
    },
    {
      "epoch": 0.511232,
      "grad_norm": 0.38535629182107584,
      "learning_rate": 1.960996666326068e-05,
      "loss": 1.3261,
      "step": 1997
    },
    {
      "epoch": 0.511488,
      "grad_norm": 0.3918910022989723,
      "learning_rate": 1.959371746532726e-05,
      "loss": 1.4068,
      "step": 1998
    },
    {
      "epoch": 0.511744,
      "grad_norm": 0.3941888414970392,
      "learning_rate": 1.9577468535683656e-05,
      "loss": 1.3977,
      "step": 1999
    },
    {
      "epoch": 0.512,
      "grad_norm": 0.3776578134866879,
      "learning_rate": 1.9561219885059882e-05,
      "loss": 1.3409,
      "step": 2000
    },
    {
      "epoch": 0.512256,
      "grad_norm": 0.3805173482554356,
      "learning_rate": 1.9544971524185788e-05,
      "loss": 1.3669,
      "step": 2001
    },
    {
      "epoch": 0.512512,
      "grad_norm": 0.4026985262759454,
      "learning_rate": 1.9528723463791018e-05,
      "loss": 1.3806,
      "step": 2002
    },
    {
      "epoch": 0.512768,
      "grad_norm": 0.3801088690882476,
      "learning_rate": 1.9512475714605026e-05,
      "loss": 1.3475,
      "step": 2003
    },
    {
      "epoch": 0.513024,
      "grad_norm": 0.3791935978374013,
      "learning_rate": 1.949622828735706e-05,
      "loss": 1.3496,
      "step": 2004
    },
    {
      "epoch": 0.51328,
      "grad_norm": 0.3875364802742619,
      "learning_rate": 1.9479981192776155e-05,
      "loss": 1.3804,
      "step": 2005
    },
    {
      "epoch": 0.513536,
      "grad_norm": 0.36535804881634,
      "learning_rate": 1.946373444159112e-05,
      "loss": 1.3265,
      "step": 2006
    },
    {
      "epoch": 0.513792,
      "grad_norm": 0.379400328996959,
      "learning_rate": 1.9447488044530562e-05,
      "loss": 1.3409,
      "step": 2007
    },
    {
      "epoch": 0.514048,
      "grad_norm": 0.3966491889236639,
      "learning_rate": 1.9431242012322804e-05,
      "loss": 1.356,
      "step": 2008
    },
    {
      "epoch": 0.514304,
      "grad_norm": 0.40753606751500926,
      "learning_rate": 1.9414996355695977e-05,
      "loss": 1.416,
      "step": 2009
    },
    {
      "epoch": 0.51456,
      "grad_norm": 0.3729933466235488,
      "learning_rate": 1.9398751085377947e-05,
      "loss": 1.3442,
      "step": 2010
    },
    {
      "epoch": 0.514816,
      "grad_norm": 0.46908178989442895,
      "learning_rate": 1.9382506212096314e-05,
      "loss": 1.3587,
      "step": 2011
    },
    {
      "epoch": 0.515072,
      "grad_norm": 0.3822963648272708,
      "learning_rate": 1.9366261746578436e-05,
      "loss": 1.3568,
      "step": 2012
    },
    {
      "epoch": 0.515328,
      "grad_norm": 0.4203938774143448,
      "learning_rate": 1.9350017699551388e-05,
      "loss": 1.3751,
      "step": 2013
    },
    {
      "epoch": 0.515584,
      "grad_norm": 0.3835878632140864,
      "learning_rate": 1.9333774081741966e-05,
      "loss": 1.3767,
      "step": 2014
    },
    {
      "epoch": 0.51584,
      "grad_norm": 0.3854311270816547,
      "learning_rate": 1.9317530903876693e-05,
      "loss": 1.3392,
      "step": 2015
    },
    {
      "epoch": 0.516096,
      "grad_norm": 0.37716617835807054,
      "learning_rate": 1.9301288176681804e-05,
      "loss": 1.3351,
      "step": 2016
    },
    {
      "epoch": 0.516352,
      "grad_norm": 0.37510208283474183,
      "learning_rate": 1.9285045910883215e-05,
      "loss": 1.36,
      "step": 2017
    },
    {
      "epoch": 0.516608,
      "grad_norm": 0.3889786974781347,
      "learning_rate": 1.926880411720656e-05,
      "loss": 1.3456,
      "step": 2018
    },
    {
      "epoch": 0.516864,
      "grad_norm": 0.38281895229950763,
      "learning_rate": 1.9252562806377157e-05,
      "loss": 1.3694,
      "step": 2019
    },
    {
      "epoch": 0.51712,
      "grad_norm": 0.3764444722565575,
      "learning_rate": 1.923632198911999e-05,
      "loss": 1.3526,
      "step": 2020
    },
    {
      "epoch": 0.517376,
      "grad_norm": 0.37803337524374225,
      "learning_rate": 1.9220081676159734e-05,
      "loss": 1.3277,
      "step": 2021
    },
    {
      "epoch": 0.517632,
      "grad_norm": 0.37486847839364956,
      "learning_rate": 1.920384187822073e-05,
      "loss": 1.3522,
      "step": 2022
    },
    {
      "epoch": 0.517888,
      "grad_norm": 0.3793818125835509,
      "learning_rate": 1.9187602606026962e-05,
      "loss": 1.35,
      "step": 2023
    },
    {
      "epoch": 0.518144,
      "grad_norm": 0.37553117272021497,
      "learning_rate": 1.917136387030209e-05,
      "loss": 1.305,
      "step": 2024
    },
    {
      "epoch": 0.5184,
      "grad_norm": 0.38402371778975825,
      "learning_rate": 1.915512568176941e-05,
      "loss": 1.3668,
      "step": 2025
    },
    {
      "epoch": 0.518656,
      "grad_norm": 0.3724125018094645,
      "learning_rate": 1.9138888051151843e-05,
      "loss": 1.3753,
      "step": 2026
    },
    {
      "epoch": 0.518912,
      "grad_norm": 0.3916883428772269,
      "learning_rate": 1.9122650989171963e-05,
      "loss": 1.3586,
      "step": 2027
    },
    {
      "epoch": 0.519168,
      "grad_norm": 0.38366679902568507,
      "learning_rate": 1.9106414506551953e-05,
      "loss": 1.3504,
      "step": 2028
    },
    {
      "epoch": 0.519424,
      "grad_norm": 0.3737928390889648,
      "learning_rate": 1.9090178614013623e-05,
      "loss": 1.3402,
      "step": 2029
    },
    {
      "epoch": 0.51968,
      "grad_norm": 0.37944351880829946,
      "learning_rate": 1.9073943322278392e-05,
      "loss": 1.3564,
      "step": 2030
    },
    {
      "epoch": 0.519936,
      "grad_norm": 0.38686519820469406,
      "learning_rate": 1.905770864206729e-05,
      "loss": 1.3531,
      "step": 2031
    },
    {
      "epoch": 0.520192,
      "grad_norm": 0.38161934980693907,
      "learning_rate": 1.9041474584100906e-05,
      "loss": 1.3843,
      "step": 2032
    },
    {
      "epoch": 0.520448,
      "grad_norm": 0.38096987600479093,
      "learning_rate": 1.9025241159099473e-05,
      "loss": 1.3479,
      "step": 2033
    },
    {
      "epoch": 0.520704,
      "grad_norm": 0.3799753033323743,
      "learning_rate": 1.900900837778276e-05,
      "loss": 1.3615,
      "step": 2034
    },
    {
      "epoch": 0.52096,
      "grad_norm": 0.3824694160095296,
      "learning_rate": 1.899277625087014e-05,
      "loss": 1.3461,
      "step": 2035
    },
    {
      "epoch": 0.521216,
      "grad_norm": 0.370765586732249,
      "learning_rate": 1.897654478908054e-05,
      "loss": 1.3601,
      "step": 2036
    },
    {
      "epoch": 0.521472,
      "grad_norm": 0.37434634624810453,
      "learning_rate": 1.8960314003132454e-05,
      "loss": 1.3343,
      "step": 2037
    },
    {
      "epoch": 0.521728,
      "grad_norm": 0.4090736874847413,
      "learning_rate": 1.8944083903743925e-05,
      "loss": 1.4035,
      "step": 2038
    },
    {
      "epoch": 0.521984,
      "grad_norm": 0.3761748606810081,
      "learning_rate": 1.8927854501632545e-05,
      "loss": 1.3427,
      "step": 2039
    },
    {
      "epoch": 0.52224,
      "grad_norm": 0.3914494264900448,
      "learning_rate": 1.891162580751544e-05,
      "loss": 1.3646,
      "step": 2040
    },
    {
      "epoch": 0.522496,
      "grad_norm": 0.36885682978584305,
      "learning_rate": 1.8895397832109278e-05,
      "loss": 1.3281,
      "step": 2041
    },
    {
      "epoch": 0.522752,
      "grad_norm": 0.38415728985535125,
      "learning_rate": 1.887917058613025e-05,
      "loss": 1.3353,
      "step": 2042
    },
    {
      "epoch": 0.523008,
      "grad_norm": 0.38406231515071804,
      "learning_rate": 1.8862944080294064e-05,
      "loss": 1.3523,
      "step": 2043
    },
    {
      "epoch": 0.523264,
      "grad_norm": 0.4364899047138978,
      "learning_rate": 1.8846718325315932e-05,
      "loss": 1.3524,
      "step": 2044
    },
    {
      "epoch": 0.52352,
      "grad_norm": 0.40585140573911505,
      "learning_rate": 1.8830493331910585e-05,
      "loss": 1.3624,
      "step": 2045
    },
    {
      "epoch": 0.523776,
      "grad_norm": 0.3866723892567866,
      "learning_rate": 1.881426911079223e-05,
      "loss": 1.3303,
      "step": 2046
    },
    {
      "epoch": 0.524032,
      "grad_norm": 0.37406192942239963,
      "learning_rate": 1.879804567267459e-05,
      "loss": 1.3459,
      "step": 2047
    },
    {
      "epoch": 0.524288,
      "grad_norm": 0.41742522714072416,
      "learning_rate": 1.878182302827085e-05,
      "loss": 1.4168,
      "step": 2048
    },
    {
      "epoch": 0.524544,
      "grad_norm": 0.372405924295808,
      "learning_rate": 1.8765601188293684e-05,
      "loss": 1.3732,
      "step": 2049
    },
    {
      "epoch": 0.5248,
      "grad_norm": 0.376264613982168,
      "learning_rate": 1.8749380163455233e-05,
      "loss": 1.3402,
      "step": 2050
    },
    {
      "epoch": 0.525056,
      "grad_norm": 0.42417981347771366,
      "learning_rate": 1.873315996446708e-05,
      "loss": 1.331,
      "step": 2051
    },
    {
      "epoch": 0.525312,
      "grad_norm": 0.37369764743074074,
      "learning_rate": 1.8716940602040288e-05,
      "loss": 1.3486,
      "step": 2052
    },
    {
      "epoch": 0.525568,
      "grad_norm": 0.38997774182167544,
      "learning_rate": 1.8700722086885362e-05,
      "loss": 1.3656,
      "step": 2053
    },
    {
      "epoch": 0.525824,
      "grad_norm": 0.37788192897730455,
      "learning_rate": 1.8684504429712243e-05,
      "loss": 1.3516,
      "step": 2054
    },
    {
      "epoch": 0.52608,
      "grad_norm": 0.3826998166530222,
      "learning_rate": 1.8668287641230297e-05,
      "loss": 1.3475,
      "step": 2055
    },
    {
      "epoch": 0.526336,
      "grad_norm": 0.38907979309832097,
      "learning_rate": 1.865207173214835e-05,
      "loss": 1.3993,
      "step": 2056
    },
    {
      "epoch": 0.526592,
      "grad_norm": 0.3958424846120189,
      "learning_rate": 1.863585671317459e-05,
      "loss": 1.3511,
      "step": 2057
    },
    {
      "epoch": 0.526848,
      "grad_norm": 0.39056588073092524,
      "learning_rate": 1.8619642595016666e-05,
      "loss": 1.4097,
      "step": 2058
    },
    {
      "epoch": 0.527104,
      "grad_norm": 0.3861322913089585,
      "learning_rate": 1.8603429388381616e-05,
      "loss": 1.3908,
      "step": 2059
    },
    {
      "epoch": 0.52736,
      "grad_norm": 0.37809193121786944,
      "learning_rate": 1.8587217103975883e-05,
      "loss": 1.3498,
      "step": 2060
    },
    {
      "epoch": 0.527616,
      "grad_norm": 0.4634022980405316,
      "learning_rate": 1.8571005752505283e-05,
      "loss": 1.3535,
      "step": 2061
    },
    {
      "epoch": 0.527872,
      "grad_norm": 0.3734249420568222,
      "learning_rate": 1.855479534467504e-05,
      "loss": 1.3504,
      "step": 2062
    },
    {
      "epoch": 0.528128,
      "grad_norm": 0.38513967648013914,
      "learning_rate": 1.8538585891189728e-05,
      "loss": 1.3613,
      "step": 2063
    },
    {
      "epoch": 0.528384,
      "grad_norm": 0.3952614760238835,
      "learning_rate": 1.852237740275331e-05,
      "loss": 1.3569,
      "step": 2064
    },
    {
      "epoch": 0.52864,
      "grad_norm": 0.3799787486364502,
      "learning_rate": 1.8506169890069107e-05,
      "loss": 1.3432,
      "step": 2065
    },
    {
      "epoch": 0.528896,
      "grad_norm": 0.3772255968606647,
      "learning_rate": 1.84899633638398e-05,
      "loss": 1.3722,
      "step": 2066
    },
    {
      "epoch": 0.529152,
      "grad_norm": 0.3952663829807274,
      "learning_rate": 1.8473757834767407e-05,
      "loss": 1.3679,
      "step": 2067
    },
    {
      "epoch": 0.529408,
      "grad_norm": 0.38699406164517774,
      "learning_rate": 1.8457553313553307e-05,
      "loss": 1.3502,
      "step": 2068
    },
    {
      "epoch": 0.529664,
      "grad_norm": 0.3747695467247886,
      "learning_rate": 1.8441349810898183e-05,
      "loss": 1.343,
      "step": 2069
    },
    {
      "epoch": 0.52992,
      "grad_norm": 0.3912642106608188,
      "learning_rate": 1.842514733750207e-05,
      "loss": 1.3621,
      "step": 2070
    },
    {
      "epoch": 0.530176,
      "grad_norm": 0.37617577873717256,
      "learning_rate": 1.8408945904064318e-05,
      "loss": 1.3442,
      "step": 2071
    },
    {
      "epoch": 0.530432,
      "grad_norm": 0.3822158921231808,
      "learning_rate": 1.8392745521283595e-05,
      "loss": 1.3835,
      "step": 2072
    },
    {
      "epoch": 0.530688,
      "grad_norm": 0.3829521686892734,
      "learning_rate": 1.8376546199857863e-05,
      "loss": 1.3116,
      "step": 2073
    },
    {
      "epoch": 0.530944,
      "grad_norm": 0.3912827432901779,
      "learning_rate": 1.8360347950484394e-05,
      "loss": 1.371,
      "step": 2074
    },
    {
      "epoch": 0.5312,
      "grad_norm": 0.38984717198741525,
      "learning_rate": 1.8344150783859735e-05,
      "loss": 1.4056,
      "step": 2075
    },
    {
      "epoch": 0.531456,
      "grad_norm": 0.3800195058843405,
      "learning_rate": 1.832795471067974e-05,
      "loss": 1.3839,
      "step": 2076
    },
    {
      "epoch": 0.531712,
      "grad_norm": 0.4058431189404512,
      "learning_rate": 1.8311759741639532e-05,
      "loss": 1.3924,
      "step": 2077
    },
    {
      "epoch": 0.531968,
      "grad_norm": 0.38818070189647086,
      "learning_rate": 1.82955658874335e-05,
      "loss": 1.3829,
      "step": 2078
    },
    {
      "epoch": 0.532224,
      "grad_norm": 0.3889718938267594,
      "learning_rate": 1.8279373158755302e-05,
      "loss": 1.3909,
      "step": 2079
    },
    {
      "epoch": 0.53248,
      "grad_norm": 0.38596348551667153,
      "learning_rate": 1.826318156629786e-05,
      "loss": 1.3399,
      "step": 2080
    },
    {
      "epoch": 0.532736,
      "grad_norm": 0.38073285372814036,
      "learning_rate": 1.824699112075332e-05,
      "loss": 1.3282,
      "step": 2081
    },
    {
      "epoch": 0.532992,
      "grad_norm": 0.3887279076016933,
      "learning_rate": 1.8230801832813098e-05,
      "loss": 1.3958,
      "step": 2082
    },
    {
      "epoch": 0.533248,
      "grad_norm": 0.38026456510914264,
      "learning_rate": 1.821461371316784e-05,
      "loss": 1.3397,
      "step": 2083
    },
    {
      "epoch": 0.533504,
      "grad_norm": 0.3840468334946173,
      "learning_rate": 1.8198426772507403e-05,
      "loss": 1.3969,
      "step": 2084
    },
    {
      "epoch": 0.53376,
      "grad_norm": 0.36735772950034606,
      "learning_rate": 1.8182241021520894e-05,
      "loss": 1.3497,
      "step": 2085
    },
    {
      "epoch": 0.534016,
      "grad_norm": 0.37727340891308475,
      "learning_rate": 1.816605647089662e-05,
      "loss": 1.3405,
      "step": 2086
    },
    {
      "epoch": 0.534272,
      "grad_norm": 0.3817167618739919,
      "learning_rate": 1.8149873131322074e-05,
      "loss": 1.3528,
      "step": 2087
    },
    {
      "epoch": 0.534528,
      "grad_norm": 0.37470028659453597,
      "learning_rate": 1.813369101348398e-05,
      "loss": 1.3704,
      "step": 2088
    },
    {
      "epoch": 0.534784,
      "grad_norm": 0.3815150268056056,
      "learning_rate": 1.8117510128068256e-05,
      "loss": 1.3744,
      "step": 2089
    },
    {
      "epoch": 0.53504,
      "grad_norm": 0.381605010560828,
      "learning_rate": 1.810133048575998e-05,
      "loss": 1.3815,
      "step": 2090
    },
    {
      "epoch": 0.535296,
      "grad_norm": 0.3719371477350257,
      "learning_rate": 1.8085152097243427e-05,
      "loss": 1.3317,
      "step": 2091
    },
    {
      "epoch": 0.535552,
      "grad_norm": 0.3775510754806728,
      "learning_rate": 1.8068974973202054e-05,
      "loss": 1.3553,
      "step": 2092
    },
    {
      "epoch": 0.535808,
      "grad_norm": 0.3856223560753888,
      "learning_rate": 1.805279912431845e-05,
      "loss": 1.3624,
      "step": 2093
    },
    {
      "epoch": 0.536064,
      "grad_norm": 0.37921495698941027,
      "learning_rate": 1.803662456127439e-05,
      "loss": 1.3477,
      "step": 2094
    },
    {
      "epoch": 0.53632,
      "grad_norm": 0.37422915813102103,
      "learning_rate": 1.8020451294750803e-05,
      "loss": 1.3367,
      "step": 2095
    },
    {
      "epoch": 0.536576,
      "grad_norm": 0.39208733182668565,
      "learning_rate": 1.8004279335427736e-05,
      "loss": 1.3699,
      "step": 2096
    },
    {
      "epoch": 0.536832,
      "grad_norm": 0.45200517766565956,
      "learning_rate": 1.798810869398439e-05,
      "loss": 1.3528,
      "step": 2097
    },
    {
      "epoch": 0.537088,
      "grad_norm": 0.38664350422940386,
      "learning_rate": 1.7971939381099096e-05,
      "loss": 1.3731,
      "step": 2098
    },
    {
      "epoch": 0.537344,
      "grad_norm": 0.3749868711865193,
      "learning_rate": 1.7955771407449318e-05,
      "loss": 1.3312,
      "step": 2099
    },
    {
      "epoch": 0.5376,
      "grad_norm": 0.3802562997281817,
      "learning_rate": 1.7939604783711604e-05,
      "loss": 1.3421,
      "step": 2100
    },
    {
      "epoch": 0.537856,
      "grad_norm": 0.3847721004167942,
      "learning_rate": 1.7923439520561633e-05,
      "loss": 1.4153,
      "step": 2101
    },
    {
      "epoch": 0.538112,
      "grad_norm": 0.38647196295576564,
      "learning_rate": 1.790727562867418e-05,
      "loss": 1.376,
      "step": 2102
    },
    {
      "epoch": 0.538368,
      "grad_norm": 0.3867842013063015,
      "learning_rate": 1.7891113118723124e-05,
      "loss": 1.3764,
      "step": 2103
    },
    {
      "epoch": 0.538624,
      "grad_norm": 0.3675020116716856,
      "learning_rate": 1.7874952001381422e-05,
      "loss": 1.3311,
      "step": 2104
    },
    {
      "epoch": 0.53888,
      "grad_norm": 0.3802479662147957,
      "learning_rate": 1.7858792287321115e-05,
      "loss": 1.3523,
      "step": 2105
    },
    {
      "epoch": 0.539136,
      "grad_norm": 0.3838485199168695,
      "learning_rate": 1.7842633987213306e-05,
      "loss": 1.3881,
      "step": 2106
    },
    {
      "epoch": 0.539392,
      "grad_norm": 0.38426644959950734,
      "learning_rate": 1.782647711172818e-05,
      "loss": 1.3757,
      "step": 2107
    },
    {
      "epoch": 0.539648,
      "grad_norm": 0.38371774530606056,
      "learning_rate": 1.781032167153497e-05,
      "loss": 1.3342,
      "step": 2108
    },
    {
      "epoch": 0.539904,
      "grad_norm": 0.3745116546620054,
      "learning_rate": 1.779416767730197e-05,
      "loss": 1.3521,
      "step": 2109
    },
    {
      "epoch": 0.54016,
      "grad_norm": 0.3855687090157187,
      "learning_rate": 1.7778015139696522e-05,
      "loss": 1.3714,
      "step": 2110
    },
    {
      "epoch": 0.540416,
      "grad_norm": 0.3675427258179148,
      "learning_rate": 1.776186406938499e-05,
      "loss": 1.3271,
      "step": 2111
    },
    {
      "epoch": 0.540672,
      "grad_norm": 0.38420824151141797,
      "learning_rate": 1.7745714477032785e-05,
      "loss": 1.3768,
      "step": 2112
    },
    {
      "epoch": 0.540928,
      "grad_norm": 0.3736316609285772,
      "learning_rate": 1.7729566373304324e-05,
      "loss": 1.3495,
      "step": 2113
    },
    {
      "epoch": 0.541184,
      "grad_norm": 0.5039960628789155,
      "learning_rate": 1.7713419768863062e-05,
      "loss": 1.3245,
      "step": 2114
    },
    {
      "epoch": 0.54144,
      "grad_norm": 0.3951841030218529,
      "learning_rate": 1.769727467437145e-05,
      "loss": 1.3773,
      "step": 2115
    },
    {
      "epoch": 0.541696,
      "grad_norm": 0.38575774498899407,
      "learning_rate": 1.768113110049096e-05,
      "loss": 1.3765,
      "step": 2116
    },
    {
      "epoch": 0.541952,
      "grad_norm": 0.4030279378238872,
      "learning_rate": 1.7664989057882028e-05,
      "loss": 1.3747,
      "step": 2117
    },
    {
      "epoch": 0.542208,
      "grad_norm": 0.3824938246922401,
      "learning_rate": 1.7648848557204108e-05,
      "loss": 1.36,
      "step": 2118
    },
    {
      "epoch": 0.542464,
      "grad_norm": 0.3813476261543493,
      "learning_rate": 1.7632709609115615e-05,
      "loss": 1.3599,
      "step": 2119
    },
    {
      "epoch": 0.54272,
      "grad_norm": 0.3854620647729533,
      "learning_rate": 1.7616572224273955e-05,
      "loss": 1.3774,
      "step": 2120
    },
    {
      "epoch": 0.542976,
      "grad_norm": 0.3872480031165972,
      "learning_rate": 1.7600436413335497e-05,
      "loss": 1.3385,
      "step": 2121
    },
    {
      "epoch": 0.543232,
      "grad_norm": 0.37459907478509025,
      "learning_rate": 1.7584302186955566e-05,
      "loss": 1.3637,
      "step": 2122
    },
    {
      "epoch": 0.543488,
      "grad_norm": 0.37003604400193363,
      "learning_rate": 1.7568169555788452e-05,
      "loss": 1.3277,
      "step": 2123
    },
    {
      "epoch": 0.543744,
      "grad_norm": 0.3897103268959548,
      "learning_rate": 1.7552038530487365e-05,
      "loss": 1.3606,
      "step": 2124
    },
    {
      "epoch": 0.544,
      "grad_norm": 0.3842180135214001,
      "learning_rate": 1.7535909121704483e-05,
      "loss": 1.3797,
      "step": 2125
    },
    {
      "epoch": 0.544256,
      "grad_norm": 0.3793962630856666,
      "learning_rate": 1.7519781340090907e-05,
      "loss": 1.3541,
      "step": 2126
    },
    {
      "epoch": 0.544512,
      "grad_norm": 0.38147543127416006,
      "learning_rate": 1.7503655196296663e-05,
      "loss": 1.3499,
      "step": 2127
    },
    {
      "epoch": 0.544768,
      "grad_norm": 0.37522914726155804,
      "learning_rate": 1.7487530700970695e-05,
      "loss": 1.3275,
      "step": 2128
    },
    {
      "epoch": 0.545024,
      "grad_norm": 0.3806746091578742,
      "learning_rate": 1.7471407864760865e-05,
      "loss": 1.3549,
      "step": 2129
    },
    {
      "epoch": 0.54528,
      "grad_norm": 0.3822651391506844,
      "learning_rate": 1.7455286698313915e-05,
      "loss": 1.336,
      "step": 2130
    },
    {
      "epoch": 0.545536,
      "grad_norm": 0.3748034070970487,
      "learning_rate": 1.7439167212275513e-05,
      "loss": 1.3661,
      "step": 2131
    },
    {
      "epoch": 0.545792,
      "grad_norm": 0.37888354545732184,
      "learning_rate": 1.742304941729021e-05,
      "loss": 1.369,
      "step": 2132
    },
    {
      "epoch": 0.546048,
      "grad_norm": 0.3856861662677533,
      "learning_rate": 1.7406933324001434e-05,
      "loss": 1.3936,
      "step": 2133
    },
    {
      "epoch": 0.546304,
      "grad_norm": 0.38421050317964134,
      "learning_rate": 1.739081894305149e-05,
      "loss": 1.3731,
      "step": 2134
    },
    {
      "epoch": 0.54656,
      "grad_norm": 0.38700673161803956,
      "learning_rate": 1.737470628508156e-05,
      "loss": 1.3835,
      "step": 2135
    },
    {
      "epoch": 0.546816,
      "grad_norm": 0.36366413651031415,
      "learning_rate": 1.7358595360731677e-05,
      "loss": 1.3355,
      "step": 2136
    },
    {
      "epoch": 0.547072,
      "grad_norm": 0.37453922051959276,
      "learning_rate": 1.7342486180640734e-05,
      "loss": 1.3516,
      "step": 2137
    },
    {
      "epoch": 0.547328,
      "grad_norm": 0.3830745867488307,
      "learning_rate": 1.7326378755446477e-05,
      "loss": 1.3693,
      "step": 2138
    },
    {
      "epoch": 0.547584,
      "grad_norm": 0.3662720842511759,
      "learning_rate": 1.7310273095785493e-05,
      "loss": 1.3627,
      "step": 2139
    },
    {
      "epoch": 0.54784,
      "grad_norm": 0.3743521462657609,
      "learning_rate": 1.729416921229319e-05,
      "loss": 1.3267,
      "step": 2140
    },
    {
      "epoch": 0.548096,
      "grad_norm": 0.37170806478825064,
      "learning_rate": 1.727806711560383e-05,
      "loss": 1.3315,
      "step": 2141
    },
    {
      "epoch": 0.548352,
      "grad_norm": 0.38371584915491314,
      "learning_rate": 1.726196681635045e-05,
      "loss": 1.356,
      "step": 2142
    },
    {
      "epoch": 0.548608,
      "grad_norm": 0.36881069578312703,
      "learning_rate": 1.724586832516495e-05,
      "loss": 1.3255,
      "step": 2143
    },
    {
      "epoch": 0.548864,
      "grad_norm": 0.36831457203626694,
      "learning_rate": 1.7229771652678e-05,
      "loss": 1.3442,
      "step": 2144
    },
    {
      "epoch": 0.54912,
      "grad_norm": 0.369309849625158,
      "learning_rate": 1.7213676809519097e-05,
      "loss": 1.391,
      "step": 2145
    },
    {
      "epoch": 0.549376,
      "grad_norm": 0.3818064674117364,
      "learning_rate": 1.7197583806316503e-05,
      "loss": 1.4232,
      "step": 2146
    },
    {
      "epoch": 0.549632,
      "grad_norm": 0.37144236216381116,
      "learning_rate": 1.7181492653697294e-05,
      "loss": 1.3693,
      "step": 2147
    },
    {
      "epoch": 0.549888,
      "grad_norm": 0.38241896266778236,
      "learning_rate": 1.7165403362287282e-05,
      "loss": 1.3803,
      "step": 2148
    },
    {
      "epoch": 0.550144,
      "grad_norm": 0.3789863922718266,
      "learning_rate": 1.7149315942711095e-05,
      "loss": 1.3494,
      "step": 2149
    },
    {
      "epoch": 0.5504,
      "grad_norm": 0.3825600834270111,
      "learning_rate": 1.7133230405592104e-05,
      "loss": 1.3613,
      "step": 2150
    },
    {
      "epoch": 0.550656,
      "grad_norm": 0.3816437222214102,
      "learning_rate": 1.7117146761552427e-05,
      "loss": 1.3612,
      "step": 2151
    },
    {
      "epoch": 0.550912,
      "grad_norm": 0.3805199233229356,
      "learning_rate": 1.7101065021212953e-05,
      "loss": 1.3333,
      "step": 2152
    },
    {
      "epoch": 0.551168,
      "grad_norm": 0.3653836458901507,
      "learning_rate": 1.7084985195193314e-05,
      "loss": 1.3175,
      "step": 2153
    },
    {
      "epoch": 0.551424,
      "grad_norm": 0.37791705846437934,
      "learning_rate": 1.7068907294111846e-05,
      "loss": 1.3723,
      "step": 2154
    },
    {
      "epoch": 0.55168,
      "grad_norm": 0.3690739756841646,
      "learning_rate": 1.7052831328585644e-05,
      "loss": 1.3116,
      "step": 2155
    },
    {
      "epoch": 0.551936,
      "grad_norm": 0.3670831997602546,
      "learning_rate": 1.7036757309230528e-05,
      "loss": 1.3333,
      "step": 2156
    },
    {
      "epoch": 0.552192,
      "grad_norm": 0.3853856160194879,
      "learning_rate": 1.7020685246661005e-05,
      "loss": 1.3806,
      "step": 2157
    },
    {
      "epoch": 0.552448,
      "grad_norm": 0.3802903542821904,
      "learning_rate": 1.7004615151490315e-05,
      "loss": 1.3389,
      "step": 2158
    },
    {
      "epoch": 0.552704,
      "grad_norm": 0.37936384609356444,
      "learning_rate": 1.6988547034330398e-05,
      "loss": 1.3684,
      "step": 2159
    },
    {
      "epoch": 0.55296,
      "grad_norm": 0.9491242525761396,
      "learning_rate": 1.6972480905791866e-05,
      "loss": 1.4162,
      "step": 2160
    },
    {
      "epoch": 0.553216,
      "grad_norm": 0.38475694090959206,
      "learning_rate": 1.6956416776484033e-05,
      "loss": 1.3494,
      "step": 2161
    },
    {
      "epoch": 0.553472,
      "grad_norm": 0.37109422392276603,
      "learning_rate": 1.6940354657014904e-05,
      "loss": 1.3229,
      "step": 2162
    },
    {
      "epoch": 0.553728,
      "grad_norm": 0.4284737805277542,
      "learning_rate": 1.6924294557991133e-05,
      "loss": 1.365,
      "step": 2163
    },
    {
      "epoch": 0.553984,
      "grad_norm": 0.39306648811881995,
      "learning_rate": 1.6908236490018057e-05,
      "loss": 1.3528,
      "step": 2164
    },
    {
      "epoch": 0.55424,
      "grad_norm": 0.3819950028296498,
      "learning_rate": 1.6892180463699662e-05,
      "loss": 1.3367,
      "step": 2165
    },
    {
      "epoch": 0.554496,
      "grad_norm": 0.3797595745228369,
      "learning_rate": 1.68761264896386e-05,
      "loss": 1.3526,
      "step": 2166
    },
    {
      "epoch": 0.554752,
      "grad_norm": 0.3810791741677055,
      "learning_rate": 1.6860074578436144e-05,
      "loss": 1.3682,
      "step": 2167
    },
    {
      "epoch": 0.555008,
      "grad_norm": 0.483730628140365,
      "learning_rate": 1.6844024740692225e-05,
      "loss": 1.3064,
      "step": 2168
    },
    {
      "epoch": 0.555264,
      "grad_norm": 0.37752333239113534,
      "learning_rate": 1.6827976987005396e-05,
      "loss": 1.3327,
      "step": 2169
    },
    {
      "epoch": 0.55552,
      "grad_norm": 0.40352985160311683,
      "learning_rate": 1.681193132797284e-05,
      "loss": 1.3625,
      "step": 2170
    },
    {
      "epoch": 0.555776,
      "grad_norm": 0.3748604787055534,
      "learning_rate": 1.6795887774190347e-05,
      "loss": 1.3689,
      "step": 2171
    },
    {
      "epoch": 0.556032,
      "grad_norm": 0.398637746166116,
      "learning_rate": 1.6779846336252338e-05,
      "loss": 1.401,
      "step": 2172
    },
    {
      "epoch": 0.556288,
      "grad_norm": 0.3760939184136376,
      "learning_rate": 1.6763807024751803e-05,
      "loss": 1.3743,
      "step": 2173
    },
    {
      "epoch": 0.556544,
      "grad_norm": 0.38519622883152477,
      "learning_rate": 1.6747769850280347e-05,
      "loss": 1.363,
      "step": 2174
    },
    {
      "epoch": 0.5568,
      "grad_norm": 0.3771531178584084,
      "learning_rate": 1.6731734823428173e-05,
      "loss": 1.3428,
      "step": 2175
    },
    {
      "epoch": 0.557056,
      "grad_norm": 0.37525144111512165,
      "learning_rate": 1.671570195478405e-05,
      "loss": 1.3512,
      "step": 2176
    },
    {
      "epoch": 0.557312,
      "grad_norm": 0.3801214309577439,
      "learning_rate": 1.6699671254935327e-05,
      "loss": 1.3465,
      "step": 2177
    },
    {
      "epoch": 0.557568,
      "grad_norm": 0.4186195347973017,
      "learning_rate": 1.668364273446794e-05,
      "loss": 1.3848,
      "step": 2178
    },
    {
      "epoch": 0.557824,
      "grad_norm": 0.37049428840564413,
      "learning_rate": 1.6667616403966336e-05,
      "loss": 1.3168,
      "step": 2179
    },
    {
      "epoch": 0.55808,
      "grad_norm": 0.3715164585563844,
      "learning_rate": 1.6651592274013566e-05,
      "loss": 1.3396,
      "step": 2180
    },
    {
      "epoch": 0.558336,
      "grad_norm": 0.3831083669912095,
      "learning_rate": 1.6635570355191203e-05,
      "loss": 1.3504,
      "step": 2181
    },
    {
      "epoch": 0.558592,
      "grad_norm": 0.36967405561187333,
      "learning_rate": 1.6619550658079367e-05,
      "loss": 1.3529,
      "step": 2182
    },
    {
      "epoch": 0.558848,
      "grad_norm": 0.3720040958031323,
      "learning_rate": 1.660353319325672e-05,
      "loss": 1.382,
      "step": 2183
    },
    {
      "epoch": 0.559104,
      "grad_norm": 0.37710264718784026,
      "learning_rate": 1.6587517971300426e-05,
      "loss": 1.3172,
      "step": 2184
    },
    {
      "epoch": 0.55936,
      "grad_norm": 0.3761260822686075,
      "learning_rate": 1.657150500278619e-05,
      "loss": 1.3763,
      "step": 2185
    },
    {
      "epoch": 0.559616,
      "grad_norm": 0.37354442154961,
      "learning_rate": 1.6555494298288214e-05,
      "loss": 1.3703,
      "step": 2186
    },
    {
      "epoch": 0.559872,
      "grad_norm": 0.38031763624367715,
      "learning_rate": 1.653948586837921e-05,
      "loss": 1.331,
      "step": 2187
    },
    {
      "epoch": 0.560128,
      "grad_norm": 0.3748668328803114,
      "learning_rate": 1.6523479723630395e-05,
      "loss": 1.3676,
      "step": 2188
    },
    {
      "epoch": 0.560384,
      "grad_norm": 0.36432902561148695,
      "learning_rate": 1.650747587461148e-05,
      "loss": 1.3251,
      "step": 2189
    },
    {
      "epoch": 0.56064,
      "grad_norm": 0.3813433918307206,
      "learning_rate": 1.6491474331890635e-05,
      "loss": 1.3768,
      "step": 2190
    },
    {
      "epoch": 0.560896,
      "grad_norm": 0.37870179057947584,
      "learning_rate": 1.6475475106034532e-05,
      "loss": 1.3708,
      "step": 2191
    },
    {
      "epoch": 0.561152,
      "grad_norm": 0.3712716619197617,
      "learning_rate": 1.64594782076083e-05,
      "loss": 1.3586,
      "step": 2192
    },
    {
      "epoch": 0.561408,
      "grad_norm": 0.37815430976426484,
      "learning_rate": 1.6443483647175543e-05,
      "loss": 1.367,
      "step": 2193
    },
    {
      "epoch": 0.561664,
      "grad_norm": 0.37732382749011745,
      "learning_rate": 1.6427491435298302e-05,
      "loss": 1.3716,
      "step": 2194
    },
    {
      "epoch": 0.56192,
      "grad_norm": 0.3701154665843065,
      "learning_rate": 1.6411501582537094e-05,
      "loss": 1.328,
      "step": 2195
    },
    {
      "epoch": 0.562176,
      "grad_norm": 0.37557034264861267,
      "learning_rate": 1.639551409945086e-05,
      "loss": 1.379,
      "step": 2196
    },
    {
      "epoch": 0.562432,
      "grad_norm": 0.3727337721669061,
      "learning_rate": 1.6379528996596968e-05,
      "loss": 1.3494,
      "step": 2197
    },
    {
      "epoch": 0.562688,
      "grad_norm": 0.3701846177575583,
      "learning_rate": 1.6363546284531228e-05,
      "loss": 1.3327,
      "step": 2198
    },
    {
      "epoch": 0.562944,
      "grad_norm": 0.38673245451596755,
      "learning_rate": 1.6347565973807874e-05,
      "loss": 1.3932,
      "step": 2199
    },
    {
      "epoch": 0.5632,
      "grad_norm": 0.38156398456502344,
      "learning_rate": 1.6331588074979548e-05,
      "loss": 1.3683,
      "step": 2200
    },
    {
      "epoch": 0.563456,
      "grad_norm": 0.3814057118745101,
      "learning_rate": 1.6315612598597303e-05,
      "loss": 1.321,
      "step": 2201
    },
    {
      "epoch": 0.563712,
      "grad_norm": 0.3906389233844145,
      "learning_rate": 1.629963955521059e-05,
      "loss": 1.4058,
      "step": 2202
    },
    {
      "epoch": 0.563968,
      "grad_norm": 0.39740377605693994,
      "learning_rate": 1.6283668955367242e-05,
      "loss": 1.3838,
      "step": 2203
    },
    {
      "epoch": 0.564224,
      "grad_norm": 0.4108933047385235,
      "learning_rate": 1.6267700809613495e-05,
      "loss": 1.3624,
      "step": 2204
    },
    {
      "epoch": 0.56448,
      "grad_norm": 0.38758479265767487,
      "learning_rate": 1.625173512849396e-05,
      "loss": 1.381,
      "step": 2205
    },
    {
      "epoch": 0.564736,
      "grad_norm": 0.36226549413256043,
      "learning_rate": 1.6235771922551625e-05,
      "loss": 1.3357,
      "step": 2206
    },
    {
      "epoch": 0.564992,
      "grad_norm": 0.4012347514162213,
      "learning_rate": 1.6219811202327823e-05,
      "loss": 1.3705,
      "step": 2207
    },
    {
      "epoch": 0.565248,
      "grad_norm": 0.4016027962429828,
      "learning_rate": 1.6203852978362283e-05,
      "loss": 1.3923,
      "step": 2208
    },
    {
      "epoch": 0.565504,
      "grad_norm": 0.38342917553150113,
      "learning_rate": 1.618789726119304e-05,
      "loss": 1.3149,
      "step": 2209
    },
    {
      "epoch": 0.56576,
      "grad_norm": 0.39146762325918033,
      "learning_rate": 1.6171944061356502e-05,
      "loss": 1.3817,
      "step": 2210
    },
    {
      "epoch": 0.566016,
      "grad_norm": 0.3876385019662338,
      "learning_rate": 1.6155993389387416e-05,
      "loss": 1.3732,
      "step": 2211
    },
    {
      "epoch": 0.566272,
      "grad_norm": 0.37806095845462007,
      "learning_rate": 1.6140045255818852e-05,
      "loss": 1.33,
      "step": 2212
    },
    {
      "epoch": 0.566528,
      "grad_norm": 0.3867062529623233,
      "learning_rate": 1.61240996711822e-05,
      "loss": 1.3505,
      "step": 2213
    },
    {
      "epoch": 0.566784,
      "grad_norm": 0.39085764650292765,
      "learning_rate": 1.610815664600718e-05,
      "loss": 1.3703,
      "step": 2214
    },
    {
      "epoch": 0.56704,
      "grad_norm": 0.3737346008548558,
      "learning_rate": 1.6092216190821797e-05,
      "loss": 1.3803,
      "step": 2215
    },
    {
      "epoch": 0.567296,
      "grad_norm": 0.393946507127276,
      "learning_rate": 1.6076278316152385e-05,
      "loss": 1.3786,
      "step": 2216
    },
    {
      "epoch": 0.567552,
      "grad_norm": 0.3792476393960428,
      "learning_rate": 1.6060343032523565e-05,
      "loss": 1.3512,
      "step": 2217
    },
    {
      "epoch": 0.567808,
      "grad_norm": 0.37555835051329145,
      "learning_rate": 1.604441035045825e-05,
      "loss": 1.3677,
      "step": 2218
    },
    {
      "epoch": 0.568064,
      "grad_norm": 0.3727881241973694,
      "learning_rate": 1.6028480280477623e-05,
      "loss": 1.3184,
      "step": 2219
    },
    {
      "epoch": 0.56832,
      "grad_norm": 0.38416225231854084,
      "learning_rate": 1.601255283310116e-05,
      "loss": 1.3418,
      "step": 2220
    },
    {
      "epoch": 0.568576,
      "grad_norm": 0.3728642396381784,
      "learning_rate": 1.5996628018846583e-05,
      "loss": 1.3196,
      "step": 2221
    },
    {
      "epoch": 0.568832,
      "grad_norm": 0.3853443838238744,
      "learning_rate": 1.598070584822989e-05,
      "loss": 1.3615,
      "step": 2222
    },
    {
      "epoch": 0.569088,
      "grad_norm": 0.3789759624321883,
      "learning_rate": 1.5964786331765338e-05,
      "loss": 1.3322,
      "step": 2223
    },
    {
      "epoch": 0.569344,
      "grad_norm": 0.3809667140495396,
      "learning_rate": 1.5948869479965427e-05,
      "loss": 1.3349,
      "step": 2224
    },
    {
      "epoch": 0.5696,
      "grad_norm": 0.36925905764877975,
      "learning_rate": 1.593295530334088e-05,
      "loss": 1.3141,
      "step": 2225
    },
    {
      "epoch": 0.569856,
      "grad_norm": 0.44187942570993816,
      "learning_rate": 1.5917043812400675e-05,
      "loss": 1.4244,
      "step": 2226
    },
    {
      "epoch": 0.570112,
      "grad_norm": 0.38114195125282335,
      "learning_rate": 1.5901135017652026e-05,
      "loss": 1.3505,
      "step": 2227
    },
    {
      "epoch": 0.570368,
      "grad_norm": 0.3936183489948876,
      "learning_rate": 1.588522892960032e-05,
      "loss": 1.3875,
      "step": 2228
    },
    {
      "epoch": 0.570624,
      "grad_norm": 0.3748487198771036,
      "learning_rate": 1.5869325558749213e-05,
      "loss": 1.3402,
      "step": 2229
    },
    {
      "epoch": 0.57088,
      "grad_norm": 0.37239379674809603,
      "learning_rate": 1.585342491560052e-05,
      "loss": 1.3343,
      "step": 2230
    },
    {
      "epoch": 0.571136,
      "grad_norm": 0.3713986366846414,
      "learning_rate": 1.5837527010654288e-05,
      "loss": 1.3379,
      "step": 2231
    },
    {
      "epoch": 0.571392,
      "grad_norm": 0.3889222732551767,
      "learning_rate": 1.5821631854408737e-05,
      "loss": 1.3892,
      "step": 2232
    },
    {
      "epoch": 0.571648,
      "grad_norm": 0.3781520062206382,
      "learning_rate": 1.580573945736029e-05,
      "loss": 1.3419,
      "step": 2233
    },
    {
      "epoch": 0.571904,
      "grad_norm": 0.43138610439627034,
      "learning_rate": 1.578984983000352e-05,
      "loss": 1.3713,
      "step": 2234
    },
    {
      "epoch": 0.57216,
      "grad_norm": 0.374183877191408,
      "learning_rate": 1.5773962982831195e-05,
      "loss": 1.3715,
      "step": 2235
    },
    {
      "epoch": 0.572416,
      "grad_norm": 0.37724382780510624,
      "learning_rate": 1.575807892633423e-05,
      "loss": 1.3236,
      "step": 2236
    },
    {
      "epoch": 0.572672,
      "grad_norm": 0.36951695134079643,
      "learning_rate": 1.5742197671001718e-05,
      "loss": 1.3278,
      "step": 2237
    },
    {
      "epoch": 0.572928,
      "grad_norm": 0.3730308173512109,
      "learning_rate": 1.5726319227320886e-05,
      "loss": 1.2887,
      "step": 2238
    },
    {
      "epoch": 0.573184,
      "grad_norm": 0.4138229444030839,
      "learning_rate": 1.5710443605777116e-05,
      "loss": 1.3717,
      "step": 2239
    },
    {
      "epoch": 0.57344,
      "grad_norm": 0.3682964854089962,
      "learning_rate": 1.56945708168539e-05,
      "loss": 1.3713,
      "step": 2240
    },
    {
      "epoch": 0.573696,
      "grad_norm": 0.3821482160361737,
      "learning_rate": 1.5678700871032897e-05,
      "loss": 1.3657,
      "step": 2241
    },
    {
      "epoch": 0.573952,
      "grad_norm": 0.3882239509569752,
      "learning_rate": 1.566283377879386e-05,
      "loss": 1.3797,
      "step": 2242
    },
    {
      "epoch": 0.574208,
      "grad_norm": 0.36504684582481467,
      "learning_rate": 1.5646969550614668e-05,
      "loss": 1.3173,
      "step": 2243
    },
    {
      "epoch": 0.574464,
      "grad_norm": 0.3715772912160699,
      "learning_rate": 1.5631108196971314e-05,
      "loss": 1.367,
      "step": 2244
    },
    {
      "epoch": 0.57472,
      "grad_norm": 0.5209604103847042,
      "learning_rate": 1.561524972833789e-05,
      "loss": 1.3432,
      "step": 2245
    },
    {
      "epoch": 0.574976,
      "grad_norm": 0.3734621425514182,
      "learning_rate": 1.5599394155186572e-05,
      "loss": 1.3482,
      "step": 2246
    },
    {
      "epoch": 0.575232,
      "grad_norm": 0.44078621941441853,
      "learning_rate": 1.5583541487987634e-05,
      "loss": 1.3893,
      "step": 2247
    },
    {
      "epoch": 0.575488,
      "grad_norm": 0.37823257441116975,
      "learning_rate": 1.5567691737209426e-05,
      "loss": 1.355,
      "step": 2248
    },
    {
      "epoch": 0.575744,
      "grad_norm": 0.36797122644153557,
      "learning_rate": 1.5551844913318384e-05,
      "loss": 1.3098,
      "step": 2249
    },
    {
      "epoch": 0.576,
      "grad_norm": 0.3737947544850818,
      "learning_rate": 1.5536001026778995e-05,
      "loss": 1.3581,
      "step": 2250
    },
    {
      "epoch": 0.576256,
      "grad_norm": 0.3806811792748175,
      "learning_rate": 1.5520160088053823e-05,
      "loss": 1.3517,
      "step": 2251
    },
    {
      "epoch": 0.576512,
      "grad_norm": 0.37623117171217324,
      "learning_rate": 1.5504322107603468e-05,
      "loss": 1.355,
      "step": 2252
    },
    {
      "epoch": 0.576768,
      "grad_norm": 0.3717659864975748,
      "learning_rate": 1.5488487095886583e-05,
      "loss": 1.3108,
      "step": 2253
    },
    {
      "epoch": 0.577024,
      "grad_norm": 0.3719871971216823,
      "learning_rate": 1.5472655063359866e-05,
      "loss": 1.3729,
      "step": 2254
    },
    {
      "epoch": 0.57728,
      "grad_norm": 0.36091153773724427,
      "learning_rate": 1.545682602047804e-05,
      "loss": 1.3094,
      "step": 2255
    },
    {
      "epoch": 0.577536,
      "grad_norm": 0.38697772679152986,
      "learning_rate": 1.544099997769387e-05,
      "loss": 1.3735,
      "step": 2256
    },
    {
      "epoch": 0.577792,
      "grad_norm": 0.3751724682900414,
      "learning_rate": 1.5425176945458115e-05,
      "loss": 1.3552,
      "step": 2257
    },
    {
      "epoch": 0.578048,
      "grad_norm": 0.3687940568201914,
      "learning_rate": 1.5409356934219565e-05,
      "loss": 1.3407,
      "step": 2258
    },
    {
      "epoch": 0.578304,
      "grad_norm": 0.3710590047489983,
      "learning_rate": 1.5393539954425003e-05,
      "loss": 1.3326,
      "step": 2259
    },
    {
      "epoch": 0.57856,
      "grad_norm": 0.3882085060289895,
      "learning_rate": 1.5377726016519217e-05,
      "loss": 1.3641,
      "step": 2260
    },
    {
      "epoch": 0.578816,
      "grad_norm": 0.3655867653954755,
      "learning_rate": 1.5361915130944992e-05,
      "loss": 1.3214,
      "step": 2261
    },
    {
      "epoch": 0.579072,
      "grad_norm": 0.3730432282349772,
      "learning_rate": 1.5346107308143086e-05,
      "loss": 1.3015,
      "step": 2262
    },
    {
      "epoch": 0.579328,
      "grad_norm": 0.38426697263394805,
      "learning_rate": 1.533030255855224e-05,
      "loss": 1.3588,
      "step": 2263
    },
    {
      "epoch": 0.579584,
      "grad_norm": 0.37662249909310863,
      "learning_rate": 1.5314500892609165e-05,
      "loss": 1.3524,
      "step": 2264
    },
    {
      "epoch": 0.57984,
      "grad_norm": 0.38510202692395645,
      "learning_rate": 1.529870232074853e-05,
      "loss": 1.359,
      "step": 2265
    },
    {
      "epoch": 0.580096,
      "grad_norm": 0.3804646416042581,
      "learning_rate": 1.528290685340297e-05,
      "loss": 1.3428,
      "step": 2266
    },
    {
      "epoch": 0.580352,
      "grad_norm": 0.3876124501784752,
      "learning_rate": 1.526711450100306e-05,
      "loss": 1.3668,
      "step": 2267
    },
    {
      "epoch": 0.580608,
      "grad_norm": 0.38287855453059777,
      "learning_rate": 1.525132527397734e-05,
      "loss": 1.3572,
      "step": 2268
    },
    {
      "epoch": 0.580864,
      "grad_norm": 0.375151612670057,
      "learning_rate": 1.523553918275226e-05,
      "loss": 1.3532,
      "step": 2269
    },
    {
      "epoch": 0.58112,
      "grad_norm": 0.3919514910840354,
      "learning_rate": 1.5219756237752208e-05,
      "loss": 1.3868,
      "step": 2270
    },
    {
      "epoch": 0.581376,
      "grad_norm": 0.3806021096952684,
      "learning_rate": 1.5203976449399496e-05,
      "loss": 1.3344,
      "step": 2271
    },
    {
      "epoch": 0.581632,
      "grad_norm": 0.44032029453162475,
      "learning_rate": 1.5188199828114351e-05,
      "loss": 1.3113,
      "step": 2272
    },
    {
      "epoch": 0.581888,
      "grad_norm": 0.3793833341445387,
      "learning_rate": 1.5172426384314913e-05,
      "loss": 1.3474,
      "step": 2273
    },
    {
      "epoch": 0.582144,
      "grad_norm": 0.36680365980370855,
      "learning_rate": 1.5156656128417222e-05,
      "loss": 1.2984,
      "step": 2274
    },
    {
      "epoch": 0.5824,
      "grad_norm": 0.3775094562078009,
      "learning_rate": 1.514088907083521e-05,
      "loss": 1.3228,
      "step": 2275
    },
    {
      "epoch": 0.582656,
      "grad_norm": 0.37919190454463625,
      "learning_rate": 1.5125125221980685e-05,
      "loss": 1.3492,
      "step": 2276
    },
    {
      "epoch": 0.582912,
      "grad_norm": 0.377088138974955,
      "learning_rate": 1.5109364592263358e-05,
      "loss": 1.3123,
      "step": 2277
    },
    {
      "epoch": 0.583168,
      "grad_norm": 0.37666893129225226,
      "learning_rate": 1.5093607192090804e-05,
      "loss": 1.3163,
      "step": 2278
    },
    {
      "epoch": 0.583424,
      "grad_norm": 0.37903665498721223,
      "learning_rate": 1.5077853031868474e-05,
      "loss": 1.3353,
      "step": 2279
    },
    {
      "epoch": 0.58368,
      "grad_norm": 0.37210419459138383,
      "learning_rate": 1.506210212199966e-05,
      "loss": 1.3698,
      "step": 2280
    },
    {
      "epoch": 0.583936,
      "grad_norm": 0.3811346156665869,
      "learning_rate": 1.5046354472885531e-05,
      "loss": 1.3625,
      "step": 2281
    },
    {
      "epoch": 0.584192,
      "grad_norm": 0.368968577575401,
      "learning_rate": 1.5030610094925075e-05,
      "loss": 1.3391,
      "step": 2282
    },
    {
      "epoch": 0.584448,
      "grad_norm": 0.36808515493658445,
      "learning_rate": 1.5014868998515149e-05,
      "loss": 1.3507,
      "step": 2283
    },
    {
      "epoch": 0.584704,
      "grad_norm": 0.3781432420689751,
      "learning_rate": 1.4999131194050422e-05,
      "loss": 1.369,
      "step": 2284
    },
    {
      "epoch": 0.58496,
      "grad_norm": 0.38214116892085476,
      "learning_rate": 1.4983396691923409e-05,
      "loss": 1.4044,
      "step": 2285
    },
    {
      "epoch": 0.585216,
      "grad_norm": 0.35929579555237,
      "learning_rate": 1.496766550252442e-05,
      "loss": 1.3208,
      "step": 2286
    },
    {
      "epoch": 0.585472,
      "grad_norm": 0.3777170341823041,
      "learning_rate": 1.4951937636241596e-05,
      "loss": 1.3933,
      "step": 2287
    },
    {
      "epoch": 0.585728,
      "grad_norm": 0.3619536341409582,
      "learning_rate": 1.4936213103460887e-05,
      "loss": 1.3382,
      "step": 2288
    },
    {
      "epoch": 0.585984,
      "grad_norm": 0.36899439216065494,
      "learning_rate": 1.4920491914566006e-05,
      "loss": 1.3235,
      "step": 2289
    },
    {
      "epoch": 0.58624,
      "grad_norm": 0.359079663981322,
      "learning_rate": 1.4904774079938506e-05,
      "loss": 1.2912,
      "step": 2290
    },
    {
      "epoch": 0.586496,
      "grad_norm": 0.3792872546558347,
      "learning_rate": 1.4889059609957701e-05,
      "loss": 1.3203,
      "step": 2291
    },
    {
      "epoch": 0.586752,
      "grad_norm": 0.39150188975039885,
      "learning_rate": 1.4873348515000674e-05,
      "loss": 1.3967,
      "step": 2292
    },
    {
      "epoch": 0.587008,
      "grad_norm": 0.3667972774714096,
      "learning_rate": 1.4857640805442299e-05,
      "loss": 1.3399,
      "step": 2293
    },
    {
      "epoch": 0.587264,
      "grad_norm": 0.3654649853645106,
      "learning_rate": 1.4841936491655214e-05,
      "loss": 1.3304,
      "step": 2294
    },
    {
      "epoch": 0.58752,
      "grad_norm": 0.37133836550508914,
      "learning_rate": 1.4826235584009787e-05,
      "loss": 1.3443,
      "step": 2295
    },
    {
      "epoch": 0.587776,
      "grad_norm": 0.3699980678136288,
      "learning_rate": 1.4810538092874166e-05,
      "loss": 1.3429,
      "step": 2296
    },
    {
      "epoch": 0.588032,
      "grad_norm": 0.3828017826190828,
      "learning_rate": 1.4794844028614241e-05,
      "loss": 1.3952,
      "step": 2297
    },
    {
      "epoch": 0.588288,
      "grad_norm": 0.3823678342606419,
      "learning_rate": 1.4779153401593615e-05,
      "loss": 1.4048,
      "step": 2298
    },
    {
      "epoch": 0.588544,
      "grad_norm": 0.3873418839487346,
      "learning_rate": 1.4763466222173651e-05,
      "loss": 1.3477,
      "step": 2299
    },
    {
      "epoch": 0.5888,
      "grad_norm": 0.42093981727887153,
      "learning_rate": 1.4747782500713424e-05,
      "loss": 1.3468,
      "step": 2300
    },
    {
      "epoch": 0.589056,
      "grad_norm": 0.37627946357442715,
      "learning_rate": 1.4732102247569707e-05,
      "loss": 1.3561,
      "step": 2301
    },
    {
      "epoch": 0.589312,
      "grad_norm": 0.3774846228628711,
      "learning_rate": 1.4716425473097014e-05,
      "loss": 1.3202,
      "step": 2302
    },
    {
      "epoch": 0.589568,
      "grad_norm": 0.39446706636088746,
      "learning_rate": 1.4700752187647536e-05,
      "loss": 1.3332,
      "step": 2303
    },
    {
      "epoch": 0.589824,
      "grad_norm": 0.4571459377747097,
      "learning_rate": 1.4685082401571175e-05,
      "loss": 1.3379,
      "step": 2304
    },
    {
      "epoch": 0.59008,
      "grad_norm": 0.3855564493166995,
      "learning_rate": 1.4669416125215521e-05,
      "loss": 1.3686,
      "step": 2305
    },
    {
      "epoch": 0.590336,
      "grad_norm": 0.3715568264605294,
      "learning_rate": 1.4653753368925849e-05,
      "loss": 1.3453,
      "step": 2306
    },
    {
      "epoch": 0.590592,
      "grad_norm": 0.37031663897396006,
      "learning_rate": 1.463809414304509e-05,
      "loss": 1.333,
      "step": 2307
    },
    {
      "epoch": 0.590848,
      "grad_norm": 0.37508863617146876,
      "learning_rate": 1.4622438457913868e-05,
      "loss": 1.347,
      "step": 2308
    },
    {
      "epoch": 0.591104,
      "grad_norm": 0.3733052979112978,
      "learning_rate": 1.4606786323870453e-05,
      "loss": 1.3266,
      "step": 2309
    },
    {
      "epoch": 0.59136,
      "grad_norm": 0.37788688941543863,
      "learning_rate": 1.4591137751250773e-05,
      "loss": 1.3701,
      "step": 2310
    },
    {
      "epoch": 0.591616,
      "grad_norm": 0.453462626404309,
      "learning_rate": 1.4575492750388414e-05,
      "loss": 1.355,
      "step": 2311
    },
    {
      "epoch": 0.591872,
      "grad_norm": 0.3795413553795468,
      "learning_rate": 1.4559851331614599e-05,
      "loss": 1.3478,
      "step": 2312
    },
    {
      "epoch": 0.592128,
      "grad_norm": 0.3823990407770366,
      "learning_rate": 1.454421350525817e-05,
      "loss": 1.3624,
      "step": 2313
    },
    {
      "epoch": 0.592384,
      "grad_norm": 0.39575781388846365,
      "learning_rate": 1.452857928164562e-05,
      "loss": 1.324,
      "step": 2314
    },
    {
      "epoch": 0.59264,
      "grad_norm": 0.3750310670682684,
      "learning_rate": 1.4512948671101044e-05,
      "loss": 1.3658,
      "step": 2315
    },
    {
      "epoch": 0.592896,
      "grad_norm": 0.3672455375013226,
      "learning_rate": 1.4497321683946164e-05,
      "loss": 1.3169,
      "step": 2316
    },
    {
      "epoch": 0.593152,
      "grad_norm": 0.3784045992827003,
      "learning_rate": 1.4481698330500305e-05,
      "loss": 1.3617,
      "step": 2317
    },
    {
      "epoch": 0.593408,
      "grad_norm": 0.38081347684495337,
      "learning_rate": 1.44660786210804e-05,
      "loss": 1.3367,
      "step": 2318
    },
    {
      "epoch": 0.593664,
      "grad_norm": 0.37537983218799875,
      "learning_rate": 1.4450462566000949e-05,
      "loss": 1.3634,
      "step": 2319
    },
    {
      "epoch": 0.59392,
      "grad_norm": 0.3710171373325578,
      "learning_rate": 1.4434850175574075e-05,
      "loss": 1.344,
      "step": 2320
    },
    {
      "epoch": 0.594176,
      "grad_norm": 0.37832574770650346,
      "learning_rate": 1.441924146010945e-05,
      "loss": 1.3282,
      "step": 2321
    },
    {
      "epoch": 0.594432,
      "grad_norm": 0.3665737755222365,
      "learning_rate": 1.4403636429914341e-05,
      "loss": 1.3412,
      "step": 2322
    },
    {
      "epoch": 0.594688,
      "grad_norm": 0.3739263205029084,
      "learning_rate": 1.4388035095293574e-05,
      "loss": 1.3116,
      "step": 2323
    },
    {
      "epoch": 0.594944,
      "grad_norm": 0.3705435601086545,
      "learning_rate": 1.4372437466549538e-05,
      "loss": 1.3449,
      "step": 2324
    },
    {
      "epoch": 0.5952,
      "grad_norm": 0.374173941420483,
      "learning_rate": 1.435684355398216e-05,
      "loss": 1.3284,
      "step": 2325
    },
    {
      "epoch": 0.595456,
      "grad_norm": 0.3902715212805494,
      "learning_rate": 1.4341253367888927e-05,
      "loss": 1.3818,
      "step": 2326
    },
    {
      "epoch": 0.595712,
      "grad_norm": 0.3787382983315016,
      "learning_rate": 1.4325666918564864e-05,
      "loss": 1.3511,
      "step": 2327
    },
    {
      "epoch": 0.595968,
      "grad_norm": 0.37324421790608675,
      "learning_rate": 1.4310084216302525e-05,
      "loss": 1.3277,
      "step": 2328
    },
    {
      "epoch": 0.596224,
      "grad_norm": 0.37216348688190976,
      "learning_rate": 1.4294505271391996e-05,
      "loss": 1.3383,
      "step": 2329
    },
    {
      "epoch": 0.59648,
      "grad_norm": 0.37608682408744293,
      "learning_rate": 1.4278930094120873e-05,
      "loss": 1.3407,
      "step": 2330
    },
    {
      "epoch": 0.596736,
      "grad_norm": 0.5292557615821755,
      "learning_rate": 1.4263358694774265e-05,
      "loss": 1.3526,
      "step": 2331
    },
    {
      "epoch": 0.596992,
      "grad_norm": 0.36832019383416703,
      "learning_rate": 1.4247791083634793e-05,
      "loss": 1.333,
      "step": 2332
    },
    {
      "epoch": 0.597248,
      "grad_norm": 0.37451138684195373,
      "learning_rate": 1.4232227270982568e-05,
      "loss": 1.3408,
      "step": 2333
    },
    {
      "epoch": 0.597504,
      "grad_norm": 0.37943289258009805,
      "learning_rate": 1.4216667267095201e-05,
      "loss": 1.4125,
      "step": 2334
    },
    {
      "epoch": 0.59776,
      "grad_norm": 0.36028885322216797,
      "learning_rate": 1.4201111082247789e-05,
      "loss": 1.2867,
      "step": 2335
    },
    {
      "epoch": 0.598016,
      "grad_norm": 0.6747125891034204,
      "learning_rate": 1.4185558726712897e-05,
      "loss": 1.326,
      "step": 2336
    },
    {
      "epoch": 0.598272,
      "grad_norm": 0.445399274655276,
      "learning_rate": 1.4170010210760565e-05,
      "loss": 1.3442,
      "step": 2337
    },
    {
      "epoch": 0.598528,
      "grad_norm": 0.3810315840204842,
      "learning_rate": 1.41544655446583e-05,
      "loss": 1.3752,
      "step": 2338
    },
    {
      "epoch": 0.598784,
      "grad_norm": 0.371717800301048,
      "learning_rate": 1.4138924738671062e-05,
      "loss": 1.3174,
      "step": 2339
    },
    {
      "epoch": 0.59904,
      "grad_norm": 0.37417685716696636,
      "learning_rate": 1.4123387803061272e-05,
      "loss": 1.344,
      "step": 2340
    },
    {
      "epoch": 0.599296,
      "grad_norm": 0.3795814689348757,
      "learning_rate": 1.4107854748088793e-05,
      "loss": 1.3694,
      "step": 2341
    },
    {
      "epoch": 0.599552,
      "grad_norm": 0.3762017395579445,
      "learning_rate": 1.409232558401091e-05,
      "loss": 1.3005,
      "step": 2342
    },
    {
      "epoch": 0.599808,
      "grad_norm": 0.37458072223306627,
      "learning_rate": 1.407680032108236e-05,
      "loss": 1.3493,
      "step": 2343
    },
    {
      "epoch": 0.600064,
      "grad_norm": 0.38458300759139086,
      "learning_rate": 1.4061278969555282e-05,
      "loss": 1.3569,
      "step": 2344
    },
    {
      "epoch": 0.60032,
      "grad_norm": 0.42039908284339145,
      "learning_rate": 1.4045761539679252e-05,
      "loss": 1.3337,
      "step": 2345
    },
    {
      "epoch": 0.600576,
      "grad_norm": 0.3791067291940985,
      "learning_rate": 1.403024804170125e-05,
      "loss": 1.3501,
      "step": 2346
    },
    {
      "epoch": 0.600832,
      "grad_norm": 0.38267077164447566,
      "learning_rate": 1.4014738485865653e-05,
      "loss": 1.338,
      "step": 2347
    },
    {
      "epoch": 0.601088,
      "grad_norm": 0.3792535507507403,
      "learning_rate": 1.3999232882414243e-05,
      "loss": 1.3361,
      "step": 2348
    },
    {
      "epoch": 0.601344,
      "grad_norm": 0.36958417253225245,
      "learning_rate": 1.3983731241586191e-05,
      "loss": 1.3227,
      "step": 2349
    },
    {
      "epoch": 0.6016,
      "grad_norm": 0.3774486735552734,
      "learning_rate": 1.3968233573618037e-05,
      "loss": 1.3499,
      "step": 2350
    },
    {
      "epoch": 0.601856,
      "grad_norm": 0.37685913758937145,
      "learning_rate": 1.395273988874372e-05,
      "loss": 1.3698,
      "step": 2351
    },
    {
      "epoch": 0.602112,
      "grad_norm": 0.374781454822194,
      "learning_rate": 1.3937250197194538e-05,
      "loss": 1.32,
      "step": 2352
    },
    {
      "epoch": 0.602368,
      "grad_norm": 0.37868228915622637,
      "learning_rate": 1.3921764509199144e-05,
      "loss": 1.3726,
      "step": 2353
    },
    {
      "epoch": 0.602624,
      "grad_norm": 0.3624110669232201,
      "learning_rate": 1.3906282834983565e-05,
      "loss": 1.3179,
      "step": 2354
    },
    {
      "epoch": 0.60288,
      "grad_norm": 0.3743659017767568,
      "learning_rate": 1.3890805184771172e-05,
      "loss": 1.3654,
      "step": 2355
    },
    {
      "epoch": 0.603136,
      "grad_norm": 0.37005102367137127,
      "learning_rate": 1.3875331568782655e-05,
      "loss": 1.3314,
      "step": 2356
    },
    {
      "epoch": 0.603392,
      "grad_norm": 0.3591237958684529,
      "learning_rate": 1.3859861997236072e-05,
      "loss": 1.3368,
      "step": 2357
    },
    {
      "epoch": 0.603648,
      "grad_norm": 0.37166373867628405,
      "learning_rate": 1.3844396480346802e-05,
      "loss": 1.3414,
      "step": 2358
    },
    {
      "epoch": 0.603904,
      "grad_norm": 0.37209451746878963,
      "learning_rate": 1.3828935028327534e-05,
      "loss": 1.3797,
      "step": 2359
    },
    {
      "epoch": 0.60416,
      "grad_norm": 0.37067600859803523,
      "learning_rate": 1.3813477651388284e-05,
      "loss": 1.3844,
      "step": 2360
    },
    {
      "epoch": 0.604416,
      "grad_norm": 0.3658287213180722,
      "learning_rate": 1.379802435973638e-05,
      "loss": 1.3276,
      "step": 2361
    },
    {
      "epoch": 0.604672,
      "grad_norm": 0.3752694937051555,
      "learning_rate": 1.3782575163576435e-05,
      "loss": 1.35,
      "step": 2362
    },
    {
      "epoch": 0.604928,
      "grad_norm": 0.3690764058793174,
      "learning_rate": 1.3767130073110376e-05,
      "loss": 1.3307,
      "step": 2363
    },
    {
      "epoch": 0.605184,
      "grad_norm": 0.3855250475312962,
      "learning_rate": 1.3751689098537414e-05,
      "loss": 1.3843,
      "step": 2364
    },
    {
      "epoch": 0.60544,
      "grad_norm": 0.3738038114002888,
      "learning_rate": 1.373625225005403e-05,
      "loss": 1.348,
      "step": 2365
    },
    {
      "epoch": 0.605696,
      "grad_norm": 0.37280230715610485,
      "learning_rate": 1.3720819537853998e-05,
      "loss": 1.3397,
      "step": 2366
    },
    {
      "epoch": 0.605952,
      "grad_norm": 0.37986374886976393,
      "learning_rate": 1.370539097212836e-05,
      "loss": 1.3603,
      "step": 2367
    },
    {
      "epoch": 0.606208,
      "grad_norm": 0.3721790725838548,
      "learning_rate": 1.3689966563065394e-05,
      "loss": 1.3604,
      "step": 2368
    },
    {
      "epoch": 0.606464,
      "grad_norm": 0.3728762171397826,
      "learning_rate": 1.3674546320850661e-05,
      "loss": 1.3065,
      "step": 2369
    },
    {
      "epoch": 0.60672,
      "grad_norm": 0.37633729879751454,
      "learning_rate": 1.3659130255666968e-05,
      "loss": 1.3772,
      "step": 2370
    },
    {
      "epoch": 0.606976,
      "grad_norm": 0.368260103901516,
      "learning_rate": 1.3643718377694341e-05,
      "loss": 1.322,
      "step": 2371
    },
    {
      "epoch": 0.607232,
      "grad_norm": 0.3776350696404196,
      "learning_rate": 1.3628310697110073e-05,
      "loss": 1.3018,
      "step": 2372
    },
    {
      "epoch": 0.607488,
      "grad_norm": 0.3657882477273965,
      "learning_rate": 1.3612907224088665e-05,
      "loss": 1.3251,
      "step": 2373
    },
    {
      "epoch": 0.607744,
      "grad_norm": 0.3629726611169219,
      "learning_rate": 1.3597507968801834e-05,
      "loss": 1.3011,
      "step": 2374
    },
    {
      "epoch": 0.608,
      "grad_norm": 0.37218339996723465,
      "learning_rate": 1.3582112941418531e-05,
      "loss": 1.3177,
      "step": 2375
    },
    {
      "epoch": 0.608256,
      "grad_norm": 0.36714667901070663,
      "learning_rate": 1.3566722152104897e-05,
      "loss": 1.3316,
      "step": 2376
    },
    {
      "epoch": 0.608512,
      "grad_norm": 0.3696568545783475,
      "learning_rate": 1.355133561102429e-05,
      "loss": 1.3337,
      "step": 2377
    },
    {
      "epoch": 0.608768,
      "grad_norm": 0.37907363729955995,
      "learning_rate": 1.353595332833725e-05,
      "loss": 1.3536,
      "step": 2378
    },
    {
      "epoch": 0.609024,
      "grad_norm": 0.40353173435338824,
      "learning_rate": 1.3520575314201524e-05,
      "loss": 1.3612,
      "step": 2379
    },
    {
      "epoch": 0.60928,
      "grad_norm": 0.3838257373005245,
      "learning_rate": 1.3505201578772009e-05,
      "loss": 1.3037,
      "step": 2380
    },
    {
      "epoch": 0.609536,
      "grad_norm": 0.3717807734805008,
      "learning_rate": 1.3489832132200804e-05,
      "loss": 1.3915,
      "step": 2381
    },
    {
      "epoch": 0.609792,
      "grad_norm": 0.36629125595214146,
      "learning_rate": 1.347446698463716e-05,
      "loss": 1.3088,
      "step": 2382
    },
    {
      "epoch": 0.610048,
      "grad_norm": 0.3735512860638466,
      "learning_rate": 1.34591061462275e-05,
      "loss": 1.3407,
      "step": 2383
    },
    {
      "epoch": 0.610304,
      "grad_norm": 0.3791904421149582,
      "learning_rate": 1.3443749627115397e-05,
      "loss": 1.3676,
      "step": 2384
    },
    {
      "epoch": 0.61056,
      "grad_norm": 0.44888286869136595,
      "learning_rate": 1.3428397437441573e-05,
      "loss": 1.3701,
      "step": 2385
    },
    {
      "epoch": 0.610816,
      "grad_norm": 0.4051327501559484,
      "learning_rate": 1.341304958734388e-05,
      "loss": 1.3546,
      "step": 2386
    },
    {
      "epoch": 0.611072,
      "grad_norm": 0.3692572990142681,
      "learning_rate": 1.3397706086957328e-05,
      "loss": 1.3407,
      "step": 2387
    },
    {
      "epoch": 0.611328,
      "grad_norm": 0.3709398849271903,
      "learning_rate": 1.3382366946414023e-05,
      "loss": 1.3157,
      "step": 2388
    },
    {
      "epoch": 0.611584,
      "grad_norm": 0.4380538631303332,
      "learning_rate": 1.336703217584322e-05,
      "loss": 1.3468,
      "step": 2389
    },
    {
      "epoch": 0.61184,
      "grad_norm": 0.37058405000496025,
      "learning_rate": 1.3351701785371275e-05,
      "loss": 1.3113,
      "step": 2390
    },
    {
      "epoch": 0.612096,
      "grad_norm": 0.3759052796556128,
      "learning_rate": 1.3336375785121667e-05,
      "loss": 1.3478,
      "step": 2391
    },
    {
      "epoch": 0.612352,
      "grad_norm": 0.6591295868015719,
      "learning_rate": 1.3321054185214941e-05,
      "loss": 1.4023,
      "step": 2392
    },
    {
      "epoch": 0.612608,
      "grad_norm": 0.366612370193042,
      "learning_rate": 1.3305736995768774e-05,
      "loss": 1.3113,
      "step": 2393
    },
    {
      "epoch": 0.612864,
      "grad_norm": 0.36778332470740066,
      "learning_rate": 1.3290424226897907e-05,
      "loss": 1.3419,
      "step": 2394
    },
    {
      "epoch": 0.61312,
      "grad_norm": 0.3816875684197744,
      "learning_rate": 1.327511588871417e-05,
      "loss": 1.3816,
      "step": 2395
    },
    {
      "epoch": 0.613376,
      "grad_norm": 0.37590865401895246,
      "learning_rate": 1.3259811991326472e-05,
      "loss": 1.3036,
      "step": 2396
    },
    {
      "epoch": 0.613632,
      "grad_norm": 0.3863213964953608,
      "learning_rate": 1.324451254484079e-05,
      "loss": 1.3585,
      "step": 2397
    },
    {
      "epoch": 0.613888,
      "grad_norm": 0.3997095797953172,
      "learning_rate": 1.3229217559360143e-05,
      "loss": 1.3528,
      "step": 2398
    },
    {
      "epoch": 0.614144,
      "grad_norm": 0.5150809974820241,
      "learning_rate": 1.3213927044984622e-05,
      "loss": 1.3494,
      "step": 2399
    },
    {
      "epoch": 0.6144,
      "grad_norm": 0.3812743969443645,
      "learning_rate": 1.319864101181136e-05,
      "loss": 1.364,
      "step": 2400
    },
    {
      "epoch": 0.614656,
      "grad_norm": 0.38039188150094044,
      "learning_rate": 1.318335946993453e-05,
      "loss": 1.3237,
      "step": 2401
    },
    {
      "epoch": 0.614912,
      "grad_norm": 0.3789602190113306,
      "learning_rate": 1.3168082429445348e-05,
      "loss": 1.356,
      "step": 2402
    },
    {
      "epoch": 0.615168,
      "grad_norm": 0.3814043160420131,
      "learning_rate": 1.3152809900432058e-05,
      "loss": 1.3067,
      "step": 2403
    },
    {
      "epoch": 0.615424,
      "grad_norm": 0.40827277514333093,
      "learning_rate": 1.3137541892979897e-05,
      "loss": 1.3554,
      "step": 2404
    },
    {
      "epoch": 0.61568,
      "grad_norm": 0.3757701933025881,
      "learning_rate": 1.3122278417171144e-05,
      "loss": 1.3321,
      "step": 2405
    },
    {
      "epoch": 0.615936,
      "grad_norm": 0.3750020847378091,
      "learning_rate": 1.3107019483085077e-05,
      "loss": 1.3395,
      "step": 2406
    },
    {
      "epoch": 0.616192,
      "grad_norm": 0.3978653529305366,
      "learning_rate": 1.3091765100797981e-05,
      "loss": 1.383,
      "step": 2407
    },
    {
      "epoch": 0.616448,
      "grad_norm": 0.37417484241798354,
      "learning_rate": 1.307651528038313e-05,
      "loss": 1.322,
      "step": 2408
    },
    {
      "epoch": 0.616704,
      "grad_norm": 0.387357799174085,
      "learning_rate": 1.3061270031910787e-05,
      "loss": 1.3924,
      "step": 2409
    },
    {
      "epoch": 0.61696,
      "grad_norm": 0.37612256373040276,
      "learning_rate": 1.3046029365448187e-05,
      "loss": 1.321,
      "step": 2410
    },
    {
      "epoch": 0.617216,
      "grad_norm": 0.38176111821312564,
      "learning_rate": 1.3030793291059545e-05,
      "loss": 1.3785,
      "step": 2411
    },
    {
      "epoch": 0.617472,
      "grad_norm": 0.39034776712012653,
      "learning_rate": 1.3015561818806055e-05,
      "loss": 1.3771,
      "step": 2412
    },
    {
      "epoch": 0.617728,
      "grad_norm": 0.38693605966312683,
      "learning_rate": 1.3000334958745856e-05,
      "loss": 1.3389,
      "step": 2413
    },
    {
      "epoch": 0.617984,
      "grad_norm": 0.37293516162162627,
      "learning_rate": 1.298511272093405e-05,
      "loss": 1.3434,
      "step": 2414
    },
    {
      "epoch": 0.61824,
      "grad_norm": 0.37014206659438237,
      "learning_rate": 1.2969895115422682e-05,
      "loss": 1.3783,
      "step": 2415
    },
    {
      "epoch": 0.618496,
      "grad_norm": 0.40638163500763214,
      "learning_rate": 1.2954682152260751e-05,
      "loss": 1.3487,
      "step": 2416
    },
    {
      "epoch": 0.618752,
      "grad_norm": 0.38528768815849423,
      "learning_rate": 1.2939473841494163e-05,
      "loss": 1.3774,
      "step": 2417
    },
    {
      "epoch": 0.619008,
      "grad_norm": 0.3645233473274598,
      "learning_rate": 1.2924270193165778e-05,
      "loss": 1.3145,
      "step": 2418
    },
    {
      "epoch": 0.619264,
      "grad_norm": 0.3813598552265583,
      "learning_rate": 1.2909071217315363e-05,
      "loss": 1.3623,
      "step": 2419
    },
    {
      "epoch": 0.61952,
      "grad_norm": 0.3715443213762051,
      "learning_rate": 1.2893876923979614e-05,
      "loss": 1.3497,
      "step": 2420
    },
    {
      "epoch": 0.619776,
      "grad_norm": 0.3684889192753593,
      "learning_rate": 1.2878687323192111e-05,
      "loss": 1.325,
      "step": 2421
    },
    {
      "epoch": 0.620032,
      "grad_norm": 0.3626529713744613,
      "learning_rate": 1.2863502424983369e-05,
      "loss": 1.3325,
      "step": 2422
    },
    {
      "epoch": 0.620288,
      "grad_norm": 0.3744268581448939,
      "learning_rate": 1.2848322239380758e-05,
      "loss": 1.3326,
      "step": 2423
    },
    {
      "epoch": 0.620544,
      "grad_norm": 0.3714715977246351,
      "learning_rate": 1.2833146776408563e-05,
      "loss": 1.3554,
      "step": 2424
    },
    {
      "epoch": 0.6208,
      "grad_norm": 0.3669863015274799,
      "learning_rate": 1.2817976046087943e-05,
      "loss": 1.339,
      "step": 2425
    },
    {
      "epoch": 0.621056,
      "grad_norm": 0.3845944202617204,
      "learning_rate": 1.2802810058436938e-05,
      "loss": 1.3947,
      "step": 2426
    },
    {
      "epoch": 0.621312,
      "grad_norm": 0.38195003261765537,
      "learning_rate": 1.2787648823470443e-05,
      "loss": 1.3138,
      "step": 2427
    },
    {
      "epoch": 0.621568,
      "grad_norm": 0.4801487254183432,
      "learning_rate": 1.2772492351200234e-05,
      "loss": 1.3308,
      "step": 2428
    },
    {
      "epoch": 0.621824,
      "grad_norm": 0.36456717822357065,
      "learning_rate": 1.2757340651634911e-05,
      "loss": 1.3421,
      "step": 2429
    },
    {
      "epoch": 0.62208,
      "grad_norm": 0.36694776482222324,
      "learning_rate": 1.274219373477995e-05,
      "loss": 1.3473,
      "step": 2430
    },
    {
      "epoch": 0.622336,
      "grad_norm": 0.35975667076551615,
      "learning_rate": 1.2727051610637666e-05,
      "loss": 1.3024,
      "step": 2431
    },
    {
      "epoch": 0.622592,
      "grad_norm": 0.36701367134315854,
      "learning_rate": 1.2711914289207195e-05,
      "loss": 1.3302,
      "step": 2432
    },
    {
      "epoch": 0.622848,
      "grad_norm": 0.3621491665612143,
      "learning_rate": 1.2696781780484513e-05,
      "loss": 1.343,
      "step": 2433
    },
    {
      "epoch": 0.623104,
      "grad_norm": 0.3709943712731596,
      "learning_rate": 1.268165409446242e-05,
      "loss": 1.3832,
      "step": 2434
    },
    {
      "epoch": 0.62336,
      "grad_norm": 0.3683479178811537,
      "learning_rate": 1.2666531241130513e-05,
      "loss": 1.3692,
      "step": 2435
    },
    {
      "epoch": 0.623616,
      "grad_norm": 0.374270580812587,
      "learning_rate": 1.265141323047522e-05,
      "loss": 1.3404,
      "step": 2436
    },
    {
      "epoch": 0.623872,
      "grad_norm": 0.36712350819584494,
      "learning_rate": 1.2636300072479758e-05,
      "loss": 1.3256,
      "step": 2437
    },
    {
      "epoch": 0.624128,
      "grad_norm": 0.370980076661176,
      "learning_rate": 1.2621191777124144e-05,
      "loss": 1.39,
      "step": 2438
    },
    {
      "epoch": 0.624384,
      "grad_norm": 0.37362924252463325,
      "learning_rate": 1.260608835438518e-05,
      "loss": 1.3666,
      "step": 2439
    },
    {
      "epoch": 0.62464,
      "grad_norm": 0.37500140491616984,
      "learning_rate": 1.2590989814236467e-05,
      "loss": 1.3466,
      "step": 2440
    },
    {
      "epoch": 0.624896,
      "grad_norm": 0.3694879866305653,
      "learning_rate": 1.2575896166648349e-05,
      "loss": 1.3569,
      "step": 2441
    },
    {
      "epoch": 0.625152,
      "grad_norm": 0.3588799700443844,
      "learning_rate": 1.2560807421587967e-05,
      "loss": 1.3271,
      "step": 2442
    },
    {
      "epoch": 0.625408,
      "grad_norm": 0.3699283775171232,
      "learning_rate": 1.254572358901922e-05,
      "loss": 1.3551,
      "step": 2443
    },
    {
      "epoch": 0.625664,
      "grad_norm": 0.3750788075269564,
      "learning_rate": 1.2530644678902752e-05,
      "loss": 1.3554,
      "step": 2444
    },
    {
      "epoch": 0.62592,
      "grad_norm": 0.37021240652505016,
      "learning_rate": 1.251557070119597e-05,
      "loss": 1.3747,
      "step": 2445
    },
    {
      "epoch": 0.626176,
      "grad_norm": 0.37318762849645426,
      "learning_rate": 1.2500501665853016e-05,
      "loss": 1.3243,
      "step": 2446
    },
    {
      "epoch": 0.626432,
      "grad_norm": 0.3774553720827331,
      "learning_rate": 1.2485437582824764e-05,
      "loss": 1.3807,
      "step": 2447
    },
    {
      "epoch": 0.626688,
      "grad_norm": 0.36914543877060885,
      "learning_rate": 1.2470378462058826e-05,
      "loss": 1.3534,
      "step": 2448
    },
    {
      "epoch": 0.626944,
      "grad_norm": 0.3659265243042741,
      "learning_rate": 1.245532431349954e-05,
      "loss": 1.3517,
      "step": 2449
    },
    {
      "epoch": 0.6272,
      "grad_norm": 0.37556825176104697,
      "learning_rate": 1.2440275147087947e-05,
      "loss": 1.3744,
      "step": 2450
    },
    {
      "epoch": 0.627456,
      "grad_norm": 0.3780268240331811,
      "learning_rate": 1.2425230972761808e-05,
      "loss": 1.3666,
      "step": 2451
    },
    {
      "epoch": 0.627712,
      "grad_norm": 0.35378470476343993,
      "learning_rate": 1.24101918004556e-05,
      "loss": 1.3044,
      "step": 2452
    },
    {
      "epoch": 0.627968,
      "grad_norm": 0.35438232422028915,
      "learning_rate": 1.2395157640100459e-05,
      "loss": 1.2613,
      "step": 2453
    },
    {
      "epoch": 0.628224,
      "grad_norm": 0.48253293345715553,
      "learning_rate": 1.2380128501624248e-05,
      "loss": 1.3316,
      "step": 2454
    },
    {
      "epoch": 0.62848,
      "grad_norm": 0.35982381292033405,
      "learning_rate": 1.2365104394951498e-05,
      "loss": 1.3177,
      "step": 2455
    },
    {
      "epoch": 0.628736,
      "grad_norm": 0.36793445905793276,
      "learning_rate": 1.2350085330003419e-05,
      "loss": 1.3284,
      "step": 2456
    },
    {
      "epoch": 0.628992,
      "grad_norm": 0.37712446631785285,
      "learning_rate": 1.2335071316697895e-05,
      "loss": 1.3464,
      "step": 2457
    },
    {
      "epoch": 0.629248,
      "grad_norm": 0.36849438037885773,
      "learning_rate": 1.2320062364949478e-05,
      "loss": 1.3737,
      "step": 2458
    },
    {
      "epoch": 0.629504,
      "grad_norm": 0.3691787225149286,
      "learning_rate": 1.2305058484669356e-05,
      "loss": 1.3795,
      "step": 2459
    },
    {
      "epoch": 0.62976,
      "grad_norm": 0.37579884824642285,
      "learning_rate": 1.2290059685765395e-05,
      "loss": 1.3317,
      "step": 2460
    },
    {
      "epoch": 0.630016,
      "grad_norm": 0.3676521605019993,
      "learning_rate": 1.2275065978142089e-05,
      "loss": 1.3408,
      "step": 2461
    },
    {
      "epoch": 0.630272,
      "grad_norm": 0.373238026670823,
      "learning_rate": 1.2260077371700572e-05,
      "loss": 1.3484,
      "step": 2462
    },
    {
      "epoch": 0.630528,
      "grad_norm": 0.3779606872881511,
      "learning_rate": 1.2245093876338618e-05,
      "loss": 1.3448,
      "step": 2463
    },
    {
      "epoch": 0.630784,
      "grad_norm": 0.36882533051104166,
      "learning_rate": 1.2230115501950626e-05,
      "loss": 1.3559,
      "step": 2464
    },
    {
      "epoch": 0.63104,
      "grad_norm": 0.38219081221027207,
      "learning_rate": 1.221514225842759e-05,
      "loss": 1.3727,
      "step": 2465
    },
    {
      "epoch": 0.631296,
      "grad_norm": 0.36692717807232483,
      "learning_rate": 1.2200174155657147e-05,
      "loss": 1.2875,
      "step": 2466
    },
    {
      "epoch": 0.631552,
      "grad_norm": 0.36256590798081884,
      "learning_rate": 1.2185211203523521e-05,
      "loss": 1.3075,
      "step": 2467
    },
    {
      "epoch": 0.631808,
      "grad_norm": 0.37238053244652786,
      "learning_rate": 1.2170253411907536e-05,
      "loss": 1.3216,
      "step": 2468
    },
    {
      "epoch": 0.632064,
      "grad_norm": 0.3803700367181227,
      "learning_rate": 1.2155300790686617e-05,
      "loss": 1.3568,
      "step": 2469
    },
    {
      "epoch": 0.63232,
      "grad_norm": 0.3699879598789386,
      "learning_rate": 1.2140353349734776e-05,
      "loss": 1.3665,
      "step": 2470
    },
    {
      "epoch": 0.632576,
      "grad_norm": 0.3665872620915169,
      "learning_rate": 1.212541109892258e-05,
      "loss": 1.3672,
      "step": 2471
    },
    {
      "epoch": 0.632832,
      "grad_norm": 0.6248408615273903,
      "learning_rate": 1.2110474048117198e-05,
      "loss": 1.4082,
      "step": 2472
    },
    {
      "epoch": 0.633088,
      "grad_norm": 0.35987766608041954,
      "learning_rate": 1.2095542207182348e-05,
      "loss": 1.3281,
      "step": 2473
    },
    {
      "epoch": 0.633344,
      "grad_norm": 0.4225789733224598,
      "learning_rate": 1.2080615585978315e-05,
      "loss": 1.4126,
      "step": 2474
    },
    {
      "epoch": 0.6336,
      "grad_norm": 0.37744910494769535,
      "learning_rate": 1.2065694194361934e-05,
      "loss": 1.3688,
      "step": 2475
    },
    {
      "epoch": 0.633856,
      "grad_norm": 0.36496449774197853,
      "learning_rate": 1.2050778042186594e-05,
      "loss": 1.3437,
      "step": 2476
    },
    {
      "epoch": 0.634112,
      "grad_norm": 0.36376727878783244,
      "learning_rate": 1.2035867139302218e-05,
      "loss": 1.3301,
      "step": 2477
    },
    {
      "epoch": 0.634368,
      "grad_norm": 0.3705950905186145,
      "learning_rate": 1.2020961495555247e-05,
      "loss": 1.3528,
      "step": 2478
    },
    {
      "epoch": 0.634624,
      "grad_norm": 0.36750560593131604,
      "learning_rate": 1.2006061120788677e-05,
      "loss": 1.3343,
      "step": 2479
    },
    {
      "epoch": 0.63488,
      "grad_norm": 0.3701806907697241,
      "learning_rate": 1.199116602484201e-05,
      "loss": 1.3635,
      "step": 2480
    },
    {
      "epoch": 0.635136,
      "grad_norm": 0.4250064284134545,
      "learning_rate": 1.1976276217551268e-05,
      "loss": 1.3371,
      "step": 2481
    },
    {
      "epoch": 0.635392,
      "grad_norm": 0.37521588071779033,
      "learning_rate": 1.1961391708748966e-05,
      "loss": 1.353,
      "step": 2482
    },
    {
      "epoch": 0.635648,
      "grad_norm": 0.366321107588632,
      "learning_rate": 1.1946512508264152e-05,
      "loss": 1.3123,
      "step": 2483
    },
    {
      "epoch": 0.635904,
      "grad_norm": 0.36953132595669375,
      "learning_rate": 1.1931638625922322e-05,
      "loss": 1.3345,
      "step": 2484
    },
    {
      "epoch": 0.63616,
      "grad_norm": 0.38256769973430976,
      "learning_rate": 1.1916770071545499e-05,
      "loss": 1.3771,
      "step": 2485
    },
    {
      "epoch": 0.636416,
      "grad_norm": 0.3795423361075958,
      "learning_rate": 1.1901906854952172e-05,
      "loss": 1.3453,
      "step": 2486
    },
    {
      "epoch": 0.636672,
      "grad_norm": 0.3718747294220241,
      "learning_rate": 1.1887048985957312e-05,
      "loss": 1.3418,
      "step": 2487
    },
    {
      "epoch": 0.636928,
      "grad_norm": 0.3687617976520658,
      "learning_rate": 1.187219647437235e-05,
      "loss": 1.3237,
      "step": 2488
    },
    {
      "epoch": 0.637184,
      "grad_norm": 0.37306731208951827,
      "learning_rate": 1.1857349330005193e-05,
      "loss": 1.3491,
      "step": 2489
    },
    {
      "epoch": 0.63744,
      "grad_norm": 0.36446125738551793,
      "learning_rate": 1.1842507562660176e-05,
      "loss": 1.3229,
      "step": 2490
    },
    {
      "epoch": 0.637696,
      "grad_norm": 0.38058161988826456,
      "learning_rate": 1.1827671182138114e-05,
      "loss": 1.3966,
      "step": 2491
    },
    {
      "epoch": 0.637952,
      "grad_norm": 0.3615258209781036,
      "learning_rate": 1.1812840198236248e-05,
      "loss": 1.3174,
      "step": 2492
    },
    {
      "epoch": 0.638208,
      "grad_norm": 0.3660253994651249,
      "learning_rate": 1.1798014620748266e-05,
      "loss": 1.3457,
      "step": 2493
    },
    {
      "epoch": 0.638464,
      "grad_norm": 0.38180918419106485,
      "learning_rate": 1.1783194459464272e-05,
      "loss": 1.3484,
      "step": 2494
    },
    {
      "epoch": 0.63872,
      "grad_norm": 0.36764697921499084,
      "learning_rate": 1.1768379724170811e-05,
      "loss": 1.333,
      "step": 2495
    },
    {
      "epoch": 0.638976,
      "grad_norm": 0.37414000131425107,
      "learning_rate": 1.1753570424650822e-05,
      "loss": 1.382,
      "step": 2496
    },
    {
      "epoch": 0.639232,
      "grad_norm": 0.3757510701645589,
      "learning_rate": 1.1738766570683673e-05,
      "loss": 1.3089,
      "step": 2497
    },
    {
      "epoch": 0.639488,
      "grad_norm": 0.36368832775713555,
      "learning_rate": 1.172396817204513e-05,
      "loss": 1.3409,
      "step": 2498
    },
    {
      "epoch": 0.639744,
      "grad_norm": 0.38513835531869484,
      "learning_rate": 1.1709175238507364e-05,
      "loss": 1.3399,
      "step": 2499
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3632332495589021,
      "learning_rate": 1.169438777983892e-05,
      "loss": 1.2926,
      "step": 2500
    },
    {
      "epoch": 0.640256,
      "grad_norm": 0.37079684532784696,
      "learning_rate": 1.1679605805804748e-05,
      "loss": 1.3431,
      "step": 2501
    },
    {
      "epoch": 0.640512,
      "grad_norm": 0.38943015771695316,
      "learning_rate": 1.1664829326166154e-05,
      "loss": 1.3706,
      "step": 2502
    },
    {
      "epoch": 0.640768,
      "grad_norm": 0.39794729038396065,
      "learning_rate": 1.1650058350680831e-05,
      "loss": 1.3487,
      "step": 2503
    },
    {
      "epoch": 0.641024,
      "grad_norm": 0.37976486635730594,
      "learning_rate": 1.1635292889102834e-05,
      "loss": 1.3599,
      "step": 2504
    },
    {
      "epoch": 0.64128,
      "grad_norm": 0.3697016444976321,
      "learning_rate": 1.1620532951182584e-05,
      "loss": 1.3526,
      "step": 2505
    },
    {
      "epoch": 0.641536,
      "grad_norm": 0.3610187394753097,
      "learning_rate": 1.1605778546666846e-05,
      "loss": 1.287,
      "step": 2506
    },
    {
      "epoch": 0.641792,
      "grad_norm": 0.3663719272650304,
      "learning_rate": 1.1591029685298725e-05,
      "loss": 1.351,
      "step": 2507
    },
    {
      "epoch": 0.642048,
      "grad_norm": 0.3721606211988694,
      "learning_rate": 1.1576286376817682e-05,
      "loss": 1.3599,
      "step": 2508
    },
    {
      "epoch": 0.642304,
      "grad_norm": 0.36786691504653046,
      "learning_rate": 1.156154863095949e-05,
      "loss": 1.3467,
      "step": 2509
    },
    {
      "epoch": 0.64256,
      "grad_norm": 0.3698288625287084,
      "learning_rate": 1.1546816457456265e-05,
      "loss": 1.3285,
      "step": 2510
    },
    {
      "epoch": 0.642816,
      "grad_norm": 0.36945551749245975,
      "learning_rate": 1.1532089866036442e-05,
      "loss": 1.3281,
      "step": 2511
    },
    {
      "epoch": 0.643072,
      "grad_norm": 0.3741585770325421,
      "learning_rate": 1.1517368866424767e-05,
      "loss": 1.3665,
      "step": 2512
    },
    {
      "epoch": 0.643328,
      "grad_norm": 0.36202612951993735,
      "learning_rate": 1.15026534683423e-05,
      "loss": 1.3007,
      "step": 2513
    },
    {
      "epoch": 0.643584,
      "grad_norm": 0.37860722875961206,
      "learning_rate": 1.1487943681506382e-05,
      "loss": 1.3333,
      "step": 2514
    },
    {
      "epoch": 0.64384,
      "grad_norm": 0.36328215072216624,
      "learning_rate": 1.1473239515630663e-05,
      "loss": 1.2957,
      "step": 2515
    },
    {
      "epoch": 0.644096,
      "grad_norm": 0.36936753269892786,
      "learning_rate": 1.1458540980425088e-05,
      "loss": 1.3152,
      "step": 2516
    },
    {
      "epoch": 0.644352,
      "grad_norm": 0.3618079974800423,
      "learning_rate": 1.144384808559587e-05,
      "loss": 1.3456,
      "step": 2517
    },
    {
      "epoch": 0.644608,
      "grad_norm": 0.36632031994452485,
      "learning_rate": 1.1429160840845515e-05,
      "loss": 1.354,
      "step": 2518
    },
    {
      "epoch": 0.644864,
      "grad_norm": 0.3632247198297989,
      "learning_rate": 1.141447925587277e-05,
      "loss": 1.3367,
      "step": 2519
    },
    {
      "epoch": 0.64512,
      "grad_norm": 0.36619419224537886,
      "learning_rate": 1.1399803340372672e-05,
      "loss": 1.334,
      "step": 2520
    },
    {
      "epoch": 0.645376,
      "grad_norm": 0.3739425129979195,
      "learning_rate": 1.1385133104036497e-05,
      "loss": 1.3497,
      "step": 2521
    },
    {
      "epoch": 0.645632,
      "grad_norm": 0.37499884738984035,
      "learning_rate": 1.1370468556551773e-05,
      "loss": 1.3819,
      "step": 2522
    },
    {
      "epoch": 0.645888,
      "grad_norm": 0.4727524029447252,
      "learning_rate": 1.1355809707602278e-05,
      "loss": 1.3388,
      "step": 2523
    },
    {
      "epoch": 0.646144,
      "grad_norm": 0.3662282274608861,
      "learning_rate": 1.1341156566868032e-05,
      "loss": 1.3192,
      "step": 2524
    },
    {
      "epoch": 0.6464,
      "grad_norm": 0.3686639498506288,
      "learning_rate": 1.1326509144025275e-05,
      "loss": 1.3473,
      "step": 2525
    },
    {
      "epoch": 0.646656,
      "grad_norm": 0.36795463671361095,
      "learning_rate": 1.1311867448746464e-05,
      "loss": 1.3234,
      "step": 2526
    },
    {
      "epoch": 0.646912,
      "grad_norm": 0.37243344052142013,
      "learning_rate": 1.1297231490700287e-05,
      "loss": 1.3501,
      "step": 2527
    },
    {
      "epoch": 0.647168,
      "grad_norm": 0.37367443842724596,
      "learning_rate": 1.1282601279551644e-05,
      "loss": 1.3368,
      "step": 2528
    },
    {
      "epoch": 0.647424,
      "grad_norm": 0.36441216204962235,
      "learning_rate": 1.1267976824961644e-05,
      "loss": 1.3246,
      "step": 2529
    },
    {
      "epoch": 0.64768,
      "grad_norm": 0.37130674618671683,
      "learning_rate": 1.1253358136587566e-05,
      "loss": 1.3336,
      "step": 2530
    },
    {
      "epoch": 0.647936,
      "grad_norm": 0.3625541384602983,
      "learning_rate": 1.1238745224082921e-05,
      "loss": 1.3144,
      "step": 2531
    },
    {
      "epoch": 0.648192,
      "grad_norm": 0.3631736161580889,
      "learning_rate": 1.1224138097097371e-05,
      "loss": 1.3599,
      "step": 2532
    },
    {
      "epoch": 0.648448,
      "grad_norm": 0.3700356782965741,
      "learning_rate": 1.120953676527678e-05,
      "loss": 1.3565,
      "step": 2533
    },
    {
      "epoch": 0.648704,
      "grad_norm": 0.37945403154520446,
      "learning_rate": 1.1194941238263177e-05,
      "loss": 1.2984,
      "step": 2534
    },
    {
      "epoch": 0.64896,
      "grad_norm": 0.36933115391430277,
      "learning_rate": 1.1180351525694761e-05,
      "loss": 1.3585,
      "step": 2535
    },
    {
      "epoch": 0.649216,
      "grad_norm": 0.36729710913029534,
      "learning_rate": 1.1165767637205894e-05,
      "loss": 1.3458,
      "step": 2536
    },
    {
      "epoch": 0.649472,
      "grad_norm": 0.3720520281340042,
      "learning_rate": 1.1151189582427077e-05,
      "loss": 1.3554,
      "step": 2537
    },
    {
      "epoch": 0.649728,
      "grad_norm": 0.38327392000773447,
      "learning_rate": 1.1136617370984984e-05,
      "loss": 1.3424,
      "step": 2538
    },
    {
      "epoch": 0.649984,
      "grad_norm": 0.3650362121886063,
      "learning_rate": 1.1122051012502402e-05,
      "loss": 1.3282,
      "step": 2539
    },
    {
      "epoch": 0.65024,
      "grad_norm": 0.371719709182339,
      "learning_rate": 1.1107490516598267e-05,
      "loss": 1.3516,
      "step": 2540
    },
    {
      "epoch": 0.650496,
      "grad_norm": 0.3586628438388398,
      "learning_rate": 1.1092935892887654e-05,
      "loss": 1.3298,
      "step": 2541
    },
    {
      "epoch": 0.650752,
      "grad_norm": 0.3719561341281398,
      "learning_rate": 1.1078387150981742e-05,
      "loss": 1.3313,
      "step": 2542
    },
    {
      "epoch": 0.651008,
      "grad_norm": 0.3727239166934467,
      "learning_rate": 1.106384430048783e-05,
      "loss": 1.3521,
      "step": 2543
    },
    {
      "epoch": 0.651264,
      "grad_norm": 0.3785858866751082,
      "learning_rate": 1.1049307351009342e-05,
      "loss": 1.3529,
      "step": 2544
    },
    {
      "epoch": 0.65152,
      "grad_norm": 0.3699255170251369,
      "learning_rate": 1.1034776312145779e-05,
      "loss": 1.3264,
      "step": 2545
    },
    {
      "epoch": 0.651776,
      "grad_norm": 0.370841786556642,
      "learning_rate": 1.1020251193492757e-05,
      "loss": 1.3455,
      "step": 2546
    },
    {
      "epoch": 0.652032,
      "grad_norm": 0.38493894578772453,
      "learning_rate": 1.1005732004641984e-05,
      "loss": 1.3635,
      "step": 2547
    },
    {
      "epoch": 0.652288,
      "grad_norm": 0.3688671495603611,
      "learning_rate": 1.0991218755181243e-05,
      "loss": 1.3832,
      "step": 2548
    },
    {
      "epoch": 0.652544,
      "grad_norm": 0.37302506509268296,
      "learning_rate": 1.0976711454694402e-05,
      "loss": 1.3361,
      "step": 2549
    },
    {
      "epoch": 0.6528,
      "grad_norm": 0.37125101625358176,
      "learning_rate": 1.0962210112761402e-05,
      "loss": 1.3177,
      "step": 2550
    },
    {
      "epoch": 0.653056,
      "grad_norm": 0.3688831866811569,
      "learning_rate": 1.0947714738958233e-05,
      "loss": 1.3671,
      "step": 2551
    },
    {
      "epoch": 0.653312,
      "grad_norm": 0.3680758547705766,
      "learning_rate": 1.0933225342856966e-05,
      "loss": 1.3123,
      "step": 2552
    },
    {
      "epoch": 0.653568,
      "grad_norm": 0.3580660580460009,
      "learning_rate": 1.0918741934025702e-05,
      "loss": 1.3228,
      "step": 2553
    },
    {
      "epoch": 0.653824,
      "grad_norm": 0.36853007839695645,
      "learning_rate": 1.090426452202861e-05,
      "loss": 1.329,
      "step": 2554
    },
    {
      "epoch": 0.65408,
      "grad_norm": 0.3774580085406426,
      "learning_rate": 1.0889793116425884e-05,
      "loss": 1.3878,
      "step": 2555
    },
    {
      "epoch": 0.654336,
      "grad_norm": 0.3633518154575558,
      "learning_rate": 1.0875327726773772e-05,
      "loss": 1.3181,
      "step": 2556
    },
    {
      "epoch": 0.654592,
      "grad_norm": 0.37751466255702537,
      "learning_rate": 1.0860868362624516e-05,
      "loss": 1.3469,
      "step": 2557
    },
    {
      "epoch": 0.654848,
      "grad_norm": 0.3788976851452432,
      "learning_rate": 1.08464150335264e-05,
      "loss": 1.3505,
      "step": 2558
    },
    {
      "epoch": 0.655104,
      "grad_norm": 0.38121911947790077,
      "learning_rate": 1.083196774902373e-05,
      "loss": 1.3783,
      "step": 2559
    },
    {
      "epoch": 0.65536,
      "grad_norm": 0.3670954629671705,
      "learning_rate": 1.0817526518656802e-05,
      "loss": 1.3488,
      "step": 2560
    },
    {
      "epoch": 0.655616,
      "grad_norm": 0.37010883596481103,
      "learning_rate": 1.0803091351961927e-05,
      "loss": 1.3305,
      "step": 2561
    },
    {
      "epoch": 0.655872,
      "grad_norm": 0.3713332390391647,
      "learning_rate": 1.0788662258471418e-05,
      "loss": 1.3636,
      "step": 2562
    },
    {
      "epoch": 0.656128,
      "grad_norm": 0.37981586637855286,
      "learning_rate": 1.0774239247713546e-05,
      "loss": 1.3667,
      "step": 2563
    },
    {
      "epoch": 0.656384,
      "grad_norm": 0.3711320649142627,
      "learning_rate": 1.0759822329212608e-05,
      "loss": 1.3573,
      "step": 2564
    },
    {
      "epoch": 0.65664,
      "grad_norm": 0.4434392626812667,
      "learning_rate": 1.0745411512488835e-05,
      "loss": 1.3154,
      "step": 2565
    },
    {
      "epoch": 0.656896,
      "grad_norm": 0.36822617722850787,
      "learning_rate": 1.0731006807058461e-05,
      "loss": 1.3204,
      "step": 2566
    },
    {
      "epoch": 0.657152,
      "grad_norm": 0.3735435377092979,
      "learning_rate": 1.0716608222433673e-05,
      "loss": 1.3833,
      "step": 2567
    },
    {
      "epoch": 0.657408,
      "grad_norm": 0.3687736873412092,
      "learning_rate": 1.0702215768122626e-05,
      "loss": 1.3484,
      "step": 2568
    },
    {
      "epoch": 0.657664,
      "grad_norm": 0.3634430428751392,
      "learning_rate": 1.0687829453629402e-05,
      "loss": 1.3347,
      "step": 2569
    },
    {
      "epoch": 0.65792,
      "grad_norm": 0.36575876073850627,
      "learning_rate": 1.0673449288454047e-05,
      "loss": 1.3377,
      "step": 2570
    },
    {
      "epoch": 0.658176,
      "grad_norm": 0.36138016967844705,
      "learning_rate": 1.0659075282092549e-05,
      "loss": 1.3255,
      "step": 2571
    },
    {
      "epoch": 0.658432,
      "grad_norm": 0.5883643275874262,
      "learning_rate": 1.064470744403682e-05,
      "loss": 1.3528,
      "step": 2572
    },
    {
      "epoch": 0.658688,
      "grad_norm": 0.3781053920212659,
      "learning_rate": 1.0630345783774703e-05,
      "loss": 1.3638,
      "step": 2573
    },
    {
      "epoch": 0.658944,
      "grad_norm": 0.35916298018797127,
      "learning_rate": 1.061599031078997e-05,
      "loss": 1.3198,
      "step": 2574
    },
    {
      "epoch": 0.6592,
      "grad_norm": 0.3887816985229547,
      "learning_rate": 1.0601641034562284e-05,
      "loss": 1.3598,
      "step": 2575
    },
    {
      "epoch": 0.659456,
      "grad_norm": 0.3748928241617236,
      "learning_rate": 1.0587297964567228e-05,
      "loss": 1.3647,
      "step": 2576
    },
    {
      "epoch": 0.659712,
      "grad_norm": 0.3738669618812053,
      "learning_rate": 1.0572961110276299e-05,
      "loss": 1.3661,
      "step": 2577
    },
    {
      "epoch": 0.659968,
      "grad_norm": 0.46500249195164156,
      "learning_rate": 1.055863048115687e-05,
      "loss": 1.3853,
      "step": 2578
    },
    {
      "epoch": 0.660224,
      "grad_norm": 0.3776039888451268,
      "learning_rate": 1.0544306086672216e-05,
      "loss": 1.3343,
      "step": 2579
    },
    {
      "epoch": 0.66048,
      "grad_norm": 0.3807101545844821,
      "learning_rate": 1.05299879362815e-05,
      "loss": 1.3605,
      "step": 2580
    },
    {
      "epoch": 0.660736,
      "grad_norm": 0.3694774074729018,
      "learning_rate": 1.0515676039439737e-05,
      "loss": 1.3275,
      "step": 2581
    },
    {
      "epoch": 0.660992,
      "grad_norm": 0.537277442177915,
      "learning_rate": 1.0501370405597834e-05,
      "loss": 1.287,
      "step": 2582
    },
    {
      "epoch": 0.661248,
      "grad_norm": 0.36570689840417997,
      "learning_rate": 1.0487071044202559e-05,
      "loss": 1.3191,
      "step": 2583
    },
    {
      "epoch": 0.661504,
      "grad_norm": 0.3600435539908686,
      "learning_rate": 1.0472777964696533e-05,
      "loss": 1.3324,
      "step": 2584
    },
    {
      "epoch": 0.66176,
      "grad_norm": 0.37070217471806244,
      "learning_rate": 1.0458491176518238e-05,
      "loss": 1.3046,
      "step": 2585
    },
    {
      "epoch": 0.662016,
      "grad_norm": 0.3700480681943844,
      "learning_rate": 1.0444210689101982e-05,
      "loss": 1.3538,
      "step": 2586
    },
    {
      "epoch": 0.662272,
      "grad_norm": 0.3761803678188354,
      "learning_rate": 1.0429936511877936e-05,
      "loss": 1.364,
      "step": 2587
    },
    {
      "epoch": 0.662528,
      "grad_norm": 0.370711766744739,
      "learning_rate": 1.0415668654272084e-05,
      "loss": 1.369,
      "step": 2588
    },
    {
      "epoch": 0.662784,
      "grad_norm": 0.37494912916105905,
      "learning_rate": 1.0401407125706248e-05,
      "loss": 1.3679,
      "step": 2589
    },
    {
      "epoch": 0.66304,
      "grad_norm": 0.3807586583252639,
      "learning_rate": 1.038715193559807e-05,
      "loss": 1.3647,
      "step": 2590
    },
    {
      "epoch": 0.663296,
      "grad_norm": 0.36888881639562676,
      "learning_rate": 1.0372903093361005e-05,
      "loss": 1.3758,
      "step": 2591
    },
    {
      "epoch": 0.663552,
      "grad_norm": 0.3802422282877783,
      "learning_rate": 1.0358660608404323e-05,
      "loss": 1.3602,
      "step": 2592
    },
    {
      "epoch": 0.663808,
      "grad_norm": 0.3667329159317049,
      "learning_rate": 1.0344424490133073e-05,
      "loss": 1.3277,
      "step": 2593
    },
    {
      "epoch": 0.664064,
      "grad_norm": 0.3704556516556962,
      "learning_rate": 1.0330194747948124e-05,
      "loss": 1.3443,
      "step": 2594
    },
    {
      "epoch": 0.66432,
      "grad_norm": 0.35781349805789997,
      "learning_rate": 1.0315971391246124e-05,
      "loss": 1.2863,
      "step": 2595
    },
    {
      "epoch": 0.664576,
      "grad_norm": 0.3733009183878282,
      "learning_rate": 1.030175442941951e-05,
      "loss": 1.366,
      "step": 2596
    },
    {
      "epoch": 0.664832,
      "grad_norm": 0.3714855036540785,
      "learning_rate": 1.02875438718565e-05,
      "loss": 1.3282,
      "step": 2597
    },
    {
      "epoch": 0.665088,
      "grad_norm": 0.37577403707192186,
      "learning_rate": 1.0273339727941059e-05,
      "loss": 1.3676,
      "step": 2598
    },
    {
      "epoch": 0.665344,
      "grad_norm": 0.5424670131785854,
      "learning_rate": 1.0259142007052945e-05,
      "loss": 1.3357,
      "step": 2599
    },
    {
      "epoch": 0.6656,
      "grad_norm": 0.36799579590693093,
      "learning_rate": 1.024495071856765e-05,
      "loss": 1.3207,
      "step": 2600
    },
    {
      "epoch": 0.665856,
      "grad_norm": 0.3834979434969095,
      "learning_rate": 1.023076587185644e-05,
      "loss": 1.3836,
      "step": 2601
    },
    {
      "epoch": 0.666112,
      "grad_norm": 0.3883966855141535,
      "learning_rate": 1.0216587476286319e-05,
      "loss": 1.3939,
      "step": 2602
    },
    {
      "epoch": 0.666368,
      "grad_norm": 0.3703229154264701,
      "learning_rate": 1.0202415541220025e-05,
      "loss": 1.3663,
      "step": 2603
    },
    {
      "epoch": 0.666624,
      "grad_norm": 0.3694268353212482,
      "learning_rate": 1.0188250076016042e-05,
      "loss": 1.3276,
      "step": 2604
    },
    {
      "epoch": 0.66688,
      "grad_norm": 0.379663923230283,
      "learning_rate": 1.0174091090028577e-05,
      "loss": 1.3145,
      "step": 2605
    },
    {
      "epoch": 0.667136,
      "grad_norm": 0.3689204488906091,
      "learning_rate": 1.0159938592607543e-05,
      "loss": 1.3508,
      "step": 2606
    },
    {
      "epoch": 0.667392,
      "grad_norm": 0.3647690934389834,
      "learning_rate": 1.0145792593098589e-05,
      "loss": 1.3776,
      "step": 2607
    },
    {
      "epoch": 0.667648,
      "grad_norm": 0.37532380234612484,
      "learning_rate": 1.013165310084307e-05,
      "loss": 1.3116,
      "step": 2608
    },
    {
      "epoch": 0.667904,
      "grad_norm": 0.3703790134183054,
      "learning_rate": 1.0117520125178028e-05,
      "loss": 1.2854,
      "step": 2609
    },
    {
      "epoch": 0.66816,
      "grad_norm": 0.36367138986957337,
      "learning_rate": 1.010339367543622e-05,
      "loss": 1.3395,
      "step": 2610
    },
    {
      "epoch": 0.668416,
      "grad_norm": 0.3682736714356572,
      "learning_rate": 1.008927376094609e-05,
      "loss": 1.3446,
      "step": 2611
    },
    {
      "epoch": 0.668672,
      "grad_norm": 0.3822665533177629,
      "learning_rate": 1.0075160391031752e-05,
      "loss": 1.3465,
      "step": 2612
    },
    {
      "epoch": 0.668928,
      "grad_norm": 0.37983676001982464,
      "learning_rate": 1.0061053575013018e-05,
      "loss": 1.3687,
      "step": 2613
    },
    {
      "epoch": 0.669184,
      "grad_norm": 0.36198830719917674,
      "learning_rate": 1.004695332220536e-05,
      "loss": 1.321,
      "step": 2614
    },
    {
      "epoch": 0.66944,
      "grad_norm": 0.977659132777435,
      "learning_rate": 1.0032859641919919e-05,
      "loss": 1.3528,
      "step": 2615
    },
    {
      "epoch": 0.669696,
      "grad_norm": 0.365790615774564,
      "learning_rate": 1.00187725434635e-05,
      "loss": 1.3051,
      "step": 2616
    },
    {
      "epoch": 0.669952,
      "grad_norm": 0.3786600766705962,
      "learning_rate": 1.0004692036138563e-05,
      "loss": 1.3577,
      "step": 2617
    },
    {
      "epoch": 0.670208,
      "grad_norm": 0.39380255388215823,
      "learning_rate": 9.990618129243196e-06,
      "loss": 1.3066,
      "step": 2618
    },
    {
      "epoch": 0.670464,
      "grad_norm": 0.3859151004351654,
      "learning_rate": 9.976550832071147e-06,
      "loss": 1.3609,
      "step": 2619
    },
    {
      "epoch": 0.67072,
      "grad_norm": 0.3840296411750152,
      "learning_rate": 9.962490153911808e-06,
      "loss": 1.3455,
      "step": 2620
    },
    {
      "epoch": 0.670976,
      "grad_norm": 0.3756080929744528,
      "learning_rate": 9.94843610405017e-06,
      "loss": 1.293,
      "step": 2621
    },
    {
      "epoch": 0.671232,
      "grad_norm": 0.37502739976127286,
      "learning_rate": 9.934388691766871e-06,
      "loss": 1.3299,
      "step": 2622
    },
    {
      "epoch": 0.671488,
      "grad_norm": 0.3857547508060106,
      "learning_rate": 9.920347926338168e-06,
      "loss": 1.3686,
      "step": 2623
    },
    {
      "epoch": 0.671744,
      "grad_norm": 0.3798305117183366,
      "learning_rate": 9.906313817035901e-06,
      "loss": 1.3142,
      "step": 2624
    },
    {
      "epoch": 0.672,
      "grad_norm": 0.3800915820206502,
      "learning_rate": 9.892286373127544e-06,
      "loss": 1.3679,
      "step": 2625
    },
    {
      "epoch": 0.672256,
      "grad_norm": 0.3861312320713112,
      "learning_rate": 9.87826560387616e-06,
      "loss": 1.3243,
      "step": 2626
    },
    {
      "epoch": 0.672512,
      "grad_norm": 0.36582924559078434,
      "learning_rate": 9.8642515185404e-06,
      "loss": 1.3056,
      "step": 2627
    },
    {
      "epoch": 0.672768,
      "grad_norm": 0.40242192935836535,
      "learning_rate": 9.850244126374507e-06,
      "loss": 1.3158,
      "step": 2628
    },
    {
      "epoch": 0.673024,
      "grad_norm": 0.37892670999798633,
      "learning_rate": 9.836243436628308e-06,
      "loss": 1.378,
      "step": 2629
    },
    {
      "epoch": 0.67328,
      "grad_norm": 0.36650655072840554,
      "learning_rate": 9.822249458547181e-06,
      "loss": 1.3186,
      "step": 2630
    },
    {
      "epoch": 0.673536,
      "grad_norm": 0.38076781390993314,
      "learning_rate": 9.808262201372105e-06,
      "loss": 1.3783,
      "step": 2631
    },
    {
      "epoch": 0.673792,
      "grad_norm": 0.366740434741912,
      "learning_rate": 9.794281674339592e-06,
      "loss": 1.3345,
      "step": 2632
    },
    {
      "epoch": 0.674048,
      "grad_norm": 0.37138330520365054,
      "learning_rate": 9.780307886681725e-06,
      "loss": 1.3704,
      "step": 2633
    },
    {
      "epoch": 0.674304,
      "grad_norm": 0.37385656318772026,
      "learning_rate": 9.766340847626137e-06,
      "loss": 1.371,
      "step": 2634
    },
    {
      "epoch": 0.67456,
      "grad_norm": 0.3598827409244142,
      "learning_rate": 9.752380566396009e-06,
      "loss": 1.2962,
      "step": 2635
    },
    {
      "epoch": 0.674816,
      "grad_norm": 0.35804557056735475,
      "learning_rate": 9.738427052210034e-06,
      "loss": 1.284,
      "step": 2636
    },
    {
      "epoch": 0.675072,
      "grad_norm": 0.36629201743992723,
      "learning_rate": 9.724480314282463e-06,
      "loss": 1.3249,
      "step": 2637
    },
    {
      "epoch": 0.675328,
      "grad_norm": 0.36613584428443136,
      "learning_rate": 9.710540361823063e-06,
      "loss": 1.3601,
      "step": 2638
    },
    {
      "epoch": 0.675584,
      "grad_norm": 0.3708551405826036,
      "learning_rate": 9.696607204037124e-06,
      "loss": 1.3774,
      "step": 2639
    },
    {
      "epoch": 0.67584,
      "grad_norm": 0.36797737809950704,
      "learning_rate": 9.68268085012544e-06,
      "loss": 1.3419,
      "step": 2640
    },
    {
      "epoch": 0.676096,
      "grad_norm": 0.37093191446741586,
      "learning_rate": 9.668761309284331e-06,
      "loss": 1.3574,
      "step": 2641
    },
    {
      "epoch": 0.676352,
      "grad_norm": 0.3597602873692287,
      "learning_rate": 9.654848590705586e-06,
      "loss": 1.324,
      "step": 2642
    },
    {
      "epoch": 0.676608,
      "grad_norm": 0.35802403622133666,
      "learning_rate": 9.640942703576523e-06,
      "loss": 1.3318,
      "step": 2643
    },
    {
      "epoch": 0.676864,
      "grad_norm": 0.3677302768290871,
      "learning_rate": 9.627043657079919e-06,
      "loss": 1.3332,
      "step": 2644
    },
    {
      "epoch": 0.67712,
      "grad_norm": 0.37406146287412306,
      "learning_rate": 9.613151460394056e-06,
      "loss": 1.3698,
      "step": 2645
    },
    {
      "epoch": 0.677376,
      "grad_norm": 0.38207001485102654,
      "learning_rate": 9.599266122692685e-06,
      "loss": 1.3434,
      "step": 2646
    },
    {
      "epoch": 0.677632,
      "grad_norm": 0.3577688961063157,
      "learning_rate": 9.585387653145035e-06,
      "loss": 1.3355,
      "step": 2647
    },
    {
      "epoch": 0.677888,
      "grad_norm": 0.36934086676171113,
      "learning_rate": 9.571516060915777e-06,
      "loss": 1.36,
      "step": 2648
    },
    {
      "epoch": 0.678144,
      "grad_norm": 0.36254847639041815,
      "learning_rate": 9.557651355165063e-06,
      "loss": 1.3247,
      "step": 2649
    },
    {
      "epoch": 0.6784,
      "grad_norm": 0.3618082848720538,
      "learning_rate": 9.543793545048492e-06,
      "loss": 1.3149,
      "step": 2650
    },
    {
      "epoch": 0.678656,
      "grad_norm": 0.36334611485521706,
      "learning_rate": 9.529942639717105e-06,
      "loss": 1.3294,
      "step": 2651
    },
    {
      "epoch": 0.678912,
      "grad_norm": 0.37409551708703354,
      "learning_rate": 9.516098648317386e-06,
      "loss": 1.3414,
      "step": 2652
    },
    {
      "epoch": 0.679168,
      "grad_norm": 0.3664262978619659,
      "learning_rate": 9.502261579991266e-06,
      "loss": 1.3293,
      "step": 2653
    },
    {
      "epoch": 0.679424,
      "grad_norm": 0.3631581741889457,
      "learning_rate": 9.488431443876076e-06,
      "loss": 1.3151,
      "step": 2654
    },
    {
      "epoch": 0.67968,
      "grad_norm": 0.3706366330549185,
      "learning_rate": 9.474608249104584e-06,
      "loss": 1.3381,
      "step": 2655
    },
    {
      "epoch": 0.679936,
      "grad_norm": 0.36464991204874425,
      "learning_rate": 9.460792004804981e-06,
      "loss": 1.3672,
      "step": 2656
    },
    {
      "epoch": 0.680192,
      "grad_norm": 0.36127269295840575,
      "learning_rate": 9.446982720100864e-06,
      "loss": 1.3086,
      "step": 2657
    },
    {
      "epoch": 0.680448,
      "grad_norm": 0.3595141547582816,
      "learning_rate": 9.433180404111228e-06,
      "loss": 1.3204,
      "step": 2658
    },
    {
      "epoch": 0.680704,
      "grad_norm": 0.36304361675869173,
      "learning_rate": 9.419385065950477e-06,
      "loss": 1.3462,
      "step": 2659
    },
    {
      "epoch": 0.68096,
      "grad_norm": 0.3643391855609854,
      "learning_rate": 9.405596714728404e-06,
      "loss": 1.3254,
      "step": 2660
    },
    {
      "epoch": 0.681216,
      "grad_norm": 0.3670255704769003,
      "learning_rate": 9.391815359550173e-06,
      "loss": 1.3325,
      "step": 2661
    },
    {
      "epoch": 0.681472,
      "grad_norm": 0.357269760668828,
      "learning_rate": 9.37804100951635e-06,
      "loss": 1.3154,
      "step": 2662
    },
    {
      "epoch": 0.681728,
      "grad_norm": 0.364239896412167,
      "learning_rate": 9.36427367372286e-06,
      "loss": 1.341,
      "step": 2663
    },
    {
      "epoch": 0.681984,
      "grad_norm": 0.3693497742785979,
      "learning_rate": 9.35051336126101e-06,
      "loss": 1.3318,
      "step": 2664
    },
    {
      "epoch": 0.68224,
      "grad_norm": 0.3710250050560966,
      "learning_rate": 9.336760081217452e-06,
      "loss": 1.3181,
      "step": 2665
    },
    {
      "epoch": 0.682496,
      "grad_norm": 0.3653262761697295,
      "learning_rate": 9.323013842674212e-06,
      "loss": 1.3712,
      "step": 2666
    },
    {
      "epoch": 0.682752,
      "grad_norm": 0.38611736819797404,
      "learning_rate": 9.309274654708643e-06,
      "loss": 1.3694,
      "step": 2667
    },
    {
      "epoch": 0.683008,
      "grad_norm": 0.3744647647875601,
      "learning_rate": 9.295542526393464e-06,
      "loss": 1.3163,
      "step": 2668
    },
    {
      "epoch": 0.683264,
      "grad_norm": 0.3669198424090417,
      "learning_rate": 9.281817466796728e-06,
      "loss": 1.3465,
      "step": 2669
    },
    {
      "epoch": 0.68352,
      "grad_norm": 0.36070003418527274,
      "learning_rate": 9.26809948498181e-06,
      "loss": 1.3255,
      "step": 2670
    },
    {
      "epoch": 0.683776,
      "grad_norm": 0.3634309404692573,
      "learning_rate": 9.254388590007424e-06,
      "loss": 1.3433,
      "step": 2671
    },
    {
      "epoch": 0.684032,
      "grad_norm": 0.36699180197385417,
      "learning_rate": 9.240684790927598e-06,
      "loss": 1.3324,
      "step": 2672
    },
    {
      "epoch": 0.684288,
      "grad_norm": 0.36280058134141696,
      "learning_rate": 9.22698809679167e-06,
      "loss": 1.3217,
      "step": 2673
    },
    {
      "epoch": 0.684544,
      "grad_norm": 0.36000615128526053,
      "learning_rate": 9.213298516644287e-06,
      "loss": 1.2957,
      "step": 2674
    },
    {
      "epoch": 0.6848,
      "grad_norm": 0.3712449515860552,
      "learning_rate": 9.199616059525404e-06,
      "loss": 1.3526,
      "step": 2675
    },
    {
      "epoch": 0.685056,
      "grad_norm": 0.3732568886667032,
      "learning_rate": 9.185940734470282e-06,
      "loss": 1.3235,
      "step": 2676
    },
    {
      "epoch": 0.685312,
      "grad_norm": 0.35894064878767934,
      "learning_rate": 9.172272550509442e-06,
      "loss": 1.3477,
      "step": 2677
    },
    {
      "epoch": 0.685568,
      "grad_norm": 0.36645436831610567,
      "learning_rate": 9.158611516668723e-06,
      "loss": 1.3157,
      "step": 2678
    },
    {
      "epoch": 0.685824,
      "grad_norm": 0.3712963788261121,
      "learning_rate": 9.144957641969212e-06,
      "loss": 1.3927,
      "step": 2679
    },
    {
      "epoch": 0.68608,
      "grad_norm": 0.3612558927210817,
      "learning_rate": 9.13131093542729e-06,
      "loss": 1.3106,
      "step": 2680
    },
    {
      "epoch": 0.686336,
      "grad_norm": 0.36323600308068715,
      "learning_rate": 9.117671406054598e-06,
      "loss": 1.3176,
      "step": 2681
    },
    {
      "epoch": 0.686592,
      "grad_norm": 0.3619876173914447,
      "learning_rate": 9.104039062858038e-06,
      "loss": 1.3247,
      "step": 2682
    },
    {
      "epoch": 0.686848,
      "grad_norm": 0.37555033906580426,
      "learning_rate": 9.090413914839767e-06,
      "loss": 1.3311,
      "step": 2683
    },
    {
      "epoch": 0.687104,
      "grad_norm": 0.35821336698172096,
      "learning_rate": 9.076795970997194e-06,
      "loss": 1.315,
      "step": 2684
    },
    {
      "epoch": 0.68736,
      "grad_norm": 0.3591643943124592,
      "learning_rate": 9.063185240322956e-06,
      "loss": 1.3312,
      "step": 2685
    },
    {
      "epoch": 0.687616,
      "grad_norm": 0.3718901452981283,
      "learning_rate": 9.04958173180494e-06,
      "loss": 1.346,
      "step": 2686
    },
    {
      "epoch": 0.687872,
      "grad_norm": 0.35944733077925384,
      "learning_rate": 9.03598545442627e-06,
      "loss": 1.3504,
      "step": 2687
    },
    {
      "epoch": 0.688128,
      "grad_norm": 0.36914490488082985,
      "learning_rate": 9.02239641716527e-06,
      "loss": 1.3406,
      "step": 2688
    },
    {
      "epoch": 0.688384,
      "grad_norm": 0.37147969084760707,
      "learning_rate": 9.00881462899551e-06,
      "loss": 1.3704,
      "step": 2689
    },
    {
      "epoch": 0.68864,
      "grad_norm": 0.3679369775193561,
      "learning_rate": 8.995240098885765e-06,
      "loss": 1.3604,
      "step": 2690
    },
    {
      "epoch": 0.688896,
      "grad_norm": 0.37102353818815614,
      "learning_rate": 8.981672835800002e-06,
      "loss": 1.3806,
      "step": 2691
    },
    {
      "epoch": 0.689152,
      "grad_norm": 0.3708482987972002,
      "learning_rate": 8.968112848697405e-06,
      "loss": 1.3537,
      "step": 2692
    },
    {
      "epoch": 0.689408,
      "grad_norm": 0.3644548553130399,
      "learning_rate": 8.954560146532352e-06,
      "loss": 1.3458,
      "step": 2693
    },
    {
      "epoch": 0.689664,
      "grad_norm": 0.3680747224598955,
      "learning_rate": 8.941014738254408e-06,
      "loss": 1.3123,
      "step": 2694
    },
    {
      "epoch": 0.68992,
      "grad_norm": 0.3667592434036538,
      "learning_rate": 8.927476632808321e-06,
      "loss": 1.3119,
      "step": 2695
    },
    {
      "epoch": 0.690176,
      "grad_norm": 0.3684561443081247,
      "learning_rate": 8.913945839134024e-06,
      "loss": 1.3607,
      "step": 2696
    },
    {
      "epoch": 0.690432,
      "grad_norm": 0.36274277056443466,
      "learning_rate": 8.900422366166597e-06,
      "loss": 1.3173,
      "step": 2697
    },
    {
      "epoch": 0.690688,
      "grad_norm": 0.36917944741693287,
      "learning_rate": 8.886906222836316e-06,
      "loss": 1.3453,
      "step": 2698
    },
    {
      "epoch": 0.690944,
      "grad_norm": 0.37373778877944863,
      "learning_rate": 8.873397418068607e-06,
      "loss": 1.3717,
      "step": 2699
    },
    {
      "epoch": 0.6912,
      "grad_norm": 0.37887534752226204,
      "learning_rate": 8.859895960784035e-06,
      "loss": 1.3662,
      "step": 2700
    },
    {
      "epoch": 0.691456,
      "grad_norm": 0.37010635786786256,
      "learning_rate": 8.84640185989833e-06,
      "loss": 1.387,
      "step": 2701
    },
    {
      "epoch": 0.691712,
      "grad_norm": 0.3642657035372625,
      "learning_rate": 8.832915124322367e-06,
      "loss": 1.3494,
      "step": 2702
    },
    {
      "epoch": 0.691968,
      "grad_norm": 0.36593322220417407,
      "learning_rate": 8.819435762962139e-06,
      "loss": 1.3331,
      "step": 2703
    },
    {
      "epoch": 0.692224,
      "grad_norm": 0.3725780521511076,
      "learning_rate": 8.805963784718784e-06,
      "loss": 1.3326,
      "step": 2704
    },
    {
      "epoch": 0.69248,
      "grad_norm": 0.358535989937182,
      "learning_rate": 8.792499198488558e-06,
      "loss": 1.298,
      "step": 2705
    },
    {
      "epoch": 0.692736,
      "grad_norm": 0.3783157675522095,
      "learning_rate": 8.77904201316284e-06,
      "loss": 1.3582,
      "step": 2706
    },
    {
      "epoch": 0.692992,
      "grad_norm": 0.37385377141941734,
      "learning_rate": 8.765592237628122e-06,
      "loss": 1.3801,
      "step": 2707
    },
    {
      "epoch": 0.693248,
      "grad_norm": 0.36088150334117597,
      "learning_rate": 8.752149880766008e-06,
      "loss": 1.3029,
      "step": 2708
    },
    {
      "epoch": 0.693504,
      "grad_norm": 0.3622307663711269,
      "learning_rate": 8.73871495145318e-06,
      "loss": 1.3265,
      "step": 2709
    },
    {
      "epoch": 0.69376,
      "grad_norm": 0.3608388450352028,
      "learning_rate": 8.725287458561447e-06,
      "loss": 1.3093,
      "step": 2710
    },
    {
      "epoch": 0.694016,
      "grad_norm": 0.35895239323077144,
      "learning_rate": 8.711867410957675e-06,
      "loss": 1.3291,
      "step": 2711
    },
    {
      "epoch": 0.694272,
      "grad_norm": 0.3724069974085379,
      "learning_rate": 8.698454817503838e-06,
      "loss": 1.3255,
      "step": 2712
    },
    {
      "epoch": 0.694528,
      "grad_norm": 0.3756068886340145,
      "learning_rate": 8.685049687056978e-06,
      "loss": 1.344,
      "step": 2713
    },
    {
      "epoch": 0.694784,
      "grad_norm": 0.3583311020199094,
      "learning_rate": 8.671652028469224e-06,
      "loss": 1.3085,
      "step": 2714
    },
    {
      "epoch": 0.69504,
      "grad_norm": 0.36166882521852034,
      "learning_rate": 8.658261850587734e-06,
      "loss": 1.3275,
      "step": 2715
    },
    {
      "epoch": 0.695296,
      "grad_norm": 0.35960985376933485,
      "learning_rate": 8.644879162254765e-06,
      "loss": 1.3634,
      "step": 2716
    },
    {
      "epoch": 0.695552,
      "grad_norm": 0.36245227571916905,
      "learning_rate": 8.631503972307606e-06,
      "loss": 1.3154,
      "step": 2717
    },
    {
      "epoch": 0.695808,
      "grad_norm": 0.3723271206685477,
      "learning_rate": 8.618136289578604e-06,
      "loss": 1.3653,
      "step": 2718
    },
    {
      "epoch": 0.696064,
      "grad_norm": 0.36436910230196,
      "learning_rate": 8.604776122895147e-06,
      "loss": 1.3349,
      "step": 2719
    },
    {
      "epoch": 0.69632,
      "grad_norm": 0.3642398946562998,
      "learning_rate": 8.591423481079664e-06,
      "loss": 1.3454,
      "step": 2720
    },
    {
      "epoch": 0.696576,
      "grad_norm": 0.36850049668145113,
      "learning_rate": 8.578078372949594e-06,
      "loss": 1.3406,
      "step": 2721
    },
    {
      "epoch": 0.696832,
      "grad_norm": 0.37386734346523903,
      "learning_rate": 8.564740807317433e-06,
      "loss": 1.3553,
      "step": 2722
    },
    {
      "epoch": 0.697088,
      "grad_norm": 0.36606994545663074,
      "learning_rate": 8.551410792990664e-06,
      "loss": 1.3306,
      "step": 2723
    },
    {
      "epoch": 0.697344,
      "grad_norm": 0.3695826996197134,
      "learning_rate": 8.538088338771806e-06,
      "loss": 1.3406,
      "step": 2724
    },
    {
      "epoch": 0.6976,
      "grad_norm": 0.36798594266626083,
      "learning_rate": 8.524773453458381e-06,
      "loss": 1.3355,
      "step": 2725
    },
    {
      "epoch": 0.697856,
      "grad_norm": 0.3727504509789047,
      "learning_rate": 8.511466145842909e-06,
      "loss": 1.3769,
      "step": 2726
    },
    {
      "epoch": 0.698112,
      "grad_norm": 0.36389633466409566,
      "learning_rate": 8.498166424712914e-06,
      "loss": 1.3475,
      "step": 2727
    },
    {
      "epoch": 0.698368,
      "grad_norm": 0.3608045830256791,
      "learning_rate": 8.484874298850894e-06,
      "loss": 1.2943,
      "step": 2728
    },
    {
      "epoch": 0.698624,
      "grad_norm": 0.3666100568208091,
      "learning_rate": 8.471589777034343e-06,
      "loss": 1.3152,
      "step": 2729
    },
    {
      "epoch": 0.69888,
      "grad_norm": 0.3935670412789717,
      "learning_rate": 8.458312868035739e-06,
      "loss": 1.3224,
      "step": 2730
    },
    {
      "epoch": 0.699136,
      "grad_norm": 0.3620247065282502,
      "learning_rate": 8.445043580622527e-06,
      "loss": 1.3244,
      "step": 2731
    },
    {
      "epoch": 0.699392,
      "grad_norm": 0.358201525530314,
      "learning_rate": 8.431781923557105e-06,
      "loss": 1.3301,
      "step": 2732
    },
    {
      "epoch": 0.699648,
      "grad_norm": 0.372965858495583,
      "learning_rate": 8.418527905596862e-06,
      "loss": 1.3515,
      "step": 2733
    },
    {
      "epoch": 0.699904,
      "grad_norm": 0.37757065670463225,
      "learning_rate": 8.40528153549411e-06,
      "loss": 1.3714,
      "step": 2734
    },
    {
      "epoch": 0.70016,
      "grad_norm": 0.3622097062107426,
      "learning_rate": 8.392042821996134e-06,
      "loss": 1.3389,
      "step": 2735
    },
    {
      "epoch": 0.700416,
      "grad_norm": 0.36302224598366645,
      "learning_rate": 8.378811773845155e-06,
      "loss": 1.3382,
      "step": 2736
    },
    {
      "epoch": 0.700672,
      "grad_norm": 0.3763654356691901,
      "learning_rate": 8.365588399778333e-06,
      "loss": 1.351,
      "step": 2737
    },
    {
      "epoch": 0.700928,
      "grad_norm": 0.3641028098842834,
      "learning_rate": 8.352372708527758e-06,
      "loss": 1.3369,
      "step": 2738
    },
    {
      "epoch": 0.701184,
      "grad_norm": 0.3591910992755397,
      "learning_rate": 8.33916470882046e-06,
      "loss": 1.3207,
      "step": 2739
    },
    {
      "epoch": 0.70144,
      "grad_norm": 0.3584194985287829,
      "learning_rate": 8.325964409378362e-06,
      "loss": 1.3244,
      "step": 2740
    },
    {
      "epoch": 0.701696,
      "grad_norm": 0.37103968026485573,
      "learning_rate": 8.312771818918326e-06,
      "loss": 1.3366,
      "step": 2741
    },
    {
      "epoch": 0.701952,
      "grad_norm": 0.3640122897082763,
      "learning_rate": 8.299586946152117e-06,
      "loss": 1.326,
      "step": 2742
    },
    {
      "epoch": 0.702208,
      "grad_norm": 0.36560827498088616,
      "learning_rate": 8.286409799786409e-06,
      "loss": 1.3507,
      "step": 2743
    },
    {
      "epoch": 0.702464,
      "grad_norm": 0.3634600113894542,
      "learning_rate": 8.273240388522749e-06,
      "loss": 1.3353,
      "step": 2744
    },
    {
      "epoch": 0.70272,
      "grad_norm": 0.36797789897040883,
      "learning_rate": 8.260078721057615e-06,
      "loss": 1.3441,
      "step": 2745
    },
    {
      "epoch": 0.702976,
      "grad_norm": 0.3652575457221787,
      "learning_rate": 8.246924806082338e-06,
      "loss": 1.363,
      "step": 2746
    },
    {
      "epoch": 0.703232,
      "grad_norm": 0.3633835172109707,
      "learning_rate": 8.233778652283143e-06,
      "loss": 1.3252,
      "step": 2747
    },
    {
      "epoch": 0.703488,
      "grad_norm": 0.3622707077283908,
      "learning_rate": 8.220640268341132e-06,
      "loss": 1.341,
      "step": 2748
    },
    {
      "epoch": 0.703744,
      "grad_norm": 0.3620740280694645,
      "learning_rate": 8.207509662932276e-06,
      "loss": 1.3187,
      "step": 2749
    },
    {
      "epoch": 0.704,
      "grad_norm": 0.45984208038303825,
      "learning_rate": 8.194386844727402e-06,
      "loss": 1.3236,
      "step": 2750
    },
    {
      "epoch": 0.704256,
      "grad_norm": 0.36700097689721134,
      "learning_rate": 8.181271822392213e-06,
      "loss": 1.3664,
      "step": 2751
    },
    {
      "epoch": 0.704512,
      "grad_norm": 0.36531681607576694,
      "learning_rate": 8.168164604587234e-06,
      "loss": 1.3405,
      "step": 2752
    },
    {
      "epoch": 0.704768,
      "grad_norm": 0.3721593833582203,
      "learning_rate": 8.155065199967857e-06,
      "loss": 1.3646,
      "step": 2753
    },
    {
      "epoch": 0.705024,
      "grad_norm": 0.36140264020776575,
      "learning_rate": 8.141973617184322e-06,
      "loss": 1.3334,
      "step": 2754
    },
    {
      "epoch": 0.70528,
      "grad_norm": 0.35267597267187806,
      "learning_rate": 8.128889864881676e-06,
      "loss": 1.3118,
      "step": 2755
    },
    {
      "epoch": 0.705536,
      "grad_norm": 0.3628531758055449,
      "learning_rate": 8.115813951699818e-06,
      "loss": 1.3489,
      "step": 2756
    },
    {
      "epoch": 0.705792,
      "grad_norm": 0.3566190851198683,
      "learning_rate": 8.102745886273472e-06,
      "loss": 1.2909,
      "step": 2757
    },
    {
      "epoch": 0.706048,
      "grad_norm": 0.4251694084038126,
      "learning_rate": 8.089685677232158e-06,
      "loss": 1.3154,
      "step": 2758
    },
    {
      "epoch": 0.706304,
      "grad_norm": 0.36430182823945445,
      "learning_rate": 8.076633333200225e-06,
      "loss": 1.329,
      "step": 2759
    },
    {
      "epoch": 0.70656,
      "grad_norm": 0.38807206857161175,
      "learning_rate": 8.063588862796832e-06,
      "loss": 1.3412,
      "step": 2760
    },
    {
      "epoch": 0.706816,
      "grad_norm": 0.3614892915017986,
      "learning_rate": 8.050552274635923e-06,
      "loss": 1.3435,
      "step": 2761
    },
    {
      "epoch": 0.707072,
      "grad_norm": 0.3827429608484736,
      "learning_rate": 8.037523577326254e-06,
      "loss": 1.3233,
      "step": 2762
    },
    {
      "epoch": 0.707328,
      "grad_norm": 0.36540664272157325,
      "learning_rate": 8.024502779471361e-06,
      "loss": 1.3626,
      "step": 2763
    },
    {
      "epoch": 0.707584,
      "grad_norm": 0.3644753721133332,
      "learning_rate": 8.011489889669554e-06,
      "loss": 1.3351,
      "step": 2764
    },
    {
      "epoch": 0.70784,
      "grad_norm": 0.35542273495093196,
      "learning_rate": 7.99848491651394e-06,
      "loss": 1.3004,
      "step": 2765
    },
    {
      "epoch": 0.708096,
      "grad_norm": 0.36107881848533613,
      "learning_rate": 7.985487868592392e-06,
      "loss": 1.3152,
      "step": 2766
    },
    {
      "epoch": 0.708352,
      "grad_norm": 0.4079202799746153,
      "learning_rate": 7.972498754487537e-06,
      "loss": 1.301,
      "step": 2767
    },
    {
      "epoch": 0.708608,
      "grad_norm": 0.3550897685212053,
      "learning_rate": 7.959517582776776e-06,
      "loss": 1.3318,
      "step": 2768
    },
    {
      "epoch": 0.708864,
      "grad_norm": 0.36217723870521634,
      "learning_rate": 7.946544362032274e-06,
      "loss": 1.3411,
      "step": 2769
    },
    {
      "epoch": 0.70912,
      "grad_norm": 0.3753939546296843,
      "learning_rate": 7.933579100820914e-06,
      "loss": 1.3805,
      "step": 2770
    },
    {
      "epoch": 0.709376,
      "grad_norm": 0.36143791329258396,
      "learning_rate": 7.920621807704355e-06,
      "loss": 1.3198,
      "step": 2771
    },
    {
      "epoch": 0.709632,
      "grad_norm": 0.36903241967884287,
      "learning_rate": 7.907672491238976e-06,
      "loss": 1.3438,
      "step": 2772
    },
    {
      "epoch": 0.709888,
      "grad_norm": 0.3634585958784751,
      "learning_rate": 7.894731159975896e-06,
      "loss": 1.3276,
      "step": 2773
    },
    {
      "epoch": 0.710144,
      "grad_norm": 0.36197502182941976,
      "learning_rate": 7.881797822460965e-06,
      "loss": 1.3291,
      "step": 2774
    },
    {
      "epoch": 0.7104,
      "grad_norm": 0.3750991396589606,
      "learning_rate": 7.86887248723475e-06,
      "loss": 1.3218,
      "step": 2775
    },
    {
      "epoch": 0.710656,
      "grad_norm": 0.3682203494174357,
      "learning_rate": 7.855955162832519e-06,
      "loss": 1.3594,
      "step": 2776
    },
    {
      "epoch": 0.710912,
      "grad_norm": 0.3659267671016427,
      "learning_rate": 7.843045857784278e-06,
      "loss": 1.3595,
      "step": 2777
    },
    {
      "epoch": 0.711168,
      "grad_norm": 0.3578802813680482,
      "learning_rate": 7.830144580614709e-06,
      "loss": 1.3338,
      "step": 2778
    },
    {
      "epoch": 0.711424,
      "grad_norm": 0.371244997667776,
      "learning_rate": 7.817251339843213e-06,
      "loss": 1.2991,
      "step": 2779
    },
    {
      "epoch": 0.71168,
      "grad_norm": 0.36245530073378024,
      "learning_rate": 7.804366143983878e-06,
      "loss": 1.358,
      "step": 2780
    },
    {
      "epoch": 0.711936,
      "grad_norm": 0.3598542358759934,
      "learning_rate": 7.791489001545483e-06,
      "loss": 1.3379,
      "step": 2781
    },
    {
      "epoch": 0.712192,
      "grad_norm": 0.37305002680947785,
      "learning_rate": 7.778619921031476e-06,
      "loss": 1.3059,
      "step": 2782
    },
    {
      "epoch": 0.712448,
      "grad_norm": 0.37256635303135505,
      "learning_rate": 7.765758910939995e-06,
      "loss": 1.3244,
      "step": 2783
    },
    {
      "epoch": 0.712704,
      "grad_norm": 0.36312275140892275,
      "learning_rate": 7.752905979763846e-06,
      "loss": 1.3299,
      "step": 2784
    },
    {
      "epoch": 0.71296,
      "grad_norm": 0.36263784012530204,
      "learning_rate": 7.740061135990493e-06,
      "loss": 1.3683,
      "step": 2785
    },
    {
      "epoch": 0.713216,
      "grad_norm": 0.3683660394471804,
      "learning_rate": 7.727224388102069e-06,
      "loss": 1.3565,
      "step": 2786
    },
    {
      "epoch": 0.713472,
      "grad_norm": 0.37191035633220004,
      "learning_rate": 7.714395744575362e-06,
      "loss": 1.3046,
      "step": 2787
    },
    {
      "epoch": 0.713728,
      "grad_norm": 0.3610043726950672,
      "learning_rate": 7.701575213881788e-06,
      "loss": 1.3341,
      "step": 2788
    },
    {
      "epoch": 0.713984,
      "grad_norm": 0.3575970011151364,
      "learning_rate": 7.688762804487437e-06,
      "loss": 1.3023,
      "step": 2789
    },
    {
      "epoch": 0.71424,
      "grad_norm": 0.36240701882029364,
      "learning_rate": 7.675958524853003e-06,
      "loss": 1.3017,
      "step": 2790
    },
    {
      "epoch": 0.714496,
      "grad_norm": 0.3829393418781689,
      "learning_rate": 7.663162383433834e-06,
      "loss": 1.3743,
      "step": 2791
    },
    {
      "epoch": 0.714752,
      "grad_norm": 0.3722489821194553,
      "learning_rate": 7.6503743886799e-06,
      "loss": 1.3718,
      "step": 2792
    },
    {
      "epoch": 0.715008,
      "grad_norm": 0.36808697856222033,
      "learning_rate": 7.637594549035787e-06,
      "loss": 1.3181,
      "step": 2793
    },
    {
      "epoch": 0.715264,
      "grad_norm": 0.3623251315224491,
      "learning_rate": 7.624822872940707e-06,
      "loss": 1.3074,
      "step": 2794
    },
    {
      "epoch": 0.71552,
      "grad_norm": 0.3606671799287942,
      "learning_rate": 7.612059368828457e-06,
      "loss": 1.3287,
      "step": 2795
    },
    {
      "epoch": 0.715776,
      "grad_norm": 0.36263257851615077,
      "learning_rate": 7.59930404512746e-06,
      "loss": 1.3188,
      "step": 2796
    },
    {
      "epoch": 0.716032,
      "grad_norm": 0.37161468820475085,
      "learning_rate": 7.5865569102607295e-06,
      "loss": 1.3312,
      "step": 2797
    },
    {
      "epoch": 0.716288,
      "grad_norm": 0.410201410714656,
      "learning_rate": 7.573817972645872e-06,
      "loss": 1.326,
      "step": 2798
    },
    {
      "epoch": 0.716544,
      "grad_norm": 0.36860411138989085,
      "learning_rate": 7.561087240695086e-06,
      "loss": 1.3233,
      "step": 2799
    },
    {
      "epoch": 0.7168,
      "grad_norm": 0.37456948624033176,
      "learning_rate": 7.548364722815142e-06,
      "loss": 1.3487,
      "step": 2800
    },
    {
      "epoch": 0.717056,
      "grad_norm": 0.3645861327314719,
      "learning_rate": 7.535650427407379e-06,
      "loss": 1.3108,
      "step": 2801
    },
    {
      "epoch": 0.717312,
      "grad_norm": 0.36824676013177354,
      "learning_rate": 7.52294436286773e-06,
      "loss": 1.3419,
      "step": 2802
    },
    {
      "epoch": 0.717568,
      "grad_norm": 0.3677961332399227,
      "learning_rate": 7.5102465375866765e-06,
      "loss": 1.331,
      "step": 2803
    },
    {
      "epoch": 0.717824,
      "grad_norm": 0.36064888460405026,
      "learning_rate": 7.497556959949262e-06,
      "loss": 1.3132,
      "step": 2804
    },
    {
      "epoch": 0.71808,
      "grad_norm": 0.3654641040746755,
      "learning_rate": 7.484875638335087e-06,
      "loss": 1.3489,
      "step": 2805
    },
    {
      "epoch": 0.718336,
      "grad_norm": 0.36333436096603144,
      "learning_rate": 7.472202581118304e-06,
      "loss": 1.3331,
      "step": 2806
    },
    {
      "epoch": 0.718592,
      "grad_norm": 0.3643158040582172,
      "learning_rate": 7.459537796667589e-06,
      "loss": 1.3097,
      "step": 2807
    },
    {
      "epoch": 0.718848,
      "grad_norm": 0.35816073504176726,
      "learning_rate": 7.446881293346171e-06,
      "loss": 1.2853,
      "step": 2808
    },
    {
      "epoch": 0.719104,
      "grad_norm": 0.3672466278772919,
      "learning_rate": 7.434233079511812e-06,
      "loss": 1.3585,
      "step": 2809
    },
    {
      "epoch": 0.71936,
      "grad_norm": 0.3673194788782525,
      "learning_rate": 7.4215931635168e-06,
      "loss": 1.3099,
      "step": 2810
    },
    {
      "epoch": 0.719616,
      "grad_norm": 0.36050103145502316,
      "learning_rate": 7.408961553707925e-06,
      "loss": 1.3425,
      "step": 2811
    },
    {
      "epoch": 0.719872,
      "grad_norm": 0.35633560634428346,
      "learning_rate": 7.396338258426521e-06,
      "loss": 1.3069,
      "step": 2812
    },
    {
      "epoch": 0.720128,
      "grad_norm": 0.36993081620145385,
      "learning_rate": 7.383723286008402e-06,
      "loss": 1.3661,
      "step": 2813
    },
    {
      "epoch": 0.720384,
      "grad_norm": 0.36454955138759243,
      "learning_rate": 7.371116644783905e-06,
      "loss": 1.3135,
      "step": 2814
    },
    {
      "epoch": 0.72064,
      "grad_norm": 0.3567921988220691,
      "learning_rate": 7.358518343077863e-06,
      "loss": 1.3066,
      "step": 2815
    },
    {
      "epoch": 0.720896,
      "grad_norm": 0.3563686964617256,
      "learning_rate": 7.3459283892096e-06,
      "loss": 1.2991,
      "step": 2816
    },
    {
      "epoch": 0.721152,
      "grad_norm": 0.3659090530769971,
      "learning_rate": 7.333346791492928e-06,
      "loss": 1.3247,
      "step": 2817
    },
    {
      "epoch": 0.721408,
      "grad_norm": 0.36442974143067813,
      "learning_rate": 7.320773558236145e-06,
      "loss": 1.3523,
      "step": 2818
    },
    {
      "epoch": 0.721664,
      "grad_norm": 0.3608091835990466,
      "learning_rate": 7.308208697742005e-06,
      "loss": 1.2936,
      "step": 2819
    },
    {
      "epoch": 0.72192,
      "grad_norm": 0.36703596126098786,
      "learning_rate": 7.295652218307763e-06,
      "loss": 1.3391,
      "step": 2820
    },
    {
      "epoch": 0.722176,
      "grad_norm": 0.3629516574988406,
      "learning_rate": 7.283104128225118e-06,
      "loss": 1.3389,
      "step": 2821
    },
    {
      "epoch": 0.722432,
      "grad_norm": 0.36207248603407655,
      "learning_rate": 7.270564435780247e-06,
      "loss": 1.3178,
      "step": 2822
    },
    {
      "epoch": 0.722688,
      "grad_norm": 0.3736835078519365,
      "learning_rate": 7.2580331492537555e-06,
      "loss": 1.3309,
      "step": 2823
    },
    {
      "epoch": 0.722944,
      "grad_norm": 0.3575569422064091,
      "learning_rate": 7.245510276920729e-06,
      "loss": 1.3333,
      "step": 2824
    },
    {
      "epoch": 0.7232,
      "grad_norm": 0.3659348719471312,
      "learning_rate": 7.232995827050666e-06,
      "loss": 1.3041,
      "step": 2825
    },
    {
      "epoch": 0.723456,
      "grad_norm": 0.3571507760616756,
      "learning_rate": 7.220489807907527e-06,
      "loss": 1.3441,
      "step": 2826
    },
    {
      "epoch": 0.723712,
      "grad_norm": 0.36173909812408583,
      "learning_rate": 7.207992227749694e-06,
      "loss": 1.3322,
      "step": 2827
    },
    {
      "epoch": 0.723968,
      "grad_norm": 0.36476573571224014,
      "learning_rate": 7.195503094829983e-06,
      "loss": 1.3529,
      "step": 2828
    },
    {
      "epoch": 0.724224,
      "grad_norm": 0.47874967975608623,
      "learning_rate": 7.1830224173956245e-06,
      "loss": 1.3253,
      "step": 2829
    },
    {
      "epoch": 0.72448,
      "grad_norm": 0.36674333850932334,
      "learning_rate": 7.170550203688278e-06,
      "loss": 1.3274,
      "step": 2830
    },
    {
      "epoch": 0.724736,
      "grad_norm": 0.3632950048806807,
      "learning_rate": 7.158086461943987e-06,
      "loss": 1.3257,
      "step": 2831
    },
    {
      "epoch": 0.724992,
      "grad_norm": 0.36140070591146417,
      "learning_rate": 7.14563120039323e-06,
      "loss": 1.318,
      "step": 2832
    },
    {
      "epoch": 0.725248,
      "grad_norm": 0.3674532997097461,
      "learning_rate": 7.1331844272608795e-06,
      "loss": 1.3397,
      "step": 2833
    },
    {
      "epoch": 0.725504,
      "grad_norm": 0.36770102503611196,
      "learning_rate": 7.1207461507661805e-06,
      "loss": 1.3155,
      "step": 2834
    },
    {
      "epoch": 0.72576,
      "grad_norm": 0.3572166504144376,
      "learning_rate": 7.108316379122793e-06,
      "loss": 1.3311,
      "step": 2835
    },
    {
      "epoch": 0.726016,
      "grad_norm": 0.47435280272680036,
      "learning_rate": 7.095895120538758e-06,
      "loss": 1.358,
      "step": 2836
    },
    {
      "epoch": 0.726272,
      "grad_norm": 0.3936589216050705,
      "learning_rate": 7.083482383216473e-06,
      "loss": 1.349,
      "step": 2837
    },
    {
      "epoch": 0.726528,
      "grad_norm": 0.3652460774782325,
      "learning_rate": 7.0710781753527325e-06,
      "loss": 1.3498,
      "step": 2838
    },
    {
      "epoch": 0.726784,
      "grad_norm": 0.3590831536592986,
      "learning_rate": 7.05868250513869e-06,
      "loss": 1.3381,
      "step": 2839
    },
    {
      "epoch": 0.72704,
      "grad_norm": 0.37046210410925323,
      "learning_rate": 7.04629538075986e-06,
      "loss": 1.3779,
      "step": 2840
    },
    {
      "epoch": 0.727296,
      "grad_norm": 0.3643138457888806,
      "learning_rate": 7.033916810396115e-06,
      "loss": 1.3314,
      "step": 2841
    },
    {
      "epoch": 0.727552,
      "grad_norm": 0.3632987223574396,
      "learning_rate": 7.021546802221686e-06,
      "loss": 1.303,
      "step": 2842
    },
    {
      "epoch": 0.727808,
      "grad_norm": 0.36910389445971414,
      "learning_rate": 7.009185364405129e-06,
      "loss": 1.3039,
      "step": 2843
    },
    {
      "epoch": 0.728064,
      "grad_norm": 0.35925845938019696,
      "learning_rate": 6.996832505109359e-06,
      "loss": 1.2821,
      "step": 2844
    },
    {
      "epoch": 0.72832,
      "grad_norm": 0.3659988405102717,
      "learning_rate": 6.984488232491628e-06,
      "loss": 1.3253,
      "step": 2845
    },
    {
      "epoch": 0.728576,
      "grad_norm": 0.37413426362838,
      "learning_rate": 6.972152554703499e-06,
      "loss": 1.3692,
      "step": 2846
    },
    {
      "epoch": 0.728832,
      "grad_norm": 0.40473669481767854,
      "learning_rate": 6.95982547989088e-06,
      "loss": 1.3242,
      "step": 2847
    },
    {
      "epoch": 0.729088,
      "grad_norm": 0.35701231562340285,
      "learning_rate": 6.947507016193986e-06,
      "loss": 1.3551,
      "step": 2848
    },
    {
      "epoch": 0.729344,
      "grad_norm": 0.36385206734505193,
      "learning_rate": 6.935197171747357e-06,
      "loss": 1.2984,
      "step": 2849
    },
    {
      "epoch": 0.7296,
      "grad_norm": 0.3640112070358793,
      "learning_rate": 6.922895954679818e-06,
      "loss": 1.3006,
      "step": 2850
    },
    {
      "epoch": 0.729856,
      "grad_norm": 0.3613111516553966,
      "learning_rate": 6.910603373114522e-06,
      "loss": 1.29,
      "step": 2851
    },
    {
      "epoch": 0.730112,
      "grad_norm": 0.35288833759229754,
      "learning_rate": 6.898319435168905e-06,
      "loss": 1.2919,
      "step": 2852
    },
    {
      "epoch": 0.730368,
      "grad_norm": 0.3933725513873792,
      "learning_rate": 6.886044148954707e-06,
      "loss": 1.3342,
      "step": 2853
    },
    {
      "epoch": 0.730624,
      "grad_norm": 0.36008652081657955,
      "learning_rate": 6.8737775225779405e-06,
      "loss": 1.307,
      "step": 2854
    },
    {
      "epoch": 0.73088,
      "grad_norm": 0.3695936770786274,
      "learning_rate": 6.8615195641389165e-06,
      "loss": 1.3487,
      "step": 2855
    },
    {
      "epoch": 0.731136,
      "grad_norm": 0.36239648938541164,
      "learning_rate": 6.849270281732206e-06,
      "loss": 1.3184,
      "step": 2856
    },
    {
      "epoch": 0.731392,
      "grad_norm": 0.3660138449744174,
      "learning_rate": 6.837029683446647e-06,
      "loss": 1.3433,
      "step": 2857
    },
    {
      "epoch": 0.731648,
      "grad_norm": 0.35969062572953525,
      "learning_rate": 6.824797777365364e-06,
      "loss": 1.3206,
      "step": 2858
    },
    {
      "epoch": 0.731904,
      "grad_norm": 0.35773110872044217,
      "learning_rate": 6.812574571565729e-06,
      "loss": 1.3208,
      "step": 2859
    },
    {
      "epoch": 0.73216,
      "grad_norm": 0.361529121509267,
      "learning_rate": 6.800360074119367e-06,
      "loss": 1.3335,
      "step": 2860
    },
    {
      "epoch": 0.732416,
      "grad_norm": 0.3749156458300707,
      "learning_rate": 6.788154293092166e-06,
      "loss": 1.3689,
      "step": 2861
    },
    {
      "epoch": 0.732672,
      "grad_norm": 0.37801284819404646,
      "learning_rate": 6.775957236544231e-06,
      "loss": 1.3548,
      "step": 2862
    },
    {
      "epoch": 0.732928,
      "grad_norm": 0.37054189757157663,
      "learning_rate": 6.763768912529934e-06,
      "loss": 1.3654,
      "step": 2863
    },
    {
      "epoch": 0.733184,
      "grad_norm": 0.3779285328064393,
      "learning_rate": 6.751589329097863e-06,
      "loss": 1.371,
      "step": 2864
    },
    {
      "epoch": 0.73344,
      "grad_norm": 0.3631957486230213,
      "learning_rate": 6.739418494290844e-06,
      "loss": 1.3323,
      "step": 2865
    },
    {
      "epoch": 0.733696,
      "grad_norm": 0.371760107273983,
      "learning_rate": 6.727256416145926e-06,
      "loss": 1.3802,
      "step": 2866
    },
    {
      "epoch": 0.733952,
      "grad_norm": 0.38476313850643973,
      "learning_rate": 6.71510310269436e-06,
      "loss": 1.3649,
      "step": 2867
    },
    {
      "epoch": 0.734208,
      "grad_norm": 0.3685614627211531,
      "learning_rate": 6.702958561961635e-06,
      "loss": 1.3407,
      "step": 2868
    },
    {
      "epoch": 0.734464,
      "grad_norm": 0.3606807338826802,
      "learning_rate": 6.690822801967414e-06,
      "loss": 1.2999,
      "step": 2869
    },
    {
      "epoch": 0.73472,
      "grad_norm": 0.36831944718808013,
      "learning_rate": 6.678695830725593e-06,
      "loss": 1.362,
      "step": 2870
    },
    {
      "epoch": 0.734976,
      "grad_norm": 0.47357312152165876,
      "learning_rate": 6.666577656244246e-06,
      "loss": 1.3113,
      "step": 2871
    },
    {
      "epoch": 0.735232,
      "grad_norm": 0.3654274251928827,
      "learning_rate": 6.654468286525646e-06,
      "loss": 1.3518,
      "step": 2872
    },
    {
      "epoch": 0.735488,
      "grad_norm": 0.3546914140566232,
      "learning_rate": 6.642367729566257e-06,
      "loss": 1.3479,
      "step": 2873
    },
    {
      "epoch": 0.735744,
      "grad_norm": 0.36387467639163257,
      "learning_rate": 6.6302759933567e-06,
      "loss": 1.3208,
      "step": 2874
    },
    {
      "epoch": 0.736,
      "grad_norm": 0.37587832089033957,
      "learning_rate": 6.618193085881794e-06,
      "loss": 1.3988,
      "step": 2875
    },
    {
      "epoch": 0.736256,
      "grad_norm": 0.37431798671056304,
      "learning_rate": 6.606119015120522e-06,
      "loss": 1.3577,
      "step": 2876
    },
    {
      "epoch": 0.736512,
      "grad_norm": 0.36748530063422613,
      "learning_rate": 6.594053789046031e-06,
      "loss": 1.3452,
      "step": 2877
    },
    {
      "epoch": 0.736768,
      "grad_norm": 0.3608372617898391,
      "learning_rate": 6.58199741562563e-06,
      "loss": 1.2782,
      "step": 2878
    },
    {
      "epoch": 0.737024,
      "grad_norm": 0.3723507905354946,
      "learning_rate": 6.569949902820776e-06,
      "loss": 1.3597,
      "step": 2879
    },
    {
      "epoch": 0.73728,
      "grad_norm": 0.36049699614161146,
      "learning_rate": 6.557911258587069e-06,
      "loss": 1.3059,
      "step": 2880
    },
    {
      "epoch": 0.737536,
      "grad_norm": 0.3712984095532251,
      "learning_rate": 6.545881490874267e-06,
      "loss": 1.3375,
      "step": 2881
    },
    {
      "epoch": 0.737792,
      "grad_norm": 0.36773145249240613,
      "learning_rate": 6.533860607626263e-06,
      "loss": 1.3383,
      "step": 2882
    },
    {
      "epoch": 0.738048,
      "grad_norm": 0.3602790819426775,
      "learning_rate": 6.521848616781079e-06,
      "loss": 1.3168,
      "step": 2883
    },
    {
      "epoch": 0.738304,
      "grad_norm": 0.3663960669176364,
      "learning_rate": 6.50984552627087e-06,
      "loss": 1.31,
      "step": 2884
    },
    {
      "epoch": 0.73856,
      "grad_norm": 0.3653924975761182,
      "learning_rate": 6.4978513440219125e-06,
      "loss": 1.3455,
      "step": 2885
    },
    {
      "epoch": 0.738816,
      "grad_norm": 0.35453178324529455,
      "learning_rate": 6.48586607795459e-06,
      "loss": 1.2849,
      "step": 2886
    },
    {
      "epoch": 0.739072,
      "grad_norm": 0.3533894471528382,
      "learning_rate": 6.47388973598341e-06,
      "loss": 1.2776,
      "step": 2887
    },
    {
      "epoch": 0.739328,
      "grad_norm": 0.3738095578280008,
      "learning_rate": 6.461922326016983e-06,
      "loss": 1.3508,
      "step": 2888
    },
    {
      "epoch": 0.739584,
      "grad_norm": 0.36506522560978083,
      "learning_rate": 6.449963855958031e-06,
      "loss": 1.3544,
      "step": 2889
    },
    {
      "epoch": 0.73984,
      "grad_norm": 0.36474066235915004,
      "learning_rate": 6.438014333703346e-06,
      "loss": 1.3264,
      "step": 2890
    },
    {
      "epoch": 0.740096,
      "grad_norm": 0.362413765001432,
      "learning_rate": 6.426073767143845e-06,
      "loss": 1.3366,
      "step": 2891
    },
    {
      "epoch": 0.740352,
      "grad_norm": 0.37652237871704547,
      "learning_rate": 6.414142164164501e-06,
      "loss": 1.3598,
      "step": 2892
    },
    {
      "epoch": 0.740608,
      "grad_norm": 0.36708740171109727,
      "learning_rate": 6.402219532644385e-06,
      "loss": 1.3666,
      "step": 2893
    },
    {
      "epoch": 0.740864,
      "grad_norm": 0.3574138690855728,
      "learning_rate": 6.3903058804566445e-06,
      "loss": 1.3346,
      "step": 2894
    },
    {
      "epoch": 0.74112,
      "grad_norm": 0.35911040744824196,
      "learning_rate": 6.3784012154684885e-06,
      "loss": 1.2751,
      "step": 2895
    },
    {
      "epoch": 0.741376,
      "grad_norm": 0.3570029046643827,
      "learning_rate": 6.3665055455412e-06,
      "loss": 1.2907,
      "step": 2896
    },
    {
      "epoch": 0.741632,
      "grad_norm": 0.365321453743712,
      "learning_rate": 6.354618878530123e-06,
      "loss": 1.3358,
      "step": 2897
    },
    {
      "epoch": 0.741888,
      "grad_norm": 0.3667263140885956,
      "learning_rate": 6.342741222284636e-06,
      "loss": 1.3622,
      "step": 2898
    },
    {
      "epoch": 0.742144,
      "grad_norm": 0.3618842106095794,
      "learning_rate": 6.330872584648193e-06,
      "loss": 1.3202,
      "step": 2899
    },
    {
      "epoch": 0.7424,
      "grad_norm": 0.35834299798938146,
      "learning_rate": 6.319012973458278e-06,
      "loss": 1.3469,
      "step": 2900
    },
    {
      "epoch": 0.742656,
      "grad_norm": 0.3778759831944958,
      "learning_rate": 6.307162396546429e-06,
      "loss": 1.3392,
      "step": 2901
    },
    {
      "epoch": 0.742912,
      "grad_norm": 0.3702861655149208,
      "learning_rate": 6.29532086173819e-06,
      "loss": 1.3629,
      "step": 2902
    },
    {
      "epoch": 0.743168,
      "grad_norm": 0.36856202080908806,
      "learning_rate": 6.283488376853168e-06,
      "loss": 1.345,
      "step": 2903
    },
    {
      "epoch": 0.743424,
      "grad_norm": 0.3558241031055548,
      "learning_rate": 6.2716649497049654e-06,
      "loss": 1.2677,
      "step": 2904
    },
    {
      "epoch": 0.74368,
      "grad_norm": 0.3654000406464242,
      "learning_rate": 6.2598505881012195e-06,
      "loss": 1.3086,
      "step": 2905
    },
    {
      "epoch": 0.743936,
      "grad_norm": 0.4211504297314756,
      "learning_rate": 6.248045299843577e-06,
      "loss": 1.3277,
      "step": 2906
    },
    {
      "epoch": 0.744192,
      "grad_norm": 0.590067654321583,
      "learning_rate": 6.236249092727693e-06,
      "loss": 1.3242,
      "step": 2907
    },
    {
      "epoch": 0.744448,
      "grad_norm": 0.36751186365142957,
      "learning_rate": 6.224461974543227e-06,
      "loss": 1.3118,
      "step": 2908
    },
    {
      "epoch": 0.744704,
      "grad_norm": 0.36372208342595935,
      "learning_rate": 6.212683953073837e-06,
      "loss": 1.2856,
      "step": 2909
    },
    {
      "epoch": 0.74496,
      "grad_norm": 0.36884207500523186,
      "learning_rate": 6.200915036097177e-06,
      "loss": 1.3546,
      "step": 2910
    },
    {
      "epoch": 0.745216,
      "grad_norm": 0.3641748520713031,
      "learning_rate": 6.189155231384869e-06,
      "loss": 1.3364,
      "step": 2911
    },
    {
      "epoch": 0.745472,
      "grad_norm": 0.389141505359334,
      "learning_rate": 6.17740454670255e-06,
      "loss": 1.364,
      "step": 2912
    },
    {
      "epoch": 0.745728,
      "grad_norm": 0.36840891706486234,
      "learning_rate": 6.165662989809802e-06,
      "loss": 1.3252,
      "step": 2913
    },
    {
      "epoch": 0.745984,
      "grad_norm": 0.4772291668007155,
      "learning_rate": 6.1539305684602e-06,
      "loss": 1.3209,
      "step": 2914
    },
    {
      "epoch": 0.74624,
      "grad_norm": 0.3596862934891284,
      "learning_rate": 6.1422072904012855e-06,
      "loss": 1.3303,
      "step": 2915
    },
    {
      "epoch": 0.746496,
      "grad_norm": 0.3658366689634869,
      "learning_rate": 6.130493163374562e-06,
      "loss": 1.3529,
      "step": 2916
    },
    {
      "epoch": 0.746752,
      "grad_norm": 0.3701227665095814,
      "learning_rate": 6.1187881951154745e-06,
      "loss": 1.3091,
      "step": 2917
    },
    {
      "epoch": 0.747008,
      "grad_norm": 0.3592989037539962,
      "learning_rate": 6.107092393353436e-06,
      "loss": 1.3253,
      "step": 2918
    },
    {
      "epoch": 0.747264,
      "grad_norm": 0.3570047933982698,
      "learning_rate": 6.095405765811804e-06,
      "loss": 1.345,
      "step": 2919
    },
    {
      "epoch": 0.74752,
      "grad_norm": 0.356744672855946,
      "learning_rate": 6.083728320207873e-06,
      "loss": 1.3315,
      "step": 2920
    },
    {
      "epoch": 0.747776,
      "grad_norm": 0.3666243470858031,
      "learning_rate": 6.072060064252878e-06,
      "loss": 1.3365,
      "step": 2921
    },
    {
      "epoch": 0.748032,
      "grad_norm": 0.36308971719881444,
      "learning_rate": 6.060401005651992e-06,
      "loss": 1.3555,
      "step": 2922
    },
    {
      "epoch": 0.748288,
      "grad_norm": 0.3874423844897525,
      "learning_rate": 6.0487511521042954e-06,
      "loss": 1.3359,
      "step": 2923
    },
    {
      "epoch": 0.748544,
      "grad_norm": 0.3673286927353204,
      "learning_rate": 6.037110511302811e-06,
      "loss": 1.3475,
      "step": 2924
    },
    {
      "epoch": 0.7488,
      "grad_norm": 0.3751307398897808,
      "learning_rate": 6.025479090934456e-06,
      "loss": 1.3532,
      "step": 2925
    },
    {
      "epoch": 0.749056,
      "grad_norm": 0.35812236855672913,
      "learning_rate": 6.013856898680081e-06,
      "loss": 1.2774,
      "step": 2926
    },
    {
      "epoch": 0.749312,
      "grad_norm": 0.36090308788878,
      "learning_rate": 6.00224394221443e-06,
      "loss": 1.2854,
      "step": 2927
    },
    {
      "epoch": 0.749568,
      "grad_norm": 0.36338299449641465,
      "learning_rate": 5.990640229206159e-06,
      "loss": 1.325,
      "step": 2928
    },
    {
      "epoch": 0.749824,
      "grad_norm": 0.36376627624726776,
      "learning_rate": 5.9790457673177995e-06,
      "loss": 1.3271,
      "step": 2929
    },
    {
      "epoch": 0.75008,
      "grad_norm": 0.36133410891253326,
      "learning_rate": 5.9674605642057914e-06,
      "loss": 1.3238,
      "step": 2930
    },
    {
      "epoch": 0.750336,
      "grad_norm": 0.3606896741827976,
      "learning_rate": 5.9558846275204605e-06,
      "loss": 1.3031,
      "step": 2931
    },
    {
      "epoch": 0.750592,
      "grad_norm": 0.3576954878141839,
      "learning_rate": 5.944317964906004e-06,
      "loss": 1.3183,
      "step": 2932
    },
    {
      "epoch": 0.750848,
      "grad_norm": 0.36007885955600133,
      "learning_rate": 5.932760584000509e-06,
      "loss": 1.3381,
      "step": 2933
    },
    {
      "epoch": 0.751104,
      "grad_norm": 0.3611404133984041,
      "learning_rate": 5.921212492435913e-06,
      "loss": 1.3308,
      "step": 2934
    },
    {
      "epoch": 0.75136,
      "grad_norm": 0.36027785631795367,
      "learning_rate": 5.9096736978380384e-06,
      "loss": 1.2989,
      "step": 2935
    },
    {
      "epoch": 0.751616,
      "grad_norm": 0.36571115154734557,
      "learning_rate": 5.898144207826553e-06,
      "loss": 1.333,
      "step": 2936
    },
    {
      "epoch": 0.751872,
      "grad_norm": 0.369863640491649,
      "learning_rate": 5.886624030014989e-06,
      "loss": 1.3047,
      "step": 2937
    },
    {
      "epoch": 0.752128,
      "grad_norm": 0.35356179798573417,
      "learning_rate": 5.875113172010733e-06,
      "loss": 1.3057,
      "step": 2938
    },
    {
      "epoch": 0.752384,
      "grad_norm": 0.36632536065659166,
      "learning_rate": 5.86361164141501e-06,
      "loss": 1.3205,
      "step": 2939
    },
    {
      "epoch": 0.75264,
      "grad_norm": 0.35460903004533356,
      "learning_rate": 5.852119445822895e-06,
      "loss": 1.3417,
      "step": 2940
    },
    {
      "epoch": 0.752896,
      "grad_norm": 0.3612903572922312,
      "learning_rate": 5.8406365928232764e-06,
      "loss": 1.3426,
      "step": 2941
    },
    {
      "epoch": 0.753152,
      "grad_norm": 0.35598890963150864,
      "learning_rate": 5.8291630899988995e-06,
      "loss": 1.3225,
      "step": 2942
    },
    {
      "epoch": 0.753408,
      "grad_norm": 0.3606739649879641,
      "learning_rate": 5.8176989449263176e-06,
      "loss": 1.3255,
      "step": 2943
    },
    {
      "epoch": 0.753664,
      "grad_norm": 0.3598927715313153,
      "learning_rate": 5.8062441651759165e-06,
      "loss": 1.303,
      "step": 2944
    },
    {
      "epoch": 0.75392,
      "grad_norm": 0.3714596240047786,
      "learning_rate": 5.794798758311894e-06,
      "loss": 1.3658,
      "step": 2945
    },
    {
      "epoch": 0.754176,
      "grad_norm": 0.3535270033350636,
      "learning_rate": 5.783362731892248e-06,
      "loss": 1.2817,
      "step": 2946
    },
    {
      "epoch": 0.754432,
      "grad_norm": 0.3542279010133975,
      "learning_rate": 5.7719360934688015e-06,
      "loss": 1.2832,
      "step": 2947
    },
    {
      "epoch": 0.754688,
      "grad_norm": 0.3589351354008305,
      "learning_rate": 5.760518850587154e-06,
      "loss": 1.2969,
      "step": 2948
    },
    {
      "epoch": 0.754944,
      "grad_norm": 0.3586419025829876,
      "learning_rate": 5.749111010786721e-06,
      "loss": 1.3182,
      "step": 2949
    },
    {
      "epoch": 0.7552,
      "grad_norm": 0.36149707658334906,
      "learning_rate": 5.7377125816007005e-06,
      "loss": 1.3072,
      "step": 2950
    },
    {
      "epoch": 0.755456,
      "grad_norm": 0.3631350656592278,
      "learning_rate": 5.726323570556076e-06,
      "loss": 1.3345,
      "step": 2951
    },
    {
      "epoch": 0.755712,
      "grad_norm": 0.3630057715266046,
      "learning_rate": 5.714943985173622e-06,
      "loss": 1.3368,
      "step": 2952
    },
    {
      "epoch": 0.755968,
      "grad_norm": 0.3674003743456213,
      "learning_rate": 5.7035738329678654e-06,
      "loss": 1.3475,
      "step": 2953
    },
    {
      "epoch": 0.756224,
      "grad_norm": 0.4908129732031185,
      "learning_rate": 5.692213121447126e-06,
      "loss": 1.3553,
      "step": 2954
    },
    {
      "epoch": 0.75648,
      "grad_norm": 0.36788077649295653,
      "learning_rate": 5.680861858113476e-06,
      "loss": 1.3499,
      "step": 2955
    },
    {
      "epoch": 0.756736,
      "grad_norm": 0.35557312197856145,
      "learning_rate": 5.6695200504627665e-06,
      "loss": 1.2952,
      "step": 2956
    },
    {
      "epoch": 0.756992,
      "grad_norm": 0.357811886157127,
      "learning_rate": 5.658187705984577e-06,
      "loss": 1.294,
      "step": 2957
    },
    {
      "epoch": 0.757248,
      "grad_norm": 0.3618713217685472,
      "learning_rate": 5.646864832162264e-06,
      "loss": 1.294,
      "step": 2958
    },
    {
      "epoch": 0.757504,
      "grad_norm": 0.3689907926411865,
      "learning_rate": 5.635551436472908e-06,
      "loss": 1.3435,
      "step": 2959
    },
    {
      "epoch": 0.75776,
      "grad_norm": 0.3607877470701937,
      "learning_rate": 5.62424752638735e-06,
      "loss": 1.2987,
      "step": 2960
    },
    {
      "epoch": 0.758016,
      "grad_norm": 0.35904133610852673,
      "learning_rate": 5.612953109370154e-06,
      "loss": 1.3131,
      "step": 2961
    },
    {
      "epoch": 0.758272,
      "grad_norm": 0.3592936221385696,
      "learning_rate": 5.601668192879622e-06,
      "loss": 1.3319,
      "step": 2962
    },
    {
      "epoch": 0.758528,
      "grad_norm": 0.37066696377475744,
      "learning_rate": 5.59039278436778e-06,
      "loss": 1.3344,
      "step": 2963
    },
    {
      "epoch": 0.758784,
      "grad_norm": 0.36033752899405347,
      "learning_rate": 5.579126891280382e-06,
      "loss": 1.3441,
      "step": 2964
    },
    {
      "epoch": 0.75904,
      "grad_norm": 0.36039021829793527,
      "learning_rate": 5.56787052105688e-06,
      "loss": 1.3562,
      "step": 2965
    },
    {
      "epoch": 0.759296,
      "grad_norm": 0.3600361299524374,
      "learning_rate": 5.556623681130453e-06,
      "loss": 1.3199,
      "step": 2966
    },
    {
      "epoch": 0.759552,
      "grad_norm": 0.35254370864844703,
      "learning_rate": 5.545386378927988e-06,
      "loss": 1.2876,
      "step": 2967
    },
    {
      "epoch": 0.759808,
      "grad_norm": 0.40244468564546243,
      "learning_rate": 5.534158621870071e-06,
      "loss": 1.3081,
      "step": 2968
    },
    {
      "epoch": 0.760064,
      "grad_norm": 0.38579074021309034,
      "learning_rate": 5.522940417370975e-06,
      "loss": 1.3377,
      "step": 2969
    },
    {
      "epoch": 0.76032,
      "grad_norm": 0.3670230014473168,
      "learning_rate": 5.51173177283868e-06,
      "loss": 1.3504,
      "step": 2970
    },
    {
      "epoch": 0.760576,
      "grad_norm": 0.37103367905305157,
      "learning_rate": 5.500532695674836e-06,
      "loss": 1.2842,
      "step": 2971
    },
    {
      "epoch": 0.760832,
      "grad_norm": 0.3531156073938096,
      "learning_rate": 5.48934319327479e-06,
      "loss": 1.2919,
      "step": 2972
    },
    {
      "epoch": 0.761088,
      "grad_norm": 0.3604569565213273,
      "learning_rate": 5.478163273027561e-06,
      "loss": 1.3262,
      "step": 2973
    },
    {
      "epoch": 0.761344,
      "grad_norm": 0.35510970486958565,
      "learning_rate": 5.466992942315843e-06,
      "loss": 1.317,
      "step": 2974
    },
    {
      "epoch": 0.7616,
      "grad_norm": 0.36493346935104093,
      "learning_rate": 5.455832208515994e-06,
      "loss": 1.3428,
      "step": 2975
    },
    {
      "epoch": 0.761856,
      "grad_norm": 0.36089640883655555,
      "learning_rate": 5.444681078998035e-06,
      "loss": 1.3137,
      "step": 2976
    },
    {
      "epoch": 0.762112,
      "grad_norm": 0.3513323364892846,
      "learning_rate": 5.433539561125652e-06,
      "loss": 1.2841,
      "step": 2977
    },
    {
      "epoch": 0.762368,
      "grad_norm": 0.5213081392785177,
      "learning_rate": 5.4224076622561665e-06,
      "loss": 1.3131,
      "step": 2978
    },
    {
      "epoch": 0.762624,
      "grad_norm": 0.36112417976360284,
      "learning_rate": 5.411285389740568e-06,
      "loss": 1.2842,
      "step": 2979
    },
    {
      "epoch": 0.76288,
      "grad_norm": 0.36975607279547196,
      "learning_rate": 5.400172750923472e-06,
      "loss": 1.3382,
      "step": 2980
    },
    {
      "epoch": 0.763136,
      "grad_norm": 0.35979062094475533,
      "learning_rate": 5.389069753143143e-06,
      "loss": 1.3465,
      "step": 2981
    },
    {
      "epoch": 0.763392,
      "grad_norm": 0.35835454438140196,
      "learning_rate": 5.37797640373148e-06,
      "loss": 1.2973,
      "step": 2982
    },
    {
      "epoch": 0.763648,
      "grad_norm": 0.3593876419585883,
      "learning_rate": 5.3668927100140136e-06,
      "loss": 1.3099,
      "step": 2983
    },
    {
      "epoch": 0.763904,
      "grad_norm": 0.3575624526026183,
      "learning_rate": 5.3558186793098744e-06,
      "loss": 1.2857,
      "step": 2984
    },
    {
      "epoch": 0.76416,
      "grad_norm": 0.3687036099013438,
      "learning_rate": 5.344754318931842e-06,
      "loss": 1.3453,
      "step": 2985
    },
    {
      "epoch": 0.764416,
      "grad_norm": 0.37214103097873885,
      "learning_rate": 5.333699636186296e-06,
      "loss": 1.3811,
      "step": 2986
    },
    {
      "epoch": 0.764672,
      "grad_norm": 0.4013738907885611,
      "learning_rate": 5.322654638373224e-06,
      "loss": 1.3639,
      "step": 2987
    },
    {
      "epoch": 0.764928,
      "grad_norm": 0.3525476526160913,
      "learning_rate": 5.311619332786224e-06,
      "loss": 1.3056,
      "step": 2988
    },
    {
      "epoch": 0.765184,
      "grad_norm": 0.35760920276407876,
      "learning_rate": 5.300593726712497e-06,
      "loss": 1.3149,
      "step": 2989
    },
    {
      "epoch": 0.76544,
      "grad_norm": 0.4102311154254802,
      "learning_rate": 5.289577827432819e-06,
      "loss": 1.3621,
      "step": 2990
    },
    {
      "epoch": 0.765696,
      "grad_norm": 0.35748288581948084,
      "learning_rate": 5.278571642221584e-06,
      "loss": 1.3005,
      "step": 2991
    },
    {
      "epoch": 0.765952,
      "grad_norm": 0.3540493531722397,
      "learning_rate": 5.2675751783467445e-06,
      "loss": 1.2995,
      "step": 2992
    },
    {
      "epoch": 0.766208,
      "grad_norm": 0.3683871517837391,
      "learning_rate": 5.256588443069853e-06,
      "loss": 1.3318,
      "step": 2993
    },
    {
      "epoch": 0.766464,
      "grad_norm": 0.35562426147612786,
      "learning_rate": 5.245611443646028e-06,
      "loss": 1.3032,
      "step": 2994
    },
    {
      "epoch": 0.76672,
      "grad_norm": 0.3610374896117163,
      "learning_rate": 5.23464418732397e-06,
      "loss": 1.3595,
      "step": 2995
    },
    {
      "epoch": 0.766976,
      "grad_norm": 0.3534251717563875,
      "learning_rate": 5.223686681345926e-06,
      "loss": 1.3228,
      "step": 2996
    },
    {
      "epoch": 0.767232,
      "grad_norm": 0.3628044160204886,
      "learning_rate": 5.212738932947718e-06,
      "loss": 1.3188,
      "step": 2997
    },
    {
      "epoch": 0.767488,
      "grad_norm": 0.3652190338148601,
      "learning_rate": 5.2018009493587285e-06,
      "loss": 1.3328,
      "step": 2998
    },
    {
      "epoch": 0.767744,
      "grad_norm": 0.3621398562741921,
      "learning_rate": 5.1908727378018796e-06,
      "loss": 1.3365,
      "step": 2999
    },
    {
      "epoch": 0.768,
      "grad_norm": 1.7120380965373587,
      "learning_rate": 5.179954305493651e-06,
      "loss": 1.3356,
      "step": 3000
    },
    {
      "epoch": 0.768256,
      "grad_norm": 0.35834378169321823,
      "learning_rate": 5.169045659644061e-06,
      "loss": 1.3285,
      "step": 3001
    },
    {
      "epoch": 0.768512,
      "grad_norm": 0.3610209931988971,
      "learning_rate": 5.158146807456663e-06,
      "loss": 1.2865,
      "step": 3002
    },
    {
      "epoch": 0.768768,
      "grad_norm": 0.3732257662595727,
      "learning_rate": 5.147257756128538e-06,
      "loss": 1.3799,
      "step": 3003
    },
    {
      "epoch": 0.769024,
      "grad_norm": 0.37134882217526066,
      "learning_rate": 5.136378512850304e-06,
      "loss": 1.315,
      "step": 3004
    },
    {
      "epoch": 0.76928,
      "grad_norm": 0.3705311725333239,
      "learning_rate": 5.125509084806104e-06,
      "loss": 1.3327,
      "step": 3005
    },
    {
      "epoch": 0.769536,
      "grad_norm": 0.3637577177691028,
      "learning_rate": 5.114649479173592e-06,
      "loss": 1.3448,
      "step": 3006
    },
    {
      "epoch": 0.769792,
      "grad_norm": 0.3620461827635551,
      "learning_rate": 5.1037997031239485e-06,
      "loss": 1.336,
      "step": 3007
    },
    {
      "epoch": 0.770048,
      "grad_norm": 0.3593466735035229,
      "learning_rate": 5.092959763821836e-06,
      "loss": 1.3295,
      "step": 3008
    },
    {
      "epoch": 0.770304,
      "grad_norm": 0.36375987192045367,
      "learning_rate": 5.082129668425451e-06,
      "loss": 1.3387,
      "step": 3009
    },
    {
      "epoch": 0.77056,
      "grad_norm": 0.3534892536722086,
      "learning_rate": 5.071309424086472e-06,
      "loss": 1.3161,
      "step": 3010
    },
    {
      "epoch": 0.770816,
      "grad_norm": 0.35490045714028706,
      "learning_rate": 5.0604990379500795e-06,
      "loss": 1.3109,
      "step": 3011
    },
    {
      "epoch": 0.771072,
      "grad_norm": 0.35703471750054855,
      "learning_rate": 5.049698517154951e-06,
      "loss": 1.3107,
      "step": 3012
    },
    {
      "epoch": 0.771328,
      "grad_norm": 0.3575705202311698,
      "learning_rate": 5.038907868833225e-06,
      "loss": 1.3066,
      "step": 3013
    },
    {
      "epoch": 0.771584,
      "grad_norm": 0.36708405680932127,
      "learning_rate": 5.0281271001105495e-06,
      "loss": 1.3391,
      "step": 3014
    },
    {
      "epoch": 0.77184,
      "grad_norm": 0.36610052132503385,
      "learning_rate": 5.017356218106022e-06,
      "loss": 1.369,
      "step": 3015
    },
    {
      "epoch": 0.772096,
      "grad_norm": 0.35423563760381266,
      "learning_rate": 5.006595229932234e-06,
      "loss": 1.2638,
      "step": 3016
    },
    {
      "epoch": 0.772352,
      "grad_norm": 0.3745075190053957,
      "learning_rate": 4.99584414269523e-06,
      "loss": 1.3633,
      "step": 3017
    },
    {
      "epoch": 0.772608,
      "grad_norm": 0.377354794170555,
      "learning_rate": 4.985102963494524e-06,
      "loss": 1.35,
      "step": 3018
    },
    {
      "epoch": 0.772864,
      "grad_norm": 0.3607129342072355,
      "learning_rate": 4.974371699423088e-06,
      "loss": 1.3005,
      "step": 3019
    },
    {
      "epoch": 0.77312,
      "grad_norm": 0.36155354105682896,
      "learning_rate": 4.9636503575673315e-06,
      "loss": 1.3372,
      "step": 3020
    },
    {
      "epoch": 0.773376,
      "grad_norm": 0.36597648814147365,
      "learning_rate": 4.952938945007127e-06,
      "loss": 1.3266,
      "step": 3021
    },
    {
      "epoch": 0.773632,
      "grad_norm": 0.3663782796535604,
      "learning_rate": 4.942237468815785e-06,
      "loss": 1.3768,
      "step": 3022
    },
    {
      "epoch": 0.773888,
      "grad_norm": 0.353994081689147,
      "learning_rate": 4.931545936060058e-06,
      "loss": 1.2683,
      "step": 3023
    },
    {
      "epoch": 0.774144,
      "grad_norm": 0.3790368190041,
      "learning_rate": 4.920864353800135e-06,
      "loss": 1.3706,
      "step": 3024
    },
    {
      "epoch": 0.7744,
      "grad_norm": 0.656109240984305,
      "learning_rate": 4.91019272908962e-06,
      "loss": 1.3788,
      "step": 3025
    },
    {
      "epoch": 0.774656,
      "grad_norm": 0.3599626949606184,
      "learning_rate": 4.899531068975547e-06,
      "loss": 1.3111,
      "step": 3026
    },
    {
      "epoch": 0.774912,
      "grad_norm": 0.35429384979386275,
      "learning_rate": 4.888879380498377e-06,
      "loss": 1.2693,
      "step": 3027
    },
    {
      "epoch": 0.775168,
      "grad_norm": 0.3578535066663329,
      "learning_rate": 4.878237670691985e-06,
      "loss": 1.3241,
      "step": 3028
    },
    {
      "epoch": 0.775424,
      "grad_norm": 0.3668962234222316,
      "learning_rate": 4.867605946583653e-06,
      "loss": 1.3518,
      "step": 3029
    },
    {
      "epoch": 0.77568,
      "grad_norm": 0.36596079299208684,
      "learning_rate": 4.856984215194067e-06,
      "loss": 1.3239,
      "step": 3030
    },
    {
      "epoch": 0.775936,
      "grad_norm": 0.36986631828555855,
      "learning_rate": 4.846372483537327e-06,
      "loss": 1.3706,
      "step": 3031
    },
    {
      "epoch": 0.776192,
      "grad_norm": 0.35252381816918543,
      "learning_rate": 4.835770758620906e-06,
      "loss": 1.2767,
      "step": 3032
    },
    {
      "epoch": 0.776448,
      "grad_norm": 0.36031430417787896,
      "learning_rate": 4.8251790474456875e-06,
      "loss": 1.3344,
      "step": 3033
    },
    {
      "epoch": 0.776704,
      "grad_norm": 0.36004813424433374,
      "learning_rate": 4.814597357005939e-06,
      "loss": 1.3104,
      "step": 3034
    },
    {
      "epoch": 0.77696,
      "grad_norm": 0.36729935517547574,
      "learning_rate": 4.804025694289316e-06,
      "loss": 1.3297,
      "step": 3035
    },
    {
      "epoch": 0.777216,
      "grad_norm": 0.3560090275587045,
      "learning_rate": 4.793464066276831e-06,
      "loss": 1.3167,
      "step": 3036
    },
    {
      "epoch": 0.777472,
      "grad_norm": 0.36207999379669126,
      "learning_rate": 4.782912479942894e-06,
      "loss": 1.3114,
      "step": 3037
    },
    {
      "epoch": 0.777728,
      "grad_norm": 0.3591899433769941,
      "learning_rate": 4.7723709422552755e-06,
      "loss": 1.2958,
      "step": 3038
    },
    {
      "epoch": 0.777984,
      "grad_norm": 0.3977382515063168,
      "learning_rate": 4.761839460175104e-06,
      "loss": 1.3696,
      "step": 3039
    },
    {
      "epoch": 0.77824,
      "grad_norm": 0.35620292489734334,
      "learning_rate": 4.751318040656874e-06,
      "loss": 1.3346,
      "step": 3040
    },
    {
      "epoch": 0.778496,
      "grad_norm": 0.3650217523570914,
      "learning_rate": 4.740806690648438e-06,
      "loss": 1.3353,
      "step": 3041
    },
    {
      "epoch": 0.778752,
      "grad_norm": 0.3684425763049362,
      "learning_rate": 4.730305417090992e-06,
      "loss": 1.3276,
      "step": 3042
    },
    {
      "epoch": 0.779008,
      "grad_norm": 0.3712144903317603,
      "learning_rate": 4.719814226919084e-06,
      "loss": 1.3491,
      "step": 3043
    },
    {
      "epoch": 0.779264,
      "grad_norm": 0.3561238382584264,
      "learning_rate": 4.709333127060605e-06,
      "loss": 1.2943,
      "step": 3044
    },
    {
      "epoch": 0.77952,
      "grad_norm": 0.37205110042317996,
      "learning_rate": 4.698862124436767e-06,
      "loss": 1.3687,
      "step": 3045
    },
    {
      "epoch": 0.779776,
      "grad_norm": 0.3695584912763628,
      "learning_rate": 4.6884012259621316e-06,
      "loss": 1.3684,
      "step": 3046
    },
    {
      "epoch": 0.780032,
      "grad_norm": 0.36329097685625406,
      "learning_rate": 4.67795043854459e-06,
      "loss": 1.3214,
      "step": 3047
    },
    {
      "epoch": 0.780288,
      "grad_norm": 0.3798693015592709,
      "learning_rate": 4.667509769085334e-06,
      "loss": 1.3809,
      "step": 3048
    },
    {
      "epoch": 0.780544,
      "grad_norm": 0.3601097934242932,
      "learning_rate": 4.6570792244789e-06,
      "loss": 1.3306,
      "step": 3049
    },
    {
      "epoch": 0.7808,
      "grad_norm": 0.35446125225039316,
      "learning_rate": 4.646658811613127e-06,
      "loss": 1.3347,
      "step": 3050
    },
    {
      "epoch": 0.781056,
      "grad_norm": 0.36405550438257583,
      "learning_rate": 4.636248537369156e-06,
      "loss": 1.3372,
      "step": 3051
    },
    {
      "epoch": 0.781312,
      "grad_norm": 0.3617185366756447,
      "learning_rate": 4.625848408621447e-06,
      "loss": 1.3214,
      "step": 3052
    },
    {
      "epoch": 0.781568,
      "grad_norm": 0.3584437977949366,
      "learning_rate": 4.615458432237751e-06,
      "loss": 1.3416,
      "step": 3053
    },
    {
      "epoch": 0.781824,
      "grad_norm": 0.37686068517129695,
      "learning_rate": 4.6050786150791216e-06,
      "loss": 1.3802,
      "step": 3054
    },
    {
      "epoch": 0.78208,
      "grad_norm": 0.3557512741068298,
      "learning_rate": 4.594708963999897e-06,
      "loss": 1.3275,
      "step": 3055
    },
    {
      "epoch": 0.782336,
      "grad_norm": 0.35954325587381625,
      "learning_rate": 4.58434948584771e-06,
      "loss": 1.33,
      "step": 3056
    },
    {
      "epoch": 0.782592,
      "grad_norm": 0.3599903432148184,
      "learning_rate": 4.574000187463466e-06,
      "loss": 1.3064,
      "step": 3057
    },
    {
      "epoch": 0.782848,
      "grad_norm": 0.35780778909053024,
      "learning_rate": 4.563661075681356e-06,
      "loss": 1.3003,
      "step": 3058
    },
    {
      "epoch": 0.783104,
      "grad_norm": 0.360807357309411,
      "learning_rate": 4.553332157328836e-06,
      "loss": 1.3272,
      "step": 3059
    },
    {
      "epoch": 0.78336,
      "grad_norm": 0.3691923761012712,
      "learning_rate": 4.54301343922664e-06,
      "loss": 1.3378,
      "step": 3060
    },
    {
      "epoch": 0.783616,
      "grad_norm": 0.408705525371534,
      "learning_rate": 4.532704928188763e-06,
      "loss": 1.3532,
      "step": 3061
    },
    {
      "epoch": 0.783872,
      "grad_norm": 0.36928712972583044,
      "learning_rate": 4.522406631022464e-06,
      "loss": 1.3937,
      "step": 3062
    },
    {
      "epoch": 0.784128,
      "grad_norm": 0.3805677952592156,
      "learning_rate": 4.512118554528242e-06,
      "loss": 1.3025,
      "step": 3063
    },
    {
      "epoch": 0.784384,
      "grad_norm": 0.369907891553738,
      "learning_rate": 4.50184070549986e-06,
      "loss": 1.3244,
      "step": 3064
    },
    {
      "epoch": 0.78464,
      "grad_norm": 0.356138064912598,
      "learning_rate": 4.491573090724328e-06,
      "loss": 1.2956,
      "step": 3065
    },
    {
      "epoch": 0.784896,
      "grad_norm": 0.3586200227620091,
      "learning_rate": 4.481315716981891e-06,
      "loss": 1.3312,
      "step": 3066
    },
    {
      "epoch": 0.785152,
      "grad_norm": 0.4061509126427212,
      "learning_rate": 4.4710685910460375e-06,
      "loss": 1.3655,
      "step": 3067
    },
    {
      "epoch": 0.785408,
      "grad_norm": 0.35601729097642537,
      "learning_rate": 4.46083171968349e-06,
      "loss": 1.2744,
      "step": 3068
    },
    {
      "epoch": 0.785664,
      "grad_norm": 0.3655100115357683,
      "learning_rate": 4.450605109654182e-06,
      "loss": 1.317,
      "step": 3069
    },
    {
      "epoch": 0.78592,
      "grad_norm": 0.3761978358436829,
      "learning_rate": 4.440388767711299e-06,
      "loss": 1.3388,
      "step": 3070
    },
    {
      "epoch": 0.786176,
      "grad_norm": 0.3554722523987874,
      "learning_rate": 4.430182700601218e-06,
      "loss": 1.314,
      "step": 3071
    },
    {
      "epoch": 0.786432,
      "grad_norm": 0.3575984379723292,
      "learning_rate": 4.419986915063548e-06,
      "loss": 1.3274,
      "step": 3072
    },
    {
      "epoch": 0.786688,
      "grad_norm": 0.3565129851797361,
      "learning_rate": 4.409801417831105e-06,
      "loss": 1.3432,
      "step": 3073
    },
    {
      "epoch": 0.786944,
      "grad_norm": 0.45537671516641026,
      "learning_rate": 4.399626215629917e-06,
      "loss": 1.3382,
      "step": 3074
    },
    {
      "epoch": 0.7872,
      "grad_norm": 0.35988155865265975,
      "learning_rate": 4.389461315179198e-06,
      "loss": 1.3515,
      "step": 3075
    },
    {
      "epoch": 0.787456,
      "grad_norm": 0.36114324128440006,
      "learning_rate": 4.379306723191372e-06,
      "loss": 1.3569,
      "step": 3076
    },
    {
      "epoch": 0.787712,
      "grad_norm": 0.3673284889231303,
      "learning_rate": 4.369162446372052e-06,
      "loss": 1.3031,
      "step": 3077
    },
    {
      "epoch": 0.787968,
      "grad_norm": 0.35692768806112357,
      "learning_rate": 4.359028491420039e-06,
      "loss": 1.3032,
      "step": 3078
    },
    {
      "epoch": 0.788224,
      "grad_norm": 0.35945720879929977,
      "learning_rate": 4.348904865027321e-06,
      "loss": 1.3131,
      "step": 3079
    },
    {
      "epoch": 0.78848,
      "grad_norm": 0.35461340813246045,
      "learning_rate": 4.3387915738790666e-06,
      "loss": 1.3094,
      "step": 3080
    },
    {
      "epoch": 0.788736,
      "grad_norm": 0.36165300799149924,
      "learning_rate": 4.3286886246536096e-06,
      "loss": 1.3376,
      "step": 3081
    },
    {
      "epoch": 0.788992,
      "grad_norm": 0.38868977657211395,
      "learning_rate": 4.318596024022457e-06,
      "loss": 1.2953,
      "step": 3082
    },
    {
      "epoch": 0.789248,
      "grad_norm": 0.35858624252787263,
      "learning_rate": 4.308513778650292e-06,
      "loss": 1.3337,
      "step": 3083
    },
    {
      "epoch": 0.789504,
      "grad_norm": 0.35453334381185514,
      "learning_rate": 4.298441895194952e-06,
      "loss": 1.3027,
      "step": 3084
    },
    {
      "epoch": 0.78976,
      "grad_norm": 0.3505993486544977,
      "learning_rate": 4.288380380307433e-06,
      "loss": 1.2952,
      "step": 3085
    },
    {
      "epoch": 0.790016,
      "grad_norm": 0.3594786774181754,
      "learning_rate": 4.278329240631891e-06,
      "loss": 1.2943,
      "step": 3086
    },
    {
      "epoch": 0.790272,
      "grad_norm": 0.35702731393577514,
      "learning_rate": 4.268288482805614e-06,
      "loss": 1.3134,
      "step": 3087
    },
    {
      "epoch": 0.790528,
      "grad_norm": 0.36179036682440446,
      "learning_rate": 4.2582581134590465e-06,
      "loss": 1.3464,
      "step": 3088
    },
    {
      "epoch": 0.790784,
      "grad_norm": 0.37112645772937886,
      "learning_rate": 4.248238139215775e-06,
      "loss": 1.3394,
      "step": 3089
    },
    {
      "epoch": 0.79104,
      "grad_norm": 0.3524229502227345,
      "learning_rate": 4.238228566692517e-06,
      "loss": 1.3004,
      "step": 3090
    },
    {
      "epoch": 0.791296,
      "grad_norm": 0.3700762586081203,
      "learning_rate": 4.2282294024991245e-06,
      "loss": 1.3474,
      "step": 3091
    },
    {
      "epoch": 0.791552,
      "grad_norm": 0.3707112495739379,
      "learning_rate": 4.218240653238566e-06,
      "loss": 1.3368,
      "step": 3092
    },
    {
      "epoch": 0.791808,
      "grad_norm": 0.3731556688911466,
      "learning_rate": 4.208262325506946e-06,
      "loss": 1.3325,
      "step": 3093
    },
    {
      "epoch": 0.792064,
      "grad_norm": 0.6725157802416758,
      "learning_rate": 4.198294425893476e-06,
      "loss": 1.3339,
      "step": 3094
    },
    {
      "epoch": 0.79232,
      "grad_norm": 0.3592065906070425,
      "learning_rate": 4.18833696098049e-06,
      "loss": 1.3136,
      "step": 3095
    },
    {
      "epoch": 0.792576,
      "grad_norm": 0.36647542390567395,
      "learning_rate": 4.178389937343425e-06,
      "loss": 1.3224,
      "step": 3096
    },
    {
      "epoch": 0.792832,
      "grad_norm": 0.36374453954357416,
      "learning_rate": 4.16845336155083e-06,
      "loss": 1.3099,
      "step": 3097
    },
    {
      "epoch": 0.793088,
      "grad_norm": 0.36455629125768607,
      "learning_rate": 4.158527240164347e-06,
      "loss": 1.3394,
      "step": 3098
    },
    {
      "epoch": 0.793344,
      "grad_norm": 0.36534109842783513,
      "learning_rate": 4.148611579738726e-06,
      "loss": 1.3744,
      "step": 3099
    },
    {
      "epoch": 0.7936,
      "grad_norm": 0.3721046078308637,
      "learning_rate": 4.138706386821789e-06,
      "loss": 1.3546,
      "step": 3100
    },
    {
      "epoch": 0.793856,
      "grad_norm": 0.35627732268861445,
      "learning_rate": 4.128811667954464e-06,
      "loss": 1.3114,
      "step": 3101
    },
    {
      "epoch": 0.794112,
      "grad_norm": 0.3681762149827043,
      "learning_rate": 4.118927429670756e-06,
      "loss": 1.3422,
      "step": 3102
    },
    {
      "epoch": 0.794368,
      "grad_norm": 0.3993286289146559,
      "learning_rate": 4.109053678497754e-06,
      "loss": 1.3146,
      "step": 3103
    },
    {
      "epoch": 0.794624,
      "grad_norm": 0.36074878999159404,
      "learning_rate": 4.099190420955607e-06,
      "loss": 1.3379,
      "step": 3104
    },
    {
      "epoch": 0.79488,
      "grad_norm": 0.3568809961976823,
      "learning_rate": 4.089337663557558e-06,
      "loss": 1.3019,
      "step": 3105
    },
    {
      "epoch": 0.795136,
      "grad_norm": 0.3769418321764292,
      "learning_rate": 4.079495412809886e-06,
      "loss": 1.3586,
      "step": 3106
    },
    {
      "epoch": 0.795392,
      "grad_norm": 0.36594227241105376,
      "learning_rate": 4.069663675211961e-06,
      "loss": 1.3327,
      "step": 3107
    },
    {
      "epoch": 0.795648,
      "grad_norm": 0.4714751645666523,
      "learning_rate": 4.059842457256193e-06,
      "loss": 1.3036,
      "step": 3108
    },
    {
      "epoch": 0.795904,
      "grad_norm": 0.3664049636985469,
      "learning_rate": 4.050031765428055e-06,
      "loss": 1.3418,
      "step": 3109
    },
    {
      "epoch": 0.79616,
      "grad_norm": 0.360443563845632,
      "learning_rate": 4.040231606206062e-06,
      "loss": 1.3159,
      "step": 3110
    },
    {
      "epoch": 0.796416,
      "grad_norm": 0.363910078885372,
      "learning_rate": 4.0304419860617835e-06,
      "loss": 1.347,
      "step": 3111
    },
    {
      "epoch": 0.796672,
      "grad_norm": 0.3670346736568895,
      "learning_rate": 4.020662911459812e-06,
      "loss": 1.3349,
      "step": 3112
    },
    {
      "epoch": 0.796928,
      "grad_norm": 0.3841657840535063,
      "learning_rate": 4.010894388857789e-06,
      "loss": 1.3613,
      "step": 3113
    },
    {
      "epoch": 0.797184,
      "grad_norm": 0.3593177124527256,
      "learning_rate": 4.0011364247063955e-06,
      "loss": 1.2885,
      "step": 3114
    },
    {
      "epoch": 0.79744,
      "grad_norm": 0.3525040279886287,
      "learning_rate": 3.991389025449317e-06,
      "loss": 1.3016,
      "step": 3115
    },
    {
      "epoch": 0.797696,
      "grad_norm": 0.37989940121636073,
      "learning_rate": 3.9816521975232805e-06,
      "loss": 1.3253,
      "step": 3116
    },
    {
      "epoch": 0.797952,
      "grad_norm": 0.3729689130394653,
      "learning_rate": 3.971925947358035e-06,
      "loss": 1.371,
      "step": 3117
    },
    {
      "epoch": 0.798208,
      "grad_norm": 0.3589802435167401,
      "learning_rate": 3.962210281376326e-06,
      "loss": 1.3222,
      "step": 3118
    },
    {
      "epoch": 0.798464,
      "grad_norm": 0.35248666608324913,
      "learning_rate": 3.952505205993926e-06,
      "loss": 1.301,
      "step": 3119
    },
    {
      "epoch": 0.79872,
      "grad_norm": 0.3552866384274521,
      "learning_rate": 3.942810727619608e-06,
      "loss": 1.3053,
      "step": 3120
    },
    {
      "epoch": 0.798976,
      "grad_norm": 0.3654536331927668,
      "learning_rate": 3.933126852655147e-06,
      "loss": 1.3536,
      "step": 3121
    },
    {
      "epoch": 0.799232,
      "grad_norm": 0.3540552016708653,
      "learning_rate": 3.923453587495319e-06,
      "loss": 1.3106,
      "step": 3122
    },
    {
      "epoch": 0.799488,
      "grad_norm": 0.3645784152942407,
      "learning_rate": 3.913790938527897e-06,
      "loss": 1.3593,
      "step": 3123
    },
    {
      "epoch": 0.799744,
      "grad_norm": 1.0262478738957521,
      "learning_rate": 3.9041389121336255e-06,
      "loss": 1.325,
      "step": 3124
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.36324752042476083,
      "learning_rate": 3.894497514686255e-06,
      "loss": 1.3634,
      "step": 3125
    },
    {
      "epoch": 0.800256,
      "grad_norm": 0.35404200570858796,
      "learning_rate": 3.8848667525525115e-06,
      "loss": 1.3083,
      "step": 3126
    },
    {
      "epoch": 0.800512,
      "grad_norm": 0.3594512835939081,
      "learning_rate": 3.875246632092089e-06,
      "loss": 1.3372,
      "step": 3127
    },
    {
      "epoch": 0.800768,
      "grad_norm": 0.3592508194905306,
      "learning_rate": 3.865637159657662e-06,
      "loss": 1.3265,
      "step": 3128
    },
    {
      "epoch": 0.801024,
      "grad_norm": 0.3584998620531033,
      "learning_rate": 3.856038341594883e-06,
      "loss": 1.3534,
      "step": 3129
    },
    {
      "epoch": 0.80128,
      "grad_norm": 0.3655001505769903,
      "learning_rate": 3.846450184242343e-06,
      "loss": 1.3045,
      "step": 3130
    },
    {
      "epoch": 0.801536,
      "grad_norm": 0.35884368539448896,
      "learning_rate": 3.836872693931615e-06,
      "loss": 1.359,
      "step": 3131
    },
    {
      "epoch": 0.801792,
      "grad_norm": 0.35540195677178904,
      "learning_rate": 3.827305876987224e-06,
      "loss": 1.3314,
      "step": 3132
    },
    {
      "epoch": 0.802048,
      "grad_norm": 0.9087730178002286,
      "learning_rate": 3.817749739726642e-06,
      "loss": 1.371,
      "step": 3133
    },
    {
      "epoch": 0.802304,
      "grad_norm": 0.3710648127870917,
      "learning_rate": 3.808204288460289e-06,
      "loss": 1.3496,
      "step": 3134
    },
    {
      "epoch": 0.80256,
      "grad_norm": 0.36483582362133865,
      "learning_rate": 3.798669529491541e-06,
      "loss": 1.2893,
      "step": 3135
    },
    {
      "epoch": 0.802816,
      "grad_norm": 0.3629681329191201,
      "learning_rate": 3.7891454691166885e-06,
      "loss": 1.3251,
      "step": 3136
    },
    {
      "epoch": 0.803072,
      "grad_norm": 0.35627904565012886,
      "learning_rate": 3.7796321136249825e-06,
      "loss": 1.3355,
      "step": 3137
    },
    {
      "epoch": 0.803328,
      "grad_norm": 0.36556979995829236,
      "learning_rate": 3.7701294692985844e-06,
      "loss": 1.3245,
      "step": 3138
    },
    {
      "epoch": 0.803584,
      "grad_norm": 0.35579056682486027,
      "learning_rate": 3.7606375424125953e-06,
      "loss": 1.3199,
      "step": 3139
    },
    {
      "epoch": 0.80384,
      "grad_norm": 0.35310407482070594,
      "learning_rate": 3.751156339235038e-06,
      "loss": 1.3081,
      "step": 3140
    },
    {
      "epoch": 0.804096,
      "grad_norm": 0.35767516665581484,
      "learning_rate": 3.7416858660268563e-06,
      "loss": 1.2938,
      "step": 3141
    },
    {
      "epoch": 0.804352,
      "grad_norm": 0.36217721165714883,
      "learning_rate": 3.7322261290418915e-06,
      "loss": 1.3282,
      "step": 3142
    },
    {
      "epoch": 0.804608,
      "grad_norm": 0.3619076136991956,
      "learning_rate": 3.722777134526914e-06,
      "loss": 1.3362,
      "step": 3143
    },
    {
      "epoch": 0.804864,
      "grad_norm": 0.35390781125231585,
      "learning_rate": 3.713338888721594e-06,
      "loss": 1.3034,
      "step": 3144
    },
    {
      "epoch": 0.80512,
      "grad_norm": 0.3972857130932456,
      "learning_rate": 3.7039113978585017e-06,
      "loss": 1.3285,
      "step": 3145
    },
    {
      "epoch": 0.805376,
      "grad_norm": 0.3710543134383761,
      "learning_rate": 3.694494668163109e-06,
      "loss": 1.3393,
      "step": 3146
    },
    {
      "epoch": 0.805632,
      "grad_norm": 0.3549013195571846,
      "learning_rate": 3.685088705853783e-06,
      "loss": 1.3023,
      "step": 3147
    },
    {
      "epoch": 0.805888,
      "grad_norm": 0.35597383801760296,
      "learning_rate": 3.6756935171417695e-06,
      "loss": 1.3135,
      "step": 3148
    },
    {
      "epoch": 0.806144,
      "grad_norm": 0.3618699620579227,
      "learning_rate": 3.6663091082312183e-06,
      "loss": 1.3352,
      "step": 3149
    },
    {
      "epoch": 0.8064,
      "grad_norm": 0.36112201626608265,
      "learning_rate": 3.6569354853191353e-06,
      "loss": 1.3084,
      "step": 3150
    },
    {
      "epoch": 0.806656,
      "grad_norm": 0.3639824258678471,
      "learning_rate": 3.6475726545954283e-06,
      "loss": 1.3489,
      "step": 3151
    },
    {
      "epoch": 0.806912,
      "grad_norm": 0.3524328259369818,
      "learning_rate": 3.6382206222428674e-06,
      "loss": 1.305,
      "step": 3152
    },
    {
      "epoch": 0.807168,
      "grad_norm": 0.3582988998964472,
      "learning_rate": 3.628879394437095e-06,
      "loss": 1.3129,
      "step": 3153
    },
    {
      "epoch": 0.807424,
      "grad_norm": 0.3492270683518275,
      "learning_rate": 3.6195489773466117e-06,
      "loss": 1.2985,
      "step": 3154
    },
    {
      "epoch": 0.80768,
      "grad_norm": 0.35918442726578187,
      "learning_rate": 3.610229377132788e-06,
      "loss": 1.3109,
      "step": 3155
    },
    {
      "epoch": 0.807936,
      "grad_norm": 0.3483311741768826,
      "learning_rate": 3.6009205999498488e-06,
      "loss": 1.2635,
      "step": 3156
    },
    {
      "epoch": 0.808192,
      "grad_norm": 0.3576191488662313,
      "learning_rate": 3.5916226519448684e-06,
      "loss": 1.3471,
      "step": 3157
    },
    {
      "epoch": 0.808448,
      "grad_norm": 0.35773411100695934,
      "learning_rate": 3.5823355392577795e-06,
      "loss": 1.3014,
      "step": 3158
    },
    {
      "epoch": 0.808704,
      "grad_norm": 0.3523857178861155,
      "learning_rate": 3.5730592680213437e-06,
      "loss": 1.2739,
      "step": 3159
    },
    {
      "epoch": 0.80896,
      "grad_norm": 0.3547318520331463,
      "learning_rate": 3.563793844361183e-06,
      "loss": 1.3088,
      "step": 3160
    },
    {
      "epoch": 0.809216,
      "grad_norm": 0.35638832452938773,
      "learning_rate": 3.554539274395734e-06,
      "loss": 1.3276,
      "step": 3161
    },
    {
      "epoch": 0.809472,
      "grad_norm": 0.3700067930580701,
      "learning_rate": 3.5452955642362817e-06,
      "loss": 1.2674,
      "step": 3162
    },
    {
      "epoch": 0.809728,
      "grad_norm": 0.35053562894613155,
      "learning_rate": 3.536062719986939e-06,
      "loss": 1.2571,
      "step": 3163
    },
    {
      "epoch": 0.809984,
      "grad_norm": 0.360168570437748,
      "learning_rate": 3.5268407477446397e-06,
      "loss": 1.3479,
      "step": 3164
    },
    {
      "epoch": 0.81024,
      "grad_norm": 0.3601469099028122,
      "learning_rate": 3.5176296535991393e-06,
      "loss": 1.3302,
      "step": 3165
    },
    {
      "epoch": 0.810496,
      "grad_norm": 0.3505621329869148,
      "learning_rate": 3.508429443633012e-06,
      "loss": 1.2728,
      "step": 3166
    },
    {
      "epoch": 0.810752,
      "grad_norm": 0.35654691884651646,
      "learning_rate": 3.4992401239216344e-06,
      "loss": 1.3426,
      "step": 3167
    },
    {
      "epoch": 0.811008,
      "grad_norm": 0.3481711831110968,
      "learning_rate": 3.490061700533205e-06,
      "loss": 1.3078,
      "step": 3168
    },
    {
      "epoch": 0.811264,
      "grad_norm": 0.3575555633068931,
      "learning_rate": 3.48089417952872e-06,
      "loss": 1.321,
      "step": 3169
    },
    {
      "epoch": 0.81152,
      "grad_norm": 0.3563789857495105,
      "learning_rate": 3.4717375669619836e-06,
      "loss": 1.3228,
      "step": 3170
    },
    {
      "epoch": 0.811776,
      "grad_norm": 0.353567446602589,
      "learning_rate": 3.462591868879579e-06,
      "loss": 1.3118,
      "step": 3171
    },
    {
      "epoch": 0.812032,
      "grad_norm": 0.364570510846547,
      "learning_rate": 3.4534570913209064e-06,
      "loss": 1.3662,
      "step": 3172
    },
    {
      "epoch": 0.812288,
      "grad_norm": 0.36135528057612365,
      "learning_rate": 3.4443332403181274e-06,
      "loss": 1.3331,
      "step": 3173
    },
    {
      "epoch": 0.812544,
      "grad_norm": 0.36456315571620285,
      "learning_rate": 3.435220321896211e-06,
      "loss": 1.3717,
      "step": 3174
    },
    {
      "epoch": 0.8128,
      "grad_norm": 0.36902814769745346,
      "learning_rate": 3.4261183420728973e-06,
      "loss": 1.3518,
      "step": 3175
    },
    {
      "epoch": 0.813056,
      "grad_norm": 0.3662245019399536,
      "learning_rate": 3.4170273068587044e-06,
      "loss": 1.3101,
      "step": 3176
    },
    {
      "epoch": 0.813312,
      "grad_norm": 0.3712202445845513,
      "learning_rate": 3.407947222256922e-06,
      "loss": 1.3775,
      "step": 3177
    },
    {
      "epoch": 0.813568,
      "grad_norm": 0.3621948131073107,
      "learning_rate": 3.398878094263616e-06,
      "loss": 1.3331,
      "step": 3178
    },
    {
      "epoch": 0.813824,
      "grad_norm": 0.3548347908253831,
      "learning_rate": 3.3898199288675972e-06,
      "loss": 1.3433,
      "step": 3179
    },
    {
      "epoch": 0.81408,
      "grad_norm": 0.35195107350278804,
      "learning_rate": 3.380772732050459e-06,
      "loss": 1.3076,
      "step": 3180
    },
    {
      "epoch": 0.814336,
      "grad_norm": 0.357501334363723,
      "learning_rate": 3.371736509786547e-06,
      "loss": 1.3289,
      "step": 3181
    },
    {
      "epoch": 0.814592,
      "grad_norm": 0.36166807012869606,
      "learning_rate": 3.362711268042944e-06,
      "loss": 1.3547,
      "step": 3182
    },
    {
      "epoch": 0.814848,
      "grad_norm": 0.3548140991611999,
      "learning_rate": 3.353697012779502e-06,
      "loss": 1.3021,
      "step": 3183
    },
    {
      "epoch": 0.815104,
      "grad_norm": 0.3568677865210114,
      "learning_rate": 3.344693749948813e-06,
      "loss": 1.3237,
      "step": 3184
    },
    {
      "epoch": 0.81536,
      "grad_norm": 0.35965158273244024,
      "learning_rate": 3.335701485496197e-06,
      "loss": 1.3507,
      "step": 3185
    },
    {
      "epoch": 0.815616,
      "grad_norm": 0.371103105303262,
      "learning_rate": 3.326720225359723e-06,
      "loss": 1.3272,
      "step": 3186
    },
    {
      "epoch": 0.815872,
      "grad_norm": 0.352526575851749,
      "learning_rate": 3.317749975470197e-06,
      "loss": 1.2738,
      "step": 3187
    },
    {
      "epoch": 0.816128,
      "grad_norm": 0.36596250426269034,
      "learning_rate": 3.3087907417511443e-06,
      "loss": 1.3588,
      "step": 3188
    },
    {
      "epoch": 0.816384,
      "grad_norm": 0.35611919763538197,
      "learning_rate": 3.299842530118822e-06,
      "loss": 1.2851,
      "step": 3189
    },
    {
      "epoch": 0.81664,
      "grad_norm": 0.35591105014991437,
      "learning_rate": 3.2909053464822093e-06,
      "loss": 1.31,
      "step": 3190
    },
    {
      "epoch": 0.816896,
      "grad_norm": 0.3597566203194,
      "learning_rate": 3.281979196742995e-06,
      "loss": 1.304,
      "step": 3191
    },
    {
      "epoch": 0.817152,
      "grad_norm": 0.3491135942161179,
      "learning_rate": 3.2730640867955854e-06,
      "loss": 1.3015,
      "step": 3192
    },
    {
      "epoch": 0.817408,
      "grad_norm": 0.3587440743190293,
      "learning_rate": 3.26416002252711e-06,
      "loss": 1.3187,
      "step": 3193
    },
    {
      "epoch": 0.817664,
      "grad_norm": 0.35309370850293453,
      "learning_rate": 3.255267009817378e-06,
      "loss": 1.3266,
      "step": 3194
    },
    {
      "epoch": 0.81792,
      "grad_norm": 0.35979789067375306,
      "learning_rate": 3.246385054538923e-06,
      "loss": 1.3274,
      "step": 3195
    },
    {
      "epoch": 0.818176,
      "grad_norm": 0.3592782212990604,
      "learning_rate": 3.237514162556972e-06,
      "loss": 1.3567,
      "step": 3196
    },
    {
      "epoch": 0.818432,
      "grad_norm": 0.3511406056643651,
      "learning_rate": 3.2286543397294336e-06,
      "loss": 1.274,
      "step": 3197
    },
    {
      "epoch": 0.818688,
      "grad_norm": 0.3490984920523951,
      "learning_rate": 3.2198055919069217e-06,
      "loss": 1.3127,
      "step": 3198
    },
    {
      "epoch": 0.818944,
      "grad_norm": 0.3652287514432127,
      "learning_rate": 3.210967924932733e-06,
      "loss": 1.4128,
      "step": 3199
    },
    {
      "epoch": 0.8192,
      "grad_norm": 0.3637268186353478,
      "learning_rate": 3.202141344642844e-06,
      "loss": 1.3575,
      "step": 3200
    },
    {
      "epoch": 0.819456,
      "grad_norm": 0.3567280680340106,
      "learning_rate": 3.1933258568659143e-06,
      "loss": 1.306,
      "step": 3201
    },
    {
      "epoch": 0.819712,
      "grad_norm": 0.35438668700471676,
      "learning_rate": 3.184521467423276e-06,
      "loss": 1.3308,
      "step": 3202
    },
    {
      "epoch": 0.819968,
      "grad_norm": 0.36006515188234234,
      "learning_rate": 3.1757281821289276e-06,
      "loss": 1.3404,
      "step": 3203
    },
    {
      "epoch": 0.820224,
      "grad_norm": 0.35606437645870986,
      "learning_rate": 3.1669460067895464e-06,
      "loss": 1.3096,
      "step": 3204
    },
    {
      "epoch": 0.82048,
      "grad_norm": 0.35469815719372927,
      "learning_rate": 3.1581749472044576e-06,
      "loss": 1.3188,
      "step": 3205
    },
    {
      "epoch": 0.820736,
      "grad_norm": 0.35744934754609076,
      "learning_rate": 3.149415009165662e-06,
      "loss": 1.3282,
      "step": 3206
    },
    {
      "epoch": 0.820992,
      "grad_norm": 0.35466137067872394,
      "learning_rate": 3.1406661984578046e-06,
      "loss": 1.2772,
      "step": 3207
    },
    {
      "epoch": 0.821248,
      "grad_norm": 0.36291999134771713,
      "learning_rate": 3.131928520858194e-06,
      "loss": 1.3448,
      "step": 3208
    },
    {
      "epoch": 0.821504,
      "grad_norm": 0.3492149689626682,
      "learning_rate": 3.1232019821367698e-06,
      "loss": 1.299,
      "step": 3209
    },
    {
      "epoch": 0.82176,
      "grad_norm": 0.37018793803544103,
      "learning_rate": 3.1144865880561338e-06,
      "loss": 1.3334,
      "step": 3210
    },
    {
      "epoch": 0.822016,
      "grad_norm": 0.35952726397129353,
      "learning_rate": 3.1057823443715174e-06,
      "loss": 1.3188,
      "step": 3211
    },
    {
      "epoch": 0.822272,
      "grad_norm": 0.3596188946658317,
      "learning_rate": 3.0970892568307965e-06,
      "loss": 1.3695,
      "step": 3212
    },
    {
      "epoch": 0.822528,
      "grad_norm": 0.3584933500309711,
      "learning_rate": 3.08840733117447e-06,
      "loss": 1.3159,
      "step": 3213
    },
    {
      "epoch": 0.822784,
      "grad_norm": 0.3512487325159134,
      "learning_rate": 3.079736573135681e-06,
      "loss": 1.3182,
      "step": 3214
    },
    {
      "epoch": 0.82304,
      "grad_norm": 0.35855855893905003,
      "learning_rate": 3.071076988440178e-06,
      "loss": 1.3226,
      "step": 3215
    },
    {
      "epoch": 0.823296,
      "grad_norm": 0.3523029347121198,
      "learning_rate": 3.0624285828063515e-06,
      "loss": 1.3054,
      "step": 3216
    },
    {
      "epoch": 0.823552,
      "grad_norm": 0.3737446954204356,
      "learning_rate": 3.0537913619451907e-06,
      "loss": 1.3107,
      "step": 3217
    },
    {
      "epoch": 0.823808,
      "grad_norm": 0.3507053391681704,
      "learning_rate": 3.0451653315603138e-06,
      "loss": 1.2848,
      "step": 3218
    },
    {
      "epoch": 0.824064,
      "grad_norm": 0.3620545220183772,
      "learning_rate": 3.036550497347943e-06,
      "loss": 1.332,
      "step": 3219
    },
    {
      "epoch": 0.82432,
      "grad_norm": 0.34687524630785277,
      "learning_rate": 3.0279468649969133e-06,
      "loss": 1.291,
      "step": 3220
    },
    {
      "epoch": 0.824576,
      "grad_norm": 0.36213902925500585,
      "learning_rate": 3.019354440188649e-06,
      "loss": 1.3015,
      "step": 3221
    },
    {
      "epoch": 0.824832,
      "grad_norm": 0.3520260335127815,
      "learning_rate": 3.0107732285971835e-06,
      "loss": 1.2859,
      "step": 3222
    },
    {
      "epoch": 0.825088,
      "grad_norm": 0.35233972463483415,
      "learning_rate": 3.0022032358891493e-06,
      "loss": 1.333,
      "step": 3223
    },
    {
      "epoch": 0.825344,
      "grad_norm": 0.352629779806838,
      "learning_rate": 2.9936444677237597e-06,
      "loss": 1.3167,
      "step": 3224
    },
    {
      "epoch": 0.8256,
      "grad_norm": 0.360336855323851,
      "learning_rate": 2.9850969297528266e-06,
      "loss": 1.3369,
      "step": 3225
    },
    {
      "epoch": 0.825856,
      "grad_norm": 0.3548136748548445,
      "learning_rate": 2.9765606276207392e-06,
      "loss": 1.35,
      "step": 3226
    },
    {
      "epoch": 0.826112,
      "grad_norm": 0.3750484356225634,
      "learning_rate": 2.9680355669644667e-06,
      "loss": 1.3403,
      "step": 3227
    },
    {
      "epoch": 0.826368,
      "grad_norm": 0.3586284762084083,
      "learning_rate": 2.9595217534135543e-06,
      "loss": 1.3691,
      "step": 3228
    },
    {
      "epoch": 0.826624,
      "grad_norm": 0.36579343700420064,
      "learning_rate": 2.951019192590123e-06,
      "loss": 1.3722,
      "step": 3229
    },
    {
      "epoch": 0.82688,
      "grad_norm": 0.36688932585244266,
      "learning_rate": 2.942527890108866e-06,
      "loss": 1.3749,
      "step": 3230
    },
    {
      "epoch": 0.827136,
      "grad_norm": 0.35766663016787575,
      "learning_rate": 2.9340478515770376e-06,
      "loss": 1.3549,
      "step": 3231
    },
    {
      "epoch": 0.827392,
      "grad_norm": 0.35735930166466434,
      "learning_rate": 2.925579082594454e-06,
      "loss": 1.3213,
      "step": 3232
    },
    {
      "epoch": 0.827648,
      "grad_norm": 0.3565379385668704,
      "learning_rate": 2.9171215887534955e-06,
      "loss": 1.3248,
      "step": 3233
    },
    {
      "epoch": 0.827904,
      "grad_norm": 0.354843011897833,
      "learning_rate": 2.908675375639085e-06,
      "loss": 1.3177,
      "step": 3234
    },
    {
      "epoch": 0.82816,
      "grad_norm": 0.35439225713215344,
      "learning_rate": 2.900240448828704e-06,
      "loss": 1.2733,
      "step": 3235
    },
    {
      "epoch": 0.828416,
      "grad_norm": 0.3543610190322335,
      "learning_rate": 2.891816813892381e-06,
      "loss": 1.3261,
      "step": 3236
    },
    {
      "epoch": 0.828672,
      "grad_norm": 0.3616501518274734,
      "learning_rate": 2.883404476392693e-06,
      "loss": 1.361,
      "step": 3237
    },
    {
      "epoch": 0.828928,
      "grad_norm": 0.3618171538856441,
      "learning_rate": 2.8750034418847404e-06,
      "loss": 1.3252,
      "step": 3238
    },
    {
      "epoch": 0.829184,
      "grad_norm": 0.3505232809998268,
      "learning_rate": 2.8666137159161776e-06,
      "loss": 1.2903,
      "step": 3239
    },
    {
      "epoch": 0.82944,
      "grad_norm": 0.3591926211742982,
      "learning_rate": 2.8582353040271772e-06,
      "loss": 1.3346,
      "step": 3240
    },
    {
      "epoch": 0.829696,
      "grad_norm": 0.3679269499774998,
      "learning_rate": 2.8498682117504504e-06,
      "loss": 1.3529,
      "step": 3241
    },
    {
      "epoch": 0.829952,
      "grad_norm": 0.3592837410241286,
      "learning_rate": 2.8415124446112296e-06,
      "loss": 1.3394,
      "step": 3242
    },
    {
      "epoch": 0.830208,
      "grad_norm": 0.3569505682825715,
      "learning_rate": 2.8331680081272695e-06,
      "loss": 1.3537,
      "step": 3243
    },
    {
      "epoch": 0.830464,
      "grad_norm": 0.3567746997930505,
      "learning_rate": 2.824834907808842e-06,
      "loss": 1.3207,
      "step": 3244
    },
    {
      "epoch": 0.83072,
      "grad_norm": 0.36051559209834994,
      "learning_rate": 2.816513149158737e-06,
      "loss": 1.3228,
      "step": 3245
    },
    {
      "epoch": 0.830976,
      "grad_norm": 0.3637180484884654,
      "learning_rate": 2.808202737672243e-06,
      "loss": 1.3583,
      "step": 3246
    },
    {
      "epoch": 0.831232,
      "grad_norm": 0.36040810056694006,
      "learning_rate": 2.799903678837168e-06,
      "loss": 1.3253,
      "step": 3247
    },
    {
      "epoch": 0.831488,
      "grad_norm": 0.3551052398695939,
      "learning_rate": 2.791615978133817e-06,
      "loss": 1.2968,
      "step": 3248
    },
    {
      "epoch": 0.831744,
      "grad_norm": 0.353241238582077,
      "learning_rate": 2.7833396410350034e-06,
      "loss": 1.2907,
      "step": 3249
    },
    {
      "epoch": 0.832,
      "grad_norm": 0.3593897105491773,
      "learning_rate": 2.7750746730060176e-06,
      "loss": 1.36,
      "step": 3250
    },
    {
      "epoch": 0.832256,
      "grad_norm": 0.3529382850734224,
      "learning_rate": 2.7668210795046667e-06,
      "loss": 1.3315,
      "step": 3251
    },
    {
      "epoch": 0.832512,
      "grad_norm": 0.36022402819842014,
      "learning_rate": 2.758578865981223e-06,
      "loss": 1.3218,
      "step": 3252
    },
    {
      "epoch": 0.832768,
      "grad_norm": 0.3553246019573726,
      "learning_rate": 2.7503480378784607e-06,
      "loss": 1.306,
      "step": 3253
    },
    {
      "epoch": 0.833024,
      "grad_norm": 0.35552399012855007,
      "learning_rate": 2.7421286006316307e-06,
      "loss": 1.345,
      "step": 3254
    },
    {
      "epoch": 0.83328,
      "grad_norm": 0.3588214718251527,
      "learning_rate": 2.7339205596684593e-06,
      "loss": 1.3494,
      "step": 3255
    },
    {
      "epoch": 0.833536,
      "grad_norm": 0.3547076684771563,
      "learning_rate": 2.725723920409149e-06,
      "loss": 1.3339,
      "step": 3256
    },
    {
      "epoch": 0.833792,
      "grad_norm": 0.36393544108983333,
      "learning_rate": 2.7175386882663812e-06,
      "loss": 1.3159,
      "step": 3257
    },
    {
      "epoch": 0.834048,
      "grad_norm": 0.35325059456831887,
      "learning_rate": 2.709364868645288e-06,
      "loss": 1.2842,
      "step": 3258
    },
    {
      "epoch": 0.834304,
      "grad_norm": 0.3515229610426091,
      "learning_rate": 2.7012024669434756e-06,
      "loss": 1.2873,
      "step": 3259
    },
    {
      "epoch": 0.83456,
      "grad_norm": 0.3608482040674047,
      "learning_rate": 2.693051488551013e-06,
      "loss": 1.3397,
      "step": 3260
    },
    {
      "epoch": 0.834816,
      "grad_norm": 0.356432821428022,
      "learning_rate": 2.684911938850414e-06,
      "loss": 1.2818,
      "step": 3261
    },
    {
      "epoch": 0.835072,
      "grad_norm": 0.3586012965977556,
      "learning_rate": 2.67678382321666e-06,
      "loss": 1.3088,
      "step": 3262
    },
    {
      "epoch": 0.835328,
      "grad_norm": 0.3601238076750384,
      "learning_rate": 2.6686671470171743e-06,
      "loss": 1.3498,
      "step": 3263
    },
    {
      "epoch": 0.835584,
      "grad_norm": 0.35789518414771604,
      "learning_rate": 2.6605619156118212e-06,
      "loss": 1.3096,
      "step": 3264
    },
    {
      "epoch": 0.83584,
      "grad_norm": 0.37219448831863466,
      "learning_rate": 2.652468134352917e-06,
      "loss": 1.3208,
      "step": 3265
    },
    {
      "epoch": 0.836096,
      "grad_norm": 0.37966498918813824,
      "learning_rate": 2.6443858085852128e-06,
      "loss": 1.3274,
      "step": 3266
    },
    {
      "epoch": 0.836352,
      "grad_norm": 0.35852627846317614,
      "learning_rate": 2.6363149436458924e-06,
      "loss": 1.3334,
      "step": 3267
    },
    {
      "epoch": 0.836608,
      "grad_norm": 0.34967803948392395,
      "learning_rate": 2.6282555448645796e-06,
      "loss": 1.296,
      "step": 3268
    },
    {
      "epoch": 0.836864,
      "grad_norm": 0.3537289409537512,
      "learning_rate": 2.6202076175633196e-06,
      "loss": 1.3433,
      "step": 3269
    },
    {
      "epoch": 0.83712,
      "grad_norm": 0.3572828079681934,
      "learning_rate": 2.6121711670565787e-06,
      "loss": 1.3136,
      "step": 3270
    },
    {
      "epoch": 0.837376,
      "grad_norm": 0.352368919160701,
      "learning_rate": 2.6041461986512517e-06,
      "loss": 1.2942,
      "step": 3271
    },
    {
      "epoch": 0.837632,
      "grad_norm": 0.35728433147215205,
      "learning_rate": 2.5961327176466533e-06,
      "loss": 1.2995,
      "step": 3272
    },
    {
      "epoch": 0.837888,
      "grad_norm": 0.35438995005259727,
      "learning_rate": 2.5881307293345016e-06,
      "loss": 1.2887,
      "step": 3273
    },
    {
      "epoch": 0.838144,
      "grad_norm": 0.35941236032423135,
      "learning_rate": 2.580140238998934e-06,
      "loss": 1.367,
      "step": 3274
    },
    {
      "epoch": 0.8384,
      "grad_norm": 0.3697952834906845,
      "learning_rate": 2.5721612519164984e-06,
      "loss": 1.3503,
      "step": 3275
    },
    {
      "epoch": 0.838656,
      "grad_norm": 0.3930174076858665,
      "learning_rate": 2.564193773356134e-06,
      "loss": 1.3145,
      "step": 3276
    },
    {
      "epoch": 0.838912,
      "grad_norm": 0.3521793233615387,
      "learning_rate": 2.5562378085791873e-06,
      "loss": 1.2878,
      "step": 3277
    },
    {
      "epoch": 0.839168,
      "grad_norm": 0.3562823524706217,
      "learning_rate": 2.5482933628394068e-06,
      "loss": 1.3428,
      "step": 3278
    },
    {
      "epoch": 0.839424,
      "grad_norm": 0.36349685126140996,
      "learning_rate": 2.54036044138293e-06,
      "loss": 1.3493,
      "step": 3279
    },
    {
      "epoch": 0.83968,
      "grad_norm": 0.35344029169121743,
      "learning_rate": 2.532439049448279e-06,
      "loss": 1.3353,
      "step": 3280
    },
    {
      "epoch": 0.839936,
      "grad_norm": 0.3580095830177698,
      "learning_rate": 2.524529192266374e-06,
      "loss": 1.3308,
      "step": 3281
    },
    {
      "epoch": 0.840192,
      "grad_norm": 0.3543119028077944,
      "learning_rate": 2.516630875060502e-06,
      "loss": 1.3134,
      "step": 3282
    },
    {
      "epoch": 0.840448,
      "grad_norm": 0.3562750741047484,
      "learning_rate": 2.50874410304635e-06,
      "loss": 1.31,
      "step": 3283
    },
    {
      "epoch": 0.840704,
      "grad_norm": 0.3519229372077795,
      "learning_rate": 2.500868881431957e-06,
      "loss": 1.3085,
      "step": 3284
    },
    {
      "epoch": 0.84096,
      "grad_norm": 0.3585854371805405,
      "learning_rate": 2.4930052154177563e-06,
      "loss": 1.3322,
      "step": 3285
    },
    {
      "epoch": 0.841216,
      "grad_norm": 0.4852777688319952,
      "learning_rate": 2.4851531101965408e-06,
      "loss": 1.348,
      "step": 3286
    },
    {
      "epoch": 0.841472,
      "grad_norm": 0.354126007467709,
      "learning_rate": 2.4773125709534673e-06,
      "loss": 1.3231,
      "step": 3287
    },
    {
      "epoch": 0.841728,
      "grad_norm": 0.4195825594615701,
      "learning_rate": 2.469483602866063e-06,
      "loss": 1.297,
      "step": 3288
    },
    {
      "epoch": 0.841984,
      "grad_norm": 0.35274347666532474,
      "learning_rate": 2.4616662111042033e-06,
      "loss": 1.3157,
      "step": 3289
    },
    {
      "epoch": 0.84224,
      "grad_norm": 0.39144282841089934,
      "learning_rate": 2.453860400830126e-06,
      "loss": 1.3402,
      "step": 3290
    },
    {
      "epoch": 0.842496,
      "grad_norm": 0.3517713709042496,
      "learning_rate": 2.4460661771984227e-06,
      "loss": 1.2923,
      "step": 3291
    },
    {
      "epoch": 0.842752,
      "grad_norm": 0.43123755438975486,
      "learning_rate": 2.438283545356026e-06,
      "loss": 1.3013,
      "step": 3292
    },
    {
      "epoch": 0.843008,
      "grad_norm": 0.3517854347493126,
      "learning_rate": 2.4305125104422267e-06,
      "loss": 1.342,
      "step": 3293
    },
    {
      "epoch": 0.843264,
      "grad_norm": 0.3630344744429917,
      "learning_rate": 2.422753077588642e-06,
      "loss": 1.3318,
      "step": 3294
    },
    {
      "epoch": 0.84352,
      "grad_norm": 0.3524454342412477,
      "learning_rate": 2.415005251919238e-06,
      "loss": 1.3435,
      "step": 3295
    },
    {
      "epoch": 0.843776,
      "grad_norm": 0.35841122485576776,
      "learning_rate": 2.4072690385503105e-06,
      "loss": 1.314,
      "step": 3296
    },
    {
      "epoch": 0.844032,
      "grad_norm": 0.3576542543848138,
      "learning_rate": 2.3995444425904914e-06,
      "loss": 1.3459,
      "step": 3297
    },
    {
      "epoch": 0.844288,
      "grad_norm": 0.35719328704951125,
      "learning_rate": 2.3918314691407373e-06,
      "loss": 1.3145,
      "step": 3298
    },
    {
      "epoch": 0.844544,
      "grad_norm": 0.3633786460398092,
      "learning_rate": 2.384130123294337e-06,
      "loss": 1.3444,
      "step": 3299
    },
    {
      "epoch": 0.8448,
      "grad_norm": 0.35512217140606295,
      "learning_rate": 2.3764404101368954e-06,
      "loss": 1.3224,
      "step": 3300
    },
    {
      "epoch": 0.845056,
      "grad_norm": 0.3595177169862895,
      "learning_rate": 2.3687623347463284e-06,
      "loss": 1.3421,
      "step": 3301
    },
    {
      "epoch": 0.845312,
      "grad_norm": 0.35613220541053237,
      "learning_rate": 2.3610959021928803e-06,
      "loss": 1.3272,
      "step": 3302
    },
    {
      "epoch": 0.845568,
      "grad_norm": 0.35548634555637093,
      "learning_rate": 2.3534411175390994e-06,
      "loss": 1.2932,
      "step": 3303
    },
    {
      "epoch": 0.845824,
      "grad_norm": 0.3604640644085452,
      "learning_rate": 2.345797985839846e-06,
      "loss": 1.3243,
      "step": 3304
    },
    {
      "epoch": 0.84608,
      "grad_norm": 0.40654114497705796,
      "learning_rate": 2.338166512142286e-06,
      "loss": 1.3079,
      "step": 3305
    },
    {
      "epoch": 0.846336,
      "grad_norm": 0.3565981329165396,
      "learning_rate": 2.330546701485883e-06,
      "loss": 1.3189,
      "step": 3306
    },
    {
      "epoch": 0.846592,
      "grad_norm": 0.35645165068430334,
      "learning_rate": 2.322938558902392e-06,
      "loss": 1.343,
      "step": 3307
    },
    {
      "epoch": 0.846848,
      "grad_norm": 0.34664805759763895,
      "learning_rate": 2.3153420894158794e-06,
      "loss": 1.2661,
      "step": 3308
    },
    {
      "epoch": 0.847104,
      "grad_norm": 0.3956037562638198,
      "learning_rate": 2.3077572980426922e-06,
      "loss": 1.3481,
      "step": 3309
    },
    {
      "epoch": 0.84736,
      "grad_norm": 0.35143504470878856,
      "learning_rate": 2.30018418979147e-06,
      "loss": 1.2837,
      "step": 3310
    },
    {
      "epoch": 0.847616,
      "grad_norm": 0.35159127460323447,
      "learning_rate": 2.2926227696631376e-06,
      "loss": 1.334,
      "step": 3311
    },
    {
      "epoch": 0.847872,
      "grad_norm": 0.3625689890323539,
      "learning_rate": 2.2850730426509005e-06,
      "loss": 1.3284,
      "step": 3312
    },
    {
      "epoch": 0.848128,
      "grad_norm": 0.35997068830307666,
      "learning_rate": 2.2775350137402353e-06,
      "loss": 1.3244,
      "step": 3313
    },
    {
      "epoch": 0.848384,
      "grad_norm": 0.363924826846468,
      "learning_rate": 2.2700086879089046e-06,
      "loss": 1.3386,
      "step": 3314
    },
    {
      "epoch": 0.84864,
      "grad_norm": 0.36148199677564,
      "learning_rate": 2.262494070126939e-06,
      "loss": 1.3276,
      "step": 3315
    },
    {
      "epoch": 0.848896,
      "grad_norm": 0.3526181905708771,
      "learning_rate": 2.2549911653566416e-06,
      "loss": 1.2908,
      "step": 3316
    },
    {
      "epoch": 0.849152,
      "grad_norm": 0.3584663992368235,
      "learning_rate": 2.2474999785525698e-06,
      "loss": 1.3391,
      "step": 3317
    },
    {
      "epoch": 0.849408,
      "grad_norm": 0.3474946229767795,
      "learning_rate": 2.2400205146615537e-06,
      "loss": 1.2792,
      "step": 3318
    },
    {
      "epoch": 0.849664,
      "grad_norm": 0.34599832423622284,
      "learning_rate": 2.2325527786226743e-06,
      "loss": 1.2805,
      "step": 3319
    },
    {
      "epoch": 0.84992,
      "grad_norm": 0.3595009114769627,
      "learning_rate": 2.2250967753672747e-06,
      "loss": 1.3166,
      "step": 3320
    },
    {
      "epoch": 0.850176,
      "grad_norm": 0.3597749993349547,
      "learning_rate": 2.217652509818946e-06,
      "loss": 1.3243,
      "step": 3321
    },
    {
      "epoch": 0.850432,
      "grad_norm": 0.3519757595317033,
      "learning_rate": 2.2102199868935316e-06,
      "loss": 1.2978,
      "step": 3322
    },
    {
      "epoch": 0.850688,
      "grad_norm": 0.3525089962860643,
      "learning_rate": 2.202799211499118e-06,
      "loss": 1.3273,
      "step": 3323
    },
    {
      "epoch": 0.850944,
      "grad_norm": 0.36404519626503606,
      "learning_rate": 2.1953901885360395e-06,
      "loss": 1.3545,
      "step": 3324
    },
    {
      "epoch": 0.8512,
      "grad_norm": 0.549722139858535,
      "learning_rate": 2.1879929228968554e-06,
      "loss": 1.3217,
      "step": 3325
    },
    {
      "epoch": 0.851456,
      "grad_norm": 0.3582505849168342,
      "learning_rate": 2.1806074194663783e-06,
      "loss": 1.3416,
      "step": 3326
    },
    {
      "epoch": 0.851712,
      "grad_norm": 0.3486846766638767,
      "learning_rate": 2.173233683121643e-06,
      "loss": 1.3244,
      "step": 3327
    },
    {
      "epoch": 0.851968,
      "grad_norm": 0.36564168075447623,
      "learning_rate": 2.1658717187319224e-06,
      "loss": 1.3189,
      "step": 3328
    },
    {
      "epoch": 0.852224,
      "grad_norm": 0.3514161506865785,
      "learning_rate": 2.1585215311587014e-06,
      "loss": 1.3081,
      "step": 3329
    },
    {
      "epoch": 0.85248,
      "grad_norm": 0.3540921564763548,
      "learning_rate": 2.1511831252557048e-06,
      "loss": 1.3048,
      "step": 3330
    },
    {
      "epoch": 0.852736,
      "grad_norm": 0.3501199097110855,
      "learning_rate": 2.1438565058688623e-06,
      "loss": 1.3054,
      "step": 3331
    },
    {
      "epoch": 0.852992,
      "grad_norm": 0.35815357873398707,
      "learning_rate": 2.1365416778363325e-06,
      "loss": 1.3367,
      "step": 3332
    },
    {
      "epoch": 0.853248,
      "grad_norm": 0.354454076503121,
      "learning_rate": 2.1292386459884804e-06,
      "loss": 1.3135,
      "step": 3333
    },
    {
      "epoch": 0.853504,
      "grad_norm": 0.3508782898263423,
      "learning_rate": 2.1219474151478823e-06,
      "loss": 1.3177,
      "step": 3334
    },
    {
      "epoch": 0.85376,
      "grad_norm": 0.34883249728020327,
      "learning_rate": 2.1146679901293267e-06,
      "loss": 1.2836,
      "step": 3335
    },
    {
      "epoch": 0.854016,
      "grad_norm": 0.35817804489055227,
      "learning_rate": 2.1074003757398055e-06,
      "loss": 1.3393,
      "step": 3336
    },
    {
      "epoch": 0.854272,
      "grad_norm": 0.35322042753110866,
      "learning_rate": 2.1001445767784978e-06,
      "loss": 1.2869,
      "step": 3337
    },
    {
      "epoch": 0.854528,
      "grad_norm": 0.35888638623148406,
      "learning_rate": 2.0929005980367957e-06,
      "loss": 1.3572,
      "step": 3338
    },
    {
      "epoch": 0.854784,
      "grad_norm": 0.3509383155618107,
      "learning_rate": 2.085668444298288e-06,
      "loss": 1.2898,
      "step": 3339
    },
    {
      "epoch": 0.85504,
      "grad_norm": 0.35369281115667717,
      "learning_rate": 2.078448120338734e-06,
      "loss": 1.2957,
      "step": 3340
    },
    {
      "epoch": 0.855296,
      "grad_norm": 0.3483203959099041,
      "learning_rate": 2.0712396309261028e-06,
      "loss": 1.2841,
      "step": 3341
    },
    {
      "epoch": 0.855552,
      "grad_norm": 0.3449622697279141,
      "learning_rate": 2.0640429808205442e-06,
      "loss": 1.2864,
      "step": 3342
    },
    {
      "epoch": 0.855808,
      "grad_norm": 0.3609443953755346,
      "learning_rate": 2.0568581747743764e-06,
      "loss": 1.3091,
      "step": 3343
    },
    {
      "epoch": 0.856064,
      "grad_norm": 0.35530059622567073,
      "learning_rate": 2.0496852175321113e-06,
      "loss": 1.3229,
      "step": 3344
    },
    {
      "epoch": 0.85632,
      "grad_norm": 0.3631972632891438,
      "learning_rate": 2.0425241138304307e-06,
      "loss": 1.3325,
      "step": 3345
    },
    {
      "epoch": 0.856576,
      "grad_norm": 0.35550296364796985,
      "learning_rate": 2.0353748683981922e-06,
      "loss": 1.2978,
      "step": 3346
    },
    {
      "epoch": 0.856832,
      "grad_norm": 0.3508494337988298,
      "learning_rate": 2.028237485956417e-06,
      "loss": 1.308,
      "step": 3347
    },
    {
      "epoch": 0.857088,
      "grad_norm": 0.3522478921437514,
      "learning_rate": 2.0211119712182947e-06,
      "loss": 1.3455,
      "step": 3348
    },
    {
      "epoch": 0.857344,
      "grad_norm": 0.350390970968294,
      "learning_rate": 2.0139983288891864e-06,
      "loss": 1.2977,
      "step": 3349
    },
    {
      "epoch": 0.8576,
      "grad_norm": 0.3539239884214114,
      "learning_rate": 2.006896563666596e-06,
      "loss": 1.3096,
      "step": 3350
    },
    {
      "epoch": 0.857856,
      "grad_norm": 0.3716172380676982,
      "learning_rate": 1.9998066802402016e-06,
      "loss": 1.3136,
      "step": 3351
    },
    {
      "epoch": 0.858112,
      "grad_norm": 0.3548652937590851,
      "learning_rate": 1.992728683291818e-06,
      "loss": 1.3218,
      "step": 3352
    },
    {
      "epoch": 0.858368,
      "grad_norm": 0.4524341497519749,
      "learning_rate": 1.9856625774954287e-06,
      "loss": 1.3292,
      "step": 3353
    },
    {
      "epoch": 0.858624,
      "grad_norm": 0.34730402883355477,
      "learning_rate": 1.9786083675171542e-06,
      "loss": 1.3152,
      "step": 3354
    },
    {
      "epoch": 0.85888,
      "grad_norm": 0.3533815715889272,
      "learning_rate": 1.971566058015264e-06,
      "loss": 1.3164,
      "step": 3355
    },
    {
      "epoch": 0.859136,
      "grad_norm": 0.3782844915418903,
      "learning_rate": 1.9645356536401604e-06,
      "loss": 1.28,
      "step": 3356
    },
    {
      "epoch": 0.859392,
      "grad_norm": 0.3504076570542578,
      "learning_rate": 1.9575171590343967e-06,
      "loss": 1.2863,
      "step": 3357
    },
    {
      "epoch": 0.859648,
      "grad_norm": 0.35037007333160175,
      "learning_rate": 1.950510578832652e-06,
      "loss": 1.2866,
      "step": 3358
    },
    {
      "epoch": 0.859904,
      "grad_norm": 0.3570274592821321,
      "learning_rate": 1.9435159176617404e-06,
      "loss": 1.3024,
      "step": 3359
    },
    {
      "epoch": 0.86016,
      "grad_norm": 0.3527714563477848,
      "learning_rate": 1.936533180140614e-06,
      "loss": 1.329,
      "step": 3360
    },
    {
      "epoch": 0.860416,
      "grad_norm": 0.3535506934545577,
      "learning_rate": 1.9295623708803334e-06,
      "loss": 1.3078,
      "step": 3361
    },
    {
      "epoch": 0.860672,
      "grad_norm": 0.3487638513465772,
      "learning_rate": 1.922603494484101e-06,
      "loss": 1.3127,
      "step": 3362
    },
    {
      "epoch": 0.860928,
      "grad_norm": 0.3442575734944706,
      "learning_rate": 1.915656555547218e-06,
      "loss": 1.2731,
      "step": 3363
    },
    {
      "epoch": 0.861184,
      "grad_norm": 0.35095124260675203,
      "learning_rate": 1.908721558657125e-06,
      "loss": 1.3159,
      "step": 3364
    },
    {
      "epoch": 0.86144,
      "grad_norm": 0.34965537091352855,
      "learning_rate": 1.9017985083933621e-06,
      "loss": 1.2864,
      "step": 3365
    },
    {
      "epoch": 0.861696,
      "grad_norm": 0.36213743542148036,
      "learning_rate": 1.894887409327586e-06,
      "loss": 1.3423,
      "step": 3366
    },
    {
      "epoch": 0.861952,
      "grad_norm": 0.3478151133571996,
      "learning_rate": 1.8879882660235638e-06,
      "loss": 1.2867,
      "step": 3367
    },
    {
      "epoch": 0.862208,
      "grad_norm": 0.3469720716160055,
      "learning_rate": 1.88110108303716e-06,
      "loss": 1.2801,
      "step": 3368
    },
    {
      "epoch": 0.862464,
      "grad_norm": 0.3520254234451795,
      "learning_rate": 1.8742258649163435e-06,
      "loss": 1.3283,
      "step": 3369
    },
    {
      "epoch": 0.86272,
      "grad_norm": 0.3609139388670736,
      "learning_rate": 1.867362616201187e-06,
      "loss": 1.2954,
      "step": 3370
    },
    {
      "epoch": 0.862976,
      "grad_norm": 0.3543433865946269,
      "learning_rate": 1.8605113414238563e-06,
      "loss": 1.3006,
      "step": 3371
    },
    {
      "epoch": 0.863232,
      "grad_norm": 0.35541616886921124,
      "learning_rate": 1.8536720451086121e-06,
      "loss": 1.3502,
      "step": 3372
    },
    {
      "epoch": 0.863488,
      "grad_norm": 0.35902328104214193,
      "learning_rate": 1.8468447317717974e-06,
      "loss": 1.3506,
      "step": 3373
    },
    {
      "epoch": 0.863744,
      "grad_norm": 0.3587515537533156,
      "learning_rate": 1.8400294059218526e-06,
      "loss": 1.3675,
      "step": 3374
    },
    {
      "epoch": 0.864,
      "grad_norm": 0.35790808505538046,
      "learning_rate": 1.8332260720592908e-06,
      "loss": 1.3337,
      "step": 3375
    },
    {
      "epoch": 0.864256,
      "grad_norm": 0.3519865353025258,
      "learning_rate": 1.8264347346767164e-06,
      "loss": 1.296,
      "step": 3376
    },
    {
      "epoch": 0.864512,
      "grad_norm": 0.35808651838077504,
      "learning_rate": 1.8196553982588083e-06,
      "loss": 1.3464,
      "step": 3377
    },
    {
      "epoch": 0.864768,
      "grad_norm": 0.3774158199850852,
      "learning_rate": 1.812888067282319e-06,
      "loss": 1.297,
      "step": 3378
    },
    {
      "epoch": 0.865024,
      "grad_norm": 0.4149531386033982,
      "learning_rate": 1.8061327462160804e-06,
      "loss": 1.3575,
      "step": 3379
    },
    {
      "epoch": 0.86528,
      "grad_norm": 0.36080464349652713,
      "learning_rate": 1.7993894395209776e-06,
      "loss": 1.3266,
      "step": 3380
    },
    {
      "epoch": 0.865536,
      "grad_norm": 0.3506050539591894,
      "learning_rate": 1.7926581516499774e-06,
      "loss": 1.2938,
      "step": 3381
    },
    {
      "epoch": 0.865792,
      "grad_norm": 0.34821621782363205,
      "learning_rate": 1.7859388870481064e-06,
      "loss": 1.2853,
      "step": 3382
    },
    {
      "epoch": 0.866048,
      "grad_norm": 0.35201314250050414,
      "learning_rate": 1.7792316501524487e-06,
      "loss": 1.3226,
      "step": 3383
    },
    {
      "epoch": 0.866304,
      "grad_norm": 0.35219982656942955,
      "learning_rate": 1.7725364453921434e-06,
      "loss": 1.3361,
      "step": 3384
    },
    {
      "epoch": 0.86656,
      "grad_norm": 0.3501884147900269,
      "learning_rate": 1.765853277188394e-06,
      "loss": 1.3112,
      "step": 3385
    },
    {
      "epoch": 0.866816,
      "grad_norm": 0.35453291242813045,
      "learning_rate": 1.7591821499544416e-06,
      "loss": 1.3416,
      "step": 3386
    },
    {
      "epoch": 0.867072,
      "grad_norm": 0.3561478272005234,
      "learning_rate": 1.7525230680955885e-06,
      "loss": 1.3418,
      "step": 3387
    },
    {
      "epoch": 0.867328,
      "grad_norm": 0.35387075475204915,
      "learning_rate": 1.7458760360091753e-06,
      "loss": 1.3066,
      "step": 3388
    },
    {
      "epoch": 0.867584,
      "grad_norm": 0.35448304826178906,
      "learning_rate": 1.739241058084593e-06,
      "loss": 1.3342,
      "step": 3389
    },
    {
      "epoch": 0.86784,
      "grad_norm": 0.3557989622097373,
      "learning_rate": 1.7326181387032636e-06,
      "loss": 1.3161,
      "step": 3390
    },
    {
      "epoch": 0.868096,
      "grad_norm": 0.3465008176436588,
      "learning_rate": 1.7260072822386554e-06,
      "loss": 1.3007,
      "step": 3391
    },
    {
      "epoch": 0.868352,
      "grad_norm": 0.3543988996800462,
      "learning_rate": 1.7194084930562582e-06,
      "loss": 1.3093,
      "step": 3392
    },
    {
      "epoch": 0.868608,
      "grad_norm": 0.3439772431098518,
      "learning_rate": 1.7128217755136046e-06,
      "loss": 1.2488,
      "step": 3393
    },
    {
      "epoch": 0.868864,
      "grad_norm": 0.3777012223225288,
      "learning_rate": 1.7062471339602548e-06,
      "loss": 1.384,
      "step": 3394
    },
    {
      "epoch": 0.86912,
      "grad_norm": 0.3508376286445271,
      "learning_rate": 1.6996845727377898e-06,
      "loss": 1.3094,
      "step": 3395
    },
    {
      "epoch": 0.869376,
      "grad_norm": 0.3640175118148113,
      "learning_rate": 1.6931340961798138e-06,
      "loss": 1.3132,
      "step": 3396
    },
    {
      "epoch": 0.869632,
      "grad_norm": 0.35221238382802705,
      "learning_rate": 1.6865957086119554e-06,
      "loss": 1.3138,
      "step": 3397
    },
    {
      "epoch": 0.869888,
      "grad_norm": 0.3555176980437666,
      "learning_rate": 1.6800694143518526e-06,
      "loss": 1.3283,
      "step": 3398
    },
    {
      "epoch": 0.870144,
      "grad_norm": 0.3513765928802746,
      "learning_rate": 1.6735552177091642e-06,
      "loss": 1.3259,
      "step": 3399
    },
    {
      "epoch": 0.8704,
      "grad_norm": 0.3524966551283487,
      "learning_rate": 1.667053122985558e-06,
      "loss": 1.2875,
      "step": 3400
    },
    {
      "epoch": 0.870656,
      "grad_norm": 0.35283646774716976,
      "learning_rate": 1.660563134474713e-06,
      "loss": 1.2975,
      "step": 3401
    },
    {
      "epoch": 0.870912,
      "grad_norm": 0.3631507666526599,
      "learning_rate": 1.6540852564623077e-06,
      "loss": 1.3618,
      "step": 3402
    },
    {
      "epoch": 0.871168,
      "grad_norm": 0.3510657752296378,
      "learning_rate": 1.6476194932260314e-06,
      "loss": 1.3385,
      "step": 3403
    },
    {
      "epoch": 0.871424,
      "grad_norm": 0.36544537424144063,
      "learning_rate": 1.6411658490355641e-06,
      "loss": 1.3706,
      "step": 3404
    },
    {
      "epoch": 0.87168,
      "grad_norm": 0.37749156847370197,
      "learning_rate": 1.6347243281525883e-06,
      "loss": 1.3532,
      "step": 3405
    },
    {
      "epoch": 0.871936,
      "grad_norm": 0.35478994049377266,
      "learning_rate": 1.6282949348307854e-06,
      "loss": 1.3404,
      "step": 3406
    },
    {
      "epoch": 0.872192,
      "grad_norm": 0.35580653294159154,
      "learning_rate": 1.621877673315817e-06,
      "loss": 1.3134,
      "step": 3407
    },
    {
      "epoch": 0.872448,
      "grad_norm": 0.35388355003394034,
      "learning_rate": 1.6154725478453426e-06,
      "loss": 1.3454,
      "step": 3408
    },
    {
      "epoch": 0.872704,
      "grad_norm": 0.350271549425445,
      "learning_rate": 1.609079562649003e-06,
      "loss": 1.2876,
      "step": 3409
    },
    {
      "epoch": 0.87296,
      "grad_norm": 0.3514018406402214,
      "learning_rate": 1.6026987219484302e-06,
      "loss": 1.3055,
      "step": 3410
    },
    {
      "epoch": 0.873216,
      "grad_norm": 0.3521687235854348,
      "learning_rate": 1.5963300299572227e-06,
      "loss": 1.3029,
      "step": 3411
    },
    {
      "epoch": 0.873472,
      "grad_norm": 0.3571864252499274,
      "learning_rate": 1.5899734908809649e-06,
      "loss": 1.3374,
      "step": 3412
    },
    {
      "epoch": 0.873728,
      "grad_norm": 0.3574598414001649,
      "learning_rate": 1.5836291089172173e-06,
      "loss": 1.3551,
      "step": 3413
    },
    {
      "epoch": 0.873984,
      "grad_norm": 0.37100542161055794,
      "learning_rate": 1.5772968882555107e-06,
      "loss": 1.335,
      "step": 3414
    },
    {
      "epoch": 0.87424,
      "grad_norm": 0.3484951446054928,
      "learning_rate": 1.5709768330773424e-06,
      "loss": 1.3177,
      "step": 3415
    },
    {
      "epoch": 0.874496,
      "grad_norm": 0.35659941910606907,
      "learning_rate": 1.5646689475561848e-06,
      "loss": 1.3128,
      "step": 3416
    },
    {
      "epoch": 0.874752,
      "grad_norm": 0.35250098450503703,
      "learning_rate": 1.55837323585746e-06,
      "loss": 1.3171,
      "step": 3417
    },
    {
      "epoch": 0.875008,
      "grad_norm": 0.35672941750811266,
      "learning_rate": 1.5520897021385638e-06,
      "loss": 1.3556,
      "step": 3418
    },
    {
      "epoch": 0.875264,
      "grad_norm": 0.3499371614761826,
      "learning_rate": 1.5458183505488421e-06,
      "loss": 1.3049,
      "step": 3419
    },
    {
      "epoch": 0.87552,
      "grad_norm": 0.35814180202520735,
      "learning_rate": 1.5395591852295998e-06,
      "loss": 1.3256,
      "step": 3420
    },
    {
      "epoch": 0.875776,
      "grad_norm": 0.35749015614497737,
      "learning_rate": 1.533312210314095e-06,
      "loss": 1.3341,
      "step": 3421
    },
    {
      "epoch": 0.876032,
      "grad_norm": 0.35388091171335445,
      "learning_rate": 1.527077429927537e-06,
      "loss": 1.306,
      "step": 3422
    },
    {
      "epoch": 0.876288,
      "grad_norm": 0.3512853982188562,
      "learning_rate": 1.5208548481870766e-06,
      "loss": 1.2815,
      "step": 3423
    },
    {
      "epoch": 0.876544,
      "grad_norm": 0.3526493537506339,
      "learning_rate": 1.514644469201816e-06,
      "loss": 1.3462,
      "step": 3424
    },
    {
      "epoch": 0.8768,
      "grad_norm": 0.3458383240363241,
      "learning_rate": 1.5084462970727942e-06,
      "loss": 1.2847,
      "step": 3425
    },
    {
      "epoch": 0.877056,
      "grad_norm": 0.3679948521400727,
      "learning_rate": 1.5022603358929955e-06,
      "loss": 1.3965,
      "step": 3426
    },
    {
      "epoch": 0.877312,
      "grad_norm": 0.3548233021387248,
      "learning_rate": 1.496086589747332e-06,
      "loss": 1.3496,
      "step": 3427
    },
    {
      "epoch": 0.877568,
      "grad_norm": 0.34709181633769964,
      "learning_rate": 1.4899250627126581e-06,
      "loss": 1.2917,
      "step": 3428
    },
    {
      "epoch": 0.877824,
      "grad_norm": 0.35053424485936585,
      "learning_rate": 1.4837757588577551e-06,
      "loss": 1.2921,
      "step": 3429
    },
    {
      "epoch": 0.87808,
      "grad_norm": 0.3561158155018892,
      "learning_rate": 1.4776386822433276e-06,
      "loss": 1.3583,
      "step": 3430
    },
    {
      "epoch": 0.878336,
      "grad_norm": 0.3599502920977389,
      "learning_rate": 1.4715138369220161e-06,
      "loss": 1.3745,
      "step": 3431
    },
    {
      "epoch": 0.878592,
      "grad_norm": 0.3577276318473569,
      "learning_rate": 1.4654012269383765e-06,
      "loss": 1.3412,
      "step": 3432
    },
    {
      "epoch": 0.878848,
      "grad_norm": 0.35579293790253486,
      "learning_rate": 1.4593008563288913e-06,
      "loss": 1.3444,
      "step": 3433
    },
    {
      "epoch": 0.879104,
      "grad_norm": 0.3536808643328128,
      "learning_rate": 1.45321272912196e-06,
      "loss": 1.301,
      "step": 3434
    },
    {
      "epoch": 0.87936,
      "grad_norm": 0.3552324607426098,
      "learning_rate": 1.4471368493378847e-06,
      "loss": 1.3511,
      "step": 3435
    },
    {
      "epoch": 0.879616,
      "grad_norm": 0.3554412733248568,
      "learning_rate": 1.4410732209888977e-06,
      "loss": 1.3372,
      "step": 3436
    },
    {
      "epoch": 0.879872,
      "grad_norm": 0.35600689966840865,
      "learning_rate": 1.4350218480791278e-06,
      "loss": 1.316,
      "step": 3437
    },
    {
      "epoch": 0.880128,
      "grad_norm": 0.36349969720593567,
      "learning_rate": 1.4289827346046204e-06,
      "loss": 1.3339,
      "step": 3438
    },
    {
      "epoch": 0.880384,
      "grad_norm": 0.35477088842985516,
      "learning_rate": 1.422955884553321e-06,
      "loss": 1.3223,
      "step": 3439
    },
    {
      "epoch": 0.88064,
      "grad_norm": 0.35489013165648886,
      "learning_rate": 1.4169413019050726e-06,
      "loss": 1.3035,
      "step": 3440
    },
    {
      "epoch": 0.880896,
      "grad_norm": 0.35423741149009724,
      "learning_rate": 1.4109389906316273e-06,
      "loss": 1.3314,
      "step": 3441
    },
    {
      "epoch": 0.881152,
      "grad_norm": 0.37710625674022186,
      "learning_rate": 1.404948954696621e-06,
      "loss": 1.3654,
      "step": 3442
    },
    {
      "epoch": 0.881408,
      "grad_norm": 0.355029291981972,
      "learning_rate": 1.3989711980555965e-06,
      "loss": 1.3552,
      "step": 3443
    },
    {
      "epoch": 0.881664,
      "grad_norm": 0.3560379635516498,
      "learning_rate": 1.3930057246559782e-06,
      "loss": 1.302,
      "step": 3444
    },
    {
      "epoch": 0.88192,
      "grad_norm": 0.35617641852467496,
      "learning_rate": 1.387052538437086e-06,
      "loss": 1.3425,
      "step": 3445
    },
    {
      "epoch": 0.882176,
      "grad_norm": 0.35361458321814293,
      "learning_rate": 1.3811116433301264e-06,
      "loss": 1.3145,
      "step": 3446
    },
    {
      "epoch": 0.882432,
      "grad_norm": 0.36181102700263085,
      "learning_rate": 1.3751830432581792e-06,
      "loss": 1.3132,
      "step": 3447
    },
    {
      "epoch": 0.882688,
      "grad_norm": 0.3506588798767743,
      "learning_rate": 1.3692667421362127e-06,
      "loss": 1.2871,
      "step": 3448
    },
    {
      "epoch": 0.882944,
      "grad_norm": 0.35932189409160614,
      "learning_rate": 1.3633627438710772e-06,
      "loss": 1.3428,
      "step": 3449
    },
    {
      "epoch": 0.8832,
      "grad_norm": 0.3522959915674311,
      "learning_rate": 1.357471052361492e-06,
      "loss": 1.3509,
      "step": 3450
    },
    {
      "epoch": 0.883456,
      "grad_norm": 0.3567314172173001,
      "learning_rate": 1.3515916714980538e-06,
      "loss": 1.3272,
      "step": 3451
    },
    {
      "epoch": 0.883712,
      "grad_norm": 0.3566599208871027,
      "learning_rate": 1.3457246051632278e-06,
      "loss": 1.3133,
      "step": 3452
    },
    {
      "epoch": 0.883968,
      "grad_norm": 0.36496200773794146,
      "learning_rate": 1.339869857231344e-06,
      "loss": 1.3008,
      "step": 3453
    },
    {
      "epoch": 0.884224,
      "grad_norm": 0.35228289608772456,
      "learning_rate": 1.3340274315686053e-06,
      "loss": 1.3367,
      "step": 3454
    },
    {
      "epoch": 0.88448,
      "grad_norm": 0.36480212002618434,
      "learning_rate": 1.32819733203307e-06,
      "loss": 1.296,
      "step": 3455
    },
    {
      "epoch": 0.884736,
      "grad_norm": 0.3527501222248405,
      "learning_rate": 1.322379562474665e-06,
      "loss": 1.3284,
      "step": 3456
    },
    {
      "epoch": 0.884992,
      "grad_norm": 0.35695834230723705,
      "learning_rate": 1.3165741267351706e-06,
      "loss": 1.3454,
      "step": 3457
    },
    {
      "epoch": 0.885248,
      "grad_norm": 0.3542935006544548,
      "learning_rate": 1.3107810286482225e-06,
      "loss": 1.3216,
      "step": 3458
    },
    {
      "epoch": 0.885504,
      "grad_norm": 0.35760333775556563,
      "learning_rate": 1.3050002720393052e-06,
      "loss": 1.3466,
      "step": 3459
    },
    {
      "epoch": 0.88576,
      "grad_norm": 0.34621530263958666,
      "learning_rate": 1.299231860725758e-06,
      "loss": 1.303,
      "step": 3460
    },
    {
      "epoch": 0.886016,
      "grad_norm": 0.36148076206005114,
      "learning_rate": 1.2934757985167723e-06,
      "loss": 1.3329,
      "step": 3461
    },
    {
      "epoch": 0.886272,
      "grad_norm": 0.35451738049496223,
      "learning_rate": 1.2877320892133782e-06,
      "loss": 1.3228,
      "step": 3462
    },
    {
      "epoch": 0.886528,
      "grad_norm": 0.36493224752850867,
      "learning_rate": 1.2820007366084486e-06,
      "loss": 1.3978,
      "step": 3463
    },
    {
      "epoch": 0.886784,
      "grad_norm": 0.3595312926926017,
      "learning_rate": 1.2762817444867005e-06,
      "loss": 1.3392,
      "step": 3464
    },
    {
      "epoch": 0.88704,
      "grad_norm": 0.3547136029895536,
      "learning_rate": 1.2705751166246816e-06,
      "loss": 1.3441,
      "step": 3465
    },
    {
      "epoch": 0.887296,
      "grad_norm": 0.3501021071214016,
      "learning_rate": 1.2648808567907823e-06,
      "loss": 1.2929,
      "step": 3466
    },
    {
      "epoch": 0.887552,
      "grad_norm": 0.3592554190689579,
      "learning_rate": 1.2591989687452232e-06,
      "loss": 1.3417,
      "step": 3467
    },
    {
      "epoch": 0.887808,
      "grad_norm": 0.3572522939389974,
      "learning_rate": 1.2535294562400547e-06,
      "loss": 1.3234,
      "step": 3468
    },
    {
      "epoch": 0.888064,
      "grad_norm": 0.3443744180496664,
      "learning_rate": 1.247872323019157e-06,
      "loss": 1.2644,
      "step": 3469
    },
    {
      "epoch": 0.88832,
      "grad_norm": 0.3557885961084128,
      "learning_rate": 1.2422275728182309e-06,
      "loss": 1.3299,
      "step": 3470
    },
    {
      "epoch": 0.888576,
      "grad_norm": 0.35207523291523296,
      "learning_rate": 1.2365952093648082e-06,
      "loss": 1.3339,
      "step": 3471
    },
    {
      "epoch": 0.888832,
      "grad_norm": 0.3536742792919578,
      "learning_rate": 1.2309752363782291e-06,
      "loss": 1.324,
      "step": 3472
    },
    {
      "epoch": 0.889088,
      "grad_norm": 0.3543059238673609,
      "learning_rate": 1.2253676575696627e-06,
      "loss": 1.3354,
      "step": 3473
    },
    {
      "epoch": 0.889344,
      "grad_norm": 0.34887546065042163,
      "learning_rate": 1.2197724766420894e-06,
      "loss": 1.2935,
      "step": 3474
    },
    {
      "epoch": 0.8896,
      "grad_norm": 0.3599923005825391,
      "learning_rate": 1.2141896972903e-06,
      "loss": 1.3192,
      "step": 3475
    },
    {
      "epoch": 0.889856,
      "grad_norm": 0.36664296668228047,
      "learning_rate": 1.2086193232008991e-06,
      "loss": 1.344,
      "step": 3476
    },
    {
      "epoch": 0.890112,
      "grad_norm": 0.3524892410936937,
      "learning_rate": 1.2030613580523021e-06,
      "loss": 1.3337,
      "step": 3477
    },
    {
      "epoch": 0.890368,
      "grad_norm": 0.35854730886610753,
      "learning_rate": 1.1975158055147218e-06,
      "loss": 1.3511,
      "step": 3478
    },
    {
      "epoch": 0.890624,
      "grad_norm": 0.35524778794202605,
      "learning_rate": 1.1919826692501823e-06,
      "loss": 1.3188,
      "step": 3479
    },
    {
      "epoch": 0.89088,
      "grad_norm": 0.35027076909184174,
      "learning_rate": 1.1864619529125055e-06,
      "loss": 1.3161,
      "step": 3480
    },
    {
      "epoch": 0.891136,
      "grad_norm": 0.5203621963323755,
      "learning_rate": 1.1809536601473103e-06,
      "loss": 1.3436,
      "step": 3481
    },
    {
      "epoch": 0.891392,
      "grad_norm": 0.35342807060355685,
      "learning_rate": 1.1754577945920142e-06,
      "loss": 1.3279,
      "step": 3482
    },
    {
      "epoch": 0.891648,
      "grad_norm": 0.35474242161684677,
      "learning_rate": 1.1699743598758317e-06,
      "loss": 1.3329,
      "step": 3483
    },
    {
      "epoch": 0.891904,
      "grad_norm": 0.35012064175757707,
      "learning_rate": 1.1645033596197575e-06,
      "loss": 1.3232,
      "step": 3484
    },
    {
      "epoch": 0.89216,
      "grad_norm": 0.3472711342171822,
      "learning_rate": 1.1590447974365881e-06,
      "loss": 1.3141,
      "step": 3485
    },
    {
      "epoch": 0.892416,
      "grad_norm": 0.35601948751968426,
      "learning_rate": 1.1535986769308915e-06,
      "loss": 1.3232,
      "step": 3486
    },
    {
      "epoch": 0.892672,
      "grad_norm": 0.3612691157127042,
      "learning_rate": 1.1481650016990375e-06,
      "loss": 1.3123,
      "step": 3487
    },
    {
      "epoch": 0.892928,
      "grad_norm": 0.3541352802209042,
      "learning_rate": 1.1427437753291627e-06,
      "loss": 1.3307,
      "step": 3488
    },
    {
      "epoch": 0.893184,
      "grad_norm": 0.35188969741657383,
      "learning_rate": 1.137335001401194e-06,
      "loss": 1.2972,
      "step": 3489
    },
    {
      "epoch": 0.89344,
      "grad_norm": 0.34468543933807433,
      "learning_rate": 1.1319386834868262e-06,
      "loss": 1.2802,
      "step": 3490
    },
    {
      "epoch": 0.893696,
      "grad_norm": 0.35079358249798087,
      "learning_rate": 1.126554825149535e-06,
      "loss": 1.299,
      "step": 3491
    },
    {
      "epoch": 0.893952,
      "grad_norm": 0.35276788466905523,
      "learning_rate": 1.1211834299445678e-06,
      "loss": 1.2909,
      "step": 3492
    },
    {
      "epoch": 0.894208,
      "grad_norm": 0.3550493056736705,
      "learning_rate": 1.1158245014189362e-06,
      "loss": 1.3214,
      "step": 3493
    },
    {
      "epoch": 0.894464,
      "grad_norm": 0.3493956441552366,
      "learning_rate": 1.110478043111427e-06,
      "loss": 1.3228,
      "step": 3494
    },
    {
      "epoch": 0.89472,
      "grad_norm": 0.36103139580512994,
      "learning_rate": 1.1051440585525896e-06,
      "loss": 1.2946,
      "step": 3495
    },
    {
      "epoch": 0.894976,
      "grad_norm": 0.3616838839935567,
      "learning_rate": 1.0998225512647332e-06,
      "loss": 1.3427,
      "step": 3496
    },
    {
      "epoch": 0.895232,
      "grad_norm": 0.352782625671208,
      "learning_rate": 1.0945135247619332e-06,
      "loss": 1.3407,
      "step": 3497
    },
    {
      "epoch": 0.895488,
      "grad_norm": 0.35310119156486997,
      "learning_rate": 1.0892169825500144e-06,
      "loss": 1.2895,
      "step": 3498
    },
    {
      "epoch": 0.895744,
      "grad_norm": 0.3565934642982159,
      "learning_rate": 1.0839329281265676e-06,
      "loss": 1.3144,
      "step": 3499
    },
    {
      "epoch": 0.896,
      "grad_norm": 0.35175896072594764,
      "learning_rate": 1.0786613649809308e-06,
      "loss": 1.323,
      "step": 3500
    },
    {
      "epoch": 0.896256,
      "grad_norm": 0.3794033558081013,
      "learning_rate": 1.0734022965942015e-06,
      "loss": 1.3436,
      "step": 3501
    },
    {
      "epoch": 0.896512,
      "grad_norm": 0.3604002157148276,
      "learning_rate": 1.0681557264392106e-06,
      "loss": 1.3459,
      "step": 3502
    },
    {
      "epoch": 0.896768,
      "grad_norm": 0.3518174571483309,
      "learning_rate": 1.0629216579805513e-06,
      "loss": 1.3243,
      "step": 3503
    },
    {
      "epoch": 0.897024,
      "grad_norm": 0.35416143127004995,
      "learning_rate": 1.0577000946745541e-06,
      "loss": 1.3263,
      "step": 3504
    },
    {
      "epoch": 0.89728,
      "grad_norm": 0.36308939075673613,
      "learning_rate": 1.0524910399692923e-06,
      "loss": 1.3179,
      "step": 3505
    },
    {
      "epoch": 0.897536,
      "grad_norm": 0.35620362365960795,
      "learning_rate": 1.0472944973045807e-06,
      "loss": 1.3421,
      "step": 3506
    },
    {
      "epoch": 0.897792,
      "grad_norm": 0.3459673500036439,
      "learning_rate": 1.0421104701119744e-06,
      "loss": 1.2885,
      "step": 3507
    },
    {
      "epoch": 0.898048,
      "grad_norm": 0.35460709434956694,
      "learning_rate": 1.0369389618147552e-06,
      "loss": 1.3387,
      "step": 3508
    },
    {
      "epoch": 0.898304,
      "grad_norm": 0.3487721503862132,
      "learning_rate": 1.0317799758279424e-06,
      "loss": 1.3073,
      "step": 3509
    },
    {
      "epoch": 0.89856,
      "grad_norm": 0.3511084869259205,
      "learning_rate": 1.0266335155582885e-06,
      "loss": 1.3183,
      "step": 3510
    },
    {
      "epoch": 0.898816,
      "grad_norm": 0.35523758126788113,
      "learning_rate": 1.021499584404273e-06,
      "loss": 1.3473,
      "step": 3511
    },
    {
      "epoch": 0.899072,
      "grad_norm": 0.3518944436803949,
      "learning_rate": 1.0163781857561017e-06,
      "loss": 1.3141,
      "step": 3512
    },
    {
      "epoch": 0.899328,
      "grad_norm": 0.35002477327491416,
      "learning_rate": 1.0112693229957093e-06,
      "loss": 1.3106,
      "step": 3513
    },
    {
      "epoch": 0.899584,
      "grad_norm": 0.35655654637448386,
      "learning_rate": 1.0061729994967374e-06,
      "loss": 1.3492,
      "step": 3514
    },
    {
      "epoch": 0.89984,
      "grad_norm": 0.34876359046269984,
      "learning_rate": 1.0010892186245647e-06,
      "loss": 1.3007,
      "step": 3515
    },
    {
      "epoch": 0.900096,
      "grad_norm": 0.3576543471055604,
      "learning_rate": 9.960179837362793e-07,
      "loss": 1.3546,
      "step": 3516
    },
    {
      "epoch": 0.900352,
      "grad_norm": 0.35304341875635104,
      "learning_rate": 9.90959298180685e-07,
      "loss": 1.3116,
      "step": 3517
    },
    {
      "epoch": 0.900608,
      "grad_norm": 0.34324990360572544,
      "learning_rate": 9.859131652982979e-07,
      "loss": 1.2684,
      "step": 3518
    },
    {
      "epoch": 0.900864,
      "grad_norm": 0.3495723058377081,
      "learning_rate": 9.808795884213462e-07,
      "loss": 1.3085,
      "step": 3519
    },
    {
      "epoch": 0.90112,
      "grad_norm": 0.3572073056979552,
      "learning_rate": 9.758585708737711e-07,
      "loss": 1.3594,
      "step": 3520
    },
    {
      "epoch": 0.901376,
      "grad_norm": 0.35194856241012956,
      "learning_rate": 9.70850115971207e-07,
      "loss": 1.3033,
      "step": 3521
    },
    {
      "epoch": 0.901632,
      "grad_norm": 0.349492515088723,
      "learning_rate": 9.658542270210058e-07,
      "loss": 1.3096,
      "step": 3522
    },
    {
      "epoch": 0.901888,
      "grad_norm": 0.35469688635601465,
      "learning_rate": 9.608709073222156e-07,
      "loss": 1.3029,
      "step": 3523
    },
    {
      "epoch": 0.902144,
      "grad_norm": 0.35312500280288117,
      "learning_rate": 9.55900160165586e-07,
      "loss": 1.348,
      "step": 3524
    },
    {
      "epoch": 0.9024,
      "grad_norm": 0.35275133993837204,
      "learning_rate": 9.509419888335692e-07,
      "loss": 1.3416,
      "step": 3525
    },
    {
      "epoch": 0.902656,
      "grad_norm": 0.37030172652101323,
      "learning_rate": 9.459963966002972e-07,
      "loss": 1.3434,
      "step": 3526
    },
    {
      "epoch": 0.902912,
      "grad_norm": 0.3548582675380162,
      "learning_rate": 9.410633867316132e-07,
      "loss": 1.3227,
      "step": 3527
    },
    {
      "epoch": 0.903168,
      "grad_norm": 0.3602397241880957,
      "learning_rate": 9.361429624850404e-07,
      "loss": 1.3169,
      "step": 3528
    },
    {
      "epoch": 0.903424,
      "grad_norm": 0.3537612699934746,
      "learning_rate": 9.312351271097953e-07,
      "loss": 1.3172,
      "step": 3529
    },
    {
      "epoch": 0.90368,
      "grad_norm": 0.3536799565779114,
      "learning_rate": 9.263398838467852e-07,
      "loss": 1.328,
      "step": 3530
    },
    {
      "epoch": 0.903936,
      "grad_norm": 0.3498399404302602,
      "learning_rate": 9.214572359285934e-07,
      "loss": 1.3076,
      "step": 3531
    },
    {
      "epoch": 0.904192,
      "grad_norm": 0.34493355855011776,
      "learning_rate": 9.165871865794895e-07,
      "loss": 1.2807,
      "step": 3532
    },
    {
      "epoch": 0.904448,
      "grad_norm": 0.35369266746038835,
      "learning_rate": 9.117297390154234e-07,
      "loss": 1.3179,
      "step": 3533
    },
    {
      "epoch": 0.904704,
      "grad_norm": 0.3582844075899968,
      "learning_rate": 9.068848964440291e-07,
      "loss": 1.347,
      "step": 3534
    },
    {
      "epoch": 0.90496,
      "grad_norm": 0.35132462241849305,
      "learning_rate": 9.020526620646075e-07,
      "loss": 1.3379,
      "step": 3535
    },
    {
      "epoch": 0.905216,
      "grad_norm": 0.3577202957237167,
      "learning_rate": 8.972330390681394e-07,
      "loss": 1.3148,
      "step": 3536
    },
    {
      "epoch": 0.905472,
      "grad_norm": 0.35662835303231133,
      "learning_rate": 8.924260306372746e-07,
      "loss": 1.3208,
      "step": 3537
    },
    {
      "epoch": 0.905728,
      "grad_norm": 0.35413662216924896,
      "learning_rate": 8.876316399463425e-07,
      "loss": 1.3589,
      "step": 3538
    },
    {
      "epoch": 0.905984,
      "grad_norm": 0.3552808857259192,
      "learning_rate": 8.82849870161322e-07,
      "loss": 1.3442,
      "step": 3539
    },
    {
      "epoch": 0.90624,
      "grad_norm": 0.36181244543573926,
      "learning_rate": 8.780807244398737e-07,
      "loss": 1.3979,
      "step": 3540
    },
    {
      "epoch": 0.906496,
      "grad_norm": 0.35560535607807114,
      "learning_rate": 8.733242059313163e-07,
      "loss": 1.3436,
      "step": 3541
    },
    {
      "epoch": 0.906752,
      "grad_norm": 0.3624070040495191,
      "learning_rate": 8.685803177766283e-07,
      "loss": 1.3387,
      "step": 3542
    },
    {
      "epoch": 0.907008,
      "grad_norm": 0.3468424947599529,
      "learning_rate": 8.638490631084484e-07,
      "loss": 1.3332,
      "step": 3543
    },
    {
      "epoch": 0.907264,
      "grad_norm": 0.3473556233891367,
      "learning_rate": 8.591304450510795e-07,
      "loss": 1.3046,
      "step": 3544
    },
    {
      "epoch": 0.90752,
      "grad_norm": 0.35274617851864754,
      "learning_rate": 8.544244667204671e-07,
      "loss": 1.3253,
      "step": 3545
    },
    {
      "epoch": 0.907776,
      "grad_norm": 0.35377422193487745,
      "learning_rate": 8.497311312242207e-07,
      "loss": 1.2953,
      "step": 3546
    },
    {
      "epoch": 0.908032,
      "grad_norm": 0.3534469150335654,
      "learning_rate": 8.45050441661599e-07,
      "loss": 1.3303,
      "step": 3547
    },
    {
      "epoch": 0.908288,
      "grad_norm": 0.3530571345460178,
      "learning_rate": 8.403824011235051e-07,
      "loss": 1.3261,
      "step": 3548
    },
    {
      "epoch": 0.908544,
      "grad_norm": 0.3541815933467247,
      "learning_rate": 8.357270126924932e-07,
      "loss": 1.2923,
      "step": 3549
    },
    {
      "epoch": 0.9088,
      "grad_norm": 0.3543523099932284,
      "learning_rate": 8.310842794427665e-07,
      "loss": 1.3631,
      "step": 3550
    },
    {
      "epoch": 0.909056,
      "grad_norm": 0.35075062904742343,
      "learning_rate": 8.264542044401614e-07,
      "loss": 1.3317,
      "step": 3551
    },
    {
      "epoch": 0.909312,
      "grad_norm": 0.34901006617657604,
      "learning_rate": 8.218367907421631e-07,
      "loss": 1.3034,
      "step": 3552
    },
    {
      "epoch": 0.909568,
      "grad_norm": 0.35269704887541753,
      "learning_rate": 8.172320413978974e-07,
      "loss": 1.3037,
      "step": 3553
    },
    {
      "epoch": 0.909824,
      "grad_norm": 0.34530473409550927,
      "learning_rate": 8.126399594481161e-07,
      "loss": 1.266,
      "step": 3554
    },
    {
      "epoch": 0.91008,
      "grad_norm": 0.36272862896057384,
      "learning_rate": 8.080605479252158e-07,
      "loss": 1.3365,
      "step": 3555
    },
    {
      "epoch": 0.910336,
      "grad_norm": 0.34708744921729207,
      "learning_rate": 8.034938098532286e-07,
      "loss": 1.3029,
      "step": 3556
    },
    {
      "epoch": 0.910592,
      "grad_norm": 0.3577031768206211,
      "learning_rate": 7.989397482478045e-07,
      "loss": 1.3458,
      "step": 3557
    },
    {
      "epoch": 0.910848,
      "grad_norm": 0.3523283772959642,
      "learning_rate": 7.94398366116238e-07,
      "loss": 1.3472,
      "step": 3558
    },
    {
      "epoch": 0.911104,
      "grad_norm": 0.3492356768066558,
      "learning_rate": 7.89869666457439e-07,
      "loss": 1.3004,
      "step": 3559
    },
    {
      "epoch": 0.91136,
      "grad_norm": 0.35548566578995267,
      "learning_rate": 7.853536522619487e-07,
      "loss": 1.3341,
      "step": 3560
    },
    {
      "epoch": 0.911616,
      "grad_norm": 0.3397966662926343,
      "learning_rate": 7.808503265119305e-07,
      "loss": 1.2506,
      "step": 3561
    },
    {
      "epoch": 0.911872,
      "grad_norm": 0.35766724906151354,
      "learning_rate": 7.763596921811701e-07,
      "loss": 1.3526,
      "step": 3562
    },
    {
      "epoch": 0.912128,
      "grad_norm": 0.36615505784750024,
      "learning_rate": 7.718817522350641e-07,
      "loss": 1.3585,
      "step": 3563
    },
    {
      "epoch": 0.912384,
      "grad_norm": 0.3609716457036324,
      "learning_rate": 7.674165096306385e-07,
      "loss": 1.3286,
      "step": 3564
    },
    {
      "epoch": 0.91264,
      "grad_norm": 0.35884694355978014,
      "learning_rate": 7.629639673165257e-07,
      "loss": 1.3681,
      "step": 3565
    },
    {
      "epoch": 0.912896,
      "grad_norm": 0.3493205127783696,
      "learning_rate": 7.585241282329759e-07,
      "loss": 1.3097,
      "step": 3566
    },
    {
      "epoch": 0.913152,
      "grad_norm": 0.35234067483955706,
      "learning_rate": 7.54096995311846e-07,
      "loss": 1.3075,
      "step": 3567
    },
    {
      "epoch": 0.913408,
      "grad_norm": 0.3422673158689805,
      "learning_rate": 7.496825714766131e-07,
      "loss": 1.3031,
      "step": 3568
    },
    {
      "epoch": 0.913664,
      "grad_norm": 0.3487649561865556,
      "learning_rate": 7.452808596423455e-07,
      "loss": 1.305,
      "step": 3569
    },
    {
      "epoch": 0.91392,
      "grad_norm": 0.3495500009257064,
      "learning_rate": 7.408918627157313e-07,
      "loss": 1.3191,
      "step": 3570
    },
    {
      "epoch": 0.914176,
      "grad_norm": 0.35515631178923585,
      "learning_rate": 7.365155835950544e-07,
      "loss": 1.3391,
      "step": 3571
    },
    {
      "epoch": 0.914432,
      "grad_norm": 0.35636128106871723,
      "learning_rate": 7.321520251702052e-07,
      "loss": 1.3668,
      "step": 3572
    },
    {
      "epoch": 0.914688,
      "grad_norm": 0.3568889919829385,
      "learning_rate": 7.2780119032267e-07,
      "loss": 1.3627,
      "step": 3573
    },
    {
      "epoch": 0.914944,
      "grad_norm": 0.34801472271629263,
      "learning_rate": 7.234630819255373e-07,
      "loss": 1.3187,
      "step": 3574
    },
    {
      "epoch": 0.9152,
      "grad_norm": 0.3500874463141885,
      "learning_rate": 7.191377028434865e-07,
      "loss": 1.3115,
      "step": 3575
    },
    {
      "epoch": 0.915456,
      "grad_norm": 0.3501867814308275,
      "learning_rate": 7.148250559327952e-07,
      "loss": 1.3007,
      "step": 3576
    },
    {
      "epoch": 0.915712,
      "grad_norm": 0.3473709029393759,
      "learning_rate": 7.105251440413297e-07,
      "loss": 1.2907,
      "step": 3577
    },
    {
      "epoch": 0.915968,
      "grad_norm": 0.35195192466079234,
      "learning_rate": 7.062379700085497e-07,
      "loss": 1.3329,
      "step": 3578
    },
    {
      "epoch": 0.916224,
      "grad_norm": 0.3498404380329524,
      "learning_rate": 7.019635366655042e-07,
      "loss": 1.2875,
      "step": 3579
    },
    {
      "epoch": 0.91648,
      "grad_norm": 0.3432855288126404,
      "learning_rate": 6.977018468348262e-07,
      "loss": 1.3058,
      "step": 3580
    },
    {
      "epoch": 0.916736,
      "grad_norm": 0.35190461485939584,
      "learning_rate": 6.934529033307336e-07,
      "loss": 1.3191,
      "step": 3581
    },
    {
      "epoch": 0.916992,
      "grad_norm": 0.36128314755081226,
      "learning_rate": 6.892167089590307e-07,
      "loss": 1.3425,
      "step": 3582
    },
    {
      "epoch": 0.917248,
      "grad_norm": 0.35179188915227366,
      "learning_rate": 6.849932665170955e-07,
      "loss": 1.3287,
      "step": 3583
    },
    {
      "epoch": 0.917504,
      "grad_norm": 0.3555656532667669,
      "learning_rate": 6.807825787938949e-07,
      "loss": 1.3421,
      "step": 3584
    },
    {
      "epoch": 0.91776,
      "grad_norm": 0.35981232075237485,
      "learning_rate": 6.76584648569969e-07,
      "loss": 1.3524,
      "step": 3585
    },
    {
      "epoch": 0.918016,
      "grad_norm": 0.34986271659257606,
      "learning_rate": 6.72399478617427e-07,
      "loss": 1.3197,
      "step": 3586
    },
    {
      "epoch": 0.918272,
      "grad_norm": 0.3495191257761887,
      "learning_rate": 6.68227071699965e-07,
      "loss": 1.335,
      "step": 3587
    },
    {
      "epoch": 0.918528,
      "grad_norm": 0.35421725254568864,
      "learning_rate": 6.640674305728368e-07,
      "loss": 1.3649,
      "step": 3588
    },
    {
      "epoch": 0.918784,
      "grad_norm": 0.349382237458626,
      "learning_rate": 6.599205579828738e-07,
      "loss": 1.3059,
      "step": 3589
    },
    {
      "epoch": 0.91904,
      "grad_norm": 0.352900572864467,
      "learning_rate": 6.557864566684791e-07,
      "loss": 1.3199,
      "step": 3590
    },
    {
      "epoch": 0.919296,
      "grad_norm": 0.3519799639777463,
      "learning_rate": 6.516651293596154e-07,
      "loss": 1.3236,
      "step": 3591
    },
    {
      "epoch": 0.919552,
      "grad_norm": 0.3402143955230197,
      "learning_rate": 6.475565787778149e-07,
      "loss": 1.2663,
      "step": 3592
    },
    {
      "epoch": 0.919808,
      "grad_norm": 0.3508149155039848,
      "learning_rate": 6.434608076361648e-07,
      "loss": 1.3068,
      "step": 3593
    },
    {
      "epoch": 0.920064,
      "grad_norm": 0.34897341981899427,
      "learning_rate": 6.393778186393263e-07,
      "loss": 1.3138,
      "step": 3594
    },
    {
      "epoch": 0.92032,
      "grad_norm": 0.35628121902740284,
      "learning_rate": 6.353076144835069e-07,
      "loss": 1.3098,
      "step": 3595
    },
    {
      "epoch": 0.920576,
      "grad_norm": 0.3488804125423955,
      "learning_rate": 6.312501978564811e-07,
      "loss": 1.2959,
      "step": 3596
    },
    {
      "epoch": 0.920832,
      "grad_norm": 0.3558162811999687,
      "learning_rate": 6.272055714375747e-07,
      "loss": 1.3137,
      "step": 3597
    },
    {
      "epoch": 0.921088,
      "grad_norm": 0.34914395992482733,
      "learning_rate": 6.231737378976665e-07,
      "loss": 1.2808,
      "step": 3598
    },
    {
      "epoch": 0.921344,
      "grad_norm": 0.3493222158036344,
      "learning_rate": 6.191546998991916e-07,
      "loss": 1.3073,
      "step": 3599
    },
    {
      "epoch": 0.9216,
      "grad_norm": 0.4128103526246106,
      "learning_rate": 6.151484600961266e-07,
      "loss": 1.375,
      "step": 3600
    },
    {
      "epoch": 0.921856,
      "grad_norm": 0.35118707265119375,
      "learning_rate": 6.111550211340112e-07,
      "loss": 1.3123,
      "step": 3601
    },
    {
      "epoch": 0.922112,
      "grad_norm": 0.3436505578914053,
      "learning_rate": 6.071743856499179e-07,
      "loss": 1.2843,
      "step": 3602
    },
    {
      "epoch": 0.922368,
      "grad_norm": 0.3594562982182878,
      "learning_rate": 6.032065562724754e-07,
      "loss": 1.3143,
      "step": 3603
    },
    {
      "epoch": 0.922624,
      "grad_norm": 0.37647005001464884,
      "learning_rate": 5.992515356218475e-07,
      "loss": 1.3222,
      "step": 3604
    },
    {
      "epoch": 0.92288,
      "grad_norm": 0.3456273548822862,
      "learning_rate": 5.953093263097476e-07,
      "loss": 1.2809,
      "step": 3605
    },
    {
      "epoch": 0.923136,
      "grad_norm": 0.3592822235877073,
      "learning_rate": 5.913799309394175e-07,
      "loss": 1.3254,
      "step": 3606
    },
    {
      "epoch": 0.923392,
      "grad_norm": 0.36279905948464897,
      "learning_rate": 5.874633521056483e-07,
      "loss": 1.3213,
      "step": 3607
    },
    {
      "epoch": 0.923648,
      "grad_norm": 0.34706910163143667,
      "learning_rate": 5.835595923947668e-07,
      "loss": 1.2785,
      "step": 3608
    },
    {
      "epoch": 0.923904,
      "grad_norm": 0.36184464960500284,
      "learning_rate": 5.796686543846264e-07,
      "loss": 1.2989,
      "step": 3609
    },
    {
      "epoch": 0.92416,
      "grad_norm": 0.4666822630572504,
      "learning_rate": 5.757905406446229e-07,
      "loss": 1.3156,
      "step": 3610
    },
    {
      "epoch": 0.924416,
      "grad_norm": 0.35158112123945845,
      "learning_rate": 5.719252537356768e-07,
      "loss": 1.3223,
      "step": 3611
    },
    {
      "epoch": 0.924672,
      "grad_norm": 0.35901118914065366,
      "learning_rate": 5.680727962102417e-07,
      "loss": 1.3346,
      "step": 3612
    },
    {
      "epoch": 0.924928,
      "grad_norm": 0.3507316866462203,
      "learning_rate": 5.642331706122983e-07,
      "loss": 1.3187,
      "step": 3613
    },
    {
      "epoch": 0.925184,
      "grad_norm": 0.3522079565993651,
      "learning_rate": 5.604063794773562e-07,
      "loss": 1.3196,
      "step": 3614
    },
    {
      "epoch": 0.92544,
      "grad_norm": 0.35484688663032976,
      "learning_rate": 5.56592425332445e-07,
      "loss": 1.3253,
      "step": 3615
    },
    {
      "epoch": 0.925696,
      "grad_norm": 0.5376993171619279,
      "learning_rate": 5.527913106961214e-07,
      "loss": 1.3195,
      "step": 3616
    },
    {
      "epoch": 0.925952,
      "grad_norm": 0.3523097804600051,
      "learning_rate": 5.490030380784661e-07,
      "loss": 1.3123,
      "step": 3617
    },
    {
      "epoch": 0.926208,
      "grad_norm": 0.3393998364806952,
      "learning_rate": 5.452276099810672e-07,
      "loss": 1.2768,
      "step": 3618
    },
    {
      "epoch": 0.926464,
      "grad_norm": 0.7972038959607683,
      "learning_rate": 5.414650288970436e-07,
      "loss": 1.3008,
      "step": 3619
    },
    {
      "epoch": 0.92672,
      "grad_norm": 0.3513779759490281,
      "learning_rate": 5.3771529731103e-07,
      "loss": 1.3265,
      "step": 3620
    },
    {
      "epoch": 0.926976,
      "grad_norm": 0.35427787038276953,
      "learning_rate": 5.339784176991636e-07,
      "loss": 1.3465,
      "step": 3621
    },
    {
      "epoch": 0.927232,
      "grad_norm": 0.3525359612351788,
      "learning_rate": 5.302543925291103e-07,
      "loss": 1.335,
      "step": 3622
    },
    {
      "epoch": 0.927488,
      "grad_norm": 0.3540421323396622,
      "learning_rate": 5.265432242600387e-07,
      "loss": 1.3197,
      "step": 3623
    },
    {
      "epoch": 0.927744,
      "grad_norm": 0.3391691953596326,
      "learning_rate": 5.228449153426263e-07,
      "loss": 1.2752,
      "step": 3624
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.45866546455341267,
      "learning_rate": 5.191594682190659e-07,
      "loss": 1.3489,
      "step": 3625
    },
    {
      "epoch": 0.928256,
      "grad_norm": 0.34174606361996795,
      "learning_rate": 5.154868853230488e-07,
      "loss": 1.2799,
      "step": 3626
    },
    {
      "epoch": 0.928512,
      "grad_norm": 0.35743804729337525,
      "learning_rate": 5.118271690797794e-07,
      "loss": 1.3501,
      "step": 3627
    },
    {
      "epoch": 0.928768,
      "grad_norm": 0.35012140001483644,
      "learning_rate": 5.0818032190596e-07,
      "loss": 1.3074,
      "step": 3628
    },
    {
      "epoch": 0.929024,
      "grad_norm": 0.34905979854782504,
      "learning_rate": 5.045463462097954e-07,
      "loss": 1.293,
      "step": 3629
    },
    {
      "epoch": 0.92928,
      "grad_norm": 0.36264173673598943,
      "learning_rate": 5.009252443909929e-07,
      "loss": 1.2885,
      "step": 3630
    },
    {
      "epoch": 0.929536,
      "grad_norm": 0.3436746431518763,
      "learning_rate": 4.973170188407573e-07,
      "loss": 1.2724,
      "step": 3631
    },
    {
      "epoch": 0.929792,
      "grad_norm": 0.34797870538608733,
      "learning_rate": 4.937216719417892e-07,
      "loss": 1.3166,
      "step": 3632
    },
    {
      "epoch": 0.930048,
      "grad_norm": 0.3523741839584143,
      "learning_rate": 4.901392060682853e-07,
      "loss": 1.3251,
      "step": 3633
    },
    {
      "epoch": 0.930304,
      "grad_norm": 0.34944842459917086,
      "learning_rate": 4.865696235859374e-07,
      "loss": 1.3032,
      "step": 3634
    },
    {
      "epoch": 0.93056,
      "grad_norm": 0.34102312564289916,
      "learning_rate": 4.830129268519334e-07,
      "loss": 1.2889,
      "step": 3635
    },
    {
      "epoch": 0.930816,
      "grad_norm": 0.3471565159973024,
      "learning_rate": 4.794691182149435e-07,
      "loss": 1.2901,
      "step": 3636
    },
    {
      "epoch": 0.931072,
      "grad_norm": 0.35192889952675166,
      "learning_rate": 4.7593820001513134e-07,
      "loss": 1.3194,
      "step": 3637
    },
    {
      "epoch": 0.931328,
      "grad_norm": 0.47049123091772027,
      "learning_rate": 4.724201745841495e-07,
      "loss": 1.3062,
      "step": 3638
    },
    {
      "epoch": 0.931584,
      "grad_norm": 0.3442210390393018,
      "learning_rate": 4.6891504424513733e-07,
      "loss": 1.2754,
      "step": 3639
    },
    {
      "epoch": 0.93184,
      "grad_norm": 0.34912641921433657,
      "learning_rate": 4.654228113127168e-07,
      "loss": 1.3301,
      "step": 3640
    },
    {
      "epoch": 0.932096,
      "grad_norm": 0.3472264652945926,
      "learning_rate": 4.619434780929966e-07,
      "loss": 1.3311,
      "step": 3641
    },
    {
      "epoch": 0.932352,
      "grad_norm": 0.3487894673998158,
      "learning_rate": 4.584770468835564e-07,
      "loss": 1.2905,
      "step": 3642
    },
    {
      "epoch": 0.932608,
      "grad_norm": 0.35570886866130924,
      "learning_rate": 4.5502351997347206e-07,
      "loss": 1.3149,
      "step": 3643
    },
    {
      "epoch": 0.932864,
      "grad_norm": 0.35757777786650197,
      "learning_rate": 4.515828996432836e-07,
      "loss": 1.3469,
      "step": 3644
    },
    {
      "epoch": 0.93312,
      "grad_norm": 0.34538228718386255,
      "learning_rate": 4.481551881650159e-07,
      "loss": 1.2926,
      "step": 3645
    },
    {
      "epoch": 0.933376,
      "grad_norm": 0.3542692201576496,
      "learning_rate": 4.4474038780216945e-07,
      "loss": 1.3555,
      "step": 3646
    },
    {
      "epoch": 0.933632,
      "grad_norm": 0.34896297061900816,
      "learning_rate": 4.413385008097204e-07,
      "loss": 1.291,
      "step": 3647
    },
    {
      "epoch": 0.933888,
      "grad_norm": 0.3582983596512249,
      "learning_rate": 4.379495294341052e-07,
      "loss": 1.3675,
      "step": 3648
    },
    {
      "epoch": 0.934144,
      "grad_norm": 0.3482172403823663,
      "learning_rate": 4.3457347591324695e-07,
      "loss": 1.3215,
      "step": 3649
    },
    {
      "epoch": 0.9344,
      "grad_norm": 0.3500905197844345,
      "learning_rate": 4.3121034247653124e-07,
      "loss": 1.3148,
      "step": 3650
    },
    {
      "epoch": 0.934656,
      "grad_norm": 0.35163533140821657,
      "learning_rate": 4.2786013134481273e-07,
      "loss": 1.3311,
      "step": 3651
    },
    {
      "epoch": 0.934912,
      "grad_norm": 0.35989463782872383,
      "learning_rate": 4.2452284473041285e-07,
      "loss": 1.2948,
      "step": 3652
    },
    {
      "epoch": 0.935168,
      "grad_norm": 0.34757994884300936,
      "learning_rate": 4.2119848483711757e-07,
      "loss": 1.3399,
      "step": 3653
    },
    {
      "epoch": 0.935424,
      "grad_norm": 0.3430665382796469,
      "learning_rate": 4.1788705386017757e-07,
      "loss": 1.2626,
      "step": 3654
    },
    {
      "epoch": 0.93568,
      "grad_norm": 0.3514060996218677,
      "learning_rate": 4.145885539863037e-07,
      "loss": 1.3337,
      "step": 3655
    },
    {
      "epoch": 0.935936,
      "grad_norm": 0.3627494674882456,
      "learning_rate": 4.1130298739367134e-07,
      "loss": 1.3405,
      "step": 3656
    },
    {
      "epoch": 0.936192,
      "grad_norm": 0.5677440041935232,
      "learning_rate": 4.080303562519117e-07,
      "loss": 1.3641,
      "step": 3657
    },
    {
      "epoch": 0.936448,
      "grad_norm": 0.35395362858250634,
      "learning_rate": 4.047706627221204e-07,
      "loss": 1.3414,
      "step": 3658
    },
    {
      "epoch": 0.936704,
      "grad_norm": 0.352600069479312,
      "learning_rate": 4.0152390895684013e-07,
      "loss": 1.3462,
      "step": 3659
    },
    {
      "epoch": 0.93696,
      "grad_norm": 0.3468000525896169,
      "learning_rate": 3.9829009710007804e-07,
      "loss": 1.2832,
      "step": 3660
    },
    {
      "epoch": 0.937216,
      "grad_norm": 0.3451692647891084,
      "learning_rate": 3.950692292872882e-07,
      "loss": 1.2733,
      "step": 3661
    },
    {
      "epoch": 0.937472,
      "grad_norm": 0.34673806818050334,
      "learning_rate": 3.918613076453781e-07,
      "loss": 1.2583,
      "step": 3662
    },
    {
      "epoch": 0.937728,
      "grad_norm": 0.3497431210258776,
      "learning_rate": 3.886663342927133e-07,
      "loss": 1.3481,
      "step": 3663
    },
    {
      "epoch": 0.937984,
      "grad_norm": 0.34610731095162506,
      "learning_rate": 3.8548431133909937e-07,
      "loss": 1.3053,
      "step": 3664
    },
    {
      "epoch": 0.93824,
      "grad_norm": 0.3473184463062814,
      "learning_rate": 3.823152408857933e-07,
      "loss": 1.3146,
      "step": 3665
    },
    {
      "epoch": 0.938496,
      "grad_norm": 0.34628643743692206,
      "learning_rate": 3.7915912502550333e-07,
      "loss": 1.3011,
      "step": 3666
    },
    {
      "epoch": 0.938752,
      "grad_norm": 0.34658919177669306,
      "learning_rate": 3.7601596584237567e-07,
      "loss": 1.2981,
      "step": 3667
    },
    {
      "epoch": 0.939008,
      "grad_norm": 0.3524424921528069,
      "learning_rate": 3.728857654120055e-07,
      "loss": 1.2861,
      "step": 3668
    },
    {
      "epoch": 0.939264,
      "grad_norm": 0.34919207181733214,
      "learning_rate": 3.6976852580142833e-07,
      "loss": 1.3512,
      "step": 3669
    },
    {
      "epoch": 0.93952,
      "grad_norm": 0.35094282376697344,
      "learning_rate": 3.666642490691241e-07,
      "loss": 1.3297,
      "step": 3670
    },
    {
      "epoch": 0.939776,
      "grad_norm": 0.35574517695435776,
      "learning_rate": 3.6357293726501097e-07,
      "loss": 1.3437,
      "step": 3671
    },
    {
      "epoch": 0.940032,
      "grad_norm": 0.35741602173527104,
      "learning_rate": 3.6049459243044483e-07,
      "loss": 1.3731,
      "step": 3672
    },
    {
      "epoch": 0.940288,
      "grad_norm": 0.3478913311254024,
      "learning_rate": 3.574292165982196e-07,
      "loss": 1.3146,
      "step": 3673
    },
    {
      "epoch": 0.940544,
      "grad_norm": 0.34489897805438563,
      "learning_rate": 3.543768117925606e-07,
      "loss": 1.3128,
      "step": 3674
    },
    {
      "epoch": 0.9408,
      "grad_norm": 0.35047016884341775,
      "learning_rate": 3.5133738002913756e-07,
      "loss": 1.3227,
      "step": 3675
    },
    {
      "epoch": 0.941056,
      "grad_norm": 0.34769112544260977,
      "learning_rate": 3.483109233150472e-07,
      "loss": 1.3085,
      "step": 3676
    },
    {
      "epoch": 0.941312,
      "grad_norm": 0.34327564729510157,
      "learning_rate": 3.4529744364881325e-07,
      "loss": 1.2747,
      "step": 3677
    },
    {
      "epoch": 0.941568,
      "grad_norm": 0.34683755489292145,
      "learning_rate": 3.422969430204015e-07,
      "loss": 1.3138,
      "step": 3678
    },
    {
      "epoch": 0.941824,
      "grad_norm": 0.34695239091406455,
      "learning_rate": 3.393094234111982e-07,
      "loss": 1.3127,
      "step": 3679
    },
    {
      "epoch": 0.94208,
      "grad_norm": 0.49883208371500354,
      "learning_rate": 3.3633488679402084e-07,
      "loss": 1.3649,
      "step": 3680
    },
    {
      "epoch": 0.942336,
      "grad_norm": 0.35123787382563937,
      "learning_rate": 3.3337333513311144e-07,
      "loss": 1.3135,
      "step": 3681
    },
    {
      "epoch": 0.942592,
      "grad_norm": 0.3510292588493751,
      "learning_rate": 3.304247703841412e-07,
      "loss": 1.3012,
      "step": 3682
    },
    {
      "epoch": 0.942848,
      "grad_norm": 0.3511237754765743,
      "learning_rate": 3.274891944942016e-07,
      "loss": 1.3405,
      "step": 3683
    },
    {
      "epoch": 0.943104,
      "grad_norm": 0.3418045164664127,
      "learning_rate": 3.245666094018107e-07,
      "loss": 1.3059,
      "step": 3684
    },
    {
      "epoch": 0.94336,
      "grad_norm": 0.3592500676442756,
      "learning_rate": 3.216570170369004e-07,
      "loss": 1.33,
      "step": 3685
    },
    {
      "epoch": 0.943616,
      "grad_norm": 0.38227780782334864,
      "learning_rate": 3.187604193208338e-07,
      "loss": 1.3226,
      "step": 3686
    },
    {
      "epoch": 0.943872,
      "grad_norm": 0.34467305191341785,
      "learning_rate": 3.158768181663874e-07,
      "loss": 1.2867,
      "step": 3687
    },
    {
      "epoch": 0.944128,
      "grad_norm": 0.34753212688026563,
      "learning_rate": 3.1300621547775136e-07,
      "loss": 1.3276,
      "step": 3688
    },
    {
      "epoch": 0.944384,
      "grad_norm": 0.34974246363777745,
      "learning_rate": 3.1014861315053825e-07,
      "loss": 1.2963,
      "step": 3689
    },
    {
      "epoch": 0.94464,
      "grad_norm": 0.35571572153488806,
      "learning_rate": 3.0730401307177414e-07,
      "loss": 1.2788,
      "step": 3690
    },
    {
      "epoch": 0.944896,
      "grad_norm": 0.36202426303980256,
      "learning_rate": 3.044724171198965e-07,
      "loss": 1.3217,
      "step": 3691
    },
    {
      "epoch": 0.945152,
      "grad_norm": 0.3570628538372465,
      "learning_rate": 3.0165382716475846e-07,
      "loss": 1.363,
      "step": 3692
    },
    {
      "epoch": 0.945408,
      "grad_norm": 0.358890884545062,
      "learning_rate": 2.988482450676267e-07,
      "loss": 1.3436,
      "step": 3693
    },
    {
      "epoch": 0.945664,
      "grad_norm": 0.35378478010708747,
      "learning_rate": 2.960556726811703e-07,
      "loss": 1.3533,
      "step": 3694
    },
    {
      "epoch": 0.94592,
      "grad_norm": 0.34934337285706285,
      "learning_rate": 2.9327611184947644e-07,
      "loss": 1.3084,
      "step": 3695
    },
    {
      "epoch": 0.946176,
      "grad_norm": 0.3457964973613574,
      "learning_rate": 2.905095644080369e-07,
      "loss": 1.3064,
      "step": 3696
    },
    {
      "epoch": 0.946432,
      "grad_norm": 0.34617646920800554,
      "learning_rate": 2.877560321837436e-07,
      "loss": 1.3112,
      "step": 3697
    },
    {
      "epoch": 0.946688,
      "grad_norm": 0.3443321129515638,
      "learning_rate": 2.8501551699490206e-07,
      "loss": 1.2978,
      "step": 3698
    },
    {
      "epoch": 0.946944,
      "grad_norm": 0.357534570546533,
      "learning_rate": 2.822880206512224e-07,
      "loss": 1.3371,
      "step": 3699
    },
    {
      "epoch": 0.9472,
      "grad_norm": 0.3464414449377667,
      "learning_rate": 2.795735449538084e-07,
      "loss": 1.286,
      "step": 3700
    },
    {
      "epoch": 0.947456,
      "grad_norm": 0.36388462021714646,
      "learning_rate": 2.7687209169517506e-07,
      "loss": 1.3667,
      "step": 3701
    },
    {
      "epoch": 0.947712,
      "grad_norm": 0.3468309764742381,
      "learning_rate": 2.7418366265923535e-07,
      "loss": 1.2784,
      "step": 3702
    },
    {
      "epoch": 0.947968,
      "grad_norm": 0.3478519166322828,
      "learning_rate": 2.715082596212981e-07,
      "loss": 1.3151,
      "step": 3703
    },
    {
      "epoch": 0.948224,
      "grad_norm": 0.3535066419612759,
      "learning_rate": 2.6884588434807456e-07,
      "loss": 1.3372,
      "step": 3704
    },
    {
      "epoch": 0.94848,
      "grad_norm": 0.3516442038861694,
      "learning_rate": 2.661965385976739e-07,
      "loss": 1.3391,
      "step": 3705
    },
    {
      "epoch": 0.948736,
      "grad_norm": 0.349091325807776,
      "learning_rate": 2.6356022411959447e-07,
      "loss": 1.3025,
      "step": 3706
    },
    {
      "epoch": 0.948992,
      "grad_norm": 0.3482138134701709,
      "learning_rate": 2.60936942654737e-07,
      "loss": 1.3055,
      "step": 3707
    },
    {
      "epoch": 0.949248,
      "grad_norm": 0.3586782407513873,
      "learning_rate": 2.5832669593539583e-07,
      "loss": 1.3103,
      "step": 3708
    },
    {
      "epoch": 0.949504,
      "grad_norm": 0.3424976332019905,
      "learning_rate": 2.557294856852477e-07,
      "loss": 1.2735,
      "step": 3709
    },
    {
      "epoch": 0.94976,
      "grad_norm": 0.3440927238857434,
      "learning_rate": 2.531453136193718e-07,
      "loss": 1.2836,
      "step": 3710
    },
    {
      "epoch": 0.950016,
      "grad_norm": 0.34619957050631434,
      "learning_rate": 2.5057418144423196e-07,
      "loss": 1.3102,
      "step": 3711
    },
    {
      "epoch": 0.950272,
      "grad_norm": 0.34480422347330414,
      "learning_rate": 2.480160908576834e-07,
      "loss": 1.3185,
      "step": 3712
    },
    {
      "epoch": 0.950528,
      "grad_norm": 0.3527175693088532,
      "learning_rate": 2.454710435489682e-07,
      "loss": 1.345,
      "step": 3713
    },
    {
      "epoch": 0.950784,
      "grad_norm": 0.3494360040131702,
      "learning_rate": 2.4293904119871535e-07,
      "loss": 1.3118,
      "step": 3714
    },
    {
      "epoch": 0.95104,
      "grad_norm": 0.3508094221365896,
      "learning_rate": 2.4042008547894067e-07,
      "loss": 1.3038,
      "step": 3715
    },
    {
      "epoch": 0.951296,
      "grad_norm": 0.36291145023559584,
      "learning_rate": 2.3791417805304028e-07,
      "loss": 1.3189,
      "step": 3716
    },
    {
      "epoch": 0.951552,
      "grad_norm": 0.3522821537551188,
      "learning_rate": 2.3542132057580159e-07,
      "loss": 1.3144,
      "step": 3717
    },
    {
      "epoch": 0.951808,
      "grad_norm": 0.35677326081905264,
      "learning_rate": 2.3294151469338777e-07,
      "loss": 1.3296,
      "step": 3718
    },
    {
      "epoch": 0.952064,
      "grad_norm": 0.3492062684946765,
      "learning_rate": 2.3047476204334452e-07,
      "loss": 1.309,
      "step": 3719
    },
    {
      "epoch": 0.95232,
      "grad_norm": 0.3519135669562277,
      "learning_rate": 2.2802106425460214e-07,
      "loss": 1.3367,
      "step": 3720
    },
    {
      "epoch": 0.952576,
      "grad_norm": 0.3543611436031582,
      "learning_rate": 2.2558042294746451e-07,
      "loss": 1.2959,
      "step": 3721
    },
    {
      "epoch": 0.952832,
      "grad_norm": 0.35237920779317133,
      "learning_rate": 2.2315283973361578e-07,
      "loss": 1.3345,
      "step": 3722
    },
    {
      "epoch": 0.953088,
      "grad_norm": 0.35645770761222434,
      "learning_rate": 2.2073831621611808e-07,
      "loss": 1.3609,
      "step": 3723
    },
    {
      "epoch": 0.953344,
      "grad_norm": 0.3592486549506144,
      "learning_rate": 2.1833685398940706e-07,
      "loss": 1.3794,
      "step": 3724
    },
    {
      "epoch": 0.9536,
      "grad_norm": 0.3546579554828751,
      "learning_rate": 2.1594845463929648e-07,
      "loss": 1.3185,
      "step": 3725
    },
    {
      "epoch": 0.953856,
      "grad_norm": 0.35656780226398344,
      "learning_rate": 2.1357311974297356e-07,
      "loss": 1.3147,
      "step": 3726
    },
    {
      "epoch": 0.954112,
      "grad_norm": 0.3432444771848927,
      "learning_rate": 2.1121085086899694e-07,
      "loss": 1.2913,
      "step": 3727
    },
    {
      "epoch": 0.954368,
      "grad_norm": 0.3519864137637237,
      "learning_rate": 2.088616495772966e-07,
      "loss": 1.2992,
      "step": 3728
    },
    {
      "epoch": 0.954624,
      "grad_norm": 0.3524566513481728,
      "learning_rate": 2.065255174191716e-07,
      "loss": 1.3428,
      "step": 3729
    },
    {
      "epoch": 0.95488,
      "grad_norm": 0.35300664407009874,
      "learning_rate": 2.0420245593729903e-07,
      "loss": 1.328,
      "step": 3730
    },
    {
      "epoch": 0.955136,
      "grad_norm": 0.3486016719590103,
      "learning_rate": 2.0189246666571405e-07,
      "loss": 1.2997,
      "step": 3731
    },
    {
      "epoch": 0.955392,
      "grad_norm": 0.355178532421379,
      "learning_rate": 1.9959555112982976e-07,
      "loss": 1.2968,
      "step": 3732
    },
    {
      "epoch": 0.955648,
      "grad_norm": 0.3465684809567164,
      "learning_rate": 1.973117108464173e-07,
      "loss": 1.3177,
      "step": 3733
    },
    {
      "epoch": 0.955904,
      "grad_norm": 0.3465817343905661,
      "learning_rate": 1.9504094732361479e-07,
      "loss": 1.2841,
      "step": 3734
    },
    {
      "epoch": 0.95616,
      "grad_norm": 0.35698292501992435,
      "learning_rate": 1.9278326206093156e-07,
      "loss": 1.3239,
      "step": 3735
    },
    {
      "epoch": 0.956416,
      "grad_norm": 0.5456809682709015,
      "learning_rate": 1.9053865654923287e-07,
      "loss": 1.3278,
      "step": 3736
    },
    {
      "epoch": 0.956672,
      "grad_norm": 0.3544702143320293,
      "learning_rate": 1.8830713227075305e-07,
      "loss": 1.3301,
      "step": 3737
    },
    {
      "epoch": 0.956928,
      "grad_norm": 0.3601081575765337,
      "learning_rate": 1.860886906990844e-07,
      "loss": 1.3762,
      "step": 3738
    },
    {
      "epoch": 0.957184,
      "grad_norm": 0.345714280447751,
      "learning_rate": 1.8388333329918185e-07,
      "loss": 1.2802,
      "step": 3739
    },
    {
      "epoch": 0.95744,
      "grad_norm": 0.34768500981504163,
      "learning_rate": 1.8169106152735815e-07,
      "loss": 1.3022,
      "step": 3740
    },
    {
      "epoch": 0.957696,
      "grad_norm": 0.35618702694685117,
      "learning_rate": 1.7951187683128645e-07,
      "loss": 1.361,
      "step": 3741
    },
    {
      "epoch": 0.957952,
      "grad_norm": 0.3465077292678786,
      "learning_rate": 1.7734578064999564e-07,
      "loss": 1.2947,
      "step": 3742
    },
    {
      "epoch": 0.958208,
      "grad_norm": 0.3493504939649855,
      "learning_rate": 1.7519277441387705e-07,
      "loss": 1.2827,
      "step": 3743
    },
    {
      "epoch": 0.958464,
      "grad_norm": 0.35354899600410666,
      "learning_rate": 1.7305285954467345e-07,
      "loss": 1.3258,
      "step": 3744
    },
    {
      "epoch": 0.95872,
      "grad_norm": 0.3438226209638275,
      "learning_rate": 1.709260374554811e-07,
      "loss": 1.2919,
      "step": 3745
    },
    {
      "epoch": 0.958976,
      "grad_norm": 0.3445092186975909,
      "learning_rate": 1.6881230955075435e-07,
      "loss": 1.2886,
      "step": 3746
    },
    {
      "epoch": 0.959232,
      "grad_norm": 0.3520236721544524,
      "learning_rate": 1.667116772262989e-07,
      "loss": 1.3493,
      "step": 3747
    },
    {
      "epoch": 0.959488,
      "grad_norm": 0.35481108577562315,
      "learning_rate": 1.64624141869274e-07,
      "loss": 1.3398,
      "step": 3748
    },
    {
      "epoch": 0.959744,
      "grad_norm": 0.3609011756511318,
      "learning_rate": 1.6254970485819032e-07,
      "loss": 1.3687,
      "step": 3749
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3589404262752977,
      "learning_rate": 1.6048836756290542e-07,
      "loss": 1.3725,
      "step": 3750
    },
    {
      "epoch": 0.960256,
      "grad_norm": 0.3588016450402552,
      "learning_rate": 1.5844013134463265e-07,
      "loss": 1.3681,
      "step": 3751
    },
    {
      "epoch": 0.960512,
      "grad_norm": 0.34706286850723644,
      "learning_rate": 1.5640499755592786e-07,
      "loss": 1.319,
      "step": 3752
    },
    {
      "epoch": 0.960768,
      "grad_norm": 0.3507443944698336,
      "learning_rate": 1.5438296754069827e-07,
      "loss": 1.3018,
      "step": 3753
    },
    {
      "epoch": 0.961024,
      "grad_norm": 0.43183290494199866,
      "learning_rate": 1.5237404263419575e-07,
      "loss": 1.2673,
      "step": 3754
    },
    {
      "epoch": 0.96128,
      "grad_norm": 0.40656043549809767,
      "learning_rate": 1.503782241630236e-07,
      "loss": 1.3371,
      "step": 3755
    },
    {
      "epoch": 0.961536,
      "grad_norm": 0.34710465562296583,
      "learning_rate": 1.4839551344512093e-07,
      "loss": 1.309,
      "step": 3756
    },
    {
      "epoch": 0.961792,
      "grad_norm": 0.3524167642896055,
      "learning_rate": 1.464259117897804e-07,
      "loss": 1.2886,
      "step": 3757
    },
    {
      "epoch": 0.962048,
      "grad_norm": 0.3515247920397948,
      "learning_rate": 1.4446942049762824e-07,
      "loss": 1.31,
      "step": 3758
    },
    {
      "epoch": 0.962304,
      "grad_norm": 0.348908550499184,
      "learning_rate": 1.425260408606466e-07,
      "loss": 1.3251,
      "step": 3759
    },
    {
      "epoch": 0.96256,
      "grad_norm": 0.34624626425083055,
      "learning_rate": 1.4059577416214442e-07,
      "loss": 1.2766,
      "step": 3760
    },
    {
      "epoch": 0.962816,
      "grad_norm": 0.3531943801194355,
      "learning_rate": 1.3867862167678438e-07,
      "loss": 1.3247,
      "step": 3761
    },
    {
      "epoch": 0.963072,
      "grad_norm": 0.353600873387186,
      "learning_rate": 1.3677458467056258e-07,
      "loss": 1.3404,
      "step": 3762
    },
    {
      "epoch": 0.963328,
      "grad_norm": 0.35528309808096603,
      "learning_rate": 1.3488366440081112e-07,
      "loss": 1.3426,
      "step": 3763
    },
    {
      "epoch": 0.963584,
      "grad_norm": 0.3635077162095375,
      "learning_rate": 1.3300586211620892e-07,
      "loss": 1.333,
      "step": 3764
    },
    {
      "epoch": 0.96384,
      "grad_norm": 0.36589848227862065,
      "learning_rate": 1.3114117905676625e-07,
      "loss": 1.3554,
      "step": 3765
    },
    {
      "epoch": 0.964096,
      "grad_norm": 0.37992993333715097,
      "learning_rate": 1.2928961645383154e-07,
      "loss": 1.3734,
      "step": 3766
    },
    {
      "epoch": 0.964352,
      "grad_norm": 0.35435015291180033,
      "learning_rate": 1.2745117553009113e-07,
      "loss": 1.3592,
      "step": 3767
    },
    {
      "epoch": 0.964608,
      "grad_norm": 0.35166604687861014,
      "learning_rate": 1.2562585749956058e-07,
      "loss": 1.2974,
      "step": 3768
    },
    {
      "epoch": 0.964864,
      "grad_norm": 0.35372962632464877,
      "learning_rate": 1.2381366356759794e-07,
      "loss": 1.3719,
      "step": 3769
    },
    {
      "epoch": 0.96512,
      "grad_norm": 0.35725880146078065,
      "learning_rate": 1.2201459493088818e-07,
      "loss": 1.3179,
      "step": 3770
    },
    {
      "epoch": 0.965376,
      "grad_norm": 0.35200565173623977,
      "learning_rate": 1.202286527774521e-07,
      "loss": 1.3383,
      "step": 3771
    },
    {
      "epoch": 0.965632,
      "grad_norm": 0.3455199299990574,
      "learning_rate": 1.1845583828664187e-07,
      "loss": 1.2903,
      "step": 3772
    },
    {
      "epoch": 0.965888,
      "grad_norm": 0.34786125419147174,
      "learning_rate": 1.1669615262914102e-07,
      "loss": 1.2915,
      "step": 3773
    },
    {
      "epoch": 0.966144,
      "grad_norm": 0.3447054310299593,
      "learning_rate": 1.1494959696696006e-07,
      "loss": 1.3032,
      "step": 3774
    },
    {
      "epoch": 0.9664,
      "grad_norm": 0.3465532618193849,
      "learning_rate": 1.132161724534453e-07,
      "loss": 1.3015,
      "step": 3775
    },
    {
      "epoch": 0.966656,
      "grad_norm": 0.34990752761191285,
      "learning_rate": 1.1149588023326773e-07,
      "loss": 1.3193,
      "step": 3776
    },
    {
      "epoch": 0.966912,
      "grad_norm": 0.3590234148770652,
      "learning_rate": 1.0978872144242536e-07,
      "loss": 1.336,
      "step": 3777
    },
    {
      "epoch": 0.967168,
      "grad_norm": 0.3397683826863262,
      "learning_rate": 1.080946972082475e-07,
      "loss": 1.285,
      "step": 3778
    },
    {
      "epoch": 0.967424,
      "grad_norm": 0.35458614593671955,
      "learning_rate": 1.06413808649386e-07,
      "loss": 1.3381,
      "step": 3779
    },
    {
      "epoch": 0.96768,
      "grad_norm": 0.3453849687078781,
      "learning_rate": 1.0474605687581962e-07,
      "loss": 1.2928,
      "step": 3780
    },
    {
      "epoch": 0.967936,
      "grad_norm": 0.35192355638990197,
      "learning_rate": 1.0309144298885631e-07,
      "loss": 1.3334,
      "step": 3781
    },
    {
      "epoch": 0.968192,
      "grad_norm": 0.35295276683997634,
      "learning_rate": 1.0144996808112207e-07,
      "loss": 1.3172,
      "step": 3782
    },
    {
      "epoch": 0.968448,
      "grad_norm": 0.35374365132960894,
      "learning_rate": 9.982163323656979e-08,
      "loss": 1.3449,
      "step": 3783
    },
    {
      "epoch": 0.968704,
      "grad_norm": 0.3572162078738479,
      "learning_rate": 9.820643953047715e-08,
      "loss": 1.3314,
      "step": 3784
    },
    {
      "epoch": 0.96896,
      "grad_norm": 0.3560131219232579,
      "learning_rate": 9.660438802943761e-08,
      "loss": 1.3432,
      "step": 3785
    },
    {
      "epoch": 0.969216,
      "grad_norm": 0.3543923590989018,
      "learning_rate": 9.501547979137383e-08,
      "loss": 1.3357,
      "step": 3786
    },
    {
      "epoch": 0.969472,
      "grad_norm": 0.3754716752990853,
      "learning_rate": 9.343971586552648e-08,
      "loss": 1.3661,
      "step": 3787
    },
    {
      "epoch": 0.969728,
      "grad_norm": 0.3530671159274057,
      "learning_rate": 9.187709729245432e-08,
      "loss": 1.3161,
      "step": 3788
    },
    {
      "epoch": 0.969984,
      "grad_norm": 0.36079833164501074,
      "learning_rate": 9.032762510403636e-08,
      "loss": 1.3572,
      "step": 3789
    },
    {
      "epoch": 0.97024,
      "grad_norm": 0.34745292722408605,
      "learning_rate": 8.879130032346973e-08,
      "loss": 1.307,
      "step": 3790
    },
    {
      "epoch": 0.970496,
      "grad_norm": 0.353658918345041,
      "learning_rate": 8.726812396527174e-08,
      "loss": 1.3273,
      "step": 3791
    },
    {
      "epoch": 0.970752,
      "grad_norm": 0.3477556545406439,
      "learning_rate": 8.575809703527782e-08,
      "loss": 1.3038,
      "step": 3792
    },
    {
      "epoch": 0.971008,
      "grad_norm": 0.35208745353302107,
      "learning_rate": 8.4261220530637e-08,
      "loss": 1.334,
      "step": 3793
    },
    {
      "epoch": 0.971264,
      "grad_norm": 0.3439914711689724,
      "learning_rate": 8.277749543981861e-08,
      "loss": 1.3121,
      "step": 3794
    },
    {
      "epoch": 0.97152,
      "grad_norm": 0.3650107700926055,
      "learning_rate": 8.130692274259888e-08,
      "loss": 1.3579,
      "step": 3795
    },
    {
      "epoch": 0.971776,
      "grad_norm": 0.359581419702592,
      "learning_rate": 7.98495034100788e-08,
      "loss": 1.3002,
      "step": 3796
    },
    {
      "epoch": 0.972032,
      "grad_norm": 0.3437992474925989,
      "learning_rate": 7.840523840467074e-08,
      "loss": 1.2895,
      "step": 3797
    },
    {
      "epoch": 0.972288,
      "grad_norm": 0.3457647055287586,
      "learning_rate": 7.697412868009402e-08,
      "loss": 1.3022,
      "step": 3798
    },
    {
      "epoch": 0.972544,
      "grad_norm": 0.35333293078351485,
      "learning_rate": 7.555617518139047e-08,
      "loss": 1.3607,
      "step": 3799
    },
    {
      "epoch": 0.9728,
      "grad_norm": 0.34217617369114206,
      "learning_rate": 7.415137884490886e-08,
      "loss": 1.2955,
      "step": 3800
    },
    {
      "epoch": 0.973056,
      "grad_norm": 0.3839225444299931,
      "learning_rate": 7.275974059830714e-08,
      "loss": 1.3439,
      "step": 3801
    },
    {
      "epoch": 0.973312,
      "grad_norm": 0.34351615172085204,
      "learning_rate": 7.138126136056134e-08,
      "loss": 1.2672,
      "step": 3802
    },
    {
      "epoch": 0.973568,
      "grad_norm": 0.35270898812469614,
      "learning_rate": 7.001594204195216e-08,
      "loss": 1.3505,
      "step": 3803
    },
    {
      "epoch": 0.973824,
      "grad_norm": 0.3528885161914668,
      "learning_rate": 6.866378354407399e-08,
      "loss": 1.3299,
      "step": 3804
    },
    {
      "epoch": 0.97408,
      "grad_norm": 0.3460153308469529,
      "learning_rate": 6.73247867598259e-08,
      "loss": 1.2987,
      "step": 3805
    },
    {
      "epoch": 0.974336,
      "grad_norm": 0.3463713192196872,
      "learning_rate": 6.599895257341837e-08,
      "loss": 1.2844,
      "step": 3806
    },
    {
      "epoch": 0.974592,
      "grad_norm": 0.3491302687297275,
      "learning_rate": 6.468628186037107e-08,
      "loss": 1.3157,
      "step": 3807
    },
    {
      "epoch": 0.974848,
      "grad_norm": 0.3500896335344266,
      "learning_rate": 6.33867754875106e-08,
      "loss": 1.319,
      "step": 3808
    },
    {
      "epoch": 0.975104,
      "grad_norm": 0.34294023316509276,
      "learning_rate": 6.210043431296608e-08,
      "loss": 1.2966,
      "step": 3809
    },
    {
      "epoch": 0.97536,
      "grad_norm": 0.358061324604674,
      "learning_rate": 6.082725918618248e-08,
      "loss": 1.3331,
      "step": 3810
    },
    {
      "epoch": 0.975616,
      "grad_norm": 0.35589401139098437,
      "learning_rate": 5.956725094789839e-08,
      "loss": 1.3371,
      "step": 3811
    },
    {
      "epoch": 0.975872,
      "grad_norm": 0.3516566823584661,
      "learning_rate": 5.832041043016601e-08,
      "loss": 1.3356,
      "step": 3812
    },
    {
      "epoch": 0.976128,
      "grad_norm": 0.36194903329381856,
      "learning_rate": 5.708673845634005e-08,
      "loss": 1.3476,
      "step": 3813
    },
    {
      "epoch": 0.976384,
      "grad_norm": 0.3578975714692887,
      "learning_rate": 5.586623584107998e-08,
      "loss": 1.3454,
      "step": 3814
    },
    {
      "epoch": 0.97664,
      "grad_norm": 0.35674902074061354,
      "learning_rate": 5.465890339034774e-08,
      "loss": 1.3614,
      "step": 3815
    },
    {
      "epoch": 0.976896,
      "grad_norm": 0.3424596876888329,
      "learning_rate": 5.3464741901407826e-08,
      "loss": 1.3084,
      "step": 3816
    },
    {
      "epoch": 0.977152,
      "grad_norm": 0.35434995759530646,
      "learning_rate": 5.228375216282944e-08,
      "loss": 1.3251,
      "step": 3817
    },
    {
      "epoch": 0.977408,
      "grad_norm": 0.34634332649603466,
      "learning_rate": 5.1115934954482086e-08,
      "loss": 1.3115,
      "step": 3818
    },
    {
      "epoch": 0.977664,
      "grad_norm": 0.3496581868508856,
      "learning_rate": 4.996129104753555e-08,
      "loss": 1.3228,
      "step": 3819
    },
    {
      "epoch": 0.97792,
      "grad_norm": 0.36266905873655914,
      "learning_rate": 4.881982120446438e-08,
      "loss": 1.3732,
      "step": 3820
    },
    {
      "epoch": 0.978176,
      "grad_norm": 0.3495836678056541,
      "learning_rate": 4.769152617904116e-08,
      "loss": 1.2845,
      "step": 3821
    },
    {
      "epoch": 0.978432,
      "grad_norm": 0.34404354553081207,
      "learning_rate": 4.657640671633656e-08,
      "loss": 1.2761,
      "step": 3822
    },
    {
      "epoch": 0.978688,
      "grad_norm": 0.35880344722926893,
      "learning_rate": 4.547446355272378e-08,
      "loss": 1.3567,
      "step": 3823
    },
    {
      "epoch": 0.978944,
      "grad_norm": 0.3548435710041313,
      "learning_rate": 4.438569741587628e-08,
      "loss": 1.3442,
      "step": 3824
    },
    {
      "epoch": 0.9792,
      "grad_norm": 0.3517953042632612,
      "learning_rate": 4.331010902476118e-08,
      "loss": 1.3246,
      "step": 3825
    },
    {
      "epoch": 0.979456,
      "grad_norm": 0.35696292875956626,
      "learning_rate": 4.2247699089648097e-08,
      "loss": 1.3278,
      "step": 3826
    },
    {
      "epoch": 0.979712,
      "grad_norm": 0.34590955714027705,
      "learning_rate": 4.119846831210028e-08,
      "loss": 1.2393,
      "step": 3827
    },
    {
      "epoch": 0.979968,
      "grad_norm": 0.3749090380556617,
      "learning_rate": 4.0162417384981275e-08,
      "loss": 1.3397,
      "step": 3828
    },
    {
      "epoch": 0.980224,
      "grad_norm": 0.3543619764178797,
      "learning_rate": 3.9139546992450484e-08,
      "loss": 1.3276,
      "step": 3829
    },
    {
      "epoch": 0.98048,
      "grad_norm": 0.3400678089862932,
      "learning_rate": 3.8129857809965365e-08,
      "loss": 1.2707,
      "step": 3830
    },
    {
      "epoch": 0.980736,
      "grad_norm": 0.35319035096323276,
      "learning_rate": 3.713335050427258e-08,
      "loss": 1.3485,
      "step": 3831
    },
    {
      "epoch": 0.980992,
      "grad_norm": 0.34793421781368084,
      "learning_rate": 3.61500257334213e-08,
      "loss": 1.3093,
      "step": 3832
    },
    {
      "epoch": 0.981248,
      "grad_norm": 0.33948996610897936,
      "learning_rate": 3.517988414675211e-08,
      "loss": 1.2442,
      "step": 3833
    },
    {
      "epoch": 0.981504,
      "grad_norm": 0.3546913139508053,
      "learning_rate": 3.4222926384899214e-08,
      "loss": 1.3515,
      "step": 3834
    },
    {
      "epoch": 0.98176,
      "grad_norm": 0.3502936772356789,
      "learning_rate": 3.32791530797949e-08,
      "loss": 1.3225,
      "step": 3835
    },
    {
      "epoch": 0.982016,
      "grad_norm": 0.3469309752003446,
      "learning_rate": 3.234856485466287e-08,
      "loss": 1.3245,
      "step": 3836
    },
    {
      "epoch": 0.982272,
      "grad_norm": 0.3514363370840732,
      "learning_rate": 3.1431162324015994e-08,
      "loss": 1.3348,
      "step": 3837
    },
    {
      "epoch": 0.982528,
      "grad_norm": 0.3474835208051609,
      "learning_rate": 3.0526946093667466e-08,
      "loss": 1.2925,
      "step": 3838
    },
    {
      "epoch": 0.982784,
      "grad_norm": 0.3514853656160855,
      "learning_rate": 2.963591676071742e-08,
      "loss": 1.3171,
      "step": 3839
    },
    {
      "epoch": 0.98304,
      "grad_norm": 0.3516273643810689,
      "learning_rate": 2.8758074913559642e-08,
      "loss": 1.3069,
      "step": 3840
    },
    {
      "epoch": 0.983296,
      "grad_norm": 0.3444475398946807,
      "learning_rate": 2.7893421131877096e-08,
      "loss": 1.3142,
      "step": 3841
    },
    {
      "epoch": 0.983552,
      "grad_norm": 0.35634014024484667,
      "learning_rate": 2.7041955986650824e-08,
      "loss": 1.3741,
      "step": 3842
    },
    {
      "epoch": 0.983808,
      "grad_norm": 0.34768763336693664,
      "learning_rate": 2.6203680040146617e-08,
      "loss": 1.3068,
      "step": 3843
    },
    {
      "epoch": 0.984064,
      "grad_norm": 0.35525404903128205,
      "learning_rate": 2.5378593845919454e-08,
      "loss": 1.3354,
      "step": 3844
    },
    {
      "epoch": 0.98432,
      "grad_norm": 0.3626379386770382,
      "learning_rate": 2.4566697948822382e-08,
      "loss": 1.3531,
      "step": 3845
    },
    {
      "epoch": 0.984576,
      "grad_norm": 0.347284377269125,
      "learning_rate": 2.3767992884988767e-08,
      "loss": 1.2826,
      "step": 3846
    },
    {
      "epoch": 0.984832,
      "grad_norm": 0.3480190393510757,
      "learning_rate": 2.2982479181847817e-08,
      "loss": 1.2948,
      "step": 3847
    },
    {
      "epoch": 0.985088,
      "grad_norm": 0.40031189243474147,
      "learning_rate": 2.2210157358113492e-08,
      "loss": 1.297,
      "step": 3848
    },
    {
      "epoch": 0.985344,
      "grad_norm": 0.3461403164240625,
      "learning_rate": 2.145102792379339e-08,
      "loss": 1.2863,
      "step": 3849
    },
    {
      "epoch": 0.9856,
      "grad_norm": 0.3520763451840515,
      "learning_rate": 2.0705091380182065e-08,
      "loss": 1.3367,
      "step": 3850
    },
    {
      "epoch": 0.985856,
      "grad_norm": 0.3539777080019041,
      "learning_rate": 1.9972348219854386e-08,
      "loss": 1.3638,
      "step": 3851
    },
    {
      "epoch": 0.986112,
      "grad_norm": 0.3594904642845286,
      "learning_rate": 1.9252798926685522e-08,
      "loss": 1.3619,
      "step": 3852
    },
    {
      "epoch": 0.986368,
      "grad_norm": 0.35024791118469184,
      "learning_rate": 1.8546443975830943e-08,
      "loss": 1.3545,
      "step": 3853
    },
    {
      "epoch": 0.986624,
      "grad_norm": 0.3470396670229881,
      "learning_rate": 1.7853283833730863e-08,
      "loss": 1.2893,
      "step": 3854
    },
    {
      "epoch": 0.98688,
      "grad_norm": 0.34403460218880644,
      "learning_rate": 1.7173318958119134e-08,
      "loss": 1.3023,
      "step": 3855
    },
    {
      "epoch": 0.987136,
      "grad_norm": 0.3544374524748081,
      "learning_rate": 1.6506549798009918e-08,
      "loss": 1.3251,
      "step": 3856
    },
    {
      "epoch": 0.987392,
      "grad_norm": 0.35555715661586185,
      "learning_rate": 1.5852976793708784e-08,
      "loss": 1.3082,
      "step": 3857
    },
    {
      "epoch": 0.987648,
      "grad_norm": 0.3511935107519599,
      "learning_rate": 1.521260037680161e-08,
      "loss": 1.3498,
      "step": 3858
    },
    {
      "epoch": 0.987904,
      "grad_norm": 0.34334834143606763,
      "learning_rate": 1.4585420970163466e-08,
      "loss": 1.2909,
      "step": 3859
    },
    {
      "epoch": 0.98816,
      "grad_norm": 0.3572379175600008,
      "learning_rate": 1.3971438987954167e-08,
      "loss": 1.3418,
      "step": 3860
    },
    {
      "epoch": 0.988416,
      "grad_norm": 0.35440490915641165,
      "learning_rate": 1.33706548356205e-08,
      "loss": 1.3511,
      "step": 3861
    },
    {
      "epoch": 0.988672,
      "grad_norm": 0.3580339355955322,
      "learning_rate": 1.2783068909889563e-08,
      "loss": 1.3147,
      "step": 3862
    },
    {
      "epoch": 0.988928,
      "grad_norm": 0.35739545462064787,
      "learning_rate": 1.2208681598775418e-08,
      "loss": 1.3625,
      "step": 3863
    },
    {
      "epoch": 0.989184,
      "grad_norm": 0.3472079378979475,
      "learning_rate": 1.1647493281576883e-08,
      "loss": 1.2752,
      "step": 3864
    },
    {
      "epoch": 0.98944,
      "grad_norm": 0.35260321075238293,
      "learning_rate": 1.109950432887752e-08,
      "loss": 1.336,
      "step": 3865
    },
    {
      "epoch": 0.989696,
      "grad_norm": 0.3535175802036694,
      "learning_rate": 1.05647151025412e-08,
      "loss": 1.3145,
      "step": 3866
    },
    {
      "epoch": 0.989952,
      "grad_norm": 0.34943144852472,
      "learning_rate": 1.0043125955718768e-08,
      "loss": 1.371,
      "step": 3867
    },
    {
      "epoch": 0.990208,
      "grad_norm": 0.33723738252045143,
      "learning_rate": 9.534737232843595e-09,
      "loss": 1.2792,
      "step": 3868
    },
    {
      "epoch": 0.990464,
      "grad_norm": 0.35362043984295866,
      "learning_rate": 9.039549269629355e-09,
      "loss": 1.3615,
      "step": 3869
    },
    {
      "epoch": 0.99072,
      "grad_norm": 0.3520687631289754,
      "learning_rate": 8.557562393076701e-09,
      "loss": 1.3399,
      "step": 3870
    },
    {
      "epoch": 0.990976,
      "grad_norm": 0.35390613023195366,
      "learning_rate": 8.08877692146659e-09,
      "loss": 1.3467,
      "step": 3871
    },
    {
      "epoch": 0.991232,
      "grad_norm": 0.34448707278488316,
      "learning_rate": 7.633193164364727e-09,
      "loss": 1.3042,
      "step": 3872
    },
    {
      "epoch": 0.991488,
      "grad_norm": 0.35563086254485854,
      "learning_rate": 7.190811422612687e-09,
      "loss": 1.3461,
      "step": 3873
    },
    {
      "epoch": 0.991744,
      "grad_norm": 0.3509166264142058,
      "learning_rate": 6.761631988341233e-09,
      "loss": 1.3177,
      "step": 3874
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.3630141134148411,
      "learning_rate": 6.345655144961438e-09,
      "loss": 1.3002,
      "step": 3875
    },
    {
      "epoch": 0.992256,
      "grad_norm": 0.3541541011937014,
      "learning_rate": 5.9428811671602415e-09,
      "loss": 1.3336,
      "step": 3876
    },
    {
      "epoch": 0.992512,
      "grad_norm": 0.35104907283775955,
      "learning_rate": 5.5533103209159945e-09,
      "loss": 1.3427,
      "step": 3877
    },
    {
      "epoch": 0.992768,
      "grad_norm": 0.3366049178999771,
      "learning_rate": 5.176942863480694e-09,
      "loss": 1.2664,
      "step": 3878
    },
    {
      "epoch": 0.993024,
      "grad_norm": 0.3501018580938512,
      "learning_rate": 4.8137790433888665e-09,
      "loss": 1.3132,
      "step": 3879
    },
    {
      "epoch": 0.99328,
      "grad_norm": 0.3738265291377611,
      "learning_rate": 4.463819100457567e-09,
      "loss": 1.3084,
      "step": 3880
    },
    {
      "epoch": 0.993536,
      "grad_norm": 0.3443566307145647,
      "learning_rate": 4.127063265781938e-09,
      "loss": 1.2984,
      "step": 3881
    },
    {
      "epoch": 0.993792,
      "grad_norm": 0.34915148412781555,
      "learning_rate": 3.803511761744094e-09,
      "loss": 1.311,
      "step": 3882
    },
    {
      "epoch": 0.994048,
      "grad_norm": 0.3650515319126288,
      "learning_rate": 3.4931648019975728e-09,
      "loss": 1.3433,
      "step": 3883
    },
    {
      "epoch": 0.994304,
      "grad_norm": 0.3542045592400462,
      "learning_rate": 3.1960225914828834e-09,
      "loss": 1.346,
      "step": 3884
    },
    {
      "epoch": 0.99456,
      "grad_norm": 0.34564596416562104,
      "learning_rate": 2.9120853264186232e-09,
      "loss": 1.3124,
      "step": 3885
    },
    {
      "epoch": 0.994816,
      "grad_norm": 0.33972248263531846,
      "learning_rate": 2.6413531943036976e-09,
      "loss": 1.2934,
      "step": 3886
    },
    {
      "epoch": 0.995072,
      "grad_norm": 0.3467753361555826,
      "learning_rate": 2.383826373915099e-09,
      "loss": 1.3308,
      "step": 3887
    },
    {
      "epoch": 0.995328,
      "grad_norm": 0.3690600203431167,
      "learning_rate": 2.1395050353145707e-09,
      "loss": 1.3299,
      "step": 3888
    },
    {
      "epoch": 0.995584,
      "grad_norm": 0.36844653127531746,
      "learning_rate": 1.908389339837502e-09,
      "loss": 1.3553,
      "step": 3889
    },
    {
      "epoch": 0.99584,
      "grad_norm": 0.34891045298644996,
      "learning_rate": 1.6904794401040313e-09,
      "loss": 1.3165,
      "step": 3890
    },
    {
      "epoch": 0.996096,
      "grad_norm": 0.34309651033192795,
      "learning_rate": 1.4857754800101654e-09,
      "loss": 1.2882,
      "step": 3891
    },
    {
      "epoch": 0.996352,
      "grad_norm": 0.3640368935262463,
      "learning_rate": 1.2942775947322184e-09,
      "loss": 1.3411,
      "step": 3892
    },
    {
      "epoch": 0.996608,
      "grad_norm": 0.3646503582741394,
      "learning_rate": 1.1159859107290337e-09,
      "loss": 1.3399,
      "step": 3893
    },
    {
      "epoch": 0.996864,
      "grad_norm": 0.3524402467497575,
      "learning_rate": 9.509005457331022e-10,
      "loss": 1.3675,
      "step": 3894
    },
    {
      "epoch": 0.99712,
      "grad_norm": 0.34898679795961723,
      "learning_rate": 7.990216087594426e-10,
      "loss": 1.322,
      "step": 3895
    },
    {
      "epoch": 0.997376,
      "grad_norm": 0.3487734891703071,
      "learning_rate": 6.603492001033828e-10,
      "loss": 1.301,
      "step": 3896
    },
    {
      "epoch": 0.997632,
      "grad_norm": 0.34534764107641575,
      "learning_rate": 5.348834113361179e-10,
      "loss": 1.2745,
      "step": 3897
    },
    {
      "epoch": 0.997888,
      "grad_norm": 0.3524019403509108,
      "learning_rate": 4.226243253091511e-10,
      "loss": 1.3533,
      "step": 3898
    },
    {
      "epoch": 0.998144,
      "grad_norm": 0.3489700492235033,
      "learning_rate": 3.2357201615429434e-10,
      "loss": 1.326,
      "step": 3899
    },
    {
      "epoch": 0.9984,
      "grad_norm": 0.3415355486302458,
      "learning_rate": 2.3772654928144733e-10,
      "loss": 1.3004,
      "step": 3900
    },
    {
      "epoch": 0.998656,
      "grad_norm": 0.35418162815142945,
      "learning_rate": 1.6508798137637727e-10,
      "loss": 1.3324,
      "step": 3901
    },
    {
      "epoch": 0.998912,
      "grad_norm": 0.3463959582994453,
      "learning_rate": 1.0565636040960059e-10,
      "loss": 1.3021,
      "step": 3902
    },
    {
      "epoch": 0.999168,
      "grad_norm": 0.34322502965955426,
      "learning_rate": 5.943172562306032e-11,
      "loss": 1.2871,
      "step": 3903
    },
    {
      "epoch": 0.999424,
      "grad_norm": 0.3523476709598277,
      "learning_rate": 2.64141075456692e-11,
      "loss": 1.3023,
      "step": 3904
    },
    {
      "epoch": 0.99968,
      "grad_norm": 0.34721163075993766,
      "learning_rate": 6.6035279755460865e-12,
      "loss": 1.3415,
      "step": 3905
    },
    {
      "epoch": 0.999936,
      "grad_norm": 0.3528875622081801,
      "learning_rate": 0.0,
      "loss": 1.3614,
      "step": 3906
    },
    {
      "epoch": 0.999936,
      "step": 3906,
      "total_flos": 4458641369333760.0,
      "train_loss": 1.408147927589192,
      "train_runtime": 51696.0221,
      "train_samples_per_second": 14.508,
      "train_steps_per_second": 0.076
    }
  ],
  "logging_steps": 1,
  "max_steps": 3906,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 400,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4458641369333760.0,
  "train_batch_size": 3,
  "trial_name": null,
  "trial_params": null
}