{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 5703,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 5.123785379357265,
      "learning_rate": 1.1627906976744187e-07,
      "loss": 0.9373,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.2104847663651706,
      "learning_rate": 2.3255813953488374e-07,
      "loss": 0.9208,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.756083943450628,
      "learning_rate": 3.488372093023256e-07,
      "loss": 0.9866,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.055372869916307,
      "learning_rate": 4.651162790697675e-07,
      "loss": 0.9716,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.469421246664214,
      "learning_rate": 5.813953488372094e-07,
      "loss": 0.9276,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.928697884589698,
      "learning_rate": 6.976744186046513e-07,
      "loss": 0.9851,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.62560547269135,
      "learning_rate": 8.139534883720931e-07,
      "loss": 0.9533,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.089097037044055,
      "learning_rate": 9.30232558139535e-07,
      "loss": 0.9517,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.100439409219032,
      "learning_rate": 1.0465116279069768e-06,
      "loss": 0.9363,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2556002889568596,
      "learning_rate": 1.1627906976744188e-06,
      "loss": 0.9148,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.306207404930218,
      "learning_rate": 1.2790697674418605e-06,
      "loss": 0.901,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.5389847167804622,
      "learning_rate": 1.3953488372093025e-06,
      "loss": 0.9759,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.700435045141162,
      "learning_rate": 1.5116279069767443e-06,
      "loss": 0.872,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5980686872778085,
      "learning_rate": 1.6279069767441862e-06,
      "loss": 0.9175,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7142623633539618,
      "learning_rate": 1.7441860465116282e-06,
      "loss": 0.8917,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4801917390433823,
      "learning_rate": 1.86046511627907e-06,
      "loss": 0.8673,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3793021087188682,
      "learning_rate": 1.976744186046512e-06,
      "loss": 0.8718,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3846309953436386,
      "learning_rate": 2.0930232558139536e-06,
      "loss": 0.8575,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2066656185882516,
      "learning_rate": 2.2093023255813954e-06,
      "loss": 0.8925,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.308663206951885,
      "learning_rate": 2.3255813953488376e-06,
      "loss": 0.8719,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2934313950714285,
      "learning_rate": 2.4418604651162793e-06,
      "loss": 0.8592,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4417810067967454,
      "learning_rate": 2.558139534883721e-06,
      "loss": 0.9104,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1126123999161326,
      "learning_rate": 2.674418604651163e-06,
      "loss": 0.8737,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9485540137003432,
      "learning_rate": 2.790697674418605e-06,
      "loss": 0.8499,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9395766006297345,
      "learning_rate": 2.9069767441860468e-06,
      "loss": 0.8718,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1275164173107504,
      "learning_rate": 3.0232558139534885e-06,
      "loss": 0.8482,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3490148785483373,
      "learning_rate": 3.1395348837209307e-06,
      "loss": 0.8689,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4580563449584825,
      "learning_rate": 3.2558139534883724e-06,
      "loss": 0.8501,
      "step": 28
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2435468146143105,
      "learning_rate": 3.372093023255814e-06,
      "loss": 0.8764,
      "step": 29
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8934267531893595,
      "learning_rate": 3.4883720930232564e-06,
      "loss": 0.8838,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3006625933904927,
      "learning_rate": 3.6046511627906977e-06,
      "loss": 0.8207,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.068620976009512,
      "learning_rate": 3.72093023255814e-06,
      "loss": 0.8439,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2248888843967605,
      "learning_rate": 3.837209302325582e-06,
      "loss": 0.8551,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8121311628094761,
      "learning_rate": 3.953488372093024e-06,
      "loss": 0.833,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.030558761815383,
      "learning_rate": 4.0697674418604655e-06,
      "loss": 0.8945,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9549142917110657,
      "learning_rate": 4.186046511627907e-06,
      "loss": 0.8378,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9718144089369075,
      "learning_rate": 4.302325581395349e-06,
      "loss": 0.8695,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7907654079361924,
      "learning_rate": 4.418604651162791e-06,
      "loss": 0.8331,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8442743610877188,
      "learning_rate": 4.5348837209302326e-06,
      "loss": 0.8723,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7766847990968313,
      "learning_rate": 4.651162790697675e-06,
      "loss": 0.8855,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9559345520186775,
      "learning_rate": 4.767441860465117e-06,
      "loss": 0.854,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9522718342241372,
      "learning_rate": 4.883720930232559e-06,
      "loss": 0.8512,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8334348584109049,
      "learning_rate": 5e-06,
      "loss": 0.8251,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.840605010141594,
      "learning_rate": 5.116279069767442e-06,
      "loss": 0.8316,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0288583399717293,
      "learning_rate": 5.232558139534885e-06,
      "loss": 0.9028,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8941486793545607,
      "learning_rate": 5.348837209302326e-06,
      "loss": 0.8032,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8500052678848606,
      "learning_rate": 5.465116279069767e-06,
      "loss": 0.8378,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8622886023968779,
      "learning_rate": 5.58139534883721e-06,
      "loss": 0.8262,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8764643614331641,
      "learning_rate": 5.697674418604652e-06,
      "loss": 0.8419,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9001363079908395,
      "learning_rate": 5.8139534883720935e-06,
      "loss": 0.839,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8799921404678155,
      "learning_rate": 5.930232558139536e-06,
      "loss": 0.8182,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8897719952696889,
      "learning_rate": 6.046511627906977e-06,
      "loss": 0.8226,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7064051831978361,
      "learning_rate": 6.162790697674419e-06,
      "loss": 0.8021,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8800348720980075,
      "learning_rate": 6.279069767441861e-06,
      "loss": 0.8479,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0428466307026807,
      "learning_rate": 6.395348837209303e-06,
      "loss": 0.8349,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8749314512723254,
      "learning_rate": 6.511627906976745e-06,
      "loss": 0.7966,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8342355709261795,
      "learning_rate": 6.627906976744186e-06,
      "loss": 0.8457,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.137035048551639,
      "learning_rate": 6.744186046511628e-06,
      "loss": 0.8196,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.006545003268565,
      "learning_rate": 6.86046511627907e-06,
      "loss": 0.7932,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7800085871164048,
      "learning_rate": 6.976744186046513e-06,
      "loss": 0.7929,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9073128060160859,
      "learning_rate": 7.0930232558139545e-06,
      "loss": 0.8921,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.063750441910888,
      "learning_rate": 7.209302325581395e-06,
      "loss": 0.8774,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9209960563331431,
      "learning_rate": 7.325581395348837e-06,
      "loss": 0.8758,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.856257812647561,
      "learning_rate": 7.44186046511628e-06,
      "loss": 0.795,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9936447395297762,
      "learning_rate": 7.5581395348837215e-06,
      "loss": 0.8115,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7629412397459028,
      "learning_rate": 7.674418604651164e-06,
      "loss": 0.7675,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0661033667469286,
      "learning_rate": 7.790697674418605e-06,
      "loss": 0.8123,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8604499475884759,
      "learning_rate": 7.906976744186048e-06,
      "loss": 0.8149,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7564891340841373,
      "learning_rate": 8.023255813953488e-06,
      "loss": 0.7682,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7329540140667775,
      "learning_rate": 8.139534883720931e-06,
      "loss": 0.8067,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.864079170484272,
      "learning_rate": 8.255813953488374e-06,
      "loss": 0.7829,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8870411371499851,
      "learning_rate": 8.372093023255815e-06,
      "loss": 0.8029,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8732154888643351,
      "learning_rate": 8.488372093023256e-06,
      "loss": 0.8097,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8458286344598275,
      "learning_rate": 8.604651162790698e-06,
      "loss": 0.8176,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8824779035363114,
      "learning_rate": 8.72093023255814e-06,
      "loss": 0.8433,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8608279610568753,
      "learning_rate": 8.837209302325582e-06,
      "loss": 0.8023,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7973380533785053,
      "learning_rate": 8.953488372093024e-06,
      "loss": 0.7997,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8671163651991461,
      "learning_rate": 9.069767441860465e-06,
      "loss": 0.8309,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7844803100354425,
      "learning_rate": 9.186046511627908e-06,
      "loss": 0.7867,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7635054950236818,
      "learning_rate": 9.30232558139535e-06,
      "loss": 0.8057,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8701139655534844,
      "learning_rate": 9.418604651162791e-06,
      "loss": 0.7853,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0703238628963716,
      "learning_rate": 9.534883720930234e-06,
      "loss": 0.7759,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9077817194450286,
      "learning_rate": 9.651162790697676e-06,
      "loss": 0.8182,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8572394984813722,
      "learning_rate": 9.767441860465117e-06,
      "loss": 0.8238,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7659803184623406,
      "learning_rate": 9.883720930232558e-06,
      "loss": 0.8,
      "step": 85
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9352670822175781,
      "learning_rate": 1e-05,
      "loss": 0.855,
      "step": 86
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9193380883373826,
      "learning_rate": 1.0116279069767443e-05,
      "loss": 0.8384,
      "step": 87
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9343933190951808,
      "learning_rate": 1.0232558139534884e-05,
      "loss": 0.7619,
      "step": 88
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8273336986325632,
      "learning_rate": 1.0348837209302327e-05,
      "loss": 0.8076,
      "step": 89
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0578461473727447,
      "learning_rate": 1.046511627906977e-05,
      "loss": 0.7684,
      "step": 90
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7567381473047289,
      "learning_rate": 1.058139534883721e-05,
      "loss": 0.7767,
      "step": 91
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1415926352256296,
      "learning_rate": 1.0697674418604651e-05,
      "loss": 0.8329,
      "step": 92
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8480414141922111,
      "learning_rate": 1.0813953488372094e-05,
      "loss": 0.7703,
      "step": 93
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9889127034483878,
      "learning_rate": 1.0930232558139535e-05,
      "loss": 0.8019,
      "step": 94
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8455987929389573,
      "learning_rate": 1.1046511627906977e-05,
      "loss": 0.7766,
      "step": 95
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9684126560558763,
      "learning_rate": 1.116279069767442e-05,
      "loss": 0.7498,
      "step": 96
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9664833907312704,
      "learning_rate": 1.1279069767441861e-05,
      "loss": 0.7774,
      "step": 97
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8962118721026515,
      "learning_rate": 1.1395348837209304e-05,
      "loss": 0.7786,
      "step": 98
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8716529667780261,
      "learning_rate": 1.1511627906976746e-05,
      "loss": 0.7645,
      "step": 99
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8321559671738246,
      "learning_rate": 1.1627906976744187e-05,
      "loss": 0.7883,
      "step": 100
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7765342321373482,
      "learning_rate": 1.174418604651163e-05,
      "loss": 0.777,
      "step": 101
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8352328783831124,
      "learning_rate": 1.1860465116279072e-05,
      "loss": 0.8032,
      "step": 102
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8553830386635267,
      "learning_rate": 1.1976744186046511e-05,
      "loss": 0.8029,
      "step": 103
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8331691093786251,
      "learning_rate": 1.2093023255813954e-05,
      "loss": 0.7697,
      "step": 104
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8582555508381573,
      "learning_rate": 1.2209302325581397e-05,
      "loss": 0.761,
      "step": 105
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.945727532510146,
      "learning_rate": 1.2325581395348838e-05,
      "loss": 0.7407,
      "step": 106
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1228885803788857,
      "learning_rate": 1.244186046511628e-05,
      "loss": 0.7679,
      "step": 107
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0553926705081207,
      "learning_rate": 1.2558139534883723e-05,
      "loss": 0.8289,
      "step": 108
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8743878267280357,
      "learning_rate": 1.2674418604651164e-05,
      "loss": 0.75,
      "step": 109
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1077406594698664,
      "learning_rate": 1.2790697674418606e-05,
      "loss": 0.7768,
      "step": 110
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2088716666843915,
      "learning_rate": 1.2906976744186049e-05,
      "loss": 0.7541,
      "step": 111
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8312347487235417,
      "learning_rate": 1.302325581395349e-05,
      "loss": 0.8121,
      "step": 112
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.542526862944443,
      "learning_rate": 1.313953488372093e-05,
      "loss": 0.7811,
      "step": 113
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1706123996049158,
      "learning_rate": 1.3255813953488372e-05,
      "loss": 0.7555,
      "step": 114
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2410765301353446,
      "learning_rate": 1.3372093023255814e-05,
      "loss": 0.7938,
      "step": 115
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2000010215731538,
      "learning_rate": 1.3488372093023257e-05,
      "loss": 0.7951,
      "step": 116
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0226069037947592,
      "learning_rate": 1.36046511627907e-05,
      "loss": 0.7782,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9796355316289806,
      "learning_rate": 1.372093023255814e-05,
      "loss": 0.795,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9267209270966663,
      "learning_rate": 1.3837209302325583e-05,
      "loss": 0.7891,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9779477032784839,
      "learning_rate": 1.3953488372093025e-05,
      "loss": 0.7788,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8904927671811802,
      "learning_rate": 1.4069767441860466e-05,
      "loss": 0.7429,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9183065702790141,
      "learning_rate": 1.4186046511627909e-05,
      "loss": 0.7828,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9709459698603393,
      "learning_rate": 1.4302325581395352e-05,
      "loss": 0.7643,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.871184707079809,
      "learning_rate": 1.441860465116279e-05,
      "loss": 0.7941,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9882279740390536,
      "learning_rate": 1.4534883720930233e-05,
      "loss": 0.8095,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9060620720106839,
      "learning_rate": 1.4651162790697674e-05,
      "loss": 0.7606,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.090577371819845,
      "learning_rate": 1.4767441860465117e-05,
      "loss": 0.8068,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.044351289978564,
      "learning_rate": 1.488372093023256e-05,
      "loss": 0.7929,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9242872736838782,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.7838,
      "step": 129
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8764152218050635,
      "learning_rate": 1.5116279069767443e-05,
      "loss": 0.7847,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9479988426643892,
      "learning_rate": 1.5232558139534886e-05,
      "loss": 0.7933,
      "step": 131
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8689459495669496,
      "learning_rate": 1.5348837209302328e-05,
      "loss": 0.7742,
      "step": 132
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8150883583722773,
      "learning_rate": 1.546511627906977e-05,
      "loss": 0.7605,
      "step": 133
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1355257517954898,
      "learning_rate": 1.558139534883721e-05,
      "loss": 0.762,
      "step": 134
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7885906456493739,
      "learning_rate": 1.569767441860465e-05,
      "loss": 0.7522,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9396233416716444,
      "learning_rate": 1.5813953488372095e-05,
      "loss": 0.8062,
      "step": 136
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9485299970804153,
      "learning_rate": 1.5930232558139536e-05,
      "loss": 0.7657,
      "step": 137
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9085988577288342,
      "learning_rate": 1.6046511627906977e-05,
      "loss": 0.7817,
      "step": 138
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8531883087308522,
      "learning_rate": 1.616279069767442e-05,
      "loss": 0.8217,
      "step": 139
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0903517858998402,
      "learning_rate": 1.6279069767441862e-05,
      "loss": 0.7782,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1622160508217594,
      "learning_rate": 1.6395348837209303e-05,
      "loss": 0.8127,
      "step": 141
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7379747675787053,
      "learning_rate": 1.6511627906976747e-05,
      "loss": 0.7893,
      "step": 142
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2466289162813258,
      "learning_rate": 1.6627906976744188e-05,
      "loss": 0.781,
      "step": 143
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9815360848678397,
      "learning_rate": 1.674418604651163e-05,
      "loss": 0.7656,
      "step": 144
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9100632675623148,
      "learning_rate": 1.6860465116279073e-05,
      "loss": 0.7856,
      "step": 145
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9184642118227849,
      "learning_rate": 1.697674418604651e-05,
      "loss": 0.8565,
      "step": 146
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9829759948127988,
      "learning_rate": 1.7093023255813955e-05,
      "loss": 0.7965,
      "step": 147
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9845096516856808,
      "learning_rate": 1.7209302325581396e-05,
      "loss": 0.8274,
      "step": 148
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9646717199605087,
      "learning_rate": 1.7325581395348837e-05,
      "loss": 0.7782,
      "step": 149
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0009121781924153,
      "learning_rate": 1.744186046511628e-05,
      "loss": 0.748,
      "step": 150
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0269906040520693,
      "learning_rate": 1.7558139534883722e-05,
      "loss": 0.7815,
      "step": 151
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9517325046922676,
      "learning_rate": 1.7674418604651163e-05,
      "loss": 0.7967,
      "step": 152
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0839055345049087,
      "learning_rate": 1.7790697674418608e-05,
      "loss": 0.7454,
      "step": 153
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0098002446046428,
      "learning_rate": 1.790697674418605e-05,
      "loss": 0.7805,
      "step": 154
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9083954364222501,
      "learning_rate": 1.802325581395349e-05,
      "loss": 0.7764,
      "step": 155
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8646007137960834,
      "learning_rate": 1.813953488372093e-05,
      "loss": 0.7589,
      "step": 156
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0064014609093634,
      "learning_rate": 1.825581395348837e-05,
      "loss": 0.8105,
      "step": 157
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8870022502113245,
      "learning_rate": 1.8372093023255815e-05,
      "loss": 0.7466,
      "step": 158
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1312001191012668,
      "learning_rate": 1.8488372093023256e-05,
      "loss": 0.7894,
      "step": 159
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9461657639704655,
      "learning_rate": 1.86046511627907e-05,
      "loss": 0.7697,
      "step": 160
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0045162986154175,
      "learning_rate": 1.872093023255814e-05,
      "loss": 0.7955,
      "step": 161
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0554113428684353,
      "learning_rate": 1.8837209302325582e-05,
      "loss": 0.8149,
      "step": 162
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8521179240773811,
      "learning_rate": 1.8953488372093027e-05,
      "loss": 0.7948,
      "step": 163
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.134501033841664,
      "learning_rate": 1.9069767441860468e-05,
      "loss": 0.7128,
      "step": 164
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1595736891729245,
      "learning_rate": 1.918604651162791e-05,
      "loss": 0.742,
      "step": 165
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9255921128494717,
      "learning_rate": 1.9302325581395353e-05,
      "loss": 0.8051,
      "step": 166
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9628573872156717,
      "learning_rate": 1.941860465116279e-05,
      "loss": 0.7713,
      "step": 167
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8667585216690853,
      "learning_rate": 1.9534883720930235e-05,
      "loss": 0.7503,
      "step": 168
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8654572010488044,
      "learning_rate": 1.9651162790697676e-05,
      "loss": 0.7353,
      "step": 169
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.846386262484889,
      "learning_rate": 1.9767441860465116e-05,
      "loss": 0.7586,
      "step": 170
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.065216121506597,
      "learning_rate": 1.988372093023256e-05,
      "loss": 0.7871,
      "step": 171
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9186522312367869,
      "learning_rate": 2e-05,
      "loss": 0.7805,
      "step": 172
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.819316228728869,
      "learning_rate": 1.9999998386895803e-05,
      "loss": 0.7551,
      "step": 173
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9368514869213895,
      "learning_rate": 1.9999993547583733e-05,
      "loss": 0.7881,
      "step": 174
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9177713416924522,
      "learning_rate": 1.9999985482065353e-05,
      "loss": 0.7561,
      "step": 175
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1149571243095722,
      "learning_rate": 1.9999974190343262e-05,
      "loss": 0.8042,
      "step": 176
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0494497937988716,
      "learning_rate": 1.9999959672421103e-05,
      "loss": 0.8022,
      "step": 177
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0821242698974232,
      "learning_rate": 1.999994192830356e-05,
      "loss": 0.781,
      "step": 178
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0399749902397244,
      "learning_rate": 1.999992095799636e-05,
      "loss": 0.7213,
      "step": 179
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0674293781287443,
      "learning_rate": 1.9999896761506266e-05,
      "loss": 0.773,
      "step": 180
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1205184297543629,
      "learning_rate": 1.9999869338841083e-05,
      "loss": 0.8365,
      "step": 181
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.940462266214596,
      "learning_rate": 1.9999838690009663e-05,
      "loss": 0.8175,
      "step": 182
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9586178313591558,
      "learning_rate": 1.9999804815021894e-05,
      "loss": 0.8002,
      "step": 183
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1853140101841466,
      "learning_rate": 1.9999767713888696e-05,
      "loss": 0.8205,
      "step": 184
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8404858936923402,
      "learning_rate": 1.999972738662205e-05,
      "loss": 0.754,
      "step": 185
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0476356458171743,
      "learning_rate": 1.9999683833234958e-05,
      "loss": 0.8009,
      "step": 186
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8547048434533767,
      "learning_rate": 1.9999637053741474e-05,
      "loss": 0.7827,
      "step": 187
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9612261183509333,
      "learning_rate": 1.9999587048156695e-05,
      "loss": 0.7448,
      "step": 188
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8713776593262271,
      "learning_rate": 1.9999533816496743e-05,
      "loss": 0.8141,
      "step": 189
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.924141504031057,
      "learning_rate": 1.9999477358778804e-05,
      "loss": 0.772,
      "step": 190
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8409559421889676,
      "learning_rate": 1.999941767502108e-05,
      "loss": 0.7602,
      "step": 191
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9036075054131932,
      "learning_rate": 1.9999354765242837e-05,
      "loss": 0.8341,
      "step": 192
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8040838241173592,
      "learning_rate": 1.9999288629464363e-05,
      "loss": 0.7871,
      "step": 193
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0427603979545126,
      "learning_rate": 1.9999219267707e-05,
      "loss": 0.852,
      "step": 194
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9818504369921642,
      "learning_rate": 1.9999146679993126e-05,
      "loss": 0.7201,
      "step": 195
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8114142280251863,
      "learning_rate": 1.9999070866346153e-05,
      "loss": 0.826,
      "step": 196
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8514323155229955,
      "learning_rate": 1.9998991826790545e-05,
      "loss": 0.7885,
      "step": 197
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8395821368673114,
      "learning_rate": 1.99989095613518e-05,
      "loss": 0.8039,
      "step": 198
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8093999016507611,
      "learning_rate": 1.9998824070056467e-05,
      "loss": 0.731,
      "step": 199
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8715220714703587,
      "learning_rate": 1.999873535293211e-05,
      "loss": 0.7973,
      "step": 200
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.898257936460322,
      "learning_rate": 1.999864341000737e-05,
      "loss": 0.7264,
      "step": 201
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.340789629109145,
      "learning_rate": 1.9998548241311894e-05,
      "loss": 0.7481,
      "step": 202
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0215234515481244,
      "learning_rate": 1.9998449846876394e-05,
      "loss": 0.7929,
      "step": 203
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8627709078225212,
      "learning_rate": 1.999834822673261e-05,
      "loss": 0.7666,
      "step": 204
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0513377027280966,
      "learning_rate": 1.999824338091333e-05,
      "loss": 0.8362,
      "step": 205
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.078386385595274,
      "learning_rate": 1.999813530945238e-05,
      "loss": 0.7825,
      "step": 206
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9419822243365016,
      "learning_rate": 1.9998024012384624e-05,
      "loss": 0.8214,
      "step": 207
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9938271788307446,
      "learning_rate": 1.9997909489745968e-05,
      "loss": 0.7638,
      "step": 208
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2577641280915903,
      "learning_rate": 1.999779174157336e-05,
      "loss": 0.7775,
      "step": 209
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9003331273859478,
      "learning_rate": 1.999767076790479e-05,
      "loss": 0.8154,
      "step": 210
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0364535526225775,
      "learning_rate": 1.9997546568779285e-05,
      "loss": 0.7551,
      "step": 211
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0373226009046603,
      "learning_rate": 1.9997419144236915e-05,
      "loss": 0.7534,
      "step": 212
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.862550525263606,
      "learning_rate": 1.999728849431879e-05,
      "loss": 0.7338,
      "step": 213
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8328091621275889,
      "learning_rate": 1.9997154619067053e-05,
      "loss": 0.7857,
      "step": 214
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.820477444157098,
      "learning_rate": 1.9997017518524905e-05,
      "loss": 0.7635,
      "step": 215
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8631528968456693,
      "learning_rate": 1.999687719273658e-05,
      "loss": 0.8215,
      "step": 216
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8175923141427798,
      "learning_rate": 1.9996733641747338e-05,
      "loss": 0.7715,
      "step": 217
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8389290084531597,
      "learning_rate": 1.9996586865603497e-05,
      "loss": 0.7487,
      "step": 218
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8473347766678769,
      "learning_rate": 1.9996436864352412e-05,
      "loss": 0.7623,
      "step": 219
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9220854282369375,
      "learning_rate": 1.9996283638042476e-05,
      "loss": 0.7892,
      "step": 220
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8957531715904382,
      "learning_rate": 1.999612718672312e-05,
      "loss": 0.8036,
      "step": 221
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8336341338989092,
      "learning_rate": 1.999596751044482e-05,
      "loss": 0.7353,
      "step": 222
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8290740180784058,
      "learning_rate": 1.9995804609259097e-05,
      "loss": 0.7795,
      "step": 223
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5743955140257853,
      "learning_rate": 1.9995638483218496e-05,
      "loss": 0.7633,
      "step": 224
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.377001229281052,
      "learning_rate": 1.999546913237662e-05,
      "loss": 0.7561,
      "step": 225
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.015801156503745,
      "learning_rate": 1.9995296556788103e-05,
      "loss": 0.7541,
      "step": 226
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9245041698060197,
      "learning_rate": 1.9995120756508618e-05,
      "loss": 0.7959,
      "step": 227
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8585043020713611,
      "learning_rate": 1.9994941731594885e-05,
      "loss": 0.7599,
      "step": 228
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1417323217173783,
      "learning_rate": 1.9994759482104663e-05,
      "loss": 0.7527,
      "step": 229
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9263994592982827,
      "learning_rate": 1.999457400809675e-05,
      "loss": 0.7263,
      "step": 230
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0436046293322503,
      "learning_rate": 1.9994385309630978e-05,
      "loss": 0.83,
      "step": 231
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.213552856830767,
      "learning_rate": 1.9994193386768227e-05,
      "loss": 0.765,
      "step": 232
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8634205049079668,
      "learning_rate": 1.999399823957042e-05,
      "loss": 0.809,
      "step": 233
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0431394588261844,
      "learning_rate": 1.999379986810051e-05,
      "loss": 0.7803,
      "step": 234
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.895987795424016,
      "learning_rate": 1.9993598272422502e-05,
      "loss": 0.7998,
      "step": 235
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8286901900927107,
      "learning_rate": 1.9993393452601425e-05,
      "loss": 0.7865,
      "step": 236
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9209354194400868,
      "learning_rate": 1.9993185408703368e-05,
      "loss": 0.8013,
      "step": 237
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9667449027740166,
      "learning_rate": 1.9992974140795448e-05,
      "loss": 0.7627,
      "step": 238
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9176102878127571,
      "learning_rate": 1.999275964894582e-05,
      "loss": 0.7815,
      "step": 239
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.86787831216263,
      "learning_rate": 1.9992541933223687e-05,
      "loss": 0.7911,
      "step": 240
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7548241105185353,
      "learning_rate": 1.9992320993699287e-05,
      "loss": 0.773,
      "step": 241
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8006933134137093,
      "learning_rate": 1.9992096830443902e-05,
      "loss": 0.7242,
      "step": 242
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8437419207241251,
      "learning_rate": 1.9991869443529852e-05,
      "loss": 0.7756,
      "step": 243
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9351477453120055,
      "learning_rate": 1.9991638833030494e-05,
      "loss": 0.818,
      "step": 244
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9755931268438567,
      "learning_rate": 1.9991404999020227e-05,
      "loss": 0.7877,
      "step": 245
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8372108035891156,
      "learning_rate": 1.9991167941574495e-05,
      "loss": 0.7792,
      "step": 246
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9703093159535989,
      "learning_rate": 1.9990927660769778e-05,
      "loss": 0.7585,
      "step": 247
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9203540833502468,
      "learning_rate": 1.999068415668359e-05,
      "loss": 0.7765,
      "step": 248
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9518414010605435,
      "learning_rate": 1.999043742939449e-05,
      "loss": 0.7981,
      "step": 249
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8519326286020753,
      "learning_rate": 1.9990187478982087e-05,
      "loss": 0.815,
      "step": 250
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9523585409271742,
      "learning_rate": 1.9989934305527013e-05,
      "loss": 0.793,
      "step": 251
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8340686806473012,
      "learning_rate": 1.9989677909110945e-05,
      "loss": 0.7284,
      "step": 252
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.069071784104213,
      "learning_rate": 1.9989418289816607e-05,
      "loss": 0.7673,
      "step": 253
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8340618308721339,
      "learning_rate": 1.9989155447727754e-05,
      "loss": 0.7373,
      "step": 254
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9351766593879471,
      "learning_rate": 1.9988889382929187e-05,
      "loss": 0.7861,
      "step": 255
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8712153138269318,
      "learning_rate": 1.9988620095506744e-05,
      "loss": 0.7587,
      "step": 256
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0777813136466325,
      "learning_rate": 1.9988347585547302e-05,
      "loss": 0.7406,
      "step": 257
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7976190552531871,
      "learning_rate": 1.9988071853138774e-05,
      "loss": 0.7471,
      "step": 258
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0013488659065124,
      "learning_rate": 1.9987792898370123e-05,
      "loss": 0.743,
      "step": 259
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9095152315741585,
      "learning_rate": 1.9987510721331346e-05,
      "loss": 0.7157,
      "step": 260
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9231724552716354,
      "learning_rate": 1.9987225322113476e-05,
      "loss": 0.739,
      "step": 261
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0587665132783133,
      "learning_rate": 1.9986936700808588e-05,
      "loss": 0.7918,
      "step": 262
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8705718111607289,
      "learning_rate": 1.99866448575098e-05,
      "loss": 0.7969,
      "step": 263
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9949534490931345,
      "learning_rate": 1.9986349792311267e-05,
      "loss": 0.7578,
      "step": 264
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9429815289656218,
      "learning_rate": 1.9986051505308178e-05,
      "loss": 0.7559,
      "step": 265
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8877005800096698,
      "learning_rate": 1.998574999659677e-05,
      "loss": 0.7624,
      "step": 266
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8425293399970402,
      "learning_rate": 1.9985445266274324e-05,
      "loss": 0.81,
      "step": 267
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9086558182062692,
      "learning_rate": 1.998513731443914e-05,
      "loss": 0.7633,
      "step": 268
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8091245178099699,
      "learning_rate": 1.9984826141190573e-05,
      "loss": 0.7797,
      "step": 269
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0378748780190088,
      "learning_rate": 1.9984511746629014e-05,
      "loss": 0.8724,
      "step": 270
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8033986223947497,
      "learning_rate": 1.99841941308559e-05,
      "loss": 0.7452,
      "step": 271
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9722143219130681,
      "learning_rate": 1.998387329397369e-05,
      "loss": 0.7479,
      "step": 272
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8458199690853218,
      "learning_rate": 1.99835492360859e-05,
      "loss": 0.7774,
      "step": 273
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8658253596584664,
      "learning_rate": 1.9983221957297077e-05,
      "loss": 0.763,
      "step": 274
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0087246895887887,
      "learning_rate": 1.9982891457712803e-05,
      "loss": 0.8572,
      "step": 275
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1239509421315388,
      "learning_rate": 1.998255773743971e-05,
      "loss": 0.7457,
      "step": 276
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0995916315318452,
      "learning_rate": 1.9982220796585462e-05,
      "loss": 0.8132,
      "step": 277
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9207859696334993,
      "learning_rate": 1.998188063525876e-05,
      "loss": 0.8008,
      "step": 278
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8878327693625236,
      "learning_rate": 1.9981537253569348e-05,
      "loss": 0.7577,
      "step": 279
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7378026901793506,
      "learning_rate": 1.998119065162801e-05,
      "loss": 0.743,
      "step": 280
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9501418037176955,
      "learning_rate": 1.9980840829546565e-05,
      "loss": 0.7465,
      "step": 281
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7715683560714279,
      "learning_rate": 1.9980487787437878e-05,
      "loss": 0.7865,
      "step": 282
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8762521911207162,
      "learning_rate": 1.998013152541584e-05,
      "loss": 0.7979,
      "step": 283
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8218406621660014,
      "learning_rate": 1.9979772043595395e-05,
      "loss": 0.7664,
      "step": 284
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.046360493938903,
      "learning_rate": 1.9979409342092513e-05,
      "loss": 0.7815,
      "step": 285
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8276805510024934,
      "learning_rate": 1.9979043421024212e-05,
      "loss": 0.8151,
      "step": 286
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8767561772737308,
      "learning_rate": 1.997867428050855e-05,
      "loss": 0.7496,
      "step": 287
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7991384269575817,
      "learning_rate": 1.9978301920664614e-05,
      "loss": 0.7604,
      "step": 288
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9627758043286011,
      "learning_rate": 1.997792634161254e-05,
      "loss": 0.7212,
      "step": 289
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8313341870334467,
      "learning_rate": 1.997754754347349e-05,
      "loss": 0.8002,
      "step": 290
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9555627885827778,
      "learning_rate": 1.997716552636968e-05,
      "loss": 0.7351,
      "step": 291
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7830815896990123,
      "learning_rate": 1.9976780290424352e-05,
      "loss": 0.7359,
      "step": 292
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1000021983227037,
      "learning_rate": 1.997639183576179e-05,
      "loss": 0.7432,
      "step": 293
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8685840664699176,
      "learning_rate": 1.9976000162507326e-05,
      "loss": 0.8275,
      "step": 294
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9157713381323864,
      "learning_rate": 1.997560527078731e-05,
      "loss": 0.7604,
      "step": 295
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.814227580245308,
      "learning_rate": 1.9975207160729154e-05,
      "loss": 0.7383,
      "step": 296
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.989453163648394,
      "learning_rate": 1.9974805832461288e-05,
      "loss": 0.7784,
      "step": 297
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8448621584638188,
      "learning_rate": 1.9974401286113194e-05,
      "loss": 0.7214,
      "step": 298
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8785773248851321,
      "learning_rate": 1.997399352181538e-05,
      "loss": 0.7801,
      "step": 299
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8868001892666376,
      "learning_rate": 1.9973582539699404e-05,
      "loss": 0.8167,
      "step": 300
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8820995686540625,
      "learning_rate": 1.9973168339897865e-05,
      "loss": 0.7775,
      "step": 301
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.832669543573008,
      "learning_rate": 1.997275092254438e-05,
      "loss": 0.7378,
      "step": 302
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8448692859390643,
      "learning_rate": 1.997233028777362e-05,
      "loss": 0.7576,
      "step": 303
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0542039260107812,
      "learning_rate": 1.9971906435721294e-05,
      "loss": 0.7901,
      "step": 304
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9139397781566436,
      "learning_rate": 1.9971479366524146e-05,
      "loss": 0.8019,
      "step": 305
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7902959740515295,
      "learning_rate": 1.997104908031995e-05,
      "loss": 0.7454,
      "step": 306
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.769975450426412,
      "learning_rate": 1.9970615577247535e-05,
      "loss": 0.7231,
      "step": 307
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9101874234956655,
      "learning_rate": 1.997017885744675e-05,
      "loss": 0.7629,
      "step": 308
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7715395681360205,
      "learning_rate": 1.9969738921058494e-05,
      "loss": 0.771,
      "step": 309
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.844319543712736,
      "learning_rate": 1.9969295768224702e-05,
      "loss": 0.7845,
      "step": 310
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7697226656126701,
      "learning_rate": 1.9968849399088337e-05,
      "loss": 0.7527,
      "step": 311
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8167977871166724,
      "learning_rate": 1.9968399813793417e-05,
      "loss": 0.7744,
      "step": 312
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9896642855018024,
      "learning_rate": 1.9967947012484975e-05,
      "loss": 0.7833,
      "step": 313
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9971706294712228,
      "learning_rate": 1.9967490995309108e-05,
      "loss": 0.7036,
      "step": 314
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7856243798764908,
      "learning_rate": 1.9967031762412925e-05,
      "loss": 0.7907,
      "step": 315
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9312618524614489,
      "learning_rate": 1.9966569313944588e-05,
      "loss": 0.7229,
      "step": 316
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8984908625114154,
      "learning_rate": 1.9966103650053297e-05,
      "loss": 0.7629,
      "step": 317
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9710960606111945,
      "learning_rate": 1.9965634770889277e-05,
      "loss": 0.7671,
      "step": 318
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.807178389447196,
      "learning_rate": 1.9965162676603805e-05,
      "loss": 0.7614,
      "step": 319
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.784297509161817,
      "learning_rate": 1.9964687367349187e-05,
      "loss": 0.7393,
      "step": 320
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8672054216755936,
      "learning_rate": 1.9964208843278765e-05,
      "loss": 0.8031,
      "step": 321
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7130877418362815,
      "learning_rate": 1.9963727104546922e-05,
      "loss": 0.702,
      "step": 322
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7848897553341526,
      "learning_rate": 1.996324215130908e-05,
      "loss": 0.774,
      "step": 323
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9420426860081439,
      "learning_rate": 1.996275398372169e-05,
      "loss": 0.8005,
      "step": 324
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8109668808453452,
      "learning_rate": 1.9962262601942246e-05,
      "loss": 0.716,
      "step": 325
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8652841290608206,
      "learning_rate": 1.996176800612928e-05,
      "loss": 0.7623,
      "step": 326
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8647543828842572,
      "learning_rate": 1.9961270196442363e-05,
      "loss": 0.7724,
      "step": 327
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8225613364996616,
      "learning_rate": 1.9960769173042088e-05,
      "loss": 0.732,
      "step": 328
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8374863438475668,
      "learning_rate": 1.9960264936090104e-05,
      "loss": 0.7783,
      "step": 329
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8559114817150187,
      "learning_rate": 1.9959757485749086e-05,
      "loss": 0.7536,
      "step": 330
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.918035836669111,
      "learning_rate": 1.9959246822182745e-05,
      "loss": 0.778,
      "step": 331
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9962416832969171,
      "learning_rate": 1.995873294555584e-05,
      "loss": 0.824,
      "step": 332
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7921623544359869,
      "learning_rate": 1.9958215856034152e-05,
      "loss": 0.7722,
      "step": 333
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8715120967252615,
      "learning_rate": 1.9957695553784503e-05,
      "loss": 0.7305,
      "step": 334
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8694787913057882,
      "learning_rate": 1.9957172038974757e-05,
      "loss": 0.7193,
      "step": 335
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7872389514657487,
      "learning_rate": 1.995664531177381e-05,
      "loss": 0.7354,
      "step": 336
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9075911684539105,
      "learning_rate": 1.9956115372351595e-05,
      "loss": 0.8674,
      "step": 337
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8261141748179646,
      "learning_rate": 1.9955582220879088e-05,
      "loss": 0.7656,
      "step": 338
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8469148468776204,
      "learning_rate": 1.9955045857528283e-05,
      "loss": 0.7537,
      "step": 339
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8157021647950843,
      "learning_rate": 1.9954506282472223e-05,
      "loss": 0.7749,
      "step": 340
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7869720124218599,
      "learning_rate": 1.9953963495884995e-05,
      "loss": 0.7683,
      "step": 341
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8943010778175937,
      "learning_rate": 1.9953417497941712e-05,
      "loss": 0.7592,
      "step": 342
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8623201822368534,
      "learning_rate": 1.995286828881852e-05,
      "loss": 0.7512,
      "step": 343
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8574321345995202,
      "learning_rate": 1.9952315868692606e-05,
      "loss": 0.7974,
      "step": 344
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7564317619211998,
      "learning_rate": 1.995176023774219e-05,
      "loss": 0.7454,
      "step": 345
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8503093675017667,
      "learning_rate": 1.9951201396146534e-05,
      "loss": 0.7402,
      "step": 346
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.771837491947763,
      "learning_rate": 1.995063934408593e-05,
      "loss": 0.7534,
      "step": 347
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9595748943674599,
      "learning_rate": 1.995007408174171e-05,
      "loss": 0.8111,
      "step": 348
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.734000528350161,
      "learning_rate": 1.994950560929624e-05,
      "loss": 0.7163,
      "step": 349
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8740766781878783,
      "learning_rate": 1.9948933926932916e-05,
      "loss": 0.8094,
      "step": 350
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7622317864689694,
      "learning_rate": 1.994835903483618e-05,
      "loss": 0.716,
      "step": 351
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7347040067995632,
      "learning_rate": 1.99477809331915e-05,
      "loss": 0.7185,
      "step": 352
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8457297668942003,
      "learning_rate": 1.9947199622185387e-05,
      "loss": 0.751,
      "step": 353
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8556086269954896,
      "learning_rate": 1.9946615102005383e-05,
      "loss": 0.7432,
      "step": 354
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9005994494688468,
      "learning_rate": 1.9946027372840063e-05,
      "loss": 0.8161,
      "step": 355
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9491489451996729,
      "learning_rate": 1.9945436434879047e-05,
      "loss": 0.7599,
      "step": 356
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.861158770692281,
      "learning_rate": 1.9944842288312977e-05,
      "loss": 0.7116,
      "step": 357
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9272666090955257,
      "learning_rate": 1.9944244933333544e-05,
      "loss": 0.8222,
      "step": 358
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9305043800370174,
      "learning_rate": 1.9943644370133462e-05,
      "loss": 0.8527,
      "step": 359
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9521778042789389,
      "learning_rate": 1.9943040598906488e-05,
      "loss": 0.7681,
      "step": 360
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7384444440360638,
      "learning_rate": 1.9942433619847406e-05,
      "loss": 0.7523,
      "step": 361
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8583175064258568,
      "learning_rate": 1.994182343315205e-05,
      "loss": 0.7533,
      "step": 362
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7179991195152676,
      "learning_rate": 1.994121003901727e-05,
      "loss": 0.7446,
      "step": 363
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9557388532877195,
      "learning_rate": 1.994059343764096e-05,
      "loss": 0.8019,
      "step": 364
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8583829789184324,
      "learning_rate": 1.9939973629222055e-05,
      "loss": 0.7084,
      "step": 365
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8272373294060107,
      "learning_rate": 1.9939350613960516e-05,
      "loss": 0.7592,
      "step": 366
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9824414897937457,
      "learning_rate": 1.993872439205734e-05,
      "loss": 0.7335,
      "step": 367
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7938265329299767,
      "learning_rate": 1.9938094963714553e-05,
      "loss": 0.8011,
      "step": 368
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.923988488558115,
      "learning_rate": 1.9937462329135227e-05,
      "loss": 0.7264,
      "step": 369
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8656076587440547,
      "learning_rate": 1.9936826488523467e-05,
      "loss": 0.7495,
      "step": 370
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.826998137916326,
      "learning_rate": 1.9936187442084403e-05,
      "loss": 0.7137,
      "step": 371
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0083657975564335,
      "learning_rate": 1.9935545190024207e-05,
      "loss": 0.7462,
      "step": 372
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9325360962571568,
      "learning_rate": 1.9934899732550083e-05,
      "loss": 0.7322,
      "step": 373
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8982281423171012,
      "learning_rate": 1.9934251069870266e-05,
      "loss": 0.7942,
      "step": 374
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9146718590995555,
      "learning_rate": 1.9933599202194034e-05,
      "loss": 0.7078,
      "step": 375
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7938957997303373,
      "learning_rate": 1.9932944129731685e-05,
      "loss": 0.7456,
      "step": 376
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8880164388752135,
      "learning_rate": 1.9932285852694568e-05,
      "loss": 0.7589,
      "step": 377
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.810794679871374,
      "learning_rate": 1.993162437129505e-05,
      "loss": 0.7099,
      "step": 378
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8454310040042922,
      "learning_rate": 1.9930959685746538e-05,
      "loss": 0.7832,
      "step": 379
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8912986008610637,
      "learning_rate": 1.993029179626348e-05,
      "loss": 0.7418,
      "step": 380
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8470391014842785,
      "learning_rate": 1.992962070306135e-05,
      "loss": 0.7583,
      "step": 381
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0360726770067683,
      "learning_rate": 1.9928946406356647e-05,
      "loss": 0.7439,
      "step": 382
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0306337985978555,
      "learning_rate": 1.9928268906366927e-05,
      "loss": 0.8297,
      "step": 383
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9021026321439637,
      "learning_rate": 1.9927588203310756e-05,
      "loss": 0.7411,
      "step": 384
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8868620811200735,
      "learning_rate": 1.9926904297407747e-05,
      "loss": 0.7996,
      "step": 385
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8484605009705304,
      "learning_rate": 1.992621718887854e-05,
      "loss": 0.7376,
      "step": 386
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7850003860205665,
      "learning_rate": 1.992552687794481e-05,
      "loss": 0.7804,
      "step": 387
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8460827136194967,
      "learning_rate": 1.992483336482927e-05,
      "loss": 0.7548,
      "step": 388
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7644667952173284,
      "learning_rate": 1.992413664975566e-05,
      "loss": 0.7184,
      "step": 389
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8314675703242032,
      "learning_rate": 1.9923436732948754e-05,
      "loss": 0.7919,
      "step": 390
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7657328303259305,
      "learning_rate": 1.992273361463436e-05,
      "loss": 0.7722,
      "step": 391
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7578391599930594,
      "learning_rate": 1.9922027295039313e-05,
      "loss": 0.7517,
      "step": 392
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8284505662895363,
      "learning_rate": 1.9921317774391498e-05,
      "loss": 0.7432,
      "step": 393
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7731279063887819,
      "learning_rate": 1.9920605052919814e-05,
      "loss": 0.7387,
      "step": 394
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7714109616958139,
      "learning_rate": 1.99198891308542e-05,
      "loss": 0.7537,
      "step": 395
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0090158026406266,
      "learning_rate": 1.9919170008425625e-05,
      "loss": 0.813,
      "step": 396
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7322779955161317,
      "learning_rate": 1.99184476858661e-05,
      "loss": 0.7172,
      "step": 397
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.976781307941285,
      "learning_rate": 1.991772216340866e-05,
      "loss": 0.7537,
      "step": 398
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8289065081846941,
      "learning_rate": 1.9916993441287364e-05,
      "loss": 0.7631,
      "step": 399
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8998868961840657,
      "learning_rate": 1.9916261519737327e-05,
      "loss": 0.7358,
      "step": 400
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8459547761018223,
      "learning_rate": 1.991552639899467e-05,
      "loss": 0.7603,
      "step": 401
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7562186861419029,
      "learning_rate": 1.9914788079296567e-05,
      "loss": 0.7565,
      "step": 402
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7623008085932237,
      "learning_rate": 1.9914046560881205e-05,
      "loss": 0.7548,
      "step": 403
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9138181335230982,
      "learning_rate": 1.9913301843987827e-05,
      "loss": 0.7612,
      "step": 404
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.864228632110966,
      "learning_rate": 1.9912553928856684e-05,
      "loss": 0.7695,
      "step": 405
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8783501356809255,
      "learning_rate": 1.9911802815729073e-05,
      "loss": 0.7335,
      "step": 406
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7917636386168562,
      "learning_rate": 1.9911048504847318e-05,
      "loss": 0.727,
      "step": 407
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8683136780698806,
      "learning_rate": 1.9910290996454774e-05,
      "loss": 0.7228,
      "step": 408
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7900846602192605,
      "learning_rate": 1.990953029079583e-05,
      "loss": 0.7645,
      "step": 409
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8795520473014077,
      "learning_rate": 1.990876638811591e-05,
      "loss": 0.769,
      "step": 410
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8052553599439309,
      "learning_rate": 1.990799928866146e-05,
      "loss": 0.7672,
      "step": 411
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8943023523091413,
      "learning_rate": 1.990722899267996e-05,
      "loss": 0.7731,
      "step": 412
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8210085958183659,
      "learning_rate": 1.9906455500419927e-05,
      "loss": 0.7742,
      "step": 413
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9530034559960427,
      "learning_rate": 1.9905678812130905e-05,
      "loss": 0.8918,
      "step": 414
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8809080640391547,
      "learning_rate": 1.990489892806347e-05,
      "loss": 0.7658,
      "step": 415
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9537989601075222,
      "learning_rate": 1.9904115848469233e-05,
      "loss": 0.7189,
      "step": 416
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8187661717570067,
      "learning_rate": 1.9903329573600824e-05,
      "loss": 0.7968,
      "step": 417
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8168613508439478,
      "learning_rate": 1.9902540103711918e-05,
      "loss": 0.7649,
      "step": 418
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.636125536859928,
      "learning_rate": 1.990174743905721e-05,
      "loss": 0.8074,
      "step": 419
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8905003794291997,
      "learning_rate": 1.990095157989243e-05,
      "loss": 0.7944,
      "step": 420
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9369478551297701,
      "learning_rate": 1.9900152526474343e-05,
      "loss": 0.7799,
      "step": 421
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7771516898357167,
      "learning_rate": 1.989935027906074e-05,
      "loss": 0.7505,
      "step": 422
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8091789992360671,
      "learning_rate": 1.9898544837910436e-05,
      "loss": 0.7241,
      "step": 423
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7261528562651761,
      "learning_rate": 1.9897736203283292e-05,
      "loss": 0.6922,
      "step": 424
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8741728311777652,
      "learning_rate": 1.9896924375440186e-05,
      "loss": 0.7745,
      "step": 425
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.841021792599743,
      "learning_rate": 1.989610935464303e-05,
      "loss": 0.777,
      "step": 426
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7926417425519007,
      "learning_rate": 1.9895291141154765e-05,
      "loss": 0.7671,
      "step": 427
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8311163606898547,
      "learning_rate": 1.9894469735239366e-05,
      "loss": 0.7696,
      "step": 428
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7673111350715692,
      "learning_rate": 1.9893645137161843e-05,
      "loss": 0.7743,
      "step": 429
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7687679476412361,
      "learning_rate": 1.9892817347188217e-05,
      "loss": 0.7597,
      "step": 430
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7739636805899913,
      "learning_rate": 1.9891986365585553e-05,
      "loss": 0.7223,
      "step": 431
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7605372233607582,
      "learning_rate": 1.989115219262195e-05,
      "loss": 0.7558,
      "step": 432
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8369358213813649,
      "learning_rate": 1.9890314828566518e-05,
      "loss": 0.7651,
      "step": 433
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9763288847852094,
      "learning_rate": 1.9889474273689415e-05,
      "loss": 0.8098,
      "step": 434
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8494432800083765,
      "learning_rate": 1.988863052826182e-05,
      "loss": 0.7765,
      "step": 435
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.822785302887837,
      "learning_rate": 1.9887783592555952e-05,
      "loss": 0.7559,
      "step": 436
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0736366843271696,
      "learning_rate": 1.9886933466845036e-05,
      "loss": 0.7799,
      "step": 437
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8080504153989466,
      "learning_rate": 1.9886080151403346e-05,
      "loss": 0.7429,
      "step": 438
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.876086703306627,
      "learning_rate": 1.988522364650618e-05,
      "loss": 0.777,
      "step": 439
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8272949669824579,
      "learning_rate": 1.9884363952429864e-05,
      "loss": 0.7747,
      "step": 440
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9823591733255342,
      "learning_rate": 1.988350106945175e-05,
      "loss": 0.7587,
      "step": 441
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8127856459024151,
      "learning_rate": 1.988263499785023e-05,
      "loss": 0.7373,
      "step": 442
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8017213817615765,
      "learning_rate": 1.988176573790471e-05,
      "loss": 0.7076,
      "step": 443
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7426799116288191,
      "learning_rate": 1.9880893289895633e-05,
      "loss": 0.734,
      "step": 444
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8452937731367595,
      "learning_rate": 1.988001765410447e-05,
      "loss": 0.7426,
      "step": 445
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8397641582221184,
      "learning_rate": 1.9879138830813714e-05,
      "loss": 0.7522,
      "step": 446
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7600164271271203,
      "learning_rate": 1.98782568203069e-05,
      "loss": 0.7436,
      "step": 447
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8185411315627878,
      "learning_rate": 1.9877371622868583e-05,
      "loss": 0.758,
      "step": 448
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0284687280247709,
      "learning_rate": 1.9876483238784336e-05,
      "loss": 0.7569,
      "step": 449
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9164245825664988,
      "learning_rate": 1.9875591668340778e-05,
      "loss": 0.7656,
      "step": 450
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8324203564469429,
      "learning_rate": 1.9874696911825547e-05,
      "loss": 0.7961,
      "step": 451
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7818976206391318,
      "learning_rate": 1.9873798969527312e-05,
      "loss": 0.7293,
      "step": 452
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.981812234349996,
      "learning_rate": 1.9872897841735765e-05,
      "loss": 0.799,
      "step": 453
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8992813985209943,
      "learning_rate": 1.9871993528741628e-05,
      "loss": 0.7848,
      "step": 454
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8367612847979895,
      "learning_rate": 1.9871086030836653e-05,
      "loss": 0.7858,
      "step": 455
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8865629152536015,
      "learning_rate": 1.9870175348313616e-05,
      "loss": 0.768,
      "step": 456
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7426504573237722,
      "learning_rate": 1.9869261481466324e-05,
      "loss": 0.7386,
      "step": 457
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8759786129633722,
      "learning_rate": 1.986834443058961e-05,
      "loss": 0.7355,
      "step": 458
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8660193152214409,
      "learning_rate": 1.9867424195979332e-05,
      "loss": 0.8011,
      "step": 459
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9071492120681568,
      "learning_rate": 1.986650077793238e-05,
      "loss": 0.8044,
      "step": 460
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7967859426296018,
      "learning_rate": 1.986557417674666e-05,
      "loss": 0.7292,
      "step": 461
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7457601622086821,
      "learning_rate": 1.986464439272112e-05,
      "loss": 0.6852,
      "step": 462
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9414665968555452,
      "learning_rate": 1.986371142615573e-05,
      "loss": 0.7992,
      "step": 463
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7306111133938757,
      "learning_rate": 1.9862775277351476e-05,
      "loss": 0.7596,
      "step": 464
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8551152737599159,
      "learning_rate": 1.9861835946610384e-05,
      "loss": 0.7015,
      "step": 465
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8092924797973844,
      "learning_rate": 1.9860893434235505e-05,
      "loss": 0.8068,
      "step": 466
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8698022481986285,
      "learning_rate": 1.9859947740530905e-05,
      "loss": 0.808,
      "step": 467
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9749619014880784,
      "learning_rate": 1.9858998865801688e-05,
      "loss": 0.7755,
      "step": 468
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.11134160481534,
      "learning_rate": 1.9858046810353986e-05,
      "loss": 0.8088,
      "step": 469
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7682795774878506,
      "learning_rate": 1.9857091574494945e-05,
      "loss": 0.7222,
      "step": 470
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8675966306309555,
      "learning_rate": 1.9856133158532748e-05,
      "loss": 0.7609,
      "step": 471
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7387688872329765,
      "learning_rate": 1.9855171562776595e-05,
      "loss": 0.7208,
      "step": 472
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9017004746211407,
      "learning_rate": 1.9854206787536726e-05,
      "loss": 0.7713,
      "step": 473
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8158422135046964,
      "learning_rate": 1.9853238833124387e-05,
      "loss": 0.7806,
      "step": 474
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.77504415793435,
      "learning_rate": 1.9852267699851865e-05,
      "loss": 0.7439,
      "step": 475
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8016697296863999,
      "learning_rate": 1.985129338803247e-05,
      "loss": 0.7428,
      "step": 476
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8294942204513672,
      "learning_rate": 1.9850315897980535e-05,
      "loss": 0.7705,
      "step": 477
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9505801828633351,
      "learning_rate": 1.9849335230011416e-05,
      "loss": 0.7746,
      "step": 478
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7615045565813228,
      "learning_rate": 1.9848351384441496e-05,
      "loss": 0.7325,
      "step": 479
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7511009681517573,
      "learning_rate": 1.9847364361588184e-05,
      "loss": 0.7464,
      "step": 480
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7719958102551266,
      "learning_rate": 1.984637416176992e-05,
      "loss": 0.7817,
      "step": 481
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8672223716889659,
      "learning_rate": 1.9845380785306157e-05,
      "loss": 0.7744,
      "step": 482
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.876084822004357,
      "learning_rate": 1.9844384232517383e-05,
      "loss": 0.7196,
      "step": 483
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7897094965928901,
      "learning_rate": 1.9843384503725103e-05,
      "loss": 0.7537,
      "step": 484
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.007353347021726,
      "learning_rate": 1.9842381599251856e-05,
      "loss": 0.8155,
      "step": 485
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8551700035708317,
      "learning_rate": 1.9841375519421194e-05,
      "loss": 0.7715,
      "step": 486
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.858289473444655,
      "learning_rate": 1.9840366264557698e-05,
      "loss": 0.7405,
      "step": 487
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7180244214938318,
      "learning_rate": 1.983935383498698e-05,
      "loss": 0.6964,
      "step": 488
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7872777840043885,
      "learning_rate": 1.983833823103567e-05,
      "loss": 0.7447,
      "step": 489
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2969134807391185,
      "learning_rate": 1.9837319453031424e-05,
      "loss": 0.817,
      "step": 490
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7351416613557185,
      "learning_rate": 1.9836297501302915e-05,
      "loss": 0.7454,
      "step": 491
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.748920966824341,
      "learning_rate": 1.983527237617985e-05,
      "loss": 0.7598,
      "step": 492
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9328531178908089,
      "learning_rate": 1.983424407799296e-05,
      "loss": 0.7741,
      "step": 493
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9215025507437631,
      "learning_rate": 1.9833212607073986e-05,
      "loss": 0.766,
      "step": 494
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8769164776130071,
      "learning_rate": 1.9832177963755707e-05,
      "loss": 0.7573,
      "step": 495
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8463112346786238,
      "learning_rate": 1.9831140148371926e-05,
      "loss": 0.7854,
      "step": 496
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6919258968671547,
      "learning_rate": 1.9830099161257454e-05,
      "loss": 0.7302,
      "step": 497
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2197117410460798,
      "learning_rate": 1.9829055002748137e-05,
      "loss": 0.7621,
      "step": 498
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7543575518054616,
      "learning_rate": 1.9828007673180847e-05,
      "loss": 0.7412,
      "step": 499
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.798829068522451,
      "learning_rate": 1.9826957172893476e-05,
      "loss": 0.7331,
      "step": 500
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7096492144982032,
      "learning_rate": 1.982590350222493e-05,
      "loss": 0.7748,
      "step": 501
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8134383572789833,
      "learning_rate": 1.982484666151515e-05,
      "loss": 0.7632,
      "step": 502
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8899575060756721,
      "learning_rate": 1.9823786651105092e-05,
      "loss": 0.7386,
      "step": 503
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9127782595419466,
      "learning_rate": 1.982272347133674e-05,
      "loss": 0.7254,
      "step": 504
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7934277100899534,
      "learning_rate": 1.9821657122553097e-05,
      "loss": 0.7818,
      "step": 505
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.808164797893132,
      "learning_rate": 1.982058760509819e-05,
      "loss": 0.7683,
      "step": 506
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7438308989496081,
      "learning_rate": 1.9819514919317066e-05,
      "loss": 0.7872,
      "step": 507
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7304858374019926,
      "learning_rate": 1.9818439065555793e-05,
      "loss": 0.7384,
      "step": 508
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6799981406023019,
      "learning_rate": 1.981736004416147e-05,
      "loss": 0.7152,
      "step": 509
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7389603698567997,
      "learning_rate": 1.9816277855482212e-05,
      "loss": 0.7058,
      "step": 510
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8802999081463164,
      "learning_rate": 1.9815192499867147e-05,
      "loss": 0.768,
      "step": 511
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9615673058983571,
      "learning_rate": 1.9814103977666443e-05,
      "loss": 0.7832,
      "step": 512
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8060413149655544,
      "learning_rate": 1.9813012289231276e-05,
      "loss": 0.7504,
      "step": 513
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.834870347708771,
      "learning_rate": 1.9811917434913846e-05,
      "loss": 0.7684,
      "step": 514
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7635428155047254,
      "learning_rate": 1.981081941506738e-05,
      "loss": 0.7821,
      "step": 515
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7296487907535214,
      "learning_rate": 1.9809718230046116e-05,
      "loss": 0.7558,
      "step": 516
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.730311596329511,
      "learning_rate": 1.9808613880205327e-05,
      "loss": 0.7576,
      "step": 517
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7068888939033778,
      "learning_rate": 1.980750636590129e-05,
      "loss": 0.7512,
      "step": 518
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9266662842456305,
      "learning_rate": 1.980639568749132e-05,
      "loss": 0.7906,
      "step": 519
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8243061561499516,
      "learning_rate": 1.980528184533374e-05,
      "loss": 0.7636,
      "step": 520
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9154261911917633,
      "learning_rate": 1.9804164839787903e-05,
      "loss": 0.7405,
      "step": 521
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.895555976839802,
      "learning_rate": 1.9803044671214176e-05,
      "loss": 0.7686,
      "step": 522
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8138936857909488,
      "learning_rate": 1.9801921339973948e-05,
      "loss": 0.7744,
      "step": 523
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7553442575492129,
      "learning_rate": 1.9800794846429628e-05,
      "loss": 0.7523,
      "step": 524
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8503077479276419,
      "learning_rate": 1.9799665190944652e-05,
      "loss": 0.7511,
      "step": 525
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7661257217724603,
      "learning_rate": 1.9798532373883465e-05,
      "loss": 0.7131,
      "step": 526
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7548176801634663,
      "learning_rate": 1.979739639561154e-05,
      "loss": 0.7635,
      "step": 527
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7674808531883642,
      "learning_rate": 1.9796257256495364e-05,
      "loss": 0.7521,
      "step": 528
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6984249999527656,
      "learning_rate": 1.979511495690245e-05,
      "loss": 0.7319,
      "step": 529
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7249978207077354,
      "learning_rate": 1.9793969497201326e-05,
      "loss": 0.7049,
      "step": 530
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7573373837073646,
      "learning_rate": 1.9792820877761543e-05,
      "loss": 0.7192,
      "step": 531
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8512096684940289,
      "learning_rate": 1.979166909895367e-05,
      "loss": 0.7772,
      "step": 532
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8324671333554267,
      "learning_rate": 1.979051416114929e-05,
      "loss": 0.7989,
      "step": 533
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.752902450756574,
      "learning_rate": 1.978935606472102e-05,
      "loss": 0.7667,
      "step": 534
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6731486246281305,
      "learning_rate": 1.9788194810042475e-05,
      "loss": 0.7119,
      "step": 535
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8012960721883187,
      "learning_rate": 1.9787030397488304e-05,
      "loss": 0.7982,
      "step": 536
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7651528887298984,
      "learning_rate": 1.978586282743417e-05,
      "loss": 0.7437,
      "step": 537
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7568889598773988,
      "learning_rate": 1.978469210025676e-05,
      "loss": 0.7144,
      "step": 538
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7595812835377417,
      "learning_rate": 1.978351821633377e-05,
      "loss": 0.7699,
      "step": 539
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7343889364111497,
      "learning_rate": 1.9782341176043923e-05,
      "loss": 0.7532,
      "step": 540
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8440756234466729,
      "learning_rate": 1.978116097976695e-05,
      "loss": 0.7404,
      "step": 541
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8447806000017273,
      "learning_rate": 1.9779977627883617e-05,
      "loss": 0.7887,
      "step": 542
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8206617830071264,
      "learning_rate": 1.977879112077569e-05,
      "loss": 0.7449,
      "step": 543
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7832428751682677,
      "learning_rate": 1.9777601458825965e-05,
      "loss": 0.7348,
      "step": 544
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8250984571494574,
      "learning_rate": 1.977640864241825e-05,
      "loss": 0.7297,
      "step": 545
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.848589405714994,
      "learning_rate": 1.977521267193737e-05,
      "loss": 0.7857,
      "step": 546
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8540680848380097,
      "learning_rate": 1.9774013547769176e-05,
      "loss": 0.7579,
      "step": 547
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7537770840647847,
      "learning_rate": 1.9772811270300524e-05,
      "loss": 0.7605,
      "step": 548
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9481246855098646,
      "learning_rate": 1.97716058399193e-05,
      "loss": 0.7842,
      "step": 549
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7055489329336314,
      "learning_rate": 1.97703972570144e-05,
      "loss": 0.7562,
      "step": 550
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.278727040807141,
      "learning_rate": 1.9769185521975732e-05,
      "loss": 0.7407,
      "step": 551
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8975871699480714,
      "learning_rate": 1.976797063519423e-05,
      "loss": 0.7953,
      "step": 552
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8759273543389297,
      "learning_rate": 1.9766752597061846e-05,
      "loss": 0.704,
      "step": 553
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8627404756637789,
      "learning_rate": 1.9765531407971543e-05,
      "loss": 0.7834,
      "step": 554
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.099407660421469,
      "learning_rate": 1.9764307068317296e-05,
      "loss": 0.828,
      "step": 555
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8775521752411964,
      "learning_rate": 1.9763079578494107e-05,
      "loss": 0.7792,
      "step": 556
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.994142090843372,
      "learning_rate": 1.9761848938897993e-05,
      "loss": 0.7312,
      "step": 557
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6975387788418437,
      "learning_rate": 1.9760615149925976e-05,
      "loss": 0.7712,
      "step": 558
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.826134829728442,
      "learning_rate": 1.975937821197611e-05,
      "loss": 0.7346,
      "step": 559
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7875180667482798,
      "learning_rate": 1.975813812544745e-05,
      "loss": 0.7686,
      "step": 560
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7354877703931323,
      "learning_rate": 1.9756894890740082e-05,
      "loss": 0.7973,
      "step": 561
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8293674848055362,
      "learning_rate": 1.9755648508255092e-05,
      "loss": 0.7446,
      "step": 562
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7684531184354124,
      "learning_rate": 1.975439897839459e-05,
      "loss": 0.7609,
      "step": 563
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8954610427291553,
      "learning_rate": 1.9753146301561705e-05,
      "loss": 0.7614,
      "step": 564
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8010195021422107,
      "learning_rate": 1.9751890478160574e-05,
      "loss": 0.738,
      "step": 565
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7264467803660443,
      "learning_rate": 1.975063150859635e-05,
      "loss": 0.7107,
      "step": 566
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8829667202655547,
      "learning_rate": 1.9749369393275205e-05,
      "loss": 0.7554,
      "step": 567
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8035079738073652,
      "learning_rate": 1.974810413260432e-05,
      "loss": 0.776,
      "step": 568
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7253334673464773,
      "learning_rate": 1.9746835726991903e-05,
      "loss": 0.7558,
      "step": 569
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8081973302113362,
      "learning_rate": 1.974556417684716e-05,
      "loss": 0.7897,
      "step": 570
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9210247034691154,
      "learning_rate": 1.974428948258032e-05,
      "loss": 0.8322,
      "step": 571
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7684248969009789,
      "learning_rate": 1.974301164460263e-05,
      "loss": 0.7743,
      "step": 572
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7970642772088219,
      "learning_rate": 1.9741730663326348e-05,
      "loss": 0.755,
      "step": 573
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7464238330575304,
      "learning_rate": 1.9740446539164735e-05,
      "loss": 0.7544,
      "step": 574
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7143368781786331,
      "learning_rate": 1.9739159272532092e-05,
      "loss": 0.7544,
      "step": 575
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.747353967939556,
      "learning_rate": 1.9737868863843704e-05,
      "loss": 0.7223,
      "step": 576
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9214918530383034,
      "learning_rate": 1.9736575313515894e-05,
      "loss": 0.7747,
      "step": 577
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7107374738748354,
      "learning_rate": 1.973527862196598e-05,
      "loss": 0.7374,
      "step": 578
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8011748004640891,
      "learning_rate": 1.973397878961231e-05,
      "loss": 0.7286,
      "step": 579
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8285663050307236,
      "learning_rate": 1.9732675816874227e-05,
      "loss": 0.7548,
      "step": 580
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7828747912904963,
      "learning_rate": 1.9731369704172107e-05,
      "loss": 0.719,
      "step": 581
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7483416817493264,
      "learning_rate": 1.973006045192732e-05,
      "loss": 0.7485,
      "step": 582
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7312356993669044,
      "learning_rate": 1.9728748060562267e-05,
      "loss": 0.7283,
      "step": 583
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7933031514826875,
      "learning_rate": 1.9727432530500348e-05,
      "loss": 0.7504,
      "step": 584
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7615497135585345,
      "learning_rate": 1.972611386216598e-05,
      "loss": 0.7187,
      "step": 585
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8639071240418686,
      "learning_rate": 1.9724792055984595e-05,
      "loss": 0.7777,
      "step": 586
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8002198283210307,
      "learning_rate": 1.9723467112382636e-05,
      "loss": 0.7645,
      "step": 587
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8362466292171453,
      "learning_rate": 1.9722139031787553e-05,
      "loss": 0.7227,
      "step": 588
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7712010037315337,
      "learning_rate": 1.9720807814627813e-05,
      "loss": 0.7359,
      "step": 589
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8938564116000176,
      "learning_rate": 1.97194734613329e-05,
      "loss": 0.8188,
      "step": 590
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7237717415515224,
      "learning_rate": 1.97181359723333e-05,
      "loss": 0.7559,
      "step": 591
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7494573536745601,
      "learning_rate": 1.9716795348060516e-05,
      "loss": 0.7477,
      "step": 592
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8552300762061101,
      "learning_rate": 1.971545158894706e-05,
      "loss": 0.7683,
      "step": 593
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7120110971434459,
      "learning_rate": 1.971410469542646e-05,
      "loss": 0.7302,
      "step": 594
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8043073429320737,
      "learning_rate": 1.9712754667933244e-05,
      "loss": 0.763,
      "step": 595
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7470812265707065,
      "learning_rate": 1.971140150690297e-05,
      "loss": 0.7326,
      "step": 596
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8086367737940354,
      "learning_rate": 1.9710045212772185e-05,
      "loss": 0.7478,
      "step": 597
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8718288877423573,
      "learning_rate": 1.970868578597847e-05,
      "loss": 0.7927,
      "step": 598
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7201783156529487,
      "learning_rate": 1.9707323226960394e-05,
      "loss": 0.7642,
      "step": 599
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8453456694905983,
      "learning_rate": 1.9705957536157547e-05,
      "loss": 0.7426,
      "step": 600
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6504997895248144,
      "learning_rate": 1.9704588714010538e-05,
      "loss": 0.7183,
      "step": 601
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8442584615492408,
      "learning_rate": 1.9703216760960968e-05,
      "loss": 0.7812,
      "step": 602
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1478460843256624,
      "learning_rate": 1.9701841677451466e-05,
      "loss": 0.8095,
      "step": 603
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7509800902611968,
      "learning_rate": 1.970046346392566e-05,
      "loss": 0.7817,
      "step": 604
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8294249423089739,
      "learning_rate": 1.9699082120828183e-05,
      "loss": 0.7741,
      "step": 605
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8580915826763138,
      "learning_rate": 1.9697697648604693e-05,
      "loss": 0.8049,
      "step": 606
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8700373624399388,
      "learning_rate": 1.969631004770185e-05,
      "loss": 0.7656,
      "step": 607
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7331766941105411,
      "learning_rate": 1.969491931856732e-05,
      "loss": 0.7097,
      "step": 608
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7606472505422243,
      "learning_rate": 1.969352546164978e-05,
      "loss": 0.7751,
      "step": 609
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9157016529821895,
      "learning_rate": 1.969212847739892e-05,
      "loss": 0.7825,
      "step": 610
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7427043187520821,
      "learning_rate": 1.9690728366265436e-05,
      "loss": 0.7383,
      "step": 611
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8440968548634247,
      "learning_rate": 1.9689325128701032e-05,
      "loss": 0.7463,
      "step": 612
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0512847090819606,
      "learning_rate": 1.968791876515842e-05,
      "loss": 0.7456,
      "step": 613
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9279768250432433,
      "learning_rate": 1.968650927609133e-05,
      "loss": 0.8418,
      "step": 614
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9232500044599615,
      "learning_rate": 1.9685096661954484e-05,
      "loss": 0.7551,
      "step": 615
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8313123568887517,
      "learning_rate": 1.968368092320362e-05,
      "loss": 0.7337,
      "step": 616
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7947495306550826,
      "learning_rate": 1.968226206029549e-05,
      "loss": 0.768,
      "step": 617
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8112831330039124,
      "learning_rate": 1.968084007368785e-05,
      "loss": 0.7343,
      "step": 618
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9562244824745674,
      "learning_rate": 1.967941496383946e-05,
      "loss": 0.8379,
      "step": 619
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7685555770256285,
      "learning_rate": 1.9677986731210084e-05,
      "loss": 0.7205,
      "step": 620
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7835450000812526,
      "learning_rate": 1.967655537626051e-05,
      "loss": 0.707,
      "step": 621
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6953327701313811,
      "learning_rate": 1.967512089945252e-05,
      "loss": 0.7384,
      "step": 622
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7287826917576935,
      "learning_rate": 1.9673683301248902e-05,
      "loss": 0.7547,
      "step": 623
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7057142511695544,
      "learning_rate": 1.9672242582113455e-05,
      "loss": 0.7418,
      "step": 624
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8040691085311827,
      "learning_rate": 1.9670798742510992e-05,
      "loss": 0.7754,
      "step": 625
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7674799730270281,
      "learning_rate": 1.966935178290732e-05,
      "loss": 0.7417,
      "step": 626
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8283420820415435,
      "learning_rate": 1.9667901703769255e-05,
      "loss": 0.773,
      "step": 627
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6931479143804953,
      "learning_rate": 1.966644850556463e-05,
      "loss": 0.7683,
      "step": 628
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7353388102914924,
      "learning_rate": 1.966499218876228e-05,
      "loss": 0.7203,
      "step": 629
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8451437998099038,
      "learning_rate": 1.966353275383203e-05,
      "loss": 0.7857,
      "step": 630
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7369261343897988,
      "learning_rate": 1.9662070201244736e-05,
      "loss": 0.7248,
      "step": 631
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7920298731653357,
      "learning_rate": 1.9660604531472246e-05,
      "loss": 0.7072,
      "step": 632
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9252563348725994,
      "learning_rate": 1.9659135744987407e-05,
      "loss": 0.7841,
      "step": 633
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7654436430622215,
      "learning_rate": 1.965766384226409e-05,
      "loss": 0.7244,
      "step": 634
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8820748665482847,
      "learning_rate": 1.9656188823777157e-05,
      "loss": 0.7678,
      "step": 635
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8368030923895007,
      "learning_rate": 1.9654710690002482e-05,
      "loss": 0.7588,
      "step": 636
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0590018104074201,
      "learning_rate": 1.965322944141694e-05,
      "loss": 0.7939,
      "step": 637
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7592766902851685,
      "learning_rate": 1.9651745078498412e-05,
      "loss": 0.7781,
      "step": 638
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.895929559055156,
      "learning_rate": 1.9650257601725785e-05,
      "loss": 0.7506,
      "step": 639
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7295566450768064,
      "learning_rate": 1.9648767011578948e-05,
      "loss": 0.7844,
      "step": 640
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.807988857781752,
      "learning_rate": 1.9647273308538802e-05,
      "loss": 0.75,
      "step": 641
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8490997395185098,
      "learning_rate": 1.9645776493087245e-05,
      "loss": 0.7306,
      "step": 642
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8397730910026341,
      "learning_rate": 1.9644276565707178e-05,
      "loss": 0.7558,
      "step": 643
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8939470484990434,
      "learning_rate": 1.9642773526882507e-05,
      "loss": 0.8794,
      "step": 644
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7261418426578365,
      "learning_rate": 1.9641267377098154e-05,
      "loss": 0.7551,
      "step": 645
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6995202321887527,
      "learning_rate": 1.9639758116840023e-05,
      "loss": 0.7722,
      "step": 646
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6989431530364268,
      "learning_rate": 1.9638245746595034e-05,
      "loss": 0.7138,
      "step": 647
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8154473439498702,
      "learning_rate": 1.9636730266851116e-05,
      "loss": 0.7433,
      "step": 648
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8657530144502702,
      "learning_rate": 1.963521167809719e-05,
      "loss": 0.8163,
      "step": 649
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8551800237869719,
      "learning_rate": 1.963368998082319e-05,
      "loss": 0.7954,
      "step": 650
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6814325641177337,
      "learning_rate": 1.9632165175520032e-05,
      "loss": 0.7582,
      "step": 651
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7356024059731523,
      "learning_rate": 1.9630637262679668e-05,
      "loss": 0.7561,
      "step": 652
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8023383413152883,
      "learning_rate": 1.9629106242795023e-05,
      "loss": 0.7432,
      "step": 653
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7554885855796972,
      "learning_rate": 1.9627572116360038e-05,
      "loss": 0.7736,
      "step": 654
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8638127320968847,
      "learning_rate": 1.962603488386966e-05,
      "loss": 0.7621,
      "step": 655
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7773569396694973,
      "learning_rate": 1.9624494545819827e-05,
      "loss": 0.7447,
      "step": 656
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7847415265722286,
      "learning_rate": 1.9622951102707482e-05,
      "loss": 0.7831,
      "step": 657
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7698427549120112,
      "learning_rate": 1.962140455503058e-05,
      "loss": 0.7451,
      "step": 658
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7467606009243652,
      "learning_rate": 1.961985490328806e-05,
      "loss": 0.7395,
      "step": 659
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7300580463927786,
      "learning_rate": 1.961830214797988e-05,
      "loss": 0.7243,
      "step": 660
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7384858315371257,
      "learning_rate": 1.9616746289606986e-05,
      "loss": 0.7352,
      "step": 661
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7445357765241579,
      "learning_rate": 1.961518732867133e-05,
      "loss": 0.7451,
      "step": 662
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5796015588976786,
      "learning_rate": 1.961362526567587e-05,
      "loss": 0.7695,
      "step": 663
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7815653887355843,
      "learning_rate": 1.961206010112456e-05,
      "loss": 0.755,
      "step": 664
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7638337574993752,
      "learning_rate": 1.9610491835522346e-05,
      "loss": 0.7973,
      "step": 665
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7321082351469685,
      "learning_rate": 1.9608920469375192e-05,
      "loss": 0.7144,
      "step": 666
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8531574315626369,
      "learning_rate": 1.9607346003190052e-05,
      "loss": 0.742,
      "step": 667
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8335275663836073,
      "learning_rate": 1.960576843747488e-05,
      "loss": 0.7607,
      "step": 668
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.784554373827368,
      "learning_rate": 1.960418777273863e-05,
      "loss": 0.7261,
      "step": 669
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9066622216989532,
      "learning_rate": 1.960260400949126e-05,
      "loss": 0.7651,
      "step": 670
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9190190474226622,
      "learning_rate": 1.9601017148243726e-05,
      "loss": 0.7179,
      "step": 671
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7924137729218209,
      "learning_rate": 1.959942718950798e-05,
      "loss": 0.6988,
      "step": 672
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8257709273400585,
      "learning_rate": 1.9597834133796976e-05,
      "loss": 0.7122,
      "step": 673
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8175005931278989,
      "learning_rate": 1.9596237981624668e-05,
      "loss": 0.7416,
      "step": 674
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7124860939812055,
      "learning_rate": 1.959463873350601e-05,
      "loss": 0.7311,
      "step": 675
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9841833044953345,
      "learning_rate": 1.9593036389956946e-05,
      "loss": 0.7356,
      "step": 676
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8238346751303917,
      "learning_rate": 1.9591430951494433e-05,
      "loss": 0.7534,
      "step": 677
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9825072154236664,
      "learning_rate": 1.9589822418636415e-05,
      "loss": 0.7675,
      "step": 678
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8426416905435004,
      "learning_rate": 1.9588210791901837e-05,
      "loss": 0.8224,
      "step": 679
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1256165577415655,
      "learning_rate": 1.9586596071810646e-05,
      "loss": 0.7697,
      "step": 680
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8404138012298858,
      "learning_rate": 1.9584978258883785e-05,
      "loss": 0.7922,
      "step": 681
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.872651223839841,
      "learning_rate": 1.9583357353643193e-05,
      "loss": 0.7974,
      "step": 682
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.71865562316907,
      "learning_rate": 1.9581733356611807e-05,
      "loss": 0.7329,
      "step": 683
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9248065792866831,
      "learning_rate": 1.958010626831356e-05,
      "loss": 0.7668,
      "step": 684
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8140289463733111,
      "learning_rate": 1.957847608927339e-05,
      "loss": 0.7458,
      "step": 685
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7264401345131417,
      "learning_rate": 1.9576842820017225e-05,
      "loss": 0.724,
      "step": 686
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7490432584858902,
      "learning_rate": 1.9575206461071992e-05,
      "loss": 0.7222,
      "step": 687
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7035512587696805,
      "learning_rate": 1.957356701296561e-05,
      "loss": 0.7103,
      "step": 688
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8689293144057544,
      "learning_rate": 1.9571924476227006e-05,
      "loss": 0.808,
      "step": 689
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8065527820963424,
      "learning_rate": 1.9570278851386088e-05,
      "loss": 0.7789,
      "step": 690
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8302363927323284,
      "learning_rate": 1.956863013897378e-05,
      "loss": 0.7414,
      "step": 691
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8570518702750012,
      "learning_rate": 1.9566978339521984e-05,
      "loss": 0.7338,
      "step": 692
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7384544401793969,
      "learning_rate": 1.9565323453563602e-05,
      "loss": 0.7451,
      "step": 693
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7161989924977574,
      "learning_rate": 1.956366548163254e-05,
      "loss": 0.7487,
      "step": 694
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7862804700556284,
      "learning_rate": 1.9562004424263693e-05,
      "loss": 0.7637,
      "step": 695
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7773685204415005,
      "learning_rate": 1.9560340281992954e-05,
      "loss": 0.7619,
      "step": 696
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7185524365580993,
      "learning_rate": 1.955867305535721e-05,
      "loss": 0.7138,
      "step": 697
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7147330285486103,
      "learning_rate": 1.9557002744894338e-05,
      "loss": 0.692,
      "step": 698
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7765990724842751,
      "learning_rate": 1.955532935114322e-05,
      "loss": 0.7562,
      "step": 699
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7477897940338193,
      "learning_rate": 1.9553652874643725e-05,
      "loss": 0.7234,
      "step": 700
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7692264007380458,
      "learning_rate": 1.9551973315936724e-05,
      "loss": 0.7534,
      "step": 701
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7198750131790563,
      "learning_rate": 1.955029067556407e-05,
      "loss": 0.7226,
      "step": 702
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8907995645265948,
      "learning_rate": 1.9548604954068624e-05,
      "loss": 0.7991,
      "step": 703
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7552763487598365,
      "learning_rate": 1.9546916151994233e-05,
      "loss": 0.73,
      "step": 704
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8877639651286385,
      "learning_rate": 1.954522426988574e-05,
      "loss": 0.7977,
      "step": 705
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9157546694203894,
      "learning_rate": 1.954352930828898e-05,
      "loss": 0.8348,
      "step": 706
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7408630699192361,
      "learning_rate": 1.9541831267750785e-05,
      "loss": 0.7415,
      "step": 707
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7624770731898197,
      "learning_rate": 1.9540130148818978e-05,
      "loss": 0.7471,
      "step": 708
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.854274021124812,
      "learning_rate": 1.9538425952042374e-05,
      "loss": 0.7341,
      "step": 709
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7932717273672455,
      "learning_rate": 1.953671867797078e-05,
      "loss": 0.745,
      "step": 710
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8141962686252763,
      "learning_rate": 1.9535008327155004e-05,
      "loss": 0.8173,
      "step": 711
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8454792388378248,
      "learning_rate": 1.953329490014684e-05,
      "loss": 0.8177,
      "step": 712
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8729256941465339,
      "learning_rate": 1.953157839749907e-05,
      "loss": 0.7204,
      "step": 713
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7521297236552301,
      "learning_rate": 1.952985881976548e-05,
      "loss": 0.713,
      "step": 714
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7510456939157029,
      "learning_rate": 1.9528136167500835e-05,
      "loss": 0.7852,
      "step": 715
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6729388972273128,
      "learning_rate": 1.9526410441260906e-05,
      "loss": 0.72,
      "step": 716
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7379414203538865,
      "learning_rate": 1.9524681641602444e-05,
      "loss": 0.7328,
      "step": 717
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7475296515458226,
      "learning_rate": 1.952294976908319e-05,
      "loss": 0.7284,
      "step": 718
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8809137616029937,
      "learning_rate": 1.9521214824261893e-05,
      "loss": 0.7764,
      "step": 719
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.839816618387551,
      "learning_rate": 1.9519476807698278e-05,
      "loss": 0.7519,
      "step": 720
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8354399746459169,
      "learning_rate": 1.9517735719953065e-05,
      "loss": 0.7372,
      "step": 721
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7858516516804052,
      "learning_rate": 1.9515991561587963e-05,
      "loss": 0.7552,
      "step": 722
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7191735057111376,
      "learning_rate": 1.951424433316568e-05,
      "loss": 0.715,
      "step": 723
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9158915621330485,
      "learning_rate": 1.95124940352499e-05,
      "loss": 0.8205,
      "step": 724
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7188878904804176,
      "learning_rate": 1.9510740668405312e-05,
      "loss": 0.7509,
      "step": 725
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8230621847778415,
      "learning_rate": 1.9508984233197585e-05,
      "loss": 0.7522,
      "step": 726
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.718567251274073,
      "learning_rate": 1.9507224730193383e-05,
      "loss": 0.7329,
      "step": 727
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8877738548602122,
      "learning_rate": 1.950546215996036e-05,
      "loss": 0.7737,
      "step": 728
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9040907673922772,
      "learning_rate": 1.9503696523067158e-05,
      "loss": 0.7244,
      "step": 729
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8208644014983358,
      "learning_rate": 1.9501927820083402e-05,
      "loss": 0.7938,
      "step": 730
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8340577123038697,
      "learning_rate": 1.950015605157972e-05,
      "loss": 0.7235,
      "step": 731
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7219745089325351,
      "learning_rate": 1.949838121812772e-05,
      "loss": 0.7084,
      "step": 732
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8742552440358988,
      "learning_rate": 1.9496603320299994e-05,
      "loss": 0.7772,
      "step": 733
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8386051024464336,
      "learning_rate": 1.9494822358670137e-05,
      "loss": 0.7605,
      "step": 734
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7695677692481361,
      "learning_rate": 1.9493038333812722e-05,
      "loss": 0.7919,
      "step": 735
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7942857330069666,
      "learning_rate": 1.949125124630331e-05,
      "loss": 0.76,
      "step": 736
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7982203347699682,
      "learning_rate": 1.9489461096718454e-05,
      "loss": 0.723,
      "step": 737
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8386424212268134,
      "learning_rate": 1.9487667885635694e-05,
      "loss": 0.7864,
      "step": 738
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7881743307849663,
      "learning_rate": 1.9485871613633557e-05,
      "loss": 0.766,
      "step": 739
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.790584643651024,
      "learning_rate": 1.9484072281291556e-05,
      "loss": 0.7759,
      "step": 740
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7718971327759712,
      "learning_rate": 1.94822698891902e-05,
      "loss": 0.8162,
      "step": 741
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7453757747439862,
      "learning_rate": 1.948046443791097e-05,
      "loss": 0.718,
      "step": 742
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7796142164839395,
      "learning_rate": 1.9478655928036345e-05,
      "loss": 0.7532,
      "step": 743
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7926341783376187,
      "learning_rate": 1.947684436014979e-05,
      "loss": 0.7298,
      "step": 744
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8331962199825166,
      "learning_rate": 1.9475029734835754e-05,
      "loss": 0.7497,
      "step": 745
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8560071728488087,
      "learning_rate": 1.9473212052679667e-05,
      "loss": 0.775,
      "step": 746
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7580470207996753,
      "learning_rate": 1.947139131426796e-05,
      "loss": 0.7784,
      "step": 747
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7265208121720204,
      "learning_rate": 1.9469567520188036e-05,
      "loss": 0.7186,
      "step": 748
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.872027530368671,
      "learning_rate": 1.9467740671028292e-05,
      "loss": 0.783,
      "step": 749
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7397228284248673,
      "learning_rate": 1.94659107673781e-05,
      "loss": 0.7359,
      "step": 750
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.816133148117238,
      "learning_rate": 1.9464077809827837e-05,
      "loss": 0.7529,
      "step": 751
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7244030072944136,
      "learning_rate": 1.9462241798968846e-05,
      "loss": 0.7454,
      "step": 752
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7734347050455592,
      "learning_rate": 1.9460402735393466e-05,
      "loss": 0.751,
      "step": 753
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7560133100468683,
      "learning_rate": 1.9458560619695007e-05,
      "loss": 0.7856,
      "step": 754
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8123980274469631,
      "learning_rate": 1.945671545246779e-05,
      "loss": 0.7598,
      "step": 755
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.7008438923259495,
      "learning_rate": 1.945486723430709e-05,
      "loss": 0.8014,
      "step": 756
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7646662650994435,
      "learning_rate": 1.945301596580919e-05,
      "loss": 0.7459,
      "step": 757
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7795209198239761,
      "learning_rate": 1.9451161647571343e-05,
      "loss": 0.7619,
      "step": 758
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7768205249404685,
      "learning_rate": 1.944930428019179e-05,
      "loss": 0.7332,
      "step": 759
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8237898771009718,
      "learning_rate": 1.9447443864269763e-05,
      "loss": 0.7851,
      "step": 760
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8103895441857897,
      "learning_rate": 1.944558040040547e-05,
      "loss": 0.7862,
      "step": 761
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7534274269896507,
      "learning_rate": 1.944371388920009e-05,
      "loss": 0.7616,
      "step": 762
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9002157667942096,
      "learning_rate": 1.9441844331255813e-05,
      "loss": 0.7875,
      "step": 763
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7006669745734746,
      "learning_rate": 1.9439971727175794e-05,
      "loss": 0.7264,
      "step": 764
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7296396161037644,
      "learning_rate": 1.943809607756417e-05,
      "loss": 0.7699,
      "step": 765
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7469177482199345,
      "learning_rate": 1.943621738302607e-05,
      "loss": 0.7323,
      "step": 766
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9077071654215104,
      "learning_rate": 1.9434335644167598e-05,
      "loss": 0.7725,
      "step": 767
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7901608032563896,
      "learning_rate": 1.9432450861595843e-05,
      "loss": 0.7942,
      "step": 768
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8117943689875206,
      "learning_rate": 1.9430563035918868e-05,
      "loss": 0.7849,
      "step": 769
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7588552667889448,
      "learning_rate": 1.9428672167745733e-05,
      "loss": 0.7424,
      "step": 770
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8513413621989011,
      "learning_rate": 1.9426778257686472e-05,
      "loss": 0.7216,
      "step": 771
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8090413132225497,
      "learning_rate": 1.9424881306352094e-05,
      "loss": 0.7585,
      "step": 772
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8784820810242319,
      "learning_rate": 1.94229813143546e-05,
      "loss": 0.7521,
      "step": 773
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7398735117337873,
      "learning_rate": 1.9421078282306963e-05,
      "loss": 0.7639,
      "step": 774
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8495484744404064,
      "learning_rate": 1.941917221082314e-05,
      "loss": 0.7823,
      "step": 775
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.858956843203514,
      "learning_rate": 1.9417263100518078e-05,
      "loss": 0.7676,
      "step": 776
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8282704026170725,
      "learning_rate": 1.941535095200769e-05,
      "loss": 0.7773,
      "step": 777
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9000269084278699,
      "learning_rate": 1.941343576590887e-05,
      "loss": 0.8096,
      "step": 778
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7604085718112832,
      "learning_rate": 1.9411517542839504e-05,
      "loss": 0.7395,
      "step": 779
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4311214763645568,
      "learning_rate": 1.9409596283418448e-05,
      "loss": 0.7236,
      "step": 780
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.792624391359825,
      "learning_rate": 1.9407671988265545e-05,
      "loss": 0.7642,
      "step": 781
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8111003365983842,
      "learning_rate": 1.9405744658001604e-05,
      "loss": 0.7476,
      "step": 782
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7500700751680208,
      "learning_rate": 1.9403814293248428e-05,
      "loss": 0.7125,
      "step": 783
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1113287864741606,
      "learning_rate": 1.940188089462879e-05,
      "loss": 0.8121,
      "step": 784
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8236277295548237,
      "learning_rate": 1.939994446276645e-05,
      "loss": 0.7391,
      "step": 785
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8007889008792068,
      "learning_rate": 1.9398004998286134e-05,
      "loss": 0.7506,
      "step": 786
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9102340130789494,
      "learning_rate": 1.9396062501813558e-05,
      "loss": 0.7343,
      "step": 787
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.759345897689732,
      "learning_rate": 1.9394116973975412e-05,
      "loss": 0.7454,
      "step": 788
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8582563179273709,
      "learning_rate": 1.9392168415399363e-05,
      "loss": 0.7431,
      "step": 789
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8504392937405674,
      "learning_rate": 1.9390216826714053e-05,
      "loss": 0.7447,
      "step": 790
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8954385083962727,
      "learning_rate": 1.938826220854911e-05,
      "loss": 0.7799,
      "step": 791
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7631232540298577,
      "learning_rate": 1.9386304561535138e-05,
      "loss": 0.8075,
      "step": 792
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8289200292552591,
      "learning_rate": 1.9384343886303703e-05,
      "loss": 0.7562,
      "step": 793
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7018660991577796,
      "learning_rate": 1.9382380183487368e-05,
      "loss": 0.7532,
      "step": 794
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7233056938549726,
      "learning_rate": 1.9380413453719663e-05,
      "loss": 0.7648,
      "step": 795
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8015718214497052,
      "learning_rate": 1.9378443697635097e-05,
      "loss": 0.7597,
      "step": 796
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8440941191253214,
      "learning_rate": 1.9376470915869154e-05,
      "loss": 0.7838,
      "step": 797
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8320170127288685,
      "learning_rate": 1.9374495109058292e-05,
      "loss": 0.7636,
      "step": 798
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7695677281204358,
      "learning_rate": 1.937251627783995e-05,
      "loss": 0.7327,
      "step": 799
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7819767644828991,
      "learning_rate": 1.9370534422852542e-05,
      "loss": 0.7805,
      "step": 800
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8072888364761335,
      "learning_rate": 1.936854954473545e-05,
      "loss": 0.8035,
      "step": 801
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7824185432744247,
      "learning_rate": 1.936656164412904e-05,
      "loss": 0.765,
      "step": 802
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7766383803496726,
      "learning_rate": 1.9364570721674655e-05,
      "loss": 0.7489,
      "step": 803
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8617386097675106,
      "learning_rate": 1.93625767780146e-05,
      "loss": 0.7648,
      "step": 804
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7035311182657393,
      "learning_rate": 1.9360579813792164e-05,
      "loss": 0.7718,
      "step": 805
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0800389717614205,
      "learning_rate": 1.9358579829651615e-05,
      "loss": 0.8152,
      "step": 806
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7233906074340729,
      "learning_rate": 1.935657682623818e-05,
      "loss": 0.7107,
      "step": 807
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8074134267248764,
      "learning_rate": 1.9354570804198082e-05,
      "loss": 0.741,
      "step": 808
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.867407865602712,
      "learning_rate": 1.93525617641785e-05,
      "loss": 0.764,
      "step": 809
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.784323951946584,
      "learning_rate": 1.935054970682759e-05,
      "loss": 0.774,
      "step": 810
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0083089734445698,
      "learning_rate": 1.9348534632794482e-05,
      "loss": 0.7885,
      "step": 811
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6525840058604645,
      "learning_rate": 1.9346516542729285e-05,
      "loss": 0.7364,
      "step": 812
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8440195230431399,
      "learning_rate": 1.9344495437283077e-05,
      "loss": 0.75,
      "step": 813
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.820408254975336,
      "learning_rate": 1.934247131710791e-05,
      "loss": 0.7782,
      "step": 814
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0846188371461798,
      "learning_rate": 1.93404441828568e-05,
      "loss": 0.8234,
      "step": 815
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7532549953034324,
      "learning_rate": 1.933841403518375e-05,
      "loss": 0.8048,
      "step": 816
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.157263055837497,
      "learning_rate": 1.9336380874743727e-05,
      "loss": 0.7603,
      "step": 817
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.665783758333789,
      "learning_rate": 1.9334344702192664e-05,
      "loss": 0.7104,
      "step": 818
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7902015752994779,
      "learning_rate": 1.9332305518187484e-05,
      "loss": 0.7882,
      "step": 819
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7205975266826072,
      "learning_rate": 1.9330263323386062e-05,
      "loss": 0.7037,
      "step": 820
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8331106660089228,
      "learning_rate": 1.9328218118447256e-05,
      "loss": 0.8102,
      "step": 821
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7768458513257069,
      "learning_rate": 1.9326169904030892e-05,
      "loss": 0.742,
      "step": 822
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8589847493899629,
      "learning_rate": 1.9324118680797764e-05,
      "loss": 0.7637,
      "step": 823
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7455110524239575,
      "learning_rate": 1.932206444940964e-05,
      "loss": 0.7815,
      "step": 824
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7752154334865642,
      "learning_rate": 1.932000721052926e-05,
      "loss": 0.8228,
      "step": 825
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6950538041620148,
      "learning_rate": 1.9317946964820332e-05,
      "loss": 0.72,
      "step": 826
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7191744915084023,
      "learning_rate": 1.931588371294753e-05,
      "loss": 0.7537,
      "step": 827
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8039591028516956,
      "learning_rate": 1.9313817455576506e-05,
      "loss": 0.7657,
      "step": 828
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7798978170399651,
      "learning_rate": 1.931174819337388e-05,
      "loss": 0.7458,
      "step": 829
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7674825966855309,
      "learning_rate": 1.930967592700723e-05,
      "loss": 0.7746,
      "step": 830
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7945981779230291,
      "learning_rate": 1.930760065714512e-05,
      "loss": 0.7562,
      "step": 831
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8359334482782033,
      "learning_rate": 1.9305522384457078e-05,
      "loss": 0.7861,
      "step": 832
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7447684704685391,
      "learning_rate": 1.9303441109613586e-05,
      "loss": 0.7469,
      "step": 833
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6888261289758142,
      "learning_rate": 1.9301356833286122e-05,
      "loss": 0.7148,
      "step": 834
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8092300689068944,
      "learning_rate": 1.9299269556147103e-05,
      "loss": 0.787,
      "step": 835
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7294126998828971,
      "learning_rate": 1.9297179278869936e-05,
      "loss": 0.7284,
      "step": 836
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7232498765268374,
      "learning_rate": 1.929508600212899e-05,
      "loss": 0.7365,
      "step": 837
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.788980131910967,
      "learning_rate": 1.9292989726599593e-05,
      "loss": 0.7659,
      "step": 838
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7573175249747485,
      "learning_rate": 1.9290890452958053e-05,
      "loss": 0.7699,
      "step": 839
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8582941584183753,
      "learning_rate": 1.928878818188163e-05,
      "loss": 0.7479,
      "step": 840
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7537484735060604,
      "learning_rate": 1.9286682914048575e-05,
      "loss": 0.7689,
      "step": 841
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7831395298302953,
      "learning_rate": 1.9284574650138076e-05,
      "loss": 0.7506,
      "step": 842
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7572542961114399,
      "learning_rate": 1.9282463390830315e-05,
      "loss": 0.7599,
      "step": 843
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8139424792363875,
      "learning_rate": 1.9280349136806422e-05,
      "loss": 0.7821,
      "step": 844
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7594542181624855,
      "learning_rate": 1.9278231888748503e-05,
      "loss": 0.7117,
      "step": 845
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9224467690471277,
      "learning_rate": 1.927611164733962e-05,
      "loss": 0.7157,
      "step": 846
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9090997891576609,
      "learning_rate": 1.9273988413263813e-05,
      "loss": 0.7277,
      "step": 847
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8884519667842612,
      "learning_rate": 1.927186218720608e-05,
      "loss": 0.7977,
      "step": 848
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1079482778534167,
      "learning_rate": 1.9269732969852386e-05,
      "loss": 0.7639,
      "step": 849
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8103074282439823,
      "learning_rate": 1.926760076188966e-05,
      "loss": 0.7327,
      "step": 850
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8040022881186183,
      "learning_rate": 1.9265465564005797e-05,
      "loss": 0.7955,
      "step": 851
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.833055586335902,
      "learning_rate": 1.9263327376889657e-05,
      "loss": 0.7145,
      "step": 852
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8263639003531955,
      "learning_rate": 1.9261186201231067e-05,
      "loss": 0.7512,
      "step": 853
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9727627936896989,
      "learning_rate": 1.9259042037720804e-05,
      "loss": 0.745,
      "step": 854
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8379804970410246,
      "learning_rate": 1.925689488705063e-05,
      "loss": 0.734,
      "step": 855
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.906483411849566,
      "learning_rate": 1.925474474991326e-05,
      "loss": 0.7952,
      "step": 856
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8423369691673629,
      "learning_rate": 1.925259162700237e-05,
      "loss": 0.7421,
      "step": 857
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8188238329385958,
      "learning_rate": 1.92504355190126e-05,
      "loss": 0.7805,
      "step": 858
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9276514842367802,
      "learning_rate": 1.924827642663956e-05,
      "loss": 0.7748,
      "step": 859
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6887265948885554,
      "learning_rate": 1.924611435057982e-05,
      "loss": 0.743,
      "step": 860
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.836629908101969,
      "learning_rate": 1.9243949291530903e-05,
      "loss": 0.7467,
      "step": 861
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8346739159040738,
      "learning_rate": 1.9241781250191307e-05,
      "loss": 0.7669,
      "step": 862
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7299294606141791,
      "learning_rate": 1.9239610227260487e-05,
      "loss": 0.7283,
      "step": 863
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8070819975563868,
      "learning_rate": 1.9237436223438863e-05,
      "loss": 0.7492,
      "step": 864
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7037457805950145,
      "learning_rate": 1.9235259239427808e-05,
      "loss": 0.7692,
      "step": 865
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6847608898504713,
      "learning_rate": 1.9233079275929666e-05,
      "loss": 0.7216,
      "step": 866
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7136076750277384,
      "learning_rate": 1.9230896333647736e-05,
      "loss": 0.7518,
      "step": 867
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7929694558791852,
      "learning_rate": 1.9228710413286286e-05,
      "loss": 0.7419,
      "step": 868
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7561940517977154,
      "learning_rate": 1.9226521515550534e-05,
      "loss": 0.7497,
      "step": 869
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8035011967558341,
      "learning_rate": 1.9224329641146666e-05,
      "loss": 0.7885,
      "step": 870
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7142304109677082,
      "learning_rate": 1.9222134790781828e-05,
      "loss": 0.7405,
      "step": 871
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7951102149966983,
      "learning_rate": 1.9219936965164123e-05,
      "loss": 0.7801,
      "step": 872
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8025763565424999,
      "learning_rate": 1.9217736165002617e-05,
      "loss": 0.78,
      "step": 873
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7361433065492453,
      "learning_rate": 1.921553239100733e-05,
      "loss": 0.7001,
      "step": 874
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7608899584273666,
      "learning_rate": 1.9213325643889245e-05,
      "loss": 0.754,
      "step": 875
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7170264306828357,
      "learning_rate": 1.9211115924360314e-05,
      "loss": 0.726,
      "step": 876
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8463444497918639,
      "learning_rate": 1.9208903233133428e-05,
      "loss": 0.7118,
      "step": 877
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8859603837273686,
      "learning_rate": 1.920668757092245e-05,
      "loss": 0.7687,
      "step": 878
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7287779158112355,
      "learning_rate": 1.9204468938442203e-05,
      "loss": 0.7558,
      "step": 879
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8600728982836613,
      "learning_rate": 1.920224733640846e-05,
      "loss": 0.7004,
      "step": 880
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8420167203443023,
      "learning_rate": 1.920002276553796e-05,
      "loss": 0.763,
      "step": 881
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8061249942932334,
      "learning_rate": 1.9197795226548388e-05,
      "loss": 0.727,
      "step": 882
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7880858185107402,
      "learning_rate": 1.91955647201584e-05,
      "loss": 0.7131,
      "step": 883
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7722174178916775,
      "learning_rate": 1.9193331247087608e-05,
      "loss": 0.7319,
      "step": 884
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7459374791986275,
      "learning_rate": 1.919109480805657e-05,
      "loss": 0.7477,
      "step": 885
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8659092577812452,
      "learning_rate": 1.9188855403786812e-05,
      "loss": 0.7839,
      "step": 886
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7688099953982104,
      "learning_rate": 1.9186613035000807e-05,
      "loss": 0.7401,
      "step": 887
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.85749015139832,
      "learning_rate": 1.9184367702422e-05,
      "loss": 0.7548,
      "step": 888
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7870278686646514,
      "learning_rate": 1.918211940677477e-05,
      "loss": 0.801,
      "step": 889
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8373341969001753,
      "learning_rate": 1.917986814878447e-05,
      "loss": 0.7551,
      "step": 890
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8182074550747407,
      "learning_rate": 1.9177613929177403e-05,
      "loss": 0.763,
      "step": 891
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8179863357862194,
      "learning_rate": 1.917535674868083e-05,
      "loss": 0.7169,
      "step": 892
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7867996136578702,
      "learning_rate": 1.9173096608022957e-05,
      "loss": 0.7409,
      "step": 893
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8272934519340924,
      "learning_rate": 1.917083350793296e-05,
      "loss": 0.7999,
      "step": 894
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7663960919871204,
      "learning_rate": 1.9168567449140956e-05,
      "loss": 0.7126,
      "step": 895
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7053123988118982,
      "learning_rate": 1.9166298432378028e-05,
      "loss": 0.7181,
      "step": 896
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9769764163935666,
      "learning_rate": 1.9164026458376205e-05,
      "loss": 0.7445,
      "step": 897
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7518515259449893,
      "learning_rate": 1.9161751527868476e-05,
      "loss": 0.7312,
      "step": 898
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7027655536927091,
      "learning_rate": 1.9159473641588777e-05,
      "loss": 0.7588,
      "step": 899
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9931930026733069,
      "learning_rate": 1.9157192800272004e-05,
      "loss": 0.7668,
      "step": 900
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8965269331410464,
      "learning_rate": 1.9154909004654002e-05,
      "loss": 0.7577,
      "step": 901
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7834497563834214,
      "learning_rate": 1.9152622255471575e-05,
      "loss": 0.7979,
      "step": 902
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8908444605401595,
      "learning_rate": 1.915033255346247e-05,
      "loss": 0.775,
      "step": 903
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8721048566377623,
      "learning_rate": 1.91480398993654e-05,
      "loss": 0.7885,
      "step": 904
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8570576782375199,
      "learning_rate": 1.9145744293920015e-05,
      "loss": 0.8156,
      "step": 905
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8038807834227626,
      "learning_rate": 1.9143445737866935e-05,
      "loss": 0.7359,
      "step": 906
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7656298678246016,
      "learning_rate": 1.9141144231947713e-05,
      "loss": 0.777,
      "step": 907
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8047501932722437,
      "learning_rate": 1.9138839776904867e-05,
      "loss": 0.772,
      "step": 908
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8219674474485854,
      "learning_rate": 1.913653237348186e-05,
      "loss": 0.768,
      "step": 909
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7779492156971773,
      "learning_rate": 1.913422202242311e-05,
      "loss": 0.7458,
      "step": 910
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7776521934679539,
      "learning_rate": 1.9131908724473986e-05,
      "loss": 0.7771,
      "step": 911
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.755133634968864,
      "learning_rate": 1.9129592480380803e-05,
      "loss": 0.7408,
      "step": 912
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8764062860760832,
      "learning_rate": 1.912727329089083e-05,
      "loss": 0.746,
      "step": 913
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8153012681362456,
      "learning_rate": 1.9124951156752287e-05,
      "loss": 0.8116,
      "step": 914
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7259146115470386,
      "learning_rate": 1.9122626078714345e-05,
      "loss": 0.7543,
      "step": 915
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7629615666477462,
      "learning_rate": 1.912029805752712e-05,
      "loss": 0.7481,
      "step": 916
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6615569565581164,
      "learning_rate": 1.9117967093941675e-05,
      "loss": 0.7273,
      "step": 917
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8513538113748502,
      "learning_rate": 1.9115633188710037e-05,
      "loss": 0.7687,
      "step": 918
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6512544680818377,
      "learning_rate": 1.9113296342585167e-05,
      "loss": 0.7087,
      "step": 919
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6964543208760609,
      "learning_rate": 1.9110956556320984e-05,
      "loss": 0.7102,
      "step": 920
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.751428492862186,
      "learning_rate": 1.9108613830672348e-05,
      "loss": 0.7522,
      "step": 921
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7564227458532978,
      "learning_rate": 1.9106268166395068e-05,
      "loss": 0.7627,
      "step": 922
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8280094105655101,
      "learning_rate": 1.9103919564245916e-05,
      "loss": 0.7129,
      "step": 923
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7637378833479229,
      "learning_rate": 1.910156802498259e-05,
      "loss": 0.7511,
      "step": 924
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8375981917708896,
      "learning_rate": 1.9099213549363747e-05,
      "loss": 0.7672,
      "step": 925
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.262098952805081,
      "learning_rate": 1.9096856138148993e-05,
      "loss": 0.7731,
      "step": 926
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7622351876585717,
      "learning_rate": 1.9094495792098874e-05,
      "loss": 0.7569,
      "step": 927
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7411815225959385,
      "learning_rate": 1.909213251197489e-05,
      "loss": 0.7893,
      "step": 928
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9021166604799222,
      "learning_rate": 1.908976629853948e-05,
      "loss": 0.7639,
      "step": 929
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7747097126805144,
      "learning_rate": 1.9087397152556043e-05,
      "loss": 0.7538,
      "step": 930
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7302733758959011,
      "learning_rate": 1.9085025074788906e-05,
      "loss": 0.7451,
      "step": 931
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6463356593459757,
      "learning_rate": 1.9082650066003354e-05,
      "loss": 0.7401,
      "step": 932
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6861210910823492,
      "learning_rate": 1.9080272126965615e-05,
      "loss": 0.7278,
      "step": 933
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7772105185097679,
      "learning_rate": 1.907789125844286e-05,
      "loss": 0.7591,
      "step": 934
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6987431579702154,
      "learning_rate": 1.9075507461203205e-05,
      "loss": 0.7307,
      "step": 935
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6962196728408497,
      "learning_rate": 1.9073120736015717e-05,
      "loss": 0.734,
      "step": 936
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7587350915361053,
      "learning_rate": 1.90707310836504e-05,
      "loss": 0.742,
      "step": 937
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6976547365783834,
      "learning_rate": 1.9068338504878212e-05,
      "loss": 0.7323,
      "step": 938
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6646864214774134,
      "learning_rate": 1.9065943000471043e-05,
      "loss": 0.723,
      "step": 939
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7176582209203917,
      "learning_rate": 1.9063544571201728e-05,
      "loss": 0.7376,
      "step": 940
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7522869789298734,
      "learning_rate": 1.906114321784406e-05,
      "loss": 0.7401,
      "step": 941
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7187092748948243,
      "learning_rate": 1.905873894117276e-05,
      "loss": 0.7341,
      "step": 942
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.817231955034069,
      "learning_rate": 1.90563317419635e-05,
      "loss": 0.7004,
      "step": 943
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6429286435648299,
      "learning_rate": 1.9053921620992893e-05,
      "loss": 0.7156,
      "step": 944
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8984477121175869,
      "learning_rate": 1.905150857903849e-05,
      "loss": 0.7177,
      "step": 945
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7329864201047219,
      "learning_rate": 1.9049092616878796e-05,
      "loss": 0.7554,
      "step": 946
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8243376867286705,
      "learning_rate": 1.9046673735293243e-05,
      "loss": 0.7257,
      "step": 947
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7319216880636694,
      "learning_rate": 1.9044251935062214e-05,
      "loss": 0.7753,
      "step": 948
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9010190608851502,
      "learning_rate": 1.904182721696704e-05,
      "loss": 0.762,
      "step": 949
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.677916422753305,
      "learning_rate": 1.9039399581789974e-05,
      "loss": 0.7002,
      "step": 950
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8427688153711639,
      "learning_rate": 1.903696903031423e-05,
      "loss": 0.7188,
      "step": 951
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7522658842867034,
      "learning_rate": 1.9034535563323952e-05,
      "loss": 0.7383,
      "step": 952
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8503630445241487,
      "learning_rate": 1.9032099181604227e-05,
      "loss": 0.7628,
      "step": 953
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5834438000726055,
      "learning_rate": 1.9029659885941087e-05,
      "loss": 0.6949,
      "step": 954
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7852526356256073,
      "learning_rate": 1.902721767712149e-05,
      "loss": 0.7271,
      "step": 955
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7826033250144177,
      "learning_rate": 1.902477255593335e-05,
      "loss": 0.7639,
      "step": 956
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6529926545591902,
      "learning_rate": 1.9022324523165514e-05,
      "loss": 0.7019,
      "step": 957
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7384990764356315,
      "learning_rate": 1.901987357960777e-05,
      "loss": 0.7142,
      "step": 958
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.686698164827462,
      "learning_rate": 1.9017419726050836e-05,
      "loss": 0.7428,
      "step": 959
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8041961544156786,
      "learning_rate": 1.9014962963286382e-05,
      "loss": 0.7655,
      "step": 960
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.770307268549736,
      "learning_rate": 1.9012503292107013e-05,
      "loss": 0.743,
      "step": 961
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.756426893089863,
      "learning_rate": 1.9010040713306264e-05,
      "loss": 0.7684,
      "step": 962
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8025947695175756,
      "learning_rate": 1.9007575227678616e-05,
      "loss": 0.7751,
      "step": 963
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8518443881004355,
      "learning_rate": 1.9005106836019488e-05,
      "loss": 0.8192,
      "step": 964
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7801440564595743,
      "learning_rate": 1.9002635539125238e-05,
      "loss": 0.7254,
      "step": 965
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8276075912400426,
      "learning_rate": 1.900016133779315e-05,
      "loss": 0.7566,
      "step": 966
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7231568049127243,
      "learning_rate": 1.8997684232821458e-05,
      "loss": 0.7079,
      "step": 967
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0161463327612232,
      "learning_rate": 1.8995204225009324e-05,
      "loss": 0.804,
      "step": 968
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8081847931035346,
      "learning_rate": 1.8992721315156856e-05,
      "loss": 0.7402,
      "step": 969
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7394328229481172,
      "learning_rate": 1.8990235504065088e-05,
      "loss": 0.7361,
      "step": 970
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7510931496194859,
      "learning_rate": 1.8987746792535996e-05,
      "loss": 0.7567,
      "step": 971
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7698441318061824,
      "learning_rate": 1.898525518137249e-05,
      "loss": 0.7622,
      "step": 972
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7516001251616994,
      "learning_rate": 1.8982760671378414e-05,
      "loss": 0.7474,
      "step": 973
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7514682719257135,
      "learning_rate": 1.8980263263358553e-05,
      "loss": 0.7426,
      "step": 974
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8230886657040679,
      "learning_rate": 1.8977762958118617e-05,
      "loss": 0.7475,
      "step": 975
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7083414112539748,
      "learning_rate": 1.8975259756465263e-05,
      "loss": 0.7229,
      "step": 976
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8274865185364843,
      "learning_rate": 1.8972753659206068e-05,
      "loss": 0.7545,
      "step": 977
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6447154249448948,
      "learning_rate": 1.8970244667149564e-05,
      "loss": 0.6958,
      "step": 978
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8492939925579488,
      "learning_rate": 1.8967732781105187e-05,
      "loss": 0.7193,
      "step": 979
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.870283062522345,
      "learning_rate": 1.896521800188334e-05,
      "loss": 0.7519,
      "step": 980
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7744260348816637,
      "learning_rate": 1.896270033029533e-05,
      "loss": 0.7154,
      "step": 981
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6813451466157958,
      "learning_rate": 1.896017976715342e-05,
      "loss": 0.7248,
      "step": 982
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9162455030471247,
      "learning_rate": 1.8957656313270795e-05,
      "loss": 0.7079,
      "step": 983
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6865379122114941,
      "learning_rate": 1.8955129969461565e-05,
      "loss": 0.726,
      "step": 984
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8292064838325677,
      "learning_rate": 1.8952600736540795e-05,
      "loss": 0.775,
      "step": 985
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7755617987802609,
      "learning_rate": 1.8950068615324455e-05,
      "loss": 0.7578,
      "step": 986
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8047560511183307,
      "learning_rate": 1.894753360662947e-05,
      "loss": 0.7652,
      "step": 987
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8945851859913141,
      "learning_rate": 1.894499571127368e-05,
      "loss": 0.7476,
      "step": 988
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6342112086924154,
      "learning_rate": 1.8942454930075865e-05,
      "loss": 0.7119,
      "step": 989
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7919195887711313,
      "learning_rate": 1.8939911263855735e-05,
      "loss": 0.6995,
      "step": 990
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7190522172356664,
      "learning_rate": 1.893736471343393e-05,
      "loss": 0.7791,
      "step": 991
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7601627989320285,
      "learning_rate": 1.8934815279632018e-05,
      "loss": 0.6913,
      "step": 992
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8591806248906775,
      "learning_rate": 1.8932262963272503e-05,
      "loss": 0.7168,
      "step": 993
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.808678543129201,
      "learning_rate": 1.8929707765178812e-05,
      "loss": 0.7432,
      "step": 994
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8509940980601528,
      "learning_rate": 1.8927149686175307e-05,
      "loss": 0.7268,
      "step": 995
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7077077677884844,
      "learning_rate": 1.8924588727087277e-05,
      "loss": 0.7557,
      "step": 996
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8487266355769154,
      "learning_rate": 1.8922024888740943e-05,
      "loss": 0.7656,
      "step": 997
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8639486568099437,
      "learning_rate": 1.891945817196345e-05,
      "loss": 0.7415,
      "step": 998
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7587480718536903,
      "learning_rate": 1.8916888577582872e-05,
      "loss": 0.7527,
      "step": 999
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8199348371213521,
      "learning_rate": 1.8914316106428216e-05,
      "loss": 0.7245,
      "step": 1000
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6554188728525571,
      "learning_rate": 1.8911740759329418e-05,
      "loss": 0.7108,
      "step": 1001
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7689017669473027,
      "learning_rate": 1.8909162537117335e-05,
      "loss": 0.7544,
      "step": 1002
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8552415772067072,
      "learning_rate": 1.890658144062376e-05,
      "loss": 0.763,
      "step": 1003
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.694806353649481,
      "learning_rate": 1.89039974706814e-05,
      "loss": 0.7659,
      "step": 1004
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7180637337651611,
      "learning_rate": 1.8901410628123898e-05,
      "loss": 0.6894,
      "step": 1005
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.826263746783263,
      "learning_rate": 1.8898820913785836e-05,
      "loss": 0.7613,
      "step": 1006
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9093297638977191,
      "learning_rate": 1.8896228328502698e-05,
      "loss": 0.8418,
      "step": 1007
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8224237528482481,
      "learning_rate": 1.889363287311091e-05,
      "loss": 0.7317,
      "step": 1008
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7215284287695675,
      "learning_rate": 1.8891034548447818e-05,
      "loss": 0.7161,
      "step": 1009
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8085375144418608,
      "learning_rate": 1.8888433355351698e-05,
      "loss": 0.7952,
      "step": 1010
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6399051064975003,
      "learning_rate": 1.888582929466175e-05,
      "loss": 0.7051,
      "step": 1011
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9036463537607607,
      "learning_rate": 1.8883222367218096e-05,
      "loss": 0.7245,
      "step": 1012
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7778998486017087,
      "learning_rate": 1.8880612573861782e-05,
      "loss": 0.7716,
      "step": 1013
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.724619401459826,
      "learning_rate": 1.887799991543479e-05,
      "loss": 0.7508,
      "step": 1014
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8688635413900918,
      "learning_rate": 1.8875384392780008e-05,
      "loss": 0.7864,
      "step": 1015
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7612905637798392,
      "learning_rate": 1.8872766006741268e-05,
      "loss": 0.6932,
      "step": 1016
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7140515559579382,
      "learning_rate": 1.8870144758163312e-05,
      "loss": 0.6759,
      "step": 1017
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8289097063133362,
      "learning_rate": 1.8867520647891805e-05,
      "loss": 0.7509,
      "step": 1018
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7657938277184583,
      "learning_rate": 1.8864893676773345e-05,
      "loss": 0.746,
      "step": 1019
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.70392613763524,
      "learning_rate": 1.8862263845655447e-05,
      "loss": 0.7082,
      "step": 1020
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8238449752437228,
      "learning_rate": 1.8859631155386547e-05,
      "loss": 0.7405,
      "step": 1021
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8398859081436716,
      "learning_rate": 1.8856995606816008e-05,
      "loss": 0.7454,
      "step": 1022
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7732393324034595,
      "learning_rate": 1.8854357200794113e-05,
      "loss": 0.7521,
      "step": 1023
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7897478340880087,
      "learning_rate": 1.8851715938172065e-05,
      "loss": 0.7329,
      "step": 1024
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7988683864649565,
      "learning_rate": 1.8849071819801993e-05,
      "loss": 0.7469,
      "step": 1025
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7850658366065575,
      "learning_rate": 1.884642484653694e-05,
      "loss": 0.754,
      "step": 1026
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7746931234387092,
      "learning_rate": 1.884377501923088e-05,
      "loss": 0.7387,
      "step": 1027
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6594406892901932,
      "learning_rate": 1.8841122338738696e-05,
      "loss": 0.6995,
      "step": 1028
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7096631281418979,
      "learning_rate": 1.8838466805916206e-05,
      "loss": 0.6991,
      "step": 1029
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7052926059592697,
      "learning_rate": 1.8835808421620135e-05,
      "loss": 0.727,
      "step": 1030
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7320198293043083,
      "learning_rate": 1.8833147186708135e-05,
      "loss": 0.6832,
      "step": 1031
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7929733699258934,
      "learning_rate": 1.8830483102038775e-05,
      "loss": 0.7304,
      "step": 1032
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.826544873404198,
      "learning_rate": 1.8827816168471543e-05,
      "loss": 0.7479,
      "step": 1033
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7401731442335532,
      "learning_rate": 1.8825146386866852e-05,
      "loss": 0.7654,
      "step": 1034
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.75880469066902,
      "learning_rate": 1.8822473758086025e-05,
      "loss": 0.7717,
      "step": 1035
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7995993208437712,
      "learning_rate": 1.8819798282991306e-05,
      "loss": 0.7701,
      "step": 1036
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8781634256468713,
      "learning_rate": 1.8817119962445867e-05,
      "loss": 0.8118,
      "step": 1037
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6775719829055298,
      "learning_rate": 1.881443879731378e-05,
      "loss": 0.709,
      "step": 1038
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7773378792412322,
      "learning_rate": 1.8811754788460052e-05,
      "loss": 0.7664,
      "step": 1039
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8004800357283477,
      "learning_rate": 1.88090679367506e-05,
      "loss": 0.7229,
      "step": 1040
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.790520995350165,
      "learning_rate": 1.8806378243052252e-05,
      "loss": 0.7731,
      "step": 1041
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7282058318688287,
      "learning_rate": 1.8803685708232765e-05,
      "loss": 0.729,
      "step": 1042
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.687385947363831,
      "learning_rate": 1.8800990333160807e-05,
      "loss": 0.7423,
      "step": 1043
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7763744055278012,
      "learning_rate": 1.879829211870596e-05,
      "loss": 0.7366,
      "step": 1044
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7586032142231582,
      "learning_rate": 1.8795591065738724e-05,
      "loss": 0.7551,
      "step": 1045
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8203984806213649,
      "learning_rate": 1.8792887175130515e-05,
      "loss": 0.7729,
      "step": 1046
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7146994089510392,
      "learning_rate": 1.8790180447753665e-05,
      "loss": 0.7545,
      "step": 1047
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.005920316434439,
      "learning_rate": 1.878747088448142e-05,
      "loss": 0.829,
      "step": 1048
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7466283309060033,
      "learning_rate": 1.8784758486187945e-05,
      "loss": 0.7199,
      "step": 1049
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8340868543267653,
      "learning_rate": 1.878204325374831e-05,
      "loss": 0.7505,
      "step": 1050
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7217488281059433,
      "learning_rate": 1.8779325188038512e-05,
      "loss": 0.7731,
      "step": 1051
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8272923207399109,
      "learning_rate": 1.8776604289935454e-05,
      "loss": 0.7131,
      "step": 1052
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7986994266661872,
      "learning_rate": 1.877388056031695e-05,
      "loss": 0.8062,
      "step": 1053
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9742922415213283,
      "learning_rate": 1.8771154000061737e-05,
      "loss": 0.8021,
      "step": 1054
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.795567333161792,
      "learning_rate": 1.876842461004946e-05,
      "loss": 0.7638,
      "step": 1055
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7445791266647259,
      "learning_rate": 1.876569239116067e-05,
      "loss": 0.773,
      "step": 1056
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7498335542901929,
      "learning_rate": 1.8762957344276845e-05,
      "loss": 0.7432,
      "step": 1057
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8128771880420195,
      "learning_rate": 1.876021947028037e-05,
      "loss": 0.7516,
      "step": 1058
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9884523731944245,
      "learning_rate": 1.875747877005453e-05,
      "loss": 0.7271,
      "step": 1059
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7142486526605412,
      "learning_rate": 1.8754735244483545e-05,
      "loss": 0.7511,
      "step": 1060
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7525881845004174,
      "learning_rate": 1.8751988894452525e-05,
      "loss": 0.7397,
      "step": 1061
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6596001478219687,
      "learning_rate": 1.8749239720847498e-05,
      "loss": 0.7065,
      "step": 1062
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7291042821195997,
      "learning_rate": 1.874648772455541e-05,
      "loss": 0.7179,
      "step": 1063
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8321581029258853,
      "learning_rate": 1.8743732906464115e-05,
      "loss": 0.8006,
      "step": 1064
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7854433987044337,
      "learning_rate": 1.8740975267462366e-05,
      "loss": 0.734,
      "step": 1065
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8379442188587956,
      "learning_rate": 1.873821480843984e-05,
      "loss": 0.7636,
      "step": 1066
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7220817863154819,
      "learning_rate": 1.873545153028712e-05,
      "loss": 0.7164,
      "step": 1067
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8612358088199471,
      "learning_rate": 1.8732685433895694e-05,
      "loss": 0.7745,
      "step": 1068
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8342722027912547,
      "learning_rate": 1.872991652015796e-05,
      "loss": 0.758,
      "step": 1069
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7561759379402743,
      "learning_rate": 1.872714478996724e-05,
      "loss": 0.7261,
      "step": 1070
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7904136222151001,
      "learning_rate": 1.8724370244217733e-05,
      "loss": 0.7179,
      "step": 1071
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7788415420297166,
      "learning_rate": 1.8721592883804577e-05,
      "loss": 0.7668,
      "step": 1072
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7002275265438141,
      "learning_rate": 1.8718812709623807e-05,
      "loss": 0.735,
      "step": 1073
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7866693478140266,
      "learning_rate": 1.871602972257236e-05,
      "loss": 0.7653,
      "step": 1074
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7045143041547405,
      "learning_rate": 1.871324392354809e-05,
      "loss": 0.7204,
      "step": 1075
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7176371337372711,
      "learning_rate": 1.871045531344975e-05,
      "loss": 0.7071,
      "step": 1076
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6704843849536632,
      "learning_rate": 1.8707663893177008e-05,
      "loss": 0.6973,
      "step": 1077
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8169559515511249,
      "learning_rate": 1.870486966363043e-05,
      "loss": 0.762,
      "step": 1078
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6835049487325052,
      "learning_rate": 1.870207262571149e-05,
      "loss": 0.73,
      "step": 1079
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9193107320053966,
      "learning_rate": 1.8699272780322583e-05,
      "loss": 0.7957,
      "step": 1080
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.791157337891335,
      "learning_rate": 1.8696470128366987e-05,
      "loss": 0.7754,
      "step": 1081
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7713739099587669,
      "learning_rate": 1.8693664670748896e-05,
      "loss": 0.7676,
      "step": 1082
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6775678001624642,
      "learning_rate": 1.8690856408373416e-05,
      "loss": 0.7212,
      "step": 1083
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7635197377519451,
      "learning_rate": 1.8688045342146543e-05,
      "loss": 0.7497,
      "step": 1084
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7261459214377627,
      "learning_rate": 1.8685231472975188e-05,
      "loss": 0.7682,
      "step": 1085
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7492633144516981,
      "learning_rate": 1.868241480176717e-05,
      "loss": 0.7582,
      "step": 1086
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7480154425531788,
      "learning_rate": 1.86795953294312e-05,
      "loss": 0.7322,
      "step": 1087
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7796121705441973,
      "learning_rate": 1.8676773056876897e-05,
      "loss": 0.768,
      "step": 1088
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7138960968376197,
      "learning_rate": 1.867394798501479e-05,
      "loss": 0.7124,
      "step": 1089
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8954399765973846,
      "learning_rate": 1.86711201147563e-05,
      "loss": 0.7162,
      "step": 1090
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7287715745907276,
      "learning_rate": 1.8668289447013763e-05,
      "loss": 0.7853,
      "step": 1091
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6790889970593916,
      "learning_rate": 1.8665455982700408e-05,
      "loss": 0.69,
      "step": 1092
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.72421883581688,
      "learning_rate": 1.866261972273037e-05,
      "loss": 0.7284,
      "step": 1093
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7155023886217237,
      "learning_rate": 1.8659780668018686e-05,
      "loss": 0.7432,
      "step": 1094
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.676722533101432,
      "learning_rate": 1.8656938819481295e-05,
      "loss": 0.674,
      "step": 1095
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7895272299387159,
      "learning_rate": 1.8654094178035035e-05,
      "loss": 0.7536,
      "step": 1096
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7269624511037626,
      "learning_rate": 1.865124674459765e-05,
      "loss": 0.7035,
      "step": 1097
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8212924929874763,
      "learning_rate": 1.8648396520087772e-05,
      "loss": 0.7273,
      "step": 1098
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8670277069807177,
      "learning_rate": 1.8645543505424953e-05,
      "loss": 0.7905,
      "step": 1099
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6688313434095239,
      "learning_rate": 1.8642687701529632e-05,
      "loss": 0.683,
      "step": 1100
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7433830345674834,
      "learning_rate": 1.863982910932315e-05,
      "loss": 0.743,
      "step": 1101
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7086023983734153,
      "learning_rate": 1.8636967729727752e-05,
      "loss": 0.6822,
      "step": 1102
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7920437851347683,
      "learning_rate": 1.8634103563666572e-05,
      "loss": 0.7902,
      "step": 1103
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6780854996028355,
      "learning_rate": 1.8631236612063654e-05,
      "loss": 0.7543,
      "step": 1104
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7403138659684492,
      "learning_rate": 1.8628366875843932e-05,
      "loss": 0.7662,
      "step": 1105
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7966639567611423,
      "learning_rate": 1.862549435593325e-05,
      "loss": 0.7237,
      "step": 1106
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7267141062375568,
      "learning_rate": 1.862261905325834e-05,
      "loss": 0.7669,
      "step": 1107
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8044752044740007,
      "learning_rate": 1.861974096874683e-05,
      "loss": 0.7442,
      "step": 1108
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7653869278481256,
      "learning_rate": 1.8616860103327257e-05,
      "loss": 0.7685,
      "step": 1109
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7153680352220333,
      "learning_rate": 1.8613976457929043e-05,
      "loss": 0.7186,
      "step": 1110
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.784961117091671,
      "learning_rate": 1.8611090033482516e-05,
      "loss": 0.7517,
      "step": 1111
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8872916770530554,
      "learning_rate": 1.860820083091889e-05,
      "loss": 0.7416,
      "step": 1112
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7724169899398557,
      "learning_rate": 1.8605308851170288e-05,
      "loss": 0.7431,
      "step": 1113
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.865956376166026,
      "learning_rate": 1.8602414095169723e-05,
      "loss": 0.7552,
      "step": 1114
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5111325096346635,
      "learning_rate": 1.85995165638511e-05,
      "loss": 0.7324,
      "step": 1115
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8000503097362549,
      "learning_rate": 1.8596616258149226e-05,
      "loss": 0.7333,
      "step": 1116
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.79382718885793,
      "learning_rate": 1.85937131789998e-05,
      "loss": 0.7751,
      "step": 1117
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7975521406987497,
      "learning_rate": 1.8590807327339414e-05,
      "loss": 0.7356,
      "step": 1118
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7356012933174411,
      "learning_rate": 1.8587898704105556e-05,
      "loss": 0.7393,
      "step": 1119
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9600454055332855,
      "learning_rate": 1.8584987310236612e-05,
      "loss": 0.7722,
      "step": 1120
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9138688064164845,
      "learning_rate": 1.8582073146671855e-05,
      "loss": 0.8032,
      "step": 1121
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8574357712093279,
      "learning_rate": 1.8579156214351458e-05,
      "loss": 0.7711,
      "step": 1122
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7506882538445572,
      "learning_rate": 1.8576236514216476e-05,
      "loss": 0.7749,
      "step": 1123
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8058908461813106,
      "learning_rate": 1.857331404720888e-05,
      "loss": 0.7361,
      "step": 1124
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7750591484505364,
      "learning_rate": 1.8570388814271503e-05,
      "loss": 0.7341,
      "step": 1125
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7548345996859028,
      "learning_rate": 1.8567460816348097e-05,
      "loss": 0.7628,
      "step": 1126
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7049993230135037,
      "learning_rate": 1.8564530054383286e-05,
      "loss": 0.7045,
      "step": 1127
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7534955822899544,
      "learning_rate": 1.8561596529322603e-05,
      "loss": 0.7682,
      "step": 1128
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8468395611674842,
      "learning_rate": 1.855866024211246e-05,
      "loss": 0.7359,
      "step": 1129
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7410208670381517,
      "learning_rate": 1.8555721193700172e-05,
      "loss": 0.725,
      "step": 1130
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9523654290676339,
      "learning_rate": 1.8552779385033926e-05,
      "loss": 0.8114,
      "step": 1131
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9090904513328822,
      "learning_rate": 1.8549834817062814e-05,
      "loss": 0.7244,
      "step": 1132
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.742170646651039,
      "learning_rate": 1.8546887490736823e-05,
      "loss": 0.7663,
      "step": 1133
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7670274170420512,
      "learning_rate": 1.854393740700681e-05,
      "loss": 0.7728,
      "step": 1134
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7525763697658396,
      "learning_rate": 1.8540984566824542e-05,
      "loss": 0.7264,
      "step": 1135
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8325851286588626,
      "learning_rate": 1.8538028971142662e-05,
      "loss": 0.7504,
      "step": 1136
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7178448566967377,
      "learning_rate": 1.8535070620914713e-05,
      "loss": 0.6925,
      "step": 1137
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8167619808334763,
      "learning_rate": 1.853210951709511e-05,
      "loss": 0.7542,
      "step": 1138
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7675766657035403,
      "learning_rate": 1.8529145660639178e-05,
      "loss": 0.7534,
      "step": 1139
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7520300964094615,
      "learning_rate": 1.8526179052503112e-05,
      "loss": 0.762,
      "step": 1140
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7600231379061789,
      "learning_rate": 1.8523209693644002e-05,
      "loss": 0.7255,
      "step": 1141
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7922333456566035,
      "learning_rate": 1.852023758501983e-05,
      "loss": 0.7528,
      "step": 1142
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6610261135108605,
      "learning_rate": 1.8517262727589456e-05,
      "loss": 0.7104,
      "step": 1143
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.739445347299883,
      "learning_rate": 1.8514285122312627e-05,
      "loss": 0.7722,
      "step": 1144
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7562998344550865,
      "learning_rate": 1.851130477014999e-05,
      "loss": 0.7777,
      "step": 1145
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7957853276525217,
      "learning_rate": 1.8508321672063063e-05,
      "loss": 0.7827,
      "step": 1146
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8236704432229167,
      "learning_rate": 1.8505335829014255e-05,
      "loss": 0.7696,
      "step": 1147
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6860995542674312,
      "learning_rate": 1.8502347241966867e-05,
      "loss": 0.726,
      "step": 1148
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6864653913119477,
      "learning_rate": 1.849935591188507e-05,
      "loss": 0.7276,
      "step": 1149
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7427981296847243,
      "learning_rate": 1.8496361839733937e-05,
      "loss": 0.7865,
      "step": 1150
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8146723224369857,
      "learning_rate": 1.8493365026479418e-05,
      "loss": 0.7273,
      "step": 1151
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7931644103464452,
      "learning_rate": 1.8490365473088344e-05,
      "loss": 0.7112,
      "step": 1152
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.719571154541641,
      "learning_rate": 1.848736318052843e-05,
      "loss": 0.728,
      "step": 1153
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7639824329665406,
      "learning_rate": 1.8484358149768287e-05,
      "loss": 0.7202,
      "step": 1154
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8393648200975895,
      "learning_rate": 1.8481350381777396e-05,
      "loss": 0.7926,
      "step": 1155
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8551869889945061,
      "learning_rate": 1.8478339877526122e-05,
      "loss": 0.7945,
      "step": 1156
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7841890578224007,
      "learning_rate": 1.8475326637985724e-05,
      "loss": 0.7554,
      "step": 1157
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8356327319823276,
      "learning_rate": 1.8472310664128332e-05,
      "loss": 0.7564,
      "step": 1158
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.700222506613228,
      "learning_rate": 1.8469291956926963e-05,
      "loss": 0.733,
      "step": 1159
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7847504540165661,
      "learning_rate": 1.8466270517355508e-05,
      "loss": 0.7226,
      "step": 1160
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7415664534788005,
      "learning_rate": 1.846324634638876e-05,
      "loss": 0.6755,
      "step": 1161
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.838838142120453,
      "learning_rate": 1.8460219445002367e-05,
      "loss": 0.8006,
      "step": 1162
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9562356493422661,
      "learning_rate": 1.8457189814172877e-05,
      "loss": 0.8256,
      "step": 1163
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7547285365755554,
      "learning_rate": 1.8454157454877712e-05,
      "loss": 0.7198,
      "step": 1164
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.860736105722227,
      "learning_rate": 1.8451122368095173e-05,
      "loss": 0.7562,
      "step": 1165
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0339469658980582,
      "learning_rate": 1.844808455480444e-05,
      "loss": 0.7283,
      "step": 1166
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8365158191052576,
      "learning_rate": 1.8445044015985576e-05,
      "loss": 0.7544,
      "step": 1167
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7373342235515724,
      "learning_rate": 1.8442000752619525e-05,
      "loss": 0.7384,
      "step": 1168
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8023254237608446,
      "learning_rate": 1.8438954765688103e-05,
      "loss": 0.7475,
      "step": 1169
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7460343740361145,
      "learning_rate": 1.8435906056174016e-05,
      "loss": 0.7566,
      "step": 1170
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7281426639780573,
      "learning_rate": 1.8432854625060833e-05,
      "loss": 0.7893,
      "step": 1171
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7692775259693888,
      "learning_rate": 1.842980047333301e-05,
      "loss": 0.7277,
      "step": 1172
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.73906504565724,
      "learning_rate": 1.8426743601975885e-05,
      "loss": 0.7329,
      "step": 1173
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.663705456260769,
      "learning_rate": 1.8423684011975668e-05,
      "loss": 0.6882,
      "step": 1174
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7490089205498403,
      "learning_rate": 1.842062170431944e-05,
      "loss": 0.7155,
      "step": 1175
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6991318873470129,
      "learning_rate": 1.8417556679995175e-05,
      "loss": 0.7309,
      "step": 1176
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9019826167082213,
      "learning_rate": 1.8414488939991705e-05,
      "loss": 0.7056,
      "step": 1177
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7293755990544968,
      "learning_rate": 1.8411418485298752e-05,
      "loss": 0.757,
      "step": 1178
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.73572001045144,
      "learning_rate": 1.8408345316906906e-05,
      "loss": 0.7124,
      "step": 1179
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8321679851581225,
      "learning_rate": 1.8405269435807637e-05,
      "loss": 0.8055,
      "step": 1180
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8579398581447657,
      "learning_rate": 1.8402190842993285e-05,
      "loss": 0.7264,
      "step": 1181
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7649053687704958,
      "learning_rate": 1.8399109539457072e-05,
      "loss": 0.7407,
      "step": 1182
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6603645165421631,
      "learning_rate": 1.8396025526193092e-05,
      "loss": 0.6866,
      "step": 1183
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7508880388119596,
      "learning_rate": 1.8392938804196308e-05,
      "loss": 0.8138,
      "step": 1184
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6841239052541156,
      "learning_rate": 1.8389849374462558e-05,
      "loss": 0.7498,
      "step": 1185
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8019286178727234,
      "learning_rate": 1.8386757237988568e-05,
      "loss": 0.7763,
      "step": 1186
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7142250087634718,
      "learning_rate": 1.8383662395771914e-05,
      "loss": 0.7828,
      "step": 1187
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7451091230936435,
      "learning_rate": 1.838056484881106e-05,
      "loss": 0.7623,
      "step": 1188
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7423323690566539,
      "learning_rate": 1.8377464598105342e-05,
      "loss": 0.734,
      "step": 1189
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7911803095528491,
      "learning_rate": 1.8374361644654965e-05,
      "loss": 0.7522,
      "step": 1190
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7021884025086534,
      "learning_rate": 1.8371255989461e-05,
      "loss": 0.7051,
      "step": 1191
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6978594144254934,
      "learning_rate": 1.8368147633525404e-05,
      "loss": 0.7576,
      "step": 1192
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8443328888376684,
      "learning_rate": 1.8365036577850996e-05,
      "loss": 0.7272,
      "step": 1193
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7399638278436399,
      "learning_rate": 1.8361922823441465e-05,
      "loss": 0.7155,
      "step": 1194
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8317299455036548,
      "learning_rate": 1.8358806371301372e-05,
      "loss": 0.773,
      "step": 1195
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7335366996803406,
      "learning_rate": 1.8355687222436157e-05,
      "loss": 0.7264,
      "step": 1196
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6532243860511874,
      "learning_rate": 1.835256537785211e-05,
      "loss": 0.7334,
      "step": 1197
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7434833523532612,
      "learning_rate": 1.8349440838556414e-05,
      "loss": 0.7745,
      "step": 1198
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7836879233659609,
      "learning_rate": 1.8346313605557105e-05,
      "loss": 0.6887,
      "step": 1199
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7479406901686436,
      "learning_rate": 1.8343183679863095e-05,
      "loss": 0.7431,
      "step": 1200
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6957101022413071,
      "learning_rate": 1.8340051062484164e-05,
      "loss": 0.7502,
      "step": 1201
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9097577767285466,
      "learning_rate": 1.8336915754430957e-05,
      "loss": 0.7794,
      "step": 1202
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8404625268273798,
      "learning_rate": 1.833377775671499e-05,
      "loss": 0.8093,
      "step": 1203
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7254139083492468,
      "learning_rate": 1.8330637070348653e-05,
      "loss": 0.7156,
      "step": 1204
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8504067950454907,
      "learning_rate": 1.8327493696345187e-05,
      "loss": 0.7054,
      "step": 1205
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8155655766315796,
      "learning_rate": 1.832434763571872e-05,
      "loss": 0.8113,
      "step": 1206
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8158315188993155,
      "learning_rate": 1.8321198889484227e-05,
      "loss": 0.746,
      "step": 1207
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7111083972240433,
      "learning_rate": 1.831804745865757e-05,
      "loss": 0.7433,
      "step": 1208
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8803373109402814,
      "learning_rate": 1.8314893344255454e-05,
      "loss": 0.7501,
      "step": 1209
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7776113912372178,
      "learning_rate": 1.831173654729547e-05,
      "loss": 0.7531,
      "step": 1210
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6929440798891552,
      "learning_rate": 1.8308577068796067e-05,
      "loss": 0.7107,
      "step": 1211
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7199069168372261,
      "learning_rate": 1.8305414909776548e-05,
      "loss": 0.757,
      "step": 1212
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6829538620226446,
      "learning_rate": 1.8302250071257107e-05,
      "loss": 0.7262,
      "step": 1213
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8213791813083442,
      "learning_rate": 1.8299082554258777e-05,
      "loss": 0.7735,
      "step": 1214
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.724535308994745,
      "learning_rate": 1.8295912359803467e-05,
      "loss": 0.7113,
      "step": 1215
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7992369518707358,
      "learning_rate": 1.8292739488913948e-05,
      "loss": 0.7807,
      "step": 1216
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7282972294964696,
      "learning_rate": 1.8289563942613857e-05,
      "loss": 0.7706,
      "step": 1217
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6611005327968683,
      "learning_rate": 1.8286385721927686e-05,
      "loss": 0.6926,
      "step": 1218
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7397523476002679,
      "learning_rate": 1.82832048278808e-05,
      "loss": 0.7367,
      "step": 1219
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7538518709455814,
      "learning_rate": 1.828002126149942e-05,
      "loss": 0.7563,
      "step": 1220
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7669858847343568,
      "learning_rate": 1.8276835023810625e-05,
      "loss": 0.7093,
      "step": 1221
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7587673947430758,
      "learning_rate": 1.8273646115842374e-05,
      "loss": 0.7583,
      "step": 1222
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6848040767252197,
      "learning_rate": 1.8270454538623465e-05,
      "loss": 0.7513,
      "step": 1223
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7194737494867387,
      "learning_rate": 1.8267260293183573e-05,
      "loss": 0.7085,
      "step": 1224
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7182178069618557,
      "learning_rate": 1.8264063380553227e-05,
      "loss": 0.7433,
      "step": 1225
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7264556909933896,
      "learning_rate": 1.8260863801763812e-05,
      "loss": 0.7689,
      "step": 1226
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.705776808918832,
      "learning_rate": 1.8257661557847588e-05,
      "loss": 0.7526,
      "step": 1227
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7692923025587816,
      "learning_rate": 1.825445664983766e-05,
      "loss": 0.7454,
      "step": 1228
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7301198095985111,
      "learning_rate": 1.8251249078767996e-05,
      "loss": 0.7407,
      "step": 1229
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6549011105365348,
      "learning_rate": 1.824803884567343e-05,
      "loss": 0.7444,
      "step": 1230
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8004584411844881,
      "learning_rate": 1.8244825951589648e-05,
      "loss": 0.7617,
      "step": 1231
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7087208375381548,
      "learning_rate": 1.82416103975532e-05,
      "loss": 0.7072,
      "step": 1232
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7180301320674589,
      "learning_rate": 1.8238392184601483e-05,
      "loss": 0.7287,
      "step": 1233
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7651916536351687,
      "learning_rate": 1.823517131377277e-05,
      "loss": 0.7765,
      "step": 1234
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7285825654059312,
      "learning_rate": 1.823194778610617e-05,
      "loss": 0.7108,
      "step": 1235
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8770598659240827,
      "learning_rate": 1.822872160264167e-05,
      "loss": 0.7912,
      "step": 1236
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9252546127532932,
      "learning_rate": 1.8225492764420098e-05,
      "loss": 0.7516,
      "step": 1237
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7622263116740613,
      "learning_rate": 1.8222261272483143e-05,
      "loss": 0.7117,
      "step": 1238
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7626335308570392,
      "learning_rate": 1.821902712787336e-05,
      "loss": 0.732,
      "step": 1239
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2319612251373464,
      "learning_rate": 1.8215790331634143e-05,
      "loss": 0.7203,
      "step": 1240
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8015420185275878,
      "learning_rate": 1.8212550884809755e-05,
      "loss": 0.7267,
      "step": 1241
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7937391439948012,
      "learning_rate": 1.8209308788445305e-05,
      "loss": 0.757,
      "step": 1242
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8503042426536581,
      "learning_rate": 1.8206064043586764e-05,
      "loss": 0.7347,
      "step": 1243
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7521065848829939,
      "learning_rate": 1.8202816651280955e-05,
      "loss": 0.7715,
      "step": 1244
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7950547015282948,
      "learning_rate": 1.819956661257555e-05,
      "loss": 0.7483,
      "step": 1245
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7604698733764718,
      "learning_rate": 1.819631392851908e-05,
      "loss": 0.756,
      "step": 1246
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7994890274825242,
      "learning_rate": 1.8193058600160935e-05,
      "loss": 0.7771,
      "step": 1247
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.73204016561875,
      "learning_rate": 1.818980062855135e-05,
      "loss": 0.7197,
      "step": 1248
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8038335427677767,
      "learning_rate": 1.8186540014741405e-05,
      "loss": 0.7859,
      "step": 1249
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8002415011656815,
      "learning_rate": 1.8183276759783055e-05,
      "loss": 0.7071,
      "step": 1250
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6975202112031618,
      "learning_rate": 1.8180010864729084e-05,
      "loss": 0.7282,
      "step": 1251
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8111671126622318,
      "learning_rate": 1.8176742330633144e-05,
      "loss": 0.7889,
      "step": 1252
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8763200607581269,
      "learning_rate": 1.8173471158549726e-05,
      "loss": 0.7584,
      "step": 1253
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7654476033783075,
      "learning_rate": 1.8170197349534186e-05,
      "loss": 0.7011,
      "step": 1254
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6847552826540045,
      "learning_rate": 1.816692090464272e-05,
      "loss": 0.7167,
      "step": 1255
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7578340230680185,
      "learning_rate": 1.8163641824932376e-05,
      "loss": 0.733,
      "step": 1256
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8924980122387501,
      "learning_rate": 1.816036011146105e-05,
      "loss": 0.8162,
      "step": 1257
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7336200928274452,
      "learning_rate": 1.81570757652875e-05,
      "loss": 0.7426,
      "step": 1258
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7612483759890136,
      "learning_rate": 1.815378878747132e-05,
      "loss": 0.6903,
      "step": 1259
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7468983068022509,
      "learning_rate": 1.8150499179072952e-05,
      "loss": 0.7269,
      "step": 1260
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7472558044497042,
      "learning_rate": 1.8147206941153704e-05,
      "loss": 0.7907,
      "step": 1261
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7440712023076372,
      "learning_rate": 1.814391207477571e-05,
      "loss": 0.761,
      "step": 1262
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7602263049915268,
      "learning_rate": 1.814061458100197e-05,
      "loss": 0.7293,
      "step": 1263
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.72859921136607,
      "learning_rate": 1.8137314460896317e-05,
      "loss": 0.7471,
      "step": 1264
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7794013856925844,
      "learning_rate": 1.813401171552345e-05,
      "loss": 0.7744,
      "step": 1265
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7243632209440674,
      "learning_rate": 1.8130706345948888e-05,
      "loss": 0.7138,
      "step": 1266
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8673439268302481,
      "learning_rate": 1.812739835323902e-05,
      "loss": 0.7643,
      "step": 1267
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6616654129079438,
      "learning_rate": 1.8124087738461076e-05,
      "loss": 0.7157,
      "step": 1268
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7909280323975353,
      "learning_rate": 1.812077450268313e-05,
      "loss": 0.7532,
      "step": 1269
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.735728087949736,
      "learning_rate": 1.811745864697409e-05,
      "loss": 0.6987,
      "step": 1270
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8179221824382982,
      "learning_rate": 1.8114140172403734e-05,
      "loss": 0.785,
      "step": 1271
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7006981695040264,
      "learning_rate": 1.8110819080042663e-05,
      "loss": 0.7652,
      "step": 1272
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6795942924457778,
      "learning_rate": 1.810749537096233e-05,
      "loss": 0.6811,
      "step": 1273
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6453429426400221,
      "learning_rate": 1.810416904623504e-05,
      "loss": 0.7228,
      "step": 1274
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7851020832157586,
      "learning_rate": 1.810084010693393e-05,
      "loss": 0.7247,
      "step": 1275
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7329777615705523,
      "learning_rate": 1.809750855413298e-05,
      "loss": 0.7096,
      "step": 1276
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8210209483941616,
      "learning_rate": 1.8094174388907025e-05,
      "loss": 0.7493,
      "step": 1277
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7852878181391496,
      "learning_rate": 1.8090837612331736e-05,
      "loss": 0.735,
      "step": 1278
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6872835087165584,
      "learning_rate": 1.8087498225483627e-05,
      "loss": 0.7333,
      "step": 1279
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7956201815298437,
      "learning_rate": 1.8084156229440048e-05,
      "loss": 0.7531,
      "step": 1280
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8577791636942308,
      "learning_rate": 1.80808116252792e-05,
      "loss": 0.7422,
      "step": 1281
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8106992600665621,
      "learning_rate": 1.807746441408013e-05,
      "loss": 0.7821,
      "step": 1282
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7465904036540154,
      "learning_rate": 1.8074114596922704e-05,
      "loss": 0.7524,
      "step": 1283
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7612898461424514,
      "learning_rate": 1.8070762174887654e-05,
      "loss": 0.6547,
      "step": 1284
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7768227559907114,
      "learning_rate": 1.8067407149056533e-05,
      "loss": 0.7781,
      "step": 1285
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6791021681202012,
      "learning_rate": 1.8064049520511747e-05,
      "loss": 0.7218,
      "step": 1286
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8006722459955136,
      "learning_rate": 1.8060689290336536e-05,
      "loss": 0.7326,
      "step": 1287
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8476169162885574,
      "learning_rate": 1.805732645961498e-05,
      "loss": 0.7968,
      "step": 1288
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7056573209214967,
      "learning_rate": 1.8053961029432e-05,
      "loss": 0.7386,
      "step": 1289
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6832871326025702,
      "learning_rate": 1.8050593000873352e-05,
      "loss": 0.7089,
      "step": 1290
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7628705195109529,
      "learning_rate": 1.8047222375025632e-05,
      "loss": 0.7151,
      "step": 1291
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7262441986219392,
      "learning_rate": 1.8043849152976277e-05,
      "loss": 0.7346,
      "step": 1292
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7544242084154198,
      "learning_rate": 1.8040473335813555e-05,
      "loss": 0.7144,
      "step": 1293
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6441637814363119,
      "learning_rate": 1.8037094924626573e-05,
      "loss": 0.6749,
      "step": 1294
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7478308657903865,
      "learning_rate": 1.8033713920505285e-05,
      "loss": 0.7286,
      "step": 1295
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7624574060661394,
      "learning_rate": 1.8030330324540468e-05,
      "loss": 0.7545,
      "step": 1296
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6973477203656536,
      "learning_rate": 1.8026944137823742e-05,
      "loss": 0.7022,
      "step": 1297
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7851650339707589,
      "learning_rate": 1.8023555361447558e-05,
      "loss": 0.7729,
      "step": 1298
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7874601140473092,
      "learning_rate": 1.802016399650521e-05,
      "loss": 0.7752,
      "step": 1299
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7100548619818904,
      "learning_rate": 1.801677004409082e-05,
      "loss": 0.7331,
      "step": 1300
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7342342824177541,
      "learning_rate": 1.8013373505299353e-05,
      "loss": 0.7697,
      "step": 1301
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8143582657752351,
      "learning_rate": 1.8009974381226596e-05,
      "loss": 0.7222,
      "step": 1302
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8876348759525704,
      "learning_rate": 1.8006572672969182e-05,
      "loss": 0.771,
      "step": 1303
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8803522770625064,
      "learning_rate": 1.8003168381624572e-05,
      "loss": 0.7658,
      "step": 1304
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7432589768071872,
      "learning_rate": 1.799976150829106e-05,
      "loss": 0.7577,
      "step": 1305
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.676948216048753,
      "learning_rate": 1.7996352054067778e-05,
      "loss": 0.735,
      "step": 1306
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7193076482738695,
      "learning_rate": 1.7992940020054682e-05,
      "loss": 0.7173,
      "step": 1307
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8086535623434694,
      "learning_rate": 1.7989525407352567e-05,
      "loss": 0.7124,
      "step": 1308
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8369103815286871,
      "learning_rate": 1.7986108217063064e-05,
      "loss": 0.8428,
      "step": 1309
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6925851777058056,
      "learning_rate": 1.798268845028862e-05,
      "loss": 0.7408,
      "step": 1310
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.73359677106135,
      "learning_rate": 1.797926610813253e-05,
      "loss": 0.7607,
      "step": 1311
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.708032767663543,
      "learning_rate": 1.7975841191698908e-05,
      "loss": 0.7291,
      "step": 1312
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7569330135053018,
      "learning_rate": 1.7972413702092708e-05,
      "loss": 0.7268,
      "step": 1313
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8627782538004081,
      "learning_rate": 1.7968983640419708e-05,
      "loss": 0.7255,
      "step": 1314
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7442692027232115,
      "learning_rate": 1.7965551007786515e-05,
      "loss": 0.7137,
      "step": 1315
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8159752903584095,
      "learning_rate": 1.7962115805300572e-05,
      "loss": 0.7682,
      "step": 1316
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.778330746358816,
      "learning_rate": 1.7958678034070143e-05,
      "loss": 0.7724,
      "step": 1317
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7670129919634199,
      "learning_rate": 1.795523769520433e-05,
      "loss": 0.7078,
      "step": 1318
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8445475111875871,
      "learning_rate": 1.795179478981305e-05,
      "loss": 0.7877,
      "step": 1319
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7180548074022983,
      "learning_rate": 1.7948349319007065e-05,
      "loss": 0.7286,
      "step": 1320
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6874728123523715,
      "learning_rate": 1.7944901283897947e-05,
      "loss": 0.7044,
      "step": 1321
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8703204213937699,
      "learning_rate": 1.7941450685598108e-05,
      "loss": 0.7332,
      "step": 1322
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7154973458444274,
      "learning_rate": 1.7937997525220785e-05,
      "loss": 0.7101,
      "step": 1323
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.702549518142173,
      "learning_rate": 1.7934541803880035e-05,
      "loss": 0.746,
      "step": 1324
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7457648322831936,
      "learning_rate": 1.7931083522690752e-05,
      "loss": 0.6896,
      "step": 1325
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.638618156794349,
      "learning_rate": 1.7927622682768645e-05,
      "loss": 0.6674,
      "step": 1326
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7668558452203226,
      "learning_rate": 1.792415928523025e-05,
      "loss": 0.787,
      "step": 1327
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7214678759960359,
      "learning_rate": 1.792069333119294e-05,
      "loss": 0.7492,
      "step": 1328
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8973117234821939,
      "learning_rate": 1.7917224821774897e-05,
      "loss": 0.8509,
      "step": 1329
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8711969361929509,
      "learning_rate": 1.7913753758095138e-05,
      "loss": 0.7517,
      "step": 1330
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6716969978518681,
      "learning_rate": 1.7910280141273494e-05,
      "loss": 0.7126,
      "step": 1331
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6532971788014965,
      "learning_rate": 1.7906803972430636e-05,
      "loss": 0.7351,
      "step": 1332
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7532776857812551,
      "learning_rate": 1.790332525268804e-05,
      "loss": 0.7166,
      "step": 1333
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6057226564298777,
      "learning_rate": 1.789984398316802e-05,
      "loss": 0.6843,
      "step": 1334
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.860729529505972,
      "learning_rate": 1.7896360164993705e-05,
      "loss": 0.8262,
      "step": 1335
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8081331136515698,
      "learning_rate": 1.7892873799289043e-05,
      "loss": 0.7786,
      "step": 1336
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8619067158805226,
      "learning_rate": 1.788938488717881e-05,
      "loss": 0.7036,
      "step": 1337
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8073997590577728,
      "learning_rate": 1.7885893429788605e-05,
      "loss": 0.7463,
      "step": 1338
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6868919022058878,
      "learning_rate": 1.7882399428244843e-05,
      "loss": 0.7343,
      "step": 1339
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8163188637738732,
      "learning_rate": 1.7878902883674757e-05,
      "loss": 0.6951,
      "step": 1340
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6872713967586096,
      "learning_rate": 1.7875403797206414e-05,
      "loss": 0.7147,
      "step": 1341
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7784020422528409,
      "learning_rate": 1.7871902169968683e-05,
      "loss": 0.6985,
      "step": 1342
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8416105288983569,
      "learning_rate": 1.786839800309127e-05,
      "loss": 0.7378,
      "step": 1343
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6592747520206479,
      "learning_rate": 1.786489129770469e-05,
      "loss": 0.71,
      "step": 1344
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8334875104087311,
      "learning_rate": 1.7861382054940274e-05,
      "loss": 0.7458,
      "step": 1345
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8364986903326895,
      "learning_rate": 1.7857870275930183e-05,
      "loss": 0.7799,
      "step": 1346
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8323425589240953,
      "learning_rate": 1.785435596180739e-05,
      "loss": 0.7594,
      "step": 1347
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.786406269725757,
      "learning_rate": 1.785083911370568e-05,
      "loss": 0.7178,
      "step": 1348
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7441814899849001,
      "learning_rate": 1.7847319732759668e-05,
      "loss": 0.7491,
      "step": 1349
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.791867606080257,
      "learning_rate": 1.7843797820104772e-05,
      "loss": 0.7014,
      "step": 1350
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7772383184903296,
      "learning_rate": 1.7840273376877246e-05,
      "loss": 0.7509,
      "step": 1351
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8263562164278344,
      "learning_rate": 1.7836746404214142e-05,
      "loss": 0.7514,
      "step": 1352
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8188149967432777,
      "learning_rate": 1.783321690325333e-05,
      "loss": 0.7735,
      "step": 1353
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7462170264569716,
      "learning_rate": 1.7829684875133512e-05,
      "loss": 0.6955,
      "step": 1354
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7023853277446592,
      "learning_rate": 1.7826150320994183e-05,
      "loss": 0.6945,
      "step": 1355
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7036000313166134,
      "learning_rate": 1.782261324197567e-05,
      "loss": 0.7472,
      "step": 1356
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7783184652440365,
      "learning_rate": 1.7819073639219105e-05,
      "loss": 0.7782,
      "step": 1357
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8159968185967744,
      "learning_rate": 1.781553151386644e-05,
      "loss": 0.7714,
      "step": 1358
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0762621694906045,
      "learning_rate": 1.781198686706044e-05,
      "loss": 0.759,
      "step": 1359
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7880196513435874,
      "learning_rate": 1.780843969994468e-05,
      "loss": 0.7735,
      "step": 1360
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.696004489591867,
      "learning_rate": 1.7804890013663545e-05,
      "loss": 0.7255,
      "step": 1361
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7149908158700372,
      "learning_rate": 1.7801337809362248e-05,
      "loss": 0.7309,
      "step": 1362
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7246305802818837,
      "learning_rate": 1.7797783088186797e-05,
      "loss": 0.7262,
      "step": 1363
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6650281131988163,
      "learning_rate": 1.779422585128402e-05,
      "loss": 0.6894,
      "step": 1364
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7592314954204435,
      "learning_rate": 1.7790666099801558e-05,
      "loss": 0.7394,
      "step": 1365
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7906348464038496,
      "learning_rate": 1.7787103834887856e-05,
      "loss": 0.7616,
      "step": 1366
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9259755791647426,
      "learning_rate": 1.7783539057692187e-05,
      "loss": 0.8053,
      "step": 1367
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7719806815437071,
      "learning_rate": 1.7779971769364603e-05,
      "loss": 0.7691,
      "step": 1368
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7435901382467853,
      "learning_rate": 1.7776401971056004e-05,
      "loss": 0.7448,
      "step": 1369
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7765686569657504,
      "learning_rate": 1.777282966391807e-05,
      "loss": 0.7525,
      "step": 1370
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7673111669109495,
      "learning_rate": 1.776925484910331e-05,
      "loss": 0.7358,
      "step": 1371
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7551828757526,
      "learning_rate": 1.7765677527765025e-05,
      "loss": 0.7237,
      "step": 1372
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8605191428215649,
      "learning_rate": 1.7762097701057338e-05,
      "loss": 0.7997,
      "step": 1373
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8449519260512914,
      "learning_rate": 1.7758515370135177e-05,
      "loss": 0.7754,
      "step": 1374
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7211630251945444,
      "learning_rate": 1.7754930536154275e-05,
      "loss": 0.7202,
      "step": 1375
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8668652903356483,
      "learning_rate": 1.7751343200271174e-05,
      "loss": 0.8472,
      "step": 1376
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6951045242773715,
      "learning_rate": 1.7747753363643226e-05,
      "loss": 0.7363,
      "step": 1377
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8142740569334616,
      "learning_rate": 1.774416102742858e-05,
      "loss": 0.7678,
      "step": 1378
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7067243069068384,
      "learning_rate": 1.7740566192786208e-05,
      "loss": 0.7614,
      "step": 1379
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.685028911861874,
      "learning_rate": 1.7736968860875867e-05,
      "loss": 0.7317,
      "step": 1380
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7856902872162957,
      "learning_rate": 1.7733369032858143e-05,
      "loss": 0.7631,
      "step": 1381
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.642469927060816,
      "learning_rate": 1.7729766709894408e-05,
      "loss": 0.6801,
      "step": 1382
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7793534204073017,
      "learning_rate": 1.772616189314685e-05,
      "loss": 0.7475,
      "step": 1383
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8192174351740954,
      "learning_rate": 1.772255458377846e-05,
      "loss": 0.7736,
      "step": 1384
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6508202834454628,
      "learning_rate": 1.7718944782953025e-05,
      "loss": 0.688,
      "step": 1385
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7174294367253088,
      "learning_rate": 1.7715332491835146e-05,
      "loss": 0.6965,
      "step": 1386
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7335128139729328,
      "learning_rate": 1.771171771159022e-05,
      "loss": 0.7285,
      "step": 1387
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7518910787841253,
      "learning_rate": 1.7708100443384455e-05,
      "loss": 0.7526,
      "step": 1388
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7973656881399546,
      "learning_rate": 1.7704480688384854e-05,
      "loss": 0.7715,
      "step": 1389
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6532681164566032,
      "learning_rate": 1.7700858447759228e-05,
      "loss": 0.6964,
      "step": 1390
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8504845256564982,
      "learning_rate": 1.7697233722676185e-05,
      "loss": 0.7305,
      "step": 1391
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.716412340130392,
      "learning_rate": 1.769360651430514e-05,
      "loss": 0.7369,
      "step": 1392
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7158785448744391,
      "learning_rate": 1.76899768238163e-05,
      "loss": 0.6794,
      "step": 1393
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7737870669286501,
      "learning_rate": 1.7686344652380684e-05,
      "loss": 0.7419,
      "step": 1394
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7432099600317918,
      "learning_rate": 1.76827100011701e-05,
      "loss": 0.7095,
      "step": 1395
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.779994432275517,
      "learning_rate": 1.7679072871357174e-05,
      "loss": 0.6995,
      "step": 1396
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7062657899227921,
      "learning_rate": 1.767543326411531e-05,
      "loss": 0.7347,
      "step": 1397
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.711440750315502,
      "learning_rate": 1.7671791180618722e-05,
      "loss": 0.7066,
      "step": 1398
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.711399017771291,
      "learning_rate": 1.7668146622042426e-05,
      "loss": 0.6788,
      "step": 1399
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6939842121445274,
      "learning_rate": 1.7664499589562227e-05,
      "loss": 0.6925,
      "step": 1400
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6988167337609682,
      "learning_rate": 1.7660850084354743e-05,
      "loss": 0.7288,
      "step": 1401
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.680263999877132,
      "learning_rate": 1.765719810759737e-05,
      "loss": 0.6947,
      "step": 1402
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6634028373507175,
      "learning_rate": 1.7653543660468317e-05,
      "loss": 0.7522,
      "step": 1403
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6758769361931172,
      "learning_rate": 1.7649886744146586e-05,
      "loss": 0.7111,
      "step": 1404
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3084487256320256,
      "learning_rate": 1.764622735981197e-05,
      "loss": 0.7981,
      "step": 1405
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7596811747259102,
      "learning_rate": 1.7642565508645067e-05,
      "loss": 0.7671,
      "step": 1406
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7378567413149598,
      "learning_rate": 1.7638901191827264e-05,
      "loss": 0.7324,
      "step": 1407
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8268960065672961,
      "learning_rate": 1.7635234410540745e-05,
      "loss": 0.8255,
      "step": 1408
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.723060352221315,
      "learning_rate": 1.7631565165968493e-05,
      "loss": 0.7367,
      "step": 1409
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.700112633798048,
      "learning_rate": 1.7627893459294283e-05,
      "loss": 0.7082,
      "step": 1410
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7760275506025097,
      "learning_rate": 1.7624219291702682e-05,
      "loss": 0.7784,
      "step": 1411
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8094556514712448,
      "learning_rate": 1.7620542664379054e-05,
      "loss": 0.7596,
      "step": 1412
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8192017153548703,
      "learning_rate": 1.7616863578509553e-05,
      "loss": 0.8516,
      "step": 1413
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7990276515867887,
      "learning_rate": 1.761318203528113e-05,
      "loss": 0.7636,
      "step": 1414
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7694498696659816,
      "learning_rate": 1.760949803588153e-05,
      "loss": 0.6742,
      "step": 1415
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1304965669801186,
      "learning_rate": 1.760581158149929e-05,
      "loss": 0.8138,
      "step": 1416
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8766000289876036,
      "learning_rate": 1.760212267332373e-05,
      "loss": 0.7356,
      "step": 1417
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.68347727879849,
      "learning_rate": 1.759843131254497e-05,
      "loss": 0.7186,
      "step": 1418
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7945192234227846,
      "learning_rate": 1.7594737500353926e-05,
      "loss": 0.7614,
      "step": 1419
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7511923148479042,
      "learning_rate": 1.7591041237942296e-05,
      "loss": 0.6801,
      "step": 1420
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6304698331729877,
      "learning_rate": 1.7587342526502566e-05,
      "loss": 0.7167,
      "step": 1421
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8180611544702946,
      "learning_rate": 1.7583641367228025e-05,
      "loss": 0.7789,
      "step": 1422
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5676947671667003,
      "learning_rate": 1.7579937761312737e-05,
      "loss": 0.8696,
      "step": 1423
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6760830481925173,
      "learning_rate": 1.757623170995157e-05,
      "loss": 0.7381,
      "step": 1424
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7548844441624366,
      "learning_rate": 1.7572523214340167e-05,
      "loss": 0.7331,
      "step": 1425
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6579158736727312,
      "learning_rate": 1.756881227567497e-05,
      "loss": 0.7348,
      "step": 1426
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6385721005315309,
      "learning_rate": 1.75650988951532e-05,
      "loss": 0.7379,
      "step": 1427
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7333982742965844,
      "learning_rate": 1.7561383073972876e-05,
      "loss": 0.7045,
      "step": 1428
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6935378406121746,
      "learning_rate": 1.75576648133328e-05,
      "loss": 0.722,
      "step": 1429
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7841624143823144,
      "learning_rate": 1.7553944114432557e-05,
      "loss": 0.7872,
      "step": 1430
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.767486433293147,
      "learning_rate": 1.7550220978472526e-05,
      "loss": 0.8257,
      "step": 1431
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5031421649920633,
      "learning_rate": 1.7546495406653862e-05,
      "loss": 0.8719,
      "step": 1432
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7075344210563942,
      "learning_rate": 1.7542767400178514e-05,
      "loss": 0.7474,
      "step": 1433
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8169103130355345,
      "learning_rate": 1.7539036960249222e-05,
      "loss": 0.815,
      "step": 1434
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.744616773273098,
      "learning_rate": 1.7535304088069494e-05,
      "loss": 0.7509,
      "step": 1435
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8361518892010283,
      "learning_rate": 1.7531568784843635e-05,
      "loss": 0.7544,
      "step": 1436
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6717744171383363,
      "learning_rate": 1.7527831051776735e-05,
      "loss": 0.7096,
      "step": 1437
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8507608848407532,
      "learning_rate": 1.7524090890074662e-05,
      "loss": 0.7287,
      "step": 1438
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7874230578052511,
      "learning_rate": 1.752034830094407e-05,
      "loss": 0.7509,
      "step": 1439
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7042124281091713,
      "learning_rate": 1.75166032855924e-05,
      "loss": 0.7067,
      "step": 1440
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6006767592364758,
      "learning_rate": 1.7512855845227866e-05,
      "loss": 0.7018,
      "step": 1441
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8688799257925469,
      "learning_rate": 1.7509105981059474e-05,
      "loss": 0.7065,
      "step": 1442
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7316939076536134,
      "learning_rate": 1.7505353694297005e-05,
      "loss": 0.7569,
      "step": 1443
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0070711470836196,
      "learning_rate": 1.750159898615103e-05,
      "loss": 0.7549,
      "step": 1444
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6515367006280961,
      "learning_rate": 1.7497841857832894e-05,
      "loss": 0.7347,
      "step": 1445
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.729092086694196,
      "learning_rate": 1.749408231055472e-05,
      "loss": 0.7665,
      "step": 1446
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7218456259209016,
      "learning_rate": 1.7490320345529424e-05,
      "loss": 0.7365,
      "step": 1447
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7305659888375168,
      "learning_rate": 1.7486555963970694e-05,
      "loss": 0.7652,
      "step": 1448
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7221108838989603,
      "learning_rate": 1.748278916709299e-05,
      "loss": 0.7519,
      "step": 1449
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6397289545699121,
      "learning_rate": 1.747901995611157e-05,
      "loss": 0.6791,
      "step": 1450
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7289007090050836,
      "learning_rate": 1.7475248332242446e-05,
      "loss": 0.7663,
      "step": 1451
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6686665266815089,
      "learning_rate": 1.7471474296702434e-05,
      "loss": 0.7058,
      "step": 1452
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8615963485012111,
      "learning_rate": 1.7467697850709115e-05,
      "loss": 0.757,
      "step": 1453
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8304739329628903,
      "learning_rate": 1.7463918995480845e-05,
      "loss": 0.7768,
      "step": 1454
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6432408500504139,
      "learning_rate": 1.7460137732236764e-05,
      "loss": 0.7246,
      "step": 1455
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8379590775150785,
      "learning_rate": 1.745635406219679e-05,
      "loss": 0.7184,
      "step": 1456
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6657795757082339,
      "learning_rate": 1.7452567986581604e-05,
      "loss": 0.7116,
      "step": 1457
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6824887887293511,
      "learning_rate": 1.7448779506612683e-05,
      "loss": 0.686,
      "step": 1458
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6970780568658766,
      "learning_rate": 1.7444988623512262e-05,
      "loss": 0.765,
      "step": 1459
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8707025924201838,
      "learning_rate": 1.7441195338503366e-05,
      "loss": 0.8073,
      "step": 1460
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7123686495216246,
      "learning_rate": 1.743739965280978e-05,
      "loss": 0.7148,
      "step": 1461
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9662516661056841,
      "learning_rate": 1.743360156765608e-05,
      "loss": 0.7765,
      "step": 1462
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.636747213059274,
      "learning_rate": 1.7429801084267598e-05,
      "loss": 0.7273,
      "step": 1463
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7353149815865959,
      "learning_rate": 1.7425998203870453e-05,
      "loss": 0.7092,
      "step": 1464
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6784037121940032,
      "learning_rate": 1.742219292769154e-05,
      "loss": 0.691,
      "step": 1465
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7545530837439987,
      "learning_rate": 1.7418385256958508e-05,
      "loss": 0.7191,
      "step": 1466
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7586222432041834,
      "learning_rate": 1.7414575192899803e-05,
      "loss": 0.7821,
      "step": 1467
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8244516088003997,
      "learning_rate": 1.7410762736744625e-05,
      "loss": 0.7298,
      "step": 1468
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7100044462035451,
      "learning_rate": 1.740694788972295e-05,
      "loss": 0.7201,
      "step": 1469
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7879301108613668,
      "learning_rate": 1.740313065306553e-05,
      "loss": 0.7265,
      "step": 1470
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7961056542810394,
      "learning_rate": 1.7399311028003886e-05,
      "loss": 0.7593,
      "step": 1471
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.743778166660551,
      "learning_rate": 1.739548901577031e-05,
      "loss": 0.76,
      "step": 1472
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6526869723745664,
      "learning_rate": 1.7391664617597858e-05,
      "loss": 0.7815,
      "step": 1473
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6923517147388485,
      "learning_rate": 1.738783783472036e-05,
      "loss": 0.7398,
      "step": 1474
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7575313659472268,
      "learning_rate": 1.738400866837242e-05,
      "loss": 0.7744,
      "step": 1475
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6951544159623877,
      "learning_rate": 1.738017711978941e-05,
      "loss": 0.7542,
      "step": 1476
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6723936270744089,
      "learning_rate": 1.7376343190207458e-05,
      "loss": 0.6883,
      "step": 1477
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7036440872851514,
      "learning_rate": 1.7372506880863477e-05,
      "loss": 0.7161,
      "step": 1478
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6667971966932783,
      "learning_rate": 1.7368668192995133e-05,
      "loss": 0.7484,
      "step": 1479
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7219492508319645,
      "learning_rate": 1.7364827127840874e-05,
      "loss": 0.7351,
      "step": 1480
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7487171665168834,
      "learning_rate": 1.736098368663991e-05,
      "loss": 0.7805,
      "step": 1481
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7247557930953763,
      "learning_rate": 1.7357137870632205e-05,
      "loss": 0.7304,
      "step": 1482
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7402013443662868,
      "learning_rate": 1.7353289681058507e-05,
      "loss": 0.7771,
      "step": 1483
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7127527478159407,
      "learning_rate": 1.734943911916032e-05,
      "loss": 0.7257,
      "step": 1484
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6852785725963915,
      "learning_rate": 1.7345586186179916e-05,
      "loss": 0.7263,
      "step": 1485
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7848548447243251,
      "learning_rate": 1.734173088336033e-05,
      "loss": 0.7196,
      "step": 1486
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7431362557862778,
      "learning_rate": 1.7337873211945365e-05,
      "loss": 0.7107,
      "step": 1487
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7268800605774455,
      "learning_rate": 1.7334013173179582e-05,
      "loss": 0.7431,
      "step": 1488
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7632035959108727,
      "learning_rate": 1.7330150768308315e-05,
      "loss": 0.7436,
      "step": 1489
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6297407699039074,
      "learning_rate": 1.7326285998577654e-05,
      "loss": 0.6915,
      "step": 1490
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7546392128514601,
      "learning_rate": 1.7322418865234457e-05,
      "loss": 0.7496,
      "step": 1491
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7251017285381203,
      "learning_rate": 1.7318549369526337e-05,
      "loss": 0.7714,
      "step": 1492
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7043883304667362,
      "learning_rate": 1.731467751270168e-05,
      "loss": 0.7225,
      "step": 1493
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7621117538423045,
      "learning_rate": 1.731080329600962e-05,
      "loss": 0.7649,
      "step": 1494
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9163464853690335,
      "learning_rate": 1.7306926720700067e-05,
      "loss": 0.778,
      "step": 1495
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7962765373934559,
      "learning_rate": 1.730304778802368e-05,
      "loss": 0.7391,
      "step": 1496
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8583136525221953,
      "learning_rate": 1.7299166499231884e-05,
      "loss": 0.7528,
      "step": 1497
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7206375854505027,
      "learning_rate": 1.7295282855576868e-05,
      "loss": 0.6785,
      "step": 1498
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7542682760335205,
      "learning_rate": 1.7291396858311575e-05,
      "loss": 0.7291,
      "step": 1499
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8541144003587724,
      "learning_rate": 1.7287508508689708e-05,
      "loss": 0.827,
      "step": 1500
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6916554120817306,
      "learning_rate": 1.7283617807965726e-05,
      "loss": 0.6948,
      "step": 1501
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8299235180873263,
      "learning_rate": 1.7279724757394854e-05,
      "loss": 0.7624,
      "step": 1502
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6455443779786671,
      "learning_rate": 1.727582935823307e-05,
      "loss": 0.7226,
      "step": 1503
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7320481035635267,
      "learning_rate": 1.7271931611737113e-05,
      "loss": 0.7222,
      "step": 1504
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7189353796916486,
      "learning_rate": 1.7268031519164474e-05,
      "loss": 0.7393,
      "step": 1505
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7280545859817267,
      "learning_rate": 1.7264129081773407e-05,
      "loss": 0.7347,
      "step": 1506
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7217587999977617,
      "learning_rate": 1.7260224300822918e-05,
      "loss": 0.7353,
      "step": 1507
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7222852385484946,
      "learning_rate": 1.725631717757277e-05,
      "loss": 0.7069,
      "step": 1508
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7801373838885238,
      "learning_rate": 1.7252407713283484e-05,
      "loss": 0.7344,
      "step": 1509
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7483089903681078,
      "learning_rate": 1.7248495909216335e-05,
      "loss": 0.7335,
      "step": 1510
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8668139504178476,
      "learning_rate": 1.724458176663335e-05,
      "loss": 0.7807,
      "step": 1511
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.734562701634865,
      "learning_rate": 1.7240665286797316e-05,
      "loss": 0.6879,
      "step": 1512
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.748912431094992,
      "learning_rate": 1.723674647097177e-05,
      "loss": 0.7597,
      "step": 1513
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7889585383725976,
      "learning_rate": 1.7232825320421e-05,
      "loss": 0.7579,
      "step": 1514
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6857193605390461,
      "learning_rate": 1.7228901836410056e-05,
      "loss": 0.6958,
      "step": 1515
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7357683315130081,
      "learning_rate": 1.7224976020204737e-05,
      "loss": 0.7551,
      "step": 1516
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7650144125312769,
      "learning_rate": 1.7221047873071586e-05,
      "loss": 0.7381,
      "step": 1517
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7420287304762396,
      "learning_rate": 1.721711739627791e-05,
      "loss": 0.7238,
      "step": 1518
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7903512760107481,
      "learning_rate": 1.7213184591091764e-05,
      "loss": 0.7728,
      "step": 1519
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7134402656927645,
      "learning_rate": 1.7209249458781947e-05,
      "loss": 0.733,
      "step": 1520
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8251416033025214,
      "learning_rate": 1.720531200061802e-05,
      "loss": 0.7556,
      "step": 1521
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6209190614076074,
      "learning_rate": 1.7201372217870285e-05,
      "loss": 0.7153,
      "step": 1522
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6286523997128947,
      "learning_rate": 1.7197430111809804e-05,
      "loss": 0.7209,
      "step": 1523
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8024434804836762,
      "learning_rate": 1.719348568370838e-05,
      "loss": 0.7226,
      "step": 1524
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7071049685593067,
      "learning_rate": 1.7189538934838564e-05,
      "loss": 0.7621,
      "step": 1525
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6736551339443139,
      "learning_rate": 1.7185589866473665e-05,
      "loss": 0.6625,
      "step": 1526
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8045672955446157,
      "learning_rate": 1.7181638479887728e-05,
      "loss": 0.8059,
      "step": 1527
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7662848558795228,
      "learning_rate": 1.717768477635556e-05,
      "loss": 0.7643,
      "step": 1528
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6912249897609034,
      "learning_rate": 1.71737287571527e-05,
      "loss": 0.6877,
      "step": 1529
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7271238711369477,
      "learning_rate": 1.716977042355545e-05,
      "loss": 0.734,
      "step": 1530
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7525342754130883,
      "learning_rate": 1.716580977684085e-05,
      "loss": 0.7845,
      "step": 1531
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7540505833186464,
      "learning_rate": 1.7161846818286683e-05,
      "loss": 0.7516,
      "step": 1532
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6680501741589535,
      "learning_rate": 1.7157881549171486e-05,
      "loss": 0.6817,
      "step": 1533
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7404765692461728,
      "learning_rate": 1.715391397077453e-05,
      "loss": 0.6957,
      "step": 1534
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8007240234311871,
      "learning_rate": 1.714994408437585e-05,
      "loss": 0.7321,
      "step": 1535
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7229562884201313,
      "learning_rate": 1.7145971891256204e-05,
      "loss": 0.7371,
      "step": 1536
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6782597653035877,
      "learning_rate": 1.714199739269711e-05,
      "loss": 0.7161,
      "step": 1537
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8553495508070331,
      "learning_rate": 1.7138020589980825e-05,
      "loss": 0.7204,
      "step": 1538
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8306299475507694,
      "learning_rate": 1.7134041484390344e-05,
      "loss": 0.7451,
      "step": 1539
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.890056971865775,
      "learning_rate": 1.713006007720941e-05,
      "loss": 0.7611,
      "step": 1540
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7745538095095827,
      "learning_rate": 1.712607636972251e-05,
      "loss": 0.7664,
      "step": 1541
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6895327327656987,
      "learning_rate": 1.7122090363214866e-05,
      "loss": 0.7138,
      "step": 1542
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.758549585398954,
      "learning_rate": 1.7118102058972457e-05,
      "loss": 0.7206,
      "step": 1543
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6964162002770254,
      "learning_rate": 1.7114111458281985e-05,
      "loss": 0.7063,
      "step": 1544
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.700207282771962,
      "learning_rate": 1.7110118562430904e-05,
      "loss": 0.7376,
      "step": 1545
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7433365429234048,
      "learning_rate": 1.71061233727074e-05,
      "loss": 0.7276,
      "step": 1546
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6464340858684804,
      "learning_rate": 1.7102125890400413e-05,
      "loss": 0.6865,
      "step": 1547
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7665288530040991,
      "learning_rate": 1.709812611679961e-05,
      "loss": 0.7471,
      "step": 1548
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7421583994573777,
      "learning_rate": 1.70941240531954e-05,
      "loss": 0.7584,
      "step": 1549
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8547421715330368,
      "learning_rate": 1.7090119700878933e-05,
      "loss": 0.7331,
      "step": 1550
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8372133047433974,
      "learning_rate": 1.70861130611421e-05,
      "loss": 0.8148,
      "step": 1551
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7226686043268731,
      "learning_rate": 1.708210413527752e-05,
      "loss": 0.7316,
      "step": 1552
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.758660056219762,
      "learning_rate": 1.707809292457856e-05,
      "loss": 0.7266,
      "step": 1553
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7193904993436278,
      "learning_rate": 1.707407943033932e-05,
      "loss": 0.7566,
      "step": 1554
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7581086000331252,
      "learning_rate": 1.7070063653854636e-05,
      "loss": 0.7103,
      "step": 1555
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8081601423926386,
      "learning_rate": 1.706604559642008e-05,
      "loss": 0.7526,
      "step": 1556
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7301041310752289,
      "learning_rate": 1.7062025259331966e-05,
      "loss": 0.7353,
      "step": 1557
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7321380109435941,
      "learning_rate": 1.7058002643887332e-05,
      "loss": 0.7104,
      "step": 1558
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7536206072522272,
      "learning_rate": 1.7053977751383964e-05,
      "loss": 0.7541,
      "step": 1559
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7054130770681657,
      "learning_rate": 1.704995058312037e-05,
      "loss": 0.6941,
      "step": 1560
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7590284871449714,
      "learning_rate": 1.7045921140395802e-05,
      "loss": 0.7389,
      "step": 1561
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7319187593901468,
      "learning_rate": 1.7041889424510243e-05,
      "loss": 0.7325,
      "step": 1562
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6653347004724218,
      "learning_rate": 1.7037855436764406e-05,
      "loss": 0.7346,
      "step": 1563
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6981961953324062,
      "learning_rate": 1.7033819178459743e-05,
      "loss": 0.7235,
      "step": 1564
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7247546511623132,
      "learning_rate": 1.702978065089843e-05,
      "loss": 0.7066,
      "step": 1565
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6918828974743905,
      "learning_rate": 1.7025739855383383e-05,
      "loss": 0.7572,
      "step": 1566
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6937841563521224,
      "learning_rate": 1.7021696793218246e-05,
      "loss": 0.7432,
      "step": 1567
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.784615985951674,
      "learning_rate": 1.7017651465707395e-05,
      "loss": 0.7377,
      "step": 1568
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7013296654359493,
      "learning_rate": 1.7013603874155942e-05,
      "loss": 0.7398,
      "step": 1569
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7423723772177212,
      "learning_rate": 1.7009554019869717e-05,
      "loss": 0.7698,
      "step": 1570
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7948108554763507,
      "learning_rate": 1.7005501904155292e-05,
      "loss": 0.7305,
      "step": 1571
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8489380796884386,
      "learning_rate": 1.7001447528319963e-05,
      "loss": 0.8044,
      "step": 1572
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7281821434393893,
      "learning_rate": 1.6997390893671754e-05,
      "loss": 0.7071,
      "step": 1573
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7651058298432656,
      "learning_rate": 1.699333200151942e-05,
      "loss": 0.763,
      "step": 1574
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9546614282269983,
      "learning_rate": 1.698927085317245e-05,
      "loss": 0.8292,
      "step": 1575
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7817371300241337,
      "learning_rate": 1.698520744994105e-05,
      "loss": 0.7872,
      "step": 1576
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8989404692789652,
      "learning_rate": 1.6981141793136154e-05,
      "loss": 0.8026,
      "step": 1577
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7620658871259062,
      "learning_rate": 1.697707388406944e-05,
      "loss": 0.7733,
      "step": 1578
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8129242871816393,
      "learning_rate": 1.697300372405329e-05,
      "loss": 0.7668,
      "step": 1579
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7359578211060912,
      "learning_rate": 1.6968931314400826e-05,
      "loss": 0.769,
      "step": 1580
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.712899315146925,
      "learning_rate": 1.696485665642589e-05,
      "loss": 0.7317,
      "step": 1581
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6265476721346054,
      "learning_rate": 1.6960779751443054e-05,
      "loss": 0.7135,
      "step": 1582
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7544325638661844,
      "learning_rate": 1.695670060076761e-05,
      "loss": 0.732,
      "step": 1583
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7512562215975038,
      "learning_rate": 1.6952619205715585e-05,
      "loss": 0.7855,
      "step": 1584
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6774587821326736,
      "learning_rate": 1.6948535567603715e-05,
      "loss": 0.7076,
      "step": 1585
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8321531111005945,
      "learning_rate": 1.6944449687749466e-05,
      "loss": 0.7391,
      "step": 1586
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8495041154632778,
      "learning_rate": 1.6940361567471025e-05,
      "loss": 0.7141,
      "step": 1587
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.712436706450894,
      "learning_rate": 1.6936271208087318e-05,
      "loss": 0.7659,
      "step": 1588
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7550323797703236,
      "learning_rate": 1.6932178610917963e-05,
      "loss": 0.7659,
      "step": 1589
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7527852037790559,
      "learning_rate": 1.6928083777283333e-05,
      "loss": 0.733,
      "step": 1590
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7230096601539913,
      "learning_rate": 1.6923986708504498e-05,
      "loss": 0.7558,
      "step": 1591
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6258243169085509,
      "learning_rate": 1.691988740590326e-05,
      "loss": 0.7106,
      "step": 1592
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.619949302715282,
      "learning_rate": 1.6915785870802136e-05,
      "loss": 0.7262,
      "step": 1593
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7872727538942906,
      "learning_rate": 1.691168210452437e-05,
      "loss": 0.745,
      "step": 1594
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6603543471535815,
      "learning_rate": 1.690757610839392e-05,
      "loss": 0.7467,
      "step": 1595
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7260225816880552,
      "learning_rate": 1.6903467883735472e-05,
      "loss": 0.6925,
      "step": 1596
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7368036723039798,
      "learning_rate": 1.689935743187442e-05,
      "loss": 0.7974,
      "step": 1597
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7954667292508565,
      "learning_rate": 1.6895244754136883e-05,
      "loss": 0.7652,
      "step": 1598
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.694328979335732,
      "learning_rate": 1.6891129851849696e-05,
      "loss": 0.7105,
      "step": 1599
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6654913170059589,
      "learning_rate": 1.6887012726340407e-05,
      "loss": 0.7248,
      "step": 1600
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.694392719957424,
      "learning_rate": 1.6882893378937297e-05,
      "loss": 0.7721,
      "step": 1601
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7540668455625681,
      "learning_rate": 1.6878771810969343e-05,
      "loss": 0.8082,
      "step": 1602
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6982971455473405,
      "learning_rate": 1.6874648023766258e-05,
      "loss": 0.7779,
      "step": 1603
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6690353006124236,
      "learning_rate": 1.6870522018658456e-05,
      "loss": 0.7587,
      "step": 1604
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7163624224667181,
      "learning_rate": 1.686639379697707e-05,
      "loss": 0.7587,
      "step": 1605
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6462083995142982,
      "learning_rate": 1.6862263360053958e-05,
      "loss": 0.7206,
      "step": 1606
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6791577162484901,
      "learning_rate": 1.6858130709221682e-05,
      "loss": 0.7363,
      "step": 1607
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6897676835754933,
      "learning_rate": 1.6853995845813516e-05,
      "loss": 0.7462,
      "step": 1608
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6483923738025086,
      "learning_rate": 1.6849858771163455e-05,
      "loss": 0.7514,
      "step": 1609
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6441693814943756,
      "learning_rate": 1.6845719486606217e-05,
      "loss": 0.7423,
      "step": 1610
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6966798815405627,
      "learning_rate": 1.6841577993477204e-05,
      "loss": 0.7273,
      "step": 1611
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.681205348671442,
      "learning_rate": 1.683743429311256e-05,
      "loss": 0.7338,
      "step": 1612
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8014487703446517,
      "learning_rate": 1.6833288386849123e-05,
      "loss": 0.7042,
      "step": 1613
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5943056853235076,
      "learning_rate": 1.682914027602445e-05,
      "loss": 0.6861,
      "step": 1614
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7556655695095701,
      "learning_rate": 1.682498996197681e-05,
      "loss": 0.7351,
      "step": 1615
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6920968197267094,
      "learning_rate": 1.6820837446045183e-05,
      "loss": 0.702,
      "step": 1616
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.630298328636567,
      "learning_rate": 1.6816682729569253e-05,
      "loss": 0.7136,
      "step": 1617
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7391889108092748,
      "learning_rate": 1.681252581388942e-05,
      "loss": 0.7087,
      "step": 1618
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7444445257799376,
      "learning_rate": 1.6808366700346787e-05,
      "loss": 0.7437,
      "step": 1619
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7355356271507535,
      "learning_rate": 1.6804205390283174e-05,
      "loss": 0.7501,
      "step": 1620
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7101921758292455,
      "learning_rate": 1.680004188504111e-05,
      "loss": 0.7358,
      "step": 1621
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8904989296285332,
      "learning_rate": 1.6795876185963824e-05,
      "loss": 0.8251,
      "step": 1622
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7896349901388369,
      "learning_rate": 1.6791708294395258e-05,
      "loss": 0.7631,
      "step": 1623
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7552480957372827,
      "learning_rate": 1.678753821168006e-05,
      "loss": 0.803,
      "step": 1624
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7763280897667751,
      "learning_rate": 1.6783365939163593e-05,
      "loss": 0.7836,
      "step": 1625
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6642419333742102,
      "learning_rate": 1.6779191478191904e-05,
      "loss": 0.736,
      "step": 1626
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7126651687337622,
      "learning_rate": 1.6775014830111776e-05,
      "loss": 0.7208,
      "step": 1627
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6963881073043047,
      "learning_rate": 1.6770835996270677e-05,
      "loss": 0.7298,
      "step": 1628
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6240590960553469,
      "learning_rate": 1.6766654978016785e-05,
      "loss": 0.7166,
      "step": 1629
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7201863433825337,
      "learning_rate": 1.6762471776698982e-05,
      "loss": 0.7122,
      "step": 1630
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7402702031759963,
      "learning_rate": 1.675828639366686e-05,
      "loss": 0.7916,
      "step": 1631
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7473629766098271,
      "learning_rate": 1.6754098830270708e-05,
      "loss": 0.7277,
      "step": 1632
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7348579180440732,
      "learning_rate": 1.6749909087861523e-05,
      "loss": 0.7306,
      "step": 1633
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7101696649633948,
      "learning_rate": 1.6745717167791004e-05,
      "loss": 0.7144,
      "step": 1634
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7258142783869698,
      "learning_rate": 1.6741523071411547e-05,
      "loss": 0.7011,
      "step": 1635
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.734560223125416,
      "learning_rate": 1.673732680007626e-05,
      "loss": 0.7505,
      "step": 1636
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7120035540876785,
      "learning_rate": 1.6733128355138943e-05,
      "loss": 0.7394,
      "step": 1637
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7271378541363523,
      "learning_rate": 1.672892773795411e-05,
      "loss": 0.6997,
      "step": 1638
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7024325870954093,
      "learning_rate": 1.672472494987696e-05,
      "loss": 0.687,
      "step": 1639
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7579682067576949,
      "learning_rate": 1.67205199922634e-05,
      "loss": 0.743,
      "step": 1640
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8125761797191711,
      "learning_rate": 1.671631286647004e-05,
      "loss": 0.8209,
      "step": 1641
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6781132540256432,
      "learning_rate": 1.6712103573854187e-05,
      "loss": 0.7049,
      "step": 1642
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7535275523031032,
      "learning_rate": 1.6707892115773843e-05,
      "loss": 0.6668,
      "step": 1643
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6609082144171392,
      "learning_rate": 1.6703678493587715e-05,
      "loss": 0.7463,
      "step": 1644
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7038413993922519,
      "learning_rate": 1.6699462708655202e-05,
      "loss": 0.7179,
      "step": 1645
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6971056159646477,
      "learning_rate": 1.669524476233641e-05,
      "loss": 0.6888,
      "step": 1646
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.772959231337151,
      "learning_rate": 1.6691024655992132e-05,
      "loss": 0.7505,
      "step": 1647
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7801239305515716,
      "learning_rate": 1.6686802390983862e-05,
      "loss": 0.7623,
      "step": 1648
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7595964038715058,
      "learning_rate": 1.6682577968673796e-05,
      "loss": 0.7141,
      "step": 1649
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7260209836075087,
      "learning_rate": 1.6678351390424813e-05,
      "loss": 0.7226,
      "step": 1650
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7766267572614689,
      "learning_rate": 1.6674122657600504e-05,
      "loss": 0.6778,
      "step": 1651
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7568440786939926,
      "learning_rate": 1.666989177156513e-05,
      "loss": 0.6943,
      "step": 1652
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6487677046115865,
      "learning_rate": 1.6665658733683683e-05,
      "loss": 0.7474,
      "step": 1653
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6928402635360607,
      "learning_rate": 1.6661423545321816e-05,
      "loss": 0.7263,
      "step": 1654
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8233435789549827,
      "learning_rate": 1.66571862078459e-05,
      "loss": 0.7067,
      "step": 1655
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6821024543915482,
      "learning_rate": 1.6652946722622978e-05,
      "loss": 0.7709,
      "step": 1656
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8410084692521922,
      "learning_rate": 1.6648705091020798e-05,
      "loss": 0.7679,
      "step": 1657
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7426333303358318,
      "learning_rate": 1.6644461314407805e-05,
      "loss": 0.729,
      "step": 1658
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7889411508218517,
      "learning_rate": 1.6640215394153122e-05,
      "loss": 0.7413,
      "step": 1659
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6917348594152969,
      "learning_rate": 1.6635967331626573e-05,
      "loss": 0.7208,
      "step": 1660
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.704703011565691,
      "learning_rate": 1.663171712819868e-05,
      "loss": 0.7256,
      "step": 1661
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7063055871961792,
      "learning_rate": 1.6627464785240635e-05,
      "loss": 0.7474,
      "step": 1662
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6224048624233247,
      "learning_rate": 1.6623210304124338e-05,
      "loss": 0.6814,
      "step": 1663
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.768407013325527,
      "learning_rate": 1.661895368622238e-05,
      "loss": 0.7453,
      "step": 1664
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8390401987266851,
      "learning_rate": 1.6614694932908023e-05,
      "loss": 0.7501,
      "step": 1665
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.773486895801635,
      "learning_rate": 1.6610434045555233e-05,
      "loss": 0.7707,
      "step": 1666
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8478274760007114,
      "learning_rate": 1.6606171025538668e-05,
      "loss": 0.7331,
      "step": 1667
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7425157618372791,
      "learning_rate": 1.6601905874233657e-05,
      "loss": 0.724,
      "step": 1668
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7584963882422961,
      "learning_rate": 1.6597638593016232e-05,
      "loss": 0.7767,
      "step": 1669
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7240817320587802,
      "learning_rate": 1.6593369183263108e-05,
      "loss": 0.6879,
      "step": 1670
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7682152552110898,
      "learning_rate": 1.6589097646351684e-05,
      "loss": 0.7812,
      "step": 1671
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.740321364874834,
      "learning_rate": 1.6584823983660046e-05,
      "loss": 0.7367,
      "step": 1672
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7130933325771466,
      "learning_rate": 1.6580548196566968e-05,
      "loss": 0.7348,
      "step": 1673
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.71650464744774,
      "learning_rate": 1.6576270286451906e-05,
      "loss": 0.7212,
      "step": 1674
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7221531594872036,
      "learning_rate": 1.6571990254695004e-05,
      "loss": 0.7238,
      "step": 1675
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.752270630844889,
      "learning_rate": 1.656770810267709e-05,
      "loss": 0.7667,
      "step": 1676
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1728964849047407,
      "learning_rate": 1.6563423831779677e-05,
      "loss": 0.8088,
      "step": 1677
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7287433116129227,
      "learning_rate": 1.6559137443384953e-05,
      "loss": 0.7027,
      "step": 1678
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6328130486829807,
      "learning_rate": 1.655484893887581e-05,
      "loss": 0.7031,
      "step": 1679
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6598246551596837,
      "learning_rate": 1.6550558319635793e-05,
      "loss": 0.7291,
      "step": 1680
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6714068608599545,
      "learning_rate": 1.6546265587049156e-05,
      "loss": 0.7277,
      "step": 1681
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7443357850753283,
      "learning_rate": 1.654197074250082e-05,
      "loss": 0.764,
      "step": 1682
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6498067150362423,
      "learning_rate": 1.6537673787376395e-05,
      "loss": 0.717,
      "step": 1683
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6778213292011972,
      "learning_rate": 1.6533374723062163e-05,
      "loss": 0.7467,
      "step": 1684
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7315422993731058,
      "learning_rate": 1.652907355094509e-05,
      "loss": 0.756,
      "step": 1685
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7227595232780747,
      "learning_rate": 1.652477027241283e-05,
      "loss": 0.7405,
      "step": 1686
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7367975061886926,
      "learning_rate": 1.652046488885371e-05,
      "loss": 0.7187,
      "step": 1687
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7966224112175703,
      "learning_rate": 1.6516157401656734e-05,
      "loss": 0.7864,
      "step": 1688
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8004063644313205,
      "learning_rate": 1.6511847812211584e-05,
      "loss": 0.7136,
      "step": 1689
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6686875947192994,
      "learning_rate": 1.650753612190863e-05,
      "loss": 0.7125,
      "step": 1690
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.666600877653839,
      "learning_rate": 1.650322233213891e-05,
      "loss": 0.7145,
      "step": 1691
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.693964098426029,
      "learning_rate": 1.6498906444294146e-05,
      "loss": 0.7187,
      "step": 1692
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7207561822274463,
      "learning_rate": 1.6494588459766726e-05,
      "loss": 0.7481,
      "step": 1693
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6890281770516284,
      "learning_rate": 1.6490268379949723e-05,
      "loss": 0.6678,
      "step": 1694
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7150397868941992,
      "learning_rate": 1.648594620623689e-05,
      "loss": 0.6987,
      "step": 1695
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7655246240205092,
      "learning_rate": 1.648162194002265e-05,
      "loss": 0.7136,
      "step": 1696
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7042608858700523,
      "learning_rate": 1.6477295582702098e-05,
      "loss": 0.7379,
      "step": 1697
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7191012484424234,
      "learning_rate": 1.6472967135671008e-05,
      "loss": 0.6722,
      "step": 1698
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7925961584562499,
      "learning_rate": 1.6468636600325828e-05,
      "loss": 0.719,
      "step": 1699
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7559729973916803,
      "learning_rate": 1.6464303978063675e-05,
      "loss": 0.7063,
      "step": 1700
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6648583697672539,
      "learning_rate": 1.6459969270282352e-05,
      "loss": 0.74,
      "step": 1701
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8605287542460726,
      "learning_rate": 1.6455632478380322e-05,
      "loss": 0.8358,
      "step": 1702
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7140178239722209,
      "learning_rate": 1.6451293603756716e-05,
      "loss": 0.725,
      "step": 1703
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6863114051040952,
      "learning_rate": 1.644695264781136e-05,
      "loss": 0.7607,
      "step": 1704
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.788387083753253,
      "learning_rate": 1.644260961194472e-05,
      "loss": 0.7554,
      "step": 1705
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6330643588837368,
      "learning_rate": 1.6438264497557968e-05,
      "loss": 0.6553,
      "step": 1706
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7587706413945704,
      "learning_rate": 1.6433917306052918e-05,
      "loss": 0.7357,
      "step": 1707
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6162192453933829,
      "learning_rate": 1.642956803883206e-05,
      "loss": 0.7095,
      "step": 1708
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7225334300031716,
      "learning_rate": 1.6425216697298573e-05,
      "loss": 0.775,
      "step": 1709
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7697485214289547,
      "learning_rate": 1.6420863282856278e-05,
      "loss": 0.7431,
      "step": 1710
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6993716286086523,
      "learning_rate": 1.641650779690968e-05,
      "loss": 0.7189,
      "step": 1711
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8108728136119241,
      "learning_rate": 1.6412150240863954e-05,
      "loss": 0.7692,
      "step": 1712
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7325571249128215,
      "learning_rate": 1.6407790616124934e-05,
      "loss": 0.7721,
      "step": 1713
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6597150967883952,
      "learning_rate": 1.6403428924099125e-05,
      "loss": 0.7727,
      "step": 1714
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.640553891083603,
      "learning_rate": 1.6399065166193703e-05,
      "loss": 0.7493,
      "step": 1715
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7416221478991782,
      "learning_rate": 1.6394699343816506e-05,
      "loss": 0.7499,
      "step": 1716
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7640916454024901,
      "learning_rate": 1.639033145837604e-05,
      "loss": 0.7118,
      "step": 1717
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7124771334698424,
      "learning_rate": 1.6385961511281473e-05,
      "loss": 0.7162,
      "step": 1718
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7116583009797277,
      "learning_rate": 1.6381589503942644e-05,
      "loss": 0.7479,
      "step": 1719
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.715054055589945,
      "learning_rate": 1.6377215437770057e-05,
      "loss": 0.7698,
      "step": 1720
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6569380574670167,
      "learning_rate": 1.6372839314174868e-05,
      "loss": 0.7244,
      "step": 1721
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6506291012265675,
      "learning_rate": 1.6368461134568908e-05,
      "loss": 0.7007,
      "step": 1722
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7619721550678522,
      "learning_rate": 1.6364080900364676e-05,
      "loss": 0.7927,
      "step": 1723
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6502591319375385,
      "learning_rate": 1.6359698612975318e-05,
      "loss": 0.7115,
      "step": 1724
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7182127829095467,
      "learning_rate": 1.635531427381466e-05,
      "loss": 0.7118,
      "step": 1725
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6711344114378908,
      "learning_rate": 1.6350927884297173e-05,
      "loss": 0.7851,
      "step": 1726
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6289528604737881,
      "learning_rate": 1.6346539445838004e-05,
      "loss": 0.7232,
      "step": 1727
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7947749214133506,
      "learning_rate": 1.634214895985295e-05,
      "loss": 0.7247,
      "step": 1728
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7330115254407826,
      "learning_rate": 1.6337756427758477e-05,
      "loss": 0.7799,
      "step": 1729
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6931376718343705,
      "learning_rate": 1.6333361850971703e-05,
      "loss": 0.7136,
      "step": 1730
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6913732038156273,
      "learning_rate": 1.6328965230910414e-05,
      "loss": 0.7309,
      "step": 1731
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7060402029736395,
      "learning_rate": 1.6324566568993047e-05,
      "loss": 0.6872,
      "step": 1732
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7493340581250791,
      "learning_rate": 1.6320165866638708e-05,
      "loss": 0.7165,
      "step": 1733
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6930763793306046,
      "learning_rate": 1.6315763125267153e-05,
      "loss": 0.7072,
      "step": 1734
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7448265361045131,
      "learning_rate": 1.6311358346298795e-05,
      "loss": 0.7251,
      "step": 1735
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7381536068940797,
      "learning_rate": 1.630695153115471e-05,
      "loss": 0.7044,
      "step": 1736
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6770168900725423,
      "learning_rate": 1.630254268125663e-05,
      "loss": 0.719,
      "step": 1737
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7592210220469693,
      "learning_rate": 1.629813179802694e-05,
      "loss": 0.7115,
      "step": 1738
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7779660715209328,
      "learning_rate": 1.6293718882888684e-05,
      "loss": 0.7322,
      "step": 1739
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7178854885322038,
      "learning_rate": 1.6289303937265556e-05,
      "loss": 0.7194,
      "step": 1740
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7604733486913685,
      "learning_rate": 1.628488696258191e-05,
      "loss": 0.7295,
      "step": 1741
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7463515595636703,
      "learning_rate": 1.628046796026276e-05,
      "loss": 0.7379,
      "step": 1742
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7732800282356235,
      "learning_rate": 1.627604693173377e-05,
      "loss": 0.7192,
      "step": 1743
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6913571439044297,
      "learning_rate": 1.6271623878421242e-05,
      "loss": 0.71,
      "step": 1744
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6913839824125327,
      "learning_rate": 1.626719880175216e-05,
      "loss": 0.7234,
      "step": 1745
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6836355493156644,
      "learning_rate": 1.626277170315414e-05,
      "loss": 0.7289,
      "step": 1746
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7860085010800116,
      "learning_rate": 1.6258342584055453e-05,
      "loss": 0.7495,
      "step": 1747
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6672694102002024,
      "learning_rate": 1.6253911445885032e-05,
      "loss": 0.7175,
      "step": 1748
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8107172178981801,
      "learning_rate": 1.6249478290072447e-05,
      "loss": 0.7551,
      "step": 1749
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7223613050272006,
      "learning_rate": 1.6245043118047935e-05,
      "loss": 0.741,
      "step": 1750
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6603456255676778,
      "learning_rate": 1.624060593124237e-05,
      "loss": 0.7123,
      "step": 1751
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6434660468058411,
      "learning_rate": 1.623616673108728e-05,
      "loss": 0.6837,
      "step": 1752
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6897128818735523,
      "learning_rate": 1.623172551901484e-05,
      "loss": 0.7361,
      "step": 1753
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7524854793400989,
      "learning_rate": 1.622728229645789e-05,
      "loss": 0.6818,
      "step": 1754
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7306830353415686,
      "learning_rate": 1.6222837064849894e-05,
      "loss": 0.7638,
      "step": 1755
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8101543819411311,
      "learning_rate": 1.6218389825624983e-05,
      "loss": 0.7805,
      "step": 1756
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7293511334904239,
      "learning_rate": 1.6213940580217925e-05,
      "loss": 0.7424,
      "step": 1757
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8009363405311527,
      "learning_rate": 1.6209489330064143e-05,
      "loss": 0.7927,
      "step": 1758
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6663656805281886,
      "learning_rate": 1.62050360765997e-05,
      "loss": 0.7326,
      "step": 1759
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6302519211047121,
      "learning_rate": 1.620058082126131e-05,
      "loss": 0.7065,
      "step": 1760
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.75955129822271,
      "learning_rate": 1.6196123565486333e-05,
      "loss": 0.7256,
      "step": 1761
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7000663692555863,
      "learning_rate": 1.619166431071277e-05,
      "loss": 0.731,
      "step": 1762
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9874024724611322,
      "learning_rate": 1.6187203058379267e-05,
      "loss": 0.7939,
      "step": 1763
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7570545706045478,
      "learning_rate": 1.6182739809925122e-05,
      "loss": 0.7517,
      "step": 1764
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8623351233128626,
      "learning_rate": 1.617827456679027e-05,
      "loss": 0.7095,
      "step": 1765
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6606789719937634,
      "learning_rate": 1.617380733041529e-05,
      "loss": 0.7731,
      "step": 1766
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7799008742784628,
      "learning_rate": 1.616933810224141e-05,
      "loss": 0.713,
      "step": 1767
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8010628399392395,
      "learning_rate": 1.616486688371049e-05,
      "loss": 0.7376,
      "step": 1768
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7052688932719632,
      "learning_rate": 1.616039367626504e-05,
      "loss": 0.7311,
      "step": 1769
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7867384094525514,
      "learning_rate": 1.6155918481348215e-05,
      "loss": 0.7413,
      "step": 1770
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7488765095728127,
      "learning_rate": 1.61514413004038e-05,
      "loss": 0.7149,
      "step": 1771
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7306998037045603,
      "learning_rate": 1.6146962134876233e-05,
      "loss": 0.7056,
      "step": 1772
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8402467445612952,
      "learning_rate": 1.6142480986210576e-05,
      "loss": 0.7052,
      "step": 1773
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7405972340899858,
      "learning_rate": 1.6137997855852555e-05,
      "loss": 0.7326,
      "step": 1774
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.696102308494348,
      "learning_rate": 1.6133512745248508e-05,
      "loss": 0.6985,
      "step": 1775
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6783367011186321,
      "learning_rate": 1.6129025655845433e-05,
      "loss": 0.7351,
      "step": 1776
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.753424798346466,
      "learning_rate": 1.6124536589090957e-05,
      "loss": 0.749,
      "step": 1777
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7063233657980029,
      "learning_rate": 1.6120045546433346e-05,
      "loss": 0.7216,
      "step": 1778
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6888697756106884,
      "learning_rate": 1.61155525293215e-05,
      "loss": 0.7481,
      "step": 1779
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.757688665550209,
      "learning_rate": 1.611105753920497e-05,
      "loss": 0.7202,
      "step": 1780
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7288909804705143,
      "learning_rate": 1.6106560577533924e-05,
      "loss": 0.7308,
      "step": 1781
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.746386787838614,
      "learning_rate": 1.610206164575918e-05,
      "loss": 0.7809,
      "step": 1782
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7914935554844562,
      "learning_rate": 1.6097560745332185e-05,
      "loss": 0.7916,
      "step": 1783
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6158347930678214,
      "learning_rate": 1.6093057877705022e-05,
      "loss": 0.7014,
      "step": 1784
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6579478939132576,
      "learning_rate": 1.6088553044330413e-05,
      "loss": 0.7371,
      "step": 1785
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7348489574326306,
      "learning_rate": 1.6084046246661713e-05,
      "loss": 0.6912,
      "step": 1786
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7602788909693398,
      "learning_rate": 1.6079537486152905e-05,
      "loss": 0.7739,
      "step": 1787
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7950431895820315,
      "learning_rate": 1.607502676425861e-05,
      "loss": 0.8039,
      "step": 1788
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6996545782145634,
      "learning_rate": 1.607051408243408e-05,
      "loss": 0.7114,
      "step": 1789
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7570830982769285,
      "learning_rate": 1.6065999442135205e-05,
      "loss": 0.735,
      "step": 1790
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7196048047770995,
      "learning_rate": 1.6061482844818494e-05,
      "loss": 0.7146,
      "step": 1791
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7409519032180568,
      "learning_rate": 1.60569642919411e-05,
      "loss": 0.7427,
      "step": 1792
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7705215863517672,
      "learning_rate": 1.6052443784960808e-05,
      "loss": 0.7093,
      "step": 1793
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7397477327670193,
      "learning_rate": 1.6047921325336014e-05,
      "loss": 0.7174,
      "step": 1794
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6877335715071068,
      "learning_rate": 1.604339691452577e-05,
      "loss": 0.6613,
      "step": 1795
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7630886499275451,
      "learning_rate": 1.603887055398974e-05,
      "loss": 0.7512,
      "step": 1796
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7121124545711951,
      "learning_rate": 1.6034342245188223e-05,
      "loss": 0.6916,
      "step": 1797
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6903376607364442,
      "learning_rate": 1.602981198958215e-05,
      "loss": 0.7209,
      "step": 1798
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7123685524298092,
      "learning_rate": 1.6025279788633064e-05,
      "loss": 0.7056,
      "step": 1799
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7208122684814061,
      "learning_rate": 1.602074564380316e-05,
      "loss": 0.7485,
      "step": 1800
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7070969046975417,
      "learning_rate": 1.6016209556555238e-05,
      "loss": 0.6748,
      "step": 1801
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7213650416987847,
      "learning_rate": 1.6011671528352747e-05,
      "loss": 0.7486,
      "step": 1802
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7424783522380406,
      "learning_rate": 1.6007131560659733e-05,
      "loss": 0.707,
      "step": 1803
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7838450411753418,
      "learning_rate": 1.60025896549409e-05,
      "loss": 0.7372,
      "step": 1804
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7481797504678842,
      "learning_rate": 1.5998045812661548e-05,
      "loss": 0.7737,
      "step": 1805
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7708799889714664,
      "learning_rate": 1.5993500035287625e-05,
      "loss": 0.8077,
      "step": 1806
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6450368806488394,
      "learning_rate": 1.598895232428569e-05,
      "loss": 0.7101,
      "step": 1807
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7050060470066803,
      "learning_rate": 1.5984402681122928e-05,
      "loss": 0.7393,
      "step": 1808
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6980612517234095,
      "learning_rate": 1.597985110726715e-05,
      "loss": 0.7365,
      "step": 1809
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7003942417784108,
      "learning_rate": 1.597529760418679e-05,
      "loss": 0.7067,
      "step": 1810
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7220506087331539,
      "learning_rate": 1.59707421733509e-05,
      "loss": 0.6977,
      "step": 1811
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6927502013402784,
      "learning_rate": 1.5966184816229157e-05,
      "loss": 0.7299,
      "step": 1812
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6954087513604053,
      "learning_rate": 1.5961625534291864e-05,
      "loss": 0.733,
      "step": 1813
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7894605940368379,
      "learning_rate": 1.5957064329009933e-05,
      "loss": 0.7254,
      "step": 1814
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7507154203872005,
      "learning_rate": 1.595250120185491e-05,
      "loss": 0.7562,
      "step": 1815
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7039814712983276,
      "learning_rate": 1.5947936154298956e-05,
      "loss": 0.7391,
      "step": 1816
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6913019691460912,
      "learning_rate": 1.5943369187814844e-05,
      "loss": 0.7194,
      "step": 1817
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7604233172401739,
      "learning_rate": 1.5938800303875977e-05,
      "loss": 0.7916,
      "step": 1818
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6959374460045886,
      "learning_rate": 1.593422950395637e-05,
      "loss": 0.7229,
      "step": 1819
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.874538464836645,
      "learning_rate": 1.592965678953066e-05,
      "loss": 0.8021,
      "step": 1820
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7899716059360099,
      "learning_rate": 1.5925082162074098e-05,
      "loss": 0.7266,
      "step": 1821
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7566789049973051,
      "learning_rate": 1.592050562306256e-05,
      "loss": 0.6934,
      "step": 1822
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.046205518056806,
      "learning_rate": 1.5915927173972522e-05,
      "loss": 0.6899,
      "step": 1823
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8462858148517756,
      "learning_rate": 1.5911346816281098e-05,
      "loss": 0.7433,
      "step": 1824
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6686454640592628,
      "learning_rate": 1.5906764551466004e-05,
      "loss": 0.7625,
      "step": 1825
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8001985458688817,
      "learning_rate": 1.5902180381005567e-05,
      "loss": 0.7442,
      "step": 1826
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8469033585100983,
      "learning_rate": 1.5897594306378748e-05,
      "loss": 0.741,
      "step": 1827
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6878540434542597,
      "learning_rate": 1.58930063290651e-05,
      "loss": 0.7228,
      "step": 1828
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8069132408469945,
      "learning_rate": 1.5888416450544803e-05,
      "loss": 0.7069,
      "step": 1829
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9409174251735448,
      "learning_rate": 1.588382467229865e-05,
      "loss": 0.8068,
      "step": 1830
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7057450979027633,
      "learning_rate": 1.587923099580804e-05,
      "loss": 0.728,
      "step": 1831
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6786751503702344,
      "learning_rate": 1.587463542255499e-05,
      "loss": 0.732,
      "step": 1832
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7261416670575954,
      "learning_rate": 1.5870037954022133e-05,
      "loss": 0.7232,
      "step": 1833
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8303550128306896,
      "learning_rate": 1.5865438591692703e-05,
      "loss": 0.7004,
      "step": 1834
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6673332073985466,
      "learning_rate": 1.586083733705055e-05,
      "loss": 0.7075,
      "step": 1835
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6614710822670398,
      "learning_rate": 1.585623419158014e-05,
      "loss": 0.7283,
      "step": 1836
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6825363944859301,
      "learning_rate": 1.5851629156766532e-05,
      "loss": 0.6827,
      "step": 1837
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6882758456232992,
      "learning_rate": 1.584702223409542e-05,
      "loss": 0.7116,
      "step": 1838
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9166312494604072,
      "learning_rate": 1.5842413425053088e-05,
      "loss": 0.759,
      "step": 1839
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6223715244954526,
      "learning_rate": 1.5837802731126427e-05,
      "loss": 0.7012,
      "step": 1840
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6672993505492546,
      "learning_rate": 1.5833190153802954e-05,
      "loss": 0.7204,
      "step": 1841
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7491882954663822,
      "learning_rate": 1.5828575694570772e-05,
      "loss": 0.7871,
      "step": 1842
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6993911648053436,
      "learning_rate": 1.5823959354918613e-05,
      "loss": 0.7204,
      "step": 1843
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7548555853583049,
      "learning_rate": 1.5819341136335796e-05,
      "loss": 0.7434,
      "step": 1844
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6480823123626158,
      "learning_rate": 1.5814721040312257e-05,
      "loss": 0.7134,
      "step": 1845
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6712698209810227,
      "learning_rate": 1.5810099068338534e-05,
      "loss": 0.7335,
      "step": 1846
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7135434205251908,
      "learning_rate": 1.5805475221905774e-05,
      "loss": 0.7389,
      "step": 1847
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.675496143415033,
      "learning_rate": 1.5800849502505725e-05,
      "loss": 0.6774,
      "step": 1848
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7074830547186598,
      "learning_rate": 1.579622191163074e-05,
      "loss": 0.7689,
      "step": 1849
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6653635070100244,
      "learning_rate": 1.5791592450773778e-05,
      "loss": 0.685,
      "step": 1850
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.666433714010464,
      "learning_rate": 1.5786961121428397e-05,
      "loss": 0.7351,
      "step": 1851
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6992225669256741,
      "learning_rate": 1.578232792508876e-05,
      "loss": 0.7196,
      "step": 1852
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7382585380595815,
      "learning_rate": 1.5777692863249638e-05,
      "loss": 0.7444,
      "step": 1853
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6451709835997319,
      "learning_rate": 1.5773055937406392e-05,
      "loss": 0.6944,
      "step": 1854
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7311375018139137,
      "learning_rate": 1.5768417149054996e-05,
      "loss": 0.7676,
      "step": 1855
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7348730114509042,
      "learning_rate": 1.5763776499692018e-05,
      "loss": 0.7454,
      "step": 1856
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7344992071653852,
      "learning_rate": 1.5759133990814625e-05,
      "loss": 0.7286,
      "step": 1857
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6633425650443611,
      "learning_rate": 1.5754489623920593e-05,
      "loss": 0.7243,
      "step": 1858
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6468479022523999,
      "learning_rate": 1.5749843400508284e-05,
      "loss": 0.6727,
      "step": 1859
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6726295576995787,
      "learning_rate": 1.5745195322076674e-05,
      "loss": 0.7175,
      "step": 1860
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6324150461623673,
      "learning_rate": 1.5740545390125326e-05,
      "loss": 0.7005,
      "step": 1861
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.658118438880032,
      "learning_rate": 1.5735893606154406e-05,
      "loss": 0.6631,
      "step": 1862
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6844170147641135,
      "learning_rate": 1.5731239971664677e-05,
      "loss": 0.7419,
      "step": 1863
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7115752301142595,
      "learning_rate": 1.5726584488157497e-05,
      "loss": 0.7699,
      "step": 1864
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7685728732666665,
      "learning_rate": 1.5721927157134823e-05,
      "loss": 0.7474,
      "step": 1865
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6774957363450368,
      "learning_rate": 1.5717267980099203e-05,
      "loss": 0.7398,
      "step": 1866
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6060622974264642,
      "learning_rate": 1.5712606958553792e-05,
      "loss": 0.6749,
      "step": 1867
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7848923765487986,
      "learning_rate": 1.5707944094002334e-05,
      "loss": 0.7753,
      "step": 1868
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7708765768277887,
      "learning_rate": 1.5703279387949152e-05,
      "loss": 0.7646,
      "step": 1869
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7780204285966534,
      "learning_rate": 1.569861284189919e-05,
      "loss": 0.7392,
      "step": 1870
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7618156791908864,
      "learning_rate": 1.5693944457357973e-05,
      "loss": 0.7788,
      "step": 1871
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8346527135282461,
      "learning_rate": 1.5689274235831612e-05,
      "loss": 0.7454,
      "step": 1872
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7124248824883883,
      "learning_rate": 1.5684602178826823e-05,
      "loss": 0.7001,
      "step": 1873
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7484721132085644,
      "learning_rate": 1.567992828785091e-05,
      "loss": 0.79,
      "step": 1874
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7496242005665792,
      "learning_rate": 1.5675252564411762e-05,
      "loss": 0.7227,
      "step": 1875
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6991159475958557,
      "learning_rate": 1.567057501001787e-05,
      "loss": 0.7676,
      "step": 1876
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7329870888734562,
      "learning_rate": 1.5665895626178307e-05,
      "loss": 0.7119,
      "step": 1877
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6806241618633562,
      "learning_rate": 1.566121441440274e-05,
      "loss": 0.7216,
      "step": 1878
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6442127958060552,
      "learning_rate": 1.5656531376201426e-05,
      "loss": 0.6832,
      "step": 1879
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.653422083928669,
      "learning_rate": 1.5651846513085218e-05,
      "loss": 0.7043,
      "step": 1880
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6886485136558183,
      "learning_rate": 1.5647159826565537e-05,
      "loss": 0.738,
      "step": 1881
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7666683793588472,
      "learning_rate": 1.5642471318154417e-05,
      "loss": 0.7386,
      "step": 1882
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.633202347078734,
      "learning_rate": 1.5637780989364464e-05,
      "loss": 0.7106,
      "step": 1883
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8231829086653383,
      "learning_rate": 1.5633088841708876e-05,
      "loss": 0.7438,
      "step": 1884
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7229258886509021,
      "learning_rate": 1.5628394876701436e-05,
      "loss": 0.7482,
      "step": 1885
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7432100224579861,
      "learning_rate": 1.5623699095856518e-05,
      "loss": 0.7342,
      "step": 1886
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7020970741848643,
      "learning_rate": 1.5619001500689076e-05,
      "loss": 0.7249,
      "step": 1887
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6758750289680571,
      "learning_rate": 1.5614302092714654e-05,
      "loss": 0.724,
      "step": 1888
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.767881074936369,
      "learning_rate": 1.5609600873449378e-05,
      "loss": 0.7817,
      "step": 1889
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8017386450213722,
      "learning_rate": 1.560489784440996e-05,
      "loss": 0.8112,
      "step": 1890
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7554469046243286,
      "learning_rate": 1.5600193007113695e-05,
      "loss": 0.7322,
      "step": 1891
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8102137822116793,
      "learning_rate": 1.559548636307846e-05,
      "loss": 0.822,
      "step": 1892
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7266622504409334,
      "learning_rate": 1.559077791382272e-05,
      "loss": 0.7331,
      "step": 1893
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8236314696978291,
      "learning_rate": 1.5586067660865517e-05,
      "loss": 0.7315,
      "step": 1894
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9073719111798381,
      "learning_rate": 1.5581355605726478e-05,
      "loss": 0.7265,
      "step": 1895
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6673654520266233,
      "learning_rate": 1.5576641749925802e-05,
      "loss": 0.7335,
      "step": 1896
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6551505165165572,
      "learning_rate": 1.557192609498429e-05,
      "loss": 0.714,
      "step": 1897
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7127133672735998,
      "learning_rate": 1.55672086424233e-05,
      "loss": 0.7067,
      "step": 1898
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5771659618549808,
      "learning_rate": 1.556248939376479e-05,
      "loss": 0.6791,
      "step": 1899
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6591285618444185,
      "learning_rate": 1.5557768350531278e-05,
      "loss": 0.6851,
      "step": 1900
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7129118595899143,
      "learning_rate": 1.5553045514245877e-05,
      "loss": 0.7347,
      "step": 1901
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7404110339990851,
      "learning_rate": 1.5548320886432274e-05,
      "loss": 0.7126,
      "step": 1902
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7820583609241064,
      "learning_rate": 1.5543594468614725e-05,
      "loss": 0.7365,
      "step": 1903
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.685974100484121,
      "learning_rate": 1.553886626231808e-05,
      "loss": 0.6807,
      "step": 1904
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8421931952642077,
      "learning_rate": 1.5534136269067747e-05,
      "loss": 0.7466,
      "step": 1905
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6567549530648689,
      "learning_rate": 1.5529404490389727e-05,
      "loss": 0.7198,
      "step": 1906
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7951364700305693,
      "learning_rate": 1.5524670927810595e-05,
      "loss": 0.708,
      "step": 1907
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7444582016187891,
      "learning_rate": 1.5519935582857485e-05,
      "loss": 0.7661,
      "step": 1908
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7175075079424397,
      "learning_rate": 1.5515198457058126e-05,
      "loss": 0.7772,
      "step": 1909
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7536624564683143,
      "learning_rate": 1.551045955194081e-05,
      "loss": 0.7065,
      "step": 1910
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6984653265623155,
      "learning_rate": 1.5505718869034413e-05,
      "loss": 0.6951,
      "step": 1911
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7357531590512498,
      "learning_rate": 1.5500976409868368e-05,
      "loss": 0.7222,
      "step": 1912
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6373570860441494,
      "learning_rate": 1.5496232175972697e-05,
      "loss": 0.7594,
      "step": 1913
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7154995245785827,
      "learning_rate": 1.549148616887799e-05,
      "loss": 0.7486,
      "step": 1914
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6961924538924301,
      "learning_rate": 1.5486738390115402e-05,
      "loss": 0.7369,
      "step": 1915
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7508730863096518,
      "learning_rate": 1.5481988841216676e-05,
      "loss": 0.7383,
      "step": 1916
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6618676905826198,
      "learning_rate": 1.54772375237141e-05,
      "loss": 0.7226,
      "step": 1917
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7043906666140048,
      "learning_rate": 1.5472484439140566e-05,
      "loss": 0.7459,
      "step": 1918
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6796200011015335,
      "learning_rate": 1.54677295890295e-05,
      "loss": 0.7125,
      "step": 1919
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7180529013515841,
      "learning_rate": 1.546297297491493e-05,
      "loss": 0.6882,
      "step": 1920
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.651686568274942,
      "learning_rate": 1.5458214598331435e-05,
      "loss": 0.7207,
      "step": 1921
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7944072657511791,
      "learning_rate": 1.5453454460814158e-05,
      "loss": 0.7021,
      "step": 1922
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6558657641758138,
      "learning_rate": 1.544869256389883e-05,
      "loss": 0.7264,
      "step": 1923
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6866109950791599,
      "learning_rate": 1.5443928909121732e-05,
      "loss": 0.7259,
      "step": 1924
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7134935733351867,
      "learning_rate": 1.543916349801972e-05,
      "loss": 0.6804,
      "step": 1925
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7166327369195801,
      "learning_rate": 1.5434396332130216e-05,
      "loss": 0.7399,
      "step": 1926
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7337132084009084,
      "learning_rate": 1.5429627412991208e-05,
      "loss": 0.707,
      "step": 1927
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9392821243663715,
      "learning_rate": 1.542485674214124e-05,
      "loss": 0.7547,
      "step": 1928
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.682557266283373,
      "learning_rate": 1.542008432111944e-05,
      "loss": 0.7237,
      "step": 1929
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7116895072851667,
      "learning_rate": 1.5415310151465492e-05,
      "loss": 0.7576,
      "step": 1930
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6816209601869964,
      "learning_rate": 1.541053423471963e-05,
      "loss": 0.7074,
      "step": 1931
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7524473093059351,
      "learning_rate": 1.5405756572422674e-05,
      "loss": 0.7369,
      "step": 1932
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6521897815958001,
      "learning_rate": 1.5400977166115996e-05,
      "loss": 0.7522,
      "step": 1933
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8770470777387647,
      "learning_rate": 1.5396196017341528e-05,
      "loss": 0.7762,
      "step": 1934
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6871408964915432,
      "learning_rate": 1.5391413127641775e-05,
      "loss": 0.7036,
      "step": 1935
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7642985635266702,
      "learning_rate": 1.5386628498559794e-05,
      "loss": 0.6925,
      "step": 1936
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7377978677191588,
      "learning_rate": 1.5381842131639204e-05,
      "loss": 0.7665,
      "step": 1937
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6122595266159292,
      "learning_rate": 1.537705402842419e-05,
      "loss": 0.6924,
      "step": 1938
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.695378768107626,
      "learning_rate": 1.5372264190459492e-05,
      "loss": 0.6968,
      "step": 1939
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5879554397453552,
      "learning_rate": 1.5367472619290407e-05,
      "loss": 0.6567,
      "step": 1940
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6850838425723593,
      "learning_rate": 1.5362679316462802e-05,
      "loss": 0.6961,
      "step": 1941
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7161293879110254,
      "learning_rate": 1.5357884283523097e-05,
      "loss": 0.7495,
      "step": 1942
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7621098860228991,
      "learning_rate": 1.5353087522018265e-05,
      "loss": 0.7762,
      "step": 1943
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7655694904627991,
      "learning_rate": 1.5348289033495843e-05,
      "loss": 0.7777,
      "step": 1944
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.692966152190597,
      "learning_rate": 1.534348881950393e-05,
      "loss": 0.7319,
      "step": 1945
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.732390788169402,
      "learning_rate": 1.533868688159116e-05,
      "loss": 0.7397,
      "step": 1946
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6486022347816912,
      "learning_rate": 1.5333883221306754e-05,
      "loss": 0.6892,
      "step": 1947
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6902685651215071,
      "learning_rate": 1.5329077840200463e-05,
      "loss": 0.7267,
      "step": 1948
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7869055847431982,
      "learning_rate": 1.5324270739822608e-05,
      "loss": 0.7514,
      "step": 1949
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8074973866712688,
      "learning_rate": 1.5319461921724056e-05,
      "loss": 0.8226,
      "step": 1950
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6901162234274941,
      "learning_rate": 1.531465138745623e-05,
      "loss": 0.7356,
      "step": 1951
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7409716231300452,
      "learning_rate": 1.5309839138571118e-05,
      "loss": 0.7516,
      "step": 1952
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6865949896345983,
      "learning_rate": 1.5305025176621245e-05,
      "loss": 0.7535,
      "step": 1953
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8733130435967762,
      "learning_rate": 1.5300209503159696e-05,
      "loss": 0.8532,
      "step": 1954
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7045106633901236,
      "learning_rate": 1.529539211974011e-05,
      "loss": 0.7424,
      "step": 1955
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6676309071028051,
      "learning_rate": 1.5290573027916673e-05,
      "loss": 0.7505,
      "step": 1956
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7258665677157482,
      "learning_rate": 1.5285752229244125e-05,
      "loss": 0.7441,
      "step": 1957
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.71288460958766,
      "learning_rate": 1.5280929725277752e-05,
      "loss": 0.7374,
      "step": 1958
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6292389235946626,
      "learning_rate": 1.5276105517573405e-05,
      "loss": 0.679,
      "step": 1959
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7200624771969867,
      "learning_rate": 1.527127960768746e-05,
      "loss": 0.7216,
      "step": 1960
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6299620320716252,
      "learning_rate": 1.526645199717687e-05,
      "loss": 0.6836,
      "step": 1961
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.693988325740059,
      "learning_rate": 1.5261622687599117e-05,
      "loss": 0.7608,
      "step": 1962
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7423686705313406,
      "learning_rate": 1.5256791680512231e-05,
      "loss": 0.7322,
      "step": 1963
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7280501651932013,
      "learning_rate": 1.52519589774748e-05,
      "loss": 0.7856,
      "step": 1964
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7332830845612114,
      "learning_rate": 1.5247124580045957e-05,
      "loss": 0.7077,
      "step": 1965
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7331775789142313,
      "learning_rate": 1.5242288489785379e-05,
      "loss": 0.7624,
      "step": 1966
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6568811547893852,
      "learning_rate": 1.5237450708253288e-05,
      "loss": 0.7139,
      "step": 1967
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7726159435544646,
      "learning_rate": 1.5232611237010455e-05,
      "loss": 0.6817,
      "step": 1968
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.620721580798881,
      "learning_rate": 1.5227770077618189e-05,
      "loss": 0.6838,
      "step": 1969
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6780444030256918,
      "learning_rate": 1.5222927231638354e-05,
      "loss": 0.7186,
      "step": 1970
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6619323604231557,
      "learning_rate": 1.5218082700633355e-05,
      "loss": 0.7341,
      "step": 1971
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7446727691438678,
      "learning_rate": 1.521323648616613e-05,
      "loss": 0.7245,
      "step": 1972
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7705211639128885,
      "learning_rate": 1.5208388589800173e-05,
      "loss": 0.7904,
      "step": 1973
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7212536243366315,
      "learning_rate": 1.5203539013099521e-05,
      "loss": 0.6996,
      "step": 1974
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6512083849489966,
      "learning_rate": 1.5198687757628745e-05,
      "loss": 0.7081,
      "step": 1975
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6917407660518554,
      "learning_rate": 1.5193834824952961e-05,
      "loss": 0.7375,
      "step": 1976
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6985719695593829,
      "learning_rate": 1.5188980216637828e-05,
      "loss": 0.7136,
      "step": 1977
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7345751227142048,
      "learning_rate": 1.5184123934249537e-05,
      "loss": 0.7394,
      "step": 1978
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7029171904807912,
      "learning_rate": 1.5179265979354837e-05,
      "loss": 0.7283,
      "step": 1979
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7027320843672292,
      "learning_rate": 1.5174406353520996e-05,
      "loss": 0.743,
      "step": 1980
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.762395911310599,
      "learning_rate": 1.516954505831583e-05,
      "loss": 0.7128,
      "step": 1981
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7437376659551443,
      "learning_rate": 1.5164682095307703e-05,
      "loss": 0.7282,
      "step": 1982
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6179752965097325,
      "learning_rate": 1.5159817466065503e-05,
      "loss": 0.7174,
      "step": 1983
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8037840910471826,
      "learning_rate": 1.5154951172158661e-05,
      "loss": 0.7903,
      "step": 1984
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7592002485243258,
      "learning_rate": 1.5150083215157146e-05,
      "loss": 0.7216,
      "step": 1985
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6808264159889833,
      "learning_rate": 1.5145213596631459e-05,
      "loss": 0.7217,
      "step": 1986
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8333529468090015,
      "learning_rate": 1.514034231815264e-05,
      "loss": 0.7734,
      "step": 1987
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7433670222150692,
      "learning_rate": 1.5135469381292275e-05,
      "loss": 0.7335,
      "step": 1988
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6930811609387546,
      "learning_rate": 1.5130594787622462e-05,
      "loss": 0.7566,
      "step": 1989
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7129616707476818,
      "learning_rate": 1.5125718538715854e-05,
      "loss": 0.7107,
      "step": 1990
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6451015336109553,
      "learning_rate": 1.512084063614563e-05,
      "loss": 0.7077,
      "step": 1991
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7599356304266249,
      "learning_rate": 1.51159610814855e-05,
      "loss": 0.7221,
      "step": 1992
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6239136181747024,
      "learning_rate": 1.511107987630971e-05,
      "loss": 0.6853,
      "step": 1993
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7284843098405713,
      "learning_rate": 1.5106197022193047e-05,
      "loss": 0.7195,
      "step": 1994
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7247203459335336,
      "learning_rate": 1.5101312520710809e-05,
      "loss": 0.753,
      "step": 1995
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6706448073996855,
      "learning_rate": 1.5096426373438843e-05,
      "loss": 0.7011,
      "step": 1996
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7712467016321712,
      "learning_rate": 1.5091538581953526e-05,
      "loss": 0.7225,
      "step": 1997
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6279182877069985,
      "learning_rate": 1.5086649147831755e-05,
      "loss": 0.6933,
      "step": 1998
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7456078684377504,
      "learning_rate": 1.5081758072650967e-05,
      "loss": 0.7225,
      "step": 1999
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6902512183781513,
      "learning_rate": 1.5076865357989126e-05,
      "loss": 0.7169,
      "step": 2000
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7360224789904406,
      "learning_rate": 1.5071971005424719e-05,
      "loss": 0.7055,
      "step": 2001
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6948627412854241,
      "learning_rate": 1.506707501653677e-05,
      "loss": 0.7145,
      "step": 2002
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7182584676931578,
      "learning_rate": 1.5062177392904832e-05,
      "loss": 0.7496,
      "step": 2003
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7018152334935637,
      "learning_rate": 1.5057278136108968e-05,
      "loss": 0.6791,
      "step": 2004
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7123042427514472,
      "learning_rate": 1.5052377247729785e-05,
      "loss": 0.7321,
      "step": 2005
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7454731513669736,
      "learning_rate": 1.5047474729348417e-05,
      "loss": 0.6868,
      "step": 2006
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8440948790909236,
      "learning_rate": 1.5042570582546513e-05,
      "loss": 0.8385,
      "step": 2007
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7242557393493116,
      "learning_rate": 1.5037664808906258e-05,
      "loss": 0.7433,
      "step": 2008
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7977179279542953,
      "learning_rate": 1.5032757410010354e-05,
      "loss": 0.7773,
      "step": 2009
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7733499272337097,
      "learning_rate": 1.5027848387442028e-05,
      "loss": 0.6933,
      "step": 2010
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7552280327114969,
      "learning_rate": 1.5022937742785036e-05,
      "loss": 0.7112,
      "step": 2011
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7435606031348168,
      "learning_rate": 1.5018025477623655e-05,
      "loss": 0.7183,
      "step": 2012
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7619123063517244,
      "learning_rate": 1.501311159354268e-05,
      "loss": 0.7329,
      "step": 2013
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8500076652794074,
      "learning_rate": 1.5008196092127434e-05,
      "loss": 0.7235,
      "step": 2014
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7646327733119529,
      "learning_rate": 1.5003278974963764e-05,
      "loss": 0.7583,
      "step": 2015
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8347281690094923,
      "learning_rate": 1.4998360243638029e-05,
      "loss": 0.8075,
      "step": 2016
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7181108800612663,
      "learning_rate": 1.4993439899737117e-05,
      "loss": 0.6934,
      "step": 2017
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6005504771116612,
      "learning_rate": 1.4988517944848436e-05,
      "loss": 0.6886,
      "step": 2018
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7123322540531,
      "learning_rate": 1.4983594380559904e-05,
      "loss": 0.7296,
      "step": 2019
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8085352961142085,
      "learning_rate": 1.4978669208459973e-05,
      "loss": 0.7458,
      "step": 2020
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5910297931904384,
      "learning_rate": 1.4973742430137603e-05,
      "loss": 0.7106,
      "step": 2021
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7889342445642351,
      "learning_rate": 1.4968814047182273e-05,
      "loss": 0.778,
      "step": 2022
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6790172595532437,
      "learning_rate": 1.4963884061183984e-05,
      "loss": 0.7014,
      "step": 2023
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.648634528993748,
      "learning_rate": 1.4958952473733256e-05,
      "loss": 0.7358,
      "step": 2024
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.760720580994878,
      "learning_rate": 1.4954019286421116e-05,
      "loss": 0.7264,
      "step": 2025
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7103172811363859,
      "learning_rate": 1.4949084500839116e-05,
      "loss": 0.7517,
      "step": 2026
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7296891252485188,
      "learning_rate": 1.4944148118579325e-05,
      "loss": 0.7338,
      "step": 2027
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7040217439456037,
      "learning_rate": 1.4939210141234312e-05,
      "loss": 0.7086,
      "step": 2028
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7632603892221763,
      "learning_rate": 1.493427057039718e-05,
      "loss": 0.7127,
      "step": 2029
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7934579763254618,
      "learning_rate": 1.4929329407661532e-05,
      "loss": 0.7636,
      "step": 2030
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6990705450006336,
      "learning_rate": 1.4924386654621496e-05,
      "loss": 0.7746,
      "step": 2031
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7554774241085195,
      "learning_rate": 1.4919442312871705e-05,
      "loss": 0.7401,
      "step": 2032
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7649190545633071,
      "learning_rate": 1.4914496384007304e-05,
      "loss": 0.7977,
      "step": 2033
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6990341985996713,
      "learning_rate": 1.4909548869623953e-05,
      "loss": 0.694,
      "step": 2034
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7221761774059925,
      "learning_rate": 1.490459977131783e-05,
      "loss": 0.7218,
      "step": 2035
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7729379267063865,
      "learning_rate": 1.4899649090685607e-05,
      "loss": 0.6897,
      "step": 2036
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7801234097557065,
      "learning_rate": 1.4894696829324482e-05,
      "loss": 0.7582,
      "step": 2037
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6695622867487041,
      "learning_rate": 1.4889742988832159e-05,
      "loss": 0.7244,
      "step": 2038
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7398609398499874,
      "learning_rate": 1.4884787570806846e-05,
      "loss": 0.7439,
      "step": 2039
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7357663488314146,
      "learning_rate": 1.4879830576847272e-05,
      "loss": 0.6893,
      "step": 2040
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7107642538850485,
      "learning_rate": 1.4874872008552658e-05,
      "loss": 0.7287,
      "step": 2041
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7353527214684926,
      "learning_rate": 1.4869911867522747e-05,
      "loss": 0.7498,
      "step": 2042
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7006849693599703,
      "learning_rate": 1.4864950155357778e-05,
      "loss": 0.7322,
      "step": 2043
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7070320079119091,
      "learning_rate": 1.485998687365851e-05,
      "loss": 0.7136,
      "step": 2044
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7044983040980524,
      "learning_rate": 1.4855022024026196e-05,
      "loss": 0.722,
      "step": 2045
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7527161481774529,
      "learning_rate": 1.4850055608062599e-05,
      "loss": 0.8111,
      "step": 2046
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7790965295818147,
      "learning_rate": 1.4845087627369995e-05,
      "loss": 0.7022,
      "step": 2047
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6189982856207513,
      "learning_rate": 1.484011808355115e-05,
      "loss": 0.6529,
      "step": 2048
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6956495446104659,
      "learning_rate": 1.4835146978209349e-05,
      "loss": 0.753,
      "step": 2049
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7101180494177186,
      "learning_rate": 1.4830174312948372e-05,
      "loss": 0.7383,
      "step": 2050
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7178519768122911,
      "learning_rate": 1.48252000893725e-05,
      "loss": 0.7113,
      "step": 2051
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6697359623447462,
      "learning_rate": 1.4820224309086527e-05,
      "loss": 0.6736,
      "step": 2052
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.690395694620125,
      "learning_rate": 1.4815246973695744e-05,
      "loss": 0.6806,
      "step": 2053
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7121209197364909,
      "learning_rate": 1.4810268084805937e-05,
      "loss": 0.7234,
      "step": 2054
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.725958024859341,
      "learning_rate": 1.4805287644023404e-05,
      "loss": 0.7114,
      "step": 2055
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6461711715685327,
      "learning_rate": 1.4800305652954937e-05,
      "loss": 0.6752,
      "step": 2056
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7593961033084474,
      "learning_rate": 1.4795322113207833e-05,
      "loss": 0.7188,
      "step": 2057
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7825458957238552,
      "learning_rate": 1.4790337026389883e-05,
      "loss": 0.7412,
      "step": 2058
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6830042399290495,
      "learning_rate": 1.4785350394109381e-05,
      "loss": 0.7163,
      "step": 2059
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5803241050703224,
      "learning_rate": 1.478036221797512e-05,
      "loss": 0.6713,
      "step": 2060
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.698100802058542,
      "learning_rate": 1.4775372499596383e-05,
      "loss": 0.7129,
      "step": 2061
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7425326382574537,
      "learning_rate": 1.4770381240582964e-05,
      "loss": 0.7583,
      "step": 2062
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8973600442849898,
      "learning_rate": 1.4765388442545148e-05,
      "loss": 0.7995,
      "step": 2063
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7069040713128016,
      "learning_rate": 1.4760394107093708e-05,
      "loss": 0.726,
      "step": 2064
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.699625996695417,
      "learning_rate": 1.4755398235839931e-05,
      "loss": 0.7498,
      "step": 2065
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6387265056322635,
      "learning_rate": 1.4750400830395577e-05,
      "loss": 0.6931,
      "step": 2066
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6319723969968118,
      "learning_rate": 1.4745401892372924e-05,
      "loss": 0.7306,
      "step": 2067
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6895581082203649,
      "learning_rate": 1.474040142338473e-05,
      "loss": 0.7345,
      "step": 2068
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6062998484630243,
      "learning_rate": 1.4735399425044248e-05,
      "loss": 0.7048,
      "step": 2069
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7257633669915735,
      "learning_rate": 1.4730395898965229e-05,
      "loss": 0.7467,
      "step": 2070
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.649202311402616,
      "learning_rate": 1.4725390846761913e-05,
      "loss": 0.7382,
      "step": 2071
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7113348502503187,
      "learning_rate": 1.472038427004904e-05,
      "loss": 0.746,
      "step": 2072
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6559122353389562,
      "learning_rate": 1.4715376170441829e-05,
      "loss": 0.7704,
      "step": 2073
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6453421828503672,
      "learning_rate": 1.4710366549555999e-05,
      "loss": 0.7236,
      "step": 2074
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6611303225878586,
      "learning_rate": 1.4705355409007758e-05,
      "loss": 0.7425,
      "step": 2075
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6919090912832708,
      "learning_rate": 1.470034275041381e-05,
      "loss": 0.724,
      "step": 2076
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7552045533960824,
      "learning_rate": 1.4695328575391337e-05,
      "loss": 0.7148,
      "step": 2077
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6814202582594446,
      "learning_rate": 1.4690312885558013e-05,
      "loss": 0.7156,
      "step": 2078
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7205892078663084,
      "learning_rate": 1.4685295682532015e-05,
      "loss": 0.7193,
      "step": 2079
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6578369714613123,
      "learning_rate": 1.4680276967931988e-05,
      "loss": 0.7143,
      "step": 2080
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.791678825656993,
      "learning_rate": 1.4675256743377078e-05,
      "loss": 0.712,
      "step": 2081
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7037080030073689,
      "learning_rate": 1.4670235010486915e-05,
      "loss": 0.7164,
      "step": 2082
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8333742208162048,
      "learning_rate": 1.4665211770881609e-05,
      "loss": 0.7469,
      "step": 2083
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7435400897767561,
      "learning_rate": 1.4660187026181766e-05,
      "loss": 0.7516,
      "step": 2084
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6897898505683674,
      "learning_rate": 1.4655160778008478e-05,
      "loss": 0.6828,
      "step": 2085
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7312156866714145,
      "learning_rate": 1.4650133027983308e-05,
      "loss": 0.7124,
      "step": 2086
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6410438746305827,
      "learning_rate": 1.464510377772832e-05,
      "loss": 0.7295,
      "step": 2087
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6987845399657587,
      "learning_rate": 1.4640073028866051e-05,
      "loss": 0.6831,
      "step": 2088
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8269171543888246,
      "learning_rate": 1.4635040783019526e-05,
      "loss": 0.7546,
      "step": 2089
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7348168344345398,
      "learning_rate": 1.4630007041812253e-05,
      "loss": 0.7167,
      "step": 2090
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6604725255771705,
      "learning_rate": 1.4624971806868224e-05,
      "loss": 0.7052,
      "step": 2091
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.685288551038183,
      "learning_rate": 1.4619935079811906e-05,
      "loss": 0.7465,
      "step": 2092
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9715181108376021,
      "learning_rate": 1.4614896862268255e-05,
      "loss": 0.7435,
      "step": 2093
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7541000429819498,
      "learning_rate": 1.4609857155862705e-05,
      "loss": 0.7327,
      "step": 2094
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.64307130244925,
      "learning_rate": 1.4604815962221173e-05,
      "loss": 0.6824,
      "step": 2095
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7246425792713121,
      "learning_rate": 1.4599773282970047e-05,
      "loss": 0.774,
      "step": 2096
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8242266242818043,
      "learning_rate": 1.4594729119736203e-05,
      "loss": 0.8015,
      "step": 2097
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8282775310077742,
      "learning_rate": 1.4589683474146994e-05,
      "loss": 0.7386,
      "step": 2098
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6444631569232817,
      "learning_rate": 1.4584636347830252e-05,
      "loss": 0.7115,
      "step": 2099
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7413012375516175,
      "learning_rate": 1.4579587742414285e-05,
      "loss": 0.7572,
      "step": 2100
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7414763020558572,
      "learning_rate": 1.4574537659527871e-05,
      "loss": 0.7878,
      "step": 2101
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7112403030047766,
      "learning_rate": 1.456948610080028e-05,
      "loss": 0.7637,
      "step": 2102
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8100998752716444,
      "learning_rate": 1.4564433067861248e-05,
      "loss": 0.7854,
      "step": 2103
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7184707305179868,
      "learning_rate": 1.455937856234099e-05,
      "loss": 0.7535,
      "step": 2104
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6921407049473625,
      "learning_rate": 1.4554322585870191e-05,
      "loss": 0.7062,
      "step": 2105
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.752560886208899,
      "learning_rate": 1.4549265140080017e-05,
      "loss": 0.7281,
      "step": 2106
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7335720821344569,
      "learning_rate": 1.4544206226602105e-05,
      "loss": 0.7442,
      "step": 2107
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.819115612290902,
      "learning_rate": 1.4539145847068568e-05,
      "loss": 0.7224,
      "step": 2108
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.676087446659887,
      "learning_rate": 1.4534084003111985e-05,
      "loss": 0.7149,
      "step": 2109
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8698679623577381,
      "learning_rate": 1.4529020696365417e-05,
      "loss": 0.7366,
      "step": 2110
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7296822718264654,
      "learning_rate": 1.452395592846239e-05,
      "loss": 0.7553,
      "step": 2111
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7226862940834334,
      "learning_rate": 1.4518889701036903e-05,
      "loss": 0.7011,
      "step": 2112
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.774451364885461,
      "learning_rate": 1.4513822015723429e-05,
      "loss": 0.6952,
      "step": 2113
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6474507781777041,
      "learning_rate": 1.4508752874156908e-05,
      "loss": 0.6955,
      "step": 2114
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7150905352099216,
      "learning_rate": 1.4503682277972748e-05,
      "loss": 0.706,
      "step": 2115
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7320017860895511,
      "learning_rate": 1.4498610228806831e-05,
      "loss": 0.7738,
      "step": 2116
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7283444232943792,
      "learning_rate": 1.4493536728295512e-05,
      "loss": 0.7216,
      "step": 2117
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8490950344561371,
      "learning_rate": 1.4488461778075595e-05,
      "loss": 0.7999,
      "step": 2118
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7612990881546638,
      "learning_rate": 1.4483385379784374e-05,
      "loss": 0.7671,
      "step": 2119
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.670153630504126,
      "learning_rate": 1.4478307535059599e-05,
      "loss": 0.6728,
      "step": 2120
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6763711162476099,
      "learning_rate": 1.4473228245539485e-05,
      "loss": 0.7434,
      "step": 2121
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6794612391489825,
      "learning_rate": 1.4468147512862725e-05,
      "loss": 0.6968,
      "step": 2122
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6693408464934483,
      "learning_rate": 1.4463065338668463e-05,
      "loss": 0.7183,
      "step": 2123
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7656847148217901,
      "learning_rate": 1.4457981724596311e-05,
      "loss": 0.7301,
      "step": 2124
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6751716409677763,
      "learning_rate": 1.4452896672286355e-05,
      "loss": 0.7048,
      "step": 2125
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6759682343424311,
      "learning_rate": 1.4447810183379138e-05,
      "loss": 0.7134,
      "step": 2126
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7636872107858956,
      "learning_rate": 1.4442722259515667e-05,
      "loss": 0.7361,
      "step": 2127
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.671640834117611,
      "learning_rate": 1.443763290233741e-05,
      "loss": 0.675,
      "step": 2128
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7687935930024894,
      "learning_rate": 1.4432542113486304e-05,
      "loss": 0.7719,
      "step": 2129
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7669356707115488,
      "learning_rate": 1.4427449894604738e-05,
      "loss": 0.7552,
      "step": 2130
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6981495109688766,
      "learning_rate": 1.4422356247335571e-05,
      "loss": 0.7415,
      "step": 2131
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7738253301810663,
      "learning_rate": 1.4417261173322121e-05,
      "loss": 0.7625,
      "step": 2132
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6991950735793045,
      "learning_rate": 1.441216467420816e-05,
      "loss": 0.6858,
      "step": 2133
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7361296106109229,
      "learning_rate": 1.440706675163793e-05,
      "loss": 0.7407,
      "step": 2134
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7182765708144832,
      "learning_rate": 1.4401967407256128e-05,
      "loss": 0.7075,
      "step": 2135
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7110897853428468,
      "learning_rate": 1.4396866642707905e-05,
      "loss": 0.7154,
      "step": 2136
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7615873112146907,
      "learning_rate": 1.4391764459638873e-05,
      "loss": 0.6979,
      "step": 2137
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6774618438561336,
      "learning_rate": 1.4386660859695107e-05,
      "loss": 0.6944,
      "step": 2138
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.710427162542709,
      "learning_rate": 1.4381555844523133e-05,
      "loss": 0.6523,
      "step": 2139
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7965061247270306,
      "learning_rate": 1.4376449415769933e-05,
      "loss": 0.7493,
      "step": 2140
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7362826737814067,
      "learning_rate": 1.437134157508295e-05,
      "loss": 0.7573,
      "step": 2141
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7379877485179942,
      "learning_rate": 1.4366232324110077e-05,
      "loss": 0.7281,
      "step": 2142
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7323668228199085,
      "learning_rate": 1.436112166449967e-05,
      "loss": 0.7151,
      "step": 2143
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8197054712167071,
      "learning_rate": 1.4356009597900526e-05,
      "loss": 0.7302,
      "step": 2144
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7771985844535266,
      "learning_rate": 1.4350896125961915e-05,
      "loss": 0.6657,
      "step": 2145
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6406337548806971,
      "learning_rate": 1.434578125033354e-05,
      "loss": 0.7141,
      "step": 2146
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7458534633175371,
      "learning_rate": 1.4340664972665572e-05,
      "loss": 0.6845,
      "step": 2147
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0118349968298501,
      "learning_rate": 1.4335547294608625e-05,
      "loss": 0.8524,
      "step": 2148
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6823615432896413,
      "learning_rate": 1.4330428217813773e-05,
      "loss": 0.707,
      "step": 2149
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8965102553002238,
      "learning_rate": 1.4325307743932535e-05,
      "loss": 0.6945,
      "step": 2150
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8369569635647724,
      "learning_rate": 1.4320185874616881e-05,
      "loss": 0.7249,
      "step": 2151
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.615975614873533,
      "learning_rate": 1.4315062611519235e-05,
      "loss": 0.7059,
      "step": 2152
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8279652513650405,
      "learning_rate": 1.4309937956292464e-05,
      "loss": 0.7274,
      "step": 2153
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.810607831456142,
      "learning_rate": 1.4304811910589896e-05,
      "loss": 0.7208,
      "step": 2154
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8649595502575466,
      "learning_rate": 1.4299684476065294e-05,
      "loss": 0.8114,
      "step": 2155
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8958670051889356,
      "learning_rate": 1.4294555654372876e-05,
      "loss": 0.7735,
      "step": 2156
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7543620629254074,
      "learning_rate": 1.4289425447167308e-05,
      "loss": 0.7523,
      "step": 2157
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8028198830588136,
      "learning_rate": 1.4284293856103703e-05,
      "loss": 0.7603,
      "step": 2158
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8255733469081601,
      "learning_rate": 1.4279160882837622e-05,
      "loss": 0.7385,
      "step": 2159
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6577618161735683,
      "learning_rate": 1.427402652902506e-05,
      "loss": 0.6811,
      "step": 2160
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7454285647093175,
      "learning_rate": 1.4268890796322475e-05,
      "loss": 0.7536,
      "step": 2161
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1067409559462653,
      "learning_rate": 1.4263753686386754e-05,
      "loss": 0.804,
      "step": 2162
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6878657536264345,
      "learning_rate": 1.4258615200875243e-05,
      "loss": 0.7047,
      "step": 2163
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6927159931381263,
      "learning_rate": 1.4253475341445722e-05,
      "loss": 0.7115,
      "step": 2164
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6792245111177933,
      "learning_rate": 1.4248334109756414e-05,
      "loss": 0.7355,
      "step": 2165
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9211252310706203,
      "learning_rate": 1.424319150746599e-05,
      "loss": 0.8263,
      "step": 2166
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6694415499479912,
      "learning_rate": 1.4238047536233559e-05,
      "loss": 0.7339,
      "step": 2167
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6727164376016272,
      "learning_rate": 1.4232902197718679e-05,
      "loss": 0.7057,
      "step": 2168
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.725865785069817,
      "learning_rate": 1.4227755493581335e-05,
      "loss": 0.7412,
      "step": 2169
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7236970601136149,
      "learning_rate": 1.4222607425481967e-05,
      "loss": 0.753,
      "step": 2170
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.736402521612048,
      "learning_rate": 1.4217457995081445e-05,
      "loss": 0.7275,
      "step": 2171
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6927836227479977,
      "learning_rate": 1.4212307204041088e-05,
      "loss": 0.7081,
      "step": 2172
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7013498501092824,
      "learning_rate": 1.4207155054022644e-05,
      "loss": 0.7207,
      "step": 2173
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7476268084614962,
      "learning_rate": 1.4202001546688302e-05,
      "loss": 0.7543,
      "step": 2174
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.713014867693021,
      "learning_rate": 1.4196846683700694e-05,
      "loss": 0.7427,
      "step": 2175
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6502133540350608,
      "learning_rate": 1.419169046672289e-05,
      "loss": 0.6913,
      "step": 2176
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7023877971994874,
      "learning_rate": 1.4186532897418388e-05,
      "loss": 0.6989,
      "step": 2177
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6749938253721791,
      "learning_rate": 1.4181373977451127e-05,
      "loss": 0.7242,
      "step": 2178
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6873977249395283,
      "learning_rate": 1.4176213708485486e-05,
      "loss": 0.728,
      "step": 2179
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6639708022496743,
      "learning_rate": 1.4171052092186271e-05,
      "loss": 0.671,
      "step": 2180
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.558632361017741,
      "learning_rate": 1.4165889130218731e-05,
      "loss": 0.733,
      "step": 2181
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6548470158696784,
      "learning_rate": 1.4160724824248542e-05,
      "loss": 0.7067,
      "step": 2182
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6774548391307514,
      "learning_rate": 1.4155559175941818e-05,
      "loss": 0.7326,
      "step": 2183
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6490306070569153,
      "learning_rate": 1.4150392186965105e-05,
      "loss": 0.7141,
      "step": 2184
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6703306060815707,
      "learning_rate": 1.414522385898538e-05,
      "loss": 0.682,
      "step": 2185
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7164795312899472,
      "learning_rate": 1.4140054193670054e-05,
      "loss": 0.7331,
      "step": 2186
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.814294821071235,
      "learning_rate": 1.413488319268697e-05,
      "loss": 0.7581,
      "step": 2187
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7127136595570086,
      "learning_rate": 1.4129710857704399e-05,
      "loss": 0.7514,
      "step": 2188
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7234094790129846,
      "learning_rate": 1.4124537190391043e-05,
      "loss": 0.7453,
      "step": 2189
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7689176100932479,
      "learning_rate": 1.411936219241604e-05,
      "loss": 0.6853,
      "step": 2190
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8394851424791324,
      "learning_rate": 1.4114185865448948e-05,
      "loss": 0.7476,
      "step": 2191
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6662651229452621,
      "learning_rate": 1.4109008211159757e-05,
      "loss": 0.7457,
      "step": 2192
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8229621000244826,
      "learning_rate": 1.4103829231218891e-05,
      "loss": 0.6943,
      "step": 2193
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7392247100360231,
      "learning_rate": 1.409864892729719e-05,
      "loss": 0.7397,
      "step": 2194
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7519823621503527,
      "learning_rate": 1.4093467301065934e-05,
      "loss": 0.6767,
      "step": 2195
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8366143938971783,
      "learning_rate": 1.4088284354196823e-05,
      "loss": 0.7293,
      "step": 2196
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6400409102172189,
      "learning_rate": 1.4083100088361978e-05,
      "loss": 0.7282,
      "step": 2197
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7966635013337957,
      "learning_rate": 1.4077914505233958e-05,
      "loss": 0.7735,
      "step": 2198
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8070313179849543,
      "learning_rate": 1.4072727606485738e-05,
      "loss": 0.7778,
      "step": 2199
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8117857249524921,
      "learning_rate": 1.4067539393790718e-05,
      "loss": 0.7778,
      "step": 2200
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6737261158845725,
      "learning_rate": 1.4062349868822724e-05,
      "loss": 0.7174,
      "step": 2201
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7779700365313287,
      "learning_rate": 1.4057159033256007e-05,
      "loss": 0.7756,
      "step": 2202
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6471716813349117,
      "learning_rate": 1.4051966888765236e-05,
      "loss": 0.6882,
      "step": 2203
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7719981424125798,
      "learning_rate": 1.4046773437025508e-05,
      "loss": 0.6995,
      "step": 2204
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7014290824295639,
      "learning_rate": 1.4041578679712335e-05,
      "loss": 0.7308,
      "step": 2205
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6722708372442774,
      "learning_rate": 1.4036382618501656e-05,
      "loss": 0.7172,
      "step": 2206
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8289432702898906,
      "learning_rate": 1.4031185255069828e-05,
      "loss": 0.724,
      "step": 2207
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6613573593154712,
      "learning_rate": 1.402598659109363e-05,
      "loss": 0.7254,
      "step": 2208
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7072426905624368,
      "learning_rate": 1.402078662825026e-05,
      "loss": 0.7515,
      "step": 2209
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.744069982082698,
      "learning_rate": 1.4015585368217332e-05,
      "loss": 0.7401,
      "step": 2210
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6443364128101892,
      "learning_rate": 1.4010382812672884e-05,
      "loss": 0.7147,
      "step": 2211
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8001396555127299,
      "learning_rate": 1.4005178963295362e-05,
      "loss": 0.7931,
      "step": 2212
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7497790115729023,
      "learning_rate": 1.3999973821763648e-05,
      "loss": 0.7163,
      "step": 2213
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0831385866205998,
      "learning_rate": 1.3994767389757018e-05,
      "loss": 0.7411,
      "step": 2214
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7579261282091356,
      "learning_rate": 1.398955966895518e-05,
      "loss": 0.708,
      "step": 2215
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7636624948343397,
      "learning_rate": 1.398435066103825e-05,
      "loss": 0.7267,
      "step": 2216
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7288568140404238,
      "learning_rate": 1.397914036768677e-05,
      "loss": 0.7495,
      "step": 2217
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7858518041283405,
      "learning_rate": 1.3973928790581683e-05,
      "loss": 0.7775,
      "step": 2218
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7331830269329576,
      "learning_rate": 1.3968715931404352e-05,
      "loss": 0.7199,
      "step": 2219
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6926937823937076,
      "learning_rate": 1.396350179183656e-05,
      "loss": 0.7105,
      "step": 2220
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7310060836181793,
      "learning_rate": 1.395828637356049e-05,
      "loss": 0.7555,
      "step": 2221
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7457909996185009,
      "learning_rate": 1.3953069678258748e-05,
      "loss": 0.7396,
      "step": 2222
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7623829229826191,
      "learning_rate": 1.394785170761435e-05,
      "loss": 0.7242,
      "step": 2223
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8162033561282758,
      "learning_rate": 1.3942632463310719e-05,
      "loss": 0.7506,
      "step": 2224
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8172242138626911,
      "learning_rate": 1.3937411947031697e-05,
      "loss": 0.7495,
      "step": 2225
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6872305997182108,
      "learning_rate": 1.3932190160461521e-05,
      "loss": 0.7067,
      "step": 2226
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6056912379667825,
      "learning_rate": 1.3926967105284859e-05,
      "loss": 0.6701,
      "step": 2227
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6734506037071749,
      "learning_rate": 1.3921742783186774e-05,
      "loss": 0.6637,
      "step": 2228
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7997869417281829,
      "learning_rate": 1.3916517195852736e-05,
      "loss": 0.7459,
      "step": 2229
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7006362701919371,
      "learning_rate": 1.3911290344968636e-05,
      "loss": 0.7293,
      "step": 2230
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6930022120424849,
      "learning_rate": 1.3906062232220762e-05,
      "loss": 0.7173,
      "step": 2231
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7477081356784656,
      "learning_rate": 1.3900832859295814e-05,
      "loss": 0.787,
      "step": 2232
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7293468746249121,
      "learning_rate": 1.389560222788089e-05,
      "loss": 0.6971,
      "step": 2233
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7819266141437773,
      "learning_rate": 1.389037033966351e-05,
      "loss": 0.7386,
      "step": 2234
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6405687236251681,
      "learning_rate": 1.3885137196331581e-05,
      "loss": 0.7174,
      "step": 2235
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6657764882117962,
      "learning_rate": 1.3879902799573435e-05,
      "loss": 0.7326,
      "step": 2236
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7238342099991129,
      "learning_rate": 1.3874667151077787e-05,
      "loss": 0.7364,
      "step": 2237
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8041460106161582,
      "learning_rate": 1.386943025253377e-05,
      "loss": 0.7612,
      "step": 2238
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6775569987776883,
      "learning_rate": 1.3864192105630916e-05,
      "loss": 0.7435,
      "step": 2239
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.688475620165104,
      "learning_rate": 1.3858952712059166e-05,
      "loss": 0.7232,
      "step": 2240
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.730748122481352,
      "learning_rate": 1.385371207350885e-05,
      "loss": 0.7609,
      "step": 2241
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6721459689899024,
      "learning_rate": 1.3848470191670709e-05,
      "loss": 0.7521,
      "step": 2242
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6829281111220477,
      "learning_rate": 1.3843227068235887e-05,
      "loss": 0.7246,
      "step": 2243
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.719431487294415,
      "learning_rate": 1.383798270489592e-05,
      "loss": 0.7001,
      "step": 2244
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7704146085467647,
      "learning_rate": 1.3832737103342753e-05,
      "loss": 0.7399,
      "step": 2245
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7172149167407539,
      "learning_rate": 1.3827490265268724e-05,
      "loss": 0.691,
      "step": 2246
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8023821990002282,
      "learning_rate": 1.3822242192366572e-05,
      "loss": 0.7545,
      "step": 2247
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7251059398331451,
      "learning_rate": 1.3816992886329436e-05,
      "loss": 0.7152,
      "step": 2248
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7070403647928714,
      "learning_rate": 1.3811742348850853e-05,
      "loss": 0.7104,
      "step": 2249
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.806052418890586,
      "learning_rate": 1.3806490581624755e-05,
      "loss": 0.7226,
      "step": 2250
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6489841013927836,
      "learning_rate": 1.3801237586345466e-05,
      "loss": 0.707,
      "step": 2251
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6884232682118497,
      "learning_rate": 1.379598336470772e-05,
      "loss": 0.6928,
      "step": 2252
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7338242566353534,
      "learning_rate": 1.3790727918406632e-05,
      "loss": 0.7354,
      "step": 2253
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7402276163874215,
      "learning_rate": 1.3785471249137722e-05,
      "loss": 0.7325,
      "step": 2254
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7846923505488829,
      "learning_rate": 1.3780213358596903e-05,
      "loss": 0.7059,
      "step": 2255
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6588803769851069,
      "learning_rate": 1.3774954248480473e-05,
      "loss": 0.6978,
      "step": 2256
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7617396194165048,
      "learning_rate": 1.3769693920485136e-05,
      "loss": 0.7222,
      "step": 2257
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6644974386060717,
      "learning_rate": 1.3764432376307984e-05,
      "loss": 0.6724,
      "step": 2258
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.681132455837786,
      "learning_rate": 1.3759169617646497e-05,
      "loss": 0.7602,
      "step": 2259
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6829738574675814,
      "learning_rate": 1.3753905646198553e-05,
      "loss": 0.6938,
      "step": 2260
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6444548185000165,
      "learning_rate": 1.3748640463662419e-05,
      "loss": 0.7201,
      "step": 2261
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7077586839512241,
      "learning_rate": 1.374337407173675e-05,
      "loss": 0.6681,
      "step": 2262
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7298465407193,
      "learning_rate": 1.3738106472120601e-05,
      "loss": 0.7302,
      "step": 2263
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7799323196081233,
      "learning_rate": 1.37328376665134e-05,
      "loss": 0.7835,
      "step": 2264
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7033756057283252,
      "learning_rate": 1.3727567656614979e-05,
      "loss": 0.677,
      "step": 2265
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6719613792347849,
      "learning_rate": 1.3722296444125554e-05,
      "loss": 0.6985,
      "step": 2266
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7293412746896747,
      "learning_rate": 1.3717024030745722e-05,
      "loss": 0.7312,
      "step": 2267
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7098142813694521,
      "learning_rate": 1.371175041817648e-05,
      "loss": 0.746,
      "step": 2268
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6944520675415881,
      "learning_rate": 1.3706475608119203e-05,
      "loss": 0.7456,
      "step": 2269
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6773032248286235,
      "learning_rate": 1.370119960227565e-05,
      "loss": 0.7023,
      "step": 2270
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6314803769392858,
      "learning_rate": 1.3695922402347976e-05,
      "loss": 0.7066,
      "step": 2271
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7221805694145836,
      "learning_rate": 1.3690644010038719e-05,
      "loss": 0.6949,
      "step": 2272
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7182170098088343,
      "learning_rate": 1.368536442705079e-05,
      "loss": 0.7066,
      "step": 2273
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6796548462858892,
      "learning_rate": 1.3680083655087495e-05,
      "loss": 0.7155,
      "step": 2274
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7176345852811481,
      "learning_rate": 1.3674801695852524e-05,
      "loss": 0.7408,
      "step": 2275
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7005792145822987,
      "learning_rate": 1.3669518551049945e-05,
      "loss": 0.7345,
      "step": 2276
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6845887182008077,
      "learning_rate": 1.3664234222384211e-05,
      "loss": 0.7373,
      "step": 2277
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6719471178066246,
      "learning_rate": 1.3658948711560158e-05,
      "loss": 0.7587,
      "step": 2278
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.636217088170897,
      "learning_rate": 1.3653662020282999e-05,
      "loss": 0.7262,
      "step": 2279
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7570233619485549,
      "learning_rate": 1.364837415025833e-05,
      "loss": 0.753,
      "step": 2280
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7141796310135061,
      "learning_rate": 1.3643085103192136e-05,
      "loss": 0.7267,
      "step": 2281
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.602311214798034,
      "learning_rate": 1.3637794880790765e-05,
      "loss": 0.7017,
      "step": 2282
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6807265309764913,
      "learning_rate": 1.3632503484760956e-05,
      "loss": 0.7241,
      "step": 2283
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6818456355111752,
      "learning_rate": 1.3627210916809824e-05,
      "loss": 0.7025,
      "step": 2284
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8326223882686892,
      "learning_rate": 1.3621917178644858e-05,
      "loss": 0.7359,
      "step": 2285
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6916724191951673,
      "learning_rate": 1.3616622271973934e-05,
      "loss": 0.7221,
      "step": 2286
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7317106665765071,
      "learning_rate": 1.3611326198505301e-05,
      "loss": 0.8041,
      "step": 2287
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7541448686933699,
      "learning_rate": 1.3606028959947575e-05,
      "loss": 0.7426,
      "step": 2288
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6134635636320508,
      "learning_rate": 1.3600730558009758e-05,
      "loss": 0.6492,
      "step": 2289
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7289292781280018,
      "learning_rate": 1.359543099440123e-05,
      "loss": 0.7116,
      "step": 2290
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6889961392992617,
      "learning_rate": 1.3590130270831733e-05,
      "loss": 0.7212,
      "step": 2291
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6527038240610591,
      "learning_rate": 1.3584828389011395e-05,
      "loss": 0.7257,
      "step": 2292
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6565879509312308,
      "learning_rate": 1.3579525350650714e-05,
      "loss": 0.6902,
      "step": 2293
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7761210606859039,
      "learning_rate": 1.3574221157460559e-05,
      "loss": 0.7505,
      "step": 2294
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8139854287245177,
      "learning_rate": 1.3568915811152177e-05,
      "loss": 0.7086,
      "step": 2295
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6847015351290453,
      "learning_rate": 1.3563609313437178e-05,
      "loss": 0.6622,
      "step": 2296
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6887425131409302,
      "learning_rate": 1.355830166602755e-05,
      "loss": 0.6938,
      "step": 2297
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6980458836383652,
      "learning_rate": 1.3552992870635653e-05,
      "loss": 0.7285,
      "step": 2298
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7195627159181451,
      "learning_rate": 1.3547682928974215e-05,
      "loss": 0.7694,
      "step": 2299
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7320559431071314,
      "learning_rate": 1.3542371842756332e-05,
      "loss": 0.7744,
      "step": 2300
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7299450934004963,
      "learning_rate": 1.353705961369547e-05,
      "loss": 0.7392,
      "step": 2301
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7400449050016393,
      "learning_rate": 1.353174624350547e-05,
      "loss": 0.6959,
      "step": 2302
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6394718401734768,
      "learning_rate": 1.352643173390053e-05,
      "loss": 0.7255,
      "step": 2303
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8205883510853714,
      "learning_rate": 1.3521116086595223e-05,
      "loss": 0.7543,
      "step": 2304
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7102579530228583,
      "learning_rate": 1.3515799303304491e-05,
      "loss": 0.7343,
      "step": 2305
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6681380752265207,
      "learning_rate": 1.3510481385743634e-05,
      "loss": 0.7182,
      "step": 2306
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6993928777601985,
      "learning_rate": 1.350516233562833e-05,
      "loss": 0.7038,
      "step": 2307
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6497288147542657,
      "learning_rate": 1.3499842154674607e-05,
      "loss": 0.6774,
      "step": 2308
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6921281276940463,
      "learning_rate": 1.3494520844598873e-05,
      "loss": 0.727,
      "step": 2309
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6676333724645953,
      "learning_rate": 1.3489198407117891e-05,
      "loss": 0.7673,
      "step": 2310
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6814929789318541,
      "learning_rate": 1.3483874843948787e-05,
      "loss": 0.7342,
      "step": 2311
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6755613185210761,
      "learning_rate": 1.3478550156809058e-05,
      "loss": 0.7031,
      "step": 2312
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7640544258591926,
      "learning_rate": 1.3473224347416561e-05,
      "loss": 0.7551,
      "step": 2313
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7768728497742771,
      "learning_rate": 1.3467897417489506e-05,
      "loss": 0.7529,
      "step": 2314
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7851402742716136,
      "learning_rate": 1.3462569368746477e-05,
      "loss": 0.8019,
      "step": 2315
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7376089125873085,
      "learning_rate": 1.3457240202906412e-05,
      "loss": 0.7009,
      "step": 2316
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6305607068461869,
      "learning_rate": 1.3451909921688609e-05,
      "loss": 0.6988,
      "step": 2317
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6802142969079492,
      "learning_rate": 1.344657852681273e-05,
      "loss": 0.7048,
      "step": 2318
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.642361316837244,
      "learning_rate": 1.3441246019998797e-05,
      "loss": 0.7455,
      "step": 2319
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6109534143501324,
      "learning_rate": 1.3435912402967182e-05,
      "loss": 0.7062,
      "step": 2320
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6987157452615502,
      "learning_rate": 1.3430577677438622e-05,
      "loss": 0.716,
      "step": 2321
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7232775043139741,
      "learning_rate": 1.3425241845134213e-05,
      "loss": 0.739,
      "step": 2322
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.708108096369702,
      "learning_rate": 1.3419904907775407e-05,
      "loss": 0.7533,
      "step": 2323
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7270532314010189,
      "learning_rate": 1.3414566867084007e-05,
      "loss": 0.7363,
      "step": 2324
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7265672226783425,
      "learning_rate": 1.3409227724782179e-05,
      "loss": 0.7297,
      "step": 2325
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7183417346709867,
      "learning_rate": 1.340388748259244e-05,
      "loss": 0.7611,
      "step": 2326
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8350082376224714,
      "learning_rate": 1.3398546142237665e-05,
      "loss": 0.7932,
      "step": 2327
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6678368715800845,
      "learning_rate": 1.3393203705441082e-05,
      "loss": 0.6724,
      "step": 2328
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.65571157845097,
      "learning_rate": 1.338786017392627e-05,
      "loss": 0.6844,
      "step": 2329
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7221769744702788,
      "learning_rate": 1.3382515549417166e-05,
      "loss": 0.6925,
      "step": 2330
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6747626071102608,
      "learning_rate": 1.3377169833638054e-05,
      "loss": 0.7325,
      "step": 2331
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7086110941316548,
      "learning_rate": 1.337182302831358e-05,
      "loss": 0.7092,
      "step": 2332
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6898999615026636,
      "learning_rate": 1.3366475135168726e-05,
      "loss": 0.6918,
      "step": 2333
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9455048736264099,
      "learning_rate": 1.3361126155928843e-05,
      "loss": 0.7754,
      "step": 2334
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6959739710528368,
      "learning_rate": 1.3355776092319613e-05,
      "loss": 0.7313,
      "step": 2335
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7120140040097467,
      "learning_rate": 1.3350424946067086e-05,
      "loss": 0.7081,
      "step": 2336
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6958876601767419,
      "learning_rate": 1.334507271889765e-05,
      "loss": 0.6935,
      "step": 2337
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7715670441318386,
      "learning_rate": 1.3339719412538046e-05,
      "loss": 0.7411,
      "step": 2338
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7045963985282403,
      "learning_rate": 1.3334365028715364e-05,
      "loss": 0.7576,
      "step": 2339
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7812009781144019,
      "learning_rate": 1.3329009569157035e-05,
      "loss": 0.8024,
      "step": 2340
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7219785546088793,
      "learning_rate": 1.3323653035590844e-05,
      "loss": 0.7399,
      "step": 2341
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7569365508190538,
      "learning_rate": 1.3318295429744921e-05,
      "loss": 0.785,
      "step": 2342
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7034514282212251,
      "learning_rate": 1.3312936753347745e-05,
      "loss": 0.6875,
      "step": 2343
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6776200322476766,
      "learning_rate": 1.3307577008128128e-05,
      "loss": 0.6981,
      "step": 2344
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6865018279972217,
      "learning_rate": 1.3302216195815244e-05,
      "loss": 0.6841,
      "step": 2345
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6699932323048641,
      "learning_rate": 1.3296854318138596e-05,
      "loss": 0.6978,
      "step": 2346
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.641702278651155,
      "learning_rate": 1.3291491376828043e-05,
      "loss": 0.7442,
      "step": 2347
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7289600109619448,
      "learning_rate": 1.328612737361378e-05,
      "loss": 0.703,
      "step": 2348
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6891202411670234,
      "learning_rate": 1.328076231022634e-05,
      "loss": 0.7304,
      "step": 2349
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7023429402195688,
      "learning_rate": 1.3275396188396615e-05,
      "loss": 0.6693,
      "step": 2350
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7099527556660274,
      "learning_rate": 1.3270029009855822e-05,
      "loss": 0.7245,
      "step": 2351
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8222040338014409,
      "learning_rate": 1.3264660776335523e-05,
      "loss": 0.8539,
      "step": 2352
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7646090840450225,
      "learning_rate": 1.3259291489567626e-05,
      "loss": 0.752,
      "step": 2353
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7077004489708681,
      "learning_rate": 1.325392115128437e-05,
      "loss": 0.6861,
      "step": 2354
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8989779757035115,
      "learning_rate": 1.3248549763218345e-05,
      "loss": 0.8174,
      "step": 2355
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7185680083867938,
      "learning_rate": 1.3243177327102465e-05,
      "loss": 0.7093,
      "step": 2356
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.758154145144408,
      "learning_rate": 1.3237803844669997e-05,
      "loss": 0.7485,
      "step": 2357
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7089891436576027,
      "learning_rate": 1.3232429317654531e-05,
      "loss": 0.7692,
      "step": 2358
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6541759137963825,
      "learning_rate": 1.3227053747790006e-05,
      "loss": 0.6973,
      "step": 2359
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8877253688383792,
      "learning_rate": 1.3221677136810698e-05,
      "loss": 0.7476,
      "step": 2360
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7286850949706205,
      "learning_rate": 1.3216299486451201e-05,
      "loss": 0.719,
      "step": 2361
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5874420332451844,
      "learning_rate": 1.3210920798446467e-05,
      "loss": 0.6889,
      "step": 2362
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6677226251942857,
      "learning_rate": 1.3205541074531769e-05,
      "loss": 0.6966,
      "step": 2363
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7296709813106913,
      "learning_rate": 1.320016031644272e-05,
      "loss": 0.7243,
      "step": 2364
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7234033603808813,
      "learning_rate": 1.3194778525915262e-05,
      "loss": 0.7068,
      "step": 2365
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6704867789154204,
      "learning_rate": 1.3189395704685677e-05,
      "loss": 0.7263,
      "step": 2366
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6477864199136458,
      "learning_rate": 1.3184011854490569e-05,
      "loss": 0.69,
      "step": 2367
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6293910138917186,
      "learning_rate": 1.3178626977066885e-05,
      "loss": 0.663,
      "step": 2368
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.689976345552514,
      "learning_rate": 1.3173241074151902e-05,
      "loss": 0.7348,
      "step": 2369
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9752829974008477,
      "learning_rate": 1.3167854147483214e-05,
      "loss": 0.7333,
      "step": 2370
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7598553515762735,
      "learning_rate": 1.3162466198798765e-05,
      "loss": 0.731,
      "step": 2371
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7712509447109908,
      "learning_rate": 1.3157077229836817e-05,
      "loss": 0.7273,
      "step": 2372
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6633454891202795,
      "learning_rate": 1.3151687242335963e-05,
      "loss": 0.6978,
      "step": 2373
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7074520448408939,
      "learning_rate": 1.3146296238035124e-05,
      "loss": 0.7067,
      "step": 2374
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6986050377459303,
      "learning_rate": 1.3140904218673552e-05,
      "loss": 0.7302,
      "step": 2375
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6510284247257021,
      "learning_rate": 1.3135511185990825e-05,
      "loss": 0.683,
      "step": 2376
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7454568775892058,
      "learning_rate": 1.3130117141726848e-05,
      "loss": 0.7203,
      "step": 2377
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7621891335488251,
      "learning_rate": 1.312472208762185e-05,
      "loss": 0.7597,
      "step": 2378
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7190573198464152,
      "learning_rate": 1.311932602541639e-05,
      "loss": 0.7367,
      "step": 2379
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6289054364108255,
      "learning_rate": 1.3113928956851351e-05,
      "loss": 0.7137,
      "step": 2380
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6148626430202413,
      "learning_rate": 1.3108530883667936e-05,
      "loss": 0.6635,
      "step": 2381
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6970307614387126,
      "learning_rate": 1.3103131807607677e-05,
      "loss": 0.7539,
      "step": 2382
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.67890119872192,
      "learning_rate": 1.3097731730412433e-05,
      "loss": 0.7284,
      "step": 2383
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.64659510831162,
      "learning_rate": 1.3092330653824377e-05,
      "loss": 0.704,
      "step": 2384
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7098230002104676,
      "learning_rate": 1.308692857958601e-05,
      "loss": 0.7357,
      "step": 2385
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6781010871287947,
      "learning_rate": 1.3081525509440151e-05,
      "loss": 0.6705,
      "step": 2386
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6539325876131098,
      "learning_rate": 1.3076121445129948e-05,
      "loss": 0.7217,
      "step": 2387
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6299518019084265,
      "learning_rate": 1.307071638839886e-05,
      "loss": 0.7032,
      "step": 2388
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6140045561885484,
      "learning_rate": 1.3065310340990678e-05,
      "loss": 0.6959,
      "step": 2389
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7841518040239945,
      "learning_rate": 1.3059903304649497e-05,
      "loss": 0.7269,
      "step": 2390
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7584955920690053,
      "learning_rate": 1.3054495281119744e-05,
      "loss": 0.7716,
      "step": 2391
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.002300221513755,
      "learning_rate": 1.304908627214616e-05,
      "loss": 0.7336,
      "step": 2392
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6660963648320816,
      "learning_rate": 1.3043676279473804e-05,
      "loss": 0.6923,
      "step": 2393
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7142337165407261,
      "learning_rate": 1.303826530484805e-05,
      "loss": 0.6882,
      "step": 2394
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6923757537343708,
      "learning_rate": 1.3032853350014597e-05,
      "loss": 0.6851,
      "step": 2395
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7201847977142652,
      "learning_rate": 1.3027440416719449e-05,
      "loss": 0.7366,
      "step": 2396
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.752411971654645,
      "learning_rate": 1.3022026506708932e-05,
      "loss": 0.736,
      "step": 2397
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.764779364947937,
      "learning_rate": 1.3016611621729686e-05,
      "loss": 0.7379,
      "step": 2398
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.693408803568944,
      "learning_rate": 1.3011195763528666e-05,
      "loss": 0.7272,
      "step": 2399
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6614633779721318,
      "learning_rate": 1.3005778933853142e-05,
      "loss": 0.7094,
      "step": 2400
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6938519997650747,
      "learning_rate": 1.3000361134450699e-05,
      "loss": 0.6871,
      "step": 2401
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6828814091619929,
      "learning_rate": 1.2994942367069224e-05,
      "loss": 0.7223,
      "step": 2402
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7413651800094773,
      "learning_rate": 1.2989522633456929e-05,
      "loss": 0.7404,
      "step": 2403
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.78029943101753,
      "learning_rate": 1.2984101935362335e-05,
      "loss": 0.8184,
      "step": 2404
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6601770425725246,
      "learning_rate": 1.2978680274534268e-05,
      "loss": 0.6828,
      "step": 2405
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6426819580390787,
      "learning_rate": 1.2973257652721869e-05,
      "loss": 0.6713,
      "step": 2406
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6494402765431891,
      "learning_rate": 1.2967834071674594e-05,
      "loss": 0.6904,
      "step": 2407
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5751640122241534,
      "learning_rate": 1.2962409533142198e-05,
      "loss": 0.6739,
      "step": 2408
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7067622840667034,
      "learning_rate": 1.295698403887475e-05,
      "loss": 0.7574,
      "step": 2409
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.796337010696576,
      "learning_rate": 1.2951557590622634e-05,
      "loss": 0.7512,
      "step": 2410
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6287948195615669,
      "learning_rate": 1.2946130190136528e-05,
      "loss": 0.6689,
      "step": 2411
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6901663629242247,
      "learning_rate": 1.2940701839167424e-05,
      "loss": 0.7173,
      "step": 2412
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7811199472969944,
      "learning_rate": 1.293527253946663e-05,
      "loss": 0.7005,
      "step": 2413
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8029462283079076,
      "learning_rate": 1.2929842292785742e-05,
      "loss": 0.7232,
      "step": 2414
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7275829239096612,
      "learning_rate": 1.2924411100876677e-05,
      "loss": 0.7507,
      "step": 2415
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7409321440367378,
      "learning_rate": 1.2918978965491646e-05,
      "loss": 0.6871,
      "step": 2416
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7757034558830632,
      "learning_rate": 1.291354588838317e-05,
      "loss": 0.7232,
      "step": 2417
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6757682004760615,
      "learning_rate": 1.2908111871304076e-05,
      "loss": 0.7054,
      "step": 2418
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7105448079641924,
      "learning_rate": 1.2902676916007491e-05,
      "loss": 0.7081,
      "step": 2419
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6842502169586996,
      "learning_rate": 1.2897241024246838e-05,
      "loss": 0.7021,
      "step": 2420
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7104564301303338,
      "learning_rate": 1.2891804197775859e-05,
      "loss": 0.7475,
      "step": 2421
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.713608135949888,
      "learning_rate": 1.2886366438348579e-05,
      "loss": 0.6939,
      "step": 2422
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7570979133305654,
      "learning_rate": 1.2880927747719333e-05,
      "loss": 0.6776,
      "step": 2423
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9041649931138871,
      "learning_rate": 1.2875488127642765e-05,
      "loss": 0.8052,
      "step": 2424
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7910652776050097,
      "learning_rate": 1.2870047579873801e-05,
      "loss": 0.7609,
      "step": 2425
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6779380963168343,
      "learning_rate": 1.2864606106167679e-05,
      "loss": 0.7071,
      "step": 2426
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6709239562566711,
      "learning_rate": 1.2859163708279929e-05,
      "loss": 0.6886,
      "step": 2427
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7485432973989077,
      "learning_rate": 1.2853720387966386e-05,
      "loss": 0.7264,
      "step": 2428
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7509066306920159,
      "learning_rate": 1.2848276146983172e-05,
      "loss": 0.7525,
      "step": 2429
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6191768795054967,
      "learning_rate": 1.284283098708672e-05,
      "loss": 0.7084,
      "step": 2430
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7278428127082829,
      "learning_rate": 1.2837384910033748e-05,
      "loss": 0.7076,
      "step": 2431
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6809586130116474,
      "learning_rate": 1.2831937917581272e-05,
      "loss": 0.7115,
      "step": 2432
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7222698473575949,
      "learning_rate": 1.2826490011486615e-05,
      "loss": 0.7184,
      "step": 2433
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6527514472304208,
      "learning_rate": 1.282104119350737e-05,
      "loss": 0.7034,
      "step": 2434
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.664847461203532,
      "learning_rate": 1.2815591465401452e-05,
      "loss": 0.7024,
      "step": 2435
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6553442758486085,
      "learning_rate": 1.2810140828927051e-05,
      "loss": 0.6794,
      "step": 2436
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6785737349385125,
      "learning_rate": 1.2804689285842657e-05,
      "loss": 0.7043,
      "step": 2437
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7017255079249497,
      "learning_rate": 1.279923683790705e-05,
      "loss": 0.7652,
      "step": 2438
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8252900337510232,
      "learning_rate": 1.2793783486879308e-05,
      "loss": 0.7272,
      "step": 2439
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5848309798228564,
      "learning_rate": 1.2788329234518788e-05,
      "loss": 0.6491,
      "step": 2440
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6897299214375676,
      "learning_rate": 1.2782874082585153e-05,
      "loss": 0.7262,
      "step": 2441
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7193473638161236,
      "learning_rate": 1.2777418032838348e-05,
      "loss": 0.7284,
      "step": 2442
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7108282565069285,
      "learning_rate": 1.27719610870386e-05,
      "loss": 0.7192,
      "step": 2443
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.703275366549796,
      "learning_rate": 1.2766503246946442e-05,
      "loss": 0.7321,
      "step": 2444
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6863627906528372,
      "learning_rate": 1.2761044514322685e-05,
      "loss": 0.7106,
      "step": 2445
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6763514553462665,
      "learning_rate": 1.2755584890928427e-05,
      "loss": 0.7242,
      "step": 2446
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8751340099590099,
      "learning_rate": 1.2750124378525062e-05,
      "loss": 0.7907,
      "step": 2447
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7095203325185853,
      "learning_rate": 1.2744662978874262e-05,
      "loss": 0.7013,
      "step": 2448
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.62491551319793,
      "learning_rate": 1.2739200693737985e-05,
      "loss": 0.7283,
      "step": 2449
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7932121180983142,
      "learning_rate": 1.273373752487848e-05,
      "loss": 0.7401,
      "step": 2450
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6962371703502113,
      "learning_rate": 1.2728273474058285e-05,
      "loss": 0.7297,
      "step": 2451
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6891702322059048,
      "learning_rate": 1.2722808543040208e-05,
      "loss": 0.7398,
      "step": 2452
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7043401245707164,
      "learning_rate": 1.2717342733587354e-05,
      "loss": 0.6975,
      "step": 2453
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5976413747727971,
      "learning_rate": 1.2711876047463107e-05,
      "loss": 0.7013,
      "step": 2454
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7863354526885893,
      "learning_rate": 1.2706408486431133e-05,
      "loss": 0.8211,
      "step": 2455
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7646308610693818,
      "learning_rate": 1.2700940052255382e-05,
      "loss": 0.7129,
      "step": 2456
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7476039486790339,
      "learning_rate": 1.2695470746700086e-05,
      "loss": 0.7464,
      "step": 2457
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6474497106702266,
      "learning_rate": 1.269000057152975e-05,
      "loss": 0.7013,
      "step": 2458
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7779635627083211,
      "learning_rate": 1.2684529528509177e-05,
      "loss": 0.7412,
      "step": 2459
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7143040459083574,
      "learning_rate": 1.2679057619403432e-05,
      "loss": 0.7708,
      "step": 2460
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.670716356244487,
      "learning_rate": 1.2673584845977869e-05,
      "loss": 0.7141,
      "step": 2461
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6656855592435986,
      "learning_rate": 1.266811120999812e-05,
      "loss": 0.7027,
      "step": 2462
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7153128025534002,
      "learning_rate": 1.266263671323009e-05,
      "loss": 0.7264,
      "step": 2463
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6655824733270616,
      "learning_rate": 1.2657161357439968e-05,
      "loss": 0.6666,
      "step": 2464
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6892386964277015,
      "learning_rate": 1.2651685144394223e-05,
      "loss": 0.7601,
      "step": 2465
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7884999340297445,
      "learning_rate": 1.264620807585959e-05,
      "loss": 0.7382,
      "step": 2466
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6686061403235437,
      "learning_rate": 1.2640730153603085e-05,
      "loss": 0.6828,
      "step": 2467
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7267559789314942,
      "learning_rate": 1.2635251379392002e-05,
      "loss": 0.7704,
      "step": 2468
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6944426675103238,
      "learning_rate": 1.262977175499391e-05,
      "loss": 0.7149,
      "step": 2469
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6575656047329317,
      "learning_rate": 1.2624291282176642e-05,
      "loss": 0.6805,
      "step": 2470
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7329960806648298,
      "learning_rate": 1.2618809962708318e-05,
      "loss": 0.742,
      "step": 2471
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6864808929277261,
      "learning_rate": 1.2613327798357329e-05,
      "loss": 0.7083,
      "step": 2472
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6192826611099626,
      "learning_rate": 1.2607844790892327e-05,
      "loss": 0.7078,
      "step": 2473
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0670318830492338,
      "learning_rate": 1.2602360942082257e-05,
      "loss": 0.7239,
      "step": 2474
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.774072163861016,
      "learning_rate": 1.259687625369631e-05,
      "loss": 0.7466,
      "step": 2475
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.685327555032074,
      "learning_rate": 1.2591390727503968e-05,
      "loss": 0.7375,
      "step": 2476
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6822488486839149,
      "learning_rate": 1.2585904365274975e-05,
      "loss": 0.7162,
      "step": 2477
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6784356654988051,
      "learning_rate": 1.2580417168779342e-05,
      "loss": 0.6932,
      "step": 2478
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6158365500872892,
      "learning_rate": 1.257492913978736e-05,
      "loss": 0.7029,
      "step": 2479
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6915595913338037,
      "learning_rate": 1.2569440280069575e-05,
      "loss": 0.7338,
      "step": 2480
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6586618401317746,
      "learning_rate": 1.2563950591396812e-05,
      "loss": 0.6989,
      "step": 2481
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7111800024197478,
      "learning_rate": 1.2558460075540156e-05,
      "loss": 0.7424,
      "step": 2482
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6475301333781439,
      "learning_rate": 1.2552968734270965e-05,
      "loss": 0.6745,
      "step": 2483
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6692668182724142,
      "learning_rate": 1.2547476569360857e-05,
      "loss": 0.7285,
      "step": 2484
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7296983960430479,
      "learning_rate": 1.2541983582581715e-05,
      "loss": 0.7787,
      "step": 2485
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7264489839817135,
      "learning_rate": 1.2536489775705703e-05,
      "loss": 0.7287,
      "step": 2486
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6802378360820478,
      "learning_rate": 1.2530995150505226e-05,
      "loss": 0.7086,
      "step": 2487
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.622803983944456,
      "learning_rate": 1.2525499708752973e-05,
      "loss": 0.6972,
      "step": 2488
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8630893302316937,
      "learning_rate": 1.2520003452221883e-05,
      "loss": 0.742,
      "step": 2489
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.735497470392842,
      "learning_rate": 1.251450638268516e-05,
      "loss": 0.7208,
      "step": 2490
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7203888095629136,
      "learning_rate": 1.250900850191628e-05,
      "loss": 0.7342,
      "step": 2491
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6020587197769756,
      "learning_rate": 1.2503509811688974e-05,
      "loss": 0.724,
      "step": 2492
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6524918479875016,
      "learning_rate": 1.2498010313777226e-05,
      "loss": 0.693,
      "step": 2493
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5524960103447677,
      "learning_rate": 1.2492510009955297e-05,
      "loss": 0.6886,
      "step": 2494
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6622245952843252,
      "learning_rate": 1.2487008901997697e-05,
      "loss": 0.7361,
      "step": 2495
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7414523400119793,
      "learning_rate": 1.2481506991679195e-05,
      "loss": 0.73,
      "step": 2496
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8726833780272764,
      "learning_rate": 1.2476004280774827e-05,
      "loss": 0.7189,
      "step": 2497
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6829724368792278,
      "learning_rate": 1.2470500771059879e-05,
      "loss": 0.7225,
      "step": 2498
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7074358755209815,
      "learning_rate": 1.2464996464309898e-05,
      "loss": 0.688,
      "step": 2499
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8180899802094852,
      "learning_rate": 1.2459491362300688e-05,
      "loss": 0.7938,
      "step": 2500
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7046634826341415,
      "learning_rate": 1.2453985466808311e-05,
      "loss": 0.7202,
      "step": 2501
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7010341634084777,
      "learning_rate": 1.2448478779609083e-05,
      "loss": 0.6906,
      "step": 2502
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7776221721541776,
      "learning_rate": 1.2442971302479577e-05,
      "loss": 0.7598,
      "step": 2503
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6661280350835119,
      "learning_rate": 1.2437463037196616e-05,
      "loss": 0.6738,
      "step": 2504
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6802776639569043,
      "learning_rate": 1.2431953985537283e-05,
      "loss": 0.6907,
      "step": 2505
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8610463120026057,
      "learning_rate": 1.2426444149278917e-05,
      "loss": 0.7603,
      "step": 2506
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7208543597417296,
      "learning_rate": 1.2420933530199105e-05,
      "loss": 0.7053,
      "step": 2507
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7001479309862667,
      "learning_rate": 1.241542213007568e-05,
      "loss": 0.6695,
      "step": 2508
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.663029923358737,
      "learning_rate": 1.2409909950686745e-05,
      "loss": 0.6895,
      "step": 2509
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7794785163611814,
      "learning_rate": 1.2404396993810637e-05,
      "loss": 0.7572,
      "step": 2510
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7011457488843825,
      "learning_rate": 1.2398883261225951e-05,
      "loss": 0.7315,
      "step": 2511
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7754329988013688,
      "learning_rate": 1.2393368754711536e-05,
      "loss": 0.7472,
      "step": 2512
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.768832665022927,
      "learning_rate": 1.2387853476046482e-05,
      "loss": 0.7536,
      "step": 2513
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6457364246219567,
      "learning_rate": 1.2382337427010138e-05,
      "loss": 0.7221,
      "step": 2514
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7056032959553097,
      "learning_rate": 1.2376820609382094e-05,
      "loss": 0.6827,
      "step": 2515
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7463769884403302,
      "learning_rate": 1.2371303024942187e-05,
      "loss": 0.7542,
      "step": 2516
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7878769975086084,
      "learning_rate": 1.2365784675470509e-05,
      "loss": 0.7815,
      "step": 2517
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8768499996005572,
      "learning_rate": 1.2360265562747393e-05,
      "loss": 0.7787,
      "step": 2518
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7519926283466613,
      "learning_rate": 1.235474568855342e-05,
      "loss": 0.7399,
      "step": 2519
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6619560525283733,
      "learning_rate": 1.2349225054669418e-05,
      "loss": 0.6986,
      "step": 2520
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.718500298315222,
      "learning_rate": 1.2343703662876456e-05,
      "loss": 0.7392,
      "step": 2521
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7088046898065011,
      "learning_rate": 1.233818151495585e-05,
      "loss": 0.7362,
      "step": 2522
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.734182344516928,
      "learning_rate": 1.2332658612689161e-05,
      "loss": 0.7215,
      "step": 2523
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8219387282270771,
      "learning_rate": 1.2327134957858195e-05,
      "loss": 0.8145,
      "step": 2524
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7165244986984408,
      "learning_rate": 1.2321610552244993e-05,
      "loss": 0.7457,
      "step": 2525
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7218319491623219,
      "learning_rate": 1.2316085397631848e-05,
      "loss": 0.7303,
      "step": 2526
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6934877205913756,
      "learning_rate": 1.2310559495801286e-05,
      "loss": 0.7028,
      "step": 2527
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7479928418726803,
      "learning_rate": 1.2305032848536081e-05,
      "loss": 0.7396,
      "step": 2528
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.681079183008768,
      "learning_rate": 1.2299505457619243e-05,
      "loss": 0.723,
      "step": 2529
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7168967524005019,
      "learning_rate": 1.2293977324834025e-05,
      "loss": 0.7139,
      "step": 2530
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6743956997022388,
      "learning_rate": 1.2288448451963916e-05,
      "loss": 0.7197,
      "step": 2531
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6350306318516435,
      "learning_rate": 1.2282918840792645e-05,
      "loss": 0.7043,
      "step": 2532
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.646922193244222,
      "learning_rate": 1.2277388493104187e-05,
      "loss": 0.6886,
      "step": 2533
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6665080608332145,
      "learning_rate": 1.227185741068274e-05,
      "loss": 0.7015,
      "step": 2534
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.679666228161228,
      "learning_rate": 1.2266325595312744e-05,
      "loss": 0.6997,
      "step": 2535
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6095118414312688,
      "learning_rate": 1.2260793048778885e-05,
      "loss": 0.7194,
      "step": 2536
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.68700037576717,
      "learning_rate": 1.2255259772866075e-05,
      "loss": 0.6963,
      "step": 2537
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7555858452187916,
      "learning_rate": 1.2249725769359464e-05,
      "loss": 0.7733,
      "step": 2538
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.740603757799908,
      "learning_rate": 1.2244191040044437e-05,
      "loss": 0.6651,
      "step": 2539
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.646690173861469,
      "learning_rate": 1.223865558670661e-05,
      "loss": 0.7115,
      "step": 2540
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7526135824296872,
      "learning_rate": 1.2233119411131844e-05,
      "loss": 0.7367,
      "step": 2541
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6816787312856507,
      "learning_rate": 1.2227582515106216e-05,
      "loss": 0.7151,
      "step": 2542
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7725925108758829,
      "learning_rate": 1.2222044900416048e-05,
      "loss": 0.7551,
      "step": 2543
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6402996661487249,
      "learning_rate": 1.221650656884789e-05,
      "loss": 0.7081,
      "step": 2544
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7239523764799829,
      "learning_rate": 1.221096752218852e-05,
      "loss": 0.7184,
      "step": 2545
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7076494064446679,
      "learning_rate": 1.2205427762224951e-05,
      "loss": 0.6854,
      "step": 2546
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.683663937893182,
      "learning_rate": 1.219988729074443e-05,
      "loss": 0.7306,
      "step": 2547
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7039974706884854,
      "learning_rate": 1.2194346109534423e-05,
      "loss": 0.6937,
      "step": 2548
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6355524793693244,
      "learning_rate": 1.2188804220382633e-05,
      "loss": 0.7115,
      "step": 2549
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6282312369399005,
      "learning_rate": 1.2183261625076987e-05,
      "loss": 0.6944,
      "step": 2550
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.760342800134097,
      "learning_rate": 1.2177718325405643e-05,
      "loss": 0.7454,
      "step": 2551
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7146595443991985,
      "learning_rate": 1.2172174323156987e-05,
      "loss": 0.7488,
      "step": 2552
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7454778261369719,
      "learning_rate": 1.2166629620119628e-05,
      "loss": 0.715,
      "step": 2553
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7658573992469987,
      "learning_rate": 1.2161084218082397e-05,
      "loss": 0.7187,
      "step": 2554
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6496028719755729,
      "learning_rate": 1.2155538118834365e-05,
      "loss": 0.7267,
      "step": 2555
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6881325787929229,
      "learning_rate": 1.2149991324164817e-05,
      "loss": 0.7062,
      "step": 2556
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7949114009021576,
      "learning_rate": 1.2144443835863262e-05,
      "loss": 0.803,
      "step": 2557
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7576903604605215,
      "learning_rate": 1.2138895655719436e-05,
      "loss": 0.7178,
      "step": 2558
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6917941454317039,
      "learning_rate": 1.21333467855233e-05,
      "loss": 0.6927,
      "step": 2559
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7533633155090416,
      "learning_rate": 1.2127797227065033e-05,
      "loss": 0.7636,
      "step": 2560
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7377718468739967,
      "learning_rate": 1.212224698213504e-05,
      "loss": 0.7211,
      "step": 2561
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7315583117889375,
      "learning_rate": 1.2116696052523942e-05,
      "loss": 0.7288,
      "step": 2562
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.737956817925368,
      "learning_rate": 1.2111144440022586e-05,
      "loss": 0.7333,
      "step": 2563
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.767513998244305,
      "learning_rate": 1.2105592146422038e-05,
      "loss": 0.7425,
      "step": 2564
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6999872220391549,
      "learning_rate": 1.210003917351359e-05,
      "loss": 0.7128,
      "step": 2565
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6624028282168123,
      "learning_rate": 1.2094485523088737e-05,
      "loss": 0.6576,
      "step": 2566
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.743906630058857,
      "learning_rate": 1.2088931196939206e-05,
      "loss": 0.719,
      "step": 2567
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6884374646372952,
      "learning_rate": 1.208337619685694e-05,
      "loss": 0.7395,
      "step": 2568
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6696075366423262,
      "learning_rate": 1.2077820524634096e-05,
      "loss": 0.71,
      "step": 2569
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6107427026391251,
      "learning_rate": 1.2072264182063052e-05,
      "loss": 0.7166,
      "step": 2570
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7517931603895875,
      "learning_rate": 1.2066707170936399e-05,
      "loss": 0.7456,
      "step": 2571
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7457088259946913,
      "learning_rate": 1.206114949304694e-05,
      "loss": 0.7253,
      "step": 2572
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7028783749191799,
      "learning_rate": 1.2055591150187704e-05,
      "loss": 0.7505,
      "step": 2573
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7471134441203098,
      "learning_rate": 1.2050032144151923e-05,
      "loss": 0.7584,
      "step": 2574
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.651752204585861,
      "learning_rate": 1.2044472476733053e-05,
      "loss": 0.6906,
      "step": 2575
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7090877633496344,
      "learning_rate": 1.2038912149724758e-05,
      "loss": 0.7096,
      "step": 2576
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7427004438124023,
      "learning_rate": 1.2033351164920913e-05,
      "loss": 0.751,
      "step": 2577
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6848322227757028,
      "learning_rate": 1.2027789524115606e-05,
      "loss": 0.7129,
      "step": 2578
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6607336321808883,
      "learning_rate": 1.2022227229103143e-05,
      "loss": 0.6939,
      "step": 2579
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7035241041500059,
      "learning_rate": 1.2016664281678035e-05,
      "loss": 0.711,
      "step": 2580
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6688081097273204,
      "learning_rate": 1.2011100683635e-05,
      "loss": 0.7273,
      "step": 2581
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7041046573662716,
      "learning_rate": 1.2005536436768976e-05,
      "loss": 0.7322,
      "step": 2582
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6869850752571246,
      "learning_rate": 1.1999971542875104e-05,
      "loss": 0.678,
      "step": 2583
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7109285629516063,
      "learning_rate": 1.1994406003748735e-05,
      "loss": 0.6461,
      "step": 2584
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6589737109288186,
      "learning_rate": 1.1988839821185428e-05,
      "loss": 0.6779,
      "step": 2585
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7384845526534729,
      "learning_rate": 1.1983272996980943e-05,
      "loss": 0.7321,
      "step": 2586
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7572532422449944,
      "learning_rate": 1.1977705532931263e-05,
      "loss": 0.7412,
      "step": 2587
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8448848077135841,
      "learning_rate": 1.1972137430832565e-05,
      "loss": 0.7121,
      "step": 2588
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6861605419218091,
      "learning_rate": 1.1966568692481233e-05,
      "loss": 0.6748,
      "step": 2589
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6902295505413473,
      "learning_rate": 1.1960999319673856e-05,
      "loss": 0.7151,
      "step": 2590
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7281280262049067,
      "learning_rate": 1.1955429314207234e-05,
      "loss": 0.6748,
      "step": 2591
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6495165295299163,
      "learning_rate": 1.1949858677878366e-05,
      "loss": 0.7117,
      "step": 2592
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6456428714810494,
      "learning_rate": 1.1944287412484454e-05,
      "loss": 0.7089,
      "step": 2593
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6417431530289259,
      "learning_rate": 1.1938715519822908e-05,
      "loss": 0.6894,
      "step": 2594
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6844240989248419,
      "learning_rate": 1.193314300169133e-05,
      "loss": 0.7259,
      "step": 2595
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7079904295108993,
      "learning_rate": 1.1927569859887534e-05,
      "loss": 0.7116,
      "step": 2596
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6542047457686662,
      "learning_rate": 1.1921996096209535e-05,
      "loss": 0.6846,
      "step": 2597
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7408392500909895,
      "learning_rate": 1.1916421712455542e-05,
      "loss": 0.7288,
      "step": 2598
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6637681474814193,
      "learning_rate": 1.1910846710423962e-05,
      "loss": 0.7202,
      "step": 2599
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7130399900391309,
      "learning_rate": 1.1905271091913417e-05,
      "loss": 0.7475,
      "step": 2600
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6766186647989126,
      "learning_rate": 1.1899694858722713e-05,
      "loss": 0.7281,
      "step": 2601
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6482351630359908,
      "learning_rate": 1.189411801265086e-05,
      "loss": 0.6768,
      "step": 2602
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6818706520526164,
      "learning_rate": 1.1888540555497064e-05,
      "loss": 0.7134,
      "step": 2603
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7905151730624117,
      "learning_rate": 1.1882962489060726e-05,
      "loss": 0.7437,
      "step": 2604
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6514308996172898,
      "learning_rate": 1.1877383815141448e-05,
      "loss": 0.73,
      "step": 2605
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6996204533261572,
      "learning_rate": 1.187180453553903e-05,
      "loss": 0.7033,
      "step": 2606
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.687804306063394,
      "learning_rate": 1.1866224652053465e-05,
      "loss": 0.7226,
      "step": 2607
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.759762122584781,
      "learning_rate": 1.1860644166484932e-05,
      "loss": 0.7794,
      "step": 2608
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6338461310405153,
      "learning_rate": 1.1855063080633818e-05,
      "loss": 0.6796,
      "step": 2609
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7205391141913385,
      "learning_rate": 1.1849481396300692e-05,
      "loss": 0.6781,
      "step": 2610
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7066614022171972,
      "learning_rate": 1.184389911528633e-05,
      "loss": 0.7138,
      "step": 2611
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6943493732615446,
      "learning_rate": 1.183831623939169e-05,
      "loss": 0.6825,
      "step": 2612
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7054800409814129,
      "learning_rate": 1.1832732770417916e-05,
      "loss": 0.7746,
      "step": 2613
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7049447703131623,
      "learning_rate": 1.182714871016636e-05,
      "loss": 0.7078,
      "step": 2614
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7404410711891249,
      "learning_rate": 1.1821564060438552e-05,
      "loss": 0.7414,
      "step": 2615
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.672170475571877,
      "learning_rate": 1.181597882303622e-05,
      "loss": 0.7133,
      "step": 2616
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6290630216160689,
      "learning_rate": 1.1810392999761273e-05,
      "loss": 0.7238,
      "step": 2617
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.716706388694791,
      "learning_rate": 1.1804806592415821e-05,
      "loss": 0.7253,
      "step": 2618
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.688756085142853,
      "learning_rate": 1.1799219602802148e-05,
      "loss": 0.7323,
      "step": 2619
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7018997618645729,
      "learning_rate": 1.1793632032722737e-05,
      "loss": 0.6838,
      "step": 2620
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6697263985106993,
      "learning_rate": 1.1788043883980256e-05,
      "loss": 0.6969,
      "step": 2621
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7534757163512334,
      "learning_rate": 1.1782455158377553e-05,
      "loss": 0.7362,
      "step": 2622
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6381662856892655,
      "learning_rate": 1.1776865857717675e-05,
      "loss": 0.6956,
      "step": 2623
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7075548371931198,
      "learning_rate": 1.1771275983803839e-05,
      "loss": 0.754,
      "step": 2624
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7641281396539292,
      "learning_rate": 1.176568553843946e-05,
      "loss": 0.7182,
      "step": 2625
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8812841570370782,
      "learning_rate": 1.176009452342813e-05,
      "loss": 0.7658,
      "step": 2626
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7060404792043924,
      "learning_rate": 1.1754502940573626e-05,
      "loss": 0.7256,
      "step": 2627
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7014813537542534,
      "learning_rate": 1.1748910791679911e-05,
      "loss": 0.7299,
      "step": 2628
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7880113623857194,
      "learning_rate": 1.1743318078551131e-05,
      "loss": 0.696,
      "step": 2629
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6907915336117115,
      "learning_rate": 1.1737724802991608e-05,
      "loss": 0.7353,
      "step": 2630
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7114927603026453,
      "learning_rate": 1.1732130966805848e-05,
      "loss": 0.7505,
      "step": 2631
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6636498623395907,
      "learning_rate": 1.1726536571798546e-05,
      "loss": 0.6768,
      "step": 2632
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7750451966017317,
      "learning_rate": 1.1720941619774562e-05,
      "loss": 0.7632,
      "step": 2633
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6529016769809391,
      "learning_rate": 1.171534611253895e-05,
      "loss": 0.6935,
      "step": 2634
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7078333077560511,
      "learning_rate": 1.1709750051896937e-05,
      "loss": 0.7146,
      "step": 2635
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7063685580861095,
      "learning_rate": 1.1704153439653925e-05,
      "loss": 0.7216,
      "step": 2636
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8639926606774052,
      "learning_rate": 1.1698556277615499e-05,
      "loss": 0.7265,
      "step": 2637
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6734112392757491,
      "learning_rate": 1.1692958567587423e-05,
      "loss": 0.6966,
      "step": 2638
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7357397585372072,
      "learning_rate": 1.1687360311375634e-05,
      "loss": 0.7277,
      "step": 2639
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7052629698501591,
      "learning_rate": 1.168176151078624e-05,
      "loss": 0.6764,
      "step": 2640
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7021018370850047,
      "learning_rate": 1.167616216762554e-05,
      "loss": 0.7005,
      "step": 2641
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7158913985640424,
      "learning_rate": 1.1670562283699993e-05,
      "loss": 0.7363,
      "step": 2642
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7433680223567416,
      "learning_rate": 1.1664961860816241e-05,
      "loss": 0.7369,
      "step": 2643
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.732675730405724,
      "learning_rate": 1.1659360900781095e-05,
      "loss": 0.728,
      "step": 2644
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.618294256228388,
      "learning_rate": 1.1653759405401544e-05,
      "loss": 0.6976,
      "step": 2645
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7276173270339732,
      "learning_rate": 1.1648157376484741e-05,
      "loss": 0.7399,
      "step": 2646
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6258664491414127,
      "learning_rate": 1.1642554815838025e-05,
      "loss": 0.6991,
      "step": 2647
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6864898047496029,
      "learning_rate": 1.16369517252689e-05,
      "loss": 0.6995,
      "step": 2648
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6768916759775832,
      "learning_rate": 1.1631348106585027e-05,
      "loss": 0.698,
      "step": 2649
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6333198182313837,
      "learning_rate": 1.1625743961594264e-05,
      "loss": 0.7099,
      "step": 2650
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7926335588438118,
      "learning_rate": 1.1620139292104617e-05,
      "loss": 0.7889,
      "step": 2651
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6068998776070016,
      "learning_rate": 1.1614534099924274e-05,
      "loss": 0.6625,
      "step": 2652
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6701688363025057,
      "learning_rate": 1.1608928386861583e-05,
      "loss": 0.72,
      "step": 2653
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6478605029493304,
      "learning_rate": 1.1603322154725065e-05,
      "loss": 0.6658,
      "step": 2654
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8085131580104724,
      "learning_rate": 1.1597715405323409e-05,
      "loss": 0.7334,
      "step": 2655
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7635136470741118,
      "learning_rate": 1.1592108140465466e-05,
      "loss": 0.7282,
      "step": 2656
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6931935277176683,
      "learning_rate": 1.158650036196026e-05,
      "loss": 0.7421,
      "step": 2657
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7408374553368173,
      "learning_rate": 1.1580892071616974e-05,
      "loss": 0.734,
      "step": 2658
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6587015407413689,
      "learning_rate": 1.1575283271244964e-05,
      "loss": 0.7151,
      "step": 2659
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7514840071845719,
      "learning_rate": 1.1569673962653737e-05,
      "loss": 0.7247,
      "step": 2660
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6456436899363663,
      "learning_rate": 1.1564064147652984e-05,
      "loss": 0.701,
      "step": 2661
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.696971846249675,
      "learning_rate": 1.1558453828052542e-05,
      "loss": 0.7376,
      "step": 2662
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6585062240609977,
      "learning_rate": 1.1552843005662417e-05,
      "loss": 0.7031,
      "step": 2663
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6768638041333195,
      "learning_rate": 1.154723168229278e-05,
      "loss": 0.709,
      "step": 2664
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6820978094731669,
      "learning_rate": 1.1541619859753954e-05,
      "loss": 0.7133,
      "step": 2665
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6860815002061117,
      "learning_rate": 1.1536007539856439e-05,
      "loss": 0.6942,
      "step": 2666
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6897465206966269,
      "learning_rate": 1.1530394724410883e-05,
      "loss": 0.6839,
      "step": 2667
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6880398603519243,
      "learning_rate": 1.1524781415228095e-05,
      "loss": 0.7325,
      "step": 2668
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6730911071840638,
      "learning_rate": 1.1519167614119044e-05,
      "loss": 0.6841,
      "step": 2669
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7088018538989049,
      "learning_rate": 1.1513553322894866e-05,
      "loss": 0.7063,
      "step": 2670
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7455386497472378,
      "learning_rate": 1.1507938543366844e-05,
      "loss": 0.7334,
      "step": 2671
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.75339073880226,
      "learning_rate": 1.1502323277346422e-05,
      "loss": 0.6953,
      "step": 2672
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.702716414046401,
      "learning_rate": 1.1496707526645205e-05,
      "loss": 0.7175,
      "step": 2673
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7148043206238749,
      "learning_rate": 1.1491091293074948e-05,
      "loss": 0.7102,
      "step": 2674
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6246455615656044,
      "learning_rate": 1.1485474578447566e-05,
      "loss": 0.7115,
      "step": 2675
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7705514814151652,
      "learning_rate": 1.147985738457513e-05,
      "loss": 0.7476,
      "step": 2676
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6651928053680907,
      "learning_rate": 1.1474239713269861e-05,
      "loss": 0.6861,
      "step": 2677
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7349399223683145,
      "learning_rate": 1.1468621566344138e-05,
      "loss": 0.7209,
      "step": 2678
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7094142796858918,
      "learning_rate": 1.1463002945610495e-05,
      "loss": 0.7474,
      "step": 2679
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7232703088050717,
      "learning_rate": 1.1457383852881613e-05,
      "loss": 0.7182,
      "step": 2680
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7564387145513923,
      "learning_rate": 1.145176428997033e-05,
      "loss": 0.7339,
      "step": 2681
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.696596100402127,
      "learning_rate": 1.1446144258689631e-05,
      "loss": 0.6942,
      "step": 2682
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6333098779669122,
      "learning_rate": 1.1440523760852657e-05,
      "loss": 0.7054,
      "step": 2683
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7519174767448613,
      "learning_rate": 1.14349027982727e-05,
      "loss": 0.743,
      "step": 2684
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7502795816664956,
      "learning_rate": 1.1429281372763198e-05,
      "loss": 0.7361,
      "step": 2685
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6748494808417854,
      "learning_rate": 1.142365948613774e-05,
      "loss": 0.7752,
      "step": 2686
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6619515072169204,
      "learning_rate": 1.1418037140210061e-05,
      "loss": 0.6857,
      "step": 2687
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6805484326952689,
      "learning_rate": 1.1412414336794052e-05,
      "loss": 0.7402,
      "step": 2688
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7563933808355008,
      "learning_rate": 1.1406791077703744e-05,
      "loss": 0.7722,
      "step": 2689
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7542795511356861,
      "learning_rate": 1.1401167364753316e-05,
      "loss": 0.7256,
      "step": 2690
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5878345650580514,
      "learning_rate": 1.1395543199757097e-05,
      "loss": 0.6764,
      "step": 2691
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7240953282258596,
      "learning_rate": 1.138991858452956e-05,
      "loss": 0.7368,
      "step": 2692
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.618755746825381,
      "learning_rate": 1.1384293520885323e-05,
      "loss": 0.7081,
      "step": 2693
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6869105352269484,
      "learning_rate": 1.1378668010639151e-05,
      "loss": 0.716,
      "step": 2694
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7063099663811014,
      "learning_rate": 1.1373042055605944e-05,
      "loss": 0.7223,
      "step": 2695
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7148963982735701,
      "learning_rate": 1.1367415657600757e-05,
      "loss": 0.7048,
      "step": 2696
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7344694746961528,
      "learning_rate": 1.1361788818438784e-05,
      "loss": 0.7467,
      "step": 2697
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6771536607654247,
      "learning_rate": 1.135616153993536e-05,
      "loss": 0.7595,
      "step": 2698
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6919642374869757,
      "learning_rate": 1.1350533823905959e-05,
      "loss": 0.7442,
      "step": 2699
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.752419772381414,
      "learning_rate": 1.1344905672166205e-05,
      "loss": 0.7531,
      "step": 2700
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6971849920813366,
      "learning_rate": 1.1339277086531853e-05,
      "loss": 0.7488,
      "step": 2701
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6222171484332547,
      "learning_rate": 1.1333648068818804e-05,
      "loss": 0.6792,
      "step": 2702
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7163940721903748,
      "learning_rate": 1.1328018620843095e-05,
      "loss": 0.739,
      "step": 2703
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6349934037502499,
      "learning_rate": 1.1322388744420902e-05,
      "loss": 0.6641,
      "step": 2704
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7162145614823003,
      "learning_rate": 1.1316758441368545e-05,
      "loss": 0.7434,
      "step": 2705
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7580632231814334,
      "learning_rate": 1.1311127713502472e-05,
      "loss": 0.7555,
      "step": 2706
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7059545826524883,
      "learning_rate": 1.1305496562639278e-05,
      "loss": 0.6864,
      "step": 2707
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7213220959021831,
      "learning_rate": 1.1299864990595689e-05,
      "loss": 0.7484,
      "step": 2708
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7647186531600428,
      "learning_rate": 1.129423299918856e-05,
      "loss": 0.7627,
      "step": 2709
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6434706841859783,
      "learning_rate": 1.1288600590234898e-05,
      "loss": 0.6903,
      "step": 2710
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7494809987721406,
      "learning_rate": 1.1282967765551832e-05,
      "loss": 0.7212,
      "step": 2711
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6823895045314353,
      "learning_rate": 1.1277334526956628e-05,
      "loss": 0.6951,
      "step": 2712
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7527909641536127,
      "learning_rate": 1.1271700876266684e-05,
      "loss": 0.753,
      "step": 2713
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6485361286589707,
      "learning_rate": 1.1266066815299541e-05,
      "loss": 0.6738,
      "step": 2714
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6368774011245344,
      "learning_rate": 1.1260432345872854e-05,
      "loss": 0.7175,
      "step": 2715
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7081370958327272,
      "learning_rate": 1.125479746980443e-05,
      "loss": 0.7307,
      "step": 2716
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.706203401327245,
      "learning_rate": 1.1249162188912193e-05,
      "loss": 0.7465,
      "step": 2717
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6975261453283592,
      "learning_rate": 1.1243526505014199e-05,
      "loss": 0.6952,
      "step": 2718
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6826044363331263,
      "learning_rate": 1.1237890419928641e-05,
      "loss": 0.7068,
      "step": 2719
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6634716316928342,
      "learning_rate": 1.1232253935473838e-05,
      "loss": 0.7156,
      "step": 2720
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7242953430226801,
      "learning_rate": 1.1226617053468237e-05,
      "loss": 0.7644,
      "step": 2721
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6854092390647029,
      "learning_rate": 1.1220979775730409e-05,
      "loss": 0.6941,
      "step": 2722
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9135791294299105,
      "learning_rate": 1.1215342104079064e-05,
      "loss": 0.7824,
      "step": 2723
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7100277991685624,
      "learning_rate": 1.1209704040333027e-05,
      "loss": 0.7126,
      "step": 2724
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6995258722737833,
      "learning_rate": 1.1204065586311259e-05,
      "loss": 0.7208,
      "step": 2725
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7484784164348334,
      "learning_rate": 1.1198426743832842e-05,
      "loss": 0.7437,
      "step": 2726
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6515412274343712,
      "learning_rate": 1.1192787514716979e-05,
      "loss": 0.7125,
      "step": 2727
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7200731237638842,
      "learning_rate": 1.1187147900783009e-05,
      "loss": 0.7031,
      "step": 2728
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6836308507893978,
      "learning_rate": 1.1181507903850388e-05,
      "loss": 0.7443,
      "step": 2729
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6334066852001051,
      "learning_rate": 1.1175867525738696e-05,
      "loss": 0.6976,
      "step": 2730
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6709627310511967,
      "learning_rate": 1.1170226768267631e-05,
      "loss": 0.6903,
      "step": 2731
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6800963315094194,
      "learning_rate": 1.1164585633257028e-05,
      "loss": 0.7086,
      "step": 2732
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6305784791789842,
      "learning_rate": 1.1158944122526827e-05,
      "loss": 0.7065,
      "step": 2733
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6559470340354706,
      "learning_rate": 1.1153302237897105e-05,
      "loss": 0.7132,
      "step": 2734
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6499445821402583,
      "learning_rate": 1.1147659981188046e-05,
      "loss": 0.7145,
      "step": 2735
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7834381561894229,
      "learning_rate": 1.1142017354219959e-05,
      "loss": 0.7742,
      "step": 2736
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7311127571256035,
      "learning_rate": 1.1136374358813273e-05,
      "loss": 0.6805,
      "step": 2737
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5995382098898412,
      "learning_rate": 1.1130730996788539e-05,
      "loss": 0.6559,
      "step": 2738
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6744613621479518,
      "learning_rate": 1.1125087269966423e-05,
      "loss": 0.673,
      "step": 2739
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7357053492629709,
      "learning_rate": 1.1119443180167706e-05,
      "loss": 0.7157,
      "step": 2740
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7231688222516701,
      "learning_rate": 1.1113798729213292e-05,
      "loss": 0.7048,
      "step": 2741
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7782090160458149,
      "learning_rate": 1.1108153918924196e-05,
      "loss": 0.7101,
      "step": 2742
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7926361843350189,
      "learning_rate": 1.1102508751121553e-05,
      "loss": 0.76,
      "step": 2743
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8569783788680596,
      "learning_rate": 1.1096863227626612e-05,
      "loss": 0.795,
      "step": 2744
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6903421067384572,
      "learning_rate": 1.1091217350260734e-05,
      "loss": 0.6914,
      "step": 2745
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7256812632073264,
      "learning_rate": 1.10855711208454e-05,
      "loss": 0.7393,
      "step": 2746
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7002652711478142,
      "learning_rate": 1.1079924541202199e-05,
      "loss": 0.698,
      "step": 2747
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7861107780449234,
      "learning_rate": 1.1074277613152838e-05,
      "loss": 0.7593,
      "step": 2748
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7245040971287392,
      "learning_rate": 1.1068630338519131e-05,
      "loss": 0.7198,
      "step": 2749
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9665351099673138,
      "learning_rate": 1.1062982719123003e-05,
      "loss": 0.7463,
      "step": 2750
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7067837861862152,
      "learning_rate": 1.1057334756786503e-05,
      "loss": 0.7254,
      "step": 2751
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6517548955900009,
      "learning_rate": 1.1051686453331774e-05,
      "loss": 0.7113,
      "step": 2752
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7488072302844772,
      "learning_rate": 1.1046037810581081e-05,
      "loss": 0.752,
      "step": 2753
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8328616242451561,
      "learning_rate": 1.1040388830356787e-05,
      "loss": 0.733,
      "step": 2754
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7652846595278713,
      "learning_rate": 1.1034739514481378e-05,
      "loss": 0.7475,
      "step": 2755
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7009817011167535,
      "learning_rate": 1.1029089864777438e-05,
      "loss": 0.6949,
      "step": 2756
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7573448884557715,
      "learning_rate": 1.1023439883067664e-05,
      "loss": 0.7211,
      "step": 2757
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6459205287078026,
      "learning_rate": 1.1017789571174855e-05,
      "loss": 0.7021,
      "step": 2758
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6984512932865691,
      "learning_rate": 1.1012138930921918e-05,
      "loss": 0.6965,
      "step": 2759
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6368129837068165,
      "learning_rate": 1.100648796413187e-05,
      "loss": 0.688,
      "step": 2760
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6682001877673517,
      "learning_rate": 1.1000836672627833e-05,
      "loss": 0.6791,
      "step": 2761
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6385094970040416,
      "learning_rate": 1.0995185058233027e-05,
      "loss": 0.6929,
      "step": 2762
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6621261960761494,
      "learning_rate": 1.098953312277078e-05,
      "loss": 0.6553,
      "step": 2763
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6895998667925124,
      "learning_rate": 1.0983880868064532e-05,
      "loss": 0.6915,
      "step": 2764
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.603268956456155,
      "learning_rate": 1.0978228295937807e-05,
      "loss": 0.6723,
      "step": 2765
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6537797696356261,
      "learning_rate": 1.0972575408214248e-05,
      "loss": 0.7419,
      "step": 2766
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7528935339183178,
      "learning_rate": 1.0966922206717602e-05,
      "loss": 0.7096,
      "step": 2767
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7069432454546059,
      "learning_rate": 1.0961268693271694e-05,
      "loss": 0.7206,
      "step": 2768
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7085505904308331,
      "learning_rate": 1.0955614869700473e-05,
      "loss": 0.7463,
      "step": 2769
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6634201609072256,
      "learning_rate": 1.0949960737827985e-05,
      "loss": 0.7064,
      "step": 2770
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6922443187203114,
      "learning_rate": 1.0944306299478367e-05,
      "loss": 0.7778,
      "step": 2771
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6742486405072988,
      "learning_rate": 1.0938651556475852e-05,
      "loss": 0.6887,
      "step": 2772
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.792476426451177,
      "learning_rate": 1.093299651064479e-05,
      "loss": 0.7135,
      "step": 2773
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6868144478350685,
      "learning_rate": 1.0927341163809608e-05,
      "loss": 0.6906,
      "step": 2774
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7102466876419056,
      "learning_rate": 1.0921685517794842e-05,
      "loss": 0.7302,
      "step": 2775
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7106565489563269,
      "learning_rate": 1.0916029574425122e-05,
      "loss": 0.7458,
      "step": 2776
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6832813963283006,
      "learning_rate": 1.0910373335525169e-05,
      "loss": 0.6963,
      "step": 2777
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6573138664136541,
      "learning_rate": 1.0904716802919808e-05,
      "loss": 0.6976,
      "step": 2778
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6244010138581234,
      "learning_rate": 1.0899059978433951e-05,
      "loss": 0.6784,
      "step": 2779
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7066182212305232,
      "learning_rate": 1.0893402863892613e-05,
      "loss": 0.6906,
      "step": 2780
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7018365956357668,
      "learning_rate": 1.0887745461120888e-05,
      "loss": 0.7228,
      "step": 2781
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6605688333961236,
      "learning_rate": 1.0882087771943981e-05,
      "loss": 0.6922,
      "step": 2782
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7096716750067581,
      "learning_rate": 1.0876429798187174e-05,
      "loss": 0.7704,
      "step": 2783
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7204135783918414,
      "learning_rate": 1.087077154167585e-05,
      "loss": 0.7292,
      "step": 2784
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5813328926073041,
      "learning_rate": 1.0865113004235484e-05,
      "loss": 0.6706,
      "step": 2785
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6407873153016231,
      "learning_rate": 1.0859454187691631e-05,
      "loss": 0.7036,
      "step": 2786
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7507221652174928,
      "learning_rate": 1.0853795093869947e-05,
      "loss": 0.7139,
      "step": 2787
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6616117240856642,
      "learning_rate": 1.0848135724596174e-05,
      "loss": 0.6792,
      "step": 2788
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8030532447895435,
      "learning_rate": 1.0842476081696142e-05,
      "loss": 0.7651,
      "step": 2789
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7631282882485894,
      "learning_rate": 1.0836816166995769e-05,
      "loss": 0.7283,
      "step": 2790
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7527381613194298,
      "learning_rate": 1.083115598232106e-05,
      "loss": 0.7712,
      "step": 2791
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6404141167399571,
      "learning_rate": 1.0825495529498111e-05,
      "loss": 0.7113,
      "step": 2792
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7404435349239418,
      "learning_rate": 1.0819834810353104e-05,
      "loss": 0.7533,
      "step": 2793
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6881252291028457,
      "learning_rate": 1.0814173826712302e-05,
      "loss": 0.7428,
      "step": 2794
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7082043600837183,
      "learning_rate": 1.0808512580402054e-05,
      "loss": 0.717,
      "step": 2795
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6968416399751947,
      "learning_rate": 1.0802851073248801e-05,
      "loss": 0.7128,
      "step": 2796
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6646071340098443,
      "learning_rate": 1.0797189307079056e-05,
      "loss": 0.7334,
      "step": 2797
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6367644314412191,
      "learning_rate": 1.0791527283719432e-05,
      "loss": 0.7043,
      "step": 2798
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.700316144454102,
      "learning_rate": 1.0785865004996614e-05,
      "loss": 0.688,
      "step": 2799
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7510593707175758,
      "learning_rate": 1.0780202472737363e-05,
      "loss": 0.7621,
      "step": 2800
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8501912818737972,
      "learning_rate": 1.0774539688768537e-05,
      "loss": 0.7736,
      "step": 2801
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7262810749907788,
      "learning_rate": 1.0768876654917068e-05,
      "loss": 0.7525,
      "step": 2802
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7373290026043067,
      "learning_rate": 1.0763213373009967e-05,
      "loss": 0.6917,
      "step": 2803
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6567365390498224,
      "learning_rate": 1.0757549844874325e-05,
      "loss": 0.7256,
      "step": 2804
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7861414882618926,
      "learning_rate": 1.0751886072337318e-05,
      "loss": 0.7541,
      "step": 2805
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7903643360213701,
      "learning_rate": 1.0746222057226194e-05,
      "loss": 0.7221,
      "step": 2806
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6719017747395004,
      "learning_rate": 1.0740557801368286e-05,
      "loss": 0.7186,
      "step": 2807
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.672216821317448,
      "learning_rate": 1.0734893306590997e-05,
      "loss": 0.6889,
      "step": 2808
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7394990929007595,
      "learning_rate": 1.0729228574721811e-05,
      "loss": 0.7326,
      "step": 2809
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6995096349308622,
      "learning_rate": 1.072356360758829e-05,
      "loss": 0.7037,
      "step": 2810
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7728299609703555,
      "learning_rate": 1.0717898407018073e-05,
      "loss": 0.7563,
      "step": 2811
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7333577919287224,
      "learning_rate": 1.0712232974838867e-05,
      "loss": 0.6995,
      "step": 2812
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6441607114657054,
      "learning_rate": 1.0706567312878461e-05,
      "loss": 0.7008,
      "step": 2813
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6534031224491351,
      "learning_rate": 1.070090142296472e-05,
      "loss": 0.6724,
      "step": 2814
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6141624958067369,
      "learning_rate": 1.069523530692557e-05,
      "loss": 0.6923,
      "step": 2815
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7145784662733589,
      "learning_rate": 1.0689568966589021e-05,
      "loss": 0.7118,
      "step": 2816
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5808789622354602,
      "learning_rate": 1.0683902403783155e-05,
      "loss": 0.6558,
      "step": 2817
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7811950388151783,
      "learning_rate": 1.0678235620336118e-05,
      "loss": 0.7468,
      "step": 2818
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6855247404740413,
      "learning_rate": 1.067256861807614e-05,
      "loss": 0.672,
      "step": 2819
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6668546652477397,
      "learning_rate": 1.0666901398831508e-05,
      "loss": 0.7294,
      "step": 2820
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6869906642558223,
      "learning_rate": 1.0661233964430588e-05,
      "loss": 0.693,
      "step": 2821
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7054356429731287,
      "learning_rate": 1.065556631670181e-05,
      "loss": 0.7453,
      "step": 2822
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7063264233382386,
      "learning_rate": 1.0649898457473678e-05,
      "loss": 0.7291,
      "step": 2823
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6702885570844186,
      "learning_rate": 1.064423038857476e-05,
      "loss": 0.6612,
      "step": 2824
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6305824421012113,
      "learning_rate": 1.0638562111833692e-05,
      "loss": 0.6913,
      "step": 2825
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7013332039023127,
      "learning_rate": 1.0632893629079181e-05,
      "loss": 0.7435,
      "step": 2826
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8278239952908487,
      "learning_rate": 1.0627224942139993e-05,
      "loss": 0.7339,
      "step": 2827
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7145008070457886,
      "learning_rate": 1.0621556052844972e-05,
      "loss": 0.7033,
      "step": 2828
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7893194399052197,
      "learning_rate": 1.0615886963023013e-05,
      "loss": 0.7389,
      "step": 2829
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7053017679872559,
      "learning_rate": 1.0610217674503083e-05,
      "loss": 0.7144,
      "step": 2830
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6698175845697973,
      "learning_rate": 1.0604548189114222e-05,
      "loss": 0.6722,
      "step": 2831
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7419643381440596,
      "learning_rate": 1.059887850868551e-05,
      "loss": 0.7161,
      "step": 2832
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6836653391633116,
      "learning_rate": 1.0593208635046112e-05,
      "loss": 0.698,
      "step": 2833
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6738355971997058,
      "learning_rate": 1.0587538570025249e-05,
      "loss": 0.6719,
      "step": 2834
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.844268452168889,
      "learning_rate": 1.0581868315452197e-05,
      "loss": 0.7577,
      "step": 2835
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.732461338357128,
      "learning_rate": 1.05761978731563e-05,
      "loss": 0.7526,
      "step": 2836
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6775544133124999,
      "learning_rate": 1.0570527244966963e-05,
      "loss": 0.7177,
      "step": 2837
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7205145020423612,
      "learning_rate": 1.0564856432713646e-05,
      "loss": 0.6554,
      "step": 2838
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6725616801561719,
      "learning_rate": 1.0559185438225874e-05,
      "loss": 0.7085,
      "step": 2839
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.999371129673703,
      "learning_rate": 1.0553514263333229e-05,
      "loss": 0.7193,
      "step": 2840
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.790983199158061,
      "learning_rate": 1.0547842909865342e-05,
      "loss": 0.7216,
      "step": 2841
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6693868120926424,
      "learning_rate": 1.0542171379651918e-05,
      "loss": 0.6843,
      "step": 2842
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7092564451054137,
      "learning_rate": 1.0536499674522711e-05,
      "loss": 0.732,
      "step": 2843
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9640677462003469,
      "learning_rate": 1.0530827796307529e-05,
      "loss": 0.7001,
      "step": 2844
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7416962484065477,
      "learning_rate": 1.0525155746836233e-05,
      "loss": 0.7894,
      "step": 2845
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6987866598640389,
      "learning_rate": 1.0519483527938751e-05,
      "loss": 0.7243,
      "step": 2846
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7524603023588484,
      "learning_rate": 1.0513811141445059e-05,
      "loss": 0.7301,
      "step": 2847
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7180544456769066,
      "learning_rate": 1.0508138589185184e-05,
      "loss": 0.7176,
      "step": 2848
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.612373614223005,
      "learning_rate": 1.0502465872989212e-05,
      "loss": 0.6656,
      "step": 2849
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7096786822590555,
      "learning_rate": 1.0496792994687276e-05,
      "loss": 0.7122,
      "step": 2850
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6900316203412138,
      "learning_rate": 1.0491119956109565e-05,
      "loss": 0.6955,
      "step": 2851
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6946954460604637,
      "learning_rate": 1.0485446759086327e-05,
      "loss": 0.7152,
      "step": 2852
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7584105948735997,
      "learning_rate": 1.0479773405447847e-05,
      "loss": 0.7146,
      "step": 2853
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6424878614199026,
      "learning_rate": 1.0474099897024465e-05,
      "loss": 0.7091,
      "step": 2854
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.703607196044149,
      "learning_rate": 1.0468426235646581e-05,
      "loss": 0.6971,
      "step": 2855
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7741010368702056,
      "learning_rate": 1.0462752423144626e-05,
      "loss": 0.7577,
      "step": 2856
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0129679062381745,
      "learning_rate": 1.0457078461349099e-05,
      "loss": 0.67,
      "step": 2857
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.679408281471285,
      "learning_rate": 1.0451404352090538e-05,
      "loss": 0.7269,
      "step": 2858
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8863361834265046,
      "learning_rate": 1.0445730097199522e-05,
      "loss": 0.7523,
      "step": 2859
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6240434662769712,
      "learning_rate": 1.0440055698506688e-05,
      "loss": 0.6915,
      "step": 2860
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.72326316396151,
      "learning_rate": 1.0434381157842717e-05,
      "loss": 0.6983,
      "step": 2861
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7230700072483903,
      "learning_rate": 1.0428706477038328e-05,
      "loss": 0.757,
      "step": 2862
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6264199914087596,
      "learning_rate": 1.04230316579243e-05,
      "loss": 0.6776,
      "step": 2863
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6532831857092505,
      "learning_rate": 1.0417356702331445e-05,
      "loss": 0.6825,
      "step": 2864
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7584829761868984,
      "learning_rate": 1.0411681612090616e-05,
      "loss": 0.7217,
      "step": 2865
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6973964214390191,
      "learning_rate": 1.0406006389032724e-05,
      "loss": 0.6773,
      "step": 2866
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7460731014569532,
      "learning_rate": 1.0400331034988706e-05,
      "loss": 0.7214,
      "step": 2867
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7805049643446456,
      "learning_rate": 1.0394655551789555e-05,
      "loss": 0.7465,
      "step": 2868
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6762835611195458,
      "learning_rate": 1.03889799412663e-05,
      "loss": 0.7402,
      "step": 2869
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6617632900725374,
      "learning_rate": 1.0383304205250007e-05,
      "loss": 0.7083,
      "step": 2870
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.631553373576675,
      "learning_rate": 1.0377628345571789e-05,
      "loss": 0.6919,
      "step": 2871
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6185586045461744,
      "learning_rate": 1.0371952364062803e-05,
      "loss": 0.696,
      "step": 2872
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7637138764805216,
      "learning_rate": 1.0366276262554227e-05,
      "loss": 0.7148,
      "step": 2873
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.682690469240136,
      "learning_rate": 1.0360600042877298e-05,
      "loss": 0.6987,
      "step": 2874
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7970988861007818,
      "learning_rate": 1.035492370686328e-05,
      "loss": 0.7219,
      "step": 2875
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8283207576996682,
      "learning_rate": 1.034924725634348e-05,
      "loss": 0.7399,
      "step": 2876
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6665626501692733,
      "learning_rate": 1.0343570693149234e-05,
      "loss": 0.6608,
      "step": 2877
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6613340266363869,
      "learning_rate": 1.0337894019111925e-05,
      "loss": 0.707,
      "step": 2878
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7266575585928552,
      "learning_rate": 1.0332217236062962e-05,
      "loss": 0.7335,
      "step": 2879
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7143181560186473,
      "learning_rate": 1.0326540345833796e-05,
      "loss": 0.7756,
      "step": 2880
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.724285193983289,
      "learning_rate": 1.0320863350255911e-05,
      "loss": 0.7236,
      "step": 2881
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6854156559488439,
      "learning_rate": 1.031518625116082e-05,
      "loss": 0.6913,
      "step": 2882
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7656531676946688,
      "learning_rate": 1.0309509050380078e-05,
      "loss": 0.7687,
      "step": 2883
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6651865018714193,
      "learning_rate": 1.0303831749745266e-05,
      "loss": 0.7274,
      "step": 2884
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.772158075290913,
      "learning_rate": 1.0298154351088e-05,
      "loss": 0.7624,
      "step": 2885
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7503930461612056,
      "learning_rate": 1.0292476856239926e-05,
      "loss": 0.7755,
      "step": 2886
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6503181857034405,
      "learning_rate": 1.0286799267032725e-05,
      "loss": 0.735,
      "step": 2887
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7086870208484286,
      "learning_rate": 1.0281121585298101e-05,
      "loss": 0.7548,
      "step": 2888
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.697188915173979,
      "learning_rate": 1.0275443812867798e-05,
      "loss": 0.7204,
      "step": 2889
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8171872080507754,
      "learning_rate": 1.026976595157358e-05,
      "loss": 0.7773,
      "step": 2890
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6733980374510048,
      "learning_rate": 1.0264088003247244e-05,
      "loss": 0.7023,
      "step": 2891
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6913558643610924,
      "learning_rate": 1.0258409969720612e-05,
      "loss": 0.7081,
      "step": 2892
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7094867182830448,
      "learning_rate": 1.0252731852825544e-05,
      "loss": 0.6945,
      "step": 2893
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8017292268235946,
      "learning_rate": 1.0247053654393908e-05,
      "loss": 0.7047,
      "step": 2894
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6599728505830732,
      "learning_rate": 1.0241375376257619e-05,
      "loss": 0.6883,
      "step": 2895
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6910567327367606,
      "learning_rate": 1.0235697020248603e-05,
      "loss": 0.7003,
      "step": 2896
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6853637519197242,
      "learning_rate": 1.0230018588198816e-05,
      "loss": 0.6731,
      "step": 2897
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6812911479677479,
      "learning_rate": 1.0224340081940238e-05,
      "loss": 0.6686,
      "step": 2898
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7560340506847467,
      "learning_rate": 1.0218661503304876e-05,
      "loss": 0.7374,
      "step": 2899
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1655884379596895,
      "learning_rate": 1.0212982854124754e-05,
      "loss": 0.7142,
      "step": 2900
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7118330259558254,
      "learning_rate": 1.0207304136231928e-05,
      "loss": 0.732,
      "step": 2901
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7225094186556846,
      "learning_rate": 1.0201625351458463e-05,
      "loss": 0.7095,
      "step": 2902
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6342480763357196,
      "learning_rate": 1.019594650163646e-05,
      "loss": 0.6901,
      "step": 2903
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8023184529444487,
      "learning_rate": 1.0190267588598033e-05,
      "loss": 0.7818,
      "step": 2904
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6744351529371273,
      "learning_rate": 1.018458861417532e-05,
      "loss": 0.7033,
      "step": 2905
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7723792593997167,
      "learning_rate": 1.0178909580200467e-05,
      "loss": 0.7044,
      "step": 2906
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6156399170020428,
      "learning_rate": 1.0173230488505659e-05,
      "loss": 0.6699,
      "step": 2907
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.65743438542043,
      "learning_rate": 1.0167551340923085e-05,
      "loss": 0.7355,
      "step": 2908
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6752900657284924,
      "learning_rate": 1.0161872139284956e-05,
      "loss": 0.7458,
      "step": 2909
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7236034034998909,
      "learning_rate": 1.0156192885423502e-05,
      "loss": 0.6959,
      "step": 2910
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0680054973020572,
      "learning_rate": 1.0150513581170966e-05,
      "loss": 0.7032,
      "step": 2911
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6598017423207226,
      "learning_rate": 1.0144834228359613e-05,
      "loss": 0.7114,
      "step": 2912
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6109098542432276,
      "learning_rate": 1.0139154828821724e-05,
      "loss": 0.6997,
      "step": 2913
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6774155698474208,
      "learning_rate": 1.0133475384389584e-05,
      "loss": 0.6825,
      "step": 2914
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6739448718415745,
      "learning_rate": 1.01277958968955e-05,
      "loss": 0.7204,
      "step": 2915
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9150734724432642,
      "learning_rate": 1.0122116368171801e-05,
      "loss": 0.8561,
      "step": 2916
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7562458322963391,
      "learning_rate": 1.0116436800050817e-05,
      "loss": 0.715,
      "step": 2917
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7933913764609585,
      "learning_rate": 1.0110757194364894e-05,
      "loss": 0.7233,
      "step": 2918
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6647073391429036,
      "learning_rate": 1.0105077552946392e-05,
      "loss": 0.7212,
      "step": 2919
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6955659469820858,
      "learning_rate": 1.009939787762768e-05,
      "loss": 0.6931,
      "step": 2920
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.698786936393277,
      "learning_rate": 1.0093718170241141e-05,
      "loss": 0.718,
      "step": 2921
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7722088505446867,
      "learning_rate": 1.0088038432619172e-05,
      "loss": 0.7232,
      "step": 2922
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7631917959056111,
      "learning_rate": 1.0082358666594166e-05,
      "loss": 0.7396,
      "step": 2923
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3764620820493751,
      "learning_rate": 1.0076678873998535e-05,
      "loss": 0.6944,
      "step": 2924
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6158398549497865,
      "learning_rate": 1.0070999056664708e-05,
      "loss": 0.6909,
      "step": 2925
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8075736471867986,
      "learning_rate": 1.00653192164251e-05,
      "loss": 0.7697,
      "step": 2926
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5997465787802853,
      "learning_rate": 1.0059639355112156e-05,
      "loss": 0.665,
      "step": 2927
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7640314526243934,
      "learning_rate": 1.0053959474558313e-05,
      "loss": 0.7511,
      "step": 2928
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7138658863607185,
      "learning_rate": 1.0048279576596016e-05,
      "loss": 0.7491,
      "step": 2929
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6528793310517919,
      "learning_rate": 1.0042599663057723e-05,
      "loss": 0.6981,
      "step": 2930
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7080994809132561,
      "learning_rate": 1.0036919735775892e-05,
      "loss": 0.7557,
      "step": 2931
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7094869592823664,
      "learning_rate": 1.0031239796582984e-05,
      "loss": 0.7051,
      "step": 2932
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6795176908263497,
      "learning_rate": 1.0025559847311464e-05,
      "loss": 0.7203,
      "step": 2933
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7142129071239831,
      "learning_rate": 1.001987988979381e-05,
      "loss": 0.7766,
      "step": 2934
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7412390373773144,
      "learning_rate": 1.0014199925862482e-05,
      "loss": 0.702,
      "step": 2935
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6592792312078375,
      "learning_rate": 1.0008519957349967e-05,
      "loss": 0.6899,
      "step": 2936
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6536505019891399,
      "learning_rate": 1.0002839986088737e-05,
      "loss": 0.6934,
      "step": 2937
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6492995912228667,
      "learning_rate": 9.997160013911268e-06,
      "loss": 0.7011,
      "step": 2938
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6989220183587851,
      "learning_rate": 9.991480042650034e-06,
      "loss": 0.696,
      "step": 2939
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7053840419896936,
      "learning_rate": 9.985800074137521e-06,
      "loss": 0.714,
      "step": 2940
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6138875064803275,
      "learning_rate": 9.980120110206196e-06,
      "loss": 0.6637,
      "step": 2941
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7712427702137129,
      "learning_rate": 9.97444015268854e-06,
      "loss": 0.7443,
      "step": 2942
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6490332986355238,
      "learning_rate": 9.968760203417021e-06,
      "loss": 0.6974,
      "step": 2943
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7154324888076159,
      "learning_rate": 9.963080264224111e-06,
      "loss": 0.7101,
      "step": 2944
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6195796607566696,
      "learning_rate": 9.957400336942279e-06,
      "loss": 0.7151,
      "step": 2945
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6539005457197838,
      "learning_rate": 9.951720423403986e-06,
      "loss": 0.7038,
      "step": 2946
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6875255046656791,
      "learning_rate": 9.946040525441692e-06,
      "loss": 0.708,
      "step": 2947
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.688542262448344,
      "learning_rate": 9.940360644887846e-06,
      "loss": 0.7121,
      "step": 2948
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6699326839112691,
      "learning_rate": 9.934680783574901e-06,
      "loss": 0.6753,
      "step": 2949
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7423986853884421,
      "learning_rate": 9.929000943335295e-06,
      "loss": 0.759,
      "step": 2950
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.699773237326604,
      "learning_rate": 9.923321126001466e-06,
      "loss": 0.7028,
      "step": 2951
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7165672667413334,
      "learning_rate": 9.917641333405838e-06,
      "loss": 0.7491,
      "step": 2952
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6231133927055954,
      "learning_rate": 9.911961567380831e-06,
      "loss": 0.6817,
      "step": 2953
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.691130879548913,
      "learning_rate": 9.90628182975886e-06,
      "loss": 0.7358,
      "step": 2954
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7775956473464739,
      "learning_rate": 9.900602122372322e-06,
      "loss": 0.7023,
      "step": 2955
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6420070493464268,
      "learning_rate": 9.894922447053613e-06,
      "loss": 0.7088,
      "step": 2956
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7130536172252836,
      "learning_rate": 9.889242805635112e-06,
      "loss": 0.7419,
      "step": 2957
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6449896777210979,
      "learning_rate": 9.883563199949188e-06,
      "loss": 0.7167,
      "step": 2958
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8415778923886844,
      "learning_rate": 9.8778836318282e-06,
      "loss": 0.749,
      "step": 2959
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7530375984612054,
      "learning_rate": 9.872204103104502e-06,
      "loss": 0.6867,
      "step": 2960
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6981573399100688,
      "learning_rate": 9.86652461561042e-06,
      "loss": 0.758,
      "step": 2961
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7264198987805495,
      "learning_rate": 9.860845171178278e-06,
      "loss": 0.711,
      "step": 2962
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7234779361577852,
      "learning_rate": 9.855165771640387e-06,
      "loss": 0.7179,
      "step": 2963
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6372828474317224,
      "learning_rate": 9.849486418829033e-06,
      "loss": 0.7356,
      "step": 2964
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6298605437461996,
      "learning_rate": 9.843807114576503e-06,
      "loss": 0.6928,
      "step": 2965
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7446392237690583,
      "learning_rate": 9.83812786071505e-06,
      "loss": 0.7211,
      "step": 2966
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6768329564564827,
      "learning_rate": 9.83244865907692e-06,
      "loss": 0.7017,
      "step": 2967
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6052394311435881,
      "learning_rate": 9.826769511494344e-06,
      "loss": 0.7074,
      "step": 2968
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6802408041604171,
      "learning_rate": 9.821090419799536e-06,
      "loss": 0.7036,
      "step": 2969
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7474241737933711,
      "learning_rate": 9.815411385824684e-06,
      "loss": 0.7382,
      "step": 2970
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8508001613055186,
      "learning_rate": 9.809732411401969e-06,
      "loss": 0.7292,
      "step": 2971
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.867113904052375,
      "learning_rate": 9.80405349836354e-06,
      "loss": 0.784,
      "step": 2972
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6992416753702727,
      "learning_rate": 9.798374648541537e-06,
      "loss": 0.6879,
      "step": 2973
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7210124637905377,
      "learning_rate": 9.792695863768075e-06,
      "loss": 0.7368,
      "step": 2974
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6148026427957894,
      "learning_rate": 9.787017145875251e-06,
      "loss": 0.6552,
      "step": 2975
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6562934738834532,
      "learning_rate": 9.78133849669513e-06,
      "loss": 0.6963,
      "step": 2976
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6767484692549744,
      "learning_rate": 9.775659918059765e-06,
      "loss": 0.7181,
      "step": 2977
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.780839955364574,
      "learning_rate": 9.769981411801187e-06,
      "loss": 0.7161,
      "step": 2978
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6850406590184578,
      "learning_rate": 9.764302979751398e-06,
      "loss": 0.7179,
      "step": 2979
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6437582212725151,
      "learning_rate": 9.758624623742383e-06,
      "loss": 0.7,
      "step": 2980
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7531196109298124,
      "learning_rate": 9.752946345606092e-06,
      "loss": 0.7074,
      "step": 2981
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6847335883175085,
      "learning_rate": 9.747268147174458e-06,
      "loss": 0.7093,
      "step": 2982
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.747307899107552,
      "learning_rate": 9.741590030279388e-06,
      "loss": 0.7281,
      "step": 2983
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6826748944456928,
      "learning_rate": 9.735911996752761e-06,
      "loss": 0.6852,
      "step": 2984
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6845900653856597,
      "learning_rate": 9.730234048426424e-06,
      "loss": 0.6817,
      "step": 2985
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7020294594501293,
      "learning_rate": 9.724556187132205e-06,
      "loss": 0.7215,
      "step": 2986
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7100138068692066,
      "learning_rate": 9.718878414701902e-06,
      "loss": 0.6759,
      "step": 2987
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7341443773216979,
      "learning_rate": 9.713200732967278e-06,
      "loss": 0.7221,
      "step": 2988
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6849507098575401,
      "learning_rate": 9.707523143760078e-06,
      "loss": 0.7035,
      "step": 2989
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7591893312931257,
      "learning_rate": 9.701845648912002e-06,
      "loss": 0.7071,
      "step": 2990
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7448070590923006,
      "learning_rate": 9.696168250254735e-06,
      "loss": 0.753,
      "step": 2991
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6397799174484448,
      "learning_rate": 9.690490949619924e-06,
      "loss": 0.716,
      "step": 2992
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6697661235162036,
      "learning_rate": 9.684813748839186e-06,
      "loss": 0.6987,
      "step": 2993
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6922334606829206,
      "learning_rate": 9.679136649744094e-06,
      "loss": 0.7249,
      "step": 2994
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8242404448909258,
      "learning_rate": 9.67345965416621e-06,
      "loss": 0.7388,
      "step": 2995
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7985721993149135,
      "learning_rate": 9.667782763937041e-06,
      "loss": 0.7735,
      "step": 2996
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7111169091294436,
      "learning_rate": 9.662105980888079e-06,
      "loss": 0.701,
      "step": 2997
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7262907288519612,
      "learning_rate": 9.656429306850769e-06,
      "loss": 0.7092,
      "step": 2998
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6666881904521723,
      "learning_rate": 9.650752743656523e-06,
      "loss": 0.7216,
      "step": 2999
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7136346456096978,
      "learning_rate": 9.645076293136722e-06,
      "loss": 0.7177,
      "step": 3000
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.743841107797694,
      "learning_rate": 9.639399957122704e-06,
      "loss": 0.7402,
      "step": 3001
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7009448551200576,
      "learning_rate": 9.633723737445778e-06,
      "loss": 0.7022,
      "step": 3002
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7075957778393175,
      "learning_rate": 9.628047635937202e-06,
      "loss": 0.7167,
      "step": 3003
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6430648107092588,
      "learning_rate": 9.622371654428214e-06,
      "loss": 0.715,
      "step": 3004
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7479730363251198,
      "learning_rate": 9.616695794749996e-06,
      "loss": 0.7899,
      "step": 3005
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.771522759702904,
      "learning_rate": 9.611020058733703e-06,
      "loss": 0.7607,
      "step": 3006
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7389932600664068,
      "learning_rate": 9.605344448210448e-06,
      "loss": 0.748,
      "step": 3007
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.735763495693723,
      "learning_rate": 9.599668965011296e-06,
      "loss": 0.7075,
      "step": 3008
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7482967784184725,
      "learning_rate": 9.59399361096728e-06,
      "loss": 0.7166,
      "step": 3009
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6715893446826081,
      "learning_rate": 9.588318387909384e-06,
      "loss": 0.7149,
      "step": 3010
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7945284047323168,
      "learning_rate": 9.58264329766856e-06,
      "loss": 0.7916,
      "step": 3011
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6988582347537284,
      "learning_rate": 9.576968342075702e-06,
      "loss": 0.7112,
      "step": 3012
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7510081948915502,
      "learning_rate": 9.571293522961674e-06,
      "loss": 0.7246,
      "step": 3013
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6990000372961187,
      "learning_rate": 9.565618842157286e-06,
      "loss": 0.7141,
      "step": 3014
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6482790824051005,
      "learning_rate": 9.559944301493314e-06,
      "loss": 0.6944,
      "step": 3015
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6383739861904434,
      "learning_rate": 9.554269902800481e-06,
      "loss": 0.6814,
      "step": 3016
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6515207765237374,
      "learning_rate": 9.548595647909466e-06,
      "loss": 0.7135,
      "step": 3017
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7496496867345362,
      "learning_rate": 9.542921538650903e-06,
      "loss": 0.73,
      "step": 3018
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6296793173963948,
      "learning_rate": 9.537247576855374e-06,
      "loss": 0.6685,
      "step": 3019
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7115762220361718,
      "learning_rate": 9.531573764353426e-06,
      "loss": 0.75,
      "step": 3020
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5524735890803717,
      "learning_rate": 9.525900102975537e-06,
      "loss": 0.6824,
      "step": 3021
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6566519613027597,
      "learning_rate": 9.520226594552158e-06,
      "loss": 0.683,
      "step": 3022
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6108551516780315,
      "learning_rate": 9.514553240913677e-06,
      "loss": 0.7133,
      "step": 3023
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6723037870173245,
      "learning_rate": 9.508880043890437e-06,
      "loss": 0.7023,
      "step": 3024
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.678811457726932,
      "learning_rate": 9.503207005312728e-06,
      "loss": 0.7056,
      "step": 3025
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6961579590839085,
      "learning_rate": 9.497534127010791e-06,
      "loss": 0.6881,
      "step": 3026
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6554487145713547,
      "learning_rate": 9.49186141081482e-06,
      "loss": 0.684,
      "step": 3027
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6262559225919642,
      "learning_rate": 9.486188858554943e-06,
      "loss": 0.6899,
      "step": 3028
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7349265544880905,
      "learning_rate": 9.480516472061254e-06,
      "loss": 0.6895,
      "step": 3029
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7090898211195074,
      "learning_rate": 9.47484425316377e-06,
      "loss": 0.7186,
      "step": 3030
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7320361343325081,
      "learning_rate": 9.469172203692478e-06,
      "loss": 0.6905,
      "step": 3031
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6762143033864013,
      "learning_rate": 9.463500325477292e-06,
      "loss": 0.6707,
      "step": 3032
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6923775593815482,
      "learning_rate": 9.457828620348083e-06,
      "loss": 0.6945,
      "step": 3033
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6921084529324079,
      "learning_rate": 9.45215709013466e-06,
      "loss": 0.7079,
      "step": 3034
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.711796586741853,
      "learning_rate": 9.446485736666774e-06,
      "loss": 0.7183,
      "step": 3035
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7357949378729888,
      "learning_rate": 9.440814561774127e-06,
      "loss": 0.7449,
      "step": 3036
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6823207060738119,
      "learning_rate": 9.435143567286354e-06,
      "loss": 0.7057,
      "step": 3037
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1081796097596324,
      "learning_rate": 9.429472755033038e-06,
      "loss": 0.6561,
      "step": 3038
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6630637565418062,
      "learning_rate": 9.423802126843706e-06,
      "loss": 0.6818,
      "step": 3039
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6892227232047483,
      "learning_rate": 9.418131684547808e-06,
      "loss": 0.7133,
      "step": 3040
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6569931079024183,
      "learning_rate": 9.412461429974756e-06,
      "loss": 0.6926,
      "step": 3041
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7529638580380184,
      "learning_rate": 9.406791364953891e-06,
      "loss": 0.7496,
      "step": 3042
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6931603340527016,
      "learning_rate": 9.401121491314493e-06,
      "loss": 0.6872,
      "step": 3043
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7295006095036071,
      "learning_rate": 9.395451810885782e-06,
      "loss": 0.7432,
      "step": 3044
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7866564574510011,
      "learning_rate": 9.389782325496917e-06,
      "loss": 0.7135,
      "step": 3045
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6230265292206093,
      "learning_rate": 9.384113036976989e-06,
      "loss": 0.7101,
      "step": 3046
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6507705535478168,
      "learning_rate": 9.378443947155031e-06,
      "loss": 0.6892,
      "step": 3047
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7054269109307946,
      "learning_rate": 9.37277505786001e-06,
      "loss": 0.7114,
      "step": 3048
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6592927067297814,
      "learning_rate": 9.367106370920824e-06,
      "loss": 0.6899,
      "step": 3049
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7166831522121646,
      "learning_rate": 9.361437888166312e-06,
      "loss": 0.7514,
      "step": 3050
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6613154296868159,
      "learning_rate": 9.355769611425246e-06,
      "loss": 0.6744,
      "step": 3051
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7104524853595993,
      "learning_rate": 9.350101542526325e-06,
      "loss": 0.7258,
      "step": 3052
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6204271264475717,
      "learning_rate": 9.344433683298192e-06,
      "loss": 0.6959,
      "step": 3053
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6633697930964958,
      "learning_rate": 9.338766035569415e-06,
      "loss": 0.7023,
      "step": 3054
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6865966980507928,
      "learning_rate": 9.333098601168493e-06,
      "loss": 0.7224,
      "step": 3055
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6187259549406448,
      "learning_rate": 9.327431381923862e-06,
      "loss": 0.6732,
      "step": 3056
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7391154549410655,
      "learning_rate": 9.321764379663887e-06,
      "loss": 0.7269,
      "step": 3057
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7092720035855693,
      "learning_rate": 9.31609759621685e-06,
      "loss": 0.7205,
      "step": 3058
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5963724004722607,
      "learning_rate": 9.310431033410982e-06,
      "loss": 0.6869,
      "step": 3059
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6459516814894025,
      "learning_rate": 9.304764693074436e-06,
      "loss": 0.6703,
      "step": 3060
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7435337519063502,
      "learning_rate": 9.299098577035283e-06,
      "loss": 0.7116,
      "step": 3061
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6297424304402939,
      "learning_rate": 9.29343268712154e-06,
      "loss": 0.7054,
      "step": 3062
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6666435763639637,
      "learning_rate": 9.287767025161135e-06,
      "loss": 0.7343,
      "step": 3063
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.806082687468403,
      "learning_rate": 9.282101592981928e-06,
      "loss": 0.8139,
      "step": 3064
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.611456030890499,
      "learning_rate": 9.276436392411711e-06,
      "loss": 0.6305,
      "step": 3065
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7152449237634148,
      "learning_rate": 9.270771425278196e-06,
      "loss": 0.7293,
      "step": 3066
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6308778579934059,
      "learning_rate": 9.26510669340901e-06,
      "loss": 0.6744,
      "step": 3067
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6255926521578901,
      "learning_rate": 9.259442198631718e-06,
      "loss": 0.6743,
      "step": 3068
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7752029940568054,
      "learning_rate": 9.25377794277381e-06,
      "loss": 0.7233,
      "step": 3069
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6457299313771301,
      "learning_rate": 9.248113927662685e-06,
      "loss": 0.694,
      "step": 3070
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7385148368582829,
      "learning_rate": 9.242450155125678e-06,
      "loss": 0.7622,
      "step": 3071
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7186467096324591,
      "learning_rate": 9.236786626990036e-06,
      "loss": 0.6843,
      "step": 3072
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6591951290130035,
      "learning_rate": 9.231123345082934e-06,
      "loss": 0.7196,
      "step": 3073
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7059323823222571,
      "learning_rate": 9.225460311231464e-06,
      "loss": 0.7123,
      "step": 3074
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6319723957975051,
      "learning_rate": 9.219797527262642e-06,
      "loss": 0.7282,
      "step": 3075
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7814703148144614,
      "learning_rate": 9.214134995003393e-06,
      "loss": 0.6867,
      "step": 3076
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7456659586687702,
      "learning_rate": 9.208472716280571e-06,
      "loss": 0.7167,
      "step": 3077
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6984634206679956,
      "learning_rate": 9.202810692920946e-06,
      "loss": 0.6779,
      "step": 3078
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7397507317937629,
      "learning_rate": 9.197148926751202e-06,
      "loss": 0.7137,
      "step": 3079
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7300528526936676,
      "learning_rate": 9.19148741959795e-06,
      "loss": 0.7695,
      "step": 3080
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6652542634070704,
      "learning_rate": 9.185826173287702e-06,
      "loss": 0.7291,
      "step": 3081
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8967666860203597,
      "learning_rate": 9.180165189646899e-06,
      "loss": 0.7266,
      "step": 3082
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6351538972301775,
      "learning_rate": 9.174504470501889e-06,
      "loss": 0.6762,
      "step": 3083
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7475618421419711,
      "learning_rate": 9.168844017678944e-06,
      "loss": 0.7315,
      "step": 3084
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8017431802935718,
      "learning_rate": 9.163183833004236e-06,
      "loss": 0.7511,
      "step": 3085
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.742151053015735,
      "learning_rate": 9.157523918303863e-06,
      "loss": 0.7297,
      "step": 3086
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7265632098410016,
      "learning_rate": 9.15186427540383e-06,
      "loss": 0.7628,
      "step": 3087
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7694562365947958,
      "learning_rate": 9.146204906130055e-06,
      "loss": 0.7567,
      "step": 3088
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6301976924185779,
      "learning_rate": 9.140545812308372e-06,
      "loss": 0.6617,
      "step": 3089
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7076030433964401,
      "learning_rate": 9.134886995764517e-06,
      "loss": 0.7314,
      "step": 3090
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6693339653539996,
      "learning_rate": 9.129228458324151e-06,
      "loss": 0.717,
      "step": 3091
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6960279889178507,
      "learning_rate": 9.123570201812826e-06,
      "loss": 0.7552,
      "step": 3092
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.636883980356098,
      "learning_rate": 9.117912228056026e-06,
      "loss": 0.7339,
      "step": 3093
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5850882267478366,
      "learning_rate": 9.112254538879115e-06,
      "loss": 0.6593,
      "step": 3094
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6401421951529693,
      "learning_rate": 9.106597136107394e-06,
      "loss": 0.7326,
      "step": 3095
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6981521788206967,
      "learning_rate": 9.10094002156605e-06,
      "loss": 0.7493,
      "step": 3096
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5408718328043658,
      "learning_rate": 9.095283197080195e-06,
      "loss": 0.6734,
      "step": 3097
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6468691010655846,
      "learning_rate": 9.089626664474834e-06,
      "loss": 0.6829,
      "step": 3098
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.744264292035705,
      "learning_rate": 9.083970425574882e-06,
      "loss": 0.7417,
      "step": 3099
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.674748772303277,
      "learning_rate": 9.078314482205161e-06,
      "loss": 0.706,
      "step": 3100
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6971925976199982,
      "learning_rate": 9.072658836190394e-06,
      "loss": 0.6774,
      "step": 3101
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6561046435987615,
      "learning_rate": 9.067003489355211e-06,
      "loss": 0.6804,
      "step": 3102
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7316659147825776,
      "learning_rate": 9.06134844352415e-06,
      "loss": 0.7208,
      "step": 3103
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7371079323383048,
      "learning_rate": 9.05569370052164e-06,
      "loss": 0.7115,
      "step": 3104
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.745131817960178,
      "learning_rate": 9.050039262172018e-06,
      "loss": 0.7043,
      "step": 3105
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7042761820420492,
      "learning_rate": 9.044385130299529e-06,
      "loss": 0.7482,
      "step": 3106
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6985937258758458,
      "learning_rate": 9.03873130672831e-06,
      "loss": 0.6866,
      "step": 3107
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6494181694483412,
      "learning_rate": 9.033077793282404e-06,
      "loss": 0.662,
      "step": 3108
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7372052955815905,
      "learning_rate": 9.027424591785754e-06,
      "loss": 0.7475,
      "step": 3109
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.800688817454749,
      "learning_rate": 9.021771704062195e-06,
      "loss": 0.8067,
      "step": 3110
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7378659235467285,
      "learning_rate": 9.01611913193547e-06,
      "loss": 0.7275,
      "step": 3111
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7071631452458035,
      "learning_rate": 9.010466877229221e-06,
      "loss": 0.743,
      "step": 3112
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7198070198920263,
      "learning_rate": 9.004814941766978e-06,
      "loss": 0.7172,
      "step": 3113
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8151766397418856,
      "learning_rate": 8.99916332737217e-06,
      "loss": 0.8027,
      "step": 3114
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7023922289905279,
      "learning_rate": 8.993512035868133e-06,
      "loss": 0.7035,
      "step": 3115
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.649465661709674,
      "learning_rate": 8.987861069078084e-06,
      "loss": 0.6934,
      "step": 3116
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6735349922785997,
      "learning_rate": 8.982210428825148e-06,
      "loss": 0.7565,
      "step": 3117
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.697761043528414,
      "learning_rate": 8.97656011693234e-06,
      "loss": 0.7481,
      "step": 3118
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6471023039548565,
      "learning_rate": 8.970910135222562e-06,
      "loss": 0.7052,
      "step": 3119
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6280755943594942,
      "learning_rate": 8.965260485518624e-06,
      "loss": 0.7124,
      "step": 3120
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.742338958885344,
      "learning_rate": 8.959611169643218e-06,
      "loss": 0.7351,
      "step": 3121
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6627411678685352,
      "learning_rate": 8.953962189418926e-06,
      "loss": 0.7148,
      "step": 3122
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6685110768468964,
      "learning_rate": 8.94831354666823e-06,
      "loss": 0.704,
      "step": 3123
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6784783264079932,
      "learning_rate": 8.942665243213502e-06,
      "loss": 0.7416,
      "step": 3124
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6601235429782621,
      "learning_rate": 8.937017280876999e-06,
      "loss": 0.7076,
      "step": 3125
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7068870266458557,
      "learning_rate": 8.931369661480874e-06,
      "loss": 0.7107,
      "step": 3126
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6166038300712262,
      "learning_rate": 8.925722386847167e-06,
      "loss": 0.7008,
      "step": 3127
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.582282728696995,
      "learning_rate": 8.920075458797803e-06,
      "loss": 0.6711,
      "step": 3128
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6854195623966178,
      "learning_rate": 8.914428879154602e-06,
      "loss": 0.7037,
      "step": 3129
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6936965266593056,
      "learning_rate": 8.908782649739271e-06,
      "loss": 0.7059,
      "step": 3130
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7576451808314996,
      "learning_rate": 8.903136772373393e-06,
      "loss": 0.7301,
      "step": 3131
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7307473431881726,
      "learning_rate": 8.897491248878449e-06,
      "loss": 0.7482,
      "step": 3132
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6755400217106982,
      "learning_rate": 8.891846081075808e-06,
      "loss": 0.7289,
      "step": 3133
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7314381166114817,
      "learning_rate": 8.88620127078671e-06,
      "loss": 0.7219,
      "step": 3134
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6671340175121777,
      "learning_rate": 8.880556819832297e-06,
      "loss": 0.6957,
      "step": 3135
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.762103585804957,
      "learning_rate": 8.874912730033579e-06,
      "loss": 0.7173,
      "step": 3136
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6289064660413481,
      "learning_rate": 8.869269003211461e-06,
      "loss": 0.7036,
      "step": 3137
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6531767565364721,
      "learning_rate": 8.863625641186729e-06,
      "loss": 0.7309,
      "step": 3138
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7587316823444283,
      "learning_rate": 8.857982645780048e-06,
      "loss": 0.711,
      "step": 3139
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7258728426171663,
      "learning_rate": 8.85234001881196e-06,
      "loss": 0.7628,
      "step": 3140
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6988599978859436,
      "learning_rate": 8.846697762102898e-06,
      "loss": 0.7023,
      "step": 3141
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6998675817446143,
      "learning_rate": 8.841055877473174e-06,
      "loss": 0.73,
      "step": 3142
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6796181797697578,
      "learning_rate": 8.835414366742974e-06,
      "loss": 0.7434,
      "step": 3143
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7313711807611419,
      "learning_rate": 8.82977323173237e-06,
      "loss": 0.7616,
      "step": 3144
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7523758809605466,
      "learning_rate": 8.82413247426131e-06,
      "loss": 0.8084,
      "step": 3145
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6981809923306003,
      "learning_rate": 8.818492096149614e-06,
      "loss": 0.7459,
      "step": 3146
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7313509622654046,
      "learning_rate": 8.812852099216991e-06,
      "loss": 0.7406,
      "step": 3147
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7368354631038556,
      "learning_rate": 8.807212485283025e-06,
      "loss": 0.7326,
      "step": 3148
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6797537358678953,
      "learning_rate": 8.801573256167165e-06,
      "loss": 0.7035,
      "step": 3149
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6656790105732313,
      "learning_rate": 8.795934413688746e-06,
      "loss": 0.7139,
      "step": 3150
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6573778300450821,
      "learning_rate": 8.790295959666976e-06,
      "loss": 0.6944,
      "step": 3151
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6647054788592275,
      "learning_rate": 8.784657895920938e-06,
      "loss": 0.6566,
      "step": 3152
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6419791638217034,
      "learning_rate": 8.779020224269594e-06,
      "loss": 0.7188,
      "step": 3153
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6843780610558504,
      "learning_rate": 8.773382946531767e-06,
      "loss": 0.6821,
      "step": 3154
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6417003731887598,
      "learning_rate": 8.767746064526162e-06,
      "loss": 0.7215,
      "step": 3155
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6854988305993699,
      "learning_rate": 8.76210958007136e-06,
      "loss": 0.7174,
      "step": 3156
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7571397275210348,
      "learning_rate": 8.756473494985806e-06,
      "loss": 0.7331,
      "step": 3157
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5986450166995977,
      "learning_rate": 8.750837811087812e-06,
      "loss": 0.7203,
      "step": 3158
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6452948777232476,
      "learning_rate": 8.745202530195575e-06,
      "loss": 0.6874,
      "step": 3159
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7986688691522695,
      "learning_rate": 8.739567654127148e-06,
      "loss": 0.7063,
      "step": 3160
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7831580752574087,
      "learning_rate": 8.733933184700462e-06,
      "loss": 0.6609,
      "step": 3161
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7047063872180486,
      "learning_rate": 8.728299123733317e-06,
      "loss": 0.7166,
      "step": 3162
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8209045209618339,
      "learning_rate": 8.722665473043376e-06,
      "loss": 0.7308,
      "step": 3163
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6960479952010561,
      "learning_rate": 8.717032234448173e-06,
      "loss": 0.7046,
      "step": 3164
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6993721779466507,
      "learning_rate": 8.711399409765104e-06,
      "loss": 0.7078,
      "step": 3165
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7206647872752087,
      "learning_rate": 8.705767000811441e-06,
      "loss": 0.7532,
      "step": 3166
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6976530127131897,
      "learning_rate": 8.700135009404318e-06,
      "loss": 0.7103,
      "step": 3167
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7172945615249098,
      "learning_rate": 8.694503437360727e-06,
      "loss": 0.7522,
      "step": 3168
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6451497333103482,
      "learning_rate": 8.68887228649753e-06,
      "loss": 0.7133,
      "step": 3169
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6445059907194665,
      "learning_rate": 8.683241558631457e-06,
      "loss": 0.6777,
      "step": 3170
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7226203952877899,
      "learning_rate": 8.6776112555791e-06,
      "loss": 0.714,
      "step": 3171
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6625151450600654,
      "learning_rate": 8.671981379156908e-06,
      "loss": 0.6964,
      "step": 3172
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7297911996582686,
      "learning_rate": 8.6663519311812e-06,
      "loss": 0.7038,
      "step": 3173
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6398076084825293,
      "learning_rate": 8.660722913468149e-06,
      "loss": 0.6913,
      "step": 3174
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8260400422013232,
      "learning_rate": 8.655094327833795e-06,
      "loss": 0.7307,
      "step": 3175
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6380375046069453,
      "learning_rate": 8.649466176094043e-06,
      "loss": 0.7265,
      "step": 3176
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8166333562636846,
      "learning_rate": 8.643838460064645e-06,
      "loss": 0.7228,
      "step": 3177
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7085817047690589,
      "learning_rate": 8.638211181561219e-06,
      "loss": 0.6745,
      "step": 3178
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.696673553030968,
      "learning_rate": 8.632584342399244e-06,
      "loss": 0.7074,
      "step": 3179
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7383554238777205,
      "learning_rate": 8.62695794439406e-06,
      "loss": 0.7455,
      "step": 3180
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.740365572571052,
      "learning_rate": 8.621331989360852e-06,
      "loss": 0.6976,
      "step": 3181
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6841653001555065,
      "learning_rate": 8.615706479114679e-06,
      "loss": 0.7659,
      "step": 3182
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6717665664321579,
      "learning_rate": 8.61008141547044e-06,
      "loss": 0.6822,
      "step": 3183
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7556179731585634,
      "learning_rate": 8.604456800242901e-06,
      "loss": 0.7558,
      "step": 3184
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6577868172375249,
      "learning_rate": 8.598832635246688e-06,
      "loss": 0.6993,
      "step": 3185
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7378377090719177,
      "learning_rate": 8.593208922296261e-06,
      "loss": 0.7118,
      "step": 3186
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6091642503808358,
      "learning_rate": 8.587585663205952e-06,
      "loss": 0.7049,
      "step": 3187
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7792363157844445,
      "learning_rate": 8.581962859789942e-06,
      "loss": 0.7621,
      "step": 3188
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7011440082222502,
      "learning_rate": 8.576340513862265e-06,
      "loss": 0.6922,
      "step": 3189
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6666689657074251,
      "learning_rate": 8.570718627236803e-06,
      "loss": 0.7051,
      "step": 3190
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6825854541320634,
      "learning_rate": 8.565097201727302e-06,
      "loss": 0.7224,
      "step": 3191
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6747243295010569,
      "learning_rate": 8.559476239147343e-06,
      "loss": 0.706,
      "step": 3192
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7164660017535778,
      "learning_rate": 8.553855741310369e-06,
      "loss": 0.7056,
      "step": 3193
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7445467706598539,
      "learning_rate": 8.548235710029674e-06,
      "loss": 0.6913,
      "step": 3194
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6980711289457655,
      "learning_rate": 8.542616147118392e-06,
      "loss": 0.7312,
      "step": 3195
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7062091301550835,
      "learning_rate": 8.536997054389508e-06,
      "loss": 0.7366,
      "step": 3196
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6693369307875315,
      "learning_rate": 8.531378433655864e-06,
      "loss": 0.7114,
      "step": 3197
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6275655617178844,
      "learning_rate": 8.52576028673014e-06,
      "loss": 0.7303,
      "step": 3198
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6577081223224852,
      "learning_rate": 8.520142615424871e-06,
      "loss": 0.7155,
      "step": 3199
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7010407089700124,
      "learning_rate": 8.514525421552436e-06,
      "loss": 0.7201,
      "step": 3200
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6576659084616722,
      "learning_rate": 8.508908706925054e-06,
      "loss": 0.7498,
      "step": 3201
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7064950986009241,
      "learning_rate": 8.503292473354798e-06,
      "loss": 0.6769,
      "step": 3202
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7007430352704472,
      "learning_rate": 8.497676722653583e-06,
      "loss": 0.6627,
      "step": 3203
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7946195581753643,
      "learning_rate": 8.492061456633161e-06,
      "loss": 0.7323,
      "step": 3204
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7148877994610223,
      "learning_rate": 8.486446677105138e-06,
      "loss": 0.7389,
      "step": 3205
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7669787756600648,
      "learning_rate": 8.48083238588096e-06,
      "loss": 0.6863,
      "step": 3206
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.787880943976036,
      "learning_rate": 8.47521858477191e-06,
      "loss": 0.7555,
      "step": 3207
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7106808389405531,
      "learning_rate": 8.46960527558912e-06,
      "loss": 0.6964,
      "step": 3208
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6551021341406851,
      "learning_rate": 8.463992460143563e-06,
      "loss": 0.6941,
      "step": 3209
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7138018263981699,
      "learning_rate": 8.458380140246046e-06,
      "loss": 0.7305,
      "step": 3210
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7808470815975769,
      "learning_rate": 8.452768317707224e-06,
      "loss": 0.7356,
      "step": 3211
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7029351978474053,
      "learning_rate": 8.447156994337588e-06,
      "loss": 0.6878,
      "step": 3212
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7845493690192326,
      "learning_rate": 8.441546171947463e-06,
      "loss": 0.664,
      "step": 3213
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6661482892607852,
      "learning_rate": 8.435935852347018e-06,
      "loss": 0.7223,
      "step": 3214
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6023273653142036,
      "learning_rate": 8.430326037346264e-06,
      "loss": 0.6832,
      "step": 3215
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6378510889594577,
      "learning_rate": 8.42471672875504e-06,
      "loss": 0.6853,
      "step": 3216
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.794579233053927,
      "learning_rate": 8.41910792838303e-06,
      "loss": 0.7467,
      "step": 3217
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6811739984579374,
      "learning_rate": 8.413499638039743e-06,
      "loss": 0.6786,
      "step": 3218
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7982091910398583,
      "learning_rate": 8.407891859534535e-06,
      "loss": 0.7365,
      "step": 3219
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7244698132795396,
      "learning_rate": 8.402284594676593e-06,
      "loss": 0.6909,
      "step": 3220
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6701861542507341,
      "learning_rate": 8.39667784527494e-06,
      "loss": 0.7136,
      "step": 3221
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7252036128662172,
      "learning_rate": 8.391071613138422e-06,
      "loss": 0.7106,
      "step": 3222
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7386453283141501,
      "learning_rate": 8.385465900075728e-06,
      "loss": 0.6946,
      "step": 3223
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6886908939249878,
      "learning_rate": 8.379860707895387e-06,
      "loss": 0.703,
      "step": 3224
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7649945953104451,
      "learning_rate": 8.37425603840574e-06,
      "loss": 0.7276,
      "step": 3225
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7116746144496365,
      "learning_rate": 8.368651893414975e-06,
      "loss": 0.7535,
      "step": 3226
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.854501698911276,
      "learning_rate": 8.363048274731105e-06,
      "loss": 0.7543,
      "step": 3227
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6426234667808957,
      "learning_rate": 8.357445184161975e-06,
      "loss": 0.6931,
      "step": 3228
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.665234921795987,
      "learning_rate": 8.351842623515259e-06,
      "loss": 0.695,
      "step": 3229
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6859181412862501,
      "learning_rate": 8.346240594598458e-06,
      "loss": 0.7101,
      "step": 3230
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7470088112606235,
      "learning_rate": 8.34063909921891e-06,
      "loss": 0.7394,
      "step": 3231
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7606225752701712,
      "learning_rate": 8.335038139183764e-06,
      "loss": 0.7837,
      "step": 3232
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6995240795182203,
      "learning_rate": 8.32943771630001e-06,
      "loss": 0.71,
      "step": 3233
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7046118866663207,
      "learning_rate": 8.323837832374462e-06,
      "loss": 0.7682,
      "step": 3234
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6972559200893806,
      "learning_rate": 8.318238489213762e-06,
      "loss": 0.7045,
      "step": 3235
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7065025198008461,
      "learning_rate": 8.312639688624372e-06,
      "loss": 0.7317,
      "step": 3236
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6326796555140619,
      "learning_rate": 8.307041432412579e-06,
      "loss": 0.6998,
      "step": 3237
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8137968279756399,
      "learning_rate": 8.301443722384503e-06,
      "loss": 0.75,
      "step": 3238
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6750382992963789,
      "learning_rate": 8.295846560346076e-06,
      "loss": 0.7167,
      "step": 3239
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7242324798721772,
      "learning_rate": 8.290249948103068e-06,
      "loss": 0.6696,
      "step": 3240
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6769756339221216,
      "learning_rate": 8.284653887461054e-06,
      "loss": 0.7104,
      "step": 3241
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.712109780890941,
      "learning_rate": 8.279058380225441e-06,
      "loss": 0.6665,
      "step": 3242
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8386218338903201,
      "learning_rate": 8.273463428201457e-06,
      "loss": 0.7785,
      "step": 3243
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6333475066763421,
      "learning_rate": 8.267869033194153e-06,
      "loss": 0.6859,
      "step": 3244
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7383035483170511,
      "learning_rate": 8.262275197008395e-06,
      "loss": 0.7468,
      "step": 3245
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6778215029780419,
      "learning_rate": 8.256681921448872e-06,
      "loss": 0.7083,
      "step": 3246
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6423032413521454,
      "learning_rate": 8.25108920832009e-06,
      "loss": 0.6724,
      "step": 3247
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6591642560187779,
      "learning_rate": 8.245497059426376e-06,
      "loss": 0.6673,
      "step": 3248
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.67531704679156,
      "learning_rate": 8.239905476571874e-06,
      "loss": 0.7191,
      "step": 3249
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6581965748534369,
      "learning_rate": 8.234314461560547e-06,
      "loss": 0.6977,
      "step": 3250
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8219008834723858,
      "learning_rate": 8.228724016196166e-06,
      "loss": 0.7266,
      "step": 3251
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7411260686964741,
      "learning_rate": 8.22313414228233e-06,
      "loss": 0.6974,
      "step": 3252
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6627503422102986,
      "learning_rate": 8.217544841622449e-06,
      "loss": 0.6965,
      "step": 3253
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7028970063862019,
      "learning_rate": 8.211956116019748e-06,
      "loss": 0.7148,
      "step": 3254
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7153949401166667,
      "learning_rate": 8.206367967277266e-06,
      "loss": 0.6694,
      "step": 3255
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.706697840088249,
      "learning_rate": 8.200780397197854e-06,
      "loss": 0.7037,
      "step": 3256
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.716452330353835,
      "learning_rate": 8.19519340758418e-06,
      "loss": 0.7223,
      "step": 3257
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6506163205815864,
      "learning_rate": 8.189607000238729e-06,
      "loss": 0.7007,
      "step": 3258
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.638011552869826,
      "learning_rate": 8.184021176963784e-06,
      "loss": 0.7003,
      "step": 3259
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7181626753851706,
      "learning_rate": 8.178435939561451e-06,
      "loss": 0.6851,
      "step": 3260
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6904738601370856,
      "learning_rate": 8.172851289833641e-06,
      "loss": 0.727,
      "step": 3261
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7226995277920272,
      "learning_rate": 8.167267229582087e-06,
      "loss": 0.694,
      "step": 3262
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6785700836396097,
      "learning_rate": 8.161683760608314e-06,
      "loss": 0.6905,
      "step": 3263
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8118906661510069,
      "learning_rate": 8.156100884713672e-06,
      "loss": 0.6862,
      "step": 3264
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7688769381075508,
      "learning_rate": 8.150518603699306e-06,
      "loss": 0.7097,
      "step": 3265
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.710139471442234,
      "learning_rate": 8.144936919366184e-06,
      "loss": 0.6963,
      "step": 3266
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6832665591366984,
      "learning_rate": 8.139355833515073e-06,
      "loss": 0.6765,
      "step": 3267
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7805647291810696,
      "learning_rate": 8.13377534794654e-06,
      "loss": 0.7682,
      "step": 3268
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7471771512132674,
      "learning_rate": 8.128195464460971e-06,
      "loss": 0.7578,
      "step": 3269
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6526251760786028,
      "learning_rate": 8.122616184858555e-06,
      "loss": 0.6927,
      "step": 3270
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6534825532105185,
      "learning_rate": 8.117037510939278e-06,
      "loss": 0.6628,
      "step": 3271
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6798668302365671,
      "learning_rate": 8.111459444502941e-06,
      "loss": 0.7003,
      "step": 3272
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8165252785006508,
      "learning_rate": 8.105881987349143e-06,
      "loss": 0.7278,
      "step": 3273
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7859173363388562,
      "learning_rate": 8.100305141277287e-06,
      "loss": 0.7417,
      "step": 3274
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7767234892147641,
      "learning_rate": 8.094728908086585e-06,
      "loss": 0.699,
      "step": 3275
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7131002083834873,
      "learning_rate": 8.08915328957604e-06,
      "loss": 0.7416,
      "step": 3276
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8947532190435318,
      "learning_rate": 8.083578287544465e-06,
      "loss": 0.7801,
      "step": 3277
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7264026718755021,
      "learning_rate": 8.07800390379047e-06,
      "loss": 0.6754,
      "step": 3278
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6932012410565431,
      "learning_rate": 8.07243014011247e-06,
      "loss": 0.7425,
      "step": 3279
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6609515728072222,
      "learning_rate": 8.066856998308673e-06,
      "loss": 0.6884,
      "step": 3280
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6372734761520102,
      "learning_rate": 8.061284480177096e-06,
      "loss": 0.7047,
      "step": 3281
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6912714503157056,
      "learning_rate": 8.055712587515548e-06,
      "loss": 0.73,
      "step": 3282
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6613742735467674,
      "learning_rate": 8.050141322121636e-06,
      "loss": 0.6591,
      "step": 3283
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7021032693080863,
      "learning_rate": 8.044570685792767e-06,
      "loss": 0.6995,
      "step": 3284
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.669031570908645,
      "learning_rate": 8.039000680326149e-06,
      "loss": 0.6953,
      "step": 3285
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5918794005077738,
      "learning_rate": 8.033431307518772e-06,
      "loss": 0.6535,
      "step": 3286
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7298258370301614,
      "learning_rate": 8.027862569167439e-06,
      "loss": 0.7212,
      "step": 3287
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6945243336943374,
      "learning_rate": 8.02229446706874e-06,
      "loss": 0.7223,
      "step": 3288
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7005384598562276,
      "learning_rate": 8.016727003019059e-06,
      "loss": 0.7422,
      "step": 3289
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.702001787144154,
      "learning_rate": 8.011160178814577e-06,
      "loss": 0.7214,
      "step": 3290
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7291780119404446,
      "learning_rate": 8.005593996251268e-06,
      "loss": 0.6978,
      "step": 3291
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8094114093973986,
      "learning_rate": 8.000028457124896e-06,
      "loss": 0.7143,
      "step": 3292
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5587063076635681,
      "learning_rate": 7.994463563231026e-06,
      "loss": 0.6369,
      "step": 3293
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.721213306977691,
      "learning_rate": 7.988899316365001e-06,
      "loss": 0.7004,
      "step": 3294
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7513443982241949,
      "learning_rate": 7.983335718321972e-06,
      "loss": 0.684,
      "step": 3295
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6577044059879045,
      "learning_rate": 7.97777277089686e-06,
      "loss": 0.6866,
      "step": 3296
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7219185278716862,
      "learning_rate": 7.972210475884398e-06,
      "loss": 0.6902,
      "step": 3297
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6663389131653885,
      "learning_rate": 7.966648835079092e-06,
      "loss": 0.6718,
      "step": 3298
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8321714590616128,
      "learning_rate": 7.961087850275247e-06,
      "loss": 0.7497,
      "step": 3299
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.681983935148707,
      "learning_rate": 7.955527523266949e-06,
      "loss": 0.6999,
      "step": 3300
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7428477613637693,
      "learning_rate": 7.949967855848078e-06,
      "loss": 0.7113,
      "step": 3301
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6821382921342434,
      "learning_rate": 7.9444088498123e-06,
      "loss": 0.7243,
      "step": 3302
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6859043293758867,
      "learning_rate": 7.938850506953062e-06,
      "loss": 0.7221,
      "step": 3303
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6987563150394814,
      "learning_rate": 7.933292829063608e-06,
      "loss": 0.7176,
      "step": 3304
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6902952164144884,
      "learning_rate": 7.927735817936952e-06,
      "loss": 0.7042,
      "step": 3305
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6641164431266118,
      "learning_rate": 7.922179475365908e-06,
      "loss": 0.7104,
      "step": 3306
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6996531588770665,
      "learning_rate": 7.916623803143062e-06,
      "loss": 0.6747,
      "step": 3307
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6942244408290257,
      "learning_rate": 7.911068803060797e-06,
      "loss": 0.702,
      "step": 3308
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7340490353744952,
      "learning_rate": 7.905514476911266e-06,
      "loss": 0.7032,
      "step": 3309
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5541476195334465,
      "learning_rate": 7.899960826486413e-06,
      "loss": 0.6873,
      "step": 3310
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6685105309475947,
      "learning_rate": 7.894407853577962e-06,
      "loss": 0.6883,
      "step": 3311
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7049948260030878,
      "learning_rate": 7.888855559977414e-06,
      "loss": 0.6801,
      "step": 3312
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7279252534332358,
      "learning_rate": 7.883303947476063e-06,
      "loss": 0.7517,
      "step": 3313
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6533821567334013,
      "learning_rate": 7.877753017864966e-06,
      "loss": 0.7114,
      "step": 3314
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6257813895268955,
      "learning_rate": 7.87220277293497e-06,
      "loss": 0.6901,
      "step": 3315
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.661024090949632,
      "learning_rate": 7.866653214476701e-06,
      "loss": 0.7072,
      "step": 3316
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7507961365059667,
      "learning_rate": 7.861104344280566e-06,
      "loss": 0.7167,
      "step": 3317
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.694207823276907,
      "learning_rate": 7.85555616413674e-06,
      "loss": 0.7162,
      "step": 3318
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7132938776720568,
      "learning_rate": 7.850008675835184e-06,
      "loss": 0.7213,
      "step": 3319
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8227410232181064,
      "learning_rate": 7.844461881165637e-06,
      "loss": 0.7883,
      "step": 3320
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5919458951340093,
      "learning_rate": 7.838915781917603e-06,
      "loss": 0.6569,
      "step": 3321
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6587946999841867,
      "learning_rate": 7.833370379880379e-06,
      "loss": 0.6631,
      "step": 3322
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7340108162605958,
      "learning_rate": 7.827825676843018e-06,
      "loss": 0.7358,
      "step": 3323
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7483829183102781,
      "learning_rate": 7.82228167459436e-06,
      "loss": 0.7155,
      "step": 3324
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8171915631950954,
      "learning_rate": 7.816738374923015e-06,
      "loss": 0.7238,
      "step": 3325
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7337159754725803,
      "learning_rate": 7.811195779617372e-06,
      "loss": 0.7312,
      "step": 3326
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7131469611391608,
      "learning_rate": 7.805653890465578e-06,
      "loss": 0.7021,
      "step": 3327
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8105793556859844,
      "learning_rate": 7.800112709255572e-06,
      "loss": 0.7279,
      "step": 3328
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9702632612465497,
      "learning_rate": 7.794572237775049e-06,
      "loss": 0.69,
      "step": 3329
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6988697584646006,
      "learning_rate": 7.78903247781148e-06,
      "loss": 0.7092,
      "step": 3330
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7375374287626989,
      "learning_rate": 7.783493431152115e-06,
      "loss": 0.7566,
      "step": 3331
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8258694682278745,
      "learning_rate": 7.777955099583957e-06,
      "loss": 0.7422,
      "step": 3332
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7128986834064747,
      "learning_rate": 7.772417484893787e-06,
      "loss": 0.724,
      "step": 3333
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6388438614395524,
      "learning_rate": 7.76688058886816e-06,
      "loss": 0.6807,
      "step": 3334
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6519137775122914,
      "learning_rate": 7.76134441329339e-06,
      "loss": 0.6872,
      "step": 3335
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6872791931272423,
      "learning_rate": 7.755808959955564e-06,
      "loss": 0.7032,
      "step": 3336
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6629140345821386,
      "learning_rate": 7.750274230640539e-06,
      "loss": 0.717,
      "step": 3337
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7437772509698569,
      "learning_rate": 7.744740227133926e-06,
      "loss": 0.7592,
      "step": 3338
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.659420945768417,
      "learning_rate": 7.739206951221115e-06,
      "loss": 0.698,
      "step": 3339
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.722833026195615,
      "learning_rate": 7.733674404687259e-06,
      "loss": 0.7056,
      "step": 3340
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6506933189213029,
      "learning_rate": 7.728142589317267e-06,
      "loss": 0.6817,
      "step": 3341
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6537942339835989,
      "learning_rate": 7.722611506895817e-06,
      "loss": 0.6868,
      "step": 3342
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.627068817025234,
      "learning_rate": 7.717081159207356e-06,
      "loss": 0.7035,
      "step": 3343
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6514564804810191,
      "learning_rate": 7.711551548036085e-06,
      "loss": 0.6664,
      "step": 3344
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6953246190767878,
      "learning_rate": 7.706022675165977e-06,
      "loss": 0.7297,
      "step": 3345
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7091531229632937,
      "learning_rate": 7.700494542380758e-06,
      "loss": 0.6906,
      "step": 3346
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5626592616501356,
      "learning_rate": 7.69496715146392e-06,
      "loss": 0.6504,
      "step": 3347
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6864871751702908,
      "learning_rate": 7.689440504198714e-06,
      "loss": 0.7024,
      "step": 3348
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7336480465994055,
      "learning_rate": 7.683914602368156e-06,
      "loss": 0.7352,
      "step": 3349
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6385242639158475,
      "learning_rate": 7.67838944775501e-06,
      "loss": 0.6935,
      "step": 3350
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7145116893617449,
      "learning_rate": 7.672865042141808e-06,
      "loss": 0.7507,
      "step": 3351
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.708434708567746,
      "learning_rate": 7.667341387310842e-06,
      "loss": 0.7368,
      "step": 3352
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6839714487870097,
      "learning_rate": 7.661818485044152e-06,
      "loss": 0.6919,
      "step": 3353
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.656366429833009,
      "learning_rate": 7.656296337123545e-06,
      "loss": 0.713,
      "step": 3354
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6988184065273012,
      "learning_rate": 7.650774945330584e-06,
      "loss": 0.6792,
      "step": 3355
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6882306557902506,
      "learning_rate": 7.64525431144658e-06,
      "loss": 0.7047,
      "step": 3356
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6687751221239941,
      "learning_rate": 7.639734437252609e-06,
      "loss": 0.7001,
      "step": 3357
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6604456037493168,
      "learning_rate": 7.634215324529491e-06,
      "loss": 0.6731,
      "step": 3358
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7252921168305162,
      "learning_rate": 7.628696975057817e-06,
      "loss": 0.6985,
      "step": 3359
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6574128599053511,
      "learning_rate": 7.62317939061791e-06,
      "loss": 0.6631,
      "step": 3360
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7044602257643034,
      "learning_rate": 7.617662572989866e-06,
      "loss": 0.6645,
      "step": 3361
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7123473702112443,
      "learning_rate": 7.612146523953518e-06,
      "loss": 0.7046,
      "step": 3362
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7297861921296619,
      "learning_rate": 7.606631245288465e-06,
      "loss": 0.7027,
      "step": 3363
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7207058694739282,
      "learning_rate": 7.60111673877405e-06,
      "loss": 0.6803,
      "step": 3364
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6812020787979006,
      "learning_rate": 7.595603006189365e-06,
      "loss": 0.735,
      "step": 3365
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7177237489750051,
      "learning_rate": 7.590090049313257e-06,
      "loss": 0.6932,
      "step": 3366
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6260119761089193,
      "learning_rate": 7.584577869924319e-06,
      "loss": 0.7041,
      "step": 3367
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7218069804025545,
      "learning_rate": 7.5790664698008995e-06,
      "loss": 0.7149,
      "step": 3368
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6656172207618545,
      "learning_rate": 7.573555850721084e-06,
      "loss": 0.7242,
      "step": 3369
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7197817385646037,
      "learning_rate": 7.5680460144627185e-06,
      "loss": 0.6826,
      "step": 3370
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6271452150248662,
      "learning_rate": 7.562536962803387e-06,
      "loss": 0.6974,
      "step": 3371
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6768455085170326,
      "learning_rate": 7.557028697520426e-06,
      "loss": 0.6819,
      "step": 3372
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7559405757869287,
      "learning_rate": 7.551521220390921e-06,
      "loss": 0.7708,
      "step": 3373
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7250401407469317,
      "learning_rate": 7.54601453319169e-06,
      "loss": 0.7042,
      "step": 3374
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8424563516565612,
      "learning_rate": 7.5405086376993154e-06,
      "loss": 0.7869,
      "step": 3375
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7012897375453875,
      "learning_rate": 7.5350035356901045e-06,
      "loss": 0.712,
      "step": 3376
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7005549718535653,
      "learning_rate": 7.529499228940126e-06,
      "loss": 0.7148,
      "step": 3377
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6792353065174003,
      "learning_rate": 7.523995719225177e-06,
      "loss": 0.6832,
      "step": 3378
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.582308812504723,
      "learning_rate": 7.518493008320809e-06,
      "loss": 0.6495,
      "step": 3379
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6543828283227073,
      "learning_rate": 7.512991098002307e-06,
      "loss": 0.7059,
      "step": 3380
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6681931858056676,
      "learning_rate": 7.507489990044706e-06,
      "loss": 0.6978,
      "step": 3381
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6266207914045867,
      "learning_rate": 7.501989686222775e-06,
      "loss": 0.7002,
      "step": 3382
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6777909580225416,
      "learning_rate": 7.4964901883110285e-06,
      "loss": 0.7095,
      "step": 3383
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7741064940769946,
      "learning_rate": 7.490991498083721e-06,
      "loss": 0.7372,
      "step": 3384
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6603597730805774,
      "learning_rate": 7.485493617314839e-06,
      "loss": 0.6848,
      "step": 3385
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6543350809272596,
      "learning_rate": 7.479996547778123e-06,
      "loss": 0.7294,
      "step": 3386
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6842411497930239,
      "learning_rate": 7.474500291247031e-06,
      "loss": 0.7122,
      "step": 3387
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6940235695139518,
      "learning_rate": 7.469004849494777e-06,
      "loss": 0.643,
      "step": 3388
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6800366345729726,
      "learning_rate": 7.4635102242942995e-06,
      "loss": 0.6837,
      "step": 3389
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.713981865366141,
      "learning_rate": 7.458016417418285e-06,
      "loss": 0.7028,
      "step": 3390
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8469028451781124,
      "learning_rate": 7.452523430639147e-06,
      "loss": 0.7265,
      "step": 3391
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1720780948189036,
      "learning_rate": 7.4470312657290364e-06,
      "loss": 0.7322,
      "step": 3392
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7326868908586083,
      "learning_rate": 7.441539924459846e-06,
      "loss": 0.7279,
      "step": 3393
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7433536273760764,
      "learning_rate": 7.436049408603188e-06,
      "loss": 0.7549,
      "step": 3394
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6908312087538637,
      "learning_rate": 7.430559719930428e-06,
      "loss": 0.7297,
      "step": 3395
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7676129462334456,
      "learning_rate": 7.425070860212644e-06,
      "loss": 0.7371,
      "step": 3396
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7782911853306689,
      "learning_rate": 7.41958283122066e-06,
      "loss": 0.6827,
      "step": 3397
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6230125611341245,
      "learning_rate": 7.41409563472503e-06,
      "loss": 0.6612,
      "step": 3398
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7234226512379147,
      "learning_rate": 7.4086092724960355e-06,
      "loss": 0.7608,
      "step": 3399
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7819668498446872,
      "learning_rate": 7.403123746303693e-06,
      "loss": 0.7657,
      "step": 3400
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6480008076164091,
      "learning_rate": 7.397639057917746e-06,
      "loss": 0.686,
      "step": 3401
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7124393128284625,
      "learning_rate": 7.392155209107673e-06,
      "loss": 0.7682,
      "step": 3402
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6356451989282638,
      "learning_rate": 7.386672201642673e-06,
      "loss": 0.6992,
      "step": 3403
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6608874445200069,
      "learning_rate": 7.381190037291686e-06,
      "loss": 0.6844,
      "step": 3404
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6692640166358949,
      "learning_rate": 7.375708717823364e-06,
      "loss": 0.7042,
      "step": 3405
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6225604534711455,
      "learning_rate": 7.370228245006097e-06,
      "loss": 0.6861,
      "step": 3406
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6614000595403748,
      "learning_rate": 7.364748620608e-06,
      "loss": 0.662,
      "step": 3407
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7363304345031013,
      "learning_rate": 7.359269846396918e-06,
      "loss": 0.7403,
      "step": 3408
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6775594963886935,
      "learning_rate": 7.353791924140412e-06,
      "loss": 0.6524,
      "step": 3409
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7131112344871606,
      "learning_rate": 7.348314855605779e-06,
      "loss": 0.7134,
      "step": 3410
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6815864693237091,
      "learning_rate": 7.342838642560031e-06,
      "loss": 0.7162,
      "step": 3411
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8149662141123207,
      "learning_rate": 7.33736328676991e-06,
      "loss": 0.781,
      "step": 3412
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8110361686811369,
      "learning_rate": 7.331888790001886e-06,
      "loss": 0.7289,
      "step": 3413
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7280603067038471,
      "learning_rate": 7.326415154022136e-06,
      "loss": 0.7627,
      "step": 3414
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5318153908514162,
      "learning_rate": 7.320942380596572e-06,
      "loss": 0.6661,
      "step": 3415
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6730791718914293,
      "learning_rate": 7.315470471490827e-06,
      "loss": 0.7085,
      "step": 3416
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.674065622819871,
      "learning_rate": 7.309999428470251e-06,
      "loss": 0.6667,
      "step": 3417
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6749923365765782,
      "learning_rate": 7.304529253299917e-06,
      "loss": 0.6545,
      "step": 3418
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.834864880371418,
      "learning_rate": 7.29905994774462e-06,
      "loss": 0.7722,
      "step": 3419
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6485698748818973,
      "learning_rate": 7.293591513568867e-06,
      "loss": 0.7094,
      "step": 3420
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7930193121304756,
      "learning_rate": 7.288123952536893e-06,
      "loss": 0.7549,
      "step": 3421
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7230173437158197,
      "learning_rate": 7.282657266412647e-06,
      "loss": 0.7702,
      "step": 3422
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7085729166119135,
      "learning_rate": 7.277191456959796e-06,
      "loss": 0.7601,
      "step": 3423
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6649396803073762,
      "learning_rate": 7.271726525941719e-06,
      "loss": 0.6998,
      "step": 3424
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6583346588078218,
      "learning_rate": 7.266262475121522e-06,
      "loss": 0.651,
      "step": 3425
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7700045933451152,
      "learning_rate": 7.2607993062620184e-06,
      "loss": 0.7357,
      "step": 3426
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7139802577742989,
      "learning_rate": 7.255337021125741e-06,
      "loss": 0.7163,
      "step": 3427
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6386615061442688,
      "learning_rate": 7.24987562147494e-06,
      "loss": 0.6746,
      "step": 3428
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6749649793120197,
      "learning_rate": 7.244415109071572e-06,
      "loss": 0.7201,
      "step": 3429
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6725156862260886,
      "learning_rate": 7.2389554856773145e-06,
      "loss": 0.6955,
      "step": 3430
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6008361984074418,
      "learning_rate": 7.233496753053559e-06,
      "loss": 0.6741,
      "step": 3431
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6433061289594628,
      "learning_rate": 7.228038912961403e-06,
      "loss": 0.7428,
      "step": 3432
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7349096398912348,
      "learning_rate": 7.222581967161658e-06,
      "loss": 0.7597,
      "step": 3433
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7146754276840791,
      "learning_rate": 7.217125917414851e-06,
      "loss": 0.6678,
      "step": 3434
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6714425866717303,
      "learning_rate": 7.211670765481214e-06,
      "loss": 0.6631,
      "step": 3435
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6573410564893297,
      "learning_rate": 7.206216513120696e-06,
      "loss": 0.6906,
      "step": 3436
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6898246405142024,
      "learning_rate": 7.200763162092953e-06,
      "loss": 0.7485,
      "step": 3437
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6798978065895742,
      "learning_rate": 7.195310714157347e-06,
      "loss": 0.705,
      "step": 3438
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7465640579243081,
      "learning_rate": 7.189859171072953e-06,
      "loss": 0.7326,
      "step": 3439
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7717183846501479,
      "learning_rate": 7.184408534598551e-06,
      "loss": 0.7821,
      "step": 3440
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6669994874604241,
      "learning_rate": 7.1789588064926344e-06,
      "loss": 0.6496,
      "step": 3441
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6441368203176757,
      "learning_rate": 7.173509988513392e-06,
      "loss": 0.6933,
      "step": 3442
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7189328619420947,
      "learning_rate": 7.168062082418731e-06,
      "loss": 0.7077,
      "step": 3443
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.76887246653179,
      "learning_rate": 7.162615089966256e-06,
      "loss": 0.7195,
      "step": 3444
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7135916601294435,
      "learning_rate": 7.157169012913283e-06,
      "loss": 0.7543,
      "step": 3445
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6491724674253859,
      "learning_rate": 7.15172385301683e-06,
      "loss": 0.6822,
      "step": 3446
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6675125356374901,
      "learning_rate": 7.146279612033618e-06,
      "loss": 0.6607,
      "step": 3447
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6598811709978598,
      "learning_rate": 7.140836291720074e-06,
      "loss": 0.6951,
      "step": 3448
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6620673306372166,
      "learning_rate": 7.1353938938323235e-06,
      "loss": 0.6803,
      "step": 3449
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6307714780644131,
      "learning_rate": 7.129952420126203e-06,
      "loss": 0.6851,
      "step": 3450
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6860317610196989,
      "learning_rate": 7.124511872357238e-06,
      "loss": 0.7255,
      "step": 3451
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.729479672794354,
      "learning_rate": 7.119072252280668e-06,
      "loss": 0.7318,
      "step": 3452
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.642995735574329,
      "learning_rate": 7.113633561651425e-06,
      "loss": 0.7222,
      "step": 3453
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6735412071633637,
      "learning_rate": 7.108195802224145e-06,
      "loss": 0.6687,
      "step": 3454
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6773039678044892,
      "learning_rate": 7.102758975753163e-06,
      "loss": 0.6921,
      "step": 3455
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6451129807233945,
      "learning_rate": 7.097323083992512e-06,
      "loss": 0.7054,
      "step": 3456
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6896762789599247,
      "learning_rate": 7.091888128695925e-06,
      "loss": 0.7008,
      "step": 3457
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6729868913922347,
      "learning_rate": 7.08645411161683e-06,
      "loss": 0.6762,
      "step": 3458
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5718881372551453,
      "learning_rate": 7.081021034508358e-06,
      "loss": 0.6684,
      "step": 3459
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6350990948106706,
      "learning_rate": 7.0755888991233264e-06,
      "loss": 0.6949,
      "step": 3460
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.688691291140142,
      "learning_rate": 7.07015770721426e-06,
      "loss": 0.6877,
      "step": 3461
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6123814596485202,
      "learning_rate": 7.064727460533373e-06,
      "loss": 0.6795,
      "step": 3462
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6630116945720456,
      "learning_rate": 7.059298160832577e-06,
      "loss": 0.7005,
      "step": 3463
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.77245909562748,
      "learning_rate": 7.053869809863476e-06,
      "loss": 0.7244,
      "step": 3464
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6618831654170163,
      "learning_rate": 7.0484424093773686e-06,
      "loss": 0.6829,
      "step": 3465
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6957384902034333,
      "learning_rate": 7.04301596112525e-06,
      "loss": 0.7269,
      "step": 3466
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6745421450234722,
      "learning_rate": 7.037590466857804e-06,
      "loss": 0.7113,
      "step": 3467
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6553567323905962,
      "learning_rate": 7.032165928325411e-06,
      "loss": 0.6554,
      "step": 3468
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6295412163332061,
      "learning_rate": 7.026742347278133e-06,
      "loss": 0.7072,
      "step": 3469
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6472255812468555,
      "learning_rate": 7.021319725465737e-06,
      "loss": 0.6978,
      "step": 3470
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6688172068315736,
      "learning_rate": 7.01589806463767e-06,
      "loss": 0.6994,
      "step": 3471
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8100666019682701,
      "learning_rate": 7.0104773665430755e-06,
      "loss": 0.7941,
      "step": 3472
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6247421055028334,
      "learning_rate": 7.0050576329307805e-06,
      "loss": 0.6703,
      "step": 3473
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7139237456451079,
      "learning_rate": 6.999638865549306e-06,
      "loss": 0.7092,
      "step": 3474
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7402746260187192,
      "learning_rate": 6.99422106614686e-06,
      "loss": 0.7441,
      "step": 3475
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7445871425521307,
      "learning_rate": 6.9888042364713345e-06,
      "loss": 0.6918,
      "step": 3476
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6916181508720427,
      "learning_rate": 6.9833883782703195e-06,
      "loss": 0.6905,
      "step": 3477
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6737336705618632,
      "learning_rate": 6.977973493291075e-06,
      "loss": 0.7121,
      "step": 3478
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6891682225069716,
      "learning_rate": 6.972559583280556e-06,
      "loss": 0.7107,
      "step": 3479
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8149934351232544,
      "learning_rate": 6.967146649985406e-06,
      "loss": 0.7669,
      "step": 3480
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6988034121613637,
      "learning_rate": 6.961734695151952e-06,
      "loss": 0.6591,
      "step": 3481
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6289297543230069,
      "learning_rate": 6.956323720526199e-06,
      "loss": 0.6978,
      "step": 3482
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7296322138610525,
      "learning_rate": 6.950913727853841e-06,
      "loss": 0.7725,
      "step": 3483
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7911227638873154,
      "learning_rate": 6.945504718880258e-06,
      "loss": 0.7295,
      "step": 3484
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6891998913419852,
      "learning_rate": 6.940096695350504e-06,
      "loss": 0.705,
      "step": 3485
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6363486569909487,
      "learning_rate": 6.934689659009325e-06,
      "loss": 0.7317,
      "step": 3486
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6929111834317353,
      "learning_rate": 6.929283611601143e-06,
      "loss": 0.7134,
      "step": 3487
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7172606059796671,
      "learning_rate": 6.923878554870056e-06,
      "loss": 0.7323,
      "step": 3488
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7639331955752773,
      "learning_rate": 6.918474490559851e-06,
      "loss": 0.7302,
      "step": 3489
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7331160663460403,
      "learning_rate": 6.913071420413995e-06,
      "loss": 0.7132,
      "step": 3490
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6619072152598279,
      "learning_rate": 6.907669346175626e-06,
      "loss": 0.744,
      "step": 3491
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7251266868036972,
      "learning_rate": 6.902268269587571e-06,
      "loss": 0.7621,
      "step": 3492
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.733650102561035,
      "learning_rate": 6.8968681923923234e-06,
      "loss": 0.7078,
      "step": 3493
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7113133055117664,
      "learning_rate": 6.891469116332066e-06,
      "loss": 0.7293,
      "step": 3494
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6063167171950443,
      "learning_rate": 6.886071043148652e-06,
      "loss": 0.6639,
      "step": 3495
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6605939168415446,
      "learning_rate": 6.880673974583614e-06,
      "loss": 0.6745,
      "step": 3496
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7109782510815349,
      "learning_rate": 6.875277912378153e-06,
      "loss": 0.7517,
      "step": 3497
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6671919145522212,
      "learning_rate": 6.869882858273155e-06,
      "loss": 0.7038,
      "step": 3498
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6645865544016955,
      "learning_rate": 6.864488814009178e-06,
      "loss": 0.7111,
      "step": 3499
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7425253228876507,
      "learning_rate": 6.859095781326451e-06,
      "loss": 0.7056,
      "step": 3500
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6056756059308238,
      "learning_rate": 6.8537037619648795e-06,
      "loss": 0.6914,
      "step": 3501
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.675618246345062,
      "learning_rate": 6.84831275766404e-06,
      "loss": 0.7331,
      "step": 3502
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7364072831098396,
      "learning_rate": 6.842922770163184e-06,
      "loss": 0.7321,
      "step": 3503
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9529792643942718,
      "learning_rate": 6.837533801201236e-06,
      "loss": 0.7109,
      "step": 3504
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6606730556334082,
      "learning_rate": 6.832145852516789e-06,
      "loss": 0.7016,
      "step": 3505
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6603722791072972,
      "learning_rate": 6.826758925848102e-06,
      "loss": 0.7172,
      "step": 3506
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6143104945676422,
      "learning_rate": 6.821373022933118e-06,
      "loss": 0.6618,
      "step": 3507
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6141769696141088,
      "learning_rate": 6.815988145509434e-06,
      "loss": 0.6547,
      "step": 3508
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.717636235121684,
      "learning_rate": 6.810604295314327e-06,
      "loss": 0.7331,
      "step": 3509
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.755001096689816,
      "learning_rate": 6.80522147408474e-06,
      "loss": 0.6657,
      "step": 3510
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.682176525565654,
      "learning_rate": 6.799839683557282e-06,
      "loss": 0.6478,
      "step": 3511
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.754598975647094,
      "learning_rate": 6.794458925468231e-06,
      "loss": 0.7393,
      "step": 3512
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7434344087292234,
      "learning_rate": 6.7890792015535345e-06,
      "loss": 0.7142,
      "step": 3513
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7211972362788732,
      "learning_rate": 6.783700513548803e-06,
      "loss": 0.737,
      "step": 3514
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7232044350767919,
      "learning_rate": 6.778322863189308e-06,
      "loss": 0.7582,
      "step": 3515
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6983250175344458,
      "learning_rate": 6.772946252209995e-06,
      "loss": 0.6797,
      "step": 3516
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6757854429444868,
      "learning_rate": 6.767570682345471e-06,
      "loss": 0.693,
      "step": 3517
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6594170481058534,
      "learning_rate": 6.7621961553300065e-06,
      "loss": 0.7385,
      "step": 3518
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6777713162121038,
      "learning_rate": 6.756822672897538e-06,
      "loss": 0.7242,
      "step": 3519
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8011934757379765,
      "learning_rate": 6.751450236781658e-06,
      "loss": 0.744,
      "step": 3520
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6499848031179054,
      "learning_rate": 6.746078848715631e-06,
      "loss": 0.7285,
      "step": 3521
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.751283458057844,
      "learning_rate": 6.740708510432377e-06,
      "loss": 0.7491,
      "step": 3522
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7209982237308328,
      "learning_rate": 6.735339223664481e-06,
      "loss": 0.7266,
      "step": 3523
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6412065937271425,
      "learning_rate": 6.729970990144182e-06,
      "loss": 0.6801,
      "step": 3524
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6648004950421383,
      "learning_rate": 6.724603811603389e-06,
      "loss": 0.6723,
      "step": 3525
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6585384349158951,
      "learning_rate": 6.719237689773661e-06,
      "loss": 0.7011,
      "step": 3526
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6930230610354272,
      "learning_rate": 6.713872626386224e-06,
      "loss": 0.6624,
      "step": 3527
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7559715673618962,
      "learning_rate": 6.708508623171961e-06,
      "loss": 0.763,
      "step": 3528
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8017104494936862,
      "learning_rate": 6.703145681861405e-06,
      "loss": 0.817,
      "step": 3529
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.801840792301606,
      "learning_rate": 6.69778380418476e-06,
      "loss": 0.6995,
      "step": 3530
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6618526057113431,
      "learning_rate": 6.692422991871873e-06,
      "loss": 0.6759,
      "step": 3531
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7203153467587843,
      "learning_rate": 6.687063246652261e-06,
      "loss": 0.7452,
      "step": 3532
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7317443594495058,
      "learning_rate": 6.681704570255081e-06,
      "loss": 0.7041,
      "step": 3533
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7681996356870799,
      "learning_rate": 6.67634696440916e-06,
      "loss": 0.7177,
      "step": 3534
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6555069193869607,
      "learning_rate": 6.670990430842969e-06,
      "loss": 0.7302,
      "step": 3535
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7605645099011664,
      "learning_rate": 6.6656349712846404e-06,
      "loss": 0.7572,
      "step": 3536
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6444102465912194,
      "learning_rate": 6.660280587461956e-06,
      "loss": 0.7078,
      "step": 3537
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6699350885670773,
      "learning_rate": 6.654927281102351e-06,
      "loss": 0.6931,
      "step": 3538
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6418678657321025,
      "learning_rate": 6.649575053932916e-06,
      "loss": 0.7247,
      "step": 3539
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7122046140955346,
      "learning_rate": 6.644223907680387e-06,
      "loss": 0.7377,
      "step": 3540
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6906829297510397,
      "learning_rate": 6.638873844071163e-06,
      "loss": 0.7199,
      "step": 3541
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6643639987606521,
      "learning_rate": 6.633524864831275e-06,
      "loss": 0.7248,
      "step": 3542
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.815356195843113,
      "learning_rate": 6.628176971686424e-06,
      "loss": 0.7392,
      "step": 3543
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7138066339445679,
      "learning_rate": 6.622830166361946e-06,
      "loss": 0.721,
      "step": 3544
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7359853054312335,
      "learning_rate": 6.617484450582837e-06,
      "loss": 0.7293,
      "step": 3545
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6700307367957531,
      "learning_rate": 6.612139826073732e-06,
      "loss": 0.7097,
      "step": 3546
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.733028994006544,
      "learning_rate": 6.606796294558919e-06,
      "loss": 0.763,
      "step": 3547
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6802847499161749,
      "learning_rate": 6.6014538577623365e-06,
      "loss": 0.7032,
      "step": 3548
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6983702893665593,
      "learning_rate": 6.59611251740756e-06,
      "loss": 0.7163,
      "step": 3549
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6391064270785828,
      "learning_rate": 6.590772275217823e-06,
      "loss": 0.7553,
      "step": 3550
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6605545836099191,
      "learning_rate": 6.585433132915996e-06,
      "loss": 0.6967,
      "step": 3551
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7616218935053916,
      "learning_rate": 6.580095092224598e-06,
      "loss": 0.6967,
      "step": 3552
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7394205109878593,
      "learning_rate": 6.574758154865789e-06,
      "loss": 0.7344,
      "step": 3553
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6863800018816526,
      "learning_rate": 6.569422322561382e-06,
      "loss": 0.7097,
      "step": 3554
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6779214882172971,
      "learning_rate": 6.564087597032822e-06,
      "loss": 0.7137,
      "step": 3555
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6730996499287865,
      "learning_rate": 6.558753980001207e-06,
      "loss": 0.7188,
      "step": 3556
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7034262067497852,
      "learning_rate": 6.5534214731872714e-06,
      "loss": 0.7033,
      "step": 3557
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6738700075722501,
      "learning_rate": 6.548090078311393e-06,
      "loss": 0.6829,
      "step": 3558
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7926751945275361,
      "learning_rate": 6.542759797093592e-06,
      "loss": 0.7035,
      "step": 3559
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6830614287649324,
      "learning_rate": 6.537430631253528e-06,
      "loss": 0.7035,
      "step": 3560
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7595964375780617,
      "learning_rate": 6.532102582510497e-06,
      "loss": 0.7086,
      "step": 3561
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7583561148587151,
      "learning_rate": 6.526775652583443e-06,
      "loss": 0.677,
      "step": 3562
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.669467024504758,
      "learning_rate": 6.521449843190944e-06,
      "loss": 0.6817,
      "step": 3563
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7199453761428952,
      "learning_rate": 6.516125156051215e-06,
      "loss": 0.7413,
      "step": 3564
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6710070894575191,
      "learning_rate": 6.510801592882112e-06,
      "loss": 0.7119,
      "step": 3565
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6958176206958299,
      "learning_rate": 6.50547915540113e-06,
      "loss": 0.7147,
      "step": 3566
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.782483019983758,
      "learning_rate": 6.500157845325395e-06,
      "loss": 0.7208,
      "step": 3567
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7538377682010605,
      "learning_rate": 6.494837664371674e-06,
      "loss": 0.6951,
      "step": 3568
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6811678019487833,
      "learning_rate": 6.489518614256369e-06,
      "loss": 0.6766,
      "step": 3569
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7010759008612599,
      "learning_rate": 6.484200696695513e-06,
      "loss": 0.7243,
      "step": 3570
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6335661092072302,
      "learning_rate": 6.478883913404779e-06,
      "loss": 0.6516,
      "step": 3571
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6820210057239217,
      "learning_rate": 6.473568266099475e-06,
      "loss": 0.7054,
      "step": 3572
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7082192837564041,
      "learning_rate": 6.468253756494533e-06,
      "loss": 0.6978,
      "step": 3573
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6815656053979829,
      "learning_rate": 6.462940386304532e-06,
      "loss": 0.7055,
      "step": 3574
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6844307183115025,
      "learning_rate": 6.45762815724367e-06,
      "loss": 0.7053,
      "step": 3575
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6471864662307,
      "learning_rate": 6.452317071025787e-06,
      "loss": 0.6757,
      "step": 3576
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7935411612904745,
      "learning_rate": 6.447007129364349e-06,
      "loss": 0.7608,
      "step": 3577
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6161809462635678,
      "learning_rate": 6.441698333972455e-06,
      "loss": 0.6918,
      "step": 3578
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7166901722820817,
      "learning_rate": 6.436390686562826e-06,
      "loss": 0.734,
      "step": 3579
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7363548395688689,
      "learning_rate": 6.431084188847827e-06,
      "loss": 0.6951,
      "step": 3580
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6699949289179976,
      "learning_rate": 6.425778842539443e-06,
      "loss": 0.7101,
      "step": 3581
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7201646031864687,
      "learning_rate": 6.420474649349289e-06,
      "loss": 0.6923,
      "step": 3582
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8349460435248445,
      "learning_rate": 6.415171610988609e-06,
      "loss": 0.7006,
      "step": 3583
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6684526874780929,
      "learning_rate": 6.40986972916827e-06,
      "loss": 0.7095,
      "step": 3584
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6882584420529697,
      "learning_rate": 6.404569005598773e-06,
      "loss": 0.7479,
      "step": 3585
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7265728048851733,
      "learning_rate": 6.399269441990244e-06,
      "loss": 0.6603,
      "step": 3586
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7172050180888494,
      "learning_rate": 6.3939710400524314e-06,
      "loss": 0.6972,
      "step": 3587
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7533045586491786,
      "learning_rate": 6.388673801494703e-06,
      "loss": 0.7254,
      "step": 3588
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6231632107264126,
      "learning_rate": 6.383377728026068e-06,
      "loss": 0.6989,
      "step": 3589
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7307613540245075,
      "learning_rate": 6.378082821355145e-06,
      "loss": 0.6798,
      "step": 3590
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.788967482542082,
      "learning_rate": 6.372789083190181e-06,
      "loss": 0.7546,
      "step": 3591
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.747798982896809,
      "learning_rate": 6.3674965152390486e-06,
      "loss": 0.7587,
      "step": 3592
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6761378842313815,
      "learning_rate": 6.3622051192092395e-06,
      "loss": 0.7007,
      "step": 3593
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8221290711441427,
      "learning_rate": 6.356914896807866e-06,
      "loss": 0.703,
      "step": 3594
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7368701363479745,
      "learning_rate": 6.351625849741669e-06,
      "loss": 0.7394,
      "step": 3595
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6932362566982139,
      "learning_rate": 6.346337979717005e-06,
      "loss": 0.7067,
      "step": 3596
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6955867002967538,
      "learning_rate": 6.341051288439845e-06,
      "loss": 0.7153,
      "step": 3597
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6487895585604765,
      "learning_rate": 6.335765777615792e-06,
      "loss": 0.6934,
      "step": 3598
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7408844018176473,
      "learning_rate": 6.330481448950058e-06,
      "loss": 0.7353,
      "step": 3599
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6547500871349252,
      "learning_rate": 6.3251983041474776e-06,
      "loss": 0.6583,
      "step": 3600
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6830161107958923,
      "learning_rate": 6.319916344912506e-06,
      "loss": 0.7159,
      "step": 3601
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6946401526592234,
      "learning_rate": 6.314635572949213e-06,
      "loss": 0.6836,
      "step": 3602
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7332321426373223,
      "learning_rate": 6.3093559899612854e-06,
      "loss": 0.7222,
      "step": 3603
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.831443181345126,
      "learning_rate": 6.304077597652023e-06,
      "loss": 0.7398,
      "step": 3604
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7071637196526387,
      "learning_rate": 6.298800397724354e-06,
      "loss": 0.7037,
      "step": 3605
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6179392527680335,
      "learning_rate": 6.293524391880803e-06,
      "loss": 0.6836,
      "step": 3606
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6688082260083551,
      "learning_rate": 6.288249581823525e-06,
      "loss": 0.6972,
      "step": 3607
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.734797098451643,
      "learning_rate": 6.282975969254281e-06,
      "loss": 0.6693,
      "step": 3608
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6798371728505118,
      "learning_rate": 6.277703555874449e-06,
      "loss": 0.7074,
      "step": 3609
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6762035108406397,
      "learning_rate": 6.2724323433850235e-06,
      "loss": 0.6926,
      "step": 3610
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6304093435260272,
      "learning_rate": 6.2671623334866005e-06,
      "loss": 0.6794,
      "step": 3611
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6521620958868556,
      "learning_rate": 6.261893527879402e-06,
      "loss": 0.7137,
      "step": 3612
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7234441541778581,
      "learning_rate": 6.256625928263249e-06,
      "loss": 0.7742,
      "step": 3613
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.608748060829844,
      "learning_rate": 6.251359536337581e-06,
      "loss": 0.6939,
      "step": 3614
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6575050191118014,
      "learning_rate": 6.246094353801451e-06,
      "loss": 0.6931,
      "step": 3615
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.607188750737669,
      "learning_rate": 6.2408303823535066e-06,
      "loss": 0.6468,
      "step": 3616
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6238861722331178,
      "learning_rate": 6.23556762369202e-06,
      "loss": 0.7116,
      "step": 3617
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6345880195038742,
      "learning_rate": 6.230306079514867e-06,
      "loss": 0.6756,
      "step": 3618
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6616477488445691,
      "learning_rate": 6.2250457515195295e-06,
      "loss": 0.6564,
      "step": 3619
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6828742644606607,
      "learning_rate": 6.219786641403101e-06,
      "loss": 0.6865,
      "step": 3620
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8754429180847298,
      "learning_rate": 6.214528750862279e-06,
      "loss": 0.6886,
      "step": 3621
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6751132846652845,
      "learning_rate": 6.209272081593369e-06,
      "loss": 0.6752,
      "step": 3622
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6129856247988313,
      "learning_rate": 6.204016635292281e-06,
      "loss": 0.6899,
      "step": 3623
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6169981480877094,
      "learning_rate": 6.198762413654537e-06,
      "loss": 0.6862,
      "step": 3624
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7189990424626926,
      "learning_rate": 6.193509418375251e-06,
      "loss": 0.7459,
      "step": 3625
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6600642934328962,
      "learning_rate": 6.18825765114915e-06,
      "loss": 0.7281,
      "step": 3626
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6903109056828168,
      "learning_rate": 6.183007113670566e-06,
      "loss": 0.7075,
      "step": 3627
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.66137599014953,
      "learning_rate": 6.17775780763343e-06,
      "loss": 0.693,
      "step": 3628
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7316790811731002,
      "learning_rate": 6.172509734731278e-06,
      "loss": 0.7718,
      "step": 3629
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6663119815930032,
      "learning_rate": 6.167262896657249e-06,
      "loss": 0.6773,
      "step": 3630
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7143908614273448,
      "learning_rate": 6.162017295104081e-06,
      "loss": 0.7297,
      "step": 3631
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5936342350699189,
      "learning_rate": 6.156772931764117e-06,
      "loss": 0.6934,
      "step": 3632
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7671129026013588,
      "learning_rate": 6.151529808329294e-06,
      "loss": 0.7427,
      "step": 3633
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6483775960059862,
      "learning_rate": 6.146287926491155e-06,
      "loss": 0.7158,
      "step": 3634
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1022217526299123,
      "learning_rate": 6.141047287940838e-06,
      "loss": 0.7438,
      "step": 3635
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7082099594531301,
      "learning_rate": 6.135807894369086e-06,
      "loss": 0.7072,
      "step": 3636
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6866927448096244,
      "learning_rate": 6.130569747466233e-06,
      "loss": 0.7231,
      "step": 3637
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.629497920971104,
      "learning_rate": 6.125332848922215e-06,
      "loss": 0.6758,
      "step": 3638
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6742917750098905,
      "learning_rate": 6.120097200426569e-06,
      "loss": 0.7233,
      "step": 3639
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6257111987547964,
      "learning_rate": 6.114862803668418e-06,
      "loss": 0.7057,
      "step": 3640
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6711491350432747,
      "learning_rate": 6.109629660336493e-06,
      "loss": 0.6788,
      "step": 3641
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7484272776848792,
      "learning_rate": 6.104397772119113e-06,
      "loss": 0.7504,
      "step": 3642
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7453203899629145,
      "learning_rate": 6.099167140704191e-06,
      "loss": 0.7276,
      "step": 3643
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6921357524059765,
      "learning_rate": 6.093937767779239e-06,
      "loss": 0.696,
      "step": 3644
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6887924676863367,
      "learning_rate": 6.088709655031366e-06,
      "loss": 0.6928,
      "step": 3645
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7138459063862419,
      "learning_rate": 6.0834828041472645e-06,
      "loss": 0.7109,
      "step": 3646
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7138066087524346,
      "learning_rate": 6.078257216813228e-06,
      "loss": 0.7359,
      "step": 3647
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7379793100523082,
      "learning_rate": 6.0730328947151425e-06,
      "loss": 0.7162,
      "step": 3648
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6567753122127806,
      "learning_rate": 6.067809839538479e-06,
      "loss": 0.6979,
      "step": 3649
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6508231687926482,
      "learning_rate": 6.062588052968308e-06,
      "loss": 0.713,
      "step": 3650
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6895285236265193,
      "learning_rate": 6.057367536689285e-06,
      "loss": 0.6432,
      "step": 3651
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6940450493461439,
      "learning_rate": 6.0521482923856536e-06,
      "loss": 0.6883,
      "step": 3652
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6995572525643989,
      "learning_rate": 6.046930321741254e-06,
      "loss": 0.6978,
      "step": 3653
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7060590390385089,
      "learning_rate": 6.041713626439514e-06,
      "loss": 0.7212,
      "step": 3654
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.737262502778645,
      "learning_rate": 6.036498208163443e-06,
      "loss": 0.733,
      "step": 3655
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6003787124169327,
      "learning_rate": 6.03128406859565e-06,
      "loss": 0.6998,
      "step": 3656
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7914754101282354,
      "learning_rate": 6.02607120941832e-06,
      "loss": 0.722,
      "step": 3657
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7110278251680815,
      "learning_rate": 6.020859632313231e-06,
      "loss": 0.7223,
      "step": 3658
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6654708169813119,
      "learning_rate": 6.01564933896175e-06,
      "loss": 0.6938,
      "step": 3659
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7529153925828365,
      "learning_rate": 6.010440331044826e-06,
      "loss": 0.7156,
      "step": 3660
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6933489300395358,
      "learning_rate": 6.005232610242986e-06,
      "loss": 0.7471,
      "step": 3661
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6589805060592346,
      "learning_rate": 6.000026178236355e-06,
      "loss": 0.7053,
      "step": 3662
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6594017977891309,
      "learning_rate": 5.9948210367046385e-06,
      "loss": 0.7076,
      "step": 3663
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6402535494537709,
      "learning_rate": 5.989617187327119e-06,
      "loss": 0.7089,
      "step": 3664
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6063886943967922,
      "learning_rate": 5.98441463178267e-06,
      "loss": 0.6812,
      "step": 3665
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7041337579115885,
      "learning_rate": 5.979213371749742e-06,
      "loss": 0.7446,
      "step": 3666
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6945316705119915,
      "learning_rate": 5.9740134089063695e-06,
      "loss": 0.7088,
      "step": 3667
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.632452059238897,
      "learning_rate": 5.968814744930174e-06,
      "loss": 0.6905,
      "step": 3668
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6734824894538969,
      "learning_rate": 5.963617381498349e-06,
      "loss": 0.7066,
      "step": 3669
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7398572956839622,
      "learning_rate": 5.958421320287669e-06,
      "loss": 0.7342,
      "step": 3670
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6113889925716475,
      "learning_rate": 5.9532265629744975e-06,
      "loss": 0.6778,
      "step": 3671
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.741130099081557,
      "learning_rate": 5.948033111234768e-06,
      "loss": 0.6766,
      "step": 3672
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6191410792872402,
      "learning_rate": 5.9428409667439955e-06,
      "loss": 0.6882,
      "step": 3673
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.741807500178219,
      "learning_rate": 5.937650131177279e-06,
      "loss": 0.685,
      "step": 3674
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0067471299888575,
      "learning_rate": 5.932460606209285e-06,
      "loss": 0.7071,
      "step": 3675
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5842270107364662,
      "learning_rate": 5.927272393514263e-06,
      "loss": 0.6432,
      "step": 3676
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7973439561306773,
      "learning_rate": 5.922085494766044e-06,
      "loss": 0.771,
      "step": 3677
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1000865227529635,
      "learning_rate": 5.916899911638021e-06,
      "loss": 0.6409,
      "step": 3678
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6775350796058979,
      "learning_rate": 5.9117156458031825e-06,
      "loss": 0.6935,
      "step": 3679
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7052719389841254,
      "learning_rate": 5.90653269893407e-06,
      "loss": 0.696,
      "step": 3680
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6709065349005693,
      "learning_rate": 5.901351072702812e-06,
      "loss": 0.6718,
      "step": 3681
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7315048440236811,
      "learning_rate": 5.896170768781112e-06,
      "loss": 0.6854,
      "step": 3682
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7733715810644142,
      "learning_rate": 5.890991788840246e-06,
      "loss": 0.7542,
      "step": 3683
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7497013683690918,
      "learning_rate": 5.8858141345510545e-06,
      "loss": 0.7324,
      "step": 3684
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.694658361204646,
      "learning_rate": 5.880637807583962e-06,
      "loss": 0.6842,
      "step": 3685
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.699531956824605,
      "learning_rate": 5.875462809608957e-06,
      "loss": 0.6772,
      "step": 3686
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.687821363204963,
      "learning_rate": 5.870289142295602e-06,
      "loss": 0.6963,
      "step": 3687
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6318667121921496,
      "learning_rate": 5.865116807313034e-06,
      "loss": 0.6971,
      "step": 3688
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6488914268508297,
      "learning_rate": 5.859945806329951e-06,
      "loss": 0.7057,
      "step": 3689
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.620716010092699,
      "learning_rate": 5.854776141014624e-06,
      "loss": 0.7096,
      "step": 3690
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5931234747257947,
      "learning_rate": 5.849607813034899e-06,
      "loss": 0.6742,
      "step": 3691
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6519308064787281,
      "learning_rate": 5.844440824058186e-06,
      "loss": 0.6885,
      "step": 3692
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7751570920064099,
      "learning_rate": 5.83927517575146e-06,
      "loss": 0.7016,
      "step": 3693
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.667787028386904,
      "learning_rate": 5.834110869781272e-06,
      "loss": 0.6757,
      "step": 3694
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6868300588547681,
      "learning_rate": 5.828947907813729e-06,
      "loss": 0.755,
      "step": 3695
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7380963955278015,
      "learning_rate": 5.823786291514515e-06,
      "loss": 0.7488,
      "step": 3696
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6831269241333745,
      "learning_rate": 5.818626022548874e-06,
      "loss": 0.6934,
      "step": 3697
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7182640039437924,
      "learning_rate": 5.813467102581613e-06,
      "loss": 0.6878,
      "step": 3698
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7164174999071558,
      "learning_rate": 5.808309533277114e-06,
      "loss": 0.7329,
      "step": 3699
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6364515481426959,
      "learning_rate": 5.803153316299308e-06,
      "loss": 0.7232,
      "step": 3700
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6789819550865234,
      "learning_rate": 5.797998453311702e-06,
      "loss": 0.7097,
      "step": 3701
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6818255908735336,
      "learning_rate": 5.7928449459773605e-06,
      "loss": 0.6861,
      "step": 3702
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.641927516860865,
      "learning_rate": 5.787692795958915e-06,
      "loss": 0.6904,
      "step": 3703
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8093988616646525,
      "learning_rate": 5.782542004918558e-06,
      "loss": 0.6979,
      "step": 3704
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6743058987883921,
      "learning_rate": 5.7773925745180355e-06,
      "loss": 0.697,
      "step": 3705
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6621115970684641,
      "learning_rate": 5.77224450641867e-06,
      "loss": 0.6704,
      "step": 3706
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6477503814253919,
      "learning_rate": 5.767097802281326e-06,
      "loss": 0.7133,
      "step": 3707
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6691966347072137,
      "learning_rate": 5.761952463766443e-06,
      "loss": 0.6902,
      "step": 3708
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7123369233876122,
      "learning_rate": 5.7568084925340115e-06,
      "loss": 0.7859,
      "step": 3709
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6584900210372273,
      "learning_rate": 5.751665890243587e-06,
      "loss": 0.7047,
      "step": 3710
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6455423863876283,
      "learning_rate": 5.7465246585542825e-06,
      "loss": 0.6844,
      "step": 3711
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7126555101231647,
      "learning_rate": 5.741384799124758e-06,
      "loss": 0.6949,
      "step": 3712
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7027563511864601,
      "learning_rate": 5.736246313613246e-06,
      "loss": 0.7445,
      "step": 3713
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6806326641535373,
      "learning_rate": 5.731109203677527e-06,
      "loss": 0.7011,
      "step": 3714
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6566685785137848,
      "learning_rate": 5.725973470974944e-06,
      "loss": 0.6925,
      "step": 3715
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6527520708156976,
      "learning_rate": 5.720839117162382e-06,
      "loss": 0.7025,
      "step": 3716
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7233925751710032,
      "learning_rate": 5.7157061438963e-06,
      "loss": 0.7407,
      "step": 3717
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7471932164376863,
      "learning_rate": 5.7105745528326935e-06,
      "loss": 0.7296,
      "step": 3718
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7302943089219985,
      "learning_rate": 5.705444345627127e-06,
      "loss": 0.6978,
      "step": 3719
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6648619023428303,
      "learning_rate": 5.700315523934709e-06,
      "loss": 0.6702,
      "step": 3720
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8410601111758593,
      "learning_rate": 5.695188089410107e-06,
      "loss": 0.8182,
      "step": 3721
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6421404082363404,
      "learning_rate": 5.690062043707538e-06,
      "loss": 0.6685,
      "step": 3722
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.661929218493477,
      "learning_rate": 5.684937388480769e-06,
      "loss": 0.7111,
      "step": 3723
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6543251042184385,
      "learning_rate": 5.679814125383125e-06,
      "loss": 0.6831,
      "step": 3724
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7755641723412785,
      "learning_rate": 5.674692256067469e-06,
      "loss": 0.7089,
      "step": 3725
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.705404338333374,
      "learning_rate": 5.66957178218623e-06,
      "loss": 0.6956,
      "step": 3726
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6742145969194477,
      "learning_rate": 5.664452705391375e-06,
      "loss": 0.7099,
      "step": 3727
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.728275550805505,
      "learning_rate": 5.659335027334433e-06,
      "loss": 0.7588,
      "step": 3728
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8022581680370245,
      "learning_rate": 5.654218749666463e-06,
      "loss": 0.7243,
      "step": 3729
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7523058046403082,
      "learning_rate": 5.649103874038089e-06,
      "loss": 0.7648,
      "step": 3730
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7203754545851695,
      "learning_rate": 5.643990402099473e-06,
      "loss": 0.7009,
      "step": 3731
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6791343482877812,
      "learning_rate": 5.638878335500331e-06,
      "loss": 0.6986,
      "step": 3732
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6060408872049132,
      "learning_rate": 5.633767675889925e-06,
      "loss": 0.6821,
      "step": 3733
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.67943236002702,
      "learning_rate": 5.628658424917057e-06,
      "loss": 0.649,
      "step": 3734
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6547715278804377,
      "learning_rate": 5.623550584230072e-06,
      "loss": 0.7202,
      "step": 3735
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.708164074948917,
      "learning_rate": 5.618444155476872e-06,
      "loss": 0.724,
      "step": 3736
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6837584649586619,
      "learning_rate": 5.613339140304894e-06,
      "loss": 0.6837,
      "step": 3737
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6741510700830021,
      "learning_rate": 5.608235540361126e-06,
      "loss": 0.6838,
      "step": 3738
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6222341375132858,
      "learning_rate": 5.603133357292096e-06,
      "loss": 0.6798,
      "step": 3739
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7274487188348453,
      "learning_rate": 5.598032592743875e-06,
      "loss": 0.6816,
      "step": 3740
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7301058573908898,
      "learning_rate": 5.59293324836207e-06,
      "loss": 0.6746,
      "step": 3741
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7210013846924617,
      "learning_rate": 5.58783532579184e-06,
      "loss": 0.7302,
      "step": 3742
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6737671424979346,
      "learning_rate": 5.582738826677885e-06,
      "loss": 0.7242,
      "step": 3743
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6965375237274973,
      "learning_rate": 5.577643752664432e-06,
      "loss": 0.7059,
      "step": 3744
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6602938551039658,
      "learning_rate": 5.572550105395265e-06,
      "loss": 0.7008,
      "step": 3745
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8345682417561775,
      "learning_rate": 5.567457886513702e-06,
      "loss": 0.6922,
      "step": 3746
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6938532024747854,
      "learning_rate": 5.562367097662593e-06,
      "loss": 0.6972,
      "step": 3747
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7532647460707428,
      "learning_rate": 5.5572777404843346e-06,
      "loss": 0.6974,
      "step": 3748
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.669690493356793,
      "learning_rate": 5.552189816620862e-06,
      "loss": 0.6444,
      "step": 3749
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6693500235865453,
      "learning_rate": 5.547103327713644e-06,
      "loss": 0.6643,
      "step": 3750
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7115888953258173,
      "learning_rate": 5.542018275403691e-06,
      "loss": 0.7182,
      "step": 3751
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7755743047879254,
      "learning_rate": 5.536934661331545e-06,
      "loss": 0.7678,
      "step": 3752
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7397469724041066,
      "learning_rate": 5.53185248713728e-06,
      "loss": 0.7204,
      "step": 3753
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6699288242515097,
      "learning_rate": 5.526771754460516e-06,
      "loss": 0.6893,
      "step": 3754
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5839284259075035,
      "learning_rate": 5.5216924649404046e-06,
      "loss": 0.6729,
      "step": 3755
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7370992763213352,
      "learning_rate": 5.516614620215628e-06,
      "loss": 0.7367,
      "step": 3756
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6681379174256832,
      "learning_rate": 5.5115382219244095e-06,
      "loss": 0.7054,
      "step": 3757
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6634008888504233,
      "learning_rate": 5.506463271704494e-06,
      "loss": 0.6853,
      "step": 3758
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7367793729251505,
      "learning_rate": 5.501389771193168e-06,
      "loss": 0.7087,
      "step": 3759
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6684249713566321,
      "learning_rate": 5.496317722027252e-06,
      "loss": 0.6904,
      "step": 3760
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6614854244692017,
      "learning_rate": 5.4912471258430966e-06,
      "loss": 0.6929,
      "step": 3761
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6813478616923928,
      "learning_rate": 5.486177984276573e-06,
      "loss": 0.6997,
      "step": 3762
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7603371061847558,
      "learning_rate": 5.4811102989630974e-06,
      "loss": 0.7584,
      "step": 3763
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6896333514007714,
      "learning_rate": 5.4760440715376136e-06,
      "loss": 0.6754,
      "step": 3764
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.640382256081795,
      "learning_rate": 5.470979303634586e-06,
      "loss": 0.6852,
      "step": 3765
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7092592558806176,
      "learning_rate": 5.465915996888016e-06,
      "loss": 0.7371,
      "step": 3766
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6402397668420263,
      "learning_rate": 5.460854152931433e-06,
      "loss": 0.7203,
      "step": 3767
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7428702539444467,
      "learning_rate": 5.4557937733978926e-06,
      "loss": 0.7146,
      "step": 3768
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.643979130413918,
      "learning_rate": 5.450734859919984e-06,
      "loss": 0.6187,
      "step": 3769
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7585789761973535,
      "learning_rate": 5.445677414129813e-06,
      "loss": 0.7635,
      "step": 3770
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7680762661699424,
      "learning_rate": 5.440621437659015e-06,
      "loss": 0.8015,
      "step": 3771
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.747766405761006,
      "learning_rate": 5.435566932138754e-06,
      "loss": 0.6796,
      "step": 3772
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7051394532522703,
      "learning_rate": 5.430513899199721e-06,
      "loss": 0.7146,
      "step": 3773
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6949089847313543,
      "learning_rate": 5.4254623404721296e-06,
      "loss": 0.7344,
      "step": 3774
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6948505282090368,
      "learning_rate": 5.420412257585721e-06,
      "loss": 0.6821,
      "step": 3775
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7564222522799866,
      "learning_rate": 5.415363652169749e-06,
      "loss": 0.7476,
      "step": 3776
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7295250739501412,
      "learning_rate": 5.410316525853006e-06,
      "loss": 0.7199,
      "step": 3777
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7058426625538013,
      "learning_rate": 5.4052708802637965e-06,
      "loss": 0.6944,
      "step": 3778
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6802501352914707,
      "learning_rate": 5.400226717029957e-06,
      "loss": 0.7412,
      "step": 3779
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.74595609461052,
      "learning_rate": 5.395184037778831e-06,
      "loss": 0.7002,
      "step": 3780
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6574353738669626,
      "learning_rate": 5.390142844137298e-06,
      "loss": 0.6845,
      "step": 3781
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7517762770301801,
      "learning_rate": 5.3851031377317485e-06,
      "loss": 0.7191,
      "step": 3782
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7167133352762778,
      "learning_rate": 5.380064920188097e-06,
      "loss": 0.7207,
      "step": 3783
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6665493790211523,
      "learning_rate": 5.375028193131779e-06,
      "loss": 0.6865,
      "step": 3784
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6636683951753829,
      "learning_rate": 5.369992958187748e-06,
      "loss": 0.7221,
      "step": 3785
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6083332953428028,
      "learning_rate": 5.364959216980478e-06,
      "loss": 0.7035,
      "step": 3786
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.704324506285884,
      "learning_rate": 5.359926971133953e-06,
      "loss": 0.6801,
      "step": 3787
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6224286457191655,
      "learning_rate": 5.354896222271686e-06,
      "loss": 0.7015,
      "step": 3788
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6551434760176825,
      "learning_rate": 5.349866972016695e-06,
      "loss": 0.6884,
      "step": 3789
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6984569202003542,
      "learning_rate": 5.344839221991526e-06,
      "loss": 0.695,
      "step": 3790
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6914434358495585,
      "learning_rate": 5.339812973818233e-06,
      "loss": 0.6857,
      "step": 3791
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7028335553597863,
      "learning_rate": 5.33478822911839e-06,
      "loss": 0.7051,
      "step": 3792
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7326798230567361,
      "learning_rate": 5.3297649895130895e-06,
      "loss": 0.7021,
      "step": 3793
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6770921019724756,
      "learning_rate": 5.324743256622924e-06,
      "loss": 0.709,
      "step": 3794
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8490460324915713,
      "learning_rate": 5.319723032068014e-06,
      "loss": 0.7038,
      "step": 3795
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7253733612956895,
      "learning_rate": 5.314704317467987e-06,
      "loss": 0.7348,
      "step": 3796
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7781015454414068,
      "learning_rate": 5.309687114441988e-06,
      "loss": 0.7078,
      "step": 3797
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7604188371261382,
      "learning_rate": 5.304671424608668e-06,
      "loss": 0.6927,
      "step": 3798
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6925294520494099,
      "learning_rate": 5.2996572495861964e-06,
      "loss": 0.7245,
      "step": 3799
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.668849326710018,
      "learning_rate": 5.294644590992244e-06,
      "loss": 0.6829,
      "step": 3800
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7854950769501681,
      "learning_rate": 5.289633450444005e-06,
      "loss": 0.7987,
      "step": 3801
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6452512361262347,
      "learning_rate": 5.2846238295581745e-06,
      "loss": 0.6825,
      "step": 3802
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6688480910420856,
      "learning_rate": 5.279615729950962e-06,
      "loss": 0.7087,
      "step": 3803
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6890494117562236,
      "learning_rate": 5.274609153238088e-06,
      "loss": 0.6891,
      "step": 3804
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6646685093384058,
      "learning_rate": 5.269604101034773e-06,
      "loss": 0.6974,
      "step": 3805
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6694631825568886,
      "learning_rate": 5.264600574955754e-06,
      "loss": 0.6627,
      "step": 3806
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6514231752706654,
      "learning_rate": 5.259598576615275e-06,
      "loss": 0.645,
      "step": 3807
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6694048366002475,
      "learning_rate": 5.254598107627078e-06,
      "loss": 0.6865,
      "step": 3808
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6708132674980001,
      "learning_rate": 5.249599169604423e-06,
      "loss": 0.6999,
      "step": 3809
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6042120229003117,
      "learning_rate": 5.2446017641600754e-06,
      "loss": 0.683,
      "step": 3810
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6164426586954925,
      "learning_rate": 5.239605892906294e-06,
      "loss": 0.672,
      "step": 3811
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6156381915418546,
      "learning_rate": 5.234611557454855e-06,
      "loss": 0.6647,
      "step": 3812
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7193681062487589,
      "learning_rate": 5.229618759417036e-06,
      "loss": 0.7196,
      "step": 3813
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6718725189307097,
      "learning_rate": 5.2246275004036164e-06,
      "loss": 0.7093,
      "step": 3814
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6948923857943885,
      "learning_rate": 5.219637782024884e-06,
      "loss": 0.7298,
      "step": 3815
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6687246630759397,
      "learning_rate": 5.214649605890625e-06,
      "loss": 0.6615,
      "step": 3816
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6104581628952634,
      "learning_rate": 5.209662973610121e-06,
      "loss": 0.697,
      "step": 3817
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6108915142527969,
      "learning_rate": 5.20467788679217e-06,
      "loss": 0.6869,
      "step": 3818
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7504762559683734,
      "learning_rate": 5.199694347045063e-06,
      "loss": 0.6941,
      "step": 3819
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7064520286747901,
      "learning_rate": 5.194712355976596e-06,
      "loss": 0.6919,
      "step": 3820
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6777608824840348,
      "learning_rate": 5.189731915194063e-06,
      "loss": 0.7041,
      "step": 3821
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7167435467179116,
      "learning_rate": 5.184753026304261e-06,
      "loss": 0.7299,
      "step": 3822
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6349898325831766,
      "learning_rate": 5.179775690913475e-06,
      "loss": 0.6881,
      "step": 3823
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6716625436872551,
      "learning_rate": 5.1747999106275005e-06,
      "loss": 0.6828,
      "step": 3824
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.69521079684104,
      "learning_rate": 5.169825687051634e-06,
      "loss": 0.7098,
      "step": 3825
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6590262866444403,
      "learning_rate": 5.1648530217906545e-06,
      "loss": 0.6963,
      "step": 3826
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8104638440228294,
      "learning_rate": 5.159881916448851e-06,
      "loss": 0.762,
      "step": 3827
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7298600878178718,
      "learning_rate": 5.154912372630011e-06,
      "loss": 0.6478,
      "step": 3828
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6637284403452532,
      "learning_rate": 5.149944391937404e-06,
      "loss": 0.7008,
      "step": 3829
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7882743389908855,
      "learning_rate": 5.1449779759738085e-06,
      "loss": 0.7662,
      "step": 3830
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6434169300159732,
      "learning_rate": 5.140013126341492e-06,
      "loss": 0.6519,
      "step": 3831
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6417461916658261,
      "learning_rate": 5.135049844642221e-06,
      "loss": 0.6664,
      "step": 3832
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7674431315440703,
      "learning_rate": 5.130088132477258e-06,
      "loss": 0.7345,
      "step": 3833
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7162543286470655,
      "learning_rate": 5.125127991447348e-06,
      "loss": 0.744,
      "step": 3834
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7620752051987778,
      "learning_rate": 5.120169423152732e-06,
      "loss": 0.7667,
      "step": 3835
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6082529818119427,
      "learning_rate": 5.115212429193155e-06,
      "loss": 0.688,
      "step": 3836
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7739604581172084,
      "learning_rate": 5.110257011167843e-06,
      "loss": 0.7441,
      "step": 3837
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.768242325020029,
      "learning_rate": 5.1053031706755184e-06,
      "loss": 0.7556,
      "step": 3838
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7356790349449117,
      "learning_rate": 5.1003509093143975e-06,
      "loss": 0.6876,
      "step": 3839
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6921361170478114,
      "learning_rate": 5.095400228682175e-06,
      "loss": 0.7232,
      "step": 3840
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6016952165733309,
      "learning_rate": 5.090451130376047e-06,
      "loss": 0.6578,
      "step": 3841
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5939114865889934,
      "learning_rate": 5.085503615992697e-06,
      "loss": 0.6917,
      "step": 3842
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8701947214391322,
      "learning_rate": 5.080557687128299e-06,
      "loss": 0.8256,
      "step": 3843
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7045726432348327,
      "learning_rate": 5.075613345378505e-06,
      "loss": 0.7261,
      "step": 3844
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.674543894829507,
      "learning_rate": 5.0706705923384715e-06,
      "loss": 0.7179,
      "step": 3845
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9706718617995038,
      "learning_rate": 5.0657294296028244e-06,
      "loss": 0.6952,
      "step": 3846
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.736243042262045,
      "learning_rate": 5.060789858765691e-06,
      "loss": 0.7167,
      "step": 3847
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7430431492984368,
      "learning_rate": 5.055851881420679e-06,
      "loss": 0.7288,
      "step": 3848
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6603799311233272,
      "learning_rate": 5.050915499160882e-06,
      "loss": 0.741,
      "step": 3849
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7843688113107631,
      "learning_rate": 5.045980713578882e-06,
      "loss": 0.7179,
      "step": 3850
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6963642129677979,
      "learning_rate": 5.041047526266747e-06,
      "loss": 0.6715,
      "step": 3851
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7180726886787616,
      "learning_rate": 5.03611593881602e-06,
      "loss": 0.7089,
      "step": 3852
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6695917716173048,
      "learning_rate": 5.031185952817732e-06,
      "loss": 0.7172,
      "step": 3853
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8447764460742302,
      "learning_rate": 5.0262575698624026e-06,
      "loss": 0.6782,
      "step": 3854
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7566683882549605,
      "learning_rate": 5.02133079154003e-06,
      "loss": 0.8001,
      "step": 3855
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6853996565554452,
      "learning_rate": 5.016405619440096e-06,
      "loss": 0.6763,
      "step": 3856
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7772617977867423,
      "learning_rate": 5.011482055151571e-06,
      "loss": 0.6975,
      "step": 3857
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6410543890495052,
      "learning_rate": 5.006560100262886e-06,
      "loss": 0.6699,
      "step": 3858
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6335982136050051,
      "learning_rate": 5.001639756361973e-06,
      "loss": 0.7087,
      "step": 3859
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7114060930853807,
      "learning_rate": 4.996721025036239e-06,
      "loss": 0.6888,
      "step": 3860
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6840949706894206,
      "learning_rate": 4.9918039078725685e-06,
      "loss": 0.7318,
      "step": 3861
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6261156380397004,
      "learning_rate": 4.986888406457323e-06,
      "loss": 0.6923,
      "step": 3862
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7257860484233973,
      "learning_rate": 4.981974522376352e-06,
      "loss": 0.7744,
      "step": 3863
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6733520041417648,
      "learning_rate": 4.977062257214968e-06,
      "loss": 0.669,
      "step": 3864
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6480139853948359,
      "learning_rate": 4.9721516125579735e-06,
      "loss": 0.6981,
      "step": 3865
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6939224944777761,
      "learning_rate": 4.967242589989648e-06,
      "loss": 0.7523,
      "step": 3866
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7119404032088039,
      "learning_rate": 4.962335191093741e-06,
      "loss": 0.6658,
      "step": 3867
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6801775365300111,
      "learning_rate": 4.9574294174534875e-06,
      "loss": 0.7009,
      "step": 3868
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7206293803736405,
      "learning_rate": 4.952525270651584e-06,
      "loss": 0.6927,
      "step": 3869
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7141212941146021,
      "learning_rate": 4.947622752270219e-06,
      "loss": 0.7342,
      "step": 3870
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.735018266942994,
      "learning_rate": 4.9427218638910375e-06,
      "loss": 0.7276,
      "step": 3871
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7310767155875324,
      "learning_rate": 4.937822607095174e-06,
      "loss": 0.691,
      "step": 3872
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7632903030581712,
      "learning_rate": 4.93292498346323e-06,
      "loss": 0.7335,
      "step": 3873
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6007912544467231,
      "learning_rate": 4.92802899457528e-06,
      "loss": 0.6634,
      "step": 3874
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6955819958547845,
      "learning_rate": 4.923134642010878e-06,
      "loss": 0.6734,
      "step": 3875
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.66878894379211,
      "learning_rate": 4.918241927349034e-06,
      "loss": 0.7068,
      "step": 3876
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6623088632117868,
      "learning_rate": 4.9133508521682464e-06,
      "loss": 0.7192,
      "step": 3877
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7105584456995008,
      "learning_rate": 4.908461418046475e-06,
      "loss": 0.6756,
      "step": 3878
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6753608938155389,
      "learning_rate": 4.903573626561156e-06,
      "loss": 0.7097,
      "step": 3879
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7451636552678931,
      "learning_rate": 4.898687479289195e-06,
      "loss": 0.6554,
      "step": 3880
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7753326965457921,
      "learning_rate": 4.89380297780696e-06,
      "loss": 0.7052,
      "step": 3881
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6435540677314671,
      "learning_rate": 4.888920123690291e-06,
      "loss": 0.663,
      "step": 3882
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6673020999174788,
      "learning_rate": 4.884038918514503e-06,
      "loss": 0.7143,
      "step": 3883
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7285281771550658,
      "learning_rate": 4.879159363854372e-06,
      "loss": 0.6953,
      "step": 3884
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.694173608957493,
      "learning_rate": 4.874281461284146e-06,
      "loss": 0.7099,
      "step": 3885
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6950531377946103,
      "learning_rate": 4.8694052123775406e-06,
      "loss": 0.7019,
      "step": 3886
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7526954187445473,
      "learning_rate": 4.864530618707728e-06,
      "loss": 0.7104,
      "step": 3887
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5805633859847171,
      "learning_rate": 4.8596576818473586e-06,
      "loss": 0.6676,
      "step": 3888
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6013091649711352,
      "learning_rate": 4.854786403368545e-06,
      "loss": 0.6399,
      "step": 3889
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.660962667044918,
      "learning_rate": 4.849916784842859e-06,
      "loss": 0.7191,
      "step": 3890
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6656048433533132,
      "learning_rate": 4.845048827841341e-06,
      "loss": 0.6844,
      "step": 3891
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8651831098950522,
      "learning_rate": 4.840182533934501e-06,
      "loss": 0.771,
      "step": 3892
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6932638465408888,
      "learning_rate": 4.835317904692299e-06,
      "loss": 0.7299,
      "step": 3893
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.632700899304217,
      "learning_rate": 4.83045494168417e-06,
      "loss": 0.6838,
      "step": 3894
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8210602352660376,
      "learning_rate": 4.825593646479007e-06,
      "loss": 0.7825,
      "step": 3895
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6227599242935085,
      "learning_rate": 4.820734020645166e-06,
      "loss": 0.6811,
      "step": 3896
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6938873828681384,
      "learning_rate": 4.815876065750464e-06,
      "loss": 0.6699,
      "step": 3897
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.694306498440378,
      "learning_rate": 4.81101978336218e-06,
      "loss": 0.7025,
      "step": 3898
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6868360962607555,
      "learning_rate": 4.806165175047043e-06,
      "loss": 0.6633,
      "step": 3899
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7162194389800982,
      "learning_rate": 4.8013122423712565e-06,
      "loss": 0.7634,
      "step": 3900
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6736335736670063,
      "learning_rate": 4.796460986900479e-06,
      "loss": 0.6791,
      "step": 3901
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7141194729654623,
      "learning_rate": 4.791611410199826e-06,
      "loss": 0.7263,
      "step": 3902
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7899384576745612,
      "learning_rate": 4.786763513833871e-06,
      "loss": 0.7034,
      "step": 3903
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6138629626129972,
      "learning_rate": 4.781917299366651e-06,
      "loss": 0.671,
      "step": 3904
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6725164664482314,
      "learning_rate": 4.7770727683616465e-06,
      "loss": 0.6772,
      "step": 3905
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6469233213991414,
      "learning_rate": 4.772229922381812e-06,
      "loss": 0.6898,
      "step": 3906
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7631567578271852,
      "learning_rate": 4.76738876298955e-06,
      "loss": 0.7556,
      "step": 3907
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6770158280140242,
      "learning_rate": 4.7625492917467145e-06,
      "loss": 0.7652,
      "step": 3908
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6608854088930747,
      "learning_rate": 4.757711510214621e-06,
      "loss": 0.6935,
      "step": 3909
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6833246379555615,
      "learning_rate": 4.752875419954046e-06,
      "loss": 0.683,
      "step": 3910
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6755786593392891,
      "learning_rate": 4.748041022525203e-06,
      "loss": 0.6916,
      "step": 3911
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.623776793187692,
      "learning_rate": 4.743208319487773e-06,
      "loss": 0.6627,
      "step": 3912
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6986150476505548,
      "learning_rate": 4.738377312400887e-06,
      "loss": 0.6772,
      "step": 3913
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8117048394174071,
      "learning_rate": 4.73354800282313e-06,
      "loss": 0.8325,
      "step": 3914
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6494509787405786,
      "learning_rate": 4.72872039231254e-06,
      "loss": 0.6669,
      "step": 3915
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7721548115882467,
      "learning_rate": 4.723894482426602e-06,
      "loss": 0.7827,
      "step": 3916
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6374713861613419,
      "learning_rate": 4.71907027472225e-06,
      "loss": 0.6628,
      "step": 3917
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7008788347200157,
      "learning_rate": 4.7142477707558795e-06,
      "loss": 0.6929,
      "step": 3918
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.719262725077465,
      "learning_rate": 4.7094269720833304e-06,
      "loss": 0.6753,
      "step": 3919
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.65161845217328,
      "learning_rate": 4.704607880259891e-06,
      "loss": 0.6716,
      "step": 3920
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.680782571418947,
      "learning_rate": 4.699790496840307e-06,
      "loss": 0.6746,
      "step": 3921
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.147455454416788,
      "learning_rate": 4.694974823378757e-06,
      "loss": 0.7293,
      "step": 3922
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8417507434733394,
      "learning_rate": 4.690160861428883e-06,
      "loss": 0.7457,
      "step": 3923
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6628827157377503,
      "learning_rate": 4.6853486125437685e-06,
      "loss": 0.7017,
      "step": 3924
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7084645457836741,
      "learning_rate": 4.680538078275949e-06,
      "loss": 0.679,
      "step": 3925
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7284298986960926,
      "learning_rate": 4.675729260177397e-06,
      "loss": 0.7281,
      "step": 3926
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7683212862065008,
      "learning_rate": 4.670922159799543e-06,
      "loss": 0.7449,
      "step": 3927
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8935893856233198,
      "learning_rate": 4.666116778693251e-06,
      "loss": 0.6877,
      "step": 3928
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6155208638645904,
      "learning_rate": 4.661313118408841e-06,
      "loss": 0.6721,
      "step": 3929
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7172702890530858,
      "learning_rate": 4.656511180496075e-06,
      "loss": 0.7187,
      "step": 3930
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6954912046735725,
      "learning_rate": 4.651710966504156e-06,
      "loss": 0.6793,
      "step": 3931
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6993206253618558,
      "learning_rate": 4.646912477981735e-06,
      "loss": 0.6981,
      "step": 3932
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.727153422435547,
      "learning_rate": 4.642115716476906e-06,
      "loss": 0.7414,
      "step": 3933
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6540648025083513,
      "learning_rate": 4.637320683537202e-06,
      "loss": 0.6683,
      "step": 3934
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7288664597113681,
      "learning_rate": 4.6325273807095975e-06,
      "loss": 0.7295,
      "step": 3935
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6285193828288874,
      "learning_rate": 4.627735809540514e-06,
      "loss": 0.6523,
      "step": 3936
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7034385122602635,
      "learning_rate": 4.622945971575814e-06,
      "loss": 0.6835,
      "step": 3937
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7341621299915216,
      "learning_rate": 4.618157868360796e-06,
      "loss": 0.7288,
      "step": 3938
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6833564038385779,
      "learning_rate": 4.613371501440209e-06,
      "loss": 0.6888,
      "step": 3939
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6708047249291859,
      "learning_rate": 4.608586872358226e-06,
      "loss": 0.6612,
      "step": 3940
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7939758481579874,
      "learning_rate": 4.603803982658472e-06,
      "loss": 0.7372,
      "step": 3941
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7795435889320006,
      "learning_rate": 4.599022833884005e-06,
      "loss": 0.642,
      "step": 3942
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6570728215770382,
      "learning_rate": 4.594243427577326e-06,
      "loss": 0.6934,
      "step": 3943
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6765631285196115,
      "learning_rate": 4.589465765280373e-06,
      "loss": 0.6812,
      "step": 3944
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6439742675806613,
      "learning_rate": 4.584689848534516e-06,
      "loss": 0.6864,
      "step": 3945
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8160413864984365,
      "learning_rate": 4.579915678880562e-06,
      "loss": 0.7106,
      "step": 3946
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8267365683303203,
      "learning_rate": 4.575143257858762e-06,
      "loss": 0.7113,
      "step": 3947
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7981274782776646,
      "learning_rate": 4.570372587008796e-06,
      "loss": 0.729,
      "step": 3948
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6846457621170189,
      "learning_rate": 4.565603667869785e-06,
      "loss": 0.7016,
      "step": 3949
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6500076081638055,
      "learning_rate": 4.560836501980283e-06,
      "loss": 0.6918,
      "step": 3950
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6406145007983253,
      "learning_rate": 4.55607109087827e-06,
      "loss": 0.6962,
      "step": 3951
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.676570725408778,
      "learning_rate": 4.551307436101172e-06,
      "loss": 0.7196,
      "step": 3952
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6721652992277668,
      "learning_rate": 4.546545539185846e-06,
      "loss": 0.6955,
      "step": 3953
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6427274533907341,
      "learning_rate": 4.541785401668572e-06,
      "loss": 0.7036,
      "step": 3954
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6508394720461732,
      "learning_rate": 4.537027025085073e-06,
      "loss": 0.7088,
      "step": 3955
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6624445009326414,
      "learning_rate": 4.5322704109705e-06,
      "loss": 0.6881,
      "step": 3956
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6277120069422828,
      "learning_rate": 4.52751556085944e-06,
      "loss": 0.6777,
      "step": 3957
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6053704513123044,
      "learning_rate": 4.5227624762859e-06,
      "loss": 0.668,
      "step": 3958
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7063263470959069,
      "learning_rate": 4.518011158783327e-06,
      "loss": 0.7369,
      "step": 3959
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7051217649171778,
      "learning_rate": 4.513261609884597e-06,
      "loss": 0.7372,
      "step": 3960
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7717968771168001,
      "learning_rate": 4.50851383112201e-06,
      "loss": 0.728,
      "step": 3961
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7669938294717987,
      "learning_rate": 4.503767824027305e-06,
      "loss": 0.7536,
      "step": 3962
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7086901376379484,
      "learning_rate": 4.499023590131637e-06,
      "loss": 0.7152,
      "step": 3963
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.65523723765447,
      "learning_rate": 4.494281130965593e-06,
      "loss": 0.717,
      "step": 3964
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7008072711442297,
      "learning_rate": 4.489540448059191e-06,
      "loss": 0.6844,
      "step": 3965
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6648701100166502,
      "learning_rate": 4.484801542941876e-06,
      "loss": 0.6914,
      "step": 3966
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7951954619503134,
      "learning_rate": 4.4800644171425155e-06,
      "loss": 0.6731,
      "step": 3967
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6616063376838991,
      "learning_rate": 4.475329072189409e-06,
      "loss": 0.6858,
      "step": 3968
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6364975615008875,
      "learning_rate": 4.4705955096102726e-06,
      "loss": 0.6959,
      "step": 3969
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7124936810528,
      "learning_rate": 4.465863730932253e-06,
      "loss": 0.7231,
      "step": 3970
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8172592761284707,
      "learning_rate": 4.461133737681926e-06,
      "loss": 0.756,
      "step": 3971
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6301959463478298,
      "learning_rate": 4.456405531385277e-06,
      "loss": 0.6749,
      "step": 3972
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7042121214641872,
      "learning_rate": 4.45167911356773e-06,
      "loss": 0.7091,
      "step": 3973
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7574947576644057,
      "learning_rate": 4.446954485754127e-06,
      "loss": 0.7054,
      "step": 3974
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6844240677967618,
      "learning_rate": 4.4422316494687255e-06,
      "loss": 0.6607,
      "step": 3975
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6949244544466432,
      "learning_rate": 4.437510606235214e-06,
      "loss": 0.67,
      "step": 3976
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7756139661151491,
      "learning_rate": 4.4327913575767e-06,
      "loss": 0.7254,
      "step": 3977
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7368394736490559,
      "learning_rate": 4.428073905015712e-06,
      "loss": 0.6958,
      "step": 3978
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6768653594930659,
      "learning_rate": 4.4233582500742e-06,
      "loss": 0.7119,
      "step": 3979
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.611418110420067,
      "learning_rate": 4.418644394273531e-06,
      "loss": 0.6983,
      "step": 3980
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7063357885124085,
      "learning_rate": 4.413932339134489e-06,
      "loss": 0.7068,
      "step": 3981
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6925585112783754,
      "learning_rate": 4.409222086177284e-06,
      "loss": 0.6779,
      "step": 3982
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6517344019147757,
      "learning_rate": 4.404513636921541e-06,
      "loss": 0.6788,
      "step": 3983
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7041831619852239,
      "learning_rate": 4.399806992886307e-06,
      "loss": 0.7104,
      "step": 3984
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7705168969758647,
      "learning_rate": 4.3951021555900406e-06,
      "loss": 0.6957,
      "step": 3985
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8306827639618176,
      "learning_rate": 4.3903991265506256e-06,
      "loss": 0.7205,
      "step": 3986
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7326720285006443,
      "learning_rate": 4.385697907285349e-06,
      "loss": 0.7452,
      "step": 3987
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6871504967409611,
      "learning_rate": 4.380998499310925e-06,
      "loss": 0.7305,
      "step": 3988
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6832408702499257,
      "learning_rate": 4.376300904143486e-06,
      "loss": 0.6983,
      "step": 3989
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.638314415648227,
      "learning_rate": 4.371605123298568e-06,
      "loss": 0.678,
      "step": 3990
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8603120334527046,
      "learning_rate": 4.3669111582911274e-06,
      "loss": 0.6881,
      "step": 3991
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6742626156558128,
      "learning_rate": 4.362219010635541e-06,
      "loss": 0.7396,
      "step": 3992
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7066612237369919,
      "learning_rate": 4.357528681845585e-06,
      "loss": 0.6919,
      "step": 3993
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7191056515381434,
      "learning_rate": 4.352840173434463e-06,
      "loss": 0.7119,
      "step": 3994
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7662799162847375,
      "learning_rate": 4.348153486914785e-06,
      "loss": 0.7053,
      "step": 3995
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6473729039926089,
      "learning_rate": 4.343468623798571e-06,
      "loss": 0.699,
      "step": 3996
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6476106181861463,
      "learning_rate": 4.338785585597263e-06,
      "loss": 0.696,
      "step": 3997
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7127786868044038,
      "learning_rate": 4.3341043738217e-06,
      "loss": 0.7053,
      "step": 3998
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7140247643548592,
      "learning_rate": 4.329424989982136e-06,
      "loss": 0.7266,
      "step": 3999
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6075025123223964,
      "learning_rate": 4.324747435588242e-06,
      "loss": 0.7015,
      "step": 4000
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7144206866893128,
      "learning_rate": 4.320071712149094e-06,
      "loss": 0.7399,
      "step": 4001
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7400632408219339,
      "learning_rate": 4.315397821173177e-06,
      "loss": 0.6905,
      "step": 4002
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.75526909205561,
      "learning_rate": 4.310725764168392e-06,
      "loss": 0.7652,
      "step": 4003
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6487284584645248,
      "learning_rate": 4.306055542642031e-06,
      "loss": 0.6946,
      "step": 4004
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7291489870260005,
      "learning_rate": 4.301387158100811e-06,
      "loss": 0.7087,
      "step": 4005
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6489591092659152,
      "learning_rate": 4.296720612050849e-06,
      "loss": 0.6739,
      "step": 4006
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.777682253569864,
      "learning_rate": 4.29205590599767e-06,
      "loss": 0.7309,
      "step": 4007
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7516945246136576,
      "learning_rate": 4.287393041446209e-06,
      "loss": 0.8011,
      "step": 4008
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7064891759073131,
      "learning_rate": 4.282732019900801e-06,
      "loss": 0.7338,
      "step": 4009
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6218176534615302,
      "learning_rate": 4.278072842865183e-06,
      "loss": 0.7146,
      "step": 4010
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6298627939371148,
      "learning_rate": 4.273415511842507e-06,
      "loss": 0.6763,
      "step": 4011
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7190160125948066,
      "learning_rate": 4.268760028335326e-06,
      "loss": 0.6912,
      "step": 4012
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7362086228336854,
      "learning_rate": 4.264106393845595e-06,
      "loss": 0.7139,
      "step": 4013
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.653600491102266,
      "learning_rate": 4.259454609874675e-06,
      "loss": 0.6817,
      "step": 4014
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6016503577473339,
      "learning_rate": 4.254804677923329e-06,
      "loss": 0.6484,
      "step": 4015
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6274787816373798,
      "learning_rate": 4.250156599491717e-06,
      "loss": 0.687,
      "step": 4016
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6595147859136136,
      "learning_rate": 4.245510376079413e-06,
      "loss": 0.6858,
      "step": 4017
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7256858917837818,
      "learning_rate": 4.240866009185378e-06,
      "loss": 0.7081,
      "step": 4018
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7012697047857332,
      "learning_rate": 4.236223500307986e-06,
      "loss": 0.6964,
      "step": 4019
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6461050279850772,
      "learning_rate": 4.2315828509450055e-06,
      "loss": 0.7108,
      "step": 4020
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6611221800624095,
      "learning_rate": 4.226944062593611e-06,
      "loss": 0.6876,
      "step": 4021
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7095670823119493,
      "learning_rate": 4.222307136750364e-06,
      "loss": 0.7021,
      "step": 4022
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6472496200030502,
      "learning_rate": 4.21767207491124e-06,
      "loss": 0.6564,
      "step": 4023
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7034928643630493,
      "learning_rate": 4.213038878571604e-06,
      "loss": 0.7611,
      "step": 4024
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6905506674503996,
      "learning_rate": 4.2084075492262226e-06,
      "loss": 0.6989,
      "step": 4025
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6778981059217278,
      "learning_rate": 4.203778088369261e-06,
      "loss": 0.7081,
      "step": 4026
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8717287303007151,
      "learning_rate": 4.19915049749428e-06,
      "loss": 0.7386,
      "step": 4027
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6633858203006394,
      "learning_rate": 4.19452477809423e-06,
      "loss": 0.6636,
      "step": 4028
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6086608947901369,
      "learning_rate": 4.189900931661469e-06,
      "loss": 0.6691,
      "step": 4029
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7436054577697347,
      "learning_rate": 4.185278959687747e-06,
      "loss": 0.6968,
      "step": 4030
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6161142239855353,
      "learning_rate": 4.180658863664205e-06,
      "loss": 0.7057,
      "step": 4031
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.732611476366949,
      "learning_rate": 4.176040645081392e-06,
      "loss": 0.7475,
      "step": 4032
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7151364623712079,
      "learning_rate": 4.171424305429229e-06,
      "loss": 0.7292,
      "step": 4033
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.685179806746212,
      "learning_rate": 4.166809846197049e-06,
      "loss": 0.7094,
      "step": 4034
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6627201639020645,
      "learning_rate": 4.162197268873577e-06,
      "loss": 0.7192,
      "step": 4035
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6024292377492336,
      "learning_rate": 4.157586574946918e-06,
      "loss": 0.6977,
      "step": 4036
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6799440318302968,
      "learning_rate": 4.1529777659045825e-06,
      "loss": 0.7258,
      "step": 4037
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6573972923197857,
      "learning_rate": 4.148370843233467e-06,
      "loss": 0.6848,
      "step": 4038
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7413596158868206,
      "learning_rate": 4.143765808419866e-06,
      "loss": 0.7126,
      "step": 4039
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6215890469459103,
      "learning_rate": 4.1391626629494515e-06,
      "loss": 0.6928,
      "step": 4040
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6380303003136703,
      "learning_rate": 4.134561408307299e-06,
      "loss": 0.6924,
      "step": 4041
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6057664561260124,
      "learning_rate": 4.129962045977868e-06,
      "loss": 0.7004,
      "step": 4042
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7067512037916545,
      "learning_rate": 4.125364577445008e-06,
      "loss": 0.686,
      "step": 4043
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6977729909286843,
      "learning_rate": 4.120769004191963e-06,
      "loss": 0.6907,
      "step": 4044
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6782617000313675,
      "learning_rate": 4.116175327701357e-06,
      "loss": 0.7002,
      "step": 4045
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7315905840456344,
      "learning_rate": 4.111583549455202e-06,
      "loss": 0.7353,
      "step": 4046
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6857781388280468,
      "learning_rate": 4.106993670934905e-06,
      "loss": 0.6781,
      "step": 4047
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7187255556302073,
      "learning_rate": 4.1024056936212556e-06,
      "loss": 0.7121,
      "step": 4048
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.769545840050583,
      "learning_rate": 4.097819618994432e-06,
      "loss": 0.7208,
      "step": 4049
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7004867816898608,
      "learning_rate": 4.093235448534e-06,
      "loss": 0.673,
      "step": 4050
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7051267313536966,
      "learning_rate": 4.088653183718903e-06,
      "loss": 0.7088,
      "step": 4051
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7081079787505027,
      "learning_rate": 4.084072826027477e-06,
      "loss": 0.7069,
      "step": 4052
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6240672562064153,
      "learning_rate": 4.079494376937445e-06,
      "loss": 0.6986,
      "step": 4053
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7025294194914378,
      "learning_rate": 4.074917837925903e-06,
      "loss": 0.702,
      "step": 4054
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6696439243854817,
      "learning_rate": 4.070343210469342e-06,
      "loss": 0.6642,
      "step": 4055
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6974677333112685,
      "learning_rate": 4.065770496043634e-06,
      "loss": 0.7312,
      "step": 4056
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6646005438956429,
      "learning_rate": 4.0611996961240276e-06,
      "loss": 0.6633,
      "step": 4057
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.821287734412913,
      "learning_rate": 4.056630812185159e-06,
      "loss": 0.7467,
      "step": 4058
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7440564930566199,
      "learning_rate": 4.0520638457010466e-06,
      "loss": 0.7446,
      "step": 4059
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6873337783315991,
      "learning_rate": 4.047498798145089e-06,
      "loss": 0.7125,
      "step": 4060
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7103067533983112,
      "learning_rate": 4.042935670990069e-06,
      "loss": 0.6912,
      "step": 4061
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6712456466714375,
      "learning_rate": 4.038374465708143e-06,
      "loss": 0.7083,
      "step": 4062
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.724889769471863,
      "learning_rate": 4.033815183770847e-06,
      "loss": 0.7398,
      "step": 4063
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6792990868455396,
      "learning_rate": 4.029257826649105e-06,
      "loss": 0.7024,
      "step": 4064
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6995132026212403,
      "learning_rate": 4.0247023958132136e-06,
      "loss": 0.6767,
      "step": 4065
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6339298374515113,
      "learning_rate": 4.020148892732851e-06,
      "loss": 0.6918,
      "step": 4066
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8917780257144801,
      "learning_rate": 4.015597318877073e-06,
      "loss": 0.7333,
      "step": 4067
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7047361941276172,
      "learning_rate": 4.0110476757143135e-06,
      "loss": 0.6604,
      "step": 4068
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6831823821697504,
      "learning_rate": 4.006499964712377e-06,
      "loss": 0.7381,
      "step": 4069
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7545247781484505,
      "learning_rate": 4.001954187338451e-06,
      "loss": 0.7748,
      "step": 4070
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7187603400865059,
      "learning_rate": 3.9974103450591015e-06,
      "loss": 0.6733,
      "step": 4071
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.752869408919982,
      "learning_rate": 3.992868439340268e-06,
      "loss": 0.7559,
      "step": 4072
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6956253888604574,
      "learning_rate": 3.988328471647257e-06,
      "loss": 0.7109,
      "step": 4073
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.622703697117216,
      "learning_rate": 3.9837904434447635e-06,
      "loss": 0.7349,
      "step": 4074
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6353703788172573,
      "learning_rate": 3.979254356196845e-06,
      "loss": 0.7245,
      "step": 4075
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6631488897410306,
      "learning_rate": 3.974720211366938e-06,
      "loss": 0.7121,
      "step": 4076
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7341204385083082,
      "learning_rate": 3.970188010417855e-06,
      "loss": 0.6912,
      "step": 4077
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6023243312376773,
      "learning_rate": 3.965657754811777e-06,
      "loss": 0.7012,
      "step": 4078
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7392345368247785,
      "learning_rate": 3.961129446010262e-06,
      "loss": 0.7005,
      "step": 4079
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7637944200535546,
      "learning_rate": 3.956603085474231e-06,
      "loss": 0.7534,
      "step": 4080
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7304559583961786,
      "learning_rate": 3.952078674663989e-06,
      "loss": 0.7179,
      "step": 4081
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6589969349506175,
      "learning_rate": 3.9475562150391975e-06,
      "loss": 0.7224,
      "step": 4082
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7366060778622724,
      "learning_rate": 3.943035708058901e-06,
      "loss": 0.7366,
      "step": 4083
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7016140382779684,
      "learning_rate": 3.9385171551815074e-06,
      "loss": 0.6922,
      "step": 4084
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6492315274910282,
      "learning_rate": 3.934000557864801e-06,
      "loss": 0.6701,
      "step": 4085
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7677861673730672,
      "learning_rate": 3.929485917565922e-06,
      "loss": 0.6975,
      "step": 4086
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6345939522879193,
      "learning_rate": 3.924973235741393e-06,
      "loss": 0.6877,
      "step": 4087
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.644454567294953,
      "learning_rate": 3.920462513847096e-06,
      "loss": 0.6713,
      "step": 4088
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6776451098319187,
      "learning_rate": 3.915953753338288e-06,
      "loss": 0.6739,
      "step": 4089
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.640049317987023,
      "learning_rate": 3.911446955669588e-06,
      "loss": 0.6649,
      "step": 4090
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6525595932219584,
      "learning_rate": 3.906942122294983e-06,
      "loss": 0.686,
      "step": 4091
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6279312381085187,
      "learning_rate": 3.9024392546678215e-06,
      "loss": 0.6924,
      "step": 4092
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7120537675247931,
      "learning_rate": 3.8979383542408254e-06,
      "loss": 0.6853,
      "step": 4093
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6582150533270535,
      "learning_rate": 3.893439422466079e-06,
      "loss": 0.7139,
      "step": 4094
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6413898789737901,
      "learning_rate": 3.888942460795032e-06,
      "loss": 0.6879,
      "step": 4095
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7312180485811128,
      "learning_rate": 3.884447470678499e-06,
      "loss": 0.7196,
      "step": 4096
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6727694257849652,
      "learning_rate": 3.879954453566657e-06,
      "loss": 0.719,
      "step": 4097
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7097732280272122,
      "learning_rate": 3.875463410909045e-06,
      "loss": 0.6982,
      "step": 4098
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7372679733846382,
      "learning_rate": 3.870974344154571e-06,
      "loss": 0.7461,
      "step": 4099
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6216035418003141,
      "learning_rate": 3.866487254751495e-06,
      "loss": 0.6906,
      "step": 4100
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6328448345740221,
      "learning_rate": 3.862002144147449e-06,
      "loss": 0.6648,
      "step": 4101
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6990975296627503,
      "learning_rate": 3.857519013789424e-06,
      "loss": 0.7073,
      "step": 4102
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6602859523941846,
      "learning_rate": 3.853037865123773e-06,
      "loss": 0.666,
      "step": 4103
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7387699039062786,
      "learning_rate": 3.848558699596202e-06,
      "loss": 0.6932,
      "step": 4104
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6706568224660141,
      "learning_rate": 3.844081518651788e-06,
      "loss": 0.7007,
      "step": 4105
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7038842346687189,
      "learning_rate": 3.83960632373496e-06,
      "loss": 0.7398,
      "step": 4106
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7028187251362598,
      "learning_rate": 3.835133116289512e-06,
      "loss": 0.6675,
      "step": 4107
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8196142073898787,
      "learning_rate": 3.830661897758595e-06,
      "loss": 0.7219,
      "step": 4108
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7040763754806472,
      "learning_rate": 3.826192669584715e-06,
      "loss": 0.7092,
      "step": 4109
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6767091019439144,
      "learning_rate": 3.821725433209735e-06,
      "loss": 0.7042,
      "step": 4110
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6891031924173102,
      "learning_rate": 3.817260190074882e-06,
      "loss": 0.7128,
      "step": 4111
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6481628826483844,
      "learning_rate": 3.8127969416207354e-06,
      "loss": 0.6863,
      "step": 4112
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7386027142216073,
      "learning_rate": 3.8083356892872337e-06,
      "loss": 0.7692,
      "step": 4113
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6313585405694262,
      "learning_rate": 3.8038764345136712e-06,
      "loss": 0.6608,
      "step": 4114
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6402379257936502,
      "learning_rate": 3.7994191787386914e-06,
      "loss": 0.6793,
      "step": 4115
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7597925382858056,
      "learning_rate": 3.794963923400301e-06,
      "loss": 0.7288,
      "step": 4116
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7287238768983444,
      "learning_rate": 3.790510669935861e-06,
      "loss": 0.7474,
      "step": 4117
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.738711728997057,
      "learning_rate": 3.7860594197820777e-06,
      "loss": 0.6955,
      "step": 4118
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7527409552193809,
      "learning_rate": 3.78161017437502e-06,
      "loss": 0.7545,
      "step": 4119
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7163892493523407,
      "learning_rate": 3.7771629351501103e-06,
      "loss": 0.6753,
      "step": 4120
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7262509227881031,
      "learning_rate": 3.772717703542115e-06,
      "loss": 0.7354,
      "step": 4121
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7072396673501435,
      "learning_rate": 3.76827448098516e-06,
      "loss": 0.7281,
      "step": 4122
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7298791653041709,
      "learning_rate": 3.763833268912723e-06,
      "loss": 0.6887,
      "step": 4123
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6623895534630554,
      "learning_rate": 3.7593940687576325e-06,
      "loss": 0.6675,
      "step": 4124
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9301847959920553,
      "learning_rate": 3.7549568819520644e-06,
      "loss": 0.6731,
      "step": 4125
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.72663164704963,
      "learning_rate": 3.7505217099275525e-06,
      "loss": 0.7518,
      "step": 4126
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7185711717908811,
      "learning_rate": 3.7460885541149727e-06,
      "loss": 0.6876,
      "step": 4127
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6891708778038718,
      "learning_rate": 3.7416574159445496e-06,
      "loss": 0.6415,
      "step": 4128
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7004585272540013,
      "learning_rate": 3.737228296845864e-06,
      "loss": 0.786,
      "step": 4129
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6943563819324008,
      "learning_rate": 3.732801198247842e-06,
      "loss": 0.675,
      "step": 4130
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6785201096757093,
      "learning_rate": 3.728376121578757e-06,
      "loss": 0.6957,
      "step": 4131
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.74538105751841,
      "learning_rate": 3.7239530682662363e-06,
      "loss": 0.6833,
      "step": 4132
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6915210235234839,
      "learning_rate": 3.71953203973724e-06,
      "loss": 0.6795,
      "step": 4133
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7067295751103678,
      "learning_rate": 3.71511303741809e-06,
      "loss": 0.6942,
      "step": 4134
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7186790492767638,
      "learning_rate": 3.710696062734447e-06,
      "loss": 0.7262,
      "step": 4135
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7453210576374013,
      "learning_rate": 3.706281117111322e-06,
      "loss": 0.6642,
      "step": 4136
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.829757203079186,
      "learning_rate": 3.7018682019730632e-06,
      "loss": 0.7231,
      "step": 4137
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.686724229174792,
      "learning_rate": 3.697457318743375e-06,
      "loss": 0.681,
      "step": 4138
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6540899809061967,
      "learning_rate": 3.6930484688452916e-06,
      "loss": 0.6927,
      "step": 4139
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7884837387593868,
      "learning_rate": 3.6886416537012073e-06,
      "loss": 0.727,
      "step": 4140
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6983306732365626,
      "learning_rate": 3.684236874732848e-06,
      "loss": 0.6818,
      "step": 4141
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6429417426337873,
      "learning_rate": 3.679834133361292e-06,
      "loss": 0.6972,
      "step": 4142
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.764259675543516,
      "learning_rate": 3.675433431006954e-06,
      "loss": 0.7452,
      "step": 4143
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6806894953890447,
      "learning_rate": 3.6710347690895885e-06,
      "loss": 0.6948,
      "step": 4144
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7498765296972095,
      "learning_rate": 3.666638149028302e-06,
      "loss": 0.7621,
      "step": 4145
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7272327049074007,
      "learning_rate": 3.6622435722415284e-06,
      "loss": 0.7396,
      "step": 4146
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6685235898208827,
      "learning_rate": 3.6578510401470526e-06,
      "loss": 0.6738,
      "step": 4147
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6852793518675613,
      "learning_rate": 3.653460554161998e-06,
      "loss": 0.7008,
      "step": 4148
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7987047350191235,
      "learning_rate": 3.6490721157028265e-06,
      "loss": 0.722,
      "step": 4149
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6752235978269928,
      "learning_rate": 3.6446857261853428e-06,
      "loss": 0.6433,
      "step": 4150
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8118613253220879,
      "learning_rate": 3.6403013870246827e-06,
      "loss": 0.7204,
      "step": 4151
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6635855937234578,
      "learning_rate": 3.6359190996353257e-06,
      "loss": 0.6673,
      "step": 4152
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.726303974277222,
      "learning_rate": 3.631538865431091e-06,
      "loss": 0.7543,
      "step": 4153
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.693839481147394,
      "learning_rate": 3.627160685825136e-06,
      "loss": 0.6787,
      "step": 4154
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6686195740702836,
      "learning_rate": 3.6227845622299474e-06,
      "loss": 0.7194,
      "step": 4155
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8551966256843198,
      "learning_rate": 3.6184104960573584e-06,
      "loss": 0.8132,
      "step": 4156
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6326882703289347,
      "learning_rate": 3.6140384887185286e-06,
      "loss": 0.7292,
      "step": 4157
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7118917430018665,
      "learning_rate": 3.609668541623962e-06,
      "loss": 0.6976,
      "step": 4158
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.741042417240332,
      "learning_rate": 3.6053006561834947e-06,
      "loss": 0.7057,
      "step": 4159
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8092978103840734,
      "learning_rate": 3.6009348338062968e-06,
      "loss": 0.7106,
      "step": 4160
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7666590802149164,
      "learning_rate": 3.5965710759008778e-06,
      "loss": 0.7432,
      "step": 4161
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7066257123276345,
      "learning_rate": 3.5922093838750684e-06,
      "loss": 0.7055,
      "step": 4162
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6690470815807529,
      "learning_rate": 3.587849759136051e-06,
      "loss": 0.6922,
      "step": 4163
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6754867205953481,
      "learning_rate": 3.583492203090322e-06,
      "loss": 0.6859,
      "step": 4164
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6765168000354858,
      "learning_rate": 3.5791367171437252e-06,
      "loss": 0.6649,
      "step": 4165
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7696327671043454,
      "learning_rate": 3.574783302701429e-06,
      "loss": 0.6903,
      "step": 4166
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7301368987337316,
      "learning_rate": 3.5704319611679407e-06,
      "loss": 0.7026,
      "step": 4167
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.679890945168546,
      "learning_rate": 3.5660826939470873e-06,
      "loss": 0.6593,
      "step": 4168
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6949728467605827,
      "learning_rate": 3.5617355024420343e-06,
      "loss": 0.7032,
      "step": 4169
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6661672877355814,
      "learning_rate": 3.5573903880552785e-06,
      "loss": 0.704,
      "step": 4170
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7212962401994266,
      "learning_rate": 3.553047352188643e-06,
      "loss": 0.7195,
      "step": 4171
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7315957946603637,
      "learning_rate": 3.5487063962432856e-06,
      "loss": 0.7011,
      "step": 4172
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7053867324031871,
      "learning_rate": 3.544367521619686e-06,
      "loss": 0.7214,
      "step": 4173
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6259413072001925,
      "learning_rate": 3.5400307297176516e-06,
      "loss": 0.6769,
      "step": 4174
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6233551483021637,
      "learning_rate": 3.5356960219363256e-06,
      "loss": 0.715,
      "step": 4175
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6876072715285857,
      "learning_rate": 3.5313633996741748e-06,
      "loss": 0.7013,
      "step": 4176
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7557484773556263,
      "learning_rate": 3.527032864328993e-06,
      "loss": 0.701,
      "step": 4177
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7195991837566319,
      "learning_rate": 3.522704417297902e-06,
      "loss": 0.719,
      "step": 4178
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7074659581026258,
      "learning_rate": 3.518378059977352e-06,
      "loss": 0.6825,
      "step": 4179
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7308340532160994,
      "learning_rate": 3.51405379376311e-06,
      "loss": 0.7102,
      "step": 4180
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8317088282358945,
      "learning_rate": 3.5097316200502807e-06,
      "loss": 0.7137,
      "step": 4181
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6495416029552771,
      "learning_rate": 3.50541154023328e-06,
      "loss": 0.6631,
      "step": 4182
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6808734204203105,
      "learning_rate": 3.501093555705859e-06,
      "loss": 0.7168,
      "step": 4183
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6875893914630794,
      "learning_rate": 3.496777667861091e-06,
      "loss": 0.6852,
      "step": 4184
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7044880038897607,
      "learning_rate": 3.4924638780913733e-06,
      "loss": 0.7403,
      "step": 4185
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6769090467780672,
      "learning_rate": 3.4881521877884185e-06,
      "loss": 0.6726,
      "step": 4186
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6795405198728878,
      "learning_rate": 3.48384259834327e-06,
      "loss": 0.6809,
      "step": 4187
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6921185948242717,
      "learning_rate": 3.4795351111462926e-06,
      "loss": 0.7244,
      "step": 4188
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6432644311754939,
      "learning_rate": 3.4752297275871706e-06,
      "loss": 0.6817,
      "step": 4189
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7014521443446441,
      "learning_rate": 3.4709264490549134e-06,
      "loss": 0.709,
      "step": 4190
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6472809782017682,
      "learning_rate": 3.466625276937845e-06,
      "loss": 0.6702,
      "step": 4191
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.691132335723365,
      "learning_rate": 3.462326212623611e-06,
      "loss": 0.6865,
      "step": 4192
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6577813293677944,
      "learning_rate": 3.4580292574991827e-06,
      "loss": 0.6632,
      "step": 4193
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7150635908464325,
      "learning_rate": 3.4537344129508453e-06,
      "loss": 0.7029,
      "step": 4194
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6125468269608902,
      "learning_rate": 3.449441680364207e-06,
      "loss": 0.676,
      "step": 4195
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6714971713342085,
      "learning_rate": 3.4451510611241957e-06,
      "loss": 0.682,
      "step": 4196
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6722023767329092,
      "learning_rate": 3.440862556615047e-06,
      "loss": 0.7325,
      "step": 4197
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.738182869577407,
      "learning_rate": 3.4365761682203258e-06,
      "loss": 0.6482,
      "step": 4198
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7691822657472277,
      "learning_rate": 3.4322918973229104e-06,
      "loss": 0.7478,
      "step": 4199
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7180203679845676,
      "learning_rate": 3.428009745304999e-06,
      "loss": 0.7286,
      "step": 4200
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7138241938313679,
      "learning_rate": 3.4237297135480964e-06,
      "loss": 0.7129,
      "step": 4201
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7428190993468228,
      "learning_rate": 3.419451803433037e-06,
      "loss": 0.7156,
      "step": 4202
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6355489524268946,
      "learning_rate": 3.4151760163399573e-06,
      "loss": 0.6685,
      "step": 4203
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6584699880549385,
      "learning_rate": 3.410902353648319e-06,
      "loss": 0.7282,
      "step": 4204
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6043057883656514,
      "learning_rate": 3.4066308167368934e-06,
      "loss": 0.7088,
      "step": 4205
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.719294658101093,
      "learning_rate": 3.4023614069837676e-06,
      "loss": 0.6841,
      "step": 4206
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8130232383194077,
      "learning_rate": 3.3980941257663424e-06,
      "loss": 0.7401,
      "step": 4207
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6759959978691861,
      "learning_rate": 3.393828974461335e-06,
      "loss": 0.6852,
      "step": 4208
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6378960331611802,
      "learning_rate": 3.3895659544447703e-06,
      "loss": 0.6603,
      "step": 4209
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9094181831246927,
      "learning_rate": 3.385305067091981e-06,
      "loss": 0.701,
      "step": 4210
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.709981636479359,
      "learning_rate": 3.3810463137776238e-06,
      "loss": 0.6625,
      "step": 4211
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6817705182048864,
      "learning_rate": 3.3767896958756605e-06,
      "loss": 0.7011,
      "step": 4212
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6084991249387433,
      "learning_rate": 3.372535214759365e-06,
      "loss": 0.6414,
      "step": 4213
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6684829864860282,
      "learning_rate": 3.368282871801325e-06,
      "loss": 0.7141,
      "step": 4214
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7375501872262327,
      "learning_rate": 3.364032668373427e-06,
      "loss": 0.6922,
      "step": 4215
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7612098623112753,
      "learning_rate": 3.359784605846881e-06,
      "loss": 0.7443,
      "step": 4216
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6900040732448244,
      "learning_rate": 3.355538685592198e-06,
      "loss": 0.6948,
      "step": 4217
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7185649593248153,
      "learning_rate": 3.3512949089792046e-06,
      "loss": 0.6966,
      "step": 4218
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.652133903980499,
      "learning_rate": 3.3470532773770257e-06,
      "loss": 0.6927,
      "step": 4219
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7236227605370223,
      "learning_rate": 3.342813792154106e-06,
      "loss": 0.7227,
      "step": 4220
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6602586558221443,
      "learning_rate": 3.338576454678184e-06,
      "loss": 0.6939,
      "step": 4221
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6685813621702712,
      "learning_rate": 3.334341266316319e-06,
      "loss": 0.6785,
      "step": 4222
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6824094533316672,
      "learning_rate": 3.3301082284348696e-06,
      "loss": 0.7031,
      "step": 4223
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6367952975831179,
      "learning_rate": 3.3258773423995005e-06,
      "loss": 0.6784,
      "step": 4224
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6711917483187669,
      "learning_rate": 3.3216486095751897e-06,
      "loss": 0.6948,
      "step": 4225
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6412845555268937,
      "learning_rate": 3.3174220313262074e-06,
      "loss": 0.6523,
      "step": 4226
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6629045980027822,
      "learning_rate": 3.313197609016141e-06,
      "loss": 0.7017,
      "step": 4227
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6620952448969619,
      "learning_rate": 3.308975344007871e-06,
      "loss": 0.7014,
      "step": 4228
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.683896407805782,
      "learning_rate": 3.3047552376635925e-06,
      "loss": 0.7212,
      "step": 4229
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6772412314318997,
      "learning_rate": 3.3005372913447986e-06,
      "loss": 0.7132,
      "step": 4230
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7391460500505719,
      "learning_rate": 3.2963215064122878e-06,
      "loss": 0.7504,
      "step": 4231
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6801246798004695,
      "learning_rate": 3.292107884226161e-06,
      "loss": 0.6521,
      "step": 4232
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6631711036122113,
      "learning_rate": 3.287896426145818e-06,
      "loss": 0.6985,
      "step": 4233
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7196931197959254,
      "learning_rate": 3.2836871335299625e-06,
      "loss": 0.7618,
      "step": 4234
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6515489293463687,
      "learning_rate": 3.279480007736602e-06,
      "loss": 0.7196,
      "step": 4235
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6638874634796682,
      "learning_rate": 3.2752750501230444e-06,
      "loss": 0.6576,
      "step": 4236
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6163533776448326,
      "learning_rate": 3.2710722620458925e-06,
      "loss": 0.7014,
      "step": 4237
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6825677699154447,
      "learning_rate": 3.2668716448610595e-06,
      "loss": 0.6778,
      "step": 4238
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6499977217470185,
      "learning_rate": 3.2626731999237436e-06,
      "loss": 0.689,
      "step": 4239
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6689385065863087,
      "learning_rate": 3.2584769285884554e-06,
      "loss": 0.6821,
      "step": 4240
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6403236928833032,
      "learning_rate": 3.254282832208998e-06,
      "loss": 0.6822,
      "step": 4241
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6835329052037585,
      "learning_rate": 3.2500909121384773e-06,
      "loss": 0.689,
      "step": 4242
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7384456543681011,
      "learning_rate": 3.245901169729294e-06,
      "loss": 0.7271,
      "step": 4243
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7175444332633186,
      "learning_rate": 3.241713606333142e-06,
      "loss": 0.6948,
      "step": 4244
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6043573371922646,
      "learning_rate": 3.237528223301022e-06,
      "loss": 0.6857,
      "step": 4245
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6170995636695833,
      "learning_rate": 3.2333450219832195e-06,
      "loss": 0.709,
      "step": 4246
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7079761904670429,
      "learning_rate": 3.2291640037293258e-06,
      "loss": 0.7323,
      "step": 4247
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.651671530801884,
      "learning_rate": 3.224985169888225e-06,
      "loss": 0.6543,
      "step": 4248
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6907117467171854,
      "learning_rate": 3.2208085218080974e-06,
      "loss": 0.7044,
      "step": 4249
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6846277970162123,
      "learning_rate": 3.2166340608364134e-06,
      "loss": 0.6626,
      "step": 4250
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6619474103855552,
      "learning_rate": 3.2124617883199405e-06,
      "loss": 0.7066,
      "step": 4251
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6711796935340077,
      "learning_rate": 3.2082917056047435e-06,
      "loss": 0.677,
      "step": 4252
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6931622347891565,
      "learning_rate": 3.2041238140361774e-06,
      "loss": 0.6873,
      "step": 4253
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7011440025221709,
      "learning_rate": 3.199958114958893e-06,
      "loss": 0.7099,
      "step": 4254
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6012658932181135,
      "learning_rate": 3.1957946097168302e-06,
      "loss": 0.635,
      "step": 4255
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6840279770342765,
      "learning_rate": 3.1916332996532185e-06,
      "loss": 0.7195,
      "step": 4256
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5947735202686021,
      "learning_rate": 3.1874741861105862e-06,
      "loss": 0.6708,
      "step": 4257
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7101407914155566,
      "learning_rate": 3.1833172704307504e-06,
      "loss": 0.7346,
      "step": 4258
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6899852164750276,
      "learning_rate": 3.1791625539548175e-06,
      "loss": 0.7372,
      "step": 4259
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5974338362590322,
      "learning_rate": 3.1750100380231874e-06,
      "loss": 0.6633,
      "step": 4260
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7154782752746012,
      "learning_rate": 3.1708597239755513e-06,
      "loss": 0.7035,
      "step": 4261
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6889100217207367,
      "learning_rate": 3.1667116131508792e-06,
      "loss": 0.7029,
      "step": 4262
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6841324023471677,
      "learning_rate": 3.1625657068874414e-06,
      "loss": 0.6637,
      "step": 4263
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7228334317828661,
      "learning_rate": 3.158422006522799e-06,
      "loss": 0.696,
      "step": 4264
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6621813566366248,
      "learning_rate": 3.154280513393787e-06,
      "loss": 0.6939,
      "step": 4265
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.719486835844998,
      "learning_rate": 3.150141228836543e-06,
      "loss": 0.7775,
      "step": 4266
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6613530417704682,
      "learning_rate": 3.1460041541864883e-06,
      "loss": 0.6892,
      "step": 4267
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6720584085834083,
      "learning_rate": 3.1418692907783235e-06,
      "loss": 0.6672,
      "step": 4268
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.748047226241028,
      "learning_rate": 3.137736639946043e-06,
      "loss": 0.7447,
      "step": 4269
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5997112117391916,
      "learning_rate": 3.1336062030229284e-06,
      "loss": 0.6906,
      "step": 4270
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7287705376116427,
      "learning_rate": 3.1294779813415456e-06,
      "loss": 0.6771,
      "step": 4271
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6748431824048075,
      "learning_rate": 3.1253519762337447e-06,
      "loss": 0.7086,
      "step": 4272
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6912742492077535,
      "learning_rate": 3.1212281890306604e-06,
      "loss": 0.6851,
      "step": 4273
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7219764151569331,
      "learning_rate": 3.1171066210627075e-06,
      "loss": 0.7197,
      "step": 4274
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6726851897770894,
      "learning_rate": 3.1129872736595947e-06,
      "loss": 0.7061,
      "step": 4275
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6948863092585174,
      "learning_rate": 3.1088701481503082e-06,
      "loss": 0.7143,
      "step": 4276
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6834453553667204,
      "learning_rate": 3.104755245863119e-06,
      "loss": 0.7005,
      "step": 4277
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6684051375014362,
      "learning_rate": 3.1006425681255826e-06,
      "loss": 0.7151,
      "step": 4278
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6148416345711465,
      "learning_rate": 3.096532116264529e-06,
      "loss": 0.6612,
      "step": 4279
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6843830212420293,
      "learning_rate": 3.092423891606079e-06,
      "loss": 0.6937,
      "step": 4280
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6924134947390549,
      "learning_rate": 3.088317895475631e-06,
      "loss": 0.7088,
      "step": 4281
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6707159256518196,
      "learning_rate": 3.0842141291978677e-06,
      "loss": 0.7296,
      "step": 4282
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6484470480390797,
      "learning_rate": 3.080112594096745e-06,
      "loss": 0.68,
      "step": 4283
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6067888618189627,
      "learning_rate": 3.076013291495508e-06,
      "loss": 0.6724,
      "step": 4284
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6612958607377855,
      "learning_rate": 3.07191622271667e-06,
      "loss": 0.7005,
      "step": 4285
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6887649750516756,
      "learning_rate": 3.0678213890820373e-06,
      "loss": 0.7145,
      "step": 4286
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6474578523127409,
      "learning_rate": 3.0637287919126856e-06,
      "loss": 0.7004,
      "step": 4287
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6999147231446484,
      "learning_rate": 3.059638432528974e-06,
      "loss": 0.7498,
      "step": 4288
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6719816663028715,
      "learning_rate": 3.055550312250536e-06,
      "loss": 0.6909,
      "step": 4289
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6855829956382038,
      "learning_rate": 3.0514644323962895e-06,
      "loss": 0.6356,
      "step": 4290
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7587808216488773,
      "learning_rate": 3.047380794284419e-06,
      "loss": 0.7537,
      "step": 4291
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6433946841169893,
      "learning_rate": 3.04329939923239e-06,
      "loss": 0.6823,
      "step": 4292
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6270676789241981,
      "learning_rate": 3.0392202485569477e-06,
      "loss": 0.6845,
      "step": 4293
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8813145994784798,
      "learning_rate": 3.035143343574112e-06,
      "loss": 0.6622,
      "step": 4294
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6558310565232773,
      "learning_rate": 3.0310686855991766e-06,
      "loss": 0.6958,
      "step": 4295
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6649812120953954,
      "learning_rate": 3.026996275946714e-06,
      "loss": 0.6896,
      "step": 4296
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.708075435844975,
      "learning_rate": 3.0229261159305634e-06,
      "loss": 0.6967,
      "step": 4297
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6574909602903621,
      "learning_rate": 3.018858206863845e-06,
      "loss": 0.6912,
      "step": 4298
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6918184580130557,
      "learning_rate": 3.0147925500589516e-06,
      "loss": 0.7022,
      "step": 4299
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.789585489410625,
      "learning_rate": 3.0107291468275523e-06,
      "loss": 0.7696,
      "step": 4300
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6476176428778349,
      "learning_rate": 3.006667998480579e-06,
      "loss": 0.7027,
      "step": 4301
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.64054403996509,
      "learning_rate": 3.002609106328249e-06,
      "loss": 0.6626,
      "step": 4302
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6894366760817895,
      "learning_rate": 2.998552471680041e-06,
      "loss": 0.7485,
      "step": 4303
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7381471617982184,
      "learning_rate": 2.994498095844709e-06,
      "loss": 0.7019,
      "step": 4304
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7387188805883755,
      "learning_rate": 2.990445980130283e-06,
      "loss": 0.6947,
      "step": 4305
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6033688937624295,
      "learning_rate": 2.986396125844058e-06,
      "loss": 0.6478,
      "step": 4306
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6757631626655024,
      "learning_rate": 2.982348534292605e-06,
      "loss": 0.691,
      "step": 4307
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6670211764171111,
      "learning_rate": 2.9783032067817554e-06,
      "loss": 0.7134,
      "step": 4308
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.715834194852012,
      "learning_rate": 2.974260144616622e-06,
      "loss": 0.7252,
      "step": 4309
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7330466872390747,
      "learning_rate": 2.9702193491015742e-06,
      "loss": 0.6904,
      "step": 4310
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6072991812760155,
      "learning_rate": 2.966180821540261e-06,
      "loss": 0.656,
      "step": 4311
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.737926732222092,
      "learning_rate": 2.962144563235595e-06,
      "loss": 0.7177,
      "step": 4312
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6674346894872141,
      "learning_rate": 2.958110575489761e-06,
      "loss": 0.6663,
      "step": 4313
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6606273305948861,
      "learning_rate": 2.9540788596042003e-06,
      "loss": 0.673,
      "step": 4314
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7265547461758818,
      "learning_rate": 2.9500494168796325e-06,
      "loss": 0.7347,
      "step": 4315
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7266140451102369,
      "learning_rate": 2.9460222486160382e-06,
      "loss": 0.7084,
      "step": 4316
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6899114465907923,
      "learning_rate": 2.941997356112668e-06,
      "loss": 0.7361,
      "step": 4317
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.672287951270178,
      "learning_rate": 2.9379747406680383e-06,
      "loss": 0.6621,
      "step": 4318
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7292874248370427,
      "learning_rate": 2.9339544035799216e-06,
      "loss": 0.6796,
      "step": 4319
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7282930032816345,
      "learning_rate": 2.92993634614537e-06,
      "loss": 0.7579,
      "step": 4320
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6986021182590472,
      "learning_rate": 2.925920569660685e-06,
      "loss": 0.6718,
      "step": 4321
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6930828404793721,
      "learning_rate": 2.921907075421444e-06,
      "loss": 0.676,
      "step": 4322
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7249082767284523,
      "learning_rate": 2.917895864722483e-06,
      "loss": 0.7123,
      "step": 4323
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7303287779382369,
      "learning_rate": 2.913886938857903e-06,
      "loss": 0.6907,
      "step": 4324
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6010593367628515,
      "learning_rate": 2.9098802991210695e-06,
      "loss": 0.6622,
      "step": 4325
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6105043396639995,
      "learning_rate": 2.9058759468046027e-06,
      "loss": 0.686,
      "step": 4326
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6450370925688146,
      "learning_rate": 2.9018738832003923e-06,
      "loss": 0.7034,
      "step": 4327
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6734581410298869,
      "learning_rate": 2.89787410959959e-06,
      "loss": 0.7152,
      "step": 4328
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6579232616134495,
      "learning_rate": 2.893876627292601e-06,
      "loss": 0.6937,
      "step": 4329
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6733451035268354,
      "learning_rate": 2.8898814375691e-06,
      "loss": 0.661,
      "step": 4330
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6917105877193951,
      "learning_rate": 2.8858885417180193e-06,
      "loss": 0.708,
      "step": 4331
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7118781516720698,
      "learning_rate": 2.881897941027546e-06,
      "loss": 0.6785,
      "step": 4332
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6852285007395738,
      "learning_rate": 2.8779096367851346e-06,
      "loss": 0.6749,
      "step": 4333
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7022562760721243,
      "learning_rate": 2.873923630277493e-06,
      "loss": 0.7241,
      "step": 4334
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7131137434976613,
      "learning_rate": 2.869939922790592e-06,
      "loss": 0.7104,
      "step": 4335
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6948315455084594,
      "learning_rate": 2.86595851560966e-06,
      "loss": 0.7233,
      "step": 4336
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7607019607764423,
      "learning_rate": 2.861979410019181e-06,
      "loss": 0.7528,
      "step": 4337
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7148931791604591,
      "learning_rate": 2.8580026073028933e-06,
      "loss": 0.676,
      "step": 4338
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6629739622866826,
      "learning_rate": 2.8540281087437983e-06,
      "loss": 0.6669,
      "step": 4339
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7136569158717904,
      "learning_rate": 2.8500559156241526e-06,
      "loss": 0.7126,
      "step": 4340
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6458336025722355,
      "learning_rate": 2.8460860292254698e-06,
      "loss": 0.7061,
      "step": 4341
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6853250752426268,
      "learning_rate": 2.842118450828516e-06,
      "loss": 0.6748,
      "step": 4342
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7058815749867753,
      "learning_rate": 2.83815318171332e-06,
      "loss": 0.7375,
      "step": 4343
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6492147780425177,
      "learning_rate": 2.834190223159152e-06,
      "loss": 0.6961,
      "step": 4344
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6405922746431567,
      "learning_rate": 2.8302295764445485e-06,
      "loss": 0.7008,
      "step": 4345
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6896270887498933,
      "learning_rate": 2.8262712428473015e-06,
      "loss": 0.6876,
      "step": 4346
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7213201180843737,
      "learning_rate": 2.822315223644444e-06,
      "loss": 0.7157,
      "step": 4347
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6460560135425492,
      "learning_rate": 2.818361520112274e-06,
      "loss": 0.7235,
      "step": 4348
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6693257136616763,
      "learning_rate": 2.8144101335263407e-06,
      "loss": 0.6803,
      "step": 4349
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7336453893625734,
      "learning_rate": 2.8104610651614383e-06,
      "loss": 0.7465,
      "step": 4350
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7219143363332473,
      "learning_rate": 2.8065143162916233e-06,
      "loss": 0.742,
      "step": 4351
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6771082084017352,
      "learning_rate": 2.8025698881901963e-06,
      "loss": 0.6922,
      "step": 4352
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6174617309361377,
      "learning_rate": 2.7986277821297137e-06,
      "loss": 0.6954,
      "step": 4353
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6715319636424123,
      "learning_rate": 2.794687999381983e-06,
      "loss": 0.7107,
      "step": 4354
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6399644824637609,
      "learning_rate": 2.790750541218058e-06,
      "loss": 0.707,
      "step": 4355
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6544944124594748,
      "learning_rate": 2.7868154089082422e-06,
      "loss": 0.6777,
      "step": 4356
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.608157979838302,
      "learning_rate": 2.7828826037220936e-06,
      "loss": 0.667,
      "step": 4357
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7175636863577015,
      "learning_rate": 2.778952126928417e-06,
      "loss": 0.6718,
      "step": 4358
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6964629710264357,
      "learning_rate": 2.775023979795266e-06,
      "loss": 0.721,
      "step": 4359
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6718009172853071,
      "learning_rate": 2.771098163589946e-06,
      "loss": 0.685,
      "step": 4360
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7860066823141806,
      "learning_rate": 2.767174679579002e-06,
      "loss": 0.7328,
      "step": 4361
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6558481617236268,
      "learning_rate": 2.7632535290282335e-06,
      "loss": 0.6687,
      "step": 4362
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7758244942395631,
      "learning_rate": 2.759334713202686e-06,
      "loss": 0.7654,
      "step": 4363
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5898269362682842,
      "learning_rate": 2.7554182333666533e-06,
      "loss": 0.651,
      "step": 4364
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5979191283319093,
      "learning_rate": 2.7515040907836677e-06,
      "loss": 0.6659,
      "step": 4365
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7990558135098621,
      "learning_rate": 2.7475922867165208e-06,
      "loss": 0.7131,
      "step": 4366
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6576632165112233,
      "learning_rate": 2.743682822427234e-06,
      "loss": 0.687,
      "step": 4367
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6238787874410033,
      "learning_rate": 2.7397756991770863e-06,
      "loss": 0.6417,
      "step": 4368
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6321519388229497,
      "learning_rate": 2.7358709182265954e-06,
      "loss": 0.6951,
      "step": 4369
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8137579098540422,
      "learning_rate": 2.7319684808355276e-06,
      "loss": 0.7507,
      "step": 4370
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6445083631366049,
      "learning_rate": 2.728068388262888e-06,
      "loss": 0.7396,
      "step": 4371
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6710396670149641,
      "learning_rate": 2.7241706417669322e-06,
      "loss": 0.688,
      "step": 4372
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6678017583435523,
      "learning_rate": 2.7202752426051506e-06,
      "loss": 0.6706,
      "step": 4373
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7175965921332618,
      "learning_rate": 2.7163821920342783e-06,
      "loss": 0.7024,
      "step": 4374
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7490131218872977,
      "learning_rate": 2.712491491310296e-06,
      "loss": 0.6852,
      "step": 4375
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6496132373206126,
      "learning_rate": 2.708603141688426e-06,
      "loss": 0.6622,
      "step": 4376
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7379769959503508,
      "learning_rate": 2.704717144423131e-06,
      "loss": 0.7348,
      "step": 4377
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7298423277495225,
      "learning_rate": 2.7008335007681174e-06,
      "loss": 0.7235,
      "step": 4378
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7148703020042013,
      "learning_rate": 2.6969522119763238e-06,
      "loss": 0.7302,
      "step": 4379
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7353012012883668,
      "learning_rate": 2.693073279299936e-06,
      "loss": 0.7378,
      "step": 4380
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7213671072948346,
      "learning_rate": 2.689196703990381e-06,
      "loss": 0.6995,
      "step": 4381
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6586157604737909,
      "learning_rate": 2.6853224872983253e-06,
      "loss": 0.6982,
      "step": 4382
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6554502916777821,
      "learning_rate": 2.681450630473664e-06,
      "loss": 0.7223,
      "step": 4383
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5917511915638496,
      "learning_rate": 2.677581134765548e-06,
      "loss": 0.6488,
      "step": 4384
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7132177638442664,
      "learning_rate": 2.673714001422348e-06,
      "loss": 0.7301,
      "step": 4385
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6752637599051652,
      "learning_rate": 2.669849231691687e-06,
      "loss": 0.6898,
      "step": 4386
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7069024161068582,
      "learning_rate": 2.6659868268204203e-06,
      "loss": 0.7209,
      "step": 4387
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6038026094588275,
      "learning_rate": 2.6621267880546388e-06,
      "loss": 0.6706,
      "step": 4388
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7722344853293152,
      "learning_rate": 2.6582691166396744e-06,
      "loss": 0.702,
      "step": 4389
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6608106654045633,
      "learning_rate": 2.654413813820088e-06,
      "loss": 0.7047,
      "step": 4390
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6405066446491424,
      "learning_rate": 2.650560880839682e-06,
      "loss": 0.6704,
      "step": 4391
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6400782605256468,
      "learning_rate": 2.6467103189414966e-06,
      "loss": 0.7114,
      "step": 4392
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6801768893843201,
      "learning_rate": 2.642862129367798e-06,
      "loss": 0.6942,
      "step": 4393
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.716169199088323,
      "learning_rate": 2.639016313360093e-06,
      "loss": 0.7311,
      "step": 4394
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7366237856782029,
      "learning_rate": 2.6351728721591287e-06,
      "loss": 0.7401,
      "step": 4395
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.632173332062806,
      "learning_rate": 2.63133180700487e-06,
      "loss": 0.6834,
      "step": 4396
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6686128681847957,
      "learning_rate": 2.627493119136528e-06,
      "loss": 0.6848,
      "step": 4397
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7375096351292418,
      "learning_rate": 2.6236568097925453e-06,
      "loss": 0.7035,
      "step": 4398
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7944011624427639,
      "learning_rate": 2.619822880210594e-06,
      "loss": 0.7809,
      "step": 4399
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6239562316391867,
      "learning_rate": 2.6159913316275786e-06,
      "loss": 0.6563,
      "step": 4400
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6741763005204546,
      "learning_rate": 2.6121621652796425e-06,
      "loss": 0.738,
      "step": 4401
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7548518584586089,
      "learning_rate": 2.6083353824021483e-06,
      "loss": 0.6809,
      "step": 4402
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6991761436446731,
      "learning_rate": 2.6045109842296944e-06,
      "loss": 0.6983,
      "step": 4403
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7331658441553062,
      "learning_rate": 2.600688971996115e-06,
      "loss": 0.7643,
      "step": 4404
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6683362521746651,
      "learning_rate": 2.596869346934471e-06,
      "loss": 0.7054,
      "step": 4405
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7563992612207664,
      "learning_rate": 2.59305211027705e-06,
      "loss": 0.7743,
      "step": 4406
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6805459187427193,
      "learning_rate": 2.58923726325538e-06,
      "loss": 0.679,
      "step": 4407
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5945768578695007,
      "learning_rate": 2.5854248071002e-06,
      "loss": 0.6922,
      "step": 4408
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6964511060895735,
      "learning_rate": 2.5816147430414917e-06,
      "loss": 0.7519,
      "step": 4409
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6753282321515117,
      "learning_rate": 2.5778070723084656e-06,
      "loss": 0.6948,
      "step": 4410
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7394107817051443,
      "learning_rate": 2.574001796129548e-06,
      "loss": 0.7001,
      "step": 4411
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6905837365216658,
      "learning_rate": 2.5701989157324048e-06,
      "loss": 0.7525,
      "step": 4412
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6327734335801021,
      "learning_rate": 2.566398432343927e-06,
      "loss": 0.6802,
      "step": 4413
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7164451820696967,
      "learning_rate": 2.5626003471902228e-06,
      "loss": 0.7099,
      "step": 4414
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6516081399484215,
      "learning_rate": 2.5588046614966377e-06,
      "loss": 0.6683,
      "step": 4415
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6618774558230182,
      "learning_rate": 2.5550113764877392e-06,
      "loss": 0.6568,
      "step": 4416
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.656293047102398,
      "learning_rate": 2.5512204933873185e-06,
      "loss": 0.6903,
      "step": 4417
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6657005833008155,
      "learning_rate": 2.547432013418398e-06,
      "loss": 0.7488,
      "step": 4418
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6222784238799268,
      "learning_rate": 2.5436459378032165e-06,
      "loss": 0.6946,
      "step": 4419
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7302355601301218,
      "learning_rate": 2.5398622677632377e-06,
      "loss": 0.7616,
      "step": 4420
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6797555188854594,
      "learning_rate": 2.5360810045191574e-06,
      "loss": 0.7164,
      "step": 4421
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.676003291630378,
      "learning_rate": 2.532302149290887e-06,
      "loss": 0.7436,
      "step": 4422
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6968607012590929,
      "learning_rate": 2.5285257032975662e-06,
      "loss": 0.6594,
      "step": 4423
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6369475091147434,
      "learning_rate": 2.5247516677575533e-06,
      "loss": 0.638,
      "step": 4424
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6995114377684288,
      "learning_rate": 2.5209800438884357e-06,
      "loss": 0.7201,
      "step": 4425
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6813982158025857,
      "learning_rate": 2.517210832907011e-06,
      "loss": 0.7043,
      "step": 4426
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6545210552855596,
      "learning_rate": 2.5134440360293078e-06,
      "loss": 0.6913,
      "step": 4427
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7436630392256248,
      "learning_rate": 2.509679654470577e-06,
      "loss": 0.7748,
      "step": 4428
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6604767388264492,
      "learning_rate": 2.50591768944528e-06,
      "loss": 0.6914,
      "step": 4429
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.736097418569258,
      "learning_rate": 2.502158142167108e-06,
      "loss": 0.7214,
      "step": 4430
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6456827815048567,
      "learning_rate": 2.4984010138489733e-06,
      "loss": 0.6678,
      "step": 4431
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7510351682698043,
      "learning_rate": 2.4946463057029966e-06,
      "loss": 0.7583,
      "step": 4432
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7014130749724792,
      "learning_rate": 2.4908940189405296e-06,
      "loss": 0.6985,
      "step": 4433
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6816463288387188,
      "learning_rate": 2.4871441547721363e-06,
      "loss": 0.6558,
      "step": 4434
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7152300795226034,
      "learning_rate": 2.483396714407601e-06,
      "loss": 0.6908,
      "step": 4435
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6187124648432665,
      "learning_rate": 2.4796516990559317e-06,
      "loss": 0.6715,
      "step": 4436
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7020206228636318,
      "learning_rate": 2.4759091099253418e-06,
      "loss": 0.7142,
      "step": 4437
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6919830588662177,
      "learning_rate": 2.472168948223268e-06,
      "loss": 0.6793,
      "step": 4438
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.691306287417338,
      "learning_rate": 2.468431215156367e-06,
      "loss": 0.7176,
      "step": 4439
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6365732479946181,
      "learning_rate": 2.4646959119305093e-06,
      "loss": 0.694,
      "step": 4440
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7758283504488048,
      "learning_rate": 2.4609630397507813e-06,
      "loss": 0.811,
      "step": 4441
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6662516509655402,
      "learning_rate": 2.4572325998214887e-06,
      "loss": 0.7135,
      "step": 4442
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6454143704141313,
      "learning_rate": 2.453504593346142e-06,
      "loss": 0.6892,
      "step": 4443
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7079257358219276,
      "learning_rate": 2.4497790215274773e-06,
      "loss": 0.7163,
      "step": 4444
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6246048771929358,
      "learning_rate": 2.446055885567443e-06,
      "loss": 0.6854,
      "step": 4445
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6681769479083386,
      "learning_rate": 2.442335186667203e-06,
      "loss": 0.6504,
      "step": 4446
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.630488453316136,
      "learning_rate": 2.4386169260271254e-06,
      "loss": 0.673,
      "step": 4447
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6659534401529311,
      "learning_rate": 2.434901104846804e-06,
      "loss": 0.7162,
      "step": 4448
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7026836680956687,
      "learning_rate": 2.431187724325036e-06,
      "loss": 0.7038,
      "step": 4449
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6807466759116871,
      "learning_rate": 2.427476785659837e-06,
      "loss": 0.6831,
      "step": 4450
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7190996853458741,
      "learning_rate": 2.4237682900484337e-06,
      "loss": 0.6675,
      "step": 4451
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.772345749949045,
      "learning_rate": 2.4200622386872643e-06,
      "loss": 0.7641,
      "step": 4452
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7327205085633297,
      "learning_rate": 2.4163586327719766e-06,
      "loss": 0.6967,
      "step": 4453
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7258374321664719,
      "learning_rate": 2.412657473497436e-06,
      "loss": 0.7132,
      "step": 4454
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9191228893655496,
      "learning_rate": 2.408958762057707e-06,
      "loss": 0.8261,
      "step": 4455
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8618411798517549,
      "learning_rate": 2.405262499646077e-06,
      "loss": 0.7316,
      "step": 4456
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.700007448189415,
      "learning_rate": 2.4015686874550304e-06,
      "loss": 0.7425,
      "step": 4457
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6508492105828403,
      "learning_rate": 2.3978773266762724e-06,
      "loss": 0.6591,
      "step": 4458
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6651944056060757,
      "learning_rate": 2.3941884185007125e-06,
      "loss": 0.6977,
      "step": 4459
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.715955328494538,
      "learning_rate": 2.3905019641184713e-06,
      "loss": 0.7089,
      "step": 4460
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7175455341859202,
      "learning_rate": 2.3868179647188715e-06,
      "loss": 0.6778,
      "step": 4461
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7157821633491877,
      "learning_rate": 2.3831364214904494e-06,
      "loss": 0.6905,
      "step": 4462
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6698301155215018,
      "learning_rate": 2.379457335620948e-06,
      "loss": 0.7668,
      "step": 4463
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.645481899728501,
      "learning_rate": 2.3757807082973184e-06,
      "loss": 0.7229,
      "step": 4464
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7600592003518846,
      "learning_rate": 2.3721065407057188e-06,
      "loss": 0.6949,
      "step": 4465
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6937986217856367,
      "learning_rate": 2.36843483403151e-06,
      "loss": 0.7055,
      "step": 4466
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7037757873797891,
      "learning_rate": 2.364765589459258e-06,
      "loss": 0.7102,
      "step": 4467
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7298276412475131,
      "learning_rate": 2.3610988081727393e-06,
      "loss": 0.7493,
      "step": 4468
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6779937018184314,
      "learning_rate": 2.357434491354935e-06,
      "loss": 0.6937,
      "step": 4469
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6547966344007566,
      "learning_rate": 2.3537726401880312e-06,
      "loss": 0.7038,
      "step": 4470
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6540165452737318,
      "learning_rate": 2.350113255853418e-06,
      "loss": 0.7252,
      "step": 4471
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6692445813554894,
      "learning_rate": 2.346456339531685e-06,
      "loss": 0.6936,
      "step": 4472
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5953100005433263,
      "learning_rate": 2.342801892402632e-06,
      "loss": 0.6838,
      "step": 4473
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.678392532831892,
      "learning_rate": 2.3391499156452626e-06,
      "loss": 0.6973,
      "step": 4474
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6097450341114111,
      "learning_rate": 2.3355004104377732e-06,
      "loss": 0.6345,
      "step": 4475
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7428681706964594,
      "learning_rate": 2.331853377957577e-06,
      "loss": 0.7513,
      "step": 4476
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6596599435156287,
      "learning_rate": 2.328208819381281e-06,
      "loss": 0.6899,
      "step": 4477
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7495897767257591,
      "learning_rate": 2.3245667358846944e-06,
      "loss": 0.7075,
      "step": 4478
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6814968300630969,
      "learning_rate": 2.3209271286428292e-06,
      "loss": 0.6805,
      "step": 4479
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7042431098650255,
      "learning_rate": 2.3172899988298993e-06,
      "loss": 0.7032,
      "step": 4480
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7152834162882614,
      "learning_rate": 2.313655347619318e-06,
      "loss": 0.7089,
      "step": 4481
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6892350299778682,
      "learning_rate": 2.3100231761837012e-06,
      "loss": 0.727,
      "step": 4482
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.671451714062499,
      "learning_rate": 2.3063934856948644e-06,
      "loss": 0.6771,
      "step": 4483
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7102701839709636,
      "learning_rate": 2.302766277323819e-06,
      "loss": 0.7014,
      "step": 4484
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6331408639605367,
      "learning_rate": 2.2991415522407743e-06,
      "loss": 0.7048,
      "step": 4485
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6557516149501416,
      "learning_rate": 2.2955193116151477e-06,
      "loss": 0.6758,
      "step": 4486
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6335297075567131,
      "learning_rate": 2.2918995566155466e-06,
      "loss": 0.7086,
      "step": 4487
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7169639820981214,
      "learning_rate": 2.28828228840978e-06,
      "loss": 0.7281,
      "step": 4488
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7124041267050691,
      "learning_rate": 2.2846675081648582e-06,
      "loss": 0.6883,
      "step": 4489
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.634135075995977,
      "learning_rate": 2.281055217046978e-06,
      "loss": 0.6978,
      "step": 4490
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6787689596836228,
      "learning_rate": 2.277445416221542e-06,
      "loss": 0.7049,
      "step": 4491
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8854846907780783,
      "learning_rate": 2.273838106853151e-06,
      "loss": 0.8259,
      "step": 4492
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.60698044967814,
      "learning_rate": 2.2702332901055925e-06,
      "loss": 0.6681,
      "step": 4493
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6739138474065548,
      "learning_rate": 2.2666309671418584e-06,
      "loss": 0.6529,
      "step": 4494
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7568486363923367,
      "learning_rate": 2.2630311391241354e-06,
      "loss": 0.7214,
      "step": 4495
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6543353225718438,
      "learning_rate": 2.259433807213797e-06,
      "loss": 0.6903,
      "step": 4496
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6674657403390726,
      "learning_rate": 2.2558389725714215e-06,
      "loss": 0.6798,
      "step": 4497
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7688137043601442,
      "learning_rate": 2.2522466363567775e-06,
      "loss": 0.695,
      "step": 4498
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6216515645252135,
      "learning_rate": 2.2486567997288255e-06,
      "loss": 0.7012,
      "step": 4499
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6645684548540303,
      "learning_rate": 2.245069463845727e-06,
      "loss": 0.6658,
      "step": 4500
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6250391395130309,
      "learning_rate": 2.2414846298648262e-06,
      "loss": 0.7013,
      "step": 4501
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6814781903124701,
      "learning_rate": 2.2379022989426645e-06,
      "loss": 0.6917,
      "step": 4502
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7187479777578677,
      "learning_rate": 2.2343224722349777e-06,
      "loss": 0.7165,
      "step": 4503
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6541262426170698,
      "learning_rate": 2.230745150896694e-06,
      "loss": 0.6494,
      "step": 4504
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6198796222088301,
      "learning_rate": 2.2271703360819297e-06,
      "loss": 0.6575,
      "step": 4505
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7428950591611629,
      "learning_rate": 2.2235980289439994e-06,
      "loss": 0.7307,
      "step": 4506
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6515232234538271,
      "learning_rate": 2.220028230635398e-06,
      "loss": 0.67,
      "step": 4507
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6983443578688461,
      "learning_rate": 2.216460942307819e-06,
      "loss": 0.7135,
      "step": 4508
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6878743266097868,
      "learning_rate": 2.212896165112143e-06,
      "loss": 0.7313,
      "step": 4509
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7030740072655034,
      "learning_rate": 2.2093339001984457e-06,
      "loss": 0.6961,
      "step": 4510
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6886688349443971,
      "learning_rate": 2.205774148715982e-06,
      "loss": 0.7001,
      "step": 4511
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6616233429551137,
      "learning_rate": 2.2022169118132053e-06,
      "loss": 0.7029,
      "step": 4512
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6385290566318708,
      "learning_rate": 2.1986621906377557e-06,
      "loss": 0.6558,
      "step": 4513
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6310737953109874,
      "learning_rate": 2.1951099863364567e-06,
      "loss": 0.7218,
      "step": 4514
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7135552564058322,
      "learning_rate": 2.1915603000553243e-06,
      "loss": 0.6777,
      "step": 4515
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6251178974862999,
      "learning_rate": 2.1880131329395614e-06,
      "loss": 0.6534,
      "step": 4516
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6879250634536112,
      "learning_rate": 2.1844684861335596e-06,
      "loss": 0.7059,
      "step": 4517
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7163623936642638,
      "learning_rate": 2.180926360780897e-06,
      "loss": 0.7386,
      "step": 4518
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.750987964471114,
      "learning_rate": 2.1773867580243337e-06,
      "loss": 0.7243,
      "step": 4519
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.611169342153591,
      "learning_rate": 2.173849679005822e-06,
      "loss": 0.6633,
      "step": 4520
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.711980383788163,
      "learning_rate": 2.170315124866493e-06,
      "loss": 0.6877,
      "step": 4521
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1257377400197228,
      "learning_rate": 2.166783096746671e-06,
      "loss": 0.691,
      "step": 4522
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.728343640574522,
      "learning_rate": 2.1632535957858615e-06,
      "loss": 0.6996,
      "step": 4523
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6743381252088636,
      "learning_rate": 2.1597266231227566e-06,
      "loss": 0.679,
      "step": 4524
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6573893955744676,
      "learning_rate": 2.156202179895227e-06,
      "loss": 0.6776,
      "step": 4525
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7089358484852818,
      "learning_rate": 2.1526802672403344e-06,
      "loss": 0.7109,
      "step": 4526
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6918295556047012,
      "learning_rate": 2.14916088629432e-06,
      "loss": 0.7546,
      "step": 4527
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7036921028332425,
      "learning_rate": 2.145644038192611e-06,
      "loss": 0.6777,
      "step": 4528
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7188076018922774,
      "learning_rate": 2.142129724069818e-06,
      "loss": 0.6677,
      "step": 4529
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.672577050364119,
      "learning_rate": 2.138617945059729e-06,
      "loss": 0.6765,
      "step": 4530
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.662145076992799,
      "learning_rate": 2.1351087022953143e-06,
      "loss": 0.6955,
      "step": 4531
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6809039898516186,
      "learning_rate": 2.131601996908731e-06,
      "loss": 0.7076,
      "step": 4532
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6818274003460432,
      "learning_rate": 2.128097830031317e-06,
      "loss": 0.7282,
      "step": 4533
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6084531516030981,
      "learning_rate": 2.1245962027935872e-06,
      "loss": 0.6855,
      "step": 4534
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7215449136074293,
      "learning_rate": 2.121097116325241e-06,
      "loss": 0.709,
      "step": 4535
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6653176932344355,
      "learning_rate": 2.1176005717551607e-06,
      "loss": 0.7144,
      "step": 4536
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6661954164066004,
      "learning_rate": 2.1141065702113963e-06,
      "loss": 0.712,
      "step": 4537
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6889900553762947,
      "learning_rate": 2.1106151128211916e-06,
      "loss": 0.6961,
      "step": 4538
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.633811772330852,
      "learning_rate": 2.1071262007109595e-06,
      "loss": 0.6855,
      "step": 4539
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.62787513770302,
      "learning_rate": 2.103639835006297e-06,
      "loss": 0.6494,
      "step": 4540
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6571966748125566,
      "learning_rate": 2.100156016831979e-06,
      "loss": 0.6892,
      "step": 4541
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6802057777876734,
      "learning_rate": 2.0966747473119607e-06,
      "loss": 0.6553,
      "step": 4542
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6437957627291225,
      "learning_rate": 2.0931960275693665e-06,
      "loss": 0.6912,
      "step": 4543
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7334563212058297,
      "learning_rate": 2.0897198587265064e-06,
      "loss": 0.7517,
      "step": 4544
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6609160359353113,
      "learning_rate": 2.0862462419048658e-06,
      "loss": 0.7029,
      "step": 4545
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6767481549071976,
      "learning_rate": 2.082775178225105e-06,
      "loss": 0.7093,
      "step": 4546
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7255890519822781,
      "learning_rate": 2.079306668807063e-06,
      "loss": 0.6508,
      "step": 4547
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7337351071913467,
      "learning_rate": 2.0758407147697537e-06,
      "loss": 0.6885,
      "step": 4548
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6952191924927043,
      "learning_rate": 2.07237731723136e-06,
      "loss": 0.6978,
      "step": 4549
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6771684522707879,
      "learning_rate": 2.068916477309252e-06,
      "loss": 0.6688,
      "step": 4550
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6493665625747481,
      "learning_rate": 2.065458196119966e-06,
      "loss": 0.6859,
      "step": 4551
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7205124810908596,
      "learning_rate": 2.0620024747792167e-06,
      "loss": 0.7468,
      "step": 4552
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6681690249215563,
      "learning_rate": 2.0585493144018953e-06,
      "loss": 0.675,
      "step": 4553
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6412073150289124,
      "learning_rate": 2.0550987161020573e-06,
      "loss": 0.6918,
      "step": 4554
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6885496903777935,
      "learning_rate": 2.0516506809929395e-06,
      "loss": 0.652,
      "step": 4555
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6904744872755174,
      "learning_rate": 2.048205210186953e-06,
      "loss": 0.7167,
      "step": 4556
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7569442392695672,
      "learning_rate": 2.0447623047956743e-06,
      "loss": 0.7832,
      "step": 4557
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6281853626667768,
      "learning_rate": 2.0413219659298577e-06,
      "loss": 0.6619,
      "step": 4558
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6530263127521929,
      "learning_rate": 2.037884194699432e-06,
      "loss": 0.7049,
      "step": 4559
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7174470480044012,
      "learning_rate": 2.034448992213488e-06,
      "loss": 0.7191,
      "step": 4560
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6680984822075945,
      "learning_rate": 2.0310163595802958e-06,
      "loss": 0.689,
      "step": 4561
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.721289422360614,
      "learning_rate": 2.0275862979072934e-06,
      "loss": 0.7399,
      "step": 4562
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6139583827196448,
      "learning_rate": 2.0241588083010932e-06,
      "loss": 0.6667,
      "step": 4563
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6654518855227894,
      "learning_rate": 2.0207338918674712e-06,
      "loss": 0.6653,
      "step": 4564
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6536664736840289,
      "learning_rate": 2.0173115497113827e-06,
      "loss": 0.6899,
      "step": 4565
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6131094765710674,
      "learning_rate": 2.013891782936942e-06,
      "loss": 0.704,
      "step": 4566
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6423798252481222,
      "learning_rate": 2.0104745926474345e-06,
      "loss": 0.6528,
      "step": 4567
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6716420256161062,
      "learning_rate": 2.0070599799453205e-06,
      "loss": 0.6852,
      "step": 4568
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8464756404756416,
      "learning_rate": 2.0036479459322245e-06,
      "loss": 0.732,
      "step": 4569
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6903102629185387,
      "learning_rate": 2.0002384917089402e-06,
      "loss": 0.7332,
      "step": 4570
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5691045863046632,
      "learning_rate": 1.996831618375431e-06,
      "loss": 0.662,
      "step": 4571
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6115987259887065,
      "learning_rate": 1.99342732703082e-06,
      "loss": 0.6803,
      "step": 4572
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.753193542017943,
      "learning_rate": 1.990025618773406e-06,
      "loss": 0.7277,
      "step": 4573
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6848806529831551,
      "learning_rate": 1.986626494700652e-06,
      "loss": 0.681,
      "step": 4574
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6785070293065137,
      "learning_rate": 1.9832299559091817e-06,
      "loss": 0.6963,
      "step": 4575
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7814033432678373,
      "learning_rate": 1.979836003494793e-06,
      "loss": 0.69,
      "step": 4576
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.697526950811242,
      "learning_rate": 1.976444638552446e-06,
      "loss": 0.6908,
      "step": 4577
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6541800687855843,
      "learning_rate": 1.973055862176263e-06,
      "loss": 0.7082,
      "step": 4578
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6640115392479015,
      "learning_rate": 1.9696696754595346e-06,
      "loss": 0.6961,
      "step": 4579
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6928032905710458,
      "learning_rate": 1.9662860794947157e-06,
      "loss": 0.7111,
      "step": 4580
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6661869461180127,
      "learning_rate": 1.9629050753734257e-06,
      "loss": 0.7062,
      "step": 4581
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6727180376070863,
      "learning_rate": 1.9595266641864495e-06,
      "loss": 0.7034,
      "step": 4582
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7589286191370245,
      "learning_rate": 1.9561508470237268e-06,
      "loss": 0.6653,
      "step": 4583
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7076533869996535,
      "learning_rate": 1.952777624974371e-06,
      "loss": 0.7376,
      "step": 4584
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6380973979545261,
      "learning_rate": 1.9494069991266505e-06,
      "loss": 0.6953,
      "step": 4585
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6809263454769167,
      "learning_rate": 1.9460389705680014e-06,
      "loss": 0.7045,
      "step": 4586
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.709129588979124,
      "learning_rate": 1.9426735403850207e-06,
      "loss": 0.7237,
      "step": 4587
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6899077238420118,
      "learning_rate": 1.9393107096634678e-06,
      "loss": 0.7509,
      "step": 4588
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.724901815299391,
      "learning_rate": 1.9359504794882567e-06,
      "loss": 0.6922,
      "step": 4589
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6765344476936033,
      "learning_rate": 1.93259285094347e-06,
      "loss": 0.7152,
      "step": 4590
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7173240610751251,
      "learning_rate": 1.92923782511235e-06,
      "loss": 0.7655,
      "step": 4591
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.738545540625782,
      "learning_rate": 1.9258854030772966e-06,
      "loss": 0.7184,
      "step": 4592
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6148111030520652,
      "learning_rate": 1.922535585919875e-06,
      "loss": 0.6579,
      "step": 4593
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7052437847322001,
      "learning_rate": 1.9191883747207994e-06,
      "loss": 0.6879,
      "step": 4594
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7378133971570272,
      "learning_rate": 1.9158437705599562e-06,
      "loss": 0.7458,
      "step": 4595
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6309029063081407,
      "learning_rate": 1.9125017745163786e-06,
      "loss": 0.6893,
      "step": 4596
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6484004404833721,
      "learning_rate": 1.9091623876682665e-06,
      "loss": 0.7015,
      "step": 4597
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6870887226090817,
      "learning_rate": 1.9058256110929763e-06,
      "loss": 0.7018,
      "step": 4598
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6300375237894748,
      "learning_rate": 1.9024914458670219e-06,
      "loss": 0.6493,
      "step": 4599
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6769309181445892,
      "learning_rate": 1.899159893066076e-06,
      "loss": 0.7025,
      "step": 4600
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6516283951022662,
      "learning_rate": 1.895830953764962e-06,
      "loss": 0.687,
      "step": 4601
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7342108618192259,
      "learning_rate": 1.8925046290376725e-06,
      "loss": 0.6513,
      "step": 4602
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6770833810890599,
      "learning_rate": 1.8891809199573408e-06,
      "loss": 0.7718,
      "step": 4603
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5902327643521433,
      "learning_rate": 1.885859827596268e-06,
      "loss": 0.6737,
      "step": 4604
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6575803030573626,
      "learning_rate": 1.8825413530259096e-06,
      "loss": 0.6887,
      "step": 4605
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.624403245133022,
      "learning_rate": 1.8792254973168754e-06,
      "loss": 0.6704,
      "step": 4606
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7159685096671614,
      "learning_rate": 1.8759122615389258e-06,
      "loss": 0.708,
      "step": 4607
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6530557642859434,
      "learning_rate": 1.8726016467609809e-06,
      "loss": 0.6632,
      "step": 4608
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6690938042221254,
      "learning_rate": 1.869293654051114e-06,
      "loss": 0.6999,
      "step": 4609
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6824907641760002,
      "learning_rate": 1.8659882844765541e-06,
      "loss": 0.7024,
      "step": 4610
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6345796922591752,
      "learning_rate": 1.8626855391036835e-06,
      "loss": 0.6802,
      "step": 4611
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6343416893894599,
      "learning_rate": 1.8593854189980343e-06,
      "loss": 0.7103,
      "step": 4612
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7428090195610553,
      "learning_rate": 1.8560879252242925e-06,
      "loss": 0.7082,
      "step": 4613
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6237848182469442,
      "learning_rate": 1.8527930588462982e-06,
      "loss": 0.7071,
      "step": 4614
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6620868431069575,
      "learning_rate": 1.8495008209270471e-06,
      "loss": 0.6243,
      "step": 4615
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.677938380654325,
      "learning_rate": 1.8462112125286824e-06,
      "loss": 0.7034,
      "step": 4616
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.66032498975549,
      "learning_rate": 1.8429242347124998e-06,
      "loss": 0.7034,
      "step": 4617
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6287616895713798,
      "learning_rate": 1.839639888538951e-06,
      "loss": 0.6746,
      "step": 4618
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7205568004595887,
      "learning_rate": 1.8363581750676274e-06,
      "loss": 0.7176,
      "step": 4619
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7329527817276801,
      "learning_rate": 1.8330790953572842e-06,
      "loss": 0.7551,
      "step": 4620
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6718537263180131,
      "learning_rate": 1.8298026504658162e-06,
      "loss": 0.7058,
      "step": 4621
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8521671987179587,
      "learning_rate": 1.8265288414502746e-06,
      "loss": 0.7273,
      "step": 4622
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6909713952528331,
      "learning_rate": 1.8232576693668591e-06,
      "loss": 0.7272,
      "step": 4623
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.689090371516081,
      "learning_rate": 1.81998913527092e-06,
      "loss": 0.7366,
      "step": 4624
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7124356527103682,
      "learning_rate": 1.8167232402169488e-06,
      "loss": 0.751,
      "step": 4625
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6846577753782936,
      "learning_rate": 1.8134599852585954e-06,
      "loss": 0.6995,
      "step": 4626
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7367052950341315,
      "learning_rate": 1.8101993714486532e-06,
      "loss": 0.7305,
      "step": 4627
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6695477412447132,
      "learning_rate": 1.806941399839064e-06,
      "loss": 0.7017,
      "step": 4628
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6485107488723956,
      "learning_rate": 1.803686071480919e-06,
      "loss": 0.6927,
      "step": 4629
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6374911816813491,
      "learning_rate": 1.800433387424455e-06,
      "loss": 0.6829,
      "step": 4630
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6740367715038027,
      "learning_rate": 1.7971833487190504e-06,
      "loss": 0.6919,
      "step": 4631
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7903679694277188,
      "learning_rate": 1.7939359564132387e-06,
      "loss": 0.7067,
      "step": 4632
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6487237804474845,
      "learning_rate": 1.7906912115546972e-06,
      "loss": 0.6845,
      "step": 4633
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7148334932065173,
      "learning_rate": 1.7874491151902474e-06,
      "loss": 0.7221,
      "step": 4634
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5977208530411333,
      "learning_rate": 1.7842096683658606e-06,
      "loss": 0.7034,
      "step": 4635
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6242499933814962,
      "learning_rate": 1.780972872126644e-06,
      "loss": 0.6707,
      "step": 4636
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6396315259322602,
      "learning_rate": 1.7777387275168577e-06,
      "loss": 0.6627,
      "step": 4637
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6916858837199753,
      "learning_rate": 1.7745072355799065e-06,
      "loss": 0.6728,
      "step": 4638
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7033237431350391,
      "learning_rate": 1.7712783973583337e-06,
      "loss": 0.7123,
      "step": 4639
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6522562216385117,
      "learning_rate": 1.7680522138938316e-06,
      "loss": 0.7049,
      "step": 4640
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6653458307907753,
      "learning_rate": 1.7648286862272358e-06,
      "loss": 0.6804,
      "step": 4641
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6851292022990259,
      "learning_rate": 1.7616078153985183e-06,
      "loss": 0.6691,
      "step": 4642
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6582605910825806,
      "learning_rate": 1.758389602446804e-06,
      "loss": 0.6727,
      "step": 4643
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.688394283164115,
      "learning_rate": 1.755174048410353e-06,
      "loss": 0.7183,
      "step": 4644
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6851553326688089,
      "learning_rate": 1.7519611543265713e-06,
      "loss": 0.6831,
      "step": 4645
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6316533175283278,
      "learning_rate": 1.7487509212320043e-06,
      "loss": 0.704,
      "step": 4646
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6549275342006233,
      "learning_rate": 1.7455433501623442e-06,
      "loss": 0.6712,
      "step": 4647
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7265867031142141,
      "learning_rate": 1.742338442152417e-06,
      "loss": 0.757,
      "step": 4648
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6564602641397871,
      "learning_rate": 1.7391361982361899e-06,
      "loss": 0.6542,
      "step": 4649
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6329647399698339,
      "learning_rate": 1.7359366194467774e-06,
      "loss": 0.6857,
      "step": 4650
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7229920100853027,
      "learning_rate": 1.7327397068164286e-06,
      "loss": 0.7154,
      "step": 4651
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7014235729318982,
      "learning_rate": 1.729545461376535e-06,
      "loss": 0.7478,
      "step": 4652
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6471534262822609,
      "learning_rate": 1.726353884157629e-06,
      "loss": 0.6716,
      "step": 4653
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7022685099805295,
      "learning_rate": 1.7231649761893742e-06,
      "loss": 0.7264,
      "step": 4654
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6612574565200947,
      "learning_rate": 1.719978738500584e-06,
      "loss": 0.6623,
      "step": 4655
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6464546812647457,
      "learning_rate": 1.7167951721192011e-06,
      "loss": 0.6857,
      "step": 4656
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7107618471752077,
      "learning_rate": 1.7136142780723164e-06,
      "loss": 0.6666,
      "step": 4657
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6553048301348487,
      "learning_rate": 1.7104360573861456e-06,
      "loss": 0.6569,
      "step": 4658
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7037444743049592,
      "learning_rate": 1.707260511086054e-06,
      "loss": 0.6937,
      "step": 4659
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.626352590503863,
      "learning_rate": 1.7040876401965355e-06,
      "loss": 0.6647,
      "step": 4660
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7176115519054681,
      "learning_rate": 1.7009174457412258e-06,
      "loss": 0.6904,
      "step": 4661
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6514160564725299,
      "learning_rate": 1.6977499287428944e-06,
      "loss": 0.6891,
      "step": 4662
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6653571258284761,
      "learning_rate": 1.6945850902234506e-06,
      "loss": 0.6743,
      "step": 4663
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6378753384490797,
      "learning_rate": 1.6914229312039387e-06,
      "loss": 0.6877,
      "step": 4664
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6957025245826159,
      "learning_rate": 1.6882634527045328e-06,
      "loss": 0.6918,
      "step": 4665
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6479913705200788,
      "learning_rate": 1.6851066557445505e-06,
      "loss": 0.6867,
      "step": 4666
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6842224702404357,
      "learning_rate": 1.6819525413424354e-06,
      "loss": 0.6954,
      "step": 4667
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6706047910707377,
      "learning_rate": 1.6788011105157731e-06,
      "loss": 0.6965,
      "step": 4668
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6342890891718624,
      "learning_rate": 1.6756523642812815e-06,
      "loss": 0.6784,
      "step": 4669
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7450188833205046,
      "learning_rate": 1.6725063036548138e-06,
      "loss": 0.7102,
      "step": 4670
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6493030117324775,
      "learning_rate": 1.6693629296513503e-06,
      "loss": 0.6721,
      "step": 4671
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7149432627475433,
      "learning_rate": 1.6662222432850096e-06,
      "loss": 0.6702,
      "step": 4672
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6479056071933678,
      "learning_rate": 1.6630842455690443e-06,
      "loss": 0.7361,
      "step": 4673
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6963659854138308,
      "learning_rate": 1.6599489375158383e-06,
      "loss": 0.7277,
      "step": 4674
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.691366333648258,
      "learning_rate": 1.6568163201369081e-06,
      "loss": 0.7185,
      "step": 4675
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5627432335971987,
      "learning_rate": 1.6536863944428982e-06,
      "loss": 0.6651,
      "step": 4676
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6530731591214513,
      "learning_rate": 1.65055916144359e-06,
      "loss": 0.7075,
      "step": 4677
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6840705630127601,
      "learning_rate": 1.6474346221478931e-06,
      "loss": 0.72,
      "step": 4678
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7036666266910252,
      "learning_rate": 1.644312777563848e-06,
      "loss": 0.7484,
      "step": 4679
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6543558437564064,
      "learning_rate": 1.6411936286986286e-06,
      "loss": 0.6698,
      "step": 4680
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.769081165891024,
      "learning_rate": 1.6380771765585357e-06,
      "loss": 0.7193,
      "step": 4681
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6913401853765636,
      "learning_rate": 1.6349634221490063e-06,
      "loss": 0.6639,
      "step": 4682
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7467016367195208,
      "learning_rate": 1.6318523664745966e-06,
      "loss": 0.7415,
      "step": 4683
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8160606272411302,
      "learning_rate": 1.6287440105390028e-06,
      "loss": 0.8014,
      "step": 4684
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6332313456953527,
      "learning_rate": 1.6256383553450394e-06,
      "loss": 0.6468,
      "step": 4685
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.696225372887348,
      "learning_rate": 1.6225354018946605e-06,
      "loss": 0.7125,
      "step": 4686
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6244208317437558,
      "learning_rate": 1.6194351511889418e-06,
      "loss": 0.6977,
      "step": 4687
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5871001950215318,
      "learning_rate": 1.6163376042280909e-06,
      "loss": 0.6466,
      "step": 4688
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6460513560438443,
      "learning_rate": 1.6132427620114366e-06,
      "loss": 0.689,
      "step": 4689
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6277024723388414,
      "learning_rate": 1.6101506255374422e-06,
      "loss": 0.6903,
      "step": 4690
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6001226464114388,
      "learning_rate": 1.6070611958036953e-06,
      "loss": 0.633,
      "step": 4691
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6773580774870095,
      "learning_rate": 1.6039744738069097e-06,
      "loss": 0.6932,
      "step": 4692
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7252279571264698,
      "learning_rate": 1.6008904605429288e-06,
      "loss": 0.7041,
      "step": 4693
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.706783101341306,
      "learning_rate": 1.597809157006719e-06,
      "loss": 0.6851,
      "step": 4694
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8894084147452546,
      "learning_rate": 1.5947305641923672e-06,
      "loss": 0.7052,
      "step": 4695
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6678441153022964,
      "learning_rate": 1.5916546830930967e-06,
      "loss": 0.6873,
      "step": 4696
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6794751496353555,
      "learning_rate": 1.5885815147012507e-06,
      "loss": 0.6999,
      "step": 4697
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7047943946190242,
      "learning_rate": 1.585511060008297e-06,
      "loss": 0.6834,
      "step": 4698
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6448194625264313,
      "learning_rate": 1.582443320004826e-06,
      "loss": 0.6795,
      "step": 4699
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7139407016806248,
      "learning_rate": 1.5793782956805603e-06,
      "loss": 0.7177,
      "step": 4700
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6245002814511571,
      "learning_rate": 1.576315988024335e-06,
      "loss": 0.687,
      "step": 4701
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7656075769112065,
      "learning_rate": 1.5732563980241178e-06,
      "loss": 0.7529,
      "step": 4702
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6137069577054421,
      "learning_rate": 1.5701995266669923e-06,
      "loss": 0.6658,
      "step": 4703
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6827902980743689,
      "learning_rate": 1.5671453749391707e-06,
      "loss": 0.6942,
      "step": 4704
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6671084523122224,
      "learning_rate": 1.5640939438259873e-06,
      "loss": 0.6785,
      "step": 4705
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6848047551076408,
      "learning_rate": 1.5610452343118986e-06,
      "loss": 0.6673,
      "step": 4706
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6880267352456603,
      "learning_rate": 1.5579992473804773e-06,
      "loss": 0.6875,
      "step": 4707
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6482312109827418,
      "learning_rate": 1.5549559840144258e-06,
      "loss": 0.69,
      "step": 4708
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7574197234090781,
      "learning_rate": 1.551915445195562e-06,
      "loss": 0.7235,
      "step": 4709
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7446928947791482,
      "learning_rate": 1.5488776319048283e-06,
      "loss": 0.7232,
      "step": 4710
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6507302819567733,
      "learning_rate": 1.5458425451222903e-06,
      "loss": 0.6829,
      "step": 4711
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.616548211450258,
      "learning_rate": 1.542810185827125e-06,
      "loss": 0.6691,
      "step": 4712
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7438372800005378,
      "learning_rate": 1.5397805549976352e-06,
      "loss": 0.7482,
      "step": 4713
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6499602216528263,
      "learning_rate": 1.5367536536112426e-06,
      "loss": 0.7167,
      "step": 4714
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6869869458943392,
      "learning_rate": 1.5337294826444915e-06,
      "loss": 0.7174,
      "step": 4715
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7607100605879255,
      "learning_rate": 1.5307080430730404e-06,
      "loss": 0.7373,
      "step": 4716
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.753360097596143,
      "learning_rate": 1.5276893358716706e-06,
      "loss": 0.7391,
      "step": 4717
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7020437663052306,
      "learning_rate": 1.5246733620142773e-06,
      "loss": 0.7224,
      "step": 4718
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6844852717992717,
      "learning_rate": 1.521660122473878e-06,
      "loss": 0.6912,
      "step": 4719
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7264917848639948,
      "learning_rate": 1.5186496182226063e-06,
      "loss": 0.7345,
      "step": 4720
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7135575753418402,
      "learning_rate": 1.5156418502317162e-06,
      "loss": 0.7258,
      "step": 4721
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6795271564149057,
      "learning_rate": 1.5126368194715712e-06,
      "loss": 0.7171,
      "step": 4722
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.584695423689043,
      "learning_rate": 1.509634526911662e-06,
      "loss": 0.6611,
      "step": 4723
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6899455832297352,
      "learning_rate": 1.5066349735205865e-06,
      "loss": 0.7216,
      "step": 4724
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6502431884450997,
      "learning_rate": 1.5036381602660643e-06,
      "loss": 0.6989,
      "step": 4725
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6953021025272058,
      "learning_rate": 1.5006440881149309e-06,
      "loss": 0.6599,
      "step": 4726
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6433418509715125,
      "learning_rate": 1.4976527580331347e-06,
      "loss": 0.7316,
      "step": 4727
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6989503841429595,
      "learning_rate": 1.4946641709857434e-06,
      "loss": 0.6716,
      "step": 4728
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6380393766185881,
      "learning_rate": 1.4916783279369384e-06,
      "loss": 0.6947,
      "step": 4729
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6646686850485624,
      "learning_rate": 1.4886952298500124e-06,
      "loss": 0.7004,
      "step": 4730
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7673337592901556,
      "learning_rate": 1.4857148776873741e-06,
      "loss": 0.7134,
      "step": 4731
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6304558670602254,
      "learning_rate": 1.4827372724105483e-06,
      "loss": 0.6971,
      "step": 4732
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6348422242143614,
      "learning_rate": 1.4797624149801726e-06,
      "loss": 0.6912,
      "step": 4733
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6889504526384641,
      "learning_rate": 1.4767903063559973e-06,
      "loss": 0.7017,
      "step": 4734
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7494002268483599,
      "learning_rate": 1.4738209474968912e-06,
      "loss": 0.742,
      "step": 4735
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6608031718235026,
      "learning_rate": 1.4708543393608244e-06,
      "loss": 0.6933,
      "step": 4736
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6726643152086824,
      "learning_rate": 1.4678904829048901e-06,
      "loss": 0.704,
      "step": 4737
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8178065344716932,
      "learning_rate": 1.4649293790852904e-06,
      "loss": 0.7838,
      "step": 4738
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6228039717815039,
      "learning_rate": 1.4619710288573397e-06,
      "loss": 0.6602,
      "step": 4739
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7179495152760568,
      "learning_rate": 1.4590154331754602e-06,
      "loss": 0.7336,
      "step": 4740
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6820620772437497,
      "learning_rate": 1.4560625929931937e-06,
      "loss": 0.6985,
      "step": 4741
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5868530113159003,
      "learning_rate": 1.4531125092631815e-06,
      "loss": 0.6428,
      "step": 4742
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6430078450460105,
      "learning_rate": 1.4501651829371865e-06,
      "loss": 0.6576,
      "step": 4743
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6652209928856478,
      "learning_rate": 1.4472206149660761e-06,
      "loss": 0.688,
      "step": 4744
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8293685680434129,
      "learning_rate": 1.4442788062998303e-06,
      "loss": 0.6964,
      "step": 4745
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.637386111070758,
      "learning_rate": 1.44133975788754e-06,
      "loss": 0.6708,
      "step": 4746
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7156513101965623,
      "learning_rate": 1.4384034706773974e-06,
      "loss": 0.7422,
      "step": 4747
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6996370615375304,
      "learning_rate": 1.4354699456167164e-06,
      "loss": 0.6847,
      "step": 4748
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6726300549136773,
      "learning_rate": 1.4325391836519087e-06,
      "loss": 0.7283,
      "step": 4749
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6886077946184247,
      "learning_rate": 1.4296111857285e-06,
      "loss": 0.6785,
      "step": 4750
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7404085913512789,
      "learning_rate": 1.4266859527911249e-06,
      "loss": 0.7339,
      "step": 4751
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.59083812382409,
      "learning_rate": 1.4237634857835248e-06,
      "loss": 0.6659,
      "step": 4752
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6130640218293704,
      "learning_rate": 1.4208437856485468e-06,
      "loss": 0.6674,
      "step": 4753
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6829965081032678,
      "learning_rate": 1.4179268533281464e-06,
      "loss": 0.6951,
      "step": 4754
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7240908029454765,
      "learning_rate": 1.4150126897633886e-06,
      "loss": 0.7009,
      "step": 4755
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7104604242004848,
      "learning_rate": 1.4121012958944436e-06,
      "loss": 0.7235,
      "step": 4756
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6415878525929153,
      "learning_rate": 1.4091926726605886e-06,
      "loss": 0.6762,
      "step": 4757
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6671879721643932,
      "learning_rate": 1.4062868210002024e-06,
      "loss": 0.7411,
      "step": 4758
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6030837536506765,
      "learning_rate": 1.4033837418507768e-06,
      "loss": 0.6423,
      "step": 4759
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7391024443091248,
      "learning_rate": 1.400483436148903e-06,
      "loss": 0.7023,
      "step": 4760
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6341042355623117,
      "learning_rate": 1.3975859048302798e-06,
      "loss": 0.6703,
      "step": 4761
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6790344616327368,
      "learning_rate": 1.3946911488297133e-06,
      "loss": 0.7281,
      "step": 4762
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6741757143952178,
      "learning_rate": 1.3917991690811117e-06,
      "loss": 0.7399,
      "step": 4763
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7878898573172304,
      "learning_rate": 1.3889099665174898e-06,
      "loss": 0.7264,
      "step": 4764
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6476082850121421,
      "learning_rate": 1.38602354207096e-06,
      "loss": 0.7183,
      "step": 4765
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7331053641895812,
      "learning_rate": 1.3831398966727473e-06,
      "loss": 0.7301,
      "step": 4766
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6301458110478818,
      "learning_rate": 1.3802590312531717e-06,
      "loss": 0.7056,
      "step": 4767
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6601684149183544,
      "learning_rate": 1.3773809467416621e-06,
      "loss": 0.6433,
      "step": 4768
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6666510503693993,
      "learning_rate": 1.3745056440667503e-06,
      "loss": 0.6832,
      "step": 4769
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6632555395646473,
      "learning_rate": 1.3716331241560688e-06,
      "loss": 0.6541,
      "step": 4770
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7205023737926639,
      "learning_rate": 1.3687633879363494e-06,
      "loss": 0.6579,
      "step": 4771
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6728635842756192,
      "learning_rate": 1.3658964363334304e-06,
      "loss": 0.7088,
      "step": 4772
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7161502097499756,
      "learning_rate": 1.3630322702722498e-06,
      "loss": 0.6607,
      "step": 4773
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7122991951406784,
      "learning_rate": 1.3601708906768486e-06,
      "loss": 0.695,
      "step": 4774
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7246323863685169,
      "learning_rate": 1.357312298470369e-06,
      "loss": 0.7659,
      "step": 4775
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7062706149731478,
      "learning_rate": 1.3544564945750495e-06,
      "loss": 0.7033,
      "step": 4776
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6701819448534938,
      "learning_rate": 1.3516034799122303e-06,
      "loss": 0.6944,
      "step": 4777
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7719666385763106,
      "learning_rate": 1.3487532554023553e-06,
      "loss": 0.7191,
      "step": 4778
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6868568370196396,
      "learning_rate": 1.3459058219649678e-06,
      "loss": 0.683,
      "step": 4779
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.713783122366391,
      "learning_rate": 1.3430611805187066e-06,
      "loss": 0.7234,
      "step": 4780
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6981344785187124,
      "learning_rate": 1.3402193319813173e-06,
      "loss": 0.6974,
      "step": 4781
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8301309162026729,
      "learning_rate": 1.337380277269632e-06,
      "loss": 0.7866,
      "step": 4782
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7537076084946169,
      "learning_rate": 1.3345440172995939e-06,
      "loss": 0.742,
      "step": 4783
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6331608088011959,
      "learning_rate": 1.3317105529862384e-06,
      "loss": 0.6814,
      "step": 4784
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7292611161087612,
      "learning_rate": 1.3288798852437025e-06,
      "loss": 0.7201,
      "step": 4785
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7784856911454575,
      "learning_rate": 1.326052014985214e-06,
      "loss": 0.7589,
      "step": 4786
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6804735530802649,
      "learning_rate": 1.323226943123105e-06,
      "loss": 0.6863,
      "step": 4787
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6658051831397126,
      "learning_rate": 1.3204046705688045e-06,
      "loss": 0.6668,
      "step": 4788
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5766032291336665,
      "learning_rate": 1.3175851982328324e-06,
      "loss": 0.6693,
      "step": 4789
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6703660877340406,
      "learning_rate": 1.3147685270248113e-06,
      "loss": 0.7041,
      "step": 4790
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6431076690714641,
      "learning_rate": 1.3119546578534582e-06,
      "loss": 0.6792,
      "step": 4791
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.747222987643262,
      "learning_rate": 1.309143591626586e-06,
      "loss": 0.7339,
      "step": 4792
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6207811673726294,
      "learning_rate": 1.3063353292511061e-06,
      "loss": 0.6403,
      "step": 4793
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6249940181339395,
      "learning_rate": 1.303529871633018e-06,
      "loss": 0.6707,
      "step": 4794
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5994443037199673,
      "learning_rate": 1.3007272196774212e-06,
      "loss": 0.6601,
      "step": 4795
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6983858673058554,
      "learning_rate": 1.29792737428851e-06,
      "loss": 0.7291,
      "step": 4796
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5971199963047447,
      "learning_rate": 1.2951303363695732e-06,
      "loss": 0.6979,
      "step": 4797
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6282339110969687,
      "learning_rate": 1.2923361068229946e-06,
      "loss": 0.6817,
      "step": 4798
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7374839388884193,
      "learning_rate": 1.2895446865502525e-06,
      "loss": 0.6931,
      "step": 4799
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6764287988479992,
      "learning_rate": 1.2867560764519126e-06,
      "loss": 0.738,
      "step": 4800
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6782477196231718,
      "learning_rate": 1.2839702774276409e-06,
      "loss": 0.6694,
      "step": 4801
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6588670093927232,
      "learning_rate": 1.2811872903761936e-06,
      "loss": 0.68,
      "step": 4802
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6474565397090951,
      "learning_rate": 1.2784071161954237e-06,
      "loss": 0.6705,
      "step": 4803
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6743271629639025,
      "learning_rate": 1.2756297557822683e-06,
      "loss": 0.6657,
      "step": 4804
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7034797040570117,
      "learning_rate": 1.2728552100327663e-06,
      "loss": 0.6878,
      "step": 4805
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6656336466153981,
      "learning_rate": 1.2700834798420403e-06,
      "loss": 0.6886,
      "step": 4806
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7075497189242923,
      "learning_rate": 1.2673145661043085e-06,
      "loss": 0.7276,
      "step": 4807
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6641957899717228,
      "learning_rate": 1.2645484697128819e-06,
      "loss": 0.6869,
      "step": 4808
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.69913131875725,
      "learning_rate": 1.2617851915601597e-06,
      "loss": 0.7046,
      "step": 4809
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6282810944399446,
      "learning_rate": 1.2590247325376337e-06,
      "loss": 0.699,
      "step": 4810
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6586695443808623,
      "learning_rate": 1.2562670935358878e-06,
      "loss": 0.6961,
      "step": 4811
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6008855700757453,
      "learning_rate": 1.2535122754445916e-06,
      "loss": 0.6147,
      "step": 4812
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6977069503186857,
      "learning_rate": 1.2507602791525042e-06,
      "loss": 0.6821,
      "step": 4813
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6524217055388718,
      "learning_rate": 1.2480111055474796e-06,
      "loss": 0.6659,
      "step": 4814
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7498900340803232,
      "learning_rate": 1.2452647555164588e-06,
      "loss": 0.6994,
      "step": 4815
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7280710879291313,
      "learning_rate": 1.2425212299454702e-06,
      "loss": 0.6675,
      "step": 4816
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7058669277689203,
      "learning_rate": 1.2397805297196353e-06,
      "loss": 0.7153,
      "step": 4817
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7294868159375801,
      "learning_rate": 1.2370426557231563e-06,
      "loss": 0.7443,
      "step": 4818
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7151294504428718,
      "learning_rate": 1.2343076088393324e-06,
      "loss": 0.7637,
      "step": 4819
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7063974785134616,
      "learning_rate": 1.2315753899505446e-06,
      "loss": 0.6987,
      "step": 4820
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7265672003523526,
      "learning_rate": 1.2288459999382663e-06,
      "loss": 0.7103,
      "step": 4821
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6530580577673792,
      "learning_rate": 1.2261194396830523e-06,
      "loss": 0.6956,
      "step": 4822
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6622418962023245,
      "learning_rate": 1.223395710064551e-06,
      "loss": 0.6513,
      "step": 4823
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.653470426488364,
      "learning_rate": 1.2206748119614898e-06,
      "loss": 0.6627,
      "step": 4824
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6388244185414118,
      "learning_rate": 1.2179567462516906e-06,
      "loss": 0.6908,
      "step": 4825
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6783308950945004,
      "learning_rate": 1.2152415138120576e-06,
      "loss": 0.7408,
      "step": 4826
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6275967730090444,
      "learning_rate": 1.2125291155185803e-06,
      "loss": 0.6659,
      "step": 4827
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6530802967993884,
      "learning_rate": 1.2098195522463386e-06,
      "loss": 0.6682,
      "step": 4828
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7243006691898933,
      "learning_rate": 1.2071128248694886e-06,
      "loss": 0.7231,
      "step": 4829
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6702433860566109,
      "learning_rate": 1.2044089342612807e-06,
      "loss": 0.6999,
      "step": 4830
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6768883387568334,
      "learning_rate": 1.201707881294044e-06,
      "loss": 0.7146,
      "step": 4831
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6161983822269897,
      "learning_rate": 1.199009666839196e-06,
      "loss": 0.658,
      "step": 4832
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6431740566055447,
      "learning_rate": 1.1963142917672354e-06,
      "loss": 0.6769,
      "step": 4833
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6296486414557712,
      "learning_rate": 1.1936217569477505e-06,
      "loss": 0.6731,
      "step": 4834
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6508217859867451,
      "learning_rate": 1.1909320632494036e-06,
      "loss": 0.6697,
      "step": 4835
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7059237156671958,
      "learning_rate": 1.1882452115399489e-06,
      "loss": 0.6973,
      "step": 4836
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7445653973510997,
      "learning_rate": 1.1855612026862206e-06,
      "loss": 0.7516,
      "step": 4837
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6603233366538676,
      "learning_rate": 1.182880037554135e-06,
      "loss": 0.6632,
      "step": 4838
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.64293288299969,
      "learning_rate": 1.1802017170086932e-06,
      "loss": 0.6789,
      "step": 4839
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7765429354700775,
      "learning_rate": 1.1775262419139788e-06,
      "loss": 0.7458,
      "step": 4840
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7045430227232269,
      "learning_rate": 1.1748536131331522e-06,
      "loss": 0.6621,
      "step": 4841
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7123200960122323,
      "learning_rate": 1.1721838315284595e-06,
      "loss": 0.7325,
      "step": 4842
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6718189132796649,
      "learning_rate": 1.1695168979612282e-06,
      "loss": 0.6625,
      "step": 4843
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6903159039175029,
      "learning_rate": 1.1668528132918677e-06,
      "loss": 0.6963,
      "step": 4844
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.666753635327419,
      "learning_rate": 1.1641915783798663e-06,
      "loss": 0.7196,
      "step": 4845
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6301011292438301,
      "learning_rate": 1.1615331940837971e-06,
      "loss": 0.6889,
      "step": 4846
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6050076302484989,
      "learning_rate": 1.1588776612613063e-06,
      "loss": 0.6931,
      "step": 4847
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7298153750825876,
      "learning_rate": 1.156224980769124e-06,
      "loss": 0.7888,
      "step": 4848
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6719390262816924,
      "learning_rate": 1.1535751534630634e-06,
      "loss": 0.7011,
      "step": 4849
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.691024870609015,
      "learning_rate": 1.150928180198011e-06,
      "loss": 0.7427,
      "step": 4850
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6275868720311245,
      "learning_rate": 1.1482840618279368e-06,
      "loss": 0.6773,
      "step": 4851
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.65788819286804,
      "learning_rate": 1.1456427992058904e-06,
      "loss": 0.646,
      "step": 4852
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6671933477432723,
      "learning_rate": 1.143004393183994e-06,
      "loss": 0.7071,
      "step": 4853
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7408073118685726,
      "learning_rate": 1.1403688446134543e-06,
      "loss": 0.7234,
      "step": 4854
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5717768379698961,
      "learning_rate": 1.1377361543445542e-06,
      "loss": 0.6532,
      "step": 4855
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6167070490471913,
      "learning_rate": 1.1351063232266558e-06,
      "loss": 0.6802,
      "step": 4856
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6977011433193291,
      "learning_rate": 1.1324793521081966e-06,
      "loss": 0.7066,
      "step": 4857
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7160993816645385,
      "learning_rate": 1.1298552418366938e-06,
      "loss": 0.6947,
      "step": 4858
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6445036018646161,
      "learning_rate": 1.127233993258735e-06,
      "loss": 0.6843,
      "step": 4859
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6715101842304472,
      "learning_rate": 1.1246156072199933e-06,
      "loss": 0.7034,
      "step": 4860
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7242974379597207,
      "learning_rate": 1.1220000845652135e-06,
      "loss": 0.7119,
      "step": 4861
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6500869868896797,
      "learning_rate": 1.1193874261382198e-06,
      "loss": 0.6703,
      "step": 4862
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6596519607991453,
      "learning_rate": 1.116777632781909e-06,
      "loss": 0.7016,
      "step": 4863
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6475290177570928,
      "learning_rate": 1.1141707053382533e-06,
      "loss": 0.7128,
      "step": 4864
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7517856219309704,
      "learning_rate": 1.1115666446483042e-06,
      "loss": 0.7196,
      "step": 4865
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7649905339930431,
      "learning_rate": 1.1089654515521831e-06,
      "loss": 0.7668,
      "step": 4866
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7274581456065664,
      "learning_rate": 1.1063671268890941e-06,
      "loss": 0.7115,
      "step": 4867
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7304124061874501,
      "learning_rate": 1.1037716714973057e-06,
      "loss": 0.706,
      "step": 4868
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5953187857500197,
      "learning_rate": 1.1011790862141669e-06,
      "loss": 0.6513,
      "step": 4869
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7382749953722643,
      "learning_rate": 1.0985893718761032e-06,
      "loss": 0.7228,
      "step": 4870
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6907849652945868,
      "learning_rate": 1.096002529318606e-06,
      "loss": 0.7017,
      "step": 4871
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6321082311270549,
      "learning_rate": 1.0934185593762458e-06,
      "loss": 0.6864,
      "step": 4872
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6454731631554506,
      "learning_rate": 1.0908374628826667e-06,
      "loss": 0.6727,
      "step": 4873
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6474850881902792,
      "learning_rate": 1.0882592406705839e-06,
      "loss": 0.6798,
      "step": 4874
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7366304386519741,
      "learning_rate": 1.0856838935717862e-06,
      "loss": 0.7453,
      "step": 4875
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.724308765521185,
      "learning_rate": 1.0831114224171334e-06,
      "loss": 0.6778,
      "step": 4876
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6715619650717174,
      "learning_rate": 1.0805418280365554e-06,
      "loss": 0.7027,
      "step": 4877
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6848498952685574,
      "learning_rate": 1.077975111259061e-06,
      "loss": 0.6769,
      "step": 4878
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6498414098950632,
      "learning_rate": 1.075411272912724e-06,
      "loss": 0.6903,
      "step": 4879
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6176596585387112,
      "learning_rate": 1.0728503138246937e-06,
      "loss": 0.7011,
      "step": 4880
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6236191408323828,
      "learning_rate": 1.0702922348211897e-06,
      "loss": 0.6685,
      "step": 4881
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5993106356865675,
      "learning_rate": 1.0677370367274987e-06,
      "loss": 0.6828,
      "step": 4882
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.666362079414263,
      "learning_rate": 1.065184720367982e-06,
      "loss": 0.7404,
      "step": 4883
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.60485778614732,
      "learning_rate": 1.0626352865660705e-06,
      "loss": 0.6811,
      "step": 4884
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7710951943923767,
      "learning_rate": 1.060088736144267e-06,
      "loss": 0.7486,
      "step": 4885
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7258074283728753,
      "learning_rate": 1.0575450699241364e-06,
      "loss": 0.6822,
      "step": 4886
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6421610518013774,
      "learning_rate": 1.0550042887263246e-06,
      "loss": 0.6958,
      "step": 4887
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6744656882030388,
      "learning_rate": 1.0524663933705348e-06,
      "loss": 0.7272,
      "step": 4888
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6427982265151857,
      "learning_rate": 1.0499313846755465e-06,
      "loss": 0.686,
      "step": 4889
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6447519324675348,
      "learning_rate": 1.0473992634592088e-06,
      "loss": 0.6475,
      "step": 4890
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6488328537102194,
      "learning_rate": 1.0448700305384341e-06,
      "loss": 0.7256,
      "step": 4891
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5980259745642345,
      "learning_rate": 1.042343686729208e-06,
      "loss": 0.6567,
      "step": 4892
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5664772589906449,
      "learning_rate": 1.0398202328465812e-06,
      "loss": 0.6583,
      "step": 4893
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7177764515045783,
      "learning_rate": 1.0372996697046723e-06,
      "loss": 0.7541,
      "step": 4894
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.695141845393571,
      "learning_rate": 1.0347819981166652e-06,
      "loss": 0.6972,
      "step": 4895
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6923209819684719,
      "learning_rate": 1.0322672188948146e-06,
      "loss": 0.7233,
      "step": 4896
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6945020151996663,
      "learning_rate": 1.0297553328504417e-06,
      "loss": 0.6933,
      "step": 4897
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7441675676896198,
      "learning_rate": 1.0272463407939314e-06,
      "loss": 0.7205,
      "step": 4898
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7110725944299673,
      "learning_rate": 1.0247402435347409e-06,
      "loss": 0.714,
      "step": 4899
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6448520211972052,
      "learning_rate": 1.0222370418813854e-06,
      "loss": 0.6932,
      "step": 4900
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6811628834476571,
      "learning_rate": 1.01973673664145e-06,
      "loss": 0.7119,
      "step": 4901
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7138603376800582,
      "learning_rate": 1.0172393286215866e-06,
      "loss": 0.7011,
      "step": 4902
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6613630642783259,
      "learning_rate": 1.014744818627511e-06,
      "loss": 0.6802,
      "step": 4903
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6809612659560299,
      "learning_rate": 1.0122532074640057e-06,
      "loss": 0.7248,
      "step": 4904
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6511937967434017,
      "learning_rate": 1.009764495934914e-06,
      "loss": 0.6698,
      "step": 4905
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7222130481501683,
      "learning_rate": 1.007278684843146e-06,
      "loss": 0.6587,
      "step": 4906
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7478976780826949,
      "learning_rate": 1.0047957749906766e-06,
      "loss": 0.7666,
      "step": 4907
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6879133183276845,
      "learning_rate": 1.0023157671785444e-06,
      "loss": 0.6863,
      "step": 4908
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6549225291676015,
      "learning_rate": 9.998386622068512e-07,
      "loss": 0.6738,
      "step": 4909
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7058123130868491,
      "learning_rate": 9.973644608747657e-07,
      "loss": 0.7221,
      "step": 4910
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7853702890761152,
      "learning_rate": 9.948931639805127e-07,
      "loss": 0.6975,
      "step": 4911
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7067801324774641,
      "learning_rate": 9.924247723213854e-07,
      "loss": 0.696,
      "step": 4912
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6844105795611678,
      "learning_rate": 9.899592866937413e-07,
      "loss": 0.7139,
      "step": 4913
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6696231816556764,
      "learning_rate": 9.87496707892992e-07,
      "loss": 0.6837,
      "step": 4914
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6164697733305771,
      "learning_rate": 9.850370367136197e-07,
      "loss": 0.689,
      "step": 4915
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6940258306740746,
      "learning_rate": 9.825802739491686e-07,
      "loss": 0.7215,
      "step": 4916
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7539289807032434,
      "learning_rate": 9.801264203922356e-07,
      "loss": 0.7023,
      "step": 4917
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6909402712957017,
      "learning_rate": 9.776754768344877e-07,
      "loss": 0.7098,
      "step": 4918
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.693728546627523,
      "learning_rate": 9.75227444066651e-07,
      "loss": 0.699,
      "step": 4919
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.651143577531007,
      "learning_rate": 9.727823228785105e-07,
      "loss": 0.7179,
      "step": 4920
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7366228545955361,
      "learning_rate": 9.703401140589153e-07,
      "loss": 0.7006,
      "step": 4921
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6946680200403874,
      "learning_rate": 9.679008183957728e-07,
      "loss": 0.7106,
      "step": 4922
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6293700690968625,
      "learning_rate": 9.6546443667605e-07,
      "loss": 0.6794,
      "step": 4923
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7421096402885868,
      "learning_rate": 9.63030969685772e-07,
      "loss": 0.7021,
      "step": 4924
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6225324722133606,
      "learning_rate": 9.606004182100271e-07,
      "loss": 0.6719,
      "step": 4925
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6282415444115741,
      "learning_rate": 9.581727830329634e-07,
      "loss": 0.662,
      "step": 4926
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7152091374304664,
      "learning_rate": 9.557480649377858e-07,
      "loss": 0.714,
      "step": 4927
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6898568713746176,
      "learning_rate": 9.533262647067609e-07,
      "loss": 0.703,
      "step": 4928
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.66937830641222,
      "learning_rate": 9.509073831212079e-07,
      "loss": 0.6748,
      "step": 4929
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6670575405327832,
      "learning_rate": 9.484914209615115e-07,
      "loss": 0.7026,
      "step": 4930
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6725675724624007,
      "learning_rate": 9.460783790071104e-07,
      "loss": 0.6979,
      "step": 4931
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6867284233537454,
      "learning_rate": 9.436682580365019e-07,
      "loss": 0.664,
      "step": 4932
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6081000426868497,
      "learning_rate": 9.412610588272408e-07,
      "loss": 0.6237,
      "step": 4933
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6326391699774856,
      "learning_rate": 9.388567821559424e-07,
      "loss": 0.6533,
      "step": 4934
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7079293461138848,
      "learning_rate": 9.36455428798273e-07,
      "loss": 0.7011,
      "step": 4935
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6700743055467192,
      "learning_rate": 9.340569995289617e-07,
      "loss": 0.7282,
      "step": 4936
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7183704799482746,
      "learning_rate": 9.316614951217895e-07,
      "loss": 0.7225,
      "step": 4937
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5819503186356885,
      "learning_rate": 9.292689163495983e-07,
      "loss": 0.6131,
      "step": 4938
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7305589742474686,
      "learning_rate": 9.268792639842839e-07,
      "loss": 0.7433,
      "step": 4939
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6585060499830437,
      "learning_rate": 9.244925387967974e-07,
      "loss": 0.6692,
      "step": 4940
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.654344126377863,
      "learning_rate": 9.221087415571439e-07,
      "loss": 0.6834,
      "step": 4941
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6827085163890979,
      "learning_rate": 9.197278730343883e-07,
      "loss": 0.7078,
      "step": 4942
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6279221848021753,
      "learning_rate": 9.173499339966474e-07,
      "loss": 0.6978,
      "step": 4943
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.651305249034445,
      "learning_rate": 9.149749252110951e-07,
      "loss": 0.6614,
      "step": 4944
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6242960767098168,
      "learning_rate": 9.126028474439596e-07,
      "loss": 0.7129,
      "step": 4945
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7669378386924466,
      "learning_rate": 9.102337014605189e-07,
      "loss": 0.7215,
      "step": 4946
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7077178233648591,
      "learning_rate": 9.078674880251126e-07,
      "loss": 0.6944,
      "step": 4947
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.65322175552763,
      "learning_rate": 9.055042079011278e-07,
      "loss": 0.7309,
      "step": 4948
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6727162987913881,
      "learning_rate": 9.031438618510102e-07,
      "loss": 0.6777,
      "step": 4949
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7392313781440198,
      "learning_rate": 9.007864506362551e-07,
      "loss": 0.7457,
      "step": 4950
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6974897084031096,
      "learning_rate": 8.984319750174119e-07,
      "loss": 0.711,
      "step": 4951
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7180106608895278,
      "learning_rate": 8.960804357540864e-07,
      "loss": 0.6579,
      "step": 4952
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6963383001832163,
      "learning_rate": 8.937318336049316e-07,
      "loss": 0.747,
      "step": 4953
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6425405594551552,
      "learning_rate": 8.913861693276549e-07,
      "loss": 0.6631,
      "step": 4954
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6950185860053312,
      "learning_rate": 8.890434436790185e-07,
      "loss": 0.7208,
      "step": 4955
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6164508383941375,
      "learning_rate": 8.867036574148336e-07,
      "loss": 0.6705,
      "step": 4956
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.743454642797909,
      "learning_rate": 8.84366811289965e-07,
      "loss": 0.6694,
      "step": 4957
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6719361682814701,
      "learning_rate": 8.820329060583277e-07,
      "loss": 0.7026,
      "step": 4958
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6765754848455626,
      "learning_rate": 8.797019424728859e-07,
      "loss": 0.672,
      "step": 4959
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5803676169275447,
      "learning_rate": 8.773739212856591e-07,
      "loss": 0.6588,
      "step": 4960
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5988240144149433,
      "learning_rate": 8.750488432477144e-07,
      "loss": 0.7321,
      "step": 4961
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7003342176413121,
      "learning_rate": 8.727267091091718e-07,
      "loss": 0.705,
      "step": 4962
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7470275192521137,
      "learning_rate": 8.704075196192008e-07,
      "loss": 0.7543,
      "step": 4963
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6503006229592941,
      "learning_rate": 8.680912755260173e-07,
      "loss": 0.6829,
      "step": 4964
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.575767601170546,
      "learning_rate": 8.657779775768916e-07,
      "loss": 0.6824,
      "step": 4965
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6444040568237167,
      "learning_rate": 8.634676265181408e-07,
      "loss": 0.7183,
      "step": 4966
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7320857456073832,
      "learning_rate": 8.611602230951343e-07,
      "loss": 0.6834,
      "step": 4967
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5971103490360162,
      "learning_rate": 8.588557680522891e-07,
      "loss": 0.6437,
      "step": 4968
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7093362353895313,
      "learning_rate": 8.56554262133068e-07,
      "loss": 0.7363,
      "step": 4969
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.760359424397373,
      "learning_rate": 8.54255706079985e-07,
      "loss": 0.7054,
      "step": 4970
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7878873664543418,
      "learning_rate": 8.519601006346023e-07,
      "loss": 0.7341,
      "step": 4971
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6340844395406424,
      "learning_rate": 8.496674465375298e-07,
      "loss": 0.6801,
      "step": 4972
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6443058047282542,
      "learning_rate": 8.47377744528427e-07,
      "loss": 0.7202,
      "step": 4973
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6470038910556581,
      "learning_rate": 8.450909953460007e-07,
      "loss": 0.693,
      "step": 4974
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6092907125581103,
      "learning_rate": 8.428071997279996e-07,
      "loss": 0.6913,
      "step": 4975
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.604723964796157,
      "learning_rate": 8.405263584112255e-07,
      "loss": 0.6734,
      "step": 4976
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7352107107078331,
      "learning_rate": 8.382484721315287e-07,
      "loss": 0.7393,
      "step": 4977
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7497338415006836,
      "learning_rate": 8.359735416237969e-07,
      "loss": 0.7511,
      "step": 4978
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6957474226022387,
      "learning_rate": 8.33701567621974e-07,
      "loss": 0.6757,
      "step": 4979
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.661691606185396,
      "learning_rate": 8.314325508590448e-07,
      "loss": 0.7141,
      "step": 4980
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7266838791839944,
      "learning_rate": 8.29166492067044e-07,
      "loss": 0.7399,
      "step": 4981
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7360360146550868,
      "learning_rate": 8.269033919770453e-07,
      "loss": 0.7139,
      "step": 4982
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6958502376252015,
      "learning_rate": 8.246432513191738e-07,
      "loss": 0.6809,
      "step": 4983
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6337509386331998,
      "learning_rate": 8.223860708225984e-07,
      "loss": 0.6636,
      "step": 4984
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6360416764916568,
      "learning_rate": 8.201318512155309e-07,
      "loss": 0.7108,
      "step": 4985
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6362319851107495,
      "learning_rate": 8.178805932252343e-07,
      "loss": 0.6461,
      "step": 4986
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6175928782039724,
      "learning_rate": 8.156322975780074e-07,
      "loss": 0.6969,
      "step": 4987
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6961194457189426,
      "learning_rate": 8.133869649991955e-07,
      "loss": 0.7148,
      "step": 4988
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7301540119320312,
      "learning_rate": 8.111445962131925e-07,
      "loss": 0.7487,
      "step": 4989
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6814505828064321,
      "learning_rate": 8.089051919434321e-07,
      "loss": 0.7686,
      "step": 4990
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6365516967484058,
      "learning_rate": 8.06668752912394e-07,
      "loss": 0.7133,
      "step": 4991
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6416760429157039,
      "learning_rate": 8.044352798416e-07,
      "loss": 0.6718,
      "step": 4992
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8188482310171137,
      "learning_rate": 8.022047734516136e-07,
      "loss": 0.6964,
      "step": 4993
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6287229916305674,
      "learning_rate": 7.999772344620437e-07,
      "loss": 0.6703,
      "step": 4994
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6997873787868881,
      "learning_rate": 7.977526635915422e-07,
      "loss": 0.6885,
      "step": 4995
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6669456989479913,
      "learning_rate": 7.955310615577993e-07,
      "loss": 0.7048,
      "step": 4996
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6438112751282876,
      "learning_rate": 7.933124290775507e-07,
      "loss": 0.6808,
      "step": 4997
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8587816340431749,
      "learning_rate": 7.910967668665759e-07,
      "loss": 0.6852,
      "step": 4998
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6317064849341534,
      "learning_rate": 7.888840756396898e-07,
      "loss": 0.6596,
      "step": 4999
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6015066886496627,
      "learning_rate": 7.866743561107548e-07,
      "loss": 0.6877,
      "step": 5000
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.674861003689793,
      "learning_rate": 7.844676089926717e-07,
      "loss": 0.6968,
      "step": 5001
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6793583441507539,
      "learning_rate": 7.822638349973854e-07,
      "loss": 0.71,
      "step": 5002
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6644660449743555,
      "learning_rate": 7.800630348358762e-07,
      "loss": 0.6693,
      "step": 5003
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6819194534321793,
      "learning_rate": 7.778652092181726e-07,
      "loss": 0.7049,
      "step": 5004
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.690266658968523,
      "learning_rate": 7.756703588533354e-07,
      "loss": 0.7038,
      "step": 5005
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6283534518482208,
      "learning_rate": 7.73478484449468e-07,
      "loss": 0.6594,
      "step": 5006
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7331245319551555,
      "learning_rate": 7.71289586713716e-07,
      "loss": 0.6832,
      "step": 5007
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6857895975580812,
      "learning_rate": 7.691036663522644e-07,
      "loss": 0.6985,
      "step": 5008
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6127390387489915,
      "learning_rate": 7.669207240703358e-07,
      "loss": 0.6782,
      "step": 5009
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7784603535718949,
      "learning_rate": 7.647407605721946e-07,
      "loss": 0.7483,
      "step": 5010
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5912844104252837,
      "learning_rate": 7.625637765611405e-07,
      "loss": 0.652,
      "step": 5011
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7349847663480231,
      "learning_rate": 7.603897727395126e-07,
      "loss": 0.7003,
      "step": 5012
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6891271266395798,
      "learning_rate": 7.582187498086946e-07,
      "loss": 0.702,
      "step": 5013
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8168526669212026,
      "learning_rate": 7.560507084690993e-07,
      "loss": 0.6965,
      "step": 5014
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6420843324525556,
      "learning_rate": 7.538856494201829e-07,
      "loss": 0.6903,
      "step": 5015
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6213623807744865,
      "learning_rate": 7.517235733604411e-07,
      "loss": 0.6581,
      "step": 5016
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6468048025760801,
      "learning_rate": 7.495644809874003e-07,
      "loss": 0.6783,
      "step": 5017
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6911198844936937,
      "learning_rate": 7.474083729976322e-07,
      "loss": 0.743,
      "step": 5018
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7650669374360607,
      "learning_rate": 7.452552500867405e-07,
      "loss": 0.6914,
      "step": 5019
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6693946687617682,
      "learning_rate": 7.431051129493682e-07,
      "loss": 0.7089,
      "step": 5020
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6850510108908252,
      "learning_rate": 7.409579622791963e-07,
      "loss": 0.6916,
      "step": 5021
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6579699658046446,
      "learning_rate": 7.388137987689381e-07,
      "loss": 0.6789,
      "step": 5022
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6585798454226429,
      "learning_rate": 7.36672623110345e-07,
      "loss": 0.7125,
      "step": 5023
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.689280138203555,
      "learning_rate": 7.345344359942042e-07,
      "loss": 0.7015,
      "step": 5024
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6218897692758315,
      "learning_rate": 7.323992381103418e-07,
      "loss": 0.6815,
      "step": 5025
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6319867259078529,
      "learning_rate": 7.302670301476156e-07,
      "loss": 0.6811,
      "step": 5026
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7300895307258279,
      "learning_rate": 7.281378127939221e-07,
      "loss": 0.7306,
      "step": 5027
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6795255653830686,
      "learning_rate": 7.260115867361895e-07,
      "loss": 0.7037,
      "step": 5028
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7026362749685073,
      "learning_rate": 7.238883526603824e-07,
      "loss": 0.6886,
      "step": 5029
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6550621318557394,
      "learning_rate": 7.217681112515007e-07,
      "loss": 0.678,
      "step": 5030
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6508997380871702,
      "learning_rate": 7.196508631935794e-07,
      "loss": 0.6653,
      "step": 5031
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6597464981857799,
      "learning_rate": 7.175366091696878e-07,
      "loss": 0.6617,
      "step": 5032
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7563859259562059,
      "learning_rate": 7.154253498619245e-07,
      "loss": 0.7065,
      "step": 5033
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7032092104059681,
      "learning_rate": 7.133170859514304e-07,
      "loss": 0.7356,
      "step": 5034
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7490460746374362,
      "learning_rate": 7.112118181183703e-07,
      "loss": 0.7467,
      "step": 5035
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6232538863714465,
      "learning_rate": 7.091095470419507e-07,
      "loss": 0.6755,
      "step": 5036
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8185962982812571,
      "learning_rate": 7.070102734004081e-07,
      "loss": 0.7712,
      "step": 5037
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6813902178345911,
      "learning_rate": 7.049139978710106e-07,
      "loss": 0.7107,
      "step": 5038
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.790868320499998,
      "learning_rate": 7.02820721130063e-07,
      "loss": 0.733,
      "step": 5039
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6523960365361848,
      "learning_rate": 7.007304438528972e-07,
      "loss": 0.6779,
      "step": 5040
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6174834521609853,
      "learning_rate": 6.986431667138815e-07,
      "loss": 0.6582,
      "step": 5041
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6412265069615184,
      "learning_rate": 6.96558890386414e-07,
      "loss": 0.6567,
      "step": 5042
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6748660429099712,
      "learning_rate": 6.944776155429256e-07,
      "loss": 0.6969,
      "step": 5043
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7161081286402309,
      "learning_rate": 6.923993428548803e-07,
      "loss": 0.6607,
      "step": 5044
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6749139834439424,
      "learning_rate": 6.90324072992773e-07,
      "loss": 0.6671,
      "step": 5045
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6561689170321549,
      "learning_rate": 6.882518066261246e-07,
      "loss": 0.6971,
      "step": 5046
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6576672327446043,
      "learning_rate": 6.861825444234959e-07,
      "loss": 0.7039,
      "step": 5047
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7029297470969077,
      "learning_rate": 6.84116287052472e-07,
      "loss": 0.7253,
      "step": 5048
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6530742469741634,
      "learning_rate": 6.820530351796706e-07,
      "loss": 0.6683,
      "step": 5049
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6687220681764383,
      "learning_rate": 6.799927894707425e-07,
      "loss": 0.6849,
      "step": 5050
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6815666785676069,
      "learning_rate": 6.779355505903629e-07,
      "loss": 0.6919,
      "step": 5051
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.634859378758188,
      "learning_rate": 6.758813192022396e-07,
      "loss": 0.6951,
      "step": 5052
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7107762500206927,
      "learning_rate": 6.738300959691113e-07,
      "loss": 0.717,
      "step": 5053
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6702180215045738,
      "learning_rate": 6.717818815527455e-07,
      "loss": 0.68,
      "step": 5054
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6231993828713926,
      "learning_rate": 6.697366766139379e-07,
      "loss": 0.6756,
      "step": 5055
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5935107664664829,
      "learning_rate": 6.676944818125187e-07,
      "loss": 0.6617,
      "step": 5056
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6405271505145459,
      "learning_rate": 6.656552978073361e-07,
      "loss": 0.6688,
      "step": 5057
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6994619209057904,
      "learning_rate": 6.636191252562763e-07,
      "loss": 0.7448,
      "step": 5058
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7516795661932462,
      "learning_rate": 6.615859648162526e-07,
      "loss": 0.6506,
      "step": 5059
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6756681774722314,
      "learning_rate": 6.595558171432026e-07,
      "loss": 0.7289,
      "step": 5060
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6736139287547559,
      "learning_rate": 6.575286828920946e-07,
      "loss": 0.7355,
      "step": 5061
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6554073721624437,
      "learning_rate": 6.555045627169243e-07,
      "loss": 0.6542,
      "step": 5062
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7239220047047097,
      "learning_rate": 6.534834572707171e-07,
      "loss": 0.7516,
      "step": 5063
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6742956254252193,
      "learning_rate": 6.514653672055204e-07,
      "loss": 0.684,
      "step": 5064
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6124987834341231,
      "learning_rate": 6.494502931724133e-07,
      "loss": 0.6587,
      "step": 5065
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6668151705372639,
      "learning_rate": 6.474382358215026e-07,
      "loss": 0.7186,
      "step": 5066
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6530610777064729,
      "learning_rate": 6.454291958019177e-07,
      "loss": 0.7063,
      "step": 5067
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6780831052040899,
      "learning_rate": 6.43423173761819e-07,
      "loss": 0.6937,
      "step": 5068
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7440453330074065,
      "learning_rate": 6.414201703483902e-07,
      "loss": 0.7241,
      "step": 5069
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7084567301809934,
      "learning_rate": 6.394201862078386e-07,
      "loss": 0.7129,
      "step": 5070
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2136770736404845,
      "learning_rate": 6.374232219854049e-07,
      "loss": 0.7095,
      "step": 5071
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6289324583974081,
      "learning_rate": 6.354292783253491e-07,
      "loss": 0.6719,
      "step": 5072
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6502456955253032,
      "learning_rate": 6.3343835587096e-07,
      "loss": 0.6768,
      "step": 5073
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6885054738152341,
      "learning_rate": 6.314504552645528e-07,
      "loss": 0.6868,
      "step": 5074
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6309001435188996,
      "learning_rate": 6.294655771474612e-07,
      "loss": 0.6634,
      "step": 5075
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6972171575520929,
      "learning_rate": 6.274837221600505e-07,
      "loss": 0.7004,
      "step": 5076
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7207801470750851,
      "learning_rate": 6.255048909417095e-07,
      "loss": 0.6888,
      "step": 5077
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.646618066970546,
      "learning_rate": 6.235290841308484e-07,
      "loss": 0.686,
      "step": 5078
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6299010903849748,
      "learning_rate": 6.21556302364904e-07,
      "loss": 0.6882,
      "step": 5079
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6465090880320002,
      "learning_rate": 6.195865462803385e-07,
      "loss": 0.6973,
      "step": 5080
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6700829905553866,
      "learning_rate": 6.176198165126335e-07,
      "loss": 0.6702,
      "step": 5081
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6413454667334564,
      "learning_rate": 6.156561136962991e-07,
      "loss": 0.6962,
      "step": 5082
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6434230996084168,
      "learning_rate": 6.136954384648663e-07,
      "loss": 0.6576,
      "step": 5083
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6536796804568686,
      "learning_rate": 6.117377914508904e-07,
      "loss": 0.714,
      "step": 5084
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7673495322704128,
      "learning_rate": 6.097831732859471e-07,
      "loss": 0.6769,
      "step": 5085
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.739246325077026,
      "learning_rate": 6.0783158460064e-07,
      "loss": 0.7268,
      "step": 5086
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6795366992985777,
      "learning_rate": 6.058830260245907e-07,
      "loss": 0.6985,
      "step": 5087
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6898973469515431,
      "learning_rate": 6.039374981864444e-07,
      "loss": 0.7415,
      "step": 5088
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6509688542476593,
      "learning_rate": 6.019950017138676e-07,
      "loss": 0.65,
      "step": 5089
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6719929573045945,
      "learning_rate": 6.000555372335526e-07,
      "loss": 0.6936,
      "step": 5090
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6207784052783373,
      "learning_rate": 5.981191053712099e-07,
      "loss": 0.7084,
      "step": 5091
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.726206580555055,
      "learning_rate": 5.961857067515742e-07,
      "loss": 0.7168,
      "step": 5092
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.754484494162379,
      "learning_rate": 5.942553419983976e-07,
      "loss": 0.7478,
      "step": 5093
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5914355769596401,
      "learning_rate": 5.923280117344576e-07,
      "loss": 0.6603,
      "step": 5094
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6573800841266276,
      "learning_rate": 5.904037165815512e-07,
      "loss": 0.6949,
      "step": 5095
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6311569146933668,
      "learning_rate": 5.884824571604974e-07,
      "loss": 0.6691,
      "step": 5096
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5824285889392142,
      "learning_rate": 5.865642340911315e-07,
      "loss": 0.6474,
      "step": 5097
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7083818226222938,
      "learning_rate": 5.84649047992315e-07,
      "loss": 0.7301,
      "step": 5098
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.637460429033412,
      "learning_rate": 5.827368994819238e-07,
      "loss": 0.6727,
      "step": 5099
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6485897681177597,
      "learning_rate": 5.808277891768599e-07,
      "loss": 0.6686,
      "step": 5100
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6506620716324071,
      "learning_rate": 5.789217176930395e-07,
      "loss": 0.6908,
      "step": 5101
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7115008245976694,
      "learning_rate": 5.77018685645403e-07,
      "loss": 0.7311,
      "step": 5102
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6594221354215759,
      "learning_rate": 5.75118693647908e-07,
      "loss": 0.717,
      "step": 5103
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5981700061434727,
      "learning_rate": 5.732217423135311e-07,
      "loss": 0.6466,
      "step": 5104
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5888191344605326,
      "learning_rate": 5.713278322542682e-07,
      "loss": 0.6598,
      "step": 5105
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6465778613998194,
      "learning_rate": 5.694369640811337e-07,
      "loss": 0.7198,
      "step": 5106
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6744947124514591,
      "learning_rate": 5.675491384041609e-07,
      "loss": 0.6806,
      "step": 5107
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6592180337186977,
      "learning_rate": 5.656643558324027e-07,
      "loss": 0.7559,
      "step": 5108
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6979195257071743,
      "learning_rate": 5.637826169739302e-07,
      "loss": 0.6944,
      "step": 5109
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6374034908030505,
      "learning_rate": 5.619039224358292e-07,
      "loss": 0.6562,
      "step": 5110
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.677400742130099,
      "learning_rate": 5.600282728242068e-07,
      "loss": 0.7285,
      "step": 5111
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6055363993626175,
      "learning_rate": 5.581556687441858e-07,
      "loss": 0.6232,
      "step": 5112
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.695714665624665,
      "learning_rate": 5.562861107999095e-07,
      "loss": 0.6899,
      "step": 5113
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6554539776657197,
      "learning_rate": 5.544195995945356e-07,
      "loss": 0.6784,
      "step": 5114
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6219620220755192,
      "learning_rate": 5.525561357302367e-07,
      "loss": 0.6436,
      "step": 5115
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6207397384671143,
      "learning_rate": 5.506957198082097e-07,
      "loss": 0.671,
      "step": 5116
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6137230544312345,
      "learning_rate": 5.48838352428659e-07,
      "loss": 0.6861,
      "step": 5117
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6155464331126521,
      "learning_rate": 5.469840341908117e-07,
      "loss": 0.6604,
      "step": 5118
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6512910762818657,
      "learning_rate": 5.451327656929106e-07,
      "loss": 0.7374,
      "step": 5119
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6825944988528989,
      "learning_rate": 5.432845475322124e-07,
      "loss": 0.7058,
      "step": 5120
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6514844659159983,
      "learning_rate": 5.414393803049933e-07,
      "loss": 0.6955,
      "step": 5121
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6045373401025126,
      "learning_rate": 5.395972646065384e-07,
      "loss": 0.6904,
      "step": 5122
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6205594939387041,
      "learning_rate": 5.377582010311555e-07,
      "loss": 0.6511,
      "step": 5123
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5955567285792948,
      "learning_rate": 5.359221901721634e-07,
      "loss": 0.6821,
      "step": 5124
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.685013666137852,
      "learning_rate": 5.340892326218983e-07,
      "loss": 0.7158,
      "step": 5125
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6573819755339418,
      "learning_rate": 5.322593289717104e-07,
      "loss": 0.6716,
      "step": 5126
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6146991542725002,
      "learning_rate": 5.304324798119664e-07,
      "loss": 0.6356,
      "step": 5127
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7014246174358304,
      "learning_rate": 5.286086857320416e-07,
      "loss": 0.726,
      "step": 5128
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6482041301817434,
      "learning_rate": 5.267879473203341e-07,
      "loss": 0.6923,
      "step": 5129
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0787217985642683,
      "learning_rate": 5.249702651642496e-07,
      "loss": 0.7384,
      "step": 5130
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6411579142880084,
      "learning_rate": 5.231556398502114e-07,
      "loss": 0.6951,
      "step": 5131
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7008672894963947,
      "learning_rate": 5.213440719636564e-07,
      "loss": 0.6581,
      "step": 5132
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.647968880896426,
      "learning_rate": 5.19535562089033e-07,
      "loss": 0.6948,
      "step": 5133
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7002244761877433,
      "learning_rate": 5.177301108098032e-07,
      "loss": 0.6671,
      "step": 5134
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7044386259084227,
      "learning_rate": 5.159277187084433e-07,
      "loss": 0.6684,
      "step": 5135
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6434325530029915,
      "learning_rate": 5.141283863664448e-07,
      "loss": 0.6729,
      "step": 5136
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6662072696449306,
      "learning_rate": 5.123321143643079e-07,
      "loss": 0.7032,
      "step": 5137
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7617014800341856,
      "learning_rate": 5.105389032815489e-07,
      "loss": 0.6729,
      "step": 5138
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6758039820760449,
      "learning_rate": 5.087487536966929e-07,
      "loss": 0.7261,
      "step": 5139
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6270496875171772,
      "learning_rate": 5.069616661872801e-07,
      "loss": 0.6955,
      "step": 5140
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6307189806645389,
      "learning_rate": 5.051776413298647e-07,
      "loss": 0.6789,
      "step": 5141
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6708695306458515,
      "learning_rate": 5.033966797000067e-07,
      "loss": 0.7052,
      "step": 5142
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6479409782448884,
      "learning_rate": 5.016187818722828e-07,
      "loss": 0.6912,
      "step": 5143
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6886477251455617,
      "learning_rate": 4.998439484202811e-07,
      "loss": 0.7473,
      "step": 5144
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.660220542492386,
      "learning_rate": 4.980721799166e-07,
      "loss": 0.7057,
      "step": 5145
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6995097048182796,
      "learning_rate": 4.963034769328457e-07,
      "loss": 0.7054,
      "step": 5146
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6849366784038846,
      "learning_rate": 4.945378400396417e-07,
      "loss": 0.7019,
      "step": 5147
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6924835898595314,
      "learning_rate": 4.92775269806618e-07,
      "loss": 0.7185,
      "step": 5148
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6703018196940217,
      "learning_rate": 4.910157668024163e-07,
      "loss": 0.7081,
      "step": 5149
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7013626617543578,
      "learning_rate": 4.892593315946914e-07,
      "loss": 0.6906,
      "step": 5150
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.712272925582211,
      "learning_rate": 4.875059647501035e-07,
      "loss": 0.7451,
      "step": 5151
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6776426477111472,
      "learning_rate": 4.857556668343244e-07,
      "loss": 0.7219,
      "step": 5152
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6264568186206098,
      "learning_rate": 4.840084384120381e-07,
      "loss": 0.6687,
      "step": 5153
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.683073728630025,
      "learning_rate": 4.822642800469368e-07,
      "loss": 0.7174,
      "step": 5154
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6648630201969823,
      "learning_rate": 4.805231923017228e-07,
      "loss": 0.6948,
      "step": 5155
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6086117454457106,
      "learning_rate": 4.787851757381079e-07,
      "loss": 0.6847,
      "step": 5156
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6066989237314606,
      "learning_rate": 4.770502309168101e-07,
      "loss": 0.6848,
      "step": 5157
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7002998834789402,
      "learning_rate": 4.753183583975596e-07,
      "loss": 0.6657,
      "step": 5158
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7186695641048381,
      "learning_rate": 4.7358955873909594e-07,
      "loss": 0.7221,
      "step": 5159
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7425682672701384,
      "learning_rate": 4.7186383249916644e-07,
      "loss": 0.7188,
      "step": 5160
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.633153837010472,
      "learning_rate": 4.701411802345224e-07,
      "loss": 0.6948,
      "step": 5161
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6985049134148822,
      "learning_rate": 4.6842160250093136e-07,
      "loss": 0.675,
      "step": 5162
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7165681989100472,
      "learning_rate": 4.66705099853163e-07,
      "loss": 0.7287,
      "step": 5163
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6043812998346797,
      "learning_rate": 4.649916728449966e-07,
      "loss": 0.6715,
      "step": 5164
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6894585855287751,
      "learning_rate": 4.632813220292209e-07,
      "loss": 0.7276,
      "step": 5165
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6826489935900882,
      "learning_rate": 4.615740479576292e-07,
      "loss": 0.7303,
      "step": 5166
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7161368948875177,
      "learning_rate": 4.598698511810251e-07,
      "loss": 0.7011,
      "step": 5167
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7156045924335728,
      "learning_rate": 4.5816873224921677e-07,
      "loss": 0.7343,
      "step": 5168
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7332992843225806,
      "learning_rate": 4.564706917110229e-07,
      "loss": 0.7445,
      "step": 5169
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7348136225620779,
      "learning_rate": 4.54775730114263e-07,
      "loss": 0.7537,
      "step": 5170
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5480062057646846,
      "learning_rate": 4.530838480057698e-07,
      "loss": 0.7118,
      "step": 5171
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6523418058712791,
      "learning_rate": 4.5139504593137763e-07,
      "loss": 0.6456,
      "step": 5172
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7834747684416298,
      "learning_rate": 4.4970932443593186e-07,
      "loss": 0.7523,
      "step": 5173
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6542506228968274,
      "learning_rate": 4.480266840632808e-07,
      "loss": 0.6769,
      "step": 5174
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6853289627902458,
      "learning_rate": 4.463471253562768e-07,
      "loss": 0.6848,
      "step": 5175
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6866860891092332,
      "learning_rate": 4.4467064885678314e-07,
      "loss": 0.7048,
      "step": 5176
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6989294718167838,
      "learning_rate": 4.4299725510566384e-07,
      "loss": 0.7425,
      "step": 5177
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6384434109611873,
      "learning_rate": 4.413269446427948e-07,
      "loss": 0.701,
      "step": 5178
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7051692230245222,
      "learning_rate": 4.396597180070483e-07,
      "loss": 0.7261,
      "step": 5179
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8007181539628061,
      "learning_rate": 4.3799557573630856e-07,
      "loss": 0.6817,
      "step": 5180
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6072024831699643,
      "learning_rate": 4.3633451836746165e-07,
      "loss": 0.7076,
      "step": 5181
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5812975762567614,
      "learning_rate": 4.346765464364e-07,
      "loss": 0.6403,
      "step": 5182
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6662109973937782,
      "learning_rate": 4.3302166047802017e-07,
      "loss": 0.6847,
      "step": 5183
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7486717173436629,
      "learning_rate": 4.313698610262218e-07,
      "loss": 0.7419,
      "step": 5184
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6717975219847047,
      "learning_rate": 4.297211486139119e-07,
      "loss": 0.6907,
      "step": 5185
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6344014498892879,
      "learning_rate": 4.280755237729972e-07,
      "loss": 0.6689,
      "step": 5186
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7123550541200384,
      "learning_rate": 4.2643298703439194e-07,
      "loss": 0.7139,
      "step": 5187
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6754168986527378,
      "learning_rate": 4.247935389280111e-07,
      "loss": 0.6766,
      "step": 5188
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6807353445148655,
      "learning_rate": 4.2315717998277603e-07,
      "loss": 0.7086,
      "step": 5189
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6591716759067102,
      "learning_rate": 4.2152391072661004e-07,
      "loss": 0.6546,
      "step": 5190
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5401367320637133,
      "learning_rate": 4.1989373168644044e-07,
      "loss": 0.6599,
      "step": 5191
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6110764584726533,
      "learning_rate": 4.182666433881966e-07,
      "loss": 0.7021,
      "step": 5192
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6387816023186227,
      "learning_rate": 4.1664264635680963e-07,
      "loss": 0.7153,
      "step": 5193
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6777775344469593,
      "learning_rate": 4.150217411162161e-07,
      "loss": 0.6722,
      "step": 5194
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7138748979699672,
      "learning_rate": 4.1340392818935426e-07,
      "loss": 0.6869,
      "step": 5195
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7099616178875814,
      "learning_rate": 4.117892080981645e-07,
      "loss": 0.6729,
      "step": 5196
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6157713215056707,
      "learning_rate": 4.101775813635878e-07,
      "loss": 0.6629,
      "step": 5197
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6788749604351574,
      "learning_rate": 4.0856904850557046e-07,
      "loss": 0.6719,
      "step": 5198
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6617874973468545,
      "learning_rate": 4.069636100430563e-07,
      "loss": 0.665,
      "step": 5199
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6957857675305978,
      "learning_rate": 4.053612664939932e-07,
      "loss": 0.6883,
      "step": 5200
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6553594419359362,
      "learning_rate": 4.0376201837533324e-07,
      "loss": 0.723,
      "step": 5201
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7416871886745575,
      "learning_rate": 4.021658662030248e-07,
      "loss": 0.7232,
      "step": 5202
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6774858073056813,
      "learning_rate": 4.0057281049202256e-07,
      "loss": 0.7368,
      "step": 5203
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6477576998527672,
      "learning_rate": 3.989828517562755e-07,
      "loss": 0.6998,
      "step": 5204
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6540415571219923,
      "learning_rate": 3.97395990508741e-07,
      "loss": 0.6547,
      "step": 5205
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6868610030398729,
      "learning_rate": 3.958122272613718e-07,
      "loss": 0.7079,
      "step": 5206
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6859186166305641,
      "learning_rate": 3.9423156252512253e-07,
      "loss": 0.6915,
      "step": 5207
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6396855913425539,
      "learning_rate": 3.9265399680994964e-07,
      "loss": 0.6819,
      "step": 5208
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7183144672540949,
      "learning_rate": 3.9107953062480943e-07,
      "loss": 0.6857,
      "step": 5209
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6849160774059312,
      "learning_rate": 3.895081644776555e-07,
      "loss": 0.7022,
      "step": 5210
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7705056837280294,
      "learning_rate": 3.8793989887544457e-07,
      "loss": 0.7018,
      "step": 5211
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.644628310550631,
      "learning_rate": 3.863747343241309e-07,
      "loss": 0.6877,
      "step": 5212
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7169902863498977,
      "learning_rate": 3.848126713286704e-07,
      "loss": 0.6839,
      "step": 5213
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6592512491058847,
      "learning_rate": 3.8325371039301673e-07,
      "loss": 0.6929,
      "step": 5214
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6973541202685539,
      "learning_rate": 3.816978520201242e-07,
      "loss": 0.6948,
      "step": 5215
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6416674405348961,
      "learning_rate": 3.8014509671194244e-07,
      "loss": 0.689,
      "step": 5216
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7522864608334948,
      "learning_rate": 3.785954449694229e-07,
      "loss": 0.6881,
      "step": 5217
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6677009854521466,
      "learning_rate": 3.7704889729251793e-07,
      "loss": 0.6844,
      "step": 5218
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.683257439476345,
      "learning_rate": 3.7550545418017506e-07,
      "loss": 0.6886,
      "step": 5219
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6302192664606404,
      "learning_rate": 3.7396511613034257e-07,
      "loss": 0.6569,
      "step": 5220
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.612703695242606,
      "learning_rate": 3.72427883639962e-07,
      "loss": 0.685,
      "step": 5221
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6469306262580434,
      "learning_rate": 3.708937572049798e-07,
      "loss": 0.7061,
      "step": 5222
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.706483412652573,
      "learning_rate": 3.6936273732033457e-07,
      "loss": 0.6335,
      "step": 5223
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6539203973247423,
      "learning_rate": 3.67834824479969e-07,
      "loss": 0.6623,
      "step": 5224
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7103488193525133,
      "learning_rate": 3.663100191768154e-07,
      "loss": 0.6741,
      "step": 5225
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.692528359129227,
      "learning_rate": 3.647883219028103e-07,
      "loss": 0.6832,
      "step": 5226
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6574266524496459,
      "learning_rate": 3.6326973314888434e-07,
      "loss": 0.6724,
      "step": 5227
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6692363061069107,
      "learning_rate": 3.617542534049656e-07,
      "loss": 0.692,
      "step": 5228
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7678900005730859,
      "learning_rate": 3.6024188315997965e-07,
      "loss": 0.7358,
      "step": 5229
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7106098651792374,
      "learning_rate": 3.587326229018484e-07,
      "loss": 0.7589,
      "step": 5230
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6227244609508701,
      "learning_rate": 3.5722647311749125e-07,
      "loss": 0.682,
      "step": 5231
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7824023646770196,
      "learning_rate": 3.55723434292824e-07,
      "loss": 0.7245,
      "step": 5232
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7013794297893006,
      "learning_rate": 3.5422350691275753e-07,
      "loss": 0.711,
      "step": 5233
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7097198619140711,
      "learning_rate": 3.527266914611993e-07,
      "loss": 0.7428,
      "step": 5234
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7449944335743958,
      "learning_rate": 3.512329884210519e-07,
      "loss": 0.7542,
      "step": 5235
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6226455816439255,
      "learning_rate": 3.497423982742176e-07,
      "loss": 0.6724,
      "step": 5236
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7103899908897485,
      "learning_rate": 3.482549215015907e-07,
      "loss": 0.689,
      "step": 5237
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6705676852488501,
      "learning_rate": 3.4677055858306385e-07,
      "loss": 0.6887,
      "step": 5238
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5846940444342903,
      "learning_rate": 3.4528930999752076e-07,
      "loss": 0.6524,
      "step": 5239
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7691389422076164,
      "learning_rate": 3.4381117622284355e-07,
      "loss": 0.7631,
      "step": 5240
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.681226364765676,
      "learning_rate": 3.4233615773591076e-07,
      "loss": 0.7144,
      "step": 5241
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6270188033554106,
      "learning_rate": 3.4086425501259403e-07,
      "loss": 0.6777,
      "step": 5242
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6456951657008014,
      "learning_rate": 3.3939546852775786e-07,
      "loss": 0.7085,
      "step": 5243
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7481318028652012,
      "learning_rate": 3.379297987552654e-07,
      "loss": 0.7081,
      "step": 5244
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6182809254332401,
      "learning_rate": 3.3646724616796966e-07,
      "loss": 0.6993,
      "step": 5245
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7096123485870652,
      "learning_rate": 3.3500781123772306e-07,
      "loss": 0.7255,
      "step": 5246
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7202420635742646,
      "learning_rate": 3.3355149443536793e-07,
      "loss": 0.659,
      "step": 5247
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6581536902648724,
      "learning_rate": 3.32098296230744e-07,
      "loss": 0.7553,
      "step": 5248
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7180676656705357,
      "learning_rate": 3.306482170926839e-07,
      "loss": 0.6794,
      "step": 5249
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7400680434709849,
      "learning_rate": 3.2920125748901135e-07,
      "loss": 0.6716,
      "step": 5250
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6404393933978099,
      "learning_rate": 3.2775741788654615e-07,
      "loss": 0.6805,
      "step": 5251
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6761145932909408,
      "learning_rate": 3.263166987511013e-07,
      "loss": 0.6857,
      "step": 5252
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8376152129116157,
      "learning_rate": 3.248791005474827e-07,
      "loss": 0.6893,
      "step": 5253
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6924119433416068,
      "learning_rate": 3.234446237394906e-07,
      "loss": 0.7076,
      "step": 5254
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7389831551054181,
      "learning_rate": 3.2201326878991466e-07,
      "loss": 0.7235,
      "step": 5255
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6817327532606716,
      "learning_rate": 3.205850361605434e-07,
      "loss": 0.7145,
      "step": 5256
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6393101294005828,
      "learning_rate": 3.1915992631215144e-07,
      "loss": 0.6967,
      "step": 5257
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6768571971112933,
      "learning_rate": 3.1773793970450883e-07,
      "loss": 0.6952,
      "step": 5258
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6660046145749697,
      "learning_rate": 3.1631907679637974e-07,
      "loss": 0.7146,
      "step": 5259
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7052856200670561,
      "learning_rate": 3.149033380455191e-07,
      "loss": 0.6962,
      "step": 5260
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9351317968217558,
      "learning_rate": 3.1349072390867266e-07,
      "loss": 0.6885,
      "step": 5261
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7501271240724482,
      "learning_rate": 3.1208123484157917e-07,
      "loss": 0.6923,
      "step": 5262
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.685496738103531,
      "learning_rate": 3.1067487129896933e-07,
      "loss": 0.6942,
      "step": 5263
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6430846374193937,
      "learning_rate": 3.092716337345647e-07,
      "loss": 0.6651,
      "step": 5264
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6915082074520287,
      "learning_rate": 3.0787152260108087e-07,
      "loss": 0.7143,
      "step": 5265
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6992352943837561,
      "learning_rate": 3.0647453835022103e-07,
      "loss": 0.6895,
      "step": 5266
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6790396857372587,
      "learning_rate": 3.0508068143268233e-07,
      "loss": 0.7672,
      "step": 5267
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6845100528168951,
      "learning_rate": 3.036899522981518e-07,
      "loss": 0.6826,
      "step": 5268
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6351502589083022,
      "learning_rate": 3.0230235139530827e-07,
      "loss": 0.6539,
      "step": 5269
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6778826857539335,
      "learning_rate": 3.0091787917181926e-07,
      "loss": 0.6875,
      "step": 5270
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7324773858779275,
      "learning_rate": 2.9953653607434517e-07,
      "loss": 0.7058,
      "step": 5271
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6539123073544474,
      "learning_rate": 2.9815832254853517e-07,
      "loss": 0.6856,
      "step": 5272
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7001219068274559,
      "learning_rate": 2.967832390390324e-07,
      "loss": 0.7642,
      "step": 5273
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7642734457748329,
      "learning_rate": 2.954112859894653e-07,
      "loss": 0.7188,
      "step": 5274
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6075845371888403,
      "learning_rate": 2.940424638424533e-07,
      "loss": 0.6405,
      "step": 5275
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6402390428449879,
      "learning_rate": 2.9267677303960963e-07,
      "loss": 0.6734,
      "step": 5276
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6686897594984523,
      "learning_rate": 2.91314214021533e-07,
      "loss": 0.7084,
      "step": 5277
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6251316476479377,
      "learning_rate": 2.899547872278152e-07,
      "loss": 0.6874,
      "step": 5278
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6798760474295191,
      "learning_rate": 2.8859849309703315e-07,
      "loss": 0.7151,
      "step": 5279
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6650334631575656,
      "learning_rate": 2.8724533206675677e-07,
      "loss": 0.701,
      "step": 5280
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.702131813608893,
      "learning_rate": 2.8589530457354353e-07,
      "loss": 0.7193,
      "step": 5281
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6983975630175427,
      "learning_rate": 2.845484110529417e-07,
      "loss": 0.7145,
      "step": 5282
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6398987738577269,
      "learning_rate": 2.832046519394849e-07,
      "loss": 0.712,
      "step": 5283
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6704847961146372,
      "learning_rate": 2.8186402766670065e-07,
      "loss": 0.6952,
      "step": 5284
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.671989660040787,
      "learning_rate": 2.805265386671008e-07,
      "loss": 0.6901,
      "step": 5285
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6633470305546442,
      "learning_rate": 2.7919218537218685e-07,
      "loss": 0.697,
      "step": 5286
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6225718100093189,
      "learning_rate": 2.778609682124489e-07,
      "loss": 0.716,
      "step": 5287
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6665048818863435,
      "learning_rate": 2.7653288761736784e-07,
      "loss": 0.6707,
      "step": 5288
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7014681265910406,
      "learning_rate": 2.752079440154065e-07,
      "loss": 0.6996,
      "step": 5289
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6550481421094896,
      "learning_rate": 2.7388613783402076e-07,
      "loss": 0.677,
      "step": 5290
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8389082777614718,
      "learning_rate": 2.7256746949965387e-07,
      "loss": 0.7548,
      "step": 5291
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6934763032354532,
      "learning_rate": 2.7125193943773444e-07,
      "loss": 0.7413,
      "step": 5292
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6452204687243271,
      "learning_rate": 2.6993954807267966e-07,
      "loss": 0.6935,
      "step": 5293
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.737444988436998,
      "learning_rate": 2.6863029582789525e-07,
      "loss": 0.7009,
      "step": 5294
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7047145414071522,
      "learning_rate": 2.6732418312577335e-07,
      "loss": 0.7419,
      "step": 5295
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7484372836539037,
      "learning_rate": 2.660212103876936e-07,
      "loss": 0.7212,
      "step": 5296
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6375132055858875,
      "learning_rate": 2.6472137803402073e-07,
      "loss": 0.6678,
      "step": 5297
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5865498389069066,
      "learning_rate": 2.6342468648410836e-07,
      "loss": 0.6449,
      "step": 5298
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6060404082395981,
      "learning_rate": 2.6213113615629613e-07,
      "loss": 0.6438,
      "step": 5299
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6504453337053913,
      "learning_rate": 2.6084072746791036e-07,
      "loss": 0.7085,
      "step": 5300
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6519244725529003,
      "learning_rate": 2.5955346083526345e-07,
      "loss": 0.7107,
      "step": 5301
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7254910910257484,
      "learning_rate": 2.582693366736555e-07,
      "loss": 0.6977,
      "step": 5302
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7482281195507772,
      "learning_rate": 2.5698835539737067e-07,
      "loss": 0.7219,
      "step": 5303
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6593995265038727,
      "learning_rate": 2.5571051741968055e-07,
      "loss": 0.732,
      "step": 5304
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.751537277999001,
      "learning_rate": 2.544358231528421e-07,
      "loss": 0.7697,
      "step": 5305
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6862845617259445,
      "learning_rate": 2.5316427300809966e-07,
      "loss": 0.6862,
      "step": 5306
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6097667208210895,
      "learning_rate": 2.518958673956795e-07,
      "loss": 0.6268,
      "step": 5307
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6833456360297879,
      "learning_rate": 2.5063060672479655e-07,
      "loss": 0.699,
      "step": 5308
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5782768204817034,
      "learning_rate": 2.4936849140365206e-07,
      "loss": 0.6191,
      "step": 5309
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6637971556904797,
      "learning_rate": 2.4810952183942805e-07,
      "loss": 0.6851,
      "step": 5310
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7273650006681454,
      "learning_rate": 2.4685369843829634e-07,
      "loss": 0.7352,
      "step": 5311
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6031768480761882,
      "learning_rate": 2.456010216054094e-07,
      "loss": 0.64,
      "step": 5312
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.606653964808978,
      "learning_rate": 2.443514917449097e-07,
      "loss": 0.6733,
      "step": 5313
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7347555273782942,
      "learning_rate": 2.431051092599213e-07,
      "loss": 0.7543,
      "step": 5314
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6647883387509251,
      "learning_rate": 2.418618745525514e-07,
      "loss": 0.6779,
      "step": 5315
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.660705648251567,
      "learning_rate": 2.406217880238937e-07,
      "loss": 0.6957,
      "step": 5316
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5922955506474055,
      "learning_rate": 2.393848500740259e-07,
      "loss": 0.6774,
      "step": 5317
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7243654673491389,
      "learning_rate": 2.3815106110201102e-07,
      "loss": 0.7307,
      "step": 5318
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6826811242581564,
      "learning_rate": 2.36920421505894e-07,
      "loss": 0.6705,
      "step": 5319
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.603614780430483,
      "learning_rate": 2.3569293168270724e-07,
      "loss": 0.6114,
      "step": 5320
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6441609257005241,
      "learning_rate": 2.3446859202846172e-07,
      "loss": 0.7168,
      "step": 5321
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6678421180030345,
      "learning_rate": 2.3324740293815596e-07,
      "loss": 0.72,
      "step": 5322
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6536284544458797,
      "learning_rate": 2.320293648057703e-07,
      "loss": 0.6764,
      "step": 5323
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6504295918063964,
      "learning_rate": 2.308144780242705e-07,
      "loss": 0.6694,
      "step": 5324
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6132297149515094,
      "learning_rate": 2.2960274298560403e-07,
      "loss": 0.6488,
      "step": 5325
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6493062799834516,
      "learning_rate": 2.283941600807016e-07,
      "loss": 0.6869,
      "step": 5326
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6857849361560552,
      "learning_rate": 2.271887296994768e-07,
      "loss": 0.6872,
      "step": 5327
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7697241837234011,
      "learning_rate": 2.2598645223082637e-07,
      "loss": 0.7318,
      "step": 5328
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7033822407821443,
      "learning_rate": 2.2478732806263114e-07,
      "loss": 0.7024,
      "step": 5329
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6325303786563078,
      "learning_rate": 2.2359135758175388e-07,
      "loss": 0.6835,
      "step": 5330
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6570855136681409,
      "learning_rate": 2.2239854117403926e-07,
      "loss": 0.6284,
      "step": 5331
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7232400004704773,
      "learning_rate": 2.212088792243128e-07,
      "loss": 0.7578,
      "step": 5332
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6721978432503914,
      "learning_rate": 2.2002237211638633e-07,
      "loss": 0.6809,
      "step": 5333
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8368467320675895,
      "learning_rate": 2.188390202330515e-07,
      "loss": 0.7522,
      "step": 5334
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6833956234177393,
      "learning_rate": 2.1765882395608174e-07,
      "loss": 0.7293,
      "step": 5335
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.675077428549778,
      "learning_rate": 2.1648178366623252e-07,
      "loss": 0.6758,
      "step": 5336
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.59199256656562,
      "learning_rate": 2.1530789974324228e-07,
      "loss": 0.6905,
      "step": 5337
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6556449634581225,
      "learning_rate": 2.1413717256583145e-07,
      "loss": 0.6839,
      "step": 5338
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6724001492706012,
      "learning_rate": 2.1296960251170006e-07,
      "loss": 0.6845,
      "step": 5339
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7076750661685112,
      "learning_rate": 2.118051899575291e-07,
      "loss": 0.7174,
      "step": 5340
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7473225196810949,
      "learning_rate": 2.1064393527898353e-07,
      "loss": 0.7675,
      "step": 5341
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7612050393454505,
      "learning_rate": 2.094858388507104e-07,
      "loss": 0.7138,
      "step": 5342
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7103366836764364,
      "learning_rate": 2.083309010463319e-07,
      "loss": 0.7236,
      "step": 5343
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7321097514861684,
      "learning_rate": 2.0717912223845894e-07,
      "loss": 0.7236,
      "step": 5344
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.692284599030079,
      "learning_rate": 2.060305027986753e-07,
      "loss": 0.6707,
      "step": 5345
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6497895531963045,
      "learning_rate": 2.048850430975524e-07,
      "loss": 0.7125,
      "step": 5346
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.653219521937856,
      "learning_rate": 2.0374274350463795e-07,
      "loss": 0.6659,
      "step": 5347
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6315017308991131,
      "learning_rate": 2.0260360438846273e-07,
      "loss": 0.7127,
      "step": 5348
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6326957297836353,
      "learning_rate": 2.0146762611653713e-07,
      "loss": 0.6708,
      "step": 5349
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7730375304677846,
      "learning_rate": 2.003348090553503e-07,
      "loss": 0.6953,
      "step": 5350
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6463240416682831,
      "learning_rate": 1.9920515357037207e-07,
      "loss": 0.7182,
      "step": 5351
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6898225593742806,
      "learning_rate": 1.9807866002605425e-07,
      "loss": 0.6998,
      "step": 5352
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6756703823555548,
      "learning_rate": 1.9695532878582614e-07,
      "loss": 0.6734,
      "step": 5353
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7366283900435714,
      "learning_rate": 1.958351602120989e-07,
      "loss": 0.7002,
      "step": 5354
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6847726887113197,
      "learning_rate": 1.947181546662613e-07,
      "loss": 0.6772,
      "step": 5355
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6278940912676385,
      "learning_rate": 1.9360431250868284e-07,
      "loss": 0.6616,
      "step": 5356
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5842397926378742,
      "learning_rate": 1.9249363409871158e-07,
      "loss": 0.6401,
      "step": 5357
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7105552374411587,
      "learning_rate": 1.9138611979467646e-07,
      "loss": 0.7019,
      "step": 5358
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6681501629359151,
      "learning_rate": 1.9028176995388503e-07,
      "loss": 0.7068,
      "step": 5359
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6305497613543287,
      "learning_rate": 1.8918058493262227e-07,
      "loss": 0.668,
      "step": 5360
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7466333048081372,
      "learning_rate": 1.8808256508615507e-07,
      "loss": 0.7171,
      "step": 5361
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6585537738923654,
      "learning_rate": 1.869877107687257e-07,
      "loss": 0.6857,
      "step": 5362
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6791053650694041,
      "learning_rate": 1.858960223335582e-07,
      "loss": 0.6921,
      "step": 5363
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6933654589367698,
      "learning_rate": 1.848075001328531e-07,
      "loss": 0.7478,
      "step": 5364
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6929465430521353,
      "learning_rate": 1.837221445177917e-07,
      "loss": 0.6961,
      "step": 5365
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6653940192058783,
      "learning_rate": 1.8263995583853056e-07,
      "loss": 0.6755,
      "step": 5366
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7241729407804185,
      "learning_rate": 1.8156093444420818e-07,
      "loss": 0.7075,
      "step": 5367
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6685563837555162,
      "learning_rate": 1.8048508068293724e-07,
      "loss": 0.6773,
      "step": 5368
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7414835438495249,
      "learning_rate": 1.7941239490181228e-07,
      "loss": 0.7123,
      "step": 5369
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.718285505687217,
      "learning_rate": 1.7834287744690427e-07,
      "loss": 0.7162,
      "step": 5370
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7211233213202328,
      "learning_rate": 1.772765286632605e-07,
      "loss": 0.6902,
      "step": 5371
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6683863172002288,
      "learning_rate": 1.7621334889490805e-07,
      "loss": 0.7517,
      "step": 5372
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.58800893657313,
      "learning_rate": 1.751533384848514e-07,
      "loss": 0.6712,
      "step": 5373
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7030546148679663,
      "learning_rate": 1.7409649777507033e-07,
      "loss": 0.6895,
      "step": 5374
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.649651583517647,
      "learning_rate": 1.7304282710652544e-07,
      "loss": 0.6953,
      "step": 5375
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6630520648904323,
      "learning_rate": 1.7199232681915147e-07,
      "loss": 0.6801,
      "step": 5376
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6915271138842615,
      "learning_rate": 1.7094499725186177e-07,
      "loss": 0.6863,
      "step": 5377
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6363107121769487,
      "learning_rate": 1.6990083874254937e-07,
      "loss": 0.7143,
      "step": 5378
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6216105974102382,
      "learning_rate": 1.6885985162807817e-07,
      "loss": 0.7159,
      "step": 5379
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7070177140889281,
      "learning_rate": 1.6782203624429394e-07,
      "loss": 0.7261,
      "step": 5380
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6868125172200412,
      "learning_rate": 1.6678739292601664e-07,
      "loss": 0.7069,
      "step": 5381
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.676021686230813,
      "learning_rate": 1.6575592200704371e-07,
      "loss": 0.7264,
      "step": 5382
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5805122717883384,
      "learning_rate": 1.6472762382015005e-07,
      "loss": 0.6554,
      "step": 5383
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6174661639239597,
      "learning_rate": 1.63702498697087e-07,
      "loss": 0.6304,
      "step": 5384
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.658960891173463,
      "learning_rate": 1.6268054696857995e-07,
      "loss": 0.6848,
      "step": 5385
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6932417552597921,
      "learning_rate": 1.6166176896433072e-07,
      "loss": 0.7377,
      "step": 5386
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.683848062736622,
      "learning_rate": 1.606461650130209e-07,
      "loss": 0.689,
      "step": 5387
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6211515237690866,
      "learning_rate": 1.5963373544230388e-07,
      "loss": 0.6213,
      "step": 5388
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6449131051877578,
      "learning_rate": 1.5862448057881065e-07,
      "loss": 0.6972,
      "step": 5389
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5785383242663593,
      "learning_rate": 1.5761840074814738e-07,
      "loss": 0.6784,
      "step": 5390
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6464724770369693,
      "learning_rate": 1.566154962748978e-07,
      "loss": 0.7152,
      "step": 5391
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6591720562861992,
      "learning_rate": 1.5561576748261863e-07,
      "loss": 0.6683,
      "step": 5392
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7171113644091159,
      "learning_rate": 1.546192146938441e-07,
      "loss": 0.732,
      "step": 5393
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7318819008314915,
      "learning_rate": 1.536258382300815e-07,
      "loss": 0.6991,
      "step": 5394
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6293937561626292,
      "learning_rate": 1.5263563841181662e-07,
      "loss": 0.6804,
      "step": 5395
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6464654854294664,
      "learning_rate": 1.5164861555850728e-07,
      "loss": 0.6694,
      "step": 5396
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6811146038603135,
      "learning_rate": 1.5066476998858882e-07,
      "loss": 0.7248,
      "step": 5397
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.606539679268634,
      "learning_rate": 1.4968410201946837e-07,
      "loss": 0.6627,
      "step": 5398
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7029138632571363,
      "learning_rate": 1.4870661196753178e-07,
      "loss": 0.7542,
      "step": 5399
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7078459740295779,
      "learning_rate": 1.4773230014813568e-07,
      "loss": 0.734,
      "step": 5400
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.702816708015938,
      "learning_rate": 1.4676116687561525e-07,
      "loss": 0.7338,
      "step": 5401
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.687227510645245,
      "learning_rate": 1.457932124632788e-07,
      "loss": 0.7437,
      "step": 5402
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6810564716651443,
      "learning_rate": 1.4482843722340544e-07,
      "loss": 0.727,
      "step": 5403
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6822261790038039,
      "learning_rate": 1.4386684146725504e-07,
      "loss": 0.6998,
      "step": 5404
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6969651744840567,
      "learning_rate": 1.4290842550505724e-07,
      "loss": 0.7339,
      "step": 5405
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6776346095184568,
      "learning_rate": 1.4195318964601689e-07,
      "loss": 0.6486,
      "step": 5406
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6870674292266882,
      "learning_rate": 1.4100113419831195e-07,
      "loss": 0.7199,
      "step": 5407
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6242175327375672,
      "learning_rate": 1.4005225946909783e-07,
      "loss": 0.676,
      "step": 5408
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7037226142899639,
      "learning_rate": 1.3910656576449966e-07,
      "loss": 0.6699,
      "step": 5409
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7120608802623337,
      "learning_rate": 1.381640533896178e-07,
      "loss": 0.7228,
      "step": 5410
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5901787869698922,
      "learning_rate": 1.372247226485257e-07,
      "loss": 0.633,
      "step": 5411
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6195575885783481,
      "learning_rate": 1.3628857384427318e-07,
      "loss": 0.6785,
      "step": 5412
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.67549776582141,
      "learning_rate": 1.3535560727888087e-07,
      "loss": 0.714,
      "step": 5413
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7267726122145755,
      "learning_rate": 1.3442582325334132e-07,
      "loss": 0.7008,
      "step": 5414
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.636555896971177,
      "learning_rate": 1.3349922206762455e-07,
      "loss": 0.6666,
      "step": 5415
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6570074279463513,
      "learning_rate": 1.325758040206704e-07,
      "loss": 0.6848,
      "step": 5416
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6743986779903356,
      "learning_rate": 1.316555694103916e-07,
      "loss": 0.6725,
      "step": 5417
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6363886940951234,
      "learning_rate": 1.3073851853367736e-07,
      "loss": 0.73,
      "step": 5418
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6413491620458593,
      "learning_rate": 1.2982465168638546e-07,
      "loss": 0.6671,
      "step": 5419
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5853681308497494,
      "learning_rate": 1.2891396916335007e-07,
      "loss": 0.671,
      "step": 5420
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7757255777893428,
      "learning_rate": 1.2800647125837396e-07,
      "loss": 0.7717,
      "step": 5421
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.622092641153492,
      "learning_rate": 1.2710215826423734e-07,
      "loss": 0.6966,
      "step": 5422
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6700193672196824,
      "learning_rate": 1.2620103047268905e-07,
      "loss": 0.6956,
      "step": 5423
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6717989465116165,
      "learning_rate": 1.2530308817445215e-07,
      "loss": 0.7075,
      "step": 5424
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6658624241255655,
      "learning_rate": 1.244083316592215e-07,
      "loss": 0.6614,
      "step": 5425
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5726927402870178,
      "learning_rate": 1.2351676121566514e-07,
      "loss": 0.6704,
      "step": 5426
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6550164951169197,
      "learning_rate": 1.2262837713142073e-07,
      "loss": 0.6795,
      "step": 5427
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7160642244361908,
      "learning_rate": 1.2174317969309902e-07,
      "loss": 0.6718,
      "step": 5428
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6367168211476035,
      "learning_rate": 1.2086116918628488e-07,
      "loss": 0.7073,
      "step": 5429
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6455559979608442,
      "learning_rate": 1.199823458955318e-07,
      "loss": 0.703,
      "step": 5430
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6733444352694116,
      "learning_rate": 1.1910671010436858e-07,
      "loss": 0.6832,
      "step": 5431
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7093572912879808,
      "learning_rate": 1.1823426209529143e-07,
      "loss": 0.7634,
      "step": 5432
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6995106190231996,
      "learning_rate": 1.1736500214977186e-07,
      "loss": 0.6898,
      "step": 5433
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8126020785607709,
      "learning_rate": 1.1649893054825002e-07,
      "loss": 0.7636,
      "step": 5434
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6440847222519523,
      "learning_rate": 1.1563604757013902e-07,
      "loss": 0.6943,
      "step": 5435
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6577235908701883,
      "learning_rate": 1.1477635349382287e-07,
      "loss": 0.649,
      "step": 5436
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6918672176021524,
      "learning_rate": 1.1391984859665751e-07,
      "loss": 0.6982,
      "step": 5437
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6773683147502787,
      "learning_rate": 1.1306653315496741e-07,
      "loss": 0.6899,
      "step": 5438
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.607910258953721,
      "learning_rate": 1.1221640744405126e-07,
      "loss": 0.7027,
      "step": 5439
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6357941165784098,
      "learning_rate": 1.1136947173817858e-07,
      "loss": 0.698,
      "step": 5440
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6765220787830165,
      "learning_rate": 1.1052572631058634e-07,
      "loss": 0.6873,
      "step": 5441
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6289713809309675,
      "learning_rate": 1.0968517143348567e-07,
      "loss": 0.6635,
      "step": 5442
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.736752890661965,
      "learning_rate": 1.0884780737805522e-07,
      "loss": 0.7389,
      "step": 5443
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6452425215022389,
      "learning_rate": 1.0801363441444889e-07,
      "loss": 0.6942,
      "step": 5444
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6576566902291915,
      "learning_rate": 1.0718265281178696e-07,
      "loss": 0.705,
      "step": 5445
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6755473646377278,
      "learning_rate": 1.0635486283816055e-07,
      "loss": 0.6615,
      "step": 5446
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6286575401958581,
      "learning_rate": 1.0553026476063266e-07,
      "loss": 0.6927,
      "step": 5447
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6218993398296805,
      "learning_rate": 1.0470885884523608e-07,
      "loss": 0.7201,
      "step": 5448
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5941705762963674,
      "learning_rate": 1.0389064535697324e-07,
      "loss": 0.6375,
      "step": 5449
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6707269164215461,
      "learning_rate": 1.0307562455981635e-07,
      "loss": 0.7064,
      "step": 5450
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.699736926800377,
      "learning_rate": 1.022637967167095e-07,
      "loss": 0.7312,
      "step": 5451
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6706483072900383,
      "learning_rate": 1.014551620895643e-07,
      "loss": 0.6823,
      "step": 5452
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6999413184463142,
      "learning_rate": 1.0064972093926317e-07,
      "loss": 0.6851,
      "step": 5453
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6173633571657301,
      "learning_rate": 9.984747352565827e-08,
      "loss": 0.6649,
      "step": 5454
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6783149393390168,
      "learning_rate": 9.904842010757143e-08,
      "loss": 0.6938,
      "step": 5455
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6667822024853537,
      "learning_rate": 9.825256094279312e-08,
      "loss": 0.6928,
      "step": 5456
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.683962660733446,
      "learning_rate": 9.745989628808572e-08,
      "loss": 0.6846,
      "step": 5457
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6540379375173747,
      "learning_rate": 9.6670426399178e-08,
      "loss": 0.6585,
      "step": 5458
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6677191222893065,
      "learning_rate": 9.588415153076958e-08,
      "loss": 0.6845,
      "step": 5459
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6467645870247477,
      "learning_rate": 9.510107193652973e-08,
      "loss": 0.6489,
      "step": 5460
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6458079781384493,
      "learning_rate": 9.432118786909639e-08,
      "loss": 0.6903,
      "step": 5461
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7366007208492513,
      "learning_rate": 9.354449958007494e-08,
      "loss": 0.6817,
      "step": 5462
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6631537568695398,
      "learning_rate": 9.277100732004273e-08,
      "loss": 0.7014,
      "step": 5463
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7228495512555279,
      "learning_rate": 9.200071133854349e-08,
      "loss": 0.6959,
      "step": 5464
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6406333034562853,
      "learning_rate": 9.123361188409175e-08,
      "loss": 0.6639,
      "step": 5465
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6852528112871983,
      "learning_rate": 9.046970920416953e-08,
      "loss": 0.708,
      "step": 5466
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7562878779958014,
      "learning_rate": 8.97090035452275e-08,
      "loss": 0.7331,
      "step": 5467
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6243821934093773,
      "learning_rate": 8.895149515268376e-08,
      "loss": 0.6952,
      "step": 5468
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7018068013029932,
      "learning_rate": 8.819718427092838e-08,
      "loss": 0.715,
      "step": 5469
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.622901467427635,
      "learning_rate": 8.744607114331782e-08,
      "loss": 0.6721,
      "step": 5470
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6192400133204734,
      "learning_rate": 8.669815601217602e-08,
      "loss": 0.6335,
      "step": 5471
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6299693044914715,
      "learning_rate": 8.59534391187955e-08,
      "loss": 0.6804,
      "step": 5472
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5998845403135361,
      "learning_rate": 8.521192070343853e-08,
      "loss": 0.6553,
      "step": 5473
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6172238627711223,
      "learning_rate": 8.447360100533264e-08,
      "loss": 0.7108,
      "step": 5474
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6899381833964708,
      "learning_rate": 8.373848026267728e-08,
      "loss": 0.7325,
      "step": 5475
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6386727569951643,
      "learning_rate": 8.300655871263608e-08,
      "loss": 0.6958,
      "step": 5476
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6076590172086611,
      "learning_rate": 8.227783659134347e-08,
      "loss": 0.6884,
      "step": 5477
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6334231259725904,
      "learning_rate": 8.155231413390031e-08,
      "loss": 0.6375,
      "step": 5478
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6992592151939718,
      "learning_rate": 8.082999157437488e-08,
      "loss": 0.7001,
      "step": 5479
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7245712388693513,
      "learning_rate": 8.011086914580302e-08,
      "loss": 0.7541,
      "step": 5480
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7241863948307017,
      "learning_rate": 7.939494708018914e-08,
      "loss": 0.6867,
      "step": 5481
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.624667387328441,
      "learning_rate": 7.868222560850402e-08,
      "loss": 0.6544,
      "step": 5482
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6444116829613343,
      "learning_rate": 7.797270496068598e-08,
      "loss": 0.6984,
      "step": 5483
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7788181361690866,
      "learning_rate": 7.726638536564413e-08,
      "loss": 0.7241,
      "step": 5484
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6171232487490448,
      "learning_rate": 7.656326705124839e-08,
      "loss": 0.6723,
      "step": 5485
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.712423075514945,
      "learning_rate": 7.586335024434178e-08,
      "loss": 0.6765,
      "step": 5486
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6737013950228641,
      "learning_rate": 7.51666351707303e-08,
      "loss": 0.6704,
      "step": 5487
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6118646559437555,
      "learning_rate": 7.447312205518976e-08,
      "loss": 0.6905,
      "step": 5488
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6486017992438559,
      "learning_rate": 7.378281112146224e-08,
      "loss": 0.6884,
      "step": 5489
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7004253281266634,
      "learning_rate": 7.309570259225629e-08,
      "loss": 0.7354,
      "step": 5490
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.672246031033486,
      "learning_rate": 7.24117966892468e-08,
      "loss": 0.667,
      "step": 5491
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6658028795447529,
      "learning_rate": 7.173109363307617e-08,
      "loss": 0.6708,
      "step": 5492
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6661353937865077,
      "learning_rate": 7.105359364335318e-08,
      "loss": 0.652,
      "step": 5493
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7733895384879172,
      "learning_rate": 7.037929693865408e-08,
      "loss": 0.7192,
      "step": 5494
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7780937399775334,
      "learning_rate": 6.970820373652154e-08,
      "loss": 0.7036,
      "step": 5495
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6710389581774653,
      "learning_rate": 6.904031425346236e-08,
      "loss": 0.7042,
      "step": 5496
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7112103254552711,
      "learning_rate": 6.837562870495307e-08,
      "loss": 0.6662,
      "step": 5497
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6571684379954628,
      "learning_rate": 6.771414730543546e-08,
      "loss": 0.6964,
      "step": 5498
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6917745000703825,
      "learning_rate": 6.705587026831551e-08,
      "loss": 0.7157,
      "step": 5499
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.654789333280689,
      "learning_rate": 6.640079780596886e-08,
      "loss": 0.7116,
      "step": 5500
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6886054394480884,
      "learning_rate": 6.574893012973427e-08,
      "loss": 0.6973,
      "step": 5501
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7149878341042419,
      "learning_rate": 6.510026744991904e-08,
      "loss": 0.7408,
      "step": 5502
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6633918888951519,
      "learning_rate": 6.445480997579356e-08,
      "loss": 0.6936,
      "step": 5503
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7033771664604066,
      "learning_rate": 6.381255791559682e-08,
      "loss": 0.7337,
      "step": 5504
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6423251938853566,
      "learning_rate": 6.317351147653305e-08,
      "loss": 0.6756,
      "step": 5505
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6298487804842499,
      "learning_rate": 6.253767086477181e-08,
      "loss": 0.6583,
      "step": 5506
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6015206424147753,
      "learning_rate": 6.190503628544897e-08,
      "loss": 0.6392,
      "step": 5507
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6728794733465611,
      "learning_rate": 6.127560794266462e-08,
      "loss": 0.6825,
      "step": 5508
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7021633765065407,
      "learning_rate": 6.064938603948633e-08,
      "loss": 0.7052,
      "step": 5509
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8020683064701568,
      "learning_rate": 6.002637077794471e-08,
      "loss": 0.6889,
      "step": 5510
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.666740883150888,
      "learning_rate": 5.940656235904008e-08,
      "loss": 0.6637,
      "step": 5511
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.655418513501751,
      "learning_rate": 5.878996098273249e-08,
      "loss": 0.7117,
      "step": 5512
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6509193483789036,
      "learning_rate": 5.8176566847953916e-08,
      "loss": 0.673,
      "step": 5513
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6282077913359626,
      "learning_rate": 5.756638015259497e-08,
      "loss": 0.7148,
      "step": 5514
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6737025470285977,
      "learning_rate": 5.6959401093515944e-08,
      "loss": 0.7112,
      "step": 5515
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6525838584142409,
      "learning_rate": 5.63556298665413e-08,
      "loss": 0.7058,
      "step": 5516
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6763715789943742,
      "learning_rate": 5.575506666645858e-08,
      "loss": 0.6911,
      "step": 5517
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6807084627385365,
      "learning_rate": 5.5157711687023884e-08,
      "loss": 0.6541,
      "step": 5518
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6531066067639685,
      "learning_rate": 5.45635651209564e-08,
      "loss": 0.6773,
      "step": 5519
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6748731250438016,
      "learning_rate": 5.397262715993834e-08,
      "loss": 0.7212,
      "step": 5520
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6680565574391911,
      "learning_rate": 5.3384897994619433e-08,
      "loss": 0.7138,
      "step": 5521
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6742687511307099,
      "learning_rate": 5.280037781461467e-08,
      "loss": 0.7481,
      "step": 5522
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7036199496400647,
      "learning_rate": 5.2219066808500976e-08,
      "loss": 0.7265,
      "step": 5523
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7331251590788749,
      "learning_rate": 5.164096516382167e-08,
      "loss": 0.7137,
      "step": 5524
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6564399318181942,
      "learning_rate": 5.106607306708422e-08,
      "loss": 0.6837,
      "step": 5525
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6617534965394755,
      "learning_rate": 5.049439070376139e-08,
      "loss": 0.6782,
      "step": 5526
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6843668503518696,
      "learning_rate": 4.992591825829007e-08,
      "loss": 0.6655,
      "step": 5527
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6192024751956472,
      "learning_rate": 4.9360655914070245e-08,
      "loss": 0.6442,
      "step": 5528
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7190940679744571,
      "learning_rate": 4.879860385346713e-08,
      "loss": 0.7058,
      "step": 5529
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6509898251074434,
      "learning_rate": 4.823976225781235e-08,
      "loss": 0.6685,
      "step": 5530
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7384313808259438,
      "learning_rate": 4.768413130739835e-08,
      "loss": 0.7364,
      "step": 5531
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6993597851830534,
      "learning_rate": 4.713171118148396e-08,
      "loss": 0.6633,
      "step": 5532
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6584431493297701,
      "learning_rate": 4.6582502058289956e-08,
      "loss": 0.6653,
      "step": 5533
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6626327007144456,
      "learning_rate": 4.60365041150046e-08,
      "loss": 0.7098,
      "step": 5534
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6075177875270455,
      "learning_rate": 4.549371752777698e-08,
      "loss": 0.6744,
      "step": 5535
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6278407232485914,
      "learning_rate": 4.4954142471721475e-08,
      "loss": 0.6674,
      "step": 5536
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6107788967364894,
      "learning_rate": 4.441777912091771e-08,
      "loss": 0.6564,
      "step": 5537
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7009218510196313,
      "learning_rate": 4.388462764840507e-08,
      "loss": 0.6939,
      "step": 5538
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6783399765325352,
      "learning_rate": 4.335468822619038e-08,
      "loss": 0.6825,
      "step": 5539
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6967567211284718,
      "learning_rate": 4.282796102524467e-08,
      "loss": 0.7133,
      "step": 5540
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7299255416291,
      "learning_rate": 4.230444621549978e-08,
      "loss": 0.7064,
      "step": 5541
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.662080336125653,
      "learning_rate": 4.178414396585173e-08,
      "loss": 0.7124,
      "step": 5542
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.68721864121572,
      "learning_rate": 4.12670544441629e-08,
      "loss": 0.7428,
      "step": 5543
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6666740345956339,
      "learning_rate": 4.0753177817255406e-08,
      "loss": 0.6459,
      "step": 5544
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7102313023014182,
      "learning_rate": 4.024251425091663e-08,
      "loss": 0.7255,
      "step": 5545
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6352479947497611,
      "learning_rate": 3.973506390989923e-08,
      "loss": 0.6251,
      "step": 5546
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6682410311757189,
      "learning_rate": 3.923082695791447e-08,
      "loss": 0.6986,
      "step": 5547
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6203552913751105,
      "learning_rate": 3.872980355764222e-08,
      "loss": 0.701,
      "step": 5548
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6166951682069134,
      "learning_rate": 3.823199387072096e-08,
      "loss": 0.6747,
      "step": 5549
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7201814776247042,
      "learning_rate": 3.773739805775556e-08,
      "loss": 0.7302,
      "step": 5550
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7311603783743563,
      "learning_rate": 3.724601627831281e-08,
      "loss": 0.6878,
      "step": 5551
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7721903203447883,
      "learning_rate": 3.675784869092258e-08,
      "loss": 0.7145,
      "step": 5552
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6878445567195126,
      "learning_rate": 3.627289545307888e-08,
      "loss": 0.7745,
      "step": 5553
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6023877723449473,
      "learning_rate": 3.5791156721236567e-08,
      "loss": 0.641,
      "step": 5554
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7864912664357397,
      "learning_rate": 3.531263265081464e-08,
      "loss": 0.7494,
      "step": 5555
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6245292753101318,
      "learning_rate": 3.4837323396195164e-08,
      "loss": 0.6661,
      "step": 5556
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8498553047530379,
      "learning_rate": 3.436522911072326e-08,
      "loss": 0.7242,
      "step": 5557
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7084446830920609,
      "learning_rate": 3.3896349946705984e-08,
      "loss": 0.7211,
      "step": 5558
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6537481507428908,
      "learning_rate": 3.343068605541233e-08,
      "loss": 0.6689,
      "step": 5559
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6998291054957169,
      "learning_rate": 3.2968237587077676e-08,
      "loss": 0.6953,
      "step": 5560
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6544705507422103,
      "learning_rate": 3.250900469089602e-08,
      "loss": 0.6872,
      "step": 5561
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7261181523196746,
      "learning_rate": 3.20529875150255e-08,
      "loss": 0.6946,
      "step": 5562
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6553953168843222,
      "learning_rate": 3.160018620658734e-08,
      "loss": 0.6327,
      "step": 5563
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7169339212533979,
      "learning_rate": 3.115060091166355e-08,
      "loss": 0.6902,
      "step": 5564
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6706494871016345,
      "learning_rate": 3.070423177530035e-08,
      "loss": 0.7259,
      "step": 5565
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6591363505346738,
      "learning_rate": 3.0261078941506986e-08,
      "loss": 0.6945,
      "step": 5566
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8309762182752156,
      "learning_rate": 2.98211425532513e-08,
      "loss": 0.7277,
      "step": 5567
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6130791252474391,
      "learning_rate": 2.9384422752467558e-08,
      "loss": 0.6466,
      "step": 5568
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6659476727852927,
      "learning_rate": 2.8950919680050814e-08,
      "loss": 0.6641,
      "step": 5569
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6805059575578812,
      "learning_rate": 2.8520633475856985e-08,
      "loss": 0.6992,
      "step": 5570
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6672156392447297,
      "learning_rate": 2.809356427870724e-08,
      "loss": 0.7011,
      "step": 5571
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.325974052365072,
      "learning_rate": 2.7669712226381374e-08,
      "loss": 0.6775,
      "step": 5572
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6606645159603727,
      "learning_rate": 2.724907745562333e-08,
      "loss": 0.6645,
      "step": 5573
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6461375985953987,
      "learning_rate": 2.6831660102139e-08,
      "loss": 0.6835,
      "step": 5574
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6812150693887683,
      "learning_rate": 2.641746030059511e-08,
      "loss": 0.6983,
      "step": 5575
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6326942933903705,
      "learning_rate": 2.600647818462143e-08,
      "loss": 0.6668,
      "step": 5576
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7217483331242144,
      "learning_rate": 2.559871388681079e-08,
      "loss": 0.7293,
      "step": 5577
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6519517630307867,
      "learning_rate": 2.519416753871462e-08,
      "loss": 0.6748,
      "step": 5578
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6811858906944799,
      "learning_rate": 2.4792839270848523e-08,
      "loss": 0.6685,
      "step": 5579
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6824146970539711,
      "learning_rate": 2.4394729212690037e-08,
      "loss": 0.714,
      "step": 5580
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6029142961677746,
      "learning_rate": 2.3999837492676427e-08,
      "loss": 0.6876,
      "step": 5581
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6667563331685223,
      "learning_rate": 2.360816423820911e-08,
      "loss": 0.7063,
      "step": 5582
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6901064776150798,
      "learning_rate": 2.321970957564923e-08,
      "loss": 0.7244,
      "step": 5583
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6516633716856788,
      "learning_rate": 2.2834473630322095e-08,
      "loss": 0.7108,
      "step": 5584
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.690202132458105,
      "learning_rate": 2.2452456526510513e-08,
      "loss": 0.7101,
      "step": 5585
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6785092486002453,
      "learning_rate": 2.207365838746256e-08,
      "loss": 0.7216,
      "step": 5586
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5967070362582922,
      "learning_rate": 2.1698079335386036e-08,
      "loss": 0.6483,
      "step": 5587
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.717265880506052,
      "learning_rate": 2.132571949145068e-08,
      "loss": 0.7158,
      "step": 5588
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.59717832573565,
      "learning_rate": 2.0956578975788177e-08,
      "loss": 0.6313,
      "step": 5589
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6059682059829004,
      "learning_rate": 2.059065790748993e-08,
      "loss": 0.6883,
      "step": 5590
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6699089102571342,
      "learning_rate": 2.0227956404609282e-08,
      "loss": 0.6841,
      "step": 5591
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6538206419330663,
      "learning_rate": 1.9868474584162633e-08,
      "loss": 0.6932,
      "step": 5592
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6178147027974642,
      "learning_rate": 1.951221256212499e-08,
      "loss": 0.6649,
      "step": 5593
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6689091824554173,
      "learning_rate": 1.9159170453435515e-08,
      "loss": 0.6918,
      "step": 5594
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6443533334043914,
      "learning_rate": 1.8809348371991997e-08,
      "loss": 0.6906,
      "step": 5595
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7345928859014631,
      "learning_rate": 1.8462746430654155e-08,
      "loss": 0.7179,
      "step": 5596
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6176032287098241,
      "learning_rate": 1.811936474124254e-08,
      "loss": 0.6995,
      "step": 5597
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6834843849128179,
      "learning_rate": 1.777920341454076e-08,
      "loss": 0.6857,
      "step": 5598
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6893804803113004,
      "learning_rate": 1.7442262560291022e-08,
      "loss": 0.6902,
      "step": 5599
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6607789810574357,
      "learning_rate": 1.7108542287197493e-08,
      "loss": 0.6758,
      "step": 5600
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6921919265606316,
      "learning_rate": 1.6778042702926266e-08,
      "loss": 0.7163,
      "step": 5601
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6876722359209989,
      "learning_rate": 1.6450763914102052e-08,
      "loss": 0.7334,
      "step": 5602
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6798548706442888,
      "learning_rate": 1.6126706026311502e-08,
      "loss": 0.6848,
      "step": 5603
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7653289765067438,
      "learning_rate": 1.5805869144103205e-08,
      "loss": 0.749,
      "step": 5604
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6820773147117589,
      "learning_rate": 1.5488253370986585e-08,
      "loss": 0.6783,
      "step": 5605
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6822425813053976,
      "learning_rate": 1.5173858809429676e-08,
      "loss": 0.6728,
      "step": 5606
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6508818535981635,
      "learning_rate": 1.4862685560863565e-08,
      "loss": 0.7089,
      "step": 5607
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6816747025597627,
      "learning_rate": 1.4554733725680171e-08,
      "loss": 0.7293,
      "step": 5608
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5927305901316613,
      "learning_rate": 1.4250003403228907e-08,
      "loss": 0.6614,
      "step": 5609
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7332773081630236,
      "learning_rate": 1.3948494691823355e-08,
      "loss": 0.7601,
      "step": 5610
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6451677762060226,
      "learning_rate": 1.3650207688735706e-08,
      "loss": 0.6785,
      "step": 5611
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8462592227347205,
      "learning_rate": 1.33551424902012e-08,
      "loss": 0.7019,
      "step": 5612
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7439854657746539,
      "learning_rate": 1.3063299191413692e-08,
      "loss": 0.7346,
      "step": 5613
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6583400200491211,
      "learning_rate": 1.2774677886526753e-08,
      "loss": 0.6715,
      "step": 5614
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7227774176985211,
      "learning_rate": 1.2489278668655901e-08,
      "loss": 0.7195,
      "step": 5615
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6113205614217184,
      "learning_rate": 1.2207101629876373e-08,
      "loss": 0.7002,
      "step": 5616
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7040848845759214,
      "learning_rate": 1.1928146861226453e-08,
      "loss": 0.7238,
      "step": 5617
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6701186810729105,
      "learning_rate": 1.1652414452701932e-08,
      "loss": 0.6947,
      "step": 5618
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6765009903599033,
      "learning_rate": 1.137990449325832e-08,
      "loss": 0.6922,
      "step": 5619
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6152819513157632,
      "learning_rate": 1.1110617070814178e-08,
      "loss": 0.6484,
      "step": 5620
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7479269557589341,
      "learning_rate": 1.0844552272246678e-08,
      "loss": 0.6922,
      "step": 5621
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6497463669298972,
      "learning_rate": 1.0581710183394934e-08,
      "loss": 0.6741,
      "step": 5622
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6704578023265587,
      "learning_rate": 1.0322090889056669e-08,
      "loss": 0.6823,
      "step": 5623
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6380405049419285,
      "learning_rate": 1.006569447299044e-08,
      "loss": 0.6647,
      "step": 5624
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7251480186673226,
      "learning_rate": 9.812521017914523e-09,
      "loss": 0.6893,
      "step": 5625
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6496013199465467,
      "learning_rate": 9.562570605509137e-09,
      "loss": 0.6896,
      "step": 5626
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7317547010303674,
      "learning_rate": 9.31584331641311e-09,
      "loss": 0.7349,
      "step": 5627
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6941541159052061,
      "learning_rate": 9.072339230224992e-09,
      "loss": 0.7274,
      "step": 5628
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6572123224573279,
      "learning_rate": 8.832058425505275e-09,
      "loss": 0.6666,
      "step": 5629
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7056504798650207,
      "learning_rate": 8.59500097977306e-09,
      "loss": 0.7147,
      "step": 5630
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7170785602251238,
      "learning_rate": 8.361166969508283e-09,
      "loss": 0.7146,
      "step": 5631
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7154492326080727,
      "learning_rate": 8.130556470149486e-09,
      "loss": 0.6976,
      "step": 5632
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6460169141662006,
      "learning_rate": 7.903169556098267e-09,
      "loss": 0.7211,
      "step": 5633
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.719410740540638,
      "learning_rate": 7.67900630071372e-09,
      "loss": 0.7217,
      "step": 5634
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6268100343483555,
      "learning_rate": 7.458066776314665e-09,
      "loss": 0.696,
      "step": 5635
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6840661464362316,
      "learning_rate": 7.24035105418186e-09,
      "loss": 0.7148,
      "step": 5636
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6963274874381727,
      "learning_rate": 7.025859204554675e-09,
      "loss": 0.6827,
      "step": 5637
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6527691498833009,
      "learning_rate": 6.814591296633311e-09,
      "loss": 0.6498,
      "step": 5638
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6972615324107898,
      "learning_rate": 6.606547398575469e-09,
      "loss": 0.7039,
      "step": 5639
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7074917816594729,
      "learning_rate": 6.401727577501904e-09,
      "loss": 0.7028,
      "step": 5640
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6734303610209647,
      "learning_rate": 6.200131899490869e-09,
      "loss": 0.6657,
      "step": 5641
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7596063316195328,
      "learning_rate": 6.001760429581449e-09,
      "loss": 0.707,
      "step": 5642
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7141100089259258,
      "learning_rate": 5.806613231773561e-09,
      "loss": 0.7044,
      "step": 5643
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6557798979954025,
      "learning_rate": 5.614690369024622e-09,
      "loss": 0.6747,
      "step": 5644
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7435377402331151,
      "learning_rate": 5.425991903252881e-09,
      "loss": 0.7045,
      "step": 5645
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6460518173894233,
      "learning_rate": 5.240517895337416e-09,
      "loss": 0.6977,
      "step": 5646
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5958300766470458,
      "learning_rate": 5.058268405114808e-09,
      "loss": 0.6783,
      "step": 5647
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6942252361999809,
      "learning_rate": 4.879243491383578e-09,
      "loss": 0.7187,
      "step": 5648
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6317507970935711,
      "learning_rate": 4.70344321190086e-09,
      "loss": 0.6854,
      "step": 5649
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6487419106880962,
      "learning_rate": 4.530867623381285e-09,
      "loss": 0.6853,
      "step": 5650
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6427862921547772,
      "learning_rate": 4.36151678150476e-09,
      "loss": 0.6799,
      "step": 5651
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6955621240362855,
      "learning_rate": 4.195390740905359e-09,
      "loss": 0.713,
      "step": 5652
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.658454078844665,
      "learning_rate": 4.0324895551779875e-09,
      "loss": 0.6997,
      "step": 5653
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6520856152064165,
      "learning_rate": 3.872813276880605e-09,
      "loss": 0.6511,
      "step": 5654
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6727730127041908,
      "learning_rate": 3.7163619575264485e-09,
      "loss": 0.7105,
      "step": 5655
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6880240607490584,
      "learning_rate": 3.5631356475895883e-09,
      "loss": 0.7037,
      "step": 5656
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7939931686203886,
      "learning_rate": 3.4131343965049244e-09,
      "loss": 0.71,
      "step": 5657
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.663413998886994,
      "learning_rate": 3.2663582526648584e-09,
      "loss": 0.7274,
      "step": 5658
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6703515192296359,
      "learning_rate": 3.122807263423733e-09,
      "loss": 0.6852,
      "step": 5659
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7908282089115537,
      "learning_rate": 2.9824814750945007e-09,
      "loss": 0.7469,
      "step": 5660
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6610715910812922,
      "learning_rate": 2.845380932947617e-09,
      "loss": 0.7247,
      "step": 5661
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6348239080514873,
      "learning_rate": 2.7115056812143657e-09,
      "loss": 0.7101,
      "step": 5662
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6912434716920728,
      "learning_rate": 2.580855763087975e-09,
      "loss": 0.7126,
      "step": 5663
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6793698310743903,
      "learning_rate": 2.453431220716951e-09,
      "loss": 0.6587,
      "step": 5664
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.612317512376912,
      "learning_rate": 2.3292320952106318e-09,
      "loss": 0.709,
      "step": 5665
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0186555713822414,
      "learning_rate": 2.2082584266402974e-09,
      "loss": 0.7297,
      "step": 5666
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6923241471031918,
      "learning_rate": 2.0905102540336173e-09,
      "loss": 0.7014,
      "step": 5667
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6663853832824121,
      "learning_rate": 1.975987615376873e-09,
      "loss": 0.649,
      "step": 5668
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6051407891403848,
      "learning_rate": 1.864690547620507e-09,
      "loss": 0.6942,
      "step": 5669
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7256563573686435,
      "learning_rate": 1.7566190866691312e-09,
      "loss": 0.7202,
      "step": 5670
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7528249654043957,
      "learning_rate": 1.6517732673904109e-09,
      "loss": 0.7547,
      "step": 5671
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6525375255576223,
      "learning_rate": 1.5501531236084e-09,
      "loss": 0.6465,
      "step": 5672
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6496737477274479,
      "learning_rate": 1.4517586881079847e-09,
      "loss": 0.6997,
      "step": 5673
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.665438568393549,
      "learning_rate": 1.3565899926337721e-09,
      "loss": 0.7057,
      "step": 5674
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6207866975370773,
      "learning_rate": 1.26464706788898e-09,
      "loss": 0.687,
      "step": 5675
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6557792100652791,
      "learning_rate": 1.1759299435365468e-09,
      "loss": 0.6807,
      "step": 5676
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6575142167006899,
      "learning_rate": 1.0904386481980222e-09,
      "loss": 0.6948,
      "step": 5677
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6953964969464204,
      "learning_rate": 1.0081732094557872e-09,
      "loss": 0.6879,
      "step": 5678
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.70671209879021,
      "learning_rate": 9.291336538486129e-10,
      "loss": 0.7287,
      "step": 5679
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.664184034482496,
      "learning_rate": 8.533200068783221e-10,
      "loss": 0.712,
      "step": 5680
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6694939436207021,
      "learning_rate": 7.807322930020179e-10,
      "loss": 0.7176,
      "step": 5681
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6491860336936887,
      "learning_rate": 7.113705356387446e-10,
      "loss": 0.6966,
      "step": 5682
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5892041087337994,
      "learning_rate": 6.452347571661577e-10,
      "loss": 0.6501,
      "step": 5683
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6380089802129617,
      "learning_rate": 5.823249789216334e-10,
      "loss": 0.716,
      "step": 5684
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6516162841713956,
      "learning_rate": 5.226412212000486e-10,
      "loss": 0.6926,
      "step": 5685
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6598421445406673,
      "learning_rate": 4.661835032582218e-10,
      "loss": 0.6837,
      "step": 5686
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6999742937275595,
      "learning_rate": 4.1295184330825135e-10,
      "loss": 0.7364,
      "step": 5687
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7204280449147162,
      "learning_rate": 3.6294625852639765e-10,
      "loss": 0.7144,
      "step": 5688
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7026042003397023,
      "learning_rate": 3.1616676504309106e-10,
      "loss": 0.6841,
      "step": 5689
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.583902088662315,
      "learning_rate": 2.7261337795181365e-10,
      "loss": 0.6676,
      "step": 5690
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6147232048798866,
      "learning_rate": 2.3228611130354795e-10,
      "loss": 0.6871,
      "step": 5691
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6724636599074791,
      "learning_rate": 1.9518497810899762e-10,
      "loss": 0.7056,
      "step": 5692
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8050494673458964,
      "learning_rate": 1.6130999033747707e-10,
      "loss": 0.7494,
      "step": 5693
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6586580424743982,
      "learning_rate": 1.3066115891691156e-10,
      "loss": 0.7141,
      "step": 5694
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7112422246450445,
      "learning_rate": 1.0323849373605754e-10,
      "loss": 0.7,
      "step": 5695
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6489913107109709,
      "learning_rate": 7.904200364228232e-11,
      "loss": 0.6951,
      "step": 5696
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.702646726069728,
      "learning_rate": 5.807169644156396e-11,
      "loss": 0.666,
      "step": 5697
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6616902934848862,
      "learning_rate": 4.0327578899601596e-11,
      "loss": 0.6658,
      "step": 5698
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7283359101001657,
      "learning_rate": 2.5809656740705123e-11,
      "loss": 0.6824,
      "step": 5699
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.629232660896775,
      "learning_rate": 1.4517934648905497e-11,
      "loss": 0.6755,
      "step": 5700
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6485218932150899,
      "learning_rate": 6.452416266844452e-12,
      "loss": 0.6873,
      "step": 5701
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6288450128046413,
      "learning_rate": 1.6131041968847628e-12,
      "loss": 0.6744,
      "step": 5702
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6746962056277772,
      "learning_rate": 0.0,
      "loss": 0.6594,
      "step": 5703
    },
    {
      "epoch": 1.0,
      "step": 5703,
      "total_flos": 0.0,
      "train_loss": 0.19594565146265794,
      "train_runtime": 5410.6584,
      "train_samples_per_second": 272.424,
      "train_steps_per_second": 1.054
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 5703,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}