{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999438107546216,
  "eval_steps": 500,
  "global_step": 4449,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 8.505369019902519,
      "learning_rate": 1.4925373134328358e-07,
      "loss": 1.0584,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.460612403025769,
      "learning_rate": 2.9850746268656716e-07,
      "loss": 1.0369,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.08028531521452,
      "learning_rate": 4.4776119402985074e-07,
      "loss": 1.0547,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.26449530311942,
      "learning_rate": 5.970149253731343e-07,
      "loss": 1.5438,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.87881967414935,
      "learning_rate": 7.462686567164179e-07,
      "loss": 1.143,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.91070226698879,
      "learning_rate": 8.955223880597015e-07,
      "loss": 1.5967,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.16528766086488,
      "learning_rate": 1.044776119402985e-06,
      "loss": 1.0637,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.38121114490295,
      "learning_rate": 1.1940298507462686e-06,
      "loss": 1.5792,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.39935471226613,
      "learning_rate": 1.3432835820895524e-06,
      "loss": 1.5904,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.336927749705877,
      "learning_rate": 1.4925373134328358e-06,
      "loss": 1.0684,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.588351905222346,
      "learning_rate": 1.6417910447761196e-06,
      "loss": 0.9508,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.743426756578889,
      "learning_rate": 1.791044776119403e-06,
      "loss": 1.1119,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.673732117230301,
      "learning_rate": 1.9402985074626867e-06,
      "loss": 1.445,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.9403413279598665,
      "learning_rate": 2.08955223880597e-06,
      "loss": 0.9677,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.847716160343175,
      "learning_rate": 2.238805970149254e-06,
      "loss": 0.9563,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5204931749940935,
      "learning_rate": 2.3880597014925373e-06,
      "loss": 0.9073,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.343311449910715,
      "learning_rate": 2.537313432835821e-06,
      "loss": 1.3288,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.334572411451441,
      "learning_rate": 2.686567164179105e-06,
      "loss": 0.7951,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.364745903031397,
      "learning_rate": 2.835820895522388e-06,
      "loss": 0.894,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.698126210880435,
      "learning_rate": 2.9850746268656716e-06,
      "loss": 0.9041,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.289637597706571,
      "learning_rate": 3.1343283582089558e-06,
      "loss": 1.316,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.69914779265071,
      "learning_rate": 3.283582089552239e-06,
      "loss": 0.9002,
      "step": 22
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.335550624770729,
      "learning_rate": 3.4328358208955225e-06,
      "loss": 1.3679,
      "step": 23
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.448766113425176,
      "learning_rate": 3.582089552238806e-06,
      "loss": 1.3909,
      "step": 24
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.346394189325518,
      "learning_rate": 3.73134328358209e-06,
      "loss": 0.82,
      "step": 25
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0155822232366165,
      "learning_rate": 3.8805970149253735e-06,
      "loss": 0.8032,
      "step": 26
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.847472965600737,
      "learning_rate": 4.029850746268657e-06,
      "loss": 0.7708,
      "step": 27
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.707732368735433,
      "learning_rate": 4.17910447761194e-06,
      "loss": 0.7985,
      "step": 28
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8081471749501614,
      "learning_rate": 4.3283582089552236e-06,
      "loss": 0.8729,
      "step": 29
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8020897817604733,
      "learning_rate": 4.477611940298508e-06,
      "loss": 0.7697,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2697448298744245,
      "learning_rate": 4.626865671641791e-06,
      "loss": 0.7441,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5682251974276156,
      "learning_rate": 4.7761194029850745e-06,
      "loss": 0.7153,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.238410056028206,
      "learning_rate": 4.925373134328359e-06,
      "loss": 1.2337,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5791613440297985,
      "learning_rate": 5.074626865671642e-06,
      "loss": 0.7112,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.437644356567243,
      "learning_rate": 5.2238805970149255e-06,
      "loss": 0.7213,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.961457023422953,
      "learning_rate": 5.37313432835821e-06,
      "loss": 0.7242,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9459308902837287,
      "learning_rate": 5.522388059701493e-06,
      "loss": 1.2992,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5579365404892402,
      "learning_rate": 5.671641791044776e-06,
      "loss": 0.8186,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.159655387537458,
      "learning_rate": 5.820895522388061e-06,
      "loss": 0.7608,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2031325556003583,
      "learning_rate": 5.970149253731343e-06,
      "loss": 0.8251,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.20009120892666,
      "learning_rate": 6.119402985074627e-06,
      "loss": 0.7404,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6643931104013094,
      "learning_rate": 6.2686567164179116e-06,
      "loss": 1.1976,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0964104023173555,
      "learning_rate": 6.417910447761194e-06,
      "loss": 0.7252,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2053680048230175,
      "learning_rate": 6.567164179104478e-06,
      "loss": 0.7042,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.066342200736478,
      "learning_rate": 6.7164179104477625e-06,
      "loss": 1.1819,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3458984801073566,
      "learning_rate": 6.865671641791045e-06,
      "loss": 0.6956,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3056131627756953,
      "learning_rate": 7.014925373134329e-06,
      "loss": 0.7561,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6066971479046015,
      "learning_rate": 7.164179104477612e-06,
      "loss": 1.196,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.023775842801237,
      "learning_rate": 7.313432835820896e-06,
      "loss": 0.7584,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0565598115557124,
      "learning_rate": 7.46268656716418e-06,
      "loss": 0.6739,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9218721779809398,
      "learning_rate": 7.611940298507463e-06,
      "loss": 0.6874,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7349139941866447,
      "learning_rate": 7.761194029850747e-06,
      "loss": 1.2392,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7289396738656957,
      "learning_rate": 7.91044776119403e-06,
      "loss": 1.2405,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.956810447631589,
      "learning_rate": 8.059701492537314e-06,
      "loss": 0.6521,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.797862860208847,
      "learning_rate": 8.208955223880599e-06,
      "loss": 1.1572,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9481495957779815,
      "learning_rate": 8.35820895522388e-06,
      "loss": 0.6344,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5521883124975173,
      "learning_rate": 8.507462686567165e-06,
      "loss": 1.2916,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0262660730378372,
      "learning_rate": 8.656716417910447e-06,
      "loss": 0.6664,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9007216768986313,
      "learning_rate": 8.805970149253732e-06,
      "loss": 1.2593,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2627644164246026,
      "learning_rate": 8.955223880597016e-06,
      "loss": 0.65,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.139796725336458,
      "learning_rate": 9.104477611940299e-06,
      "loss": 0.6577,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.176209841596166,
      "learning_rate": 9.253731343283582e-06,
      "loss": 0.7098,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9586966649220618,
      "learning_rate": 9.402985074626867e-06,
      "loss": 0.7111,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1184396040308147,
      "learning_rate": 9.552238805970149e-06,
      "loss": 0.7181,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.54796354170701,
      "learning_rate": 9.701492537313434e-06,
      "loss": 1.2102,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.76083381012228,
      "learning_rate": 9.850746268656717e-06,
      "loss": 1.2011,
      "step": 66
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0026223995983057,
      "learning_rate": 1e-05,
      "loss": 0.7102,
      "step": 67
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.094223791176943,
      "learning_rate": 1.0149253731343284e-05,
      "loss": 0.6196,
      "step": 68
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9512351669416703,
      "learning_rate": 1.029850746268657e-05,
      "loss": 0.6477,
      "step": 69
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9038405315829852,
      "learning_rate": 1.0447761194029851e-05,
      "loss": 0.6956,
      "step": 70
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.820321602172989,
      "learning_rate": 1.0597014925373134e-05,
      "loss": 1.1868,
      "step": 71
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5815783873738978,
      "learning_rate": 1.074626865671642e-05,
      "loss": 1.1777,
      "step": 72
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0389392584282704,
      "learning_rate": 1.0895522388059703e-05,
      "loss": 0.6616,
      "step": 73
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0390784941799085,
      "learning_rate": 1.1044776119402986e-05,
      "loss": 0.6706,
      "step": 74
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9352745210290212,
      "learning_rate": 1.1194029850746268e-05,
      "loss": 0.6358,
      "step": 75
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.254878192960396,
      "learning_rate": 1.1343283582089553e-05,
      "loss": 0.793,
      "step": 76
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.729048110143807,
      "learning_rate": 1.1492537313432836e-05,
      "loss": 1.2772,
      "step": 77
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9614935977155836,
      "learning_rate": 1.1641791044776121e-05,
      "loss": 0.6749,
      "step": 78
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.075875186123347,
      "learning_rate": 1.1791044776119405e-05,
      "loss": 0.6909,
      "step": 79
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.054127027217796,
      "learning_rate": 1.1940298507462686e-05,
      "loss": 0.7348,
      "step": 80
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1741327096762184,
      "learning_rate": 1.2089552238805971e-05,
      "loss": 0.7689,
      "step": 81
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0003862689224574,
      "learning_rate": 1.2238805970149255e-05,
      "loss": 0.7057,
      "step": 82
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1201078195196486,
      "learning_rate": 1.2388059701492538e-05,
      "loss": 0.704,
      "step": 83
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9439356366213623,
      "learning_rate": 1.2537313432835823e-05,
      "loss": 0.6027,
      "step": 84
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9095713117222013,
      "learning_rate": 1.2686567164179107e-05,
      "loss": 0.6374,
      "step": 85
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8754205809623905,
      "learning_rate": 1.2835820895522388e-05,
      "loss": 0.6112,
      "step": 86
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.97797476489062,
      "learning_rate": 1.2985074626865673e-05,
      "loss": 0.7016,
      "step": 87
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8077873361181087,
      "learning_rate": 1.3134328358208957e-05,
      "loss": 0.6458,
      "step": 88
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2447657366748563,
      "learning_rate": 1.328358208955224e-05,
      "loss": 0.5773,
      "step": 89
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.223357877404996,
      "learning_rate": 1.3432835820895525e-05,
      "loss": 0.6682,
      "step": 90
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8838084355419114,
      "learning_rate": 1.3582089552238807e-05,
      "loss": 0.6536,
      "step": 91
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.867645790913768,
      "learning_rate": 1.373134328358209e-05,
      "loss": 0.6881,
      "step": 92
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9325031220270306,
      "learning_rate": 1.3880597014925375e-05,
      "loss": 0.6487,
      "step": 93
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9166964640119357,
      "learning_rate": 1.4029850746268658e-05,
      "loss": 0.6691,
      "step": 94
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0125127979762367,
      "learning_rate": 1.4179104477611942e-05,
      "loss": 1.2089,
      "step": 95
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8472642829934545,
      "learning_rate": 1.4328358208955224e-05,
      "loss": 0.6185,
      "step": 96
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7928083624079982,
      "learning_rate": 1.4477611940298509e-05,
      "loss": 1.2286,
      "step": 97
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8800154807727514,
      "learning_rate": 1.4626865671641792e-05,
      "loss": 0.6187,
      "step": 98
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9600132769366878,
      "learning_rate": 1.4776119402985077e-05,
      "loss": 0.6779,
      "step": 99
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3612453814695775,
      "learning_rate": 1.492537313432836e-05,
      "loss": 0.7146,
      "step": 100
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7455690267612742,
      "learning_rate": 1.5074626865671642e-05,
      "loss": 1.2118,
      "step": 101
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.846487712223882,
      "learning_rate": 1.5223880597014925e-05,
      "loss": 1.2496,
      "step": 102
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.058335265979515,
      "learning_rate": 1.537313432835821e-05,
      "loss": 0.7075,
      "step": 103
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.436345883505203,
      "learning_rate": 1.5522388059701494e-05,
      "loss": 1.0699,
      "step": 104
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8555540632796816,
      "learning_rate": 1.5671641791044777e-05,
      "loss": 0.6568,
      "step": 105
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.162597215804945,
      "learning_rate": 1.582089552238806e-05,
      "loss": 0.7111,
      "step": 106
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8083061851402171,
      "learning_rate": 1.5970149253731344e-05,
      "loss": 0.6665,
      "step": 107
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1935882023752398,
      "learning_rate": 1.6119402985074627e-05,
      "loss": 0.7355,
      "step": 108
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9806957260630291,
      "learning_rate": 1.626865671641791e-05,
      "loss": 0.7064,
      "step": 109
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.005162566082128,
      "learning_rate": 1.6417910447761197e-05,
      "loss": 0.5643,
      "step": 110
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.10920526937281,
      "learning_rate": 1.6567164179104477e-05,
      "loss": 0.7235,
      "step": 111
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1018111394432974,
      "learning_rate": 1.671641791044776e-05,
      "loss": 0.7517,
      "step": 112
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9794386073009478,
      "learning_rate": 1.6865671641791048e-05,
      "loss": 0.7189,
      "step": 113
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9931037900638113,
      "learning_rate": 1.701492537313433e-05,
      "loss": 0.64,
      "step": 114
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0629449167472607,
      "learning_rate": 1.7164179104477614e-05,
      "loss": 0.7037,
      "step": 115
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.046534173606315,
      "learning_rate": 1.7313432835820894e-05,
      "loss": 1.2487,
      "step": 116
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9090380778720875,
      "learning_rate": 1.746268656716418e-05,
      "loss": 0.6258,
      "step": 117
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.7830849876862334,
      "learning_rate": 1.7611940298507464e-05,
      "loss": 1.2667,
      "step": 118
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.999111105282526,
      "learning_rate": 1.7761194029850748e-05,
      "loss": 0.6438,
      "step": 119
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.758081370065071,
      "learning_rate": 1.791044776119403e-05,
      "loss": 1.2324,
      "step": 120
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.082325409743831,
      "learning_rate": 1.8059701492537314e-05,
      "loss": 0.6122,
      "step": 121
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.01571963204768,
      "learning_rate": 1.8208955223880598e-05,
      "loss": 0.644,
      "step": 122
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.8762425550762685,
      "learning_rate": 1.835820895522388e-05,
      "loss": 1.147,
      "step": 123
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.1017986320675095,
      "learning_rate": 1.8507462686567165e-05,
      "loss": 1.3227,
      "step": 124
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0134511906283787,
      "learning_rate": 1.865671641791045e-05,
      "loss": 1.1617,
      "step": 125
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2164291964667693,
      "learning_rate": 1.8805970149253735e-05,
      "loss": 0.6422,
      "step": 126
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6614839623010313,
      "learning_rate": 1.8955223880597015e-05,
      "loss": 0.7239,
      "step": 127
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.085070792799839,
      "learning_rate": 1.9104477611940298e-05,
      "loss": 0.7053,
      "step": 128
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0755760693534637,
      "learning_rate": 1.9253731343283585e-05,
      "loss": 0.641,
      "step": 129
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.059628759675678,
      "learning_rate": 1.9402985074626868e-05,
      "loss": 0.6641,
      "step": 130
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.879687934768634,
      "learning_rate": 1.955223880597015e-05,
      "loss": 0.5982,
      "step": 131
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1753737228565355,
      "learning_rate": 1.9701492537313435e-05,
      "loss": 0.6434,
      "step": 132
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.157136026009731,
      "learning_rate": 1.9850746268656718e-05,
      "loss": 1.1928,
      "step": 133
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3855290081148066,
      "learning_rate": 2e-05,
      "loss": 0.7085,
      "step": 134
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.005059496613934,
      "learning_rate": 1.9999997349620096e-05,
      "loss": 1.2848,
      "step": 135
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2398611897093574,
      "learning_rate": 1.999998939848178e-05,
      "loss": 0.6902,
      "step": 136
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9216682129844016,
      "learning_rate": 1.9999976146589266e-05,
      "loss": 0.6558,
      "step": 137
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.150061843616105,
      "learning_rate": 1.9999957593949586e-05,
      "loss": 0.7076,
      "step": 138
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0545608924408776,
      "learning_rate": 1.999993374057257e-05,
      "loss": 0.5651,
      "step": 139
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.063482132054999,
      "learning_rate": 1.9999904586470863e-05,
      "loss": 0.6454,
      "step": 140
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.278053533650964,
      "learning_rate": 1.9999870131659914e-05,
      "loss": 1.2316,
      "step": 141
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.360202477568926,
      "learning_rate": 1.9999830376157994e-05,
      "loss": 1.2373,
      "step": 142
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1794339564222254,
      "learning_rate": 1.9999785319986174e-05,
      "loss": 0.7005,
      "step": 143
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.279792531367325,
      "learning_rate": 1.9999734963168334e-05,
      "loss": 0.6377,
      "step": 144
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2530836176772344,
      "learning_rate": 1.9999679305731172e-05,
      "loss": 0.674,
      "step": 145
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3900192900750286,
      "learning_rate": 1.9999618347704186e-05,
      "loss": 0.632,
      "step": 146
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2063794298978077,
      "learning_rate": 1.9999552089119693e-05,
      "loss": 0.6311,
      "step": 147
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.8910056455253854,
      "learning_rate": 1.999948053001281e-05,
      "loss": 1.2441,
      "step": 148
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.7345903389102513,
      "learning_rate": 1.9999403670421477e-05,
      "loss": 1.2174,
      "step": 149
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.043035627852491,
      "learning_rate": 1.9999321510386425e-05,
      "loss": 0.622,
      "step": 150
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.14695732703489,
      "learning_rate": 1.999923404995121e-05,
      "loss": 1.1773,
      "step": 151
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0673119689952775,
      "learning_rate": 1.9999141289162193e-05,
      "loss": 0.6358,
      "step": 152
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3040077313327543,
      "learning_rate": 1.999904322806854e-05,
      "loss": 0.6702,
      "step": 153
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.070018298326846,
      "learning_rate": 1.999893986672224e-05,
      "loss": 0.6223,
      "step": 154
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.78923439998937,
      "learning_rate": 1.9998831205178075e-05,
      "loss": 0.5866,
      "step": 155
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8022132516965002,
      "learning_rate": 1.999871724349365e-05,
      "loss": 0.6765,
      "step": 156
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.025006329850373,
      "learning_rate": 1.999859798172936e-05,
      "loss": 0.6918,
      "step": 157
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.975487667188271,
      "learning_rate": 1.999847341994844e-05,
      "loss": 0.6407,
      "step": 158
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.7751275858741615,
      "learning_rate": 1.9998343558216907e-05,
      "loss": 1.2175,
      "step": 159
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.031844917400032,
      "learning_rate": 1.99982083966036e-05,
      "loss": 0.6387,
      "step": 160
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.080519609285389,
      "learning_rate": 1.9998067935180167e-05,
      "loss": 0.7571,
      "step": 161
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.813770522162203,
      "learning_rate": 1.9997922174021056e-05,
      "loss": 1.1641,
      "step": 162
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.905864606152831,
      "learning_rate": 1.999777111320354e-05,
      "loss": 1.2243,
      "step": 163
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.742367225948728,
      "learning_rate": 1.9997614752807686e-05,
      "loss": 1.2472,
      "step": 164
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0303924937444684,
      "learning_rate": 1.9997453092916385e-05,
      "loss": 0.6584,
      "step": 165
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4647758643909166,
      "learning_rate": 1.9997286133615318e-05,
      "loss": 1.2758,
      "step": 166
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.8879657109175696,
      "learning_rate": 1.9997113874992998e-05,
      "loss": 1.2275,
      "step": 167
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7575986145854856,
      "learning_rate": 1.9996936317140726e-05,
      "loss": 0.5709,
      "step": 168
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0534911501195454,
      "learning_rate": 1.9996753460152623e-05,
      "loss": 0.6546,
      "step": 169
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0971035278486743,
      "learning_rate": 1.9996565304125623e-05,
      "loss": 0.5882,
      "step": 170
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.003449755394488,
      "learning_rate": 1.999637184915946e-05,
      "loss": 0.6406,
      "step": 171
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.944827342430971,
      "learning_rate": 1.999617309535667e-05,
      "loss": 0.69,
      "step": 172
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8806206031251016,
      "learning_rate": 1.999596904282262e-05,
      "loss": 0.723,
      "step": 173
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.3417299211787035,
      "learning_rate": 1.9995759691665472e-05,
      "loss": 1.2348,
      "step": 174
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.018486096653129,
      "learning_rate": 1.9995545041996193e-05,
      "loss": 1.2681,
      "step": 175
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.2990912450764407,
      "learning_rate": 1.9995325093928568e-05,
      "loss": 1.1645,
      "step": 176
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4356784740536943,
      "learning_rate": 1.9995099847579184e-05,
      "loss": 0.6263,
      "step": 177
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2135481091846176,
      "learning_rate": 1.9994869303067436e-05,
      "loss": 0.592,
      "step": 178
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5203563748962265,
      "learning_rate": 1.9994633460515537e-05,
      "loss": 0.6423,
      "step": 179
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1122657802531006,
      "learning_rate": 1.9994392320048493e-05,
      "loss": 0.6431,
      "step": 180
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.3988722026182003,
      "learning_rate": 1.9994145881794134e-05,
      "loss": 1.2604,
      "step": 181
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0468040868831148,
      "learning_rate": 1.999389414588309e-05,
      "loss": 0.6242,
      "step": 182
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0731459229643527,
      "learning_rate": 1.9993637112448796e-05,
      "loss": 0.6515,
      "step": 183
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.210990859176942,
      "learning_rate": 1.9993374781627502e-05,
      "loss": 1.2393,
      "step": 184
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1959733381098876,
      "learning_rate": 1.9993107153558263e-05,
      "loss": 0.687,
      "step": 185
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.8503346170876216,
      "learning_rate": 1.9992834228382945e-05,
      "loss": 1.2165,
      "step": 186
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5711667546196515,
      "learning_rate": 1.999255600624621e-05,
      "loss": 1.1752,
      "step": 187
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1944624395841097,
      "learning_rate": 1.999227248729555e-05,
      "loss": 0.6252,
      "step": 188
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.17564319756526,
      "learning_rate": 1.9991983671681238e-05,
      "loss": 0.7695,
      "step": 189
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.910711271741687,
      "learning_rate": 1.999168955955638e-05,
      "loss": 0.5838,
      "step": 190
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.111343036583741,
      "learning_rate": 1.9991390151076873e-05,
      "loss": 0.6756,
      "step": 191
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9627311177188662,
      "learning_rate": 1.999108544640142e-05,
      "loss": 0.6644,
      "step": 192
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8113455414920878,
      "learning_rate": 1.999077544569155e-05,
      "loss": 0.6179,
      "step": 193
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.10485124478239,
      "learning_rate": 1.9990460149111577e-05,
      "loss": 0.6366,
      "step": 194
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9251939046909012,
      "learning_rate": 1.9990139556828638e-05,
      "loss": 0.5893,
      "step": 195
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8888459486778963,
      "learning_rate": 1.9989813669012665e-05,
      "loss": 0.6185,
      "step": 196
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8722345763352337,
      "learning_rate": 1.998948248583641e-05,
      "loss": 0.5651,
      "step": 197
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.2596811672896964,
      "learning_rate": 1.9989146007475426e-05,
      "loss": 1.2222,
      "step": 198
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8706026761681467,
      "learning_rate": 1.9988804234108063e-05,
      "loss": 0.5972,
      "step": 199
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9672153683727858,
      "learning_rate": 1.9988457165915495e-05,
      "loss": 0.7156,
      "step": 200
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8331038267001016,
      "learning_rate": 1.998810480308169e-05,
      "loss": 0.6722,
      "step": 201
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.80948568971521,
      "learning_rate": 1.998774714579343e-05,
      "loss": 0.6069,
      "step": 202
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.039901552010398,
      "learning_rate": 1.99873841942403e-05,
      "loss": 1.239,
      "step": 203
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5513385795781547,
      "learning_rate": 1.9987015948614693e-05,
      "loss": 1.1588,
      "step": 204
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8965634279722858,
      "learning_rate": 1.9986642409111802e-05,
      "loss": 0.6016,
      "step": 205
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9062882521447402,
      "learning_rate": 1.9986263575929635e-05,
      "loss": 0.6372,
      "step": 206
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.32681826298641,
      "learning_rate": 1.9985879449269006e-05,
      "loss": 0.7325,
      "step": 207
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8958997698244993,
      "learning_rate": 1.9985490029333526e-05,
      "loss": 0.6597,
      "step": 208
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9285222731619027,
      "learning_rate": 1.9985095316329617e-05,
      "loss": 0.5896,
      "step": 209
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.845460537483003,
      "learning_rate": 1.9984695310466508e-05,
      "loss": 1.1768,
      "step": 210
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9729966486435735,
      "learning_rate": 1.9984290011956236e-05,
      "loss": 0.6824,
      "step": 211
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7846998144298152,
      "learning_rate": 1.9983879421013636e-05,
      "loss": 0.6568,
      "step": 212
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6458790285450875,
      "learning_rate": 1.9983463537856355e-05,
      "loss": 1.1715,
      "step": 213
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6210501194708997,
      "learning_rate": 1.9983042362704837e-05,
      "loss": 1.1822,
      "step": 214
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.069387606725775,
      "learning_rate": 1.9982615895782348e-05,
      "loss": 0.6654,
      "step": 215
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0505403877218584,
      "learning_rate": 1.9982184137314936e-05,
      "loss": 0.6282,
      "step": 216
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.098131061494969,
      "learning_rate": 1.998174708753147e-05,
      "loss": 0.6061,
      "step": 217
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2733811082934436,
      "learning_rate": 1.9981304746663625e-05,
      "loss": 0.7129,
      "step": 218
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.9562207238386162,
      "learning_rate": 1.9980857114945867e-05,
      "loss": 1.1841,
      "step": 219
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.7183162445870677,
      "learning_rate": 1.998040419261548e-05,
      "loss": 1.2262,
      "step": 220
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7331673373955951,
      "learning_rate": 1.9979945979912543e-05,
      "loss": 0.6124,
      "step": 221
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0629119251368904,
      "learning_rate": 1.997948247707995e-05,
      "loss": 0.6888,
      "step": 222
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.9507467211460763,
      "learning_rate": 1.9979013684363385e-05,
      "loss": 1.1466,
      "step": 223
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9767246804729763,
      "learning_rate": 1.997853960201135e-05,
      "loss": 0.5793,
      "step": 224
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6650053017014874,
      "learning_rate": 1.997806023027514e-05,
      "loss": 0.5505,
      "step": 225
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.011270898968324,
      "learning_rate": 1.997757556940886e-05,
      "loss": 0.6886,
      "step": 226
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8826543090077328,
      "learning_rate": 1.9977085619669424e-05,
      "loss": 0.6409,
      "step": 227
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.122463930027374,
      "learning_rate": 1.997659038131653e-05,
      "loss": 0.6232,
      "step": 228
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7053041174106458,
      "learning_rate": 1.9976089854612703e-05,
      "loss": 0.5638,
      "step": 229
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.950131242216505,
      "learning_rate": 1.9975584039823253e-05,
      "loss": 0.729,
      "step": 230
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8674052550958435,
      "learning_rate": 1.9975072937216306e-05,
      "loss": 0.6435,
      "step": 231
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8798758023946172,
      "learning_rate": 1.9974556547062778e-05,
      "loss": 0.6814,
      "step": 232
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9590556598286692,
      "learning_rate": 1.99740348696364e-05,
      "loss": 0.6641,
      "step": 233
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.771069626990221,
      "learning_rate": 1.9973507905213703e-05,
      "loss": 0.6984,
      "step": 234
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0615627844499507,
      "learning_rate": 1.9972975654074012e-05,
      "loss": 0.7041,
      "step": 235
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9217191100886912,
      "learning_rate": 1.9972438116499464e-05,
      "loss": 0.6381,
      "step": 236
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.111212715831134,
      "learning_rate": 1.9971895292774994e-05,
      "loss": 1.2126,
      "step": 237
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0603226976592115,
      "learning_rate": 1.9971347183188342e-05,
      "loss": 0.6062,
      "step": 238
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9060274234225754,
      "learning_rate": 1.9970793788030044e-05,
      "loss": 0.6048,
      "step": 239
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.229194416089407,
      "learning_rate": 1.997023510759344e-05,
      "loss": 0.6555,
      "step": 240
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3206311450879578,
      "learning_rate": 1.9969671142174684e-05,
      "loss": 0.7078,
      "step": 241
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5704849714031872,
      "learning_rate": 1.9969101892072706e-05,
      "loss": 0.6368,
      "step": 242
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.916174995993769,
      "learning_rate": 1.9968527357589264e-05,
      "loss": 1.201,
      "step": 243
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.136574839147845,
      "learning_rate": 1.9967947539028895e-05,
      "loss": 0.7384,
      "step": 244
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3715205558859056,
      "learning_rate": 1.9967362436698956e-05,
      "loss": 0.7345,
      "step": 245
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.077965378229934,
      "learning_rate": 1.9966772050909587e-05,
      "loss": 1.2166,
      "step": 246
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1392892540364064,
      "learning_rate": 1.9966176381973746e-05,
      "loss": 0.6151,
      "step": 247
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9330146937221815,
      "learning_rate": 1.9965575430207176e-05,
      "loss": 0.6286,
      "step": 248
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.055529626608731,
      "learning_rate": 1.996496919592843e-05,
      "loss": 0.6454,
      "step": 249
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.2934196962240336,
      "learning_rate": 1.9964357679458857e-05,
      "loss": 1.1561,
      "step": 250
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.8117437959749205,
      "learning_rate": 1.9963740881122608e-05,
      "loss": 1.1751,
      "step": 251
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7644620189420648,
      "learning_rate": 1.996311880124663e-05,
      "loss": 0.6195,
      "step": 252
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9596750119522368,
      "learning_rate": 1.9962491440160677e-05,
      "loss": 0.6617,
      "step": 253
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.921222482585801,
      "learning_rate": 1.99618587981973e-05,
      "loss": 0.6106,
      "step": 254
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2752732292708346,
      "learning_rate": 1.9961220875691842e-05,
      "loss": 0.6102,
      "step": 255
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1105393071537155,
      "learning_rate": 1.996057767298245e-05,
      "loss": 0.7667,
      "step": 256
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.828224684870718,
      "learning_rate": 1.9959929190410074e-05,
      "loss": 1.1278,
      "step": 257
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.821372942188169,
      "learning_rate": 1.9959275428318458e-05,
      "loss": 0.6065,
      "step": 258
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7369432561209013,
      "learning_rate": 1.9958616387054147e-05,
      "loss": 0.6302,
      "step": 259
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.173586156523818,
      "learning_rate": 1.995795206696648e-05,
      "loss": 1.1598,
      "step": 260
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1246022594019114,
      "learning_rate": 1.9957282468407594e-05,
      "loss": 0.6821,
      "step": 261
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8828509707436036,
      "learning_rate": 1.9956607591732437e-05,
      "loss": 0.6666,
      "step": 262
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9215836446372228,
      "learning_rate": 1.9955927437298733e-05,
      "loss": 0.6023,
      "step": 263
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0353091605252387,
      "learning_rate": 1.9955242005467025e-05,
      "loss": 0.6896,
      "step": 264
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.118041093159659,
      "learning_rate": 1.995455129660064e-05,
      "loss": 0.6776,
      "step": 265
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0383073569482577,
      "learning_rate": 1.995385531106571e-05,
      "loss": 0.6435,
      "step": 266
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8107529993564448,
      "learning_rate": 1.9953154049231152e-05,
      "loss": 0.6377,
      "step": 267
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8927516763285153,
      "learning_rate": 1.9952447511468696e-05,
      "loss": 0.65,
      "step": 268
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6655708978747543,
      "learning_rate": 1.995173569815286e-05,
      "loss": 0.5804,
      "step": 269
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9871822542922586,
      "learning_rate": 1.9951018609660954e-05,
      "loss": 0.6637,
      "step": 270
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.046084588566069,
      "learning_rate": 1.9950296246373096e-05,
      "loss": 0.6343,
      "step": 271
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8692499256779775,
      "learning_rate": 1.9949568608672186e-05,
      "loss": 0.6185,
      "step": 272
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9143065609431147,
      "learning_rate": 1.9948835696943934e-05,
      "loss": 0.623,
      "step": 273
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.614575819397116,
      "learning_rate": 1.9948097511576832e-05,
      "loss": 0.5375,
      "step": 274
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7598274856409284,
      "learning_rate": 1.994735405296218e-05,
      "loss": 0.5661,
      "step": 275
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.090736749330294,
      "learning_rate": 1.9946605321494066e-05,
      "loss": 1.2063,
      "step": 276
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.9131914302176067,
      "learning_rate": 1.9945851317569377e-05,
      "loss": 1.1934,
      "step": 277
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7732565377789304,
      "learning_rate": 1.9945092041587788e-05,
      "loss": 0.6691,
      "step": 278
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8046695589562753,
      "learning_rate": 1.9944327493951775e-05,
      "loss": 0.5975,
      "step": 279
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.378931425510121,
      "learning_rate": 1.9943557675066603e-05,
      "loss": 1.2158,
      "step": 280
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1938543232083525,
      "learning_rate": 1.994278258534034e-05,
      "loss": 0.6563,
      "step": 281
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.715526320422855,
      "learning_rate": 1.994200222518384e-05,
      "loss": 0.6025,
      "step": 282
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9462874576154647,
      "learning_rate": 1.9941216595010754e-05,
      "loss": 0.6376,
      "step": 283
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1346701831130153,
      "learning_rate": 1.9940425695237523e-05,
      "loss": 0.6337,
      "step": 284
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.934286258335673,
      "learning_rate": 1.9939629526283387e-05,
      "loss": 0.6255,
      "step": 285
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.841669210782983,
      "learning_rate": 1.9938828088570376e-05,
      "loss": 0.5949,
      "step": 286
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9909007428765866,
      "learning_rate": 1.993802138252331e-05,
      "loss": 0.6623,
      "step": 287
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0080178690743797,
      "learning_rate": 1.993720940856981e-05,
      "loss": 0.6533,
      "step": 288
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7666722780270652,
      "learning_rate": 1.9936392167140277e-05,
      "loss": 0.6579,
      "step": 289
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7112318183396484,
      "learning_rate": 1.9935569658667912e-05,
      "loss": 0.6298,
      "step": 290
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8319762209811619,
      "learning_rate": 1.9934741883588714e-05,
      "loss": 0.6873,
      "step": 291
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9705718223575701,
      "learning_rate": 1.993390884234146e-05,
      "loss": 0.5485,
      "step": 292
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0997640293704984,
      "learning_rate": 1.9933070535367725e-05,
      "loss": 0.7111,
      "step": 293
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.196001522195831,
      "learning_rate": 1.993222696311188e-05,
      "loss": 1.2222,
      "step": 294
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7976898054166033,
      "learning_rate": 1.993137812602108e-05,
      "loss": 0.6437,
      "step": 295
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6763837425641976,
      "learning_rate": 1.9930524024545274e-05,
      "loss": 1.1894,
      "step": 296
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8941528655372184,
      "learning_rate": 1.9929664659137192e-05,
      "loss": 1.2038,
      "step": 297
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.076651425678732,
      "learning_rate": 1.992880003025238e-05,
      "loss": 0.6841,
      "step": 298
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9913096662288945,
      "learning_rate": 1.9927930138349142e-05,
      "loss": 0.5558,
      "step": 299
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.863869361934061,
      "learning_rate": 1.9927054983888598e-05,
      "loss": 0.5671,
      "step": 300
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0534931814127275,
      "learning_rate": 1.9926174567334636e-05,
      "loss": 0.5695,
      "step": 301
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.0916460773198944,
      "learning_rate": 1.992528888915395e-05,
      "loss": 1.1897,
      "step": 302
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.731111522424649,
      "learning_rate": 1.9924397949816015e-05,
      "loss": 1.1944,
      "step": 303
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0714364757203967,
      "learning_rate": 1.9923501749793097e-05,
      "loss": 0.618,
      "step": 304
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3895223796630707,
      "learning_rate": 1.992260028956025e-05,
      "loss": 0.6527,
      "step": 305
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.282117881483951,
      "learning_rate": 1.9921693569595315e-05,
      "loss": 0.6116,
      "step": 306
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9702574858714912,
      "learning_rate": 1.9920781590378927e-05,
      "loss": 0.7016,
      "step": 307
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.770267653403453,
      "learning_rate": 1.99198643523945e-05,
      "loss": 0.6262,
      "step": 308
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8870652935067402,
      "learning_rate": 1.9918941856128237e-05,
      "loss": 0.6606,
      "step": 309
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8946688242327563,
      "learning_rate": 1.9918014102069137e-05,
      "loss": 0.6314,
      "step": 310
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.899280439709356,
      "learning_rate": 1.991708109070898e-05,
      "loss": 0.6272,
      "step": 311
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.844478635784078,
      "learning_rate": 1.9916142822542327e-05,
      "loss": 0.6227,
      "step": 312
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.695549839092226,
      "learning_rate": 1.991519929806654e-05,
      "loss": 0.6108,
      "step": 313
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7342610485817698,
      "learning_rate": 1.9914250517781754e-05,
      "loss": 0.5872,
      "step": 314
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7497337082318518,
      "learning_rate": 1.991329648219089e-05,
      "loss": 0.6848,
      "step": 315
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7854330880419997,
      "learning_rate": 1.991233719179967e-05,
      "loss": 0.5899,
      "step": 316
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8221051037518063,
      "learning_rate": 1.9911372647116584e-05,
      "loss": 0.5913,
      "step": 317
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.059614547479122,
      "learning_rate": 1.9910402848652912e-05,
      "loss": 0.6166,
      "step": 318
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.853074686623409,
      "learning_rate": 1.9909427796922725e-05,
      "loss": 1.2353,
      "step": 319
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7396877344896937,
      "learning_rate": 1.9908447492442874e-05,
      "loss": 0.6269,
      "step": 320
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7850436325919774,
      "learning_rate": 1.9907461935732993e-05,
      "loss": 0.5954,
      "step": 321
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0195324412770828,
      "learning_rate": 1.9906471127315504e-05,
      "loss": 0.6716,
      "step": 322
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6044907457109014,
      "learning_rate": 1.9905475067715608e-05,
      "loss": 0.5921,
      "step": 323
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8011879266020785,
      "learning_rate": 1.9904473757461295e-05,
      "loss": 0.6691,
      "step": 324
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6794450924909123,
      "learning_rate": 1.9903467197083333e-05,
      "loss": 0.6083,
      "step": 325
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7528306498332549,
      "learning_rate": 1.990245538711528e-05,
      "loss": 0.6138,
      "step": 326
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6603368108061503,
      "learning_rate": 1.9901438328093468e-05,
      "loss": 0.561,
      "step": 327
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7631676295602412,
      "learning_rate": 1.9900416020557016e-05,
      "loss": 0.5957,
      "step": 328
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8582455991798625,
      "learning_rate": 1.989938846504783e-05,
      "loss": 0.6465,
      "step": 329
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8302522937737131,
      "learning_rate": 1.9898355662110584e-05,
      "loss": 0.7004,
      "step": 330
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0306492816801636,
      "learning_rate": 1.9897317612292745e-05,
      "loss": 0.6986,
      "step": 331
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9559069042604915,
      "learning_rate": 1.9896274316144558e-05,
      "loss": 0.6297,
      "step": 332
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.063809270088481,
      "learning_rate": 1.9895225774219055e-05,
      "loss": 0.7155,
      "step": 333
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7757929173976745,
      "learning_rate": 1.9894171987072038e-05,
      "loss": 0.6303,
      "step": 334
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8291447024470837,
      "learning_rate": 1.989311295526209e-05,
      "loss": 0.6312,
      "step": 335
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7737977770239508,
      "learning_rate": 1.9892048679350585e-05,
      "loss": 0.6597,
      "step": 336
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8370829509725881,
      "learning_rate": 1.9890979159901673e-05,
      "loss": 0.687,
      "step": 337
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.944954640609632,
      "learning_rate": 1.988990439748227e-05,
      "loss": 0.6214,
      "step": 338
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6079897550645066,
      "learning_rate": 1.988882439266209e-05,
      "loss": 0.5945,
      "step": 339
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8495240546531089,
      "learning_rate": 1.9887739146013612e-05,
      "loss": 0.714,
      "step": 340
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8819722587326169,
      "learning_rate": 1.9886648658112103e-05,
      "loss": 0.659,
      "step": 341
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8290019264956314,
      "learning_rate": 1.9885552929535607e-05,
      "loss": 0.7105,
      "step": 342
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0693231805752865,
      "learning_rate": 1.9884451960864938e-05,
      "loss": 1.1426,
      "step": 343
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.895667297658207,
      "learning_rate": 1.9883345752683695e-05,
      "loss": 1.1378,
      "step": 344
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0122938529933743,
      "learning_rate": 1.9882234305578253e-05,
      "loss": 0.6072,
      "step": 345
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9056932649277842,
      "learning_rate": 1.9881117620137762e-05,
      "loss": 0.7152,
      "step": 346
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.856593553289029,
      "learning_rate": 1.987999569695415e-05,
      "loss": 1.2244,
      "step": 347
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8642030437848063,
      "learning_rate": 1.9878868536622124e-05,
      "loss": 0.6737,
      "step": 348
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.731292081116041,
      "learning_rate": 1.9877736139739162e-05,
      "loss": 1.1612,
      "step": 349
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7874162567371499,
      "learning_rate": 1.9876598506905522e-05,
      "loss": 0.6085,
      "step": 350
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9776687021755055,
      "learning_rate": 1.987545563872424e-05,
      "loss": 0.642,
      "step": 351
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5184514250882157,
      "learning_rate": 1.9874307535801112e-05,
      "loss": 0.5737,
      "step": 352
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6606483900799254,
      "learning_rate": 1.987315419874473e-05,
      "loss": 1.2345,
      "step": 353
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7323786509741863,
      "learning_rate": 1.9871995628166442e-05,
      "loss": 1.2508,
      "step": 354
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.650754258015573,
      "learning_rate": 1.987083182468039e-05,
      "loss": 1.2026,
      "step": 355
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9781554270516888,
      "learning_rate": 1.9869662788903465e-05,
      "loss": 0.674,
      "step": 356
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9023426257603882,
      "learning_rate": 1.9868488521455348e-05,
      "loss": 0.6803,
      "step": 357
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8585615702395661,
      "learning_rate": 1.98673090229585e-05,
      "loss": 0.6071,
      "step": 358
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7067461106168094,
      "learning_rate": 1.9866124294038135e-05,
      "loss": 0.6376,
      "step": 359
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6902186034235378,
      "learning_rate": 1.986493433532225e-05,
      "loss": 0.6244,
      "step": 360
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7469912701482346,
      "learning_rate": 1.986373914744162e-05,
      "loss": 0.6283,
      "step": 361
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8010694771959,
      "learning_rate": 1.9862538731029778e-05,
      "loss": 0.674,
      "step": 362
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9356930806424317,
      "learning_rate": 1.986133308672304e-05,
      "loss": 0.5929,
      "step": 363
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0978022446248445,
      "learning_rate": 1.986012221516049e-05,
      "loss": 1.1679,
      "step": 364
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5928745535562483,
      "learning_rate": 1.9858906116983975e-05,
      "loss": 0.5729,
      "step": 365
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6547982028423456,
      "learning_rate": 1.9857684792838127e-05,
      "loss": 1.2333,
      "step": 366
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7966894064114822,
      "learning_rate": 1.9856458243370337e-05,
      "loss": 0.6324,
      "step": 367
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.87491109232087,
      "learning_rate": 1.985522646923077e-05,
      "loss": 0.6045,
      "step": 368
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9116883521794503,
      "learning_rate": 1.9853989471072362e-05,
      "loss": 0.6628,
      "step": 369
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8784856823991756,
      "learning_rate": 1.9852747249550813e-05,
      "loss": 0.6528,
      "step": 370
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8749061661093194,
      "learning_rate": 1.98514998053246e-05,
      "loss": 1.1708,
      "step": 371
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9220041538325663,
      "learning_rate": 1.9850247139054953e-05,
      "loss": 0.5655,
      "step": 372
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6805354387248517,
      "learning_rate": 1.984898925140589e-05,
      "loss": 0.6989,
      "step": 373
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7645865185802045,
      "learning_rate": 1.9847726143044184e-05,
      "loss": 0.6106,
      "step": 374
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.665056807429794,
      "learning_rate": 1.9846457814639378e-05,
      "loss": 0.6567,
      "step": 375
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7586631728217,
      "learning_rate": 1.984518426686378e-05,
      "loss": 0.6726,
      "step": 376
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8245784874901188,
      "learning_rate": 1.984390550039247e-05,
      "loss": 0.6206,
      "step": 377
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8019197949565975,
      "learning_rate": 1.9842621515903292e-05,
      "loss": 0.6264,
      "step": 378
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8317386884986628,
      "learning_rate": 1.9841332314076856e-05,
      "loss": 0.6906,
      "step": 379
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8388708254011201,
      "learning_rate": 1.9840037895596532e-05,
      "loss": 0.6017,
      "step": 380
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.071740050647249,
      "learning_rate": 1.9838738261148465e-05,
      "loss": 0.6488,
      "step": 381
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.731607976148078,
      "learning_rate": 1.9837433411421556e-05,
      "loss": 0.6361,
      "step": 382
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8408566675757874,
      "learning_rate": 1.9836123347107482e-05,
      "loss": 0.6258,
      "step": 383
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.682692573898136,
      "learning_rate": 1.9834808068900666e-05,
      "loss": 0.5691,
      "step": 384
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8926777470697658,
      "learning_rate": 1.9833487577498312e-05,
      "loss": 0.6381,
      "step": 385
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.747614396757527,
      "learning_rate": 1.983216187360038e-05,
      "loss": 0.5861,
      "step": 386
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8979937038774366,
      "learning_rate": 1.9830830957909596e-05,
      "loss": 0.6067,
      "step": 387
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8588065721548537,
      "learning_rate": 1.982949483113144e-05,
      "loss": 0.6376,
      "step": 388
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.150787132750063,
      "learning_rate": 1.9828153493974164e-05,
      "loss": 1.2461,
      "step": 389
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8575256927051824,
      "learning_rate": 1.9826806947148783e-05,
      "loss": 0.5412,
      "step": 390
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4711926317324084,
      "learning_rate": 1.9825455191369065e-05,
      "loss": 1.1682,
      "step": 391
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9323228310423761,
      "learning_rate": 1.982409822735154e-05,
      "loss": 0.5876,
      "step": 392
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.011076169471263,
      "learning_rate": 1.982273605581551e-05,
      "loss": 0.6755,
      "step": 393
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7819873559820911,
      "learning_rate": 1.982136867748302e-05,
      "loss": 0.6076,
      "step": 394
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.955507209153146,
      "learning_rate": 1.981999609307889e-05,
      "loss": 1.1526,
      "step": 395
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.821727031066249,
      "learning_rate": 1.98186183033307e-05,
      "loss": 0.6423,
      "step": 396
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8217992339443552,
      "learning_rate": 1.9817235308968772e-05,
      "loss": 0.6283,
      "step": 397
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9241913086196663,
      "learning_rate": 1.9815847110726203e-05,
      "loss": 0.6413,
      "step": 398
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.272360955409776,
      "learning_rate": 1.9814453709338844e-05,
      "loss": 1.2319,
      "step": 399
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.911110536004943,
      "learning_rate": 1.98130551055453e-05,
      "loss": 0.696,
      "step": 400
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9479304307362895,
      "learning_rate": 1.9811651300086946e-05,
      "loss": 0.5999,
      "step": 401
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7768718117932525,
      "learning_rate": 1.9810242293707896e-05,
      "loss": 0.7093,
      "step": 402
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7303326624758413,
      "learning_rate": 1.9808828087155034e-05,
      "loss": 0.6437,
      "step": 403
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6963702212471425,
      "learning_rate": 1.9807408681178e-05,
      "loss": 0.5925,
      "step": 404
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7696972559190178,
      "learning_rate": 1.980598407652918e-05,
      "loss": 0.6347,
      "step": 405
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.645874529875309,
      "learning_rate": 1.9804554273963728e-05,
      "loss": 0.5809,
      "step": 406
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.71960593593967,
      "learning_rate": 1.9803119274239546e-05,
      "loss": 0.5855,
      "step": 407
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.9112527429960537,
      "learning_rate": 1.9801679078117298e-05,
      "loss": 1.169,
      "step": 408
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8493277651845803,
      "learning_rate": 1.980023368636039e-05,
      "loss": 0.6772,
      "step": 409
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6018528781127475,
      "learning_rate": 1.9798783099734993e-05,
      "loss": 1.2126,
      "step": 410
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9368360332898713,
      "learning_rate": 1.9797327319010028e-05,
      "loss": 0.6522,
      "step": 411
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0887370821850273,
      "learning_rate": 1.9795866344957168e-05,
      "loss": 1.1049,
      "step": 412
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.572820200805713,
      "learning_rate": 1.9794400178350842e-05,
      "loss": 1.2848,
      "step": 413
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8834557886572172,
      "learning_rate": 1.979292881996823e-05,
      "loss": 0.6416,
      "step": 414
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.53426927122438,
      "learning_rate": 1.979145227058926e-05,
      "loss": 1.206,
      "step": 415
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0061652908688057,
      "learning_rate": 1.9789970530996623e-05,
      "loss": 1.1975,
      "step": 416
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.921529200332813,
      "learning_rate": 1.9788483601975743e-05,
      "loss": 0.5681,
      "step": 417
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.861647673680284,
      "learning_rate": 1.9786991484314816e-05,
      "loss": 1.1938,
      "step": 418
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7950466913131442,
      "learning_rate": 1.978549417880477e-05,
      "loss": 0.6239,
      "step": 419
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.006826543260303,
      "learning_rate": 1.9783991686239295e-05,
      "loss": 0.6576,
      "step": 420
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7339940408544992,
      "learning_rate": 1.9782484007414823e-05,
      "loss": 0.6399,
      "step": 421
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9580713264125729,
      "learning_rate": 1.9780971143130543e-05,
      "loss": 0.5778,
      "step": 422
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9317974011776613,
      "learning_rate": 1.9779453094188384e-05,
      "loss": 0.5758,
      "step": 423
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9863066803717309,
      "learning_rate": 1.9777929861393027e-05,
      "loss": 0.6856,
      "step": 424
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7589149422443384,
      "learning_rate": 1.9776401445551902e-05,
      "loss": 0.5582,
      "step": 425
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8268374797814138,
      "learning_rate": 1.9774867847475186e-05,
      "loss": 0.6345,
      "step": 426
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8781028898574537,
      "learning_rate": 1.9773329067975805e-05,
      "loss": 0.6246,
      "step": 427
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.883708227502277,
      "learning_rate": 1.9771785107869427e-05,
      "loss": 1.201,
      "step": 428
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9124593012464222,
      "learning_rate": 1.977023596797446e-05,
      "loss": 0.6972,
      "step": 429
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9023145913620827,
      "learning_rate": 1.9768681649112082e-05,
      "loss": 0.5744,
      "step": 430
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.719010055288642,
      "learning_rate": 1.9767122152106188e-05,
      "loss": 1.2141,
      "step": 431
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.817428013427065,
      "learning_rate": 1.9765557477783437e-05,
      "loss": 0.6138,
      "step": 432
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.720330420374574,
      "learning_rate": 1.976398762697322e-05,
      "loss": 0.6318,
      "step": 433
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7194618549072198,
      "learning_rate": 1.9762412600507676e-05,
      "loss": 0.6099,
      "step": 434
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8096427950875988,
      "learning_rate": 1.9760832399221698e-05,
      "loss": 0.6433,
      "step": 435
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.744145886961996,
      "learning_rate": 1.97592470239529e-05,
      "loss": 0.6389,
      "step": 436
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.628315992486051,
      "learning_rate": 1.9757656475541663e-05,
      "loss": 0.6164,
      "step": 437
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7784069162355873,
      "learning_rate": 1.9756060754831093e-05,
      "loss": 0.6315,
      "step": 438
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.7593968036638885,
      "learning_rate": 1.975445986266704e-05,
      "loss": 1.1319,
      "step": 439
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.567481327014898,
      "learning_rate": 1.9752853799898105e-05,
      "loss": 1.1868,
      "step": 440
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8558838611792783,
      "learning_rate": 1.9751242567375617e-05,
      "loss": 0.6362,
      "step": 441
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.053615350834657,
      "learning_rate": 1.974962616595366e-05,
      "loss": 0.6579,
      "step": 442
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.8640668128696745,
      "learning_rate": 1.974800459648904e-05,
      "loss": 1.1827,
      "step": 443
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0335771831180267,
      "learning_rate": 1.974637785984132e-05,
      "loss": 0.6303,
      "step": 444
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.654389677100017,
      "learning_rate": 1.974474595687279e-05,
      "loss": 0.5567,
      "step": 445
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.666334889939127,
      "learning_rate": 1.974310888844848e-05,
      "loss": 0.6194,
      "step": 446
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.110897958108769,
      "learning_rate": 1.974146665543617e-05,
      "loss": 1.2347,
      "step": 447
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8938997644770843,
      "learning_rate": 1.9739819258706355e-05,
      "loss": 0.5927,
      "step": 448
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9554888346949137,
      "learning_rate": 1.9738166699132294e-05,
      "loss": 0.6488,
      "step": 449
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.635930648568153,
      "learning_rate": 1.973650897758996e-05,
      "loss": 1.2367,
      "step": 450
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9036986446114401,
      "learning_rate": 1.9734846094958074e-05,
      "loss": 0.5654,
      "step": 451
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.838845427588315,
      "learning_rate": 1.973317805211809e-05,
      "loss": 0.6749,
      "step": 452
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6843280923834856,
      "learning_rate": 1.9731504849954198e-05,
      "loss": 0.6987,
      "step": 453
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.8833683618354455,
      "learning_rate": 1.9729826489353322e-05,
      "loss": 1.2877,
      "step": 454
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7855962916812733,
      "learning_rate": 1.972814297120512e-05,
      "loss": 0.6527,
      "step": 455
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8445735629160158,
      "learning_rate": 1.9726454296401988e-05,
      "loss": 0.702,
      "step": 456
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9698539766731924,
      "learning_rate": 1.9724760465839046e-05,
      "loss": 0.592,
      "step": 457
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.668976140676766,
      "learning_rate": 1.9723061480414158e-05,
      "loss": 0.5804,
      "step": 458
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.739728723699868,
      "learning_rate": 1.9721357341027912e-05,
      "loss": 0.5846,
      "step": 459
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.149056228594489,
      "learning_rate": 1.971964804858363e-05,
      "loss": 1.2467,
      "step": 460
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9759166356044684,
      "learning_rate": 1.9717933603987377e-05,
      "loss": 0.5965,
      "step": 461
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.092037453210013,
      "learning_rate": 1.9716214008147926e-05,
      "loss": 0.6194,
      "step": 462
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7027385328194597,
      "learning_rate": 1.97144892619768e-05,
      "loss": 0.5929,
      "step": 463
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9942106018991792,
      "learning_rate": 1.9712759366388245e-05,
      "loss": 0.6641,
      "step": 464
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.733674381651157,
      "learning_rate": 1.971102432229924e-05,
      "loss": 1.1968,
      "step": 465
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9021691158323273,
      "learning_rate": 1.970928413062948e-05,
      "loss": 0.5811,
      "step": 466
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0536564541803664,
      "learning_rate": 1.9707538792301406e-05,
      "loss": 0.5955,
      "step": 467
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9566210510949646,
      "learning_rate": 1.9705788308240183e-05,
      "loss": 0.6347,
      "step": 468
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.454222678401722,
      "learning_rate": 1.9704032679373696e-05,
      "loss": 1.1532,
      "step": 469
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8101468989147187,
      "learning_rate": 1.9702271906632564e-05,
      "loss": 0.7055,
      "step": 470
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9159608118523128,
      "learning_rate": 1.9700505990950126e-05,
      "loss": 0.6623,
      "step": 471
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.6895160776725295,
      "learning_rate": 1.969873493326246e-05,
      "loss": 1.2413,
      "step": 472
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0923237529301044,
      "learning_rate": 1.969695873450835e-05,
      "loss": 0.6129,
      "step": 473
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8693115745253919,
      "learning_rate": 1.9695177395629324e-05,
      "loss": 0.588,
      "step": 474
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.0391856676829256,
      "learning_rate": 1.969339091756963e-05,
      "loss": 1.1368,
      "step": 475
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.151535207656338,
      "learning_rate": 1.9691599301276226e-05,
      "loss": 0.6843,
      "step": 476
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.6605860317739443,
      "learning_rate": 1.968980254769881e-05,
      "loss": 1.1999,
      "step": 477
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.6223106029948897,
      "learning_rate": 1.96880006577898e-05,
      "loss": 1.1289,
      "step": 478
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0275840397522606,
      "learning_rate": 1.968619363250434e-05,
      "loss": 0.6431,
      "step": 479
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0446934521983864,
      "learning_rate": 1.9684381472800275e-05,
      "loss": 0.6541,
      "step": 480
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.626619082004227,
      "learning_rate": 1.96825641796382e-05,
      "loss": 0.5879,
      "step": 481
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.9172625782589607,
      "learning_rate": 1.9680741753981416e-05,
      "loss": 1.2117,
      "step": 482
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9963734593894578,
      "learning_rate": 1.9678914196795942e-05,
      "loss": 0.5817,
      "step": 483
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9328037867334873,
      "learning_rate": 1.967708150905053e-05,
      "loss": 0.7263,
      "step": 484
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7730191562542734,
      "learning_rate": 1.967524369171664e-05,
      "loss": 0.6392,
      "step": 485
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7742391642693032,
      "learning_rate": 1.9673400745768455e-05,
      "loss": 0.644,
      "step": 486
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9587529372259582,
      "learning_rate": 1.9671552672182872e-05,
      "loss": 0.6469,
      "step": 487
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8197757974702211,
      "learning_rate": 1.9669699471939517e-05,
      "loss": 0.6404,
      "step": 488
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7852744858095524,
      "learning_rate": 1.9667841146020725e-05,
      "loss": 0.6267,
      "step": 489
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8041291711389191,
      "learning_rate": 1.9665977695411547e-05,
      "loss": 0.6306,
      "step": 490
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9037396611307151,
      "learning_rate": 1.966410912109976e-05,
      "loss": 0.6223,
      "step": 491
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8348733918754703,
      "learning_rate": 1.9662235424075843e-05,
      "loss": 0.6933,
      "step": 492
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9741399247163107,
      "learning_rate": 1.9660356605333e-05,
      "loss": 0.703,
      "step": 493
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.72269697090997,
      "learning_rate": 1.9658472665867147e-05,
      "loss": 0.6644,
      "step": 494
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.124343902155255,
      "learning_rate": 1.965658360667692e-05,
      "loss": 1.1728,
      "step": 495
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.641766248956268,
      "learning_rate": 1.965468942876366e-05,
      "loss": 1.1756,
      "step": 496
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1127753142926284,
      "learning_rate": 1.9652790133131426e-05,
      "loss": 0.6522,
      "step": 497
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.300077751965158,
      "learning_rate": 1.965088572078699e-05,
      "loss": 1.1086,
      "step": 498
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.811060376816787,
      "learning_rate": 1.9648976192739832e-05,
      "loss": 0.5866,
      "step": 499
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.711882058988913,
      "learning_rate": 1.9647061550002152e-05,
      "loss": 1.1601,
      "step": 500
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8479900676856984,
      "learning_rate": 1.964514179358885e-05,
      "loss": 0.5996,
      "step": 501
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7207601513194368,
      "learning_rate": 1.9643216924517545e-05,
      "loss": 0.5707,
      "step": 502
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.2319163852671697,
      "learning_rate": 1.9641286943808565e-05,
      "loss": 1.2266,
      "step": 503
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.978185276516353,
      "learning_rate": 1.9639351852484946e-05,
      "loss": 0.5978,
      "step": 504
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.77512729871649,
      "learning_rate": 1.9637411651572437e-05,
      "loss": 0.5728,
      "step": 505
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.7472725502904387,
      "learning_rate": 1.9635466342099486e-05,
      "loss": 1.1489,
      "step": 506
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8392463348860042,
      "learning_rate": 1.9633515925097257e-05,
      "loss": 0.5982,
      "step": 507
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7134063801588202,
      "learning_rate": 1.9631560401599622e-05,
      "loss": 0.5799,
      "step": 508
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.754438814853521,
      "learning_rate": 1.962959977264315e-05,
      "loss": 0.6293,
      "step": 509
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7463127277469122,
      "learning_rate": 1.962763403926713e-05,
      "loss": 0.587,
      "step": 510
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7268512282880915,
      "learning_rate": 1.9625663202513553e-05,
      "loss": 0.625,
      "step": 511
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8751142349110455,
      "learning_rate": 1.9623687263427102e-05,
      "loss": 0.6496,
      "step": 512
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7693974981874472,
      "learning_rate": 1.9621706223055178e-05,
      "loss": 0.6633,
      "step": 513
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7022454503197872,
      "learning_rate": 1.961972008244789e-05,
      "loss": 0.6433,
      "step": 514
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.722187820884337,
      "learning_rate": 1.9617728842658032e-05,
      "loss": 0.6029,
      "step": 515
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9875074251144569,
      "learning_rate": 1.9615732504741123e-05,
      "loss": 0.6085,
      "step": 516
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.719620808836751,
      "learning_rate": 1.9613731069755372e-05,
      "loss": 0.5568,
      "step": 517
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9128767324431817,
      "learning_rate": 1.9611724538761682e-05,
      "loss": 0.6394,
      "step": 518
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.771780453405438,
      "learning_rate": 1.960971291282368e-05,
      "loss": 0.543,
      "step": 519
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5480600197120653,
      "learning_rate": 1.9607696193007673e-05,
      "loss": 1.2054,
      "step": 520
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.485777173595935,
      "learning_rate": 1.9605674380382676e-05,
      "loss": 1.2212,
      "step": 521
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.913093539164394,
      "learning_rate": 1.9603647476020406e-05,
      "loss": 0.6639,
      "step": 522
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8312934952251556,
      "learning_rate": 1.9601615480995275e-05,
      "loss": 0.67,
      "step": 523
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4972706084011236,
      "learning_rate": 1.9599578396384393e-05,
      "loss": 1.1393,
      "step": 524
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.854779971545699,
      "learning_rate": 1.9597536223267573e-05,
      "loss": 0.6071,
      "step": 525
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7696259025866878,
      "learning_rate": 1.959548896272732e-05,
      "loss": 0.5556,
      "step": 526
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8987661702923944,
      "learning_rate": 1.9593436615848836e-05,
      "loss": 0.6511,
      "step": 527
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5677871890195156,
      "learning_rate": 1.9591379183720026e-05,
      "loss": 1.2096,
      "step": 528
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8355386994466634,
      "learning_rate": 1.9589316667431482e-05,
      "loss": 0.6397,
      "step": 529
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8497368629671336,
      "learning_rate": 1.9587249068076494e-05,
      "loss": 0.6367,
      "step": 530
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4526079445233036,
      "learning_rate": 1.958517638675104e-05,
      "loss": 1.2118,
      "step": 531
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.653553164207118,
      "learning_rate": 1.9583098624553815e-05,
      "loss": 1.1795,
      "step": 532
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.543255466488932,
      "learning_rate": 1.9581015782586177e-05,
      "loss": 1.1227,
      "step": 533
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8462122495531341,
      "learning_rate": 1.9578927861952196e-05,
      "loss": 0.5744,
      "step": 534
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.011216130048859,
      "learning_rate": 1.9576834863758628e-05,
      "loss": 0.7432,
      "step": 535
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9475482775344386,
      "learning_rate": 1.9574736789114924e-05,
      "loss": 0.6305,
      "step": 536
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8645030527139113,
      "learning_rate": 1.9572633639133213e-05,
      "loss": 0.6876,
      "step": 537
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.000577996689892,
      "learning_rate": 1.9570525414928336e-05,
      "loss": 0.6658,
      "step": 538
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.912973626285911,
      "learning_rate": 1.9568412117617806e-05,
      "loss": 0.6583,
      "step": 539
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8087668480156571,
      "learning_rate": 1.956629374832183e-05,
      "loss": 0.6222,
      "step": 540
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7730242841022839,
      "learning_rate": 1.9564170308163308e-05,
      "loss": 0.656,
      "step": 541
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.9086026769947533,
      "learning_rate": 1.9562041798267826e-05,
      "loss": 1.1983,
      "step": 542
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8278839825911772,
      "learning_rate": 1.9559908219763648e-05,
      "loss": 0.5811,
      "step": 543
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5875051367085,
      "learning_rate": 1.955776957378174e-05,
      "loss": 1.1571,
      "step": 544
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.803344650484341,
      "learning_rate": 1.9555625861455748e-05,
      "loss": 0.6344,
      "step": 545
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.082746845493575,
      "learning_rate": 1.9553477083922e-05,
      "loss": 0.6647,
      "step": 546
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.103944983531063,
      "learning_rate": 1.9551323242319507e-05,
      "loss": 1.2506,
      "step": 547
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.9446330569607695,
      "learning_rate": 1.9549164337789974e-05,
      "loss": 1.2042,
      "step": 548
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9729821177626556,
      "learning_rate": 1.9547000371477782e-05,
      "loss": 0.6369,
      "step": 549
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9114591312830038,
      "learning_rate": 1.954483134453e-05,
      "loss": 0.6651,
      "step": 550
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.884781691282806,
      "learning_rate": 1.9542657258096374e-05,
      "loss": 1.2116,
      "step": 551
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7065165049111746,
      "learning_rate": 1.954047811332934e-05,
      "loss": 0.5718,
      "step": 552
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.2814246701480334,
      "learning_rate": 1.9538293911384002e-05,
      "loss": 1.2151,
      "step": 553
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.9357079609683177,
      "learning_rate": 1.953610465341816e-05,
      "loss": 1.1695,
      "step": 554
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1304268477586077,
      "learning_rate": 1.9533910340592287e-05,
      "loss": 0.6715,
      "step": 555
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7722788608224815,
      "learning_rate": 1.953171097406953e-05,
      "loss": 0.6053,
      "step": 556
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.848293489006093,
      "learning_rate": 1.9529506555015726e-05,
      "loss": 0.6154,
      "step": 557
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9354114548690007,
      "learning_rate": 1.9527297084599384e-05,
      "loss": 0.6292,
      "step": 558
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.858922461565216,
      "learning_rate": 1.9525082563991685e-05,
      "loss": 0.6676,
      "step": 559
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.855947040925127,
      "learning_rate": 1.9522862994366504e-05,
      "loss": 0.6732,
      "step": 560
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6863890155215429,
      "learning_rate": 1.952063837690037e-05,
      "loss": 0.6755,
      "step": 561
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0303492759166417,
      "learning_rate": 1.9518408712772507e-05,
      "loss": 1.2502,
      "step": 562
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6020013377909434,
      "learning_rate": 1.9516174003164803e-05,
      "loss": 0.6017,
      "step": 563
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7280304003571196,
      "learning_rate": 1.9513934249261823e-05,
      "loss": 0.5402,
      "step": 564
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.209700006059489,
      "learning_rate": 1.9511689452250814e-05,
      "loss": 0.6789,
      "step": 565
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0837708999079734,
      "learning_rate": 1.950943961332168e-05,
      "loss": 1.2219,
      "step": 566
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.8068292557584265,
      "learning_rate": 1.950718473366701e-05,
      "loss": 1.1775,
      "step": 567
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9885623146537932,
      "learning_rate": 1.9504924814482063e-05,
      "loss": 0.643,
      "step": 568
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.02212183389083,
      "learning_rate": 1.9502659856964764e-05,
      "loss": 1.172,
      "step": 569
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6563325166366203,
      "learning_rate": 1.950038986231572e-05,
      "loss": 0.5688,
      "step": 570
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.298082303419361,
      "learning_rate": 1.9498114831738192e-05,
      "loss": 1.1544,
      "step": 571
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.853108444813667,
      "learning_rate": 1.9495834766438125e-05,
      "loss": 1.2023,
      "step": 572
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9289220972691052,
      "learning_rate": 1.9493549667624125e-05,
      "loss": 0.6217,
      "step": 573
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7120243502985624,
      "learning_rate": 1.949125953650747e-05,
      "loss": 0.5714,
      "step": 574
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1663496817633168,
      "learning_rate": 1.9488964374302095e-05,
      "loss": 0.6645,
      "step": 575
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0197387341368462,
      "learning_rate": 1.948666418222462e-05,
      "loss": 0.6927,
      "step": 576
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8586062811834712,
      "learning_rate": 1.948435896149432e-05,
      "loss": 0.5713,
      "step": 577
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7734157874513785,
      "learning_rate": 1.9482048713333132e-05,
      "loss": 0.6276,
      "step": 578
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0707722902514885,
      "learning_rate": 1.9479733438965666e-05,
      "loss": 1.2166,
      "step": 579
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.897203642111775,
      "learning_rate": 1.94774131396192e-05,
      "loss": 0.6033,
      "step": 580
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.6473325137516968,
      "learning_rate": 1.9475087816523658e-05,
      "loss": 1.2144,
      "step": 581
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.015625398941892,
      "learning_rate": 1.947275747091164e-05,
      "loss": 0.6262,
      "step": 582
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0152782898596295,
      "learning_rate": 1.947042210401841e-05,
      "loss": 1.2128,
      "step": 583
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.428041473654515,
      "learning_rate": 1.9468081717081893e-05,
      "loss": 0.6471,
      "step": 584
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1773872370253686,
      "learning_rate": 1.9465736311342668e-05,
      "loss": 0.6659,
      "step": 585
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9987061923947134,
      "learning_rate": 1.946338588804397e-05,
      "loss": 0.6323,
      "step": 586
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.7015333350643527,
      "learning_rate": 1.946103044843172e-05,
      "loss": 1.1088,
      "step": 587
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.97336459410416,
      "learning_rate": 1.9458669993754462e-05,
      "loss": 0.6197,
      "step": 588
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9273099265101512,
      "learning_rate": 1.9456304525263426e-05,
      "loss": 0.6519,
      "step": 589
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.317599307398721,
      "learning_rate": 1.9453934044212494e-05,
      "loss": 1.1651,
      "step": 590
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7759731952523523,
      "learning_rate": 1.9451558551858188e-05,
      "loss": 0.665,
      "step": 591
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.8703215546511056,
      "learning_rate": 1.944917804945971e-05,
      "loss": 1.122,
      "step": 592
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8628482304767249,
      "learning_rate": 1.9446792538278904e-05,
      "loss": 0.5882,
      "step": 593
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.884339102752723,
      "learning_rate": 1.944440201958027e-05,
      "loss": 1.195,
      "step": 594
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.738771490733278,
      "learning_rate": 1.944200649463097e-05,
      "loss": 0.6361,
      "step": 595
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.012018585622007,
      "learning_rate": 1.9439605964700806e-05,
      "loss": 0.6597,
      "step": 596
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.327539128650693,
      "learning_rate": 1.9437200431062247e-05,
      "loss": 1.1447,
      "step": 597
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7759863055441927,
      "learning_rate": 1.943478989499041e-05,
      "loss": 0.6402,
      "step": 598
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.924147819165279,
      "learning_rate": 1.943237435776306e-05,
      "loss": 0.7102,
      "step": 599
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9388355914962594,
      "learning_rate": 1.9429953820660614e-05,
      "loss": 0.6197,
      "step": 600
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8712557790780144,
      "learning_rate": 1.9427528284966142e-05,
      "loss": 0.6552,
      "step": 601
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8561926091074796,
      "learning_rate": 1.9425097751965364e-05,
      "loss": 0.6216,
      "step": 602
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.1452059089844497,
      "learning_rate": 1.9422662222946643e-05,
      "loss": 1.1068,
      "step": 603
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.745026112814656,
      "learning_rate": 1.9420221699200997e-05,
      "loss": 0.6187,
      "step": 604
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6934285205023631,
      "learning_rate": 1.9417776182022088e-05,
      "loss": 0.6004,
      "step": 605
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.389301274577714,
      "learning_rate": 1.9415325672706227e-05,
      "loss": 1.1775,
      "step": 606
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8899876257388304,
      "learning_rate": 1.9412870172552368e-05,
      "loss": 0.6861,
      "step": 607
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8066358767917574,
      "learning_rate": 1.9410409682862113e-05,
      "loss": 0.646,
      "step": 608
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6627292777278777,
      "learning_rate": 1.9407944204939718e-05,
      "loss": 1.1791,
      "step": 609
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8303730614679006,
      "learning_rate": 1.9405473740092058e-05,
      "loss": 0.6984,
      "step": 610
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9057117551444531,
      "learning_rate": 1.9402998289628678e-05,
      "loss": 0.6432,
      "step": 611
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.0920300728750716,
      "learning_rate": 1.940051785486175e-05,
      "loss": 1.1461,
      "step": 612
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7114991240485247,
      "learning_rate": 1.9398032437106095e-05,
      "loss": 0.6252,
      "step": 613
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9817525200379695,
      "learning_rate": 1.9395542037679172e-05,
      "loss": 0.6539,
      "step": 614
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9726975073391015,
      "learning_rate": 1.939304665790108e-05,
      "loss": 0.6907,
      "step": 615
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.774071968389767,
      "learning_rate": 1.9390546299094567e-05,
      "loss": 0.5989,
      "step": 616
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.957118273858022,
      "learning_rate": 1.9388040962585003e-05,
      "loss": 0.6499,
      "step": 617
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8894082886814645,
      "learning_rate": 1.9385530649700418e-05,
      "loss": 0.6049,
      "step": 618
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7911915347252485,
      "learning_rate": 1.938301536177146e-05,
      "loss": 0.6427,
      "step": 619
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.138629191197112,
      "learning_rate": 1.9380495100131425e-05,
      "loss": 1.2761,
      "step": 620
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6557325306021398,
      "learning_rate": 1.9377969866116242e-05,
      "loss": 0.6297,
      "step": 621
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9357250830933337,
      "learning_rate": 1.9375439661064483e-05,
      "loss": 0.6137,
      "step": 622
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7987254586685466,
      "learning_rate": 1.9372904486317343e-05,
      "loss": 0.5991,
      "step": 623
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9527488242723554,
      "learning_rate": 1.9370364343218656e-05,
      "loss": 0.6222,
      "step": 624
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9100586845877814,
      "learning_rate": 1.9367819233114895e-05,
      "loss": 0.6579,
      "step": 625
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.06362932687423,
      "learning_rate": 1.936526915735516e-05,
      "loss": 1.236,
      "step": 626
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5651284786379382,
      "learning_rate": 1.9362714117291185e-05,
      "loss": 0.6081,
      "step": 627
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9371436962853492,
      "learning_rate": 1.9360154114277337e-05,
      "loss": 0.5563,
      "step": 628
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7880993286238525,
      "learning_rate": 1.935758914967061e-05,
      "loss": 0.7233,
      "step": 629
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7819773872096971,
      "learning_rate": 1.935501922483063e-05,
      "loss": 0.6368,
      "step": 630
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.1837884032958463,
      "learning_rate": 1.9352444341119655e-05,
      "loss": 1.1147,
      "step": 631
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8481273096847435,
      "learning_rate": 1.9349864499902563e-05,
      "loss": 0.6179,
      "step": 632
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8108842506667042,
      "learning_rate": 1.9347279702546876e-05,
      "loss": 0.5959,
      "step": 633
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.897269943574509,
      "learning_rate": 1.9344689950422722e-05,
      "loss": 0.6184,
      "step": 634
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7140217343167634,
      "learning_rate": 1.9342095244902875e-05,
      "loss": 0.5962,
      "step": 635
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.794958745912123,
      "learning_rate": 1.9339495587362722e-05,
      "loss": 0.5586,
      "step": 636
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9577589407837666,
      "learning_rate": 1.933689097918028e-05,
      "loss": 0.624,
      "step": 637
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9506144382727881,
      "learning_rate": 1.9334281421736187e-05,
      "loss": 0.6362,
      "step": 638
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8669195545620452,
      "learning_rate": 1.9331666916413708e-05,
      "loss": 0.5702,
      "step": 639
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8595154080001284,
      "learning_rate": 1.932904746459873e-05,
      "loss": 0.6052,
      "step": 640
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.9133668318733474,
      "learning_rate": 1.9326423067679764e-05,
      "loss": 1.1143,
      "step": 641
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.101914275610915,
      "learning_rate": 1.9323793727047935e-05,
      "loss": 0.6181,
      "step": 642
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.192454805214121,
      "learning_rate": 1.9321159444096998e-05,
      "loss": 0.6526,
      "step": 643
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.204377696470718,
      "learning_rate": 1.9318520220223318e-05,
      "loss": 0.6406,
      "step": 644
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9343899526666066,
      "learning_rate": 1.931587605682589e-05,
      "loss": 0.6143,
      "step": 645
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7988983325064687,
      "learning_rate": 1.9313226955306314e-05,
      "loss": 0.6094,
      "step": 646
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.841346126718479,
      "learning_rate": 1.9310572917068824e-05,
      "loss": 1.1963,
      "step": 647
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.1054754160390345,
      "learning_rate": 1.9307913943520257e-05,
      "loss": 1.243,
      "step": 648
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0245579255447104,
      "learning_rate": 1.9305250036070067e-05,
      "loss": 0.6471,
      "step": 649
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.0330082343968963,
      "learning_rate": 1.9302581196130335e-05,
      "loss": 1.1927,
      "step": 650
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9777120524172098,
      "learning_rate": 1.9299907425115747e-05,
      "loss": 0.6755,
      "step": 651
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6999386262100138,
      "learning_rate": 1.92972287244436e-05,
      "loss": 0.6214,
      "step": 652
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8161448225816876,
      "learning_rate": 1.9294545095533817e-05,
      "loss": 0.6539,
      "step": 653
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8553989386842258,
      "learning_rate": 1.9291856539808917e-05,
      "loss": 0.6035,
      "step": 654
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.916223186129764,
      "learning_rate": 1.9289163058694044e-05,
      "loss": 0.7275,
      "step": 655
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.394313753078694,
      "learning_rate": 1.9286464653616942e-05,
      "loss": 1.1618,
      "step": 656
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.8037149590649695,
      "learning_rate": 1.9283761326007982e-05,
      "loss": 1.1815,
      "step": 657
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.6879202309793198,
      "learning_rate": 1.9281053077300124e-05,
      "loss": 1.2256,
      "step": 658
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.758455510297471,
      "learning_rate": 1.9278339908928943e-05,
      "loss": 0.6619,
      "step": 659
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3662180665455272,
      "learning_rate": 1.927562182233263e-05,
      "loss": 0.7185,
      "step": 660
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.7082151457732526,
      "learning_rate": 1.927289881895198e-05,
      "loss": 1.1743,
      "step": 661
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.4425776033716087,
      "learning_rate": 1.9270170900230386e-05,
      "loss": 1.1922,
      "step": 662
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.865095250637452,
      "learning_rate": 1.926743806761385e-05,
      "loss": 0.6373,
      "step": 663
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8678244082620803,
      "learning_rate": 1.9264700322550992e-05,
      "loss": 0.686,
      "step": 664
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7101664589120966,
      "learning_rate": 1.9261957666493013e-05,
      "loss": 0.5504,
      "step": 665
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7479048601878433,
      "learning_rate": 1.9259210100893737e-05,
      "loss": 0.6447,
      "step": 666
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9481696921920832,
      "learning_rate": 1.9256457627209578e-05,
      "loss": 0.6353,
      "step": 667
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.055630819940536,
      "learning_rate": 1.9253700246899562e-05,
      "loss": 0.6583,
      "step": 668
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7605474274135853,
      "learning_rate": 1.9250937961425303e-05,
      "loss": 0.6681,
      "step": 669
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1656825138411366,
      "learning_rate": 1.924817077225102e-05,
      "loss": 0.7407,
      "step": 670
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5492138449897812,
      "learning_rate": 1.9245398680843546e-05,
      "loss": 0.6227,
      "step": 671
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.4019929981462815,
      "learning_rate": 1.924262168867229e-05,
      "loss": 1.164,
      "step": 672
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.8317839936089384,
      "learning_rate": 1.9239839797209268e-05,
      "loss": 1.2347,
      "step": 673
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5057501730773124,
      "learning_rate": 1.9237053007929103e-05,
      "loss": 1.1863,
      "step": 674
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.267404071898381,
      "learning_rate": 1.9234261322309e-05,
      "loss": 1.2412,
      "step": 675
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4835985938583685,
      "learning_rate": 1.9231464741828757e-05,
      "loss": 0.6143,
      "step": 676
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5488040478896496,
      "learning_rate": 1.9228663267970785e-05,
      "loss": 0.5695,
      "step": 677
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3815235941751487,
      "learning_rate": 1.9225856902220076e-05,
      "loss": 0.6169,
      "step": 678
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9730240906366059,
      "learning_rate": 1.9223045646064215e-05,
      "loss": 0.6525,
      "step": 679
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.876506715139014,
      "learning_rate": 1.9220229500993377e-05,
      "loss": 0.6446,
      "step": 680
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8682986701772129,
      "learning_rate": 1.921740846850034e-05,
      "loss": 0.5976,
      "step": 681
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.921234422966361,
      "learning_rate": 1.9214582550080463e-05,
      "loss": 1.1784,
      "step": 682
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9742388735067549,
      "learning_rate": 1.9211751747231697e-05,
      "loss": 0.6821,
      "step": 683
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9275887405308776,
      "learning_rate": 1.9208916061454584e-05,
      "loss": 0.5624,
      "step": 684
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8744479621492562,
      "learning_rate": 1.920607549425225e-05,
      "loss": 0.594,
      "step": 685
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7873924596799802,
      "learning_rate": 1.9203230047130416e-05,
      "loss": 0.6922,
      "step": 686
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.978307227735342,
      "learning_rate": 1.920037972159738e-05,
      "loss": 0.6886,
      "step": 687
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9935818326423578,
      "learning_rate": 1.9197524519164033e-05,
      "loss": 0.6524,
      "step": 688
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.698026952070986,
      "learning_rate": 1.9194664441343853e-05,
      "loss": 1.2272,
      "step": 689
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.045347565522643,
      "learning_rate": 1.9191799489652893e-05,
      "loss": 1.1624,
      "step": 690
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4907988153887564,
      "learning_rate": 1.9188929665609797e-05,
      "loss": 0.5568,
      "step": 691
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8791611493010894,
      "learning_rate": 1.9186054970735797e-05,
      "loss": 0.639,
      "step": 692
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0149799201066325,
      "learning_rate": 1.9183175406554687e-05,
      "loss": 0.5994,
      "step": 693
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1934544056969845,
      "learning_rate": 1.9180290974592863e-05,
      "loss": 0.6478,
      "step": 694
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9904694783179653,
      "learning_rate": 1.9177401676379288e-05,
      "loss": 0.6643,
      "step": 695
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.082077730661695,
      "learning_rate": 1.9174507513445514e-05,
      "loss": 0.6406,
      "step": 696
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7567418629437344,
      "learning_rate": 1.9171608487325664e-05,
      "loss": 0.5629,
      "step": 697
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7183630166964565,
      "learning_rate": 1.916870459955645e-05,
      "loss": 0.5967,
      "step": 698
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9658355710574087,
      "learning_rate": 1.916579585167714e-05,
      "loss": 0.6785,
      "step": 699
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8660539962334004,
      "learning_rate": 1.9162882245229604e-05,
      "loss": 0.6266,
      "step": 700
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9464262423166379,
      "learning_rate": 1.9159963781758266e-05,
      "loss": 0.5445,
      "step": 701
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.808474956984881,
      "learning_rate": 1.9157040462810138e-05,
      "loss": 0.5776,
      "step": 702
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9023525122995006,
      "learning_rate": 1.91541122899348e-05,
      "loss": 0.6448,
      "step": 703
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.838386208954701,
      "learning_rate": 1.9151179264684402e-05,
      "loss": 0.632,
      "step": 704
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0304766590199232,
      "learning_rate": 1.9148241388613678e-05,
      "loss": 0.6707,
      "step": 705
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7628962451579766,
      "learning_rate": 1.9145298663279923e-05,
      "loss": 0.6017,
      "step": 706
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7637941418018683,
      "learning_rate": 1.9142351090243e-05,
      "loss": 0.6295,
      "step": 707
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.7026277477415315,
      "learning_rate": 1.9139398671065352e-05,
      "loss": 1.144,
      "step": 708
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.712767937532015,
      "learning_rate": 1.9136441407311985e-05,
      "loss": 0.5968,
      "step": 709
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7670560686357366,
      "learning_rate": 1.913347930055047e-05,
      "loss": 0.686,
      "step": 710
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.04948545266075,
      "learning_rate": 1.9130512352350953e-05,
      "loss": 0.6915,
      "step": 711
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9060801378406393,
      "learning_rate": 1.9127540564286142e-05,
      "loss": 0.5477,
      "step": 712
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3946786775545044,
      "learning_rate": 1.91245639379313e-05,
      "loss": 0.6519,
      "step": 713
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8332864300639626,
      "learning_rate": 1.9121582474864283e-05,
      "loss": 0.5787,
      "step": 714
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.2006835799193905,
      "learning_rate": 1.9118596176665478e-05,
      "loss": 1.1922,
      "step": 715
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9586379894121875,
      "learning_rate": 1.9115605044917856e-05,
      "loss": 0.6408,
      "step": 716
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6855452699551545,
      "learning_rate": 1.9112609081206948e-05,
      "loss": 0.6065,
      "step": 717
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0716209862938024,
      "learning_rate": 1.9109608287120834e-05,
      "loss": 0.65,
      "step": 718
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8869888872055065,
      "learning_rate": 1.9106602664250166e-05,
      "loss": 0.6449,
      "step": 719
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0932922969749126,
      "learning_rate": 1.9103592214188157e-05,
      "loss": 0.5863,
      "step": 720
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.1932286235221365,
      "learning_rate": 1.9100576938530567e-05,
      "loss": 1.2164,
      "step": 721
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8190629771073445,
      "learning_rate": 1.9097556838875726e-05,
      "loss": 0.5874,
      "step": 722
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3407839436018274,
      "learning_rate": 1.9094531916824518e-05,
      "loss": 1.1767,
      "step": 723
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9761215441702318,
      "learning_rate": 1.9091502173980375e-05,
      "loss": 0.5969,
      "step": 724
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.051248643265865,
      "learning_rate": 1.9088467611949295e-05,
      "loss": 0.6242,
      "step": 725
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.197888634039495,
      "learning_rate": 1.9085428232339828e-05,
      "loss": 1.2056,
      "step": 726
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1334600722138233,
      "learning_rate": 1.908238403676307e-05,
      "loss": 0.6642,
      "step": 727
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9060363904238338,
      "learning_rate": 1.9079335026832687e-05,
      "loss": 0.6266,
      "step": 728
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.770794945502175,
      "learning_rate": 1.9076281204164875e-05,
      "loss": 0.5999,
      "step": 729
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7035694502349659,
      "learning_rate": 1.9073222570378396e-05,
      "loss": 0.6661,
      "step": 730
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8361911572297496,
      "learning_rate": 1.907015912709456e-05,
      "loss": 0.5883,
      "step": 731
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.802094956224313,
      "learning_rate": 1.906709087593722e-05,
      "loss": 1.2418,
      "step": 732
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.960459000906455,
      "learning_rate": 1.9064017818532785e-05,
      "loss": 0.6457,
      "step": 733
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0875440957411313,
      "learning_rate": 1.906093995651021e-05,
      "loss": 0.6205,
      "step": 734
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7512722519982644,
      "learning_rate": 1.9057857291501e-05,
      "loss": 0.6676,
      "step": 735
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.01250498629494,
      "learning_rate": 1.9054769825139193e-05,
      "loss": 0.7041,
      "step": 736
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9184827652485237,
      "learning_rate": 1.9051677559061382e-05,
      "loss": 0.652,
      "step": 737
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.080649672081786,
      "learning_rate": 1.904858049490671e-05,
      "loss": 0.6701,
      "step": 738
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6437525784415714,
      "learning_rate": 1.904547863431685e-05,
      "loss": 0.6117,
      "step": 739
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8026317751872092,
      "learning_rate": 1.904237197893602e-05,
      "loss": 0.6953,
      "step": 740
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6845307247155374,
      "learning_rate": 1.9039260530410998e-05,
      "loss": 0.621,
      "step": 741
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7572188707916858,
      "learning_rate": 1.9036144290391073e-05,
      "loss": 0.6513,
      "step": 742
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.854382878672642,
      "learning_rate": 1.90330232605281e-05,
      "loss": 0.666,
      "step": 743
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.717645021194221,
      "learning_rate": 1.9029897442476453e-05,
      "loss": 1.1738,
      "step": 744
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7920895157417553,
      "learning_rate": 1.9026766837893058e-05,
      "loss": 0.6896,
      "step": 745
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9488212025489655,
      "learning_rate": 1.9023631448437376e-05,
      "loss": 0.6118,
      "step": 746
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.657273682527097,
      "learning_rate": 1.902049127577139e-05,
      "loss": 0.6675,
      "step": 747
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6511938376632265,
      "learning_rate": 1.9017346321559644e-05,
      "loss": 0.5895,
      "step": 748
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8089600765109153,
      "learning_rate": 1.9014196587469195e-05,
      "loss": 0.6897,
      "step": 749
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8595262276273195,
      "learning_rate": 1.9011042075169644e-05,
      "loss": 0.6568,
      "step": 750
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6197805161536711,
      "learning_rate": 1.9007882786333122e-05,
      "loss": 0.6465,
      "step": 751
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.894095408762971,
      "learning_rate": 1.9004718722634287e-05,
      "loss": 1.1276,
      "step": 752
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0566542300572643,
      "learning_rate": 1.9001549885750344e-05,
      "loss": 0.7268,
      "step": 753
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.687106024216675,
      "learning_rate": 1.8998376277361008e-05,
      "loss": 0.6439,
      "step": 754
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.6445713676603546,
      "learning_rate": 1.8995197899148536e-05,
      "loss": 1.1966,
      "step": 755
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7016047714277887,
      "learning_rate": 1.8992014752797708e-05,
      "loss": 0.596,
      "step": 756
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.157221087393897,
      "learning_rate": 1.8988826839995836e-05,
      "loss": 0.7331,
      "step": 757
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8468273515809652,
      "learning_rate": 1.8985634162432757e-05,
      "loss": 0.6353,
      "step": 758
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.98926378521132,
      "learning_rate": 1.898243672180083e-05,
      "loss": 0.6804,
      "step": 759
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9308318186608435,
      "learning_rate": 1.897923451979494e-05,
      "loss": 0.6158,
      "step": 760
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.5462304484789016,
      "learning_rate": 1.8976027558112502e-05,
      "loss": 1.1588,
      "step": 761
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7935978646670112,
      "learning_rate": 1.8972815838453445e-05,
      "loss": 0.6514,
      "step": 762
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7952337925290776,
      "learning_rate": 1.8969599362520224e-05,
      "loss": 0.682,
      "step": 763
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4465883455377013,
      "learning_rate": 1.896637813201782e-05,
      "loss": 1.1572,
      "step": 764
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.67254652543058,
      "learning_rate": 1.896315214865373e-05,
      "loss": 0.6521,
      "step": 765
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9534637252198725,
      "learning_rate": 1.895992141413797e-05,
      "loss": 0.6193,
      "step": 766
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8887577067152197,
      "learning_rate": 1.8956685930183068e-05,
      "loss": 0.64,
      "step": 767
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7048357456065857,
      "learning_rate": 1.8953445698504084e-05,
      "loss": 0.5366,
      "step": 768
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6083866830804876,
      "learning_rate": 1.8950200720818584e-05,
      "loss": 0.5963,
      "step": 769
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.7537414019840734,
      "learning_rate": 1.8946950998846652e-05,
      "loss": 1.2566,
      "step": 770
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.541228551392406,
      "learning_rate": 1.8943696534310892e-05,
      "loss": 0.5825,
      "step": 771
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3936132942430617,
      "learning_rate": 1.894043732893641e-05,
      "loss": 1.1925,
      "step": 772
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.89947571004921,
      "learning_rate": 1.8937173384450838e-05,
      "loss": 0.6049,
      "step": 773
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.8069453314428516,
      "learning_rate": 1.8933904702584315e-05,
      "loss": 1.2227,
      "step": 774
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.700904168188228,
      "learning_rate": 1.8930631285069487e-05,
      "loss": 1.1894,
      "step": 775
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.530004427583205,
      "learning_rate": 1.8927353133641513e-05,
      "loss": 1.1356,
      "step": 776
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2586690944378707,
      "learning_rate": 1.892407025003807e-05,
      "loss": 1.1997,
      "step": 777
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7327945408864875,
      "learning_rate": 1.8920782635999328e-05,
      "loss": 0.5878,
      "step": 778
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5263417064081588,
      "learning_rate": 1.8917490293267976e-05,
      "loss": 0.6378,
      "step": 779
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7141991716871732,
      "learning_rate": 1.8914193223589207e-05,
      "loss": 0.5819,
      "step": 780
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7674939584310627,
      "learning_rate": 1.891089142871071e-05,
      "loss": 0.5776,
      "step": 781
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9356498081646623,
      "learning_rate": 1.8907584910382702e-05,
      "loss": 0.5934,
      "step": 782
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7343254183411134,
      "learning_rate": 1.8904273670357874e-05,
      "loss": 0.6895,
      "step": 783
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8330249832841277,
      "learning_rate": 1.890095771039145e-05,
      "loss": 0.6278,
      "step": 784
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.818273554347792,
      "learning_rate": 1.889763703224112e-05,
      "loss": 0.6459,
      "step": 785
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.418579349879989,
      "learning_rate": 1.8894311637667114e-05,
      "loss": 1.2072,
      "step": 786
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8850791087215168,
      "learning_rate": 1.889098152843214e-05,
      "loss": 0.6816,
      "step": 787
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.891512786156873,
      "learning_rate": 1.8887646706301406e-05,
      "loss": 0.6279,
      "step": 788
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.060797139023358,
      "learning_rate": 1.888430717304262e-05,
      "loss": 1.221,
      "step": 789
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3849724927940104,
      "learning_rate": 1.8880962930425987e-05,
      "loss": 1.2115,
      "step": 790
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8566277197491887,
      "learning_rate": 1.8877613980224215e-05,
      "loss": 0.5553,
      "step": 791
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6951914041148413,
      "learning_rate": 1.8874260324212503e-05,
      "loss": 0.6185,
      "step": 792
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.003426493803629,
      "learning_rate": 1.887090196416854e-05,
      "loss": 0.6642,
      "step": 793
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7581394096098089,
      "learning_rate": 1.886753890187251e-05,
      "loss": 0.5909,
      "step": 794
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.1817309623164878,
      "learning_rate": 1.886417113910709e-05,
      "loss": 1.1681,
      "step": 795
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.718038831191041,
      "learning_rate": 1.8860798677657463e-05,
      "loss": 1.1891,
      "step": 796
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5596269505386764,
      "learning_rate": 1.8857421519311273e-05,
      "loss": 0.61,
      "step": 797
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.385014625073786,
      "learning_rate": 1.885403966585868e-05,
      "loss": 1.1905,
      "step": 798
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6930467813661074,
      "learning_rate": 1.8850653119092325e-05,
      "loss": 0.6399,
      "step": 799
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6542191669836435,
      "learning_rate": 1.8847261880807327e-05,
      "loss": 0.5928,
      "step": 800
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.915010259141638,
      "learning_rate": 1.884386595280131e-05,
      "loss": 0.6285,
      "step": 801
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7298970298572087,
      "learning_rate": 1.8840465336874362e-05,
      "loss": 0.5764,
      "step": 802
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9222535646319536,
      "learning_rate": 1.883706003482908e-05,
      "loss": 0.6474,
      "step": 803
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.2182114929995325,
      "learning_rate": 1.8833650048470524e-05,
      "loss": 1.1999,
      "step": 804
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0254456802966616,
      "learning_rate": 1.8830235379606252e-05,
      "loss": 0.6645,
      "step": 805
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9326164058497326,
      "learning_rate": 1.8826816030046293e-05,
      "loss": 0.6243,
      "step": 806
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9536896729121564,
      "learning_rate": 1.8823392001603164e-05,
      "loss": 0.6244,
      "step": 807
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8691299111020905,
      "learning_rate": 1.8819963296091857e-05,
      "loss": 0.645,
      "step": 808
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.3801907789553565,
      "learning_rate": 1.8816529915329855e-05,
      "loss": 1.1626,
      "step": 809
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9054422415721428,
      "learning_rate": 1.8813091861137106e-05,
      "loss": 0.5775,
      "step": 810
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.6283221599786524,
      "learning_rate": 1.8809649135336035e-05,
      "loss": 1.2439,
      "step": 811
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.932973165407275,
      "learning_rate": 1.8806201739751555e-05,
      "loss": 0.6257,
      "step": 812
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7046668843072235,
      "learning_rate": 1.8802749676211046e-05,
      "loss": 0.6106,
      "step": 813
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6423275490096885,
      "learning_rate": 1.8799292946544363e-05,
      "loss": 0.6466,
      "step": 814
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.69769225822137,
      "learning_rate": 1.8795831552583832e-05,
      "loss": 0.5863,
      "step": 815
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8332810745687536,
      "learning_rate": 1.879236549616426e-05,
      "loss": 0.6522,
      "step": 816
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.0673237204661867,
      "learning_rate": 1.878889477912292e-05,
      "loss": 1.1768,
      "step": 817
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.5524259336536823,
      "learning_rate": 1.8785419403299556e-05,
      "loss": 1.2183,
      "step": 818
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.5133749577719042,
      "learning_rate": 1.8781939370536376e-05,
      "loss": 1.2348,
      "step": 819
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.539934130640499,
      "learning_rate": 1.877845468267807e-05,
      "loss": 1.2229,
      "step": 820
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.948964396854176,
      "learning_rate": 1.877496534157178e-05,
      "loss": 0.6438,
      "step": 821
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.4397552762678796,
      "learning_rate": 1.8771471349067125e-05,
      "loss": 1.2319,
      "step": 822
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9371606223160784,
      "learning_rate": 1.8767972707016186e-05,
      "loss": 0.6614,
      "step": 823
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.712734296578037,
      "learning_rate": 1.8764469417273507e-05,
      "loss": 0.6516,
      "step": 824
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8544222911901314,
      "learning_rate": 1.8760961481696103e-05,
      "loss": 0.6559,
      "step": 825
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0452848297751673,
      "learning_rate": 1.8757448902143444e-05,
      "loss": 0.6422,
      "step": 826
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6763539079870435,
      "learning_rate": 1.875393168047746e-05,
      "loss": 0.6247,
      "step": 827
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6342785530277855,
      "learning_rate": 1.875040981856255e-05,
      "loss": 0.5553,
      "step": 828
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.3155751650209866,
      "learning_rate": 1.874688331826557e-05,
      "loss": 1.2224,
      "step": 829
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.588102242160416,
      "learning_rate": 1.874335218145583e-05,
      "loss": 1.1319,
      "step": 830
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9152098170501894,
      "learning_rate": 1.87398164100051e-05,
      "loss": 0.6447,
      "step": 831
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.062037625777986,
      "learning_rate": 1.873627600578761e-05,
      "loss": 0.67,
      "step": 832
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3576564219679836,
      "learning_rate": 1.8732730970680042e-05,
      "loss": 0.6294,
      "step": 833
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8618537639656665,
      "learning_rate": 1.8729181306561534e-05,
      "loss": 0.6843,
      "step": 834
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6661120919851624,
      "learning_rate": 1.872562701531368e-05,
      "loss": 0.6292,
      "step": 835
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8664972284272898,
      "learning_rate": 1.872206809882052e-05,
      "loss": 0.5928,
      "step": 836
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8917321138281376,
      "learning_rate": 1.8718504558968553e-05,
      "loss": 0.5808,
      "step": 837
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.693154583441528,
      "learning_rate": 1.8714936397646727e-05,
      "loss": 0.5498,
      "step": 838
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.8655732948797135,
      "learning_rate": 1.8711363616746438e-05,
      "loss": 1.2431,
      "step": 839
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.488561736709535,
      "learning_rate": 1.8707786218161527e-05,
      "loss": 1.1545,
      "step": 840
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0809440500703107,
      "learning_rate": 1.870420420378829e-05,
      "loss": 0.6264,
      "step": 841
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0227289177135157,
      "learning_rate": 1.870061757552547e-05,
      "loss": 0.6127,
      "step": 842
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8772610568157746,
      "learning_rate": 1.869702633527425e-05,
      "loss": 0.6259,
      "step": 843
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6595341301855062,
      "learning_rate": 1.8693430484938256e-05,
      "loss": 0.4999,
      "step": 844
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9689921970040263,
      "learning_rate": 1.8689830026423566e-05,
      "loss": 0.6056,
      "step": 845
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.3413391408769124,
      "learning_rate": 1.8686224961638704e-05,
      "loss": 1.16,
      "step": 846
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9528919823294377,
      "learning_rate": 1.8682615292494613e-05,
      "loss": 0.6785,
      "step": 847
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7893852315442829,
      "learning_rate": 1.86790010209047e-05,
      "loss": 0.5705,
      "step": 848
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8348344700375754,
      "learning_rate": 1.8675382148784805e-05,
      "loss": 0.6451,
      "step": 849
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.959652620953991,
      "learning_rate": 1.8671758678053206e-05,
      "loss": 0.613,
      "step": 850
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7359590865053591,
      "learning_rate": 1.8668130610630613e-05,
      "loss": 0.5918,
      "step": 851
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.962331084818572,
      "learning_rate": 1.8664497948440178e-05,
      "loss": 1.1459,
      "step": 852
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9984126172346692,
      "learning_rate": 1.866086069340749e-05,
      "loss": 0.5805,
      "step": 853
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.078596216783928,
      "learning_rate": 1.865721884746057e-05,
      "loss": 0.6077,
      "step": 854
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1232980421331598,
      "learning_rate": 1.8653572412529874e-05,
      "loss": 0.6271,
      "step": 855
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9201254098530522,
      "learning_rate": 1.8649921390548288e-05,
      "loss": 0.6145,
      "step": 856
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9548979421631796,
      "learning_rate": 1.8646265783451133e-05,
      "loss": 0.6468,
      "step": 857
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.432207164223108,
      "learning_rate": 1.8642605593176157e-05,
      "loss": 1.1995,
      "step": 858
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.883672377390732,
      "learning_rate": 1.8638940821663536e-05,
      "loss": 0.5842,
      "step": 859
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8052110361982083,
      "learning_rate": 1.8635271470855883e-05,
      "loss": 0.5634,
      "step": 860
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9415365947692709,
      "learning_rate": 1.8631597542698232e-05,
      "loss": 0.6712,
      "step": 861
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9336029591329165,
      "learning_rate": 1.8627919039138042e-05,
      "loss": 0.7137,
      "step": 862
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8723817199556325,
      "learning_rate": 1.8624235962125195e-05,
      "loss": 0.6096,
      "step": 863
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.85559639513595,
      "learning_rate": 1.862054831361201e-05,
      "loss": 0.7393,
      "step": 864
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3673496141313533,
      "learning_rate": 1.861685609555322e-05,
      "loss": 1.2012,
      "step": 865
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.605963020121471,
      "learning_rate": 1.861315930990597e-05,
      "loss": 1.2234,
      "step": 866
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4755360672639788,
      "learning_rate": 1.8609457958629854e-05,
      "loss": 0.5801,
      "step": 867
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6876056308149432,
      "learning_rate": 1.860575204368686e-05,
      "loss": 0.6472,
      "step": 868
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7105851083422212,
      "learning_rate": 1.86020415670414e-05,
      "loss": 0.6412,
      "step": 869
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0644479762938075,
      "learning_rate": 1.859832653066032e-05,
      "loss": 0.6078,
      "step": 870
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.062088303638194,
      "learning_rate": 1.8594606936512864e-05,
      "loss": 1.1863,
      "step": 871
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7897902462173474,
      "learning_rate": 1.8590882786570698e-05,
      "loss": 0.6504,
      "step": 872
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8133899120026444,
      "learning_rate": 1.858715408280791e-05,
      "loss": 0.6599,
      "step": 873
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7472452642686465,
      "learning_rate": 1.8583420827200996e-05,
      "loss": 0.6038,
      "step": 874
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8818857938617504,
      "learning_rate": 1.857968302172886e-05,
      "loss": 0.5538,
      "step": 875
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6977167781579512,
      "learning_rate": 1.857594066837283e-05,
      "loss": 0.6246,
      "step": 876
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.8334533575200926,
      "learning_rate": 1.8572193769116632e-05,
      "loss": 1.1977,
      "step": 877
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.765430885681226,
      "learning_rate": 1.8568442325946408e-05,
      "loss": 0.5368,
      "step": 878
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7179521058385572,
      "learning_rate": 1.8564686340850707e-05,
      "loss": 0.6326,
      "step": 879
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.407822859151782,
      "learning_rate": 1.8560925815820493e-05,
      "loss": 1.1549,
      "step": 880
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7009966438859583,
      "learning_rate": 1.855716075284912e-05,
      "loss": 0.6306,
      "step": 881
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7824319870957261,
      "learning_rate": 1.8553391153932364e-05,
      "loss": 0.6727,
      "step": 882
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.873600063002331,
      "learning_rate": 1.8549617021068402e-05,
      "loss": 1.1482,
      "step": 883
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9118333896051039,
      "learning_rate": 1.8545838356257804e-05,
      "loss": 0.6628,
      "step": 884
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.912600166769499,
      "learning_rate": 1.8542055161503552e-05,
      "loss": 0.6605,
      "step": 885
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7872173436933703,
      "learning_rate": 1.8538267438811027e-05,
      "loss": 0.6562,
      "step": 886
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.725376628734759,
      "learning_rate": 1.853447519018801e-05,
      "loss": 0.6141,
      "step": 887
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7380110667667692,
      "learning_rate": 1.8530678417644677e-05,
      "loss": 0.6138,
      "step": 888
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8177051774829325,
      "learning_rate": 1.8526877123193615e-05,
      "loss": 0.6366,
      "step": 889
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.1966936058210713,
      "learning_rate": 1.852307130884979e-05,
      "loss": 1.1792,
      "step": 890
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.1726570589128515,
      "learning_rate": 1.851926097663058e-05,
      "loss": 1.1447,
      "step": 891
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3450433705072005,
      "learning_rate": 1.8515446128555744e-05,
      "loss": 1.1562,
      "step": 892
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7286474028791272,
      "learning_rate": 1.8511626766647442e-05,
      "loss": 0.6264,
      "step": 893
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2469073700000255,
      "learning_rate": 1.8507802892930232e-05,
      "loss": 0.6019,
      "step": 894
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.900827365330872,
      "learning_rate": 1.8503974509431056e-05,
      "loss": 0.6234,
      "step": 895
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8174905023993568,
      "learning_rate": 1.8500141618179242e-05,
      "loss": 0.6502,
      "step": 896
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8280000100023852,
      "learning_rate": 1.849630422120652e-05,
      "loss": 0.6094,
      "step": 897
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.02087915749528,
      "learning_rate": 1.8492462320546997e-05,
      "loss": 1.1407,
      "step": 898
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8428291084344104,
      "learning_rate": 1.8488615918237175e-05,
      "loss": 0.5181,
      "step": 899
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6834668740741208,
      "learning_rate": 1.848476501631594e-05,
      "loss": 0.5728,
      "step": 900
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.450559611325547,
      "learning_rate": 1.848090961682456e-05,
      "loss": 1.1791,
      "step": 901
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7417247469769372,
      "learning_rate": 1.8477049721806695e-05,
      "loss": 0.6152,
      "step": 902
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6957176849716458,
      "learning_rate": 1.847318533330838e-05,
      "loss": 0.5646,
      "step": 903
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8895364337006744,
      "learning_rate": 1.846931645337803e-05,
      "loss": 0.5998,
      "step": 904
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6489471061497922,
      "learning_rate": 1.846544308406645e-05,
      "loss": 0.572,
      "step": 905
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.957381266014668,
      "learning_rate": 1.846156522742682e-05,
      "loss": 1.1352,
      "step": 906
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.928554291688688,
      "learning_rate": 1.8457682885514696e-05,
      "loss": 1.239,
      "step": 907
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.714100101460991,
      "learning_rate": 1.8453796060388016e-05,
      "loss": 0.6213,
      "step": 908
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.6534973445050007,
      "learning_rate": 1.8449904754107092e-05,
      "loss": 1.2654,
      "step": 909
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8747755671009438,
      "learning_rate": 1.8446008968734615e-05,
      "loss": 0.6404,
      "step": 910
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.668522076700751,
      "learning_rate": 1.8442108706335648e-05,
      "loss": 1.1577,
      "step": 911
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5320908609345,
      "learning_rate": 1.8438203968977618e-05,
      "loss": 1.1342,
      "step": 912
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7793825943469805,
      "learning_rate": 1.843429475873034e-05,
      "loss": 0.6587,
      "step": 913
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.4781113803037313,
      "learning_rate": 1.843038107766599e-05,
      "loss": 1.1717,
      "step": 914
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.183440098314604,
      "learning_rate": 1.8426462927859115e-05,
      "loss": 0.591,
      "step": 915
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8733491778407647,
      "learning_rate": 1.8422540311386633e-05,
      "loss": 0.6579,
      "step": 916
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8034431911751676,
      "learning_rate": 1.841861323032783e-05,
      "loss": 0.6275,
      "step": 917
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7508954901184286,
      "learning_rate": 1.841468168676436e-05,
      "loss": 0.5777,
      "step": 918
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1598222436167833,
      "learning_rate": 1.8410745682780234e-05,
      "loss": 0.6655,
      "step": 919
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.4387972387961026,
      "learning_rate": 1.8406805220461834e-05,
      "loss": 1.0998,
      "step": 920
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8828084998982173,
      "learning_rate": 1.8402860301897908e-05,
      "loss": 0.618,
      "step": 921
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5966126325332661,
      "learning_rate": 1.839891092917956e-05,
      "loss": 0.5692,
      "step": 922
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7634684050632583,
      "learning_rate": 1.8394957104400258e-05,
      "loss": 0.6038,
      "step": 923
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.931001905996964,
      "learning_rate": 1.8390998829655834e-05,
      "loss": 0.7002,
      "step": 924
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.2726791293441098,
      "learning_rate": 1.8387036107044467e-05,
      "loss": 1.2023,
      "step": 925
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7672141692137906,
      "learning_rate": 1.8383068938666702e-05,
      "loss": 0.5875,
      "step": 926
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.783919638342859,
      "learning_rate": 1.837909732662545e-05,
      "loss": 0.6378,
      "step": 927
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.673597779927744,
      "learning_rate": 1.837512127302595e-05,
      "loss": 0.5496,
      "step": 928
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.7811954175744233,
      "learning_rate": 1.8371140779975826e-05,
      "loss": 1.1957,
      "step": 929
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8856493963336478,
      "learning_rate": 1.836715584958504e-05,
      "loss": 0.6472,
      "step": 930
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.8748118915940215,
      "learning_rate": 1.83631664839659e-05,
      "loss": 1.2306,
      "step": 931
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.89176935041631,
      "learning_rate": 1.8359172685233078e-05,
      "loss": 1.2071,
      "step": 932
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5135783572909798,
      "learning_rate": 1.8355174455503598e-05,
      "loss": 0.6149,
      "step": 933
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6675913332957883,
      "learning_rate": 1.8351171796896815e-05,
      "loss": 0.5629,
      "step": 934
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.025784487463427,
      "learning_rate": 1.8347164711534443e-05,
      "loss": 0.6061,
      "step": 935
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.7691864884851958,
      "learning_rate": 1.8343153201540545e-05,
      "loss": 1.164,
      "step": 936
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6724648371312394,
      "learning_rate": 1.833913726904153e-05,
      "loss": 0.5839,
      "step": 937
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6612541015888396,
      "learning_rate": 1.833511691616614e-05,
      "loss": 0.5979,
      "step": 938
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8457593383207367,
      "learning_rate": 1.833109214504547e-05,
      "loss": 0.6121,
      "step": 939
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9515891321403604,
      "learning_rate": 1.8327062957812956e-05,
      "loss": 0.6744,
      "step": 940
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7127550286295694,
      "learning_rate": 1.832302935660437e-05,
      "loss": 0.5584,
      "step": 941
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6485466823227113,
      "learning_rate": 1.831899134355783e-05,
      "loss": 0.6652,
      "step": 942
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.7765602245284065,
      "learning_rate": 1.8314948920813794e-05,
      "loss": 1.2153,
      "step": 943
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8984788554700547,
      "learning_rate": 1.8310902090515045e-05,
      "loss": 0.6524,
      "step": 944
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7084646230906868,
      "learning_rate": 1.830685085480671e-05,
      "loss": 0.669,
      "step": 945
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7723336665263685,
      "learning_rate": 1.8302795215836263e-05,
      "loss": 0.5794,
      "step": 946
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7996149800724812,
      "learning_rate": 1.829873517575349e-05,
      "loss": 0.6648,
      "step": 947
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6978220297471878,
      "learning_rate": 1.8294670736710522e-05,
      "loss": 0.6156,
      "step": 948
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.8485844981636,
      "learning_rate": 1.8290601900861823e-05,
      "loss": 1.2584,
      "step": 949
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8866023488696293,
      "learning_rate": 1.8286528670364187e-05,
      "loss": 0.6369,
      "step": 950
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.8633242511091934,
      "learning_rate": 1.8282451047376734e-05,
      "loss": 1.1798,
      "step": 951
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6885520809100354,
      "learning_rate": 1.8278369034060912e-05,
      "loss": 0.6235,
      "step": 952
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.4871735059362594,
      "learning_rate": 1.82742826325805e-05,
      "loss": 1.1292,
      "step": 953
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.672785797177815,
      "learning_rate": 1.8270191845101602e-05,
      "loss": 0.6114,
      "step": 954
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.764713647872729,
      "learning_rate": 1.8266096673792643e-05,
      "loss": 0.6133,
      "step": 955
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8644481133610684,
      "learning_rate": 1.826199712082438e-05,
      "loss": 0.6179,
      "step": 956
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8027950247779745,
      "learning_rate": 1.8257893188369883e-05,
      "loss": 0.6108,
      "step": 957
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7376197406676284,
      "learning_rate": 1.825378487860455e-05,
      "loss": 0.6571,
      "step": 958
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6026510516678032,
      "learning_rate": 1.8249672193706097e-05,
      "loss": 0.5685,
      "step": 959
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6789753175370827,
      "learning_rate": 1.8245555135854558e-05,
      "loss": 0.6531,
      "step": 960
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7784420423105196,
      "learning_rate": 1.824143370723229e-05,
      "loss": 0.623,
      "step": 961
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8066130186390759,
      "learning_rate": 1.8237307910023963e-05,
      "loss": 0.629,
      "step": 962
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5862388176750108,
      "learning_rate": 1.8233177746416557e-05,
      "loss": 0.5731,
      "step": 963
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5743925718349194,
      "learning_rate": 1.8229043218599382e-05,
      "loss": 0.5937,
      "step": 964
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.826351841555935,
      "learning_rate": 1.822490432876404e-05,
      "loss": 1.1909,
      "step": 965
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.383229071828364,
      "learning_rate": 1.822076107910447e-05,
      "loss": 1.1357,
      "step": 966
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9735125919523575,
      "learning_rate": 1.82166134718169e-05,
      "loss": 0.6095,
      "step": 967
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.949715263627105,
      "learning_rate": 1.8212461509099882e-05,
      "loss": 1.1606,
      "step": 968
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.883297775179817,
      "learning_rate": 1.8208305193154266e-05,
      "loss": 0.6493,
      "step": 969
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.764582146629523,
      "learning_rate": 1.8204144526183224e-05,
      "loss": 0.6318,
      "step": 970
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.8238419630658247,
      "learning_rate": 1.8199979510392216e-05,
      "loss": 1.0817,
      "step": 971
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.358178792404035,
      "learning_rate": 1.8195810147989026e-05,
      "loss": 1.1828,
      "step": 972
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.103013924467637,
      "learning_rate": 1.8191636441183727e-05,
      "loss": 0.6602,
      "step": 973
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.353478015891184,
      "learning_rate": 1.81874583921887e-05,
      "loss": 0.7213,
      "step": 974
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0146876620117373,
      "learning_rate": 1.818327600321863e-05,
      "loss": 0.6069,
      "step": 975
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8041232997989367,
      "learning_rate": 1.8179089276490504e-05,
      "loss": 0.6723,
      "step": 976
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5917460390333102,
      "learning_rate": 1.81748982142236e-05,
      "loss": 0.5835,
      "step": 977
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.6732277028824782,
      "learning_rate": 1.8170702818639502e-05,
      "loss": 1.1383,
      "step": 978
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7755978909435184,
      "learning_rate": 1.816650309196209e-05,
      "loss": 0.5895,
      "step": 979
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1779692777453037,
      "learning_rate": 1.816229903641754e-05,
      "loss": 0.6796,
      "step": 980
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.735893991830327,
      "learning_rate": 1.815809065423431e-05,
      "loss": 0.6421,
      "step": 981
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0108904622395274,
      "learning_rate": 1.8153877947643178e-05,
      "loss": 0.6248,
      "step": 982
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6062982454527193,
      "learning_rate": 1.8149660918877185e-05,
      "loss": 0.5911,
      "step": 983
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.684633315006672,
      "learning_rate": 1.8145439570171684e-05,
      "loss": 0.6383,
      "step": 984
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7277467467933034,
      "learning_rate": 1.814121390376431e-05,
      "loss": 0.6012,
      "step": 985
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.668834519913874,
      "learning_rate": 1.8136983921894982e-05,
      "loss": 1.172,
      "step": 986
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7167260923912844,
      "learning_rate": 1.813274962680592e-05,
      "loss": 0.6549,
      "step": 987
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6396470485657344,
      "learning_rate": 1.812851102074162e-05,
      "loss": 0.5804,
      "step": 988
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.678268973485464,
      "learning_rate": 1.8124268105948857e-05,
      "loss": 0.6331,
      "step": 989
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6815590393078754,
      "learning_rate": 1.8120020884676708e-05,
      "loss": 0.6262,
      "step": 990
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8145004664191295,
      "learning_rate": 1.8115769359176517e-05,
      "loss": 0.6888,
      "step": 991
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8871656471226044,
      "learning_rate": 1.811151353170192e-05,
      "loss": 0.6411,
      "step": 992
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.952883880491969,
      "learning_rate": 1.8107253404508827e-05,
      "loss": 0.5579,
      "step": 993
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.8806798183467617,
      "learning_rate": 1.8102988979855428e-05,
      "loss": 1.1858,
      "step": 994
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.793825555728122,
      "learning_rate": 1.8098720260002192e-05,
      "loss": 0.6166,
      "step": 995
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1938215376531263,
      "learning_rate": 1.8094447247211867e-05,
      "loss": 0.6886,
      "step": 996
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9830931978134538,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 0.6284,
      "step": 997
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9274562152723929,
      "learning_rate": 1.808588835188231e-05,
      "loss": 0.6738,
      "step": 998
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.463677539464579,
      "learning_rate": 1.808160247387994e-05,
      "loss": 1.1666,
      "step": 999
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3591298051191267,
      "learning_rate": 1.807731231201421e-05,
      "loss": 1.2434,
      "step": 1000
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7025524235064056,
      "learning_rate": 1.8073017868559223e-05,
      "loss": 0.5383,
      "step": 1001
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7396898514823291,
      "learning_rate": 1.8068719145791373e-05,
      "loss": 0.6294,
      "step": 1002
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8616431104476083,
      "learning_rate": 1.80644161459893e-05,
      "loss": 0.6049,
      "step": 1003
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7067248057141111,
      "learning_rate": 1.8060108871433922e-05,
      "loss": 0.6346,
      "step": 1004
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8779915025280611,
      "learning_rate": 1.8055797324408427e-05,
      "loss": 0.6127,
      "step": 1005
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.2155842680430187,
      "learning_rate": 1.8051481507198256e-05,
      "loss": 1.194,
      "step": 1006
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5890720158162817,
      "learning_rate": 1.8047161422091126e-05,
      "loss": 0.6084,
      "step": 1007
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7135853016834544,
      "learning_rate": 1.8042837071377004e-05,
      "loss": 0.6616,
      "step": 1008
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8188325964211935,
      "learning_rate": 1.803850845734813e-05,
      "loss": 0.6492,
      "step": 1009
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.834710994556691,
      "learning_rate": 1.8034175582299e-05,
      "loss": 0.6533,
      "step": 1010
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.722924672704436,
      "learning_rate": 1.8029838448526357e-05,
      "loss": 0.5733,
      "step": 1011
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8687206130051106,
      "learning_rate": 1.8025497058329224e-05,
      "loss": 0.6227,
      "step": 1012
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0459976972386635,
      "learning_rate": 1.8021151414008856e-05,
      "loss": 0.6468,
      "step": 1013
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.4077532130614676,
      "learning_rate": 1.8016801517868784e-05,
      "loss": 1.1749,
      "step": 1014
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8633937966483354,
      "learning_rate": 1.8012447372214777e-05,
      "loss": 0.6233,
      "step": 1015
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.6820234436693933,
      "learning_rate": 1.8008088979354864e-05,
      "loss": 1.1864,
      "step": 1016
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6698838083683931,
      "learning_rate": 1.800372634159933e-05,
      "loss": 0.4778,
      "step": 1017
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.4042588608900215,
      "learning_rate": 1.79993594612607e-05,
      "loss": 1.1568,
      "step": 1018
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.579870679278826,
      "learning_rate": 1.799498834065375e-05,
      "loss": 1.1699,
      "step": 1019
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.184816935061694,
      "learning_rate": 1.799061298209551e-05,
      "loss": 0.6507,
      "step": 1020
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7170983270407747,
      "learning_rate": 1.7986233387905255e-05,
      "loss": 0.5687,
      "step": 1021
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9955886460100283,
      "learning_rate": 1.7981849560404498e-05,
      "loss": 0.594,
      "step": 1022
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5342188780163613,
      "learning_rate": 1.7977461501916998e-05,
      "loss": 0.6501,
      "step": 1023
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.864398268288002,
      "learning_rate": 1.7973069214768766e-05,
      "loss": 0.5982,
      "step": 1024
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7903274964797984,
      "learning_rate": 1.796867270128804e-05,
      "loss": 0.6226,
      "step": 1025
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7497210365142182,
      "learning_rate": 1.7964271963805317e-05,
      "loss": 0.5512,
      "step": 1026
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6909037954015653,
      "learning_rate": 1.7959867004653315e-05,
      "loss": 0.6055,
      "step": 1027
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.0361412526533926,
      "learning_rate": 1.7955457826167e-05,
      "loss": 1.1689,
      "step": 1028
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.758981901920231,
      "learning_rate": 1.7951044430683567e-05,
      "loss": 0.6624,
      "step": 1029
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9626791911459593,
      "learning_rate": 1.7946626820542456e-05,
      "loss": 0.5994,
      "step": 1030
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6932771894105372,
      "learning_rate": 1.7942204998085336e-05,
      "loss": 0.5489,
      "step": 1031
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8417170929491016,
      "learning_rate": 1.7937778965656103e-05,
      "loss": 0.6379,
      "step": 1032
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8693685663378943,
      "learning_rate": 1.7933348725600896e-05,
      "loss": 0.6228,
      "step": 1033
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9565869199944088,
      "learning_rate": 1.792891428026808e-05,
      "loss": 0.6542,
      "step": 1034
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.642213086056391,
      "learning_rate": 1.792447563200824e-05,
      "loss": 1.1717,
      "step": 1035
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8775836862426645,
      "learning_rate": 1.7920032783174207e-05,
      "loss": 0.6514,
      "step": 1036
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8027562653254454,
      "learning_rate": 1.791558573612102e-05,
      "loss": 0.6323,
      "step": 1037
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.8119916478358875,
      "learning_rate": 1.791113449320596e-05,
      "loss": 1.2164,
      "step": 1038
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.604355204018539,
      "learning_rate": 1.7906679056788516e-05,
      "loss": 0.5365,
      "step": 1039
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7555664053153675,
      "learning_rate": 1.790221942923041e-05,
      "loss": 0.5798,
      "step": 1040
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9092956134953492,
      "learning_rate": 1.7897755612895585e-05,
      "loss": 0.6705,
      "step": 1041
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.525397466395638,
      "learning_rate": 1.7893287610150207e-05,
      "loss": 1.1923,
      "step": 1042
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8571233884463898,
      "learning_rate": 1.7888815423362647e-05,
      "loss": 0.6784,
      "step": 1043
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.72519956530939,
      "learning_rate": 1.788433905490351e-05,
      "loss": 0.6745,
      "step": 1044
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8437387384361712,
      "learning_rate": 1.7879858507145615e-05,
      "loss": 0.6729,
      "step": 1045
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7318560984229023,
      "learning_rate": 1.7875373782463986e-05,
      "loss": 0.6707,
      "step": 1046
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.909884048946607,
      "learning_rate": 1.7870884883235872e-05,
      "loss": 1.1894,
      "step": 1047
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6339825247842483,
      "learning_rate": 1.7866391811840725e-05,
      "loss": 0.5876,
      "step": 1048
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7518226171333897,
      "learning_rate": 1.7861894570660218e-05,
      "loss": 0.5949,
      "step": 1049
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4890187494100435,
      "learning_rate": 1.7857393162078234e-05,
      "loss": 0.5444,
      "step": 1050
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8191163790269749,
      "learning_rate": 1.7852887588480857e-05,
      "loss": 0.7303,
      "step": 1051
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.468552444006581,
      "learning_rate": 1.7848377852256383e-05,
      "loss": 1.1003,
      "step": 1052
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.556918923834809,
      "learning_rate": 1.784386395579532e-05,
      "loss": 0.6263,
      "step": 1053
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.692537502528194,
      "learning_rate": 1.783934590149037e-05,
      "loss": 1.177,
      "step": 1054
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.745830189999208,
      "learning_rate": 1.7834823691736446e-05,
      "loss": 0.6008,
      "step": 1055
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.4862660865201875,
      "learning_rate": 1.783029732893067e-05,
      "loss": 1.2086,
      "step": 1056
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5517838129252783,
      "learning_rate": 1.7825766815472345e-05,
      "loss": 0.5862,
      "step": 1057
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.665849087809521,
      "learning_rate": 1.7821232153763e-05,
      "loss": 0.6184,
      "step": 1058
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6282692758504462,
      "learning_rate": 1.7816693346206343e-05,
      "loss": 0.5855,
      "step": 1059
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.0601029072892443,
      "learning_rate": 1.781215039520829e-05,
      "loss": 1.2508,
      "step": 1060
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6673892142281903,
      "learning_rate": 1.780760330317695e-05,
      "loss": 0.6153,
      "step": 1061
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.904540680513943,
      "learning_rate": 1.7803052072522627e-05,
      "loss": 0.6406,
      "step": 1062
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7306743085956953,
      "learning_rate": 1.7798496705657817e-05,
      "loss": 0.6826,
      "step": 1063
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.3523706691678017,
      "learning_rate": 1.779393720499721e-05,
      "loss": 1.1825,
      "step": 1064
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7887078661622116,
      "learning_rate": 1.7789373572957696e-05,
      "loss": 0.6209,
      "step": 1065
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.593450177501334,
      "learning_rate": 1.7784805811958336e-05,
      "loss": 0.565,
      "step": 1066
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.661545504975921,
      "learning_rate": 1.7780233924420395e-05,
      "loss": 0.5747,
      "step": 1067
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6836255269672282,
      "learning_rate": 1.7775657912767324e-05,
      "loss": 0.5267,
      "step": 1068
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2563125242495032,
      "learning_rate": 1.777107777942475e-05,
      "loss": 1.1454,
      "step": 1069
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6128566761611765,
      "learning_rate": 1.7766493526820498e-05,
      "loss": 0.6616,
      "step": 1070
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.730219497591321,
      "learning_rate": 1.7761905157384567e-05,
      "loss": 0.6355,
      "step": 1071
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.79849754033908,
      "learning_rate": 1.7757312673549144e-05,
      "loss": 0.5938,
      "step": 1072
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8411650429607402,
      "learning_rate": 1.775271607774859e-05,
      "loss": 0.5561,
      "step": 1073
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.4075795760755856,
      "learning_rate": 1.7748115372419453e-05,
      "loss": 1.1484,
      "step": 1074
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7631888145510417,
      "learning_rate": 1.7743510560000456e-05,
      "loss": 0.6857,
      "step": 1075
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6758582373355448,
      "learning_rate": 1.7738901642932495e-05,
      "loss": 0.6049,
      "step": 1076
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6568216998279908,
      "learning_rate": 1.7734288623658654e-05,
      "loss": 0.6125,
      "step": 1077
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.3882731394108063,
      "learning_rate": 1.7729671504624183e-05,
      "loss": 1.178,
      "step": 1078
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.3528499394370437,
      "learning_rate": 1.77250502882765e-05,
      "loss": 1.1423,
      "step": 1079
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.714601568573906,
      "learning_rate": 1.7720424977065206e-05,
      "loss": 0.6315,
      "step": 1080
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5759700061484214,
      "learning_rate": 1.771579557344206e-05,
      "loss": 0.5237,
      "step": 1081
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.682741461043965,
      "learning_rate": 1.771116207986101e-05,
      "loss": 0.666,
      "step": 1082
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.743723955511679,
      "learning_rate": 1.770652449877815e-05,
      "loss": 0.5708,
      "step": 1083
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8924688842394661,
      "learning_rate": 1.7701882832651752e-05,
      "loss": 0.6479,
      "step": 1084
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6916265103571362,
      "learning_rate": 1.7697237083942254e-05,
      "loss": 0.6156,
      "step": 1085
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6611635707427457,
      "learning_rate": 1.769258725511225e-05,
      "loss": 0.6062,
      "step": 1086
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8818345999388877,
      "learning_rate": 1.7687933348626517e-05,
      "loss": 0.701,
      "step": 1087
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.658342192186451,
      "learning_rate": 1.768327536695196e-05,
      "loss": 1.1242,
      "step": 1088
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7304120643517682,
      "learning_rate": 1.7678613312557674e-05,
      "loss": 0.5653,
      "step": 1089
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.803860524557296,
      "learning_rate": 1.7673947187914905e-05,
      "loss": 0.6152,
      "step": 1090
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8409364820696166,
      "learning_rate": 1.7669276995497045e-05,
      "loss": 0.6422,
      "step": 1091
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9014512021629058,
      "learning_rate": 1.7664602737779655e-05,
      "loss": 0.683,
      "step": 1092
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7002251291637045,
      "learning_rate": 1.765992441724045e-05,
      "loss": 0.6779,
      "step": 1093
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5991992967670463,
      "learning_rate": 1.7655242036359287e-05,
      "loss": 0.6004,
      "step": 1094
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6948030204748317,
      "learning_rate": 1.765055559761819e-05,
      "loss": 0.6711,
      "step": 1095
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6201662001201964,
      "learning_rate": 1.764586510350133e-05,
      "loss": 0.6024,
      "step": 1096
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6085431513050976,
      "learning_rate": 1.7641170556495016e-05,
      "loss": 0.5975,
      "step": 1097
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5439666978176096,
      "learning_rate": 1.7636471959087722e-05,
      "loss": 0.6082,
      "step": 1098
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.4837333507260144,
      "learning_rate": 1.763176931377006e-05,
      "loss": 1.1544,
      "step": 1099
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7995089774205122,
      "learning_rate": 1.762706262303479e-05,
      "loss": 0.6962,
      "step": 1100
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.726981629524958,
      "learning_rate": 1.7622351889376817e-05,
      "loss": 0.6342,
      "step": 1101
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2813437390385625,
      "learning_rate": 1.761763711529318e-05,
      "loss": 1.1916,
      "step": 1102
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.459021570490057,
      "learning_rate": 1.7612918303283075e-05,
      "loss": 1.1333,
      "step": 1103
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.331871309693492,
      "learning_rate": 1.760819545584783e-05,
      "loss": 1.1018,
      "step": 1104
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7813964270471123,
      "learning_rate": 1.7603468575490912e-05,
      "loss": 0.6222,
      "step": 1105
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.37954336573614,
      "learning_rate": 1.7598737664717927e-05,
      "loss": 1.1874,
      "step": 1106
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6599299573043835,
      "learning_rate": 1.7594002726036615e-05,
      "loss": 0.5712,
      "step": 1107
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.667951854440397,
      "learning_rate": 1.7589263761956856e-05,
      "loss": 0.6144,
      "step": 1108
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8510930162428314,
      "learning_rate": 1.758452077499066e-05,
      "loss": 0.6354,
      "step": 1109
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.9821295223903936,
      "learning_rate": 1.7579773767652174e-05,
      "loss": 1.1867,
      "step": 1110
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5861150137641404,
      "learning_rate": 1.7575022742457664e-05,
      "loss": 0.5966,
      "step": 1111
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9177521772660246,
      "learning_rate": 1.757026770192554e-05,
      "loss": 0.6039,
      "step": 1112
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8273813348419066,
      "learning_rate": 1.7565508648576338e-05,
      "loss": 0.6424,
      "step": 1113
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.728502454791725,
      "learning_rate": 1.7560745584932712e-05,
      "loss": 0.6085,
      "step": 1114
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8815961806961323,
      "learning_rate": 1.755597851351945e-05,
      "loss": 0.6249,
      "step": 1115
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6400762563792899,
      "learning_rate": 1.755120743686346e-05,
      "loss": 0.6117,
      "step": 1116
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6576094198988052,
      "learning_rate": 1.7546432357493782e-05,
      "loss": 0.6454,
      "step": 1117
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6844269450087588,
      "learning_rate": 1.754165327794156e-05,
      "loss": 0.6453,
      "step": 1118
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8282884794905427,
      "learning_rate": 1.7536870200740076e-05,
      "loss": 0.6648,
      "step": 1119
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8016320943266986,
      "learning_rate": 1.7532083128424728e-05,
      "loss": 0.5898,
      "step": 1120
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6303949813547367,
      "learning_rate": 1.752729206353302e-05,
      "loss": 0.5974,
      "step": 1121
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6715784038740933,
      "learning_rate": 1.7522497008604586e-05,
      "loss": 0.6192,
      "step": 1122
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6791100419264986,
      "learning_rate": 1.7517697966181167e-05,
      "loss": 0.5807,
      "step": 1123
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.4475949065707936,
      "learning_rate": 1.751289493880662e-05,
      "loss": 1.1907,
      "step": 1124
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6436905766942516,
      "learning_rate": 1.7508087929026913e-05,
      "loss": 0.5912,
      "step": 1125
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7618693783942623,
      "learning_rate": 1.7503276939390133e-05,
      "loss": 0.6678,
      "step": 1126
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6546379101685789,
      "learning_rate": 1.7498461972446463e-05,
      "loss": 0.6154,
      "step": 1127
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.3836239626021363,
      "learning_rate": 1.7493643030748203e-05,
      "loss": 1.1639,
      "step": 1128
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8621945564174298,
      "learning_rate": 1.748882011684976e-05,
      "loss": 0.6339,
      "step": 1129
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7627924618945812,
      "learning_rate": 1.7483993233307644e-05,
      "loss": 0.5921,
      "step": 1130
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6751497837410223,
      "learning_rate": 1.747916238268047e-05,
      "loss": 0.6341,
      "step": 1131
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6086849284040812,
      "learning_rate": 1.7474327567528955e-05,
      "loss": 0.6384,
      "step": 1132
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.654740578631574,
      "learning_rate": 1.746948879041592e-05,
      "loss": 1.1653,
      "step": 1133
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.6213434118470853,
      "learning_rate": 1.7464646053906283e-05,
      "loss": 1.1208,
      "step": 1134
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7909274755982116,
      "learning_rate": 1.745979936056706e-05,
      "loss": 0.641,
      "step": 1135
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6665605150352847,
      "learning_rate": 1.7454948712967377e-05,
      "loss": 0.6384,
      "step": 1136
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8600733702153325,
      "learning_rate": 1.7450094113678436e-05,
      "loss": 0.6668,
      "step": 1137
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.846725166816054,
      "learning_rate": 1.7445235565273542e-05,
      "loss": 0.652,
      "step": 1138
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8483662831480385,
      "learning_rate": 1.74403730703281e-05,
      "loss": 0.5574,
      "step": 1139
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.500617210917236,
      "learning_rate": 1.7435506631419605e-05,
      "loss": 1.1322,
      "step": 1140
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6123669419695679,
      "learning_rate": 1.7430636251127634e-05,
      "loss": 0.5746,
      "step": 1141
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.41452458379561,
      "learning_rate": 1.7425761932033857e-05,
      "loss": 1.1816,
      "step": 1142
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9445234516241312,
      "learning_rate": 1.7420883676722036e-05,
      "loss": 0.6581,
      "step": 1143
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.4964057902234926,
      "learning_rate": 1.7416001487778017e-05,
      "loss": 1.2045,
      "step": 1144
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8150072682625908,
      "learning_rate": 1.7411115367789733e-05,
      "loss": 0.6023,
      "step": 1145
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.4935342984744446,
      "learning_rate": 1.7406225319347197e-05,
      "loss": 1.1982,
      "step": 1146
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5994893597902466,
      "learning_rate": 1.7401331345042504e-05,
      "loss": 0.5902,
      "step": 1147
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.8913793268067653,
      "learning_rate": 1.7396433447469836e-05,
      "loss": 1.1907,
      "step": 1148
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6937006169416398,
      "learning_rate": 1.7391531629225448e-05,
      "loss": 0.633,
      "step": 1149
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6796545466522361,
      "learning_rate": 1.738662589290768e-05,
      "loss": 0.5294,
      "step": 1150
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6618649601058555,
      "learning_rate": 1.7381716241116945e-05,
      "loss": 0.5987,
      "step": 1151
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7226767987214822,
      "learning_rate": 1.7376802676455723e-05,
      "loss": 0.6207,
      "step": 1152
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5978304057869164,
      "learning_rate": 1.7371885201528584e-05,
      "loss": 0.5987,
      "step": 1153
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6353445883404873,
      "learning_rate": 1.736696381894216e-05,
      "loss": 0.5908,
      "step": 1154
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.326679232273842,
      "learning_rate": 1.736203853130516e-05,
      "loss": 1.1546,
      "step": 1155
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.642648887803711,
      "learning_rate": 1.7357109341228366e-05,
      "loss": 0.5774,
      "step": 1156
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.5640276219181684,
      "learning_rate": 1.7352176251324614e-05,
      "loss": 1.1506,
      "step": 1157
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7655602298204705,
      "learning_rate": 1.7347239264208823e-05,
      "loss": 0.5998,
      "step": 1158
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.429239699822731,
      "learning_rate": 1.7342298382497964e-05,
      "loss": 1.1403,
      "step": 1159
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.5704750988166936,
      "learning_rate": 1.7337353608811086e-05,
      "loss": 1.1661,
      "step": 1160
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.748367794829484,
      "learning_rate": 1.7332404945769292e-05,
      "loss": 0.6456,
      "step": 1161
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7138348937905592,
      "learning_rate": 1.7327452395995748e-05,
      "loss": 0.6453,
      "step": 1162
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6395194657030927,
      "learning_rate": 1.7322495962115687e-05,
      "loss": 0.6496,
      "step": 1163
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7098657217266555,
      "learning_rate": 1.7317535646756395e-05,
      "loss": 0.6755,
      "step": 1164
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8836747888395204,
      "learning_rate": 1.7312571452547208e-05,
      "loss": 0.5844,
      "step": 1165
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7182822375320081,
      "learning_rate": 1.7307603382119532e-05,
      "loss": 0.5646,
      "step": 1166
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8485328922508069,
      "learning_rate": 1.7302631438106824e-05,
      "loss": 0.608,
      "step": 1167
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.7842793204772653,
      "learning_rate": 1.729765562314459e-05,
      "loss": 1.1281,
      "step": 1168
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5972955944869682,
      "learning_rate": 1.729267593987039e-05,
      "loss": 0.6121,
      "step": 1169
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4806208425833645,
      "learning_rate": 1.728769239092383e-05,
      "loss": 0.5269,
      "step": 1170
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7716399643649445,
      "learning_rate": 1.728270497894658e-05,
      "loss": 0.5925,
      "step": 1171
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6440198017866678,
      "learning_rate": 1.7277713706582333e-05,
      "loss": 0.5868,
      "step": 1172
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8713315046718613,
      "learning_rate": 1.7272718576476855e-05,
      "loss": 0.5449,
      "step": 1173
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7360470966930692,
      "learning_rate": 1.726771959127794e-05,
      "loss": 0.6552,
      "step": 1174
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7312237813546296,
      "learning_rate": 1.726271675363543e-05,
      "loss": 0.5831,
      "step": 1175
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.753918495602997,
      "learning_rate": 1.7257710066201205e-05,
      "loss": 1.1771,
      "step": 1176
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5824798828123259,
      "learning_rate": 1.7252699531629197e-05,
      "loss": 0.5587,
      "step": 1177
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3516552505511106,
      "learning_rate": 1.7247685152575367e-05,
      "loss": 1.0983,
      "step": 1178
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5785638137962545,
      "learning_rate": 1.724266693169772e-05,
      "loss": 0.6421,
      "step": 1179
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.701127771179485,
      "learning_rate": 1.7237644871656286e-05,
      "loss": 1.1618,
      "step": 1180
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.619384821995541,
      "learning_rate": 1.7232618975113146e-05,
      "loss": 1.141,
      "step": 1181
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5844015156126925,
      "learning_rate": 1.7227589244732407e-05,
      "loss": 0.586,
      "step": 1182
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.583635549459223,
      "learning_rate": 1.72225556831802e-05,
      "loss": 0.6083,
      "step": 1183
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.692782307493963,
      "learning_rate": 1.7217518293124705e-05,
      "loss": 0.6497,
      "step": 1184
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.7213838513460944,
      "learning_rate": 1.7212477077236115e-05,
      "loss": 1.2115,
      "step": 1185
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.70211924562607,
      "learning_rate": 1.720743203818666e-05,
      "loss": 1.1344,
      "step": 1186
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6122593308749684,
      "learning_rate": 1.7202383178650595e-05,
      "loss": 0.5853,
      "step": 1187
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5809256783242744,
      "learning_rate": 1.7197330501304196e-05,
      "loss": 0.6627,
      "step": 1188
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.984236030323815,
      "learning_rate": 1.7192274008825772e-05,
      "loss": 0.6732,
      "step": 1189
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.0840254352669,
      "learning_rate": 1.718721370389564e-05,
      "loss": 1.1392,
      "step": 1190
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.629724740156973,
      "learning_rate": 1.7182149589196152e-05,
      "loss": 0.61,
      "step": 1191
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.8658524485792802,
      "learning_rate": 1.7177081667411672e-05,
      "loss": 1.132,
      "step": 1192
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.696987121164853,
      "learning_rate": 1.7172009941228583e-05,
      "loss": 0.5815,
      "step": 1193
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8802088704831215,
      "learning_rate": 1.7166934413335286e-05,
      "loss": 0.626,
      "step": 1194
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.6113400430223983,
      "learning_rate": 1.7161855086422195e-05,
      "loss": 1.1987,
      "step": 1195
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.821617375141145,
      "learning_rate": 1.7156771963181743e-05,
      "loss": 1.1598,
      "step": 1196
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.773253858135045,
      "learning_rate": 1.7151685046308365e-05,
      "loss": 0.5935,
      "step": 1197
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9911328728842737,
      "learning_rate": 1.7146594338498523e-05,
      "loss": 0.6456,
      "step": 1198
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7392546939256859,
      "learning_rate": 1.7141499842450667e-05,
      "loss": 0.5632,
      "step": 1199
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6756891815899673,
      "learning_rate": 1.7136401560865275e-05,
      "loss": 0.5446,
      "step": 1200
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7417017732678968,
      "learning_rate": 1.7131299496444825e-05,
      "loss": 0.6735,
      "step": 1201
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5804270204368596,
      "learning_rate": 1.7126193651893795e-05,
      "loss": 0.6689,
      "step": 1202
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7602990275096704,
      "learning_rate": 1.7121084029918668e-05,
      "loss": 0.6387,
      "step": 1203
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.527633788975348,
      "learning_rate": 1.7115970633227935e-05,
      "loss": 0.5669,
      "step": 1204
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.065220340983929,
      "learning_rate": 1.7110853464532086e-05,
      "loss": 1.149,
      "step": 1205
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.497288286234867,
      "learning_rate": 1.710573252654361e-05,
      "loss": 1.2193,
      "step": 1206
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.780304387723364,
      "learning_rate": 1.710060782197699e-05,
      "loss": 0.664,
      "step": 1207
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.88675275864799,
      "learning_rate": 1.709547935354871e-05,
      "loss": 0.6171,
      "step": 1208
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.847954631639945,
      "learning_rate": 1.7090347123977248e-05,
      "loss": 0.7226,
      "step": 1209
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6555978476858695,
      "learning_rate": 1.7085211135983076e-05,
      "loss": 0.6559,
      "step": 1210
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.712064315279315,
      "learning_rate": 1.7080071392288656e-05,
      "loss": 0.5044,
      "step": 1211
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7483356260039788,
      "learning_rate": 1.7074927895618447e-05,
      "loss": 0.5734,
      "step": 1212
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.691715376433066,
      "learning_rate": 1.7069780648698887e-05,
      "loss": 0.5948,
      "step": 1213
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7983914368333171,
      "learning_rate": 1.7064629654258415e-05,
      "loss": 0.5922,
      "step": 1214
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.677351621446738,
      "learning_rate": 1.7059474915027443e-05,
      "loss": 0.6286,
      "step": 1215
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.731539633384732,
      "learning_rate": 1.7054316433738377e-05,
      "loss": 0.6374,
      "step": 1216
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5806138944924715,
      "learning_rate": 1.7049154213125606e-05,
      "loss": 0.6385,
      "step": 1217
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6924111242182183,
      "learning_rate": 1.7043988255925495e-05,
      "loss": 0.6322,
      "step": 1218
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.764822519738195,
      "learning_rate": 1.7038818564876396e-05,
      "loss": 1.1917,
      "step": 1219
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6792174882950155,
      "learning_rate": 1.703364514271864e-05,
      "loss": 0.659,
      "step": 1220
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6144495055762322,
      "learning_rate": 1.7028467992194534e-05,
      "loss": 0.5498,
      "step": 1221
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.548188389279914,
      "learning_rate": 1.7023287116048353e-05,
      "loss": 1.1907,
      "step": 1222
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.477160434465515,
      "learning_rate": 1.7018102517026368e-05,
      "loss": 1.1368,
      "step": 1223
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0378530390322407,
      "learning_rate": 1.70129141978768e-05,
      "loss": 0.6716,
      "step": 1224
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.829644375522536,
      "learning_rate": 1.7007722161349856e-05,
      "loss": 0.6335,
      "step": 1225
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.757752839707049,
      "learning_rate": 1.700252641019771e-05,
      "loss": 1.2072,
      "step": 1226
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.741352143771202,
      "learning_rate": 1.6997326947174504e-05,
      "loss": 0.5638,
      "step": 1227
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.536046486558781,
      "learning_rate": 1.6992123775036352e-05,
      "loss": 1.1734,
      "step": 1228
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8706908718052186,
      "learning_rate": 1.6986916896541324e-05,
      "loss": 0.6561,
      "step": 1229
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6096369511821624,
      "learning_rate": 1.6981706314449466e-05,
      "loss": 0.6242,
      "step": 1230
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.63162843105215,
      "learning_rate": 1.6976492031522786e-05,
      "loss": 0.6849,
      "step": 1231
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6781571295024238,
      "learning_rate": 1.697127405052524e-05,
      "loss": 0.6893,
      "step": 1232
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.3916338405539426,
      "learning_rate": 1.696605237422276e-05,
      "loss": 1.1084,
      "step": 1233
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.671812818024552,
      "learning_rate": 1.6960827005383236e-05,
      "loss": 0.5762,
      "step": 1234
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6560221649619782,
      "learning_rate": 1.69555979467765e-05,
      "loss": 0.5646,
      "step": 1235
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5028527051845886,
      "learning_rate": 1.695036520117436e-05,
      "loss": 0.5409,
      "step": 1236
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5415575195090299,
      "learning_rate": 1.694512877135056e-05,
      "loss": 0.5487,
      "step": 1237
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.7530805804164786,
      "learning_rate": 1.6939888660080813e-05,
      "loss": 1.1493,
      "step": 1238
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9566676602906072,
      "learning_rate": 1.6934644870142777e-05,
      "loss": 0.6353,
      "step": 1239
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8092140741912346,
      "learning_rate": 1.6929397404316052e-05,
      "loss": 0.6122,
      "step": 1240
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7928457346888556,
      "learning_rate": 1.69241462653822e-05,
      "loss": 0.6885,
      "step": 1241
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.626536970721218,
      "learning_rate": 1.6918891456124718e-05,
      "loss": 0.6081,
      "step": 1242
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.521124863564944,
      "learning_rate": 1.691363297932906e-05,
      "loss": 1.2155,
      "step": 1243
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8040562610112452,
      "learning_rate": 1.690837083778261e-05,
      "loss": 0.5322,
      "step": 1244
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6043364832661422,
      "learning_rate": 1.6903105034274715e-05,
      "loss": 0.6821,
      "step": 1245
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7150446002332453,
      "learning_rate": 1.689783557159664e-05,
      "loss": 0.6328,
      "step": 1246
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.620309137394163,
      "learning_rate": 1.6892562452541607e-05,
      "loss": 0.6214,
      "step": 1247
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6102666429017956,
      "learning_rate": 1.6887285679904764e-05,
      "loss": 0.5497,
      "step": 1248
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.877102552590226,
      "learning_rate": 1.688200525648321e-05,
      "loss": 1.1566,
      "step": 1249
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4965130016506465,
      "learning_rate": 1.687672118507596e-05,
      "loss": 0.6168,
      "step": 1250
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6081946022703506,
      "learning_rate": 1.687143346848398e-05,
      "loss": 0.6748,
      "step": 1251
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.7112302822426244,
      "learning_rate": 1.6866142109510158e-05,
      "loss": 1.1679,
      "step": 1252
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7286905886209758,
      "learning_rate": 1.6860847110959325e-05,
      "loss": 0.5946,
      "step": 1253
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.774249705588917,
      "learning_rate": 1.6855548475638225e-05,
      "loss": 1.161,
      "step": 1254
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8135950538250107,
      "learning_rate": 1.6850246206355538e-05,
      "loss": 0.6049,
      "step": 1255
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8531928449366326,
      "learning_rate": 1.684494030592187e-05,
      "loss": 0.5609,
      "step": 1256
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.6612033010214207,
      "learning_rate": 1.683963077714975e-05,
      "loss": 1.1708,
      "step": 1257
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.784885752883404,
      "learning_rate": 1.6834317622853635e-05,
      "loss": 1.2179,
      "step": 1258
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6275581919444457,
      "learning_rate": 1.6829000845849897e-05,
      "loss": 0.594,
      "step": 1259
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.698336513031606,
      "learning_rate": 1.6823680448956835e-05,
      "loss": 0.647,
      "step": 1260
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5895244308306828,
      "learning_rate": 1.6818356434994662e-05,
      "loss": 0.6256,
      "step": 1261
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9712908164577576,
      "learning_rate": 1.6813028806785512e-05,
      "loss": 1.1108,
      "step": 1262
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6556751302617345,
      "learning_rate": 1.6807697567153428e-05,
      "loss": 0.6131,
      "step": 1263
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7231709875932122,
      "learning_rate": 1.6802362718924374e-05,
      "loss": 0.5711,
      "step": 1264
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.7075374169398594,
      "learning_rate": 1.6797024264926228e-05,
      "loss": 1.1269,
      "step": 1265
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.770453301796851,
      "learning_rate": 1.6791682207988775e-05,
      "loss": 0.5959,
      "step": 1266
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8289689226281425,
      "learning_rate": 1.6786336550943708e-05,
      "loss": 0.6535,
      "step": 1267
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.7501460456691422,
      "learning_rate": 1.6780987296624635e-05,
      "loss": 1.1698,
      "step": 1268
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8018463820041601,
      "learning_rate": 1.6775634447867064e-05,
      "loss": 0.5741,
      "step": 1269
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.69911581233276,
      "learning_rate": 1.6770278007508416e-05,
      "loss": 1.23,
      "step": 1270
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5687436611939032,
      "learning_rate": 1.6764917978388004e-05,
      "loss": 0.5781,
      "step": 1271
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8344210820151179,
      "learning_rate": 1.675955436334706e-05,
      "loss": 0.6251,
      "step": 1272
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9996504592394078,
      "learning_rate": 1.6754187165228702e-05,
      "loss": 0.649,
      "step": 1273
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.758591923262015,
      "learning_rate": 1.6748816386877953e-05,
      "loss": 0.5449,
      "step": 1274
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.805684539967408,
      "learning_rate": 1.6743442031141733e-05,
      "loss": 1.132,
      "step": 1275
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.842366202727308,
      "learning_rate": 1.6738064100868864e-05,
      "loss": 0.6644,
      "step": 1276
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5427834073806672,
      "learning_rate": 1.6732682598910052e-05,
      "loss": 0.6369,
      "step": 1277
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.665529445321774,
      "learning_rate": 1.6727297528117897e-05,
      "loss": 0.6358,
      "step": 1278
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6882278683144107,
      "learning_rate": 1.672190889134691e-05,
      "loss": 0.7135,
      "step": 1279
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.649590549158943,
      "learning_rate": 1.671651669145347e-05,
      "loss": 0.5762,
      "step": 1280
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.790439704645241,
      "learning_rate": 1.671112093129585e-05,
      "loss": 0.6574,
      "step": 1281
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.8500223296611256,
      "learning_rate": 1.670572161373422e-05,
      "loss": 1.2381,
      "step": 1282
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4993973098496454,
      "learning_rate": 1.6700318741630623e-05,
      "loss": 0.5827,
      "step": 1283
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6353654861926334,
      "learning_rate": 1.6694912317848993e-05,
      "loss": 0.6149,
      "step": 1284
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7518926252450142,
      "learning_rate": 1.6689502345255145e-05,
      "loss": 0.6124,
      "step": 1285
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.704282839862532,
      "learning_rate": 1.6684088826716776e-05,
      "loss": 1.1496,
      "step": 1286
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5980975967818443,
      "learning_rate": 1.6678671765103466e-05,
      "loss": 0.6801,
      "step": 1287
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.3216954065243876,
      "learning_rate": 1.667325116328666e-05,
      "loss": 1.225,
      "step": 1288
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.4302317583848088,
      "learning_rate": 1.6667827024139697e-05,
      "loss": 1.1511,
      "step": 1289
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6983649415987,
      "learning_rate": 1.666239935053778e-05,
      "loss": 0.5473,
      "step": 1290
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.622692777194373,
      "learning_rate": 1.6656968145357988e-05,
      "loss": 1.2029,
      "step": 1291
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9099442095008596,
      "learning_rate": 1.6651533411479273e-05,
      "loss": 0.6822,
      "step": 1292
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.6290175761177306,
      "learning_rate": 1.664609515178246e-05,
      "loss": 1.1869,
      "step": 1293
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5720497020755386,
      "learning_rate": 1.6640653369150233e-05,
      "loss": 0.5684,
      "step": 1294
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.52645406160175,
      "learning_rate": 1.6635208066467158e-05,
      "loss": 1.1086,
      "step": 1295
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.59773042133447,
      "learning_rate": 1.6629759246619652e-05,
      "loss": 0.6012,
      "step": 1296
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5761248198316014,
      "learning_rate": 1.6624306912496006e-05,
      "loss": 0.6182,
      "step": 1297
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6903447359827073,
      "learning_rate": 1.6618851066986375e-05,
      "loss": 0.6283,
      "step": 1298
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.5051840051937346,
      "learning_rate": 1.6613391712982765e-05,
      "loss": 1.198,
      "step": 1299
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.295638688947587,
      "learning_rate": 1.6607928853379058e-05,
      "loss": 1.1544,
      "step": 1300
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7269480291642918,
      "learning_rate": 1.6602462491070975e-05,
      "loss": 0.5398,
      "step": 1301
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.842114669593502,
      "learning_rate": 1.6596992628956104e-05,
      "loss": 0.5966,
      "step": 1302
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.766134921326459,
      "learning_rate": 1.6591519269933897e-05,
      "loss": 0.6585,
      "step": 1303
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6471475424886821,
      "learning_rate": 1.658604241690564e-05,
      "loss": 0.5875,
      "step": 1304
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.5524420593863018,
      "learning_rate": 1.658056207277449e-05,
      "loss": 1.1127,
      "step": 1305
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9516313442047828,
      "learning_rate": 1.6575078240445437e-05,
      "loss": 0.6411,
      "step": 1306
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7911416312482378,
      "learning_rate": 1.6569590922825334e-05,
      "loss": 0.6364,
      "step": 1307
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5256216062017403,
      "learning_rate": 1.6564100122822874e-05,
      "loss": 0.5953,
      "step": 1308
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.6880705213353564,
      "learning_rate": 1.65586058433486e-05,
      "loss": 1.1422,
      "step": 1309
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.747145831867303,
      "learning_rate": 1.6553108087314898e-05,
      "loss": 0.6374,
      "step": 1310
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6325920201800133,
      "learning_rate": 1.6547606857635994e-05,
      "loss": 0.6193,
      "step": 1311
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6452802360706993,
      "learning_rate": 1.654210215722796e-05,
      "loss": 0.5495,
      "step": 1312
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7573161938209922,
      "learning_rate": 1.65365939890087e-05,
      "loss": 0.5181,
      "step": 1313
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.796647165965572,
      "learning_rate": 1.653108235589797e-05,
      "loss": 0.5602,
      "step": 1314
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.777691427354836,
      "learning_rate": 1.652556726081735e-05,
      "loss": 0.6053,
      "step": 1315
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8933756347866653,
      "learning_rate": 1.6520048706690263e-05,
      "loss": 0.6553,
      "step": 1316
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9068987945566658,
      "learning_rate": 1.651452669644196e-05,
      "loss": 0.6283,
      "step": 1317
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5653997407194098,
      "learning_rate": 1.6509001232999516e-05,
      "loss": 0.5866,
      "step": 1318
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.652714217572718,
      "learning_rate": 1.6503472319291864e-05,
      "loss": 1.1896,
      "step": 1319
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9026828057200624,
      "learning_rate": 1.6497939958249732e-05,
      "loss": 0.6269,
      "step": 1320
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2664248736472703,
      "learning_rate": 1.6492404152805705e-05,
      "loss": 1.2073,
      "step": 1321
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5367827920081563,
      "learning_rate": 1.6486864905894173e-05,
      "loss": 0.5776,
      "step": 1322
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7126944018840822,
      "learning_rate": 1.648132222045136e-05,
      "loss": 0.5799,
      "step": 1323
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.023216235403456,
      "learning_rate": 1.6475776099415312e-05,
      "loss": 0.6799,
      "step": 1324
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.644352018876203,
      "learning_rate": 1.647022654572589e-05,
      "loss": 0.6265,
      "step": 1325
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6826820856906914,
      "learning_rate": 1.6464673562324782e-05,
      "loss": 0.6102,
      "step": 1326
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7856791056391985,
      "learning_rate": 1.6459117152155496e-05,
      "loss": 0.6447,
      "step": 1327
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8899614650805192,
      "learning_rate": 1.645355731816334e-05,
      "loss": 0.6609,
      "step": 1328
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.6837560393587054,
      "learning_rate": 1.6447994063295454e-05,
      "loss": 1.1659,
      "step": 1329
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7314311264257864,
      "learning_rate": 1.6442427390500794e-05,
      "loss": 0.6196,
      "step": 1330
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6797301061011585,
      "learning_rate": 1.6436857302730105e-05,
      "loss": 0.6135,
      "step": 1331
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6067732142181128,
      "learning_rate": 1.6431283802935968e-05,
      "loss": 0.5895,
      "step": 1332
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5404532123588461,
      "learning_rate": 1.6425706894072753e-05,
      "loss": 0.5671,
      "step": 1333
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.703284432678176,
      "learning_rate": 1.6420126579096656e-05,
      "loss": 1.1678,
      "step": 1334
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7908068729219062,
      "learning_rate": 1.6414542860965655e-05,
      "loss": 0.6121,
      "step": 1335
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6873780278721084,
      "learning_rate": 1.640895574263956e-05,
      "loss": 0.6144,
      "step": 1336
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.439444965539078,
      "learning_rate": 1.6403365227079955e-05,
      "loss": 1.1126,
      "step": 1337
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5546829126693142,
      "learning_rate": 1.6397771317250242e-05,
      "loss": 0.5953,
      "step": 1338
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8633812035040556,
      "learning_rate": 1.6392174016115623e-05,
      "loss": 0.6646,
      "step": 1339
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8582616217426624,
      "learning_rate": 1.6386573326643083e-05,
      "loss": 0.5747,
      "step": 1340
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.016092650116414,
      "learning_rate": 1.6380969251801424e-05,
      "loss": 1.2022,
      "step": 1341
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.5671296545000546,
      "learning_rate": 1.6375361794561224e-05,
      "loss": 1.1768,
      "step": 1342
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4996191083889001,
      "learning_rate": 1.6369750957894866e-05,
      "loss": 0.6277,
      "step": 1343
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5802417985411963,
      "learning_rate": 1.6364136744776517e-05,
      "loss": 0.6093,
      "step": 1344
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.670713305298955,
      "learning_rate": 1.6358519158182137e-05,
      "loss": 0.554,
      "step": 1345
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.812629239807632,
      "learning_rate": 1.6352898201089472e-05,
      "loss": 1.206,
      "step": 1346
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7085745195187618,
      "learning_rate": 1.634727387647806e-05,
      "loss": 0.5484,
      "step": 1347
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.020077884556137,
      "learning_rate": 1.634164618732922e-05,
      "loss": 1.1843,
      "step": 1348
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3849189208260846,
      "learning_rate": 1.633601513662605e-05,
      "loss": 1.119,
      "step": 1349
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7743133087627403,
      "learning_rate": 1.633038072735344e-05,
      "loss": 0.6518,
      "step": 1350
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.691511220305703,
      "learning_rate": 1.6324742962498054e-05,
      "loss": 0.6025,
      "step": 1351
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.548001571692995,
      "learning_rate": 1.6319101845048333e-05,
      "loss": 0.5908,
      "step": 1352
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8264837098096594,
      "learning_rate": 1.63134573779945e-05,
      "loss": 0.6383,
      "step": 1353
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8833939516749072,
      "learning_rate": 1.630780956432855e-05,
      "loss": 0.6123,
      "step": 1354
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.6122974293443004,
      "learning_rate": 1.6302158407044257e-05,
      "loss": 1.1044,
      "step": 1355
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.629970473572377,
      "learning_rate": 1.629650390913716e-05,
      "loss": 0.5632,
      "step": 1356
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9322349016962121,
      "learning_rate": 1.6290846073604577e-05,
      "loss": 0.6315,
      "step": 1357
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.605656769389546,
      "learning_rate": 1.6285184903445586e-05,
      "loss": 0.6108,
      "step": 1358
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5766329720170071,
      "learning_rate": 1.627952040166104e-05,
      "loss": 0.6029,
      "step": 1359
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7595218894873386,
      "learning_rate": 1.627385257125355e-05,
      "loss": 0.6206,
      "step": 1360
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7156925076717822,
      "learning_rate": 1.6268181415227503e-05,
      "loss": 0.6522,
      "step": 1361
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.624595851801003,
      "learning_rate": 1.626250693658904e-05,
      "loss": 0.5937,
      "step": 1362
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8886357924577526,
      "learning_rate": 1.625682913834607e-05,
      "loss": 0.6411,
      "step": 1363
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.724368952705897,
      "learning_rate": 1.625114802350825e-05,
      "loss": 0.6017,
      "step": 1364
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.76037344255256,
      "learning_rate": 1.6245463595087006e-05,
      "loss": 0.6329,
      "step": 1365
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8374292498076126,
      "learning_rate": 1.623977585609552e-05,
      "loss": 0.5782,
      "step": 1366
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.837836865186828,
      "learning_rate": 1.6234084809548724e-05,
      "loss": 0.6864,
      "step": 1367
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.5764093253318583,
      "learning_rate": 1.6228390458463303e-05,
      "loss": 1.0746,
      "step": 1368
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5332200035110615,
      "learning_rate": 1.6222692805857698e-05,
      "loss": 0.5274,
      "step": 1369
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.3539796415341505,
      "learning_rate": 1.6216991854752097e-05,
      "loss": 1.131,
      "step": 1370
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5845456742907522,
      "learning_rate": 1.6211287608168436e-05,
      "loss": 0.5931,
      "step": 1371
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8711957417534504,
      "learning_rate": 1.62055800691304e-05,
      "loss": 0.6526,
      "step": 1372
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8598249308394856,
      "learning_rate": 1.6199869240663418e-05,
      "loss": 0.6512,
      "step": 1373
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.7485804278261505,
      "learning_rate": 1.6194155125794664e-05,
      "loss": 1.1595,
      "step": 1374
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8365225910617755,
      "learning_rate": 1.618843772755305e-05,
      "loss": 0.6578,
      "step": 1375
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.665157981021099,
      "learning_rate": 1.6182717048969238e-05,
      "loss": 0.59,
      "step": 1376
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8527593959616298,
      "learning_rate": 1.6176993093075618e-05,
      "loss": 0.5893,
      "step": 1377
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7377937055550503,
      "learning_rate": 1.617126586290632e-05,
      "loss": 0.6398,
      "step": 1378
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6924227861115104,
      "learning_rate": 1.6165535361497218e-05,
      "loss": 0.654,
      "step": 1379
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.719489925501425,
      "learning_rate": 1.6159801591885905e-05,
      "loss": 0.5757,
      "step": 1380
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7080741673413173,
      "learning_rate": 1.6154064557111716e-05,
      "loss": 0.5791,
      "step": 1381
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7199846908927676,
      "learning_rate": 1.614832426021572e-05,
      "loss": 0.6029,
      "step": 1382
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7668678831462008,
      "learning_rate": 1.6142580704240708e-05,
      "loss": 0.5959,
      "step": 1383
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7029735639142454,
      "learning_rate": 1.61368338922312e-05,
      "loss": 0.5735,
      "step": 1384
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7212871258011464,
      "learning_rate": 1.6131083827233443e-05,
      "loss": 0.6697,
      "step": 1385
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7616089976353297,
      "learning_rate": 1.6125330512295408e-05,
      "loss": 0.6326,
      "step": 1386
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.714322522643667,
      "learning_rate": 1.611957395046679e-05,
      "loss": 0.6194,
      "step": 1387
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.8361068742331654,
      "learning_rate": 1.6113814144799008e-05,
      "loss": 1.0905,
      "step": 1388
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.570303655693966,
      "learning_rate": 1.610805109834519e-05,
      "loss": 1.1859,
      "step": 1389
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8820765192032967,
      "learning_rate": 1.6102284814160194e-05,
      "loss": 0.592,
      "step": 1390
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6805065502859262,
      "learning_rate": 1.6096515295300583e-05,
      "loss": 0.6212,
      "step": 1391
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6491069947480497,
      "learning_rate": 1.6090742544824646e-05,
      "loss": 0.5937,
      "step": 1392
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.948452698662528,
      "learning_rate": 1.6084966565792374e-05,
      "loss": 0.6054,
      "step": 1393
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8913382487759034,
      "learning_rate": 1.607918736126548e-05,
      "loss": 0.714,
      "step": 1394
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.662222894722117,
      "learning_rate": 1.607340493430738e-05,
      "loss": 0.6378,
      "step": 1395
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8417607164920449,
      "learning_rate": 1.60676192879832e-05,
      "loss": 0.6809,
      "step": 1396
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8207011775530924,
      "learning_rate": 1.6061830425359768e-05,
      "loss": 0.6949,
      "step": 1397
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6901682819238202,
      "learning_rate": 1.6056038349505625e-05,
      "loss": 0.6012,
      "step": 1398
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8093845819834302,
      "learning_rate": 1.605024306349101e-05,
      "loss": 0.6369,
      "step": 1399
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.9830894340790786,
      "learning_rate": 1.604444457038787e-05,
      "loss": 1.2332,
      "step": 1400
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5872087510304196,
      "learning_rate": 1.6038642873269834e-05,
      "loss": 0.5345,
      "step": 1401
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.467120669644978,
      "learning_rate": 1.6032837975212256e-05,
      "loss": 1.172,
      "step": 1402
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.839118977291933,
      "learning_rate": 1.6027029879292167e-05,
      "loss": 0.6886,
      "step": 1403
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7076026162857905,
      "learning_rate": 1.60212185885883e-05,
      "loss": 0.6372,
      "step": 1404
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7907728074504363,
      "learning_rate": 1.6015404106181076e-05,
      "loss": 0.5849,
      "step": 1405
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.636982965489027,
      "learning_rate": 1.600958643515262e-05,
      "loss": 0.6389,
      "step": 1406
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8061014367280392,
      "learning_rate": 1.6003765578586737e-05,
      "loss": 0.5901,
      "step": 1407
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5437604889426955,
      "learning_rate": 1.5997941539568926e-05,
      "loss": 1.1757,
      "step": 1408
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6539932941626823,
      "learning_rate": 1.599211432118636e-05,
      "loss": 0.635,
      "step": 1409
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.38126533390053,
      "learning_rate": 1.5986283926527918e-05,
      "loss": 1.1753,
      "step": 1410
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8912373628836547,
      "learning_rate": 1.598045035868415e-05,
      "loss": 0.647,
      "step": 1411
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8841109101389026,
      "learning_rate": 1.5974613620747286e-05,
      "loss": 0.6465,
      "step": 1412
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7702297566993515,
      "learning_rate": 1.5968773715811245e-05,
      "loss": 0.5981,
      "step": 1413
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7689295780712266,
      "learning_rate": 1.5962930646971617e-05,
      "loss": 0.6072,
      "step": 1414
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8469574525992927,
      "learning_rate": 1.5957084417325675e-05,
      "loss": 0.5979,
      "step": 1415
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7394703090351658,
      "learning_rate": 1.5951235029972366e-05,
      "loss": 0.6324,
      "step": 1416
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.2049124760393424,
      "learning_rate": 1.5945382488012306e-05,
      "loss": 1.2163,
      "step": 1417
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8458000979596398,
      "learning_rate": 1.5939526794547787e-05,
      "loss": 0.6853,
      "step": 1418
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.4474781850826597,
      "learning_rate": 1.5933667952682777e-05,
      "loss": 1.1036,
      "step": 1419
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6331976979946747,
      "learning_rate": 1.59278059655229e-05,
      "loss": 0.5857,
      "step": 1420
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6366978438332804,
      "learning_rate": 1.592194083617546e-05,
      "loss": 0.6408,
      "step": 1421
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5533629980924153,
      "learning_rate": 1.5916072567749422e-05,
      "loss": 1.2018,
      "step": 1422
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7406530525881379,
      "learning_rate": 1.5910201163355405e-05,
      "loss": 0.5744,
      "step": 1423
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.6540885306332895,
      "learning_rate": 1.590432662610571e-05,
      "loss": 1.1284,
      "step": 1424
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.694420848193439,
      "learning_rate": 1.5898448959114286e-05,
      "loss": 0.6698,
      "step": 1425
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.823531152791464,
      "learning_rate": 1.5892568165496736e-05,
      "loss": 0.6805,
      "step": 1426
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6253219042782194,
      "learning_rate": 1.5886684248370336e-05,
      "loss": 0.699,
      "step": 1427
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6056177420915079,
      "learning_rate": 1.5880797210854003e-05,
      "loss": 0.5768,
      "step": 1428
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.580480099725914,
      "learning_rate": 1.587490705606832e-05,
      "loss": 1.1931,
      "step": 1429
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.704531287479664,
      "learning_rate": 1.5869013787135513e-05,
      "loss": 0.6433,
      "step": 1430
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6686192085327878,
      "learning_rate": 1.5863117407179465e-05,
      "loss": 0.5791,
      "step": 1431
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5258490399958795,
      "learning_rate": 1.5857217919325703e-05,
      "loss": 0.6254,
      "step": 1432
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7195273777957458,
      "learning_rate": 1.58513153267014e-05,
      "loss": 0.5863,
      "step": 1433
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.677068820106545,
      "learning_rate": 1.5845409632435385e-05,
      "loss": 1.208,
      "step": 1434
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8369873488720918,
      "learning_rate": 1.583950083965812e-05,
      "loss": 0.6547,
      "step": 1435
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5563907421959502,
      "learning_rate": 1.583358895150172e-05,
      "loss": 0.641,
      "step": 1436
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6306861074707926,
      "learning_rate": 1.5827673971099927e-05,
      "loss": 0.5723,
      "step": 1437
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6351488119526665,
      "learning_rate": 1.582175590158814e-05,
      "loss": 0.6179,
      "step": 1438
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8826951174013473,
      "learning_rate": 1.5815834746103376e-05,
      "loss": 0.6325,
      "step": 1439
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.4185125404459913,
      "learning_rate": 1.5809910507784302e-05,
      "loss": 1.1299,
      "step": 1440
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6816776546555283,
      "learning_rate": 1.580398318977121e-05,
      "loss": 0.6505,
      "step": 1441
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8554463197095146,
      "learning_rate": 1.579805279520604e-05,
      "loss": 0.6219,
      "step": 1442
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5389909788507192,
      "learning_rate": 1.5792119327232337e-05,
      "loss": 0.5432,
      "step": 1443
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7284550439583575,
      "learning_rate": 1.5786182788995302e-05,
      "loss": 0.6018,
      "step": 1444
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8825042754895167,
      "learning_rate": 1.5780243183641747e-05,
      "loss": 0.672,
      "step": 1445
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8356597963984314,
      "learning_rate": 1.5774300514320112e-05,
      "loss": 0.6381,
      "step": 1446
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8109143563854717,
      "learning_rate": 1.576835478418046e-05,
      "loss": 0.6239,
      "step": 1447
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5709035650088379,
      "learning_rate": 1.576240599637449e-05,
      "loss": 0.5873,
      "step": 1448
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6623856132362331,
      "learning_rate": 1.5756454154055508e-05,
      "loss": 0.5691,
      "step": 1449
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.981349319320395,
      "learning_rate": 1.5750499260378438e-05,
      "loss": 1.1066,
      "step": 1450
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6523601045446756,
      "learning_rate": 1.5744541318499828e-05,
      "loss": 0.6163,
      "step": 1451
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7356221163550567,
      "learning_rate": 1.573858033157784e-05,
      "loss": 0.6664,
      "step": 1452
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5268120683458641,
      "learning_rate": 1.5732616302772255e-05,
      "loss": 0.5549,
      "step": 1453
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6504663926784688,
      "learning_rate": 1.5726649235244455e-05,
      "loss": 0.5746,
      "step": 1454
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5618870564530263,
      "learning_rate": 1.572067913215744e-05,
      "loss": 0.5945,
      "step": 1455
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.945706818971144,
      "learning_rate": 1.571470599667582e-05,
      "loss": 0.5525,
      "step": 1456
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.548397694540301,
      "learning_rate": 1.5708729831965814e-05,
      "loss": 1.1833,
      "step": 1457
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5606325137532036,
      "learning_rate": 1.5702750641195235e-05,
      "loss": 0.5981,
      "step": 1458
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7051709989815345,
      "learning_rate": 1.5696768427533515e-05,
      "loss": 0.6478,
      "step": 1459
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.3616599816292876,
      "learning_rate": 1.569078319415168e-05,
      "loss": 1.1849,
      "step": 1460
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0236857350929975,
      "learning_rate": 1.5684794944222358e-05,
      "loss": 0.6487,
      "step": 1461
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.112756870360602,
      "learning_rate": 1.5678803680919776e-05,
      "loss": 0.6788,
      "step": 1462
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9580729310465834,
      "learning_rate": 1.5672809407419762e-05,
      "loss": 0.6208,
      "step": 1463
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7154500040843206,
      "learning_rate": 1.5666812126899732e-05,
      "loss": 0.5701,
      "step": 1464
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.74552476117838,
      "learning_rate": 1.56608118425387e-05,
      "loss": 1.1612,
      "step": 1465
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5884683567151885,
      "learning_rate": 1.565480855751728e-05,
      "loss": 0.6004,
      "step": 1466
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6489076432069736,
      "learning_rate": 1.564880227501766e-05,
      "loss": 0.6038,
      "step": 1467
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.844250813091195,
      "learning_rate": 1.5642792998223633e-05,
      "loss": 0.6408,
      "step": 1468
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7979354710038207,
      "learning_rate": 1.563678073032057e-05,
      "loss": 0.5732,
      "step": 1469
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5644202677276071,
      "learning_rate": 1.5630765474495426e-05,
      "loss": 0.5326,
      "step": 1470
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7220077625827221,
      "learning_rate": 1.562474723393675e-05,
      "loss": 0.626,
      "step": 1471
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9527869434155747,
      "learning_rate": 1.561872601183466e-05,
      "loss": 0.6799,
      "step": 1472
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.896717970075418,
      "learning_rate": 1.561270181138087e-05,
      "loss": 0.6735,
      "step": 1473
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6581461641952546,
      "learning_rate": 1.5606674635768656e-05,
      "loss": 0.5887,
      "step": 1474
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.454098781552538,
      "learning_rate": 1.5600644488192882e-05,
      "loss": 1.1442,
      "step": 1475
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7391794469709112,
      "learning_rate": 1.5594611371849986e-05,
      "loss": 0.6005,
      "step": 1476
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7607724886542881,
      "learning_rate": 1.5588575289937975e-05,
      "loss": 0.5824,
      "step": 1477
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.3746870794656854,
      "learning_rate": 1.558253624565643e-05,
      "loss": 1.162,
      "step": 1478
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6510284147093444,
      "learning_rate": 1.557649424220651e-05,
      "loss": 0.6204,
      "step": 1479
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5729492570153507,
      "learning_rate": 1.5570449282790928e-05,
      "loss": 0.5592,
      "step": 1480
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7286163494717603,
      "learning_rate": 1.556440137061398e-05,
      "loss": 0.6178,
      "step": 1481
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5363305574599773,
      "learning_rate": 1.555835050888151e-05,
      "loss": 0.5872,
      "step": 1482
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7362930933509495,
      "learning_rate": 1.555229670080094e-05,
      "loss": 0.6002,
      "step": 1483
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6307226520953806,
      "learning_rate": 1.554623994958125e-05,
      "loss": 0.6272,
      "step": 1484
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8593013107389507,
      "learning_rate": 1.5540180258432974e-05,
      "loss": 0.6082,
      "step": 1485
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6166753186309324,
      "learning_rate": 1.553411763056821e-05,
      "loss": 0.6546,
      "step": 1486
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7975483317522654,
      "learning_rate": 1.5528052069200613e-05,
      "loss": 0.6581,
      "step": 1487
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4397360369694887,
      "learning_rate": 1.5521983577545388e-05,
      "loss": 0.633,
      "step": 1488
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9139925835347034,
      "learning_rate": 1.55159121588193e-05,
      "loss": 0.6634,
      "step": 1489
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8478615401874137,
      "learning_rate": 1.550983781624066e-05,
      "loss": 0.6557,
      "step": 1490
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.520527261664088,
      "learning_rate": 1.5503760553029332e-05,
      "loss": 0.5452,
      "step": 1491
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6186251651510286,
      "learning_rate": 1.5497680372406727e-05,
      "loss": 0.6419,
      "step": 1492
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.561259580746221,
      "learning_rate": 1.5491597277595805e-05,
      "loss": 0.6156,
      "step": 1493
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6778448930758503,
      "learning_rate": 1.5485511271821065e-05,
      "loss": 0.6092,
      "step": 1494
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7468447481355753,
      "learning_rate": 1.5479422358308557e-05,
      "loss": 0.5723,
      "step": 1495
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6643141416590244,
      "learning_rate": 1.5473330540285863e-05,
      "loss": 0.5762,
      "step": 1496
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.520321624586812,
      "learning_rate": 1.546723582098211e-05,
      "loss": 1.1835,
      "step": 1497
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8183181095485312,
      "learning_rate": 1.5461138203627965e-05,
      "loss": 0.6707,
      "step": 1498
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5548361250911125,
      "learning_rate": 1.5455037691455626e-05,
      "loss": 0.6149,
      "step": 1499
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.596763150772806,
      "learning_rate": 1.5448934287698827e-05,
      "loss": 0.5522,
      "step": 1500
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.9804241983392536,
      "learning_rate": 1.544282799559284e-05,
      "loss": 1.1752,
      "step": 1501
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5236359615313269,
      "learning_rate": 1.543671881837446e-05,
      "loss": 0.6679,
      "step": 1502
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.568143080371396,
      "learning_rate": 1.5430606759282022e-05,
      "loss": 0.5628,
      "step": 1503
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.5305157594955356,
      "learning_rate": 1.542449182155537e-05,
      "loss": 1.1975,
      "step": 1504
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8273174591658896,
      "learning_rate": 1.5418374008435896e-05,
      "loss": 0.5872,
      "step": 1505
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.4491431912156476,
      "learning_rate": 1.54122533231665e-05,
      "loss": 1.1608,
      "step": 1506
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.4722452097708287,
      "learning_rate": 1.5406129768991615e-05,
      "loss": 1.1269,
      "step": 1507
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6270353933708268,
      "learning_rate": 1.540000334915718e-05,
      "loss": 0.5799,
      "step": 1508
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6764593938507582,
      "learning_rate": 1.5393874066910677e-05,
      "loss": 0.5985,
      "step": 1509
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.711299931468356,
      "learning_rate": 1.538774192550108e-05,
      "loss": 0.6881,
      "step": 1510
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7815780933576826,
      "learning_rate": 1.53816069281789e-05,
      "loss": 0.6026,
      "step": 1511
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5795378625672067,
      "learning_rate": 1.5375469078196138e-05,
      "loss": 0.5229,
      "step": 1512
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6595291203766682,
      "learning_rate": 1.5369328378806333e-05,
      "loss": 0.6608,
      "step": 1513
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6282431094326804,
      "learning_rate": 1.536318483326452e-05,
      "loss": 0.634,
      "step": 1514
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6706904926684616,
      "learning_rate": 1.5357038444827243e-05,
      "loss": 0.6077,
      "step": 1515
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6506236467792272,
      "learning_rate": 1.535088921675255e-05,
      "loss": 0.5967,
      "step": 1516
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.879835463059038,
      "learning_rate": 1.5344737152300004e-05,
      "loss": 0.6413,
      "step": 1517
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6970131194420202,
      "learning_rate": 1.533858225473067e-05,
      "loss": 0.5887,
      "step": 1518
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.448454136175947,
      "learning_rate": 1.533242452730711e-05,
      "loss": 1.1988,
      "step": 1519
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.379253131040836,
      "learning_rate": 1.532626397329338e-05,
      "loss": 1.1423,
      "step": 1520
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.86778783329462,
      "learning_rate": 1.532010059595505e-05,
      "loss": 0.7163,
      "step": 1521
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.3431968811983945,
      "learning_rate": 1.5313934398559175e-05,
      "loss": 1.205,
      "step": 1522
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5617300870246507,
      "learning_rate": 1.5307765384374306e-05,
      "loss": 0.5959,
      "step": 1523
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.778432954746471,
      "learning_rate": 1.5301593556670497e-05,
      "loss": 0.6713,
      "step": 1524
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8449088354701497,
      "learning_rate": 1.5295418918719275e-05,
      "loss": 0.5419,
      "step": 1525
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.761447135635392,
      "learning_rate": 1.5289241473793678e-05,
      "loss": 0.6897,
      "step": 1526
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7884917228547086,
      "learning_rate": 1.5283061225168214e-05,
      "loss": 0.5672,
      "step": 1527
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.9863493326011783,
      "learning_rate": 1.5276878176118884e-05,
      "loss": 1.21,
      "step": 1528
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5574726505460976,
      "learning_rate": 1.5270692329923176e-05,
      "loss": 0.5656,
      "step": 1529
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.450015022096298,
      "learning_rate": 1.5264503689860062e-05,
      "loss": 1.1436,
      "step": 1530
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7915534205665513,
      "learning_rate": 1.5258312259209983e-05,
      "loss": 0.592,
      "step": 1531
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7893041371064786,
      "learning_rate": 1.5252118041254877e-05,
      "loss": 0.5852,
      "step": 1532
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.6074328764002908,
      "learning_rate": 1.5245921039278143e-05,
      "loss": 1.1067,
      "step": 1533
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7841888923762739,
      "learning_rate": 1.5239721256564665e-05,
      "loss": 0.6088,
      "step": 1534
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.6902747558591074,
      "learning_rate": 1.5233518696400802e-05,
      "loss": 1.1357,
      "step": 1535
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.568365481722614,
      "learning_rate": 1.522731336207438e-05,
      "loss": 1.1521,
      "step": 1536
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6952524159081495,
      "learning_rate": 1.5221105256874697e-05,
      "loss": 0.5453,
      "step": 1537
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7143995892396202,
      "learning_rate": 1.521489438409252e-05,
      "loss": 0.6614,
      "step": 1538
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7356950170621657,
      "learning_rate": 1.5208680747020083e-05,
      "loss": 0.653,
      "step": 1539
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.58400373284469,
      "learning_rate": 1.5202464348951091e-05,
      "loss": 0.522,
      "step": 1540
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.0038588601223823,
      "learning_rate": 1.5196245193180699e-05,
      "loss": 1.1667,
      "step": 1541
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7192232098788787,
      "learning_rate": 1.5190023283005539e-05,
      "loss": 0.5701,
      "step": 1542
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5440527464162679,
      "learning_rate": 1.5183798621723692e-05,
      "loss": 0.5583,
      "step": 1543
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4667284441483754,
      "learning_rate": 1.5177571212634705e-05,
      "loss": 0.5648,
      "step": 1544
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7956183740538694,
      "learning_rate": 1.5171341059039574e-05,
      "loss": 0.6518,
      "step": 1545
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.76847003455728,
      "learning_rate": 1.5165108164240758e-05,
      "loss": 0.5892,
      "step": 1546
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6817916592574231,
      "learning_rate": 1.5158872531542163e-05,
      "loss": 0.6066,
      "step": 1547
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.319448829278522,
      "learning_rate": 1.5152634164249147e-05,
      "loss": 1.2252,
      "step": 1548
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8316502139485276,
      "learning_rate": 1.5146393065668518e-05,
      "loss": 0.6193,
      "step": 1549
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.835512028482519,
      "learning_rate": 1.5140149239108538e-05,
      "loss": 0.608,
      "step": 1550
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6462004818330054,
      "learning_rate": 1.5133902687878904e-05,
      "loss": 0.5818,
      "step": 1551
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.6844429614918783,
      "learning_rate": 1.5127653415290761e-05,
      "loss": 1.1491,
      "step": 1552
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7499012737932447,
      "learning_rate": 1.5121401424656704e-05,
      "loss": 0.6431,
      "step": 1553
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.6392268580544735,
      "learning_rate": 1.5115146719290761e-05,
      "loss": 1.0669,
      "step": 1554
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5967775244356817,
      "learning_rate": 1.5108889302508398e-05,
      "loss": 0.5728,
      "step": 1555
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2415729996934837,
      "learning_rate": 1.5102629177626527e-05,
      "loss": 1.116,
      "step": 1556
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7473680730164913,
      "learning_rate": 1.5096366347963485e-05,
      "loss": 0.6391,
      "step": 1557
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6956408369681986,
      "learning_rate": 1.509010081683905e-05,
      "loss": 0.6052,
      "step": 1558
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4766564033319385,
      "learning_rate": 1.508383258757443e-05,
      "loss": 0.5817,
      "step": 1559
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8144194653691015,
      "learning_rate": 1.5077561663492259e-05,
      "loss": 0.593,
      "step": 1560
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.7715282852826024,
      "learning_rate": 1.507128804791661e-05,
      "loss": 1.1509,
      "step": 1561
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.5099516817189813,
      "learning_rate": 1.5065011744172968e-05,
      "loss": 1.151,
      "step": 1562
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7930303751088321,
      "learning_rate": 1.5058732755588259e-05,
      "loss": 0.6263,
      "step": 1563
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5668146858141323,
      "learning_rate": 1.5052451085490816e-05,
      "loss": 0.5625,
      "step": 1564
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5695296134121013,
      "learning_rate": 1.5046166737210406e-05,
      "loss": 0.6116,
      "step": 1565
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7297715495127548,
      "learning_rate": 1.5039879714078211e-05,
      "loss": 0.631,
      "step": 1566
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7117611580907057,
      "learning_rate": 1.5033590019426832e-05,
      "loss": 0.7299,
      "step": 1567
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5544696498650392,
      "learning_rate": 1.5027297656590282e-05,
      "loss": 0.5824,
      "step": 1568
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6987572470904637,
      "learning_rate": 1.5021002628903994e-05,
      "loss": 0.6207,
      "step": 1569
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.8205897217454416,
      "learning_rate": 1.5014704939704806e-05,
      "loss": 1.0917,
      "step": 1570
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.599543899313396,
      "learning_rate": 1.5008404592330981e-05,
      "loss": 1.1775,
      "step": 1571
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7393414193860257,
      "learning_rate": 1.5002101590122173e-05,
      "loss": 0.7357,
      "step": 1572
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7772684371617913,
      "learning_rate": 1.4995795936419454e-05,
      "loss": 0.6261,
      "step": 1573
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.44661634948812,
      "learning_rate": 1.4989487634565302e-05,
      "loss": 1.1115,
      "step": 1574
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6276180682430883,
      "learning_rate": 1.4983176687903598e-05,
      "loss": 0.5484,
      "step": 1575
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8206795670568627,
      "learning_rate": 1.4976863099779617e-05,
      "loss": 0.5741,
      "step": 1576
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6508932748407918,
      "learning_rate": 1.4970546873540046e-05,
      "loss": 0.6545,
      "step": 1577
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7333503570861089,
      "learning_rate": 1.4964228012532963e-05,
      "loss": 0.6735,
      "step": 1578
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6320459407691106,
      "learning_rate": 1.4957906520107846e-05,
      "loss": 0.6352,
      "step": 1579
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4563856107515796,
      "learning_rate": 1.495158239961556e-05,
      "loss": 0.5602,
      "step": 1580
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.8975309538941136,
      "learning_rate": 1.4945255654408378e-05,
      "loss": 1.1268,
      "step": 1581
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8692636970258347,
      "learning_rate": 1.493892628783995e-05,
      "loss": 0.6243,
      "step": 1582
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5601664982367807,
      "learning_rate": 1.4932594303265324e-05,
      "loss": 0.5631,
      "step": 1583
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6196834971373253,
      "learning_rate": 1.492625970404093e-05,
      "loss": 0.5678,
      "step": 1584
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7375538609477317,
      "learning_rate": 1.4919922493524588e-05,
      "loss": 0.5943,
      "step": 1585
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5793428791687176,
      "learning_rate": 1.4913582675075504e-05,
      "loss": 0.6061,
      "step": 1586
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6752726586417843,
      "learning_rate": 1.490724025205426e-05,
      "loss": 0.5716,
      "step": 1587
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4874907246645235,
      "learning_rate": 1.4900895227822823e-05,
      "loss": 0.5699,
      "step": 1588
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6935069130365747,
      "learning_rate": 1.4894547605744541e-05,
      "loss": 0.5788,
      "step": 1589
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.058784819930293,
      "learning_rate": 1.4888197389184129e-05,
      "loss": 1.1302,
      "step": 1590
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7026078555432234,
      "learning_rate": 1.4881844581507692e-05,
      "loss": 0.6548,
      "step": 1591
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6418406243156103,
      "learning_rate": 1.4875489186082696e-05,
      "loss": 0.6488,
      "step": 1592
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8124370176580287,
      "learning_rate": 1.4869131206277984e-05,
      "loss": 0.5705,
      "step": 1593
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.772078560088565,
      "learning_rate": 1.486277064546377e-05,
      "loss": 0.6324,
      "step": 1594
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.787500937170533,
      "learning_rate": 1.4856407507011635e-05,
      "loss": 0.6074,
      "step": 1595
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6717744812514703,
      "learning_rate": 1.4850041794294523e-05,
      "loss": 0.6029,
      "step": 1596
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.606812347279957,
      "learning_rate": 1.4843673510686746e-05,
      "loss": 0.621,
      "step": 1597
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7039372711805494,
      "learning_rate": 1.4837302659563981e-05,
      "loss": 0.6466,
      "step": 1598
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7042294893182408,
      "learning_rate": 1.4830929244303262e-05,
      "loss": 0.7264,
      "step": 1599
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.519072654689364,
      "learning_rate": 1.4824553268282982e-05,
      "loss": 1.1964,
      "step": 1600
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6108498952154624,
      "learning_rate": 1.481817473488289e-05,
      "loss": 0.5646,
      "step": 1601
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.3576112881100166,
      "learning_rate": 1.4811793647484103e-05,
      "loss": 1.1106,
      "step": 1602
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6243790669629024,
      "learning_rate": 1.4805410009469072e-05,
      "loss": 0.6483,
      "step": 1603
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.3987117563246247,
      "learning_rate": 1.4799023824221614e-05,
      "loss": 1.1802,
      "step": 1604
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8316076508001329,
      "learning_rate": 1.4792635095126893e-05,
      "loss": 0.664,
      "step": 1605
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4282986846866121,
      "learning_rate": 1.478624382557142e-05,
      "loss": 0.568,
      "step": 1606
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.019614963111477,
      "learning_rate": 1.4779850018943054e-05,
      "loss": 0.535,
      "step": 1607
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7353910932214895,
      "learning_rate": 1.4773453678630998e-05,
      "loss": 0.5673,
      "step": 1608
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.973429974521228,
      "learning_rate": 1.47670548080258e-05,
      "loss": 1.1056,
      "step": 1609
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.4891868898604987,
      "learning_rate": 1.4760653410519345e-05,
      "loss": 1.1985,
      "step": 1610
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6731682337984393,
      "learning_rate": 1.475424948950486e-05,
      "loss": 0.5847,
      "step": 1611
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.584772345282334,
      "learning_rate": 1.4747843048376914e-05,
      "loss": 0.5548,
      "step": 1612
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6026947543260226,
      "learning_rate": 1.4741434090531405e-05,
      "loss": 0.582,
      "step": 1613
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5293141645007085,
      "learning_rate": 1.4735022619365564e-05,
      "loss": 0.5835,
      "step": 1614
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5801722137158427,
      "learning_rate": 1.4728608638277964e-05,
      "loss": 0.5947,
      "step": 1615
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6407144472014557,
      "learning_rate": 1.4722192150668495e-05,
      "loss": 0.6061,
      "step": 1616
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8194597221429543,
      "learning_rate": 1.471577315993839e-05,
      "loss": 0.6106,
      "step": 1617
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6997922187114898,
      "learning_rate": 1.47093516694902e-05,
      "loss": 0.6101,
      "step": 1618
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6503861395626747,
      "learning_rate": 1.4702927682727798e-05,
      "loss": 0.6003,
      "step": 1619
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7489358893874773,
      "learning_rate": 1.4696501203056395e-05,
      "loss": 0.6777,
      "step": 1620
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.330097990730009,
      "learning_rate": 1.4690072233882502e-05,
      "loss": 1.1096,
      "step": 1621
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.477393053506581,
      "learning_rate": 1.468364077861397e-05,
      "loss": 1.125,
      "step": 1622
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.805578510620248,
      "learning_rate": 1.4677206840659956e-05,
      "loss": 0.5859,
      "step": 1623
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.6604357733523476,
      "learning_rate": 1.4670770423430933e-05,
      "loss": 1.1772,
      "step": 1624
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.387414621477512,
      "learning_rate": 1.4664331530338696e-05,
      "loss": 1.153,
      "step": 1625
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.719561500661748,
      "learning_rate": 1.4657890164796345e-05,
      "loss": 0.6075,
      "step": 1626
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9028292016682662,
      "learning_rate": 1.4651446330218292e-05,
      "loss": 0.6163,
      "step": 1627
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8220399069037092,
      "learning_rate": 1.464500003002026e-05,
      "loss": 0.6287,
      "step": 1628
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.488860806834101,
      "learning_rate": 1.463855126761928e-05,
      "loss": 1.1873,
      "step": 1629
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.48378741523558,
      "learning_rate": 1.4632100046433686e-05,
      "loss": 1.114,
      "step": 1630
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7459203820548128,
      "learning_rate": 1.4625646369883113e-05,
      "loss": 0.6005,
      "step": 1631
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9714747632448952,
      "learning_rate": 1.4619190241388501e-05,
      "loss": 0.5717,
      "step": 1632
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7428907813108945,
      "learning_rate": 1.461273166437209e-05,
      "loss": 0.6503,
      "step": 1633
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6698110800883077,
      "learning_rate": 1.460627064225741e-05,
      "loss": 0.6193,
      "step": 1634
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.669656384680762,
      "learning_rate": 1.4599807178469301e-05,
      "loss": 0.6513,
      "step": 1635
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5506545234465552,
      "learning_rate": 1.4593341276433888e-05,
      "loss": 0.5981,
      "step": 1636
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.762145015238282,
      "learning_rate": 1.4586872939578589e-05,
      "loss": 1.1744,
      "step": 1637
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6286544626198234,
      "learning_rate": 1.4580402171332115e-05,
      "loss": 0.5405,
      "step": 1638
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.551788727069294,
      "learning_rate": 1.4573928975124466e-05,
      "loss": 0.5946,
      "step": 1639
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5450581840545543,
      "learning_rate": 1.4567453354386925e-05,
      "loss": 0.5572,
      "step": 1640
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.510468525559182,
      "learning_rate": 1.4560975312552066e-05,
      "loss": 1.1665,
      "step": 1641
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7736582639118805,
      "learning_rate": 1.4554494853053742e-05,
      "loss": 0.6374,
      "step": 1642
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7091778862054474,
      "learning_rate": 1.4548011979327088e-05,
      "loss": 0.5448,
      "step": 1643
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6565880876994257,
      "learning_rate": 1.4541526694808524e-05,
      "loss": 0.58,
      "step": 1644
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5607147158595038,
      "learning_rate": 1.4535039002935735e-05,
      "loss": 0.5732,
      "step": 1645
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5981102321127383,
      "learning_rate": 1.4528548907147697e-05,
      "loss": 0.5699,
      "step": 1646
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7057319431633138,
      "learning_rate": 1.4522056410884653e-05,
      "loss": 0.5989,
      "step": 1647
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8774727491699255,
      "learning_rate": 1.451556151758812e-05,
      "loss": 0.667,
      "step": 1648
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.8868564235486747,
      "learning_rate": 1.4509064230700881e-05,
      "loss": 1.129,
      "step": 1649
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7623406207024626,
      "learning_rate": 1.4502564553666998e-05,
      "loss": 0.6573,
      "step": 1650
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8119006813863023,
      "learning_rate": 1.4496062489931788e-05,
      "loss": 0.6589,
      "step": 1651
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6240237707815333,
      "learning_rate": 1.4489558042941844e-05,
      "loss": 0.6002,
      "step": 1652
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6426702199816439,
      "learning_rate": 1.4483051216145012e-05,
      "loss": 0.5653,
      "step": 1653
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.5347596872438363,
      "learning_rate": 1.4476542012990412e-05,
      "loss": 1.1509,
      "step": 1654
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6532912387200778,
      "learning_rate": 1.4470030436928404e-05,
      "loss": 0.5501,
      "step": 1655
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7913288000379202,
      "learning_rate": 1.4463516491410627e-05,
      "loss": 0.6279,
      "step": 1656
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.3131754995871066,
      "learning_rate": 1.4457000179889966e-05,
      "loss": 1.1424,
      "step": 1657
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6865216031868768,
      "learning_rate": 1.4450481505820562e-05,
      "loss": 0.5881,
      "step": 1658
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.6963787473595975,
      "learning_rate": 1.4443960472657803e-05,
      "loss": 1.2193,
      "step": 1659
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.981417941538851,
      "learning_rate": 1.4437437083858337e-05,
      "loss": 0.6465,
      "step": 1660
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7714049643515444,
      "learning_rate": 1.4430911342880051e-05,
      "loss": 0.5977,
      "step": 1661
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7190448219355652,
      "learning_rate": 1.442438325318209e-05,
      "loss": 0.5845,
      "step": 1662
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6625738730676,
      "learning_rate": 1.4417852818224833e-05,
      "loss": 0.5838,
      "step": 1663
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5803601681191264,
      "learning_rate": 1.441132004146991e-05,
      "loss": 0.5306,
      "step": 1664
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7202164088403464,
      "learning_rate": 1.4404784926380183e-05,
      "loss": 0.5652,
      "step": 1665
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.674295407597812,
      "learning_rate": 1.4398247476419764e-05,
      "loss": 0.6074,
      "step": 1666
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.8871003047309864,
      "learning_rate": 1.4391707695053999e-05,
      "loss": 1.1774,
      "step": 1667
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8591153185736213,
      "learning_rate": 1.4385165585749463e-05,
      "loss": 0.6366,
      "step": 1668
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8108645804451404,
      "learning_rate": 1.437862115197398e-05,
      "loss": 0.5312,
      "step": 1669
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7952455256695479,
      "learning_rate": 1.437207439719659e-05,
      "loss": 0.592,
      "step": 1670
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9004171515216381,
      "learning_rate": 1.4365525324887573e-05,
      "loss": 0.57,
      "step": 1671
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6927769431245576,
      "learning_rate": 1.4358973938518435e-05,
      "loss": 0.6175,
      "step": 1672
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5835932200203526,
      "learning_rate": 1.435242024156191e-05,
      "loss": 0.5819,
      "step": 1673
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7053250462902285,
      "learning_rate": 1.4345864237491956e-05,
      "loss": 0.5951,
      "step": 1674
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8157474825367264,
      "learning_rate": 1.4339305929783748e-05,
      "loss": 0.5618,
      "step": 1675
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.5957065692604697,
      "learning_rate": 1.4332745321913693e-05,
      "loss": 1.1109,
      "step": 1676
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.681912502055594,
      "learning_rate": 1.4326182417359407e-05,
      "loss": 1.1469,
      "step": 1677
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.964747814463453,
      "learning_rate": 1.4319617219599731e-05,
      "loss": 0.7263,
      "step": 1678
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.3633210714791435,
      "learning_rate": 1.4313049732114716e-05,
      "loss": 1.1063,
      "step": 1679
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5594437127882703,
      "learning_rate": 1.4306479958385633e-05,
      "loss": 0.5144,
      "step": 1680
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6148842169565092,
      "learning_rate": 1.429990790189496e-05,
      "loss": 0.6462,
      "step": 1681
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7419843347594373,
      "learning_rate": 1.4293333566126386e-05,
      "loss": 0.5399,
      "step": 1682
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.3356685190367705,
      "learning_rate": 1.4286756954564808e-05,
      "loss": 1.1105,
      "step": 1683
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6632253729159359,
      "learning_rate": 1.4280178070696332e-05,
      "loss": 0.6248,
      "step": 1684
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7803401282446933,
      "learning_rate": 1.4273596918008263e-05,
      "loss": 0.6468,
      "step": 1685
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.630179484239076,
      "learning_rate": 1.4267013499989113e-05,
      "loss": 1.1475,
      "step": 1686
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.330467245982155,
      "learning_rate": 1.4260427820128594e-05,
      "loss": 1.1603,
      "step": 1687
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5881318037903318,
      "learning_rate": 1.4253839881917619e-05,
      "loss": 0.6,
      "step": 1688
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5974558075200744,
      "learning_rate": 1.4247249688848293e-05,
      "loss": 0.5745,
      "step": 1689
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8130359768375028,
      "learning_rate": 1.4240657244413918e-05,
      "loss": 0.6122,
      "step": 1690
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.748242728382771,
      "learning_rate": 1.4234062552108993e-05,
      "loss": 0.6704,
      "step": 1691
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7124627841905953,
      "learning_rate": 1.4227465615429206e-05,
      "loss": 0.5727,
      "step": 1692
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.674749956278994,
      "learning_rate": 1.4220866437871432e-05,
      "loss": 1.1068,
      "step": 1693
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6071341605559388,
      "learning_rate": 1.4214265022933744e-05,
      "loss": 0.5725,
      "step": 1694
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6399705096921953,
      "learning_rate": 1.4207661374115384e-05,
      "loss": 0.5556,
      "step": 1695
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6333530290493163,
      "learning_rate": 1.4201055494916793e-05,
      "loss": 0.5756,
      "step": 1696
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.73709155872362,
      "learning_rate": 1.4194447388839585e-05,
      "loss": 0.5605,
      "step": 1697
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.881412385279308,
      "learning_rate": 1.4187837059386564e-05,
      "loss": 0.5857,
      "step": 1698
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.687356143313933,
      "learning_rate": 1.4181224510061702e-05,
      "loss": 0.587,
      "step": 1699
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7050303649519782,
      "learning_rate": 1.4174609744370153e-05,
      "loss": 0.6144,
      "step": 1700
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8675041368294543,
      "learning_rate": 1.4167992765818248e-05,
      "loss": 0.6229,
      "step": 1701
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7520639034333685,
      "learning_rate": 1.4161373577913486e-05,
      "loss": 0.5942,
      "step": 1702
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7847422311090926,
      "learning_rate": 1.4154752184164542e-05,
      "loss": 0.5903,
      "step": 1703
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.7009908256890123,
      "learning_rate": 1.4148128588081258e-05,
      "loss": 1.1098,
      "step": 1704
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.166789738465342,
      "learning_rate": 1.4141502793174639e-05,
      "loss": 1.108,
      "step": 1705
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8470859248479188,
      "learning_rate": 1.4134874802956866e-05,
      "loss": 0.6079,
      "step": 1706
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8313698319054088,
      "learning_rate": 1.4128244620941271e-05,
      "loss": 0.6782,
      "step": 1707
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6457941421761888,
      "learning_rate": 1.4121612250642357e-05,
      "loss": 0.6333,
      "step": 1708
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8411463504268863,
      "learning_rate": 1.4114977695575784e-05,
      "loss": 0.6028,
      "step": 1709
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6736535048331722,
      "learning_rate": 1.4108340959258371e-05,
      "loss": 0.6212,
      "step": 1710
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.741572299973798,
      "learning_rate": 1.4101702045208093e-05,
      "loss": 0.6286,
      "step": 1711
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7117062195434323,
      "learning_rate": 1.4095060956944075e-05,
      "loss": 0.6092,
      "step": 1712
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7857191063682594,
      "learning_rate": 1.4088417697986603e-05,
      "loss": 0.6082,
      "step": 1713
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7928487748082802,
      "learning_rate": 1.4081772271857108e-05,
      "loss": 0.6715,
      "step": 1714
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6733097862379502,
      "learning_rate": 1.4075124682078169e-05,
      "loss": 0.6534,
      "step": 1715
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5941956747638828,
      "learning_rate": 1.4068474932173517e-05,
      "loss": 0.6444,
      "step": 1716
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6071872989735452,
      "learning_rate": 1.406182302566802e-05,
      "loss": 0.5692,
      "step": 1717
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.61330270942428,
      "learning_rate": 1.4055168966087698e-05,
      "loss": 0.5462,
      "step": 1718
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5741360403102813,
      "learning_rate": 1.4048512756959705e-05,
      "loss": 0.6716,
      "step": 1719
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.730464617260422,
      "learning_rate": 1.4041854401812338e-05,
      "loss": 0.6902,
      "step": 1720
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7124040156808527,
      "learning_rate": 1.4035193904175033e-05,
      "loss": 0.5783,
      "step": 1721
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.449800678707343,
      "learning_rate": 1.402853126757836e-05,
      "loss": 1.1515,
      "step": 1722
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5123159974801506,
      "learning_rate": 1.4021866495554021e-05,
      "loss": 0.4643,
      "step": 1723
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.311733741867702,
      "learning_rate": 1.4015199591634851e-05,
      "loss": 1.1162,
      "step": 1724
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6715589101997608,
      "learning_rate": 1.4008530559354822e-05,
      "loss": 0.5325,
      "step": 1725
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.623900304385404,
      "learning_rate": 1.4001859402249021e-05,
      "loss": 0.5706,
      "step": 1726
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5785014972344538,
      "learning_rate": 1.3995186123853669e-05,
      "loss": 0.5966,
      "step": 1727
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.4083774768712347,
      "learning_rate": 1.3988510727706107e-05,
      "loss": 1.1655,
      "step": 1728
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.798698833608501,
      "learning_rate": 1.3981833217344811e-05,
      "loss": 0.6122,
      "step": 1729
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.922879498290252,
      "learning_rate": 1.3975153596309363e-05,
      "loss": 0.5517,
      "step": 1730
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7486425328599056,
      "learning_rate": 1.3968471868140472e-05,
      "loss": 0.6122,
      "step": 1731
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6506128228175811,
      "learning_rate": 1.396178803637996e-05,
      "loss": 0.6011,
      "step": 1732
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6830494974784607,
      "learning_rate": 1.3955102104570766e-05,
      "loss": 0.5951,
      "step": 1733
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4333570990999487,
      "learning_rate": 1.3948414076256945e-05,
      "loss": 0.5173,
      "step": 1734
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.731540234759156,
      "learning_rate": 1.3941723954983656e-05,
      "loss": 0.5927,
      "step": 1735
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.600519723080211,
      "learning_rate": 1.3935031744297174e-05,
      "loss": 0.5917,
      "step": 1736
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7418014156488728,
      "learning_rate": 1.3928337447744877e-05,
      "loss": 0.582,
      "step": 1737
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7987420509373113,
      "learning_rate": 1.3921641068875254e-05,
      "loss": 0.5938,
      "step": 1738
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5891666649374203,
      "learning_rate": 1.391494261123789e-05,
      "loss": 0.5209,
      "step": 1739
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5548318473748868,
      "learning_rate": 1.3908242078383479e-05,
      "loss": 0.5818,
      "step": 1740
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8966366698609893,
      "learning_rate": 1.3901539473863813e-05,
      "loss": 0.5661,
      "step": 1741
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.531016637317232,
      "learning_rate": 1.389483480123178e-05,
      "loss": 0.6065,
      "step": 1742
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6304259927423068,
      "learning_rate": 1.3888128064041371e-05,
      "loss": 0.5878,
      "step": 1743
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6301861607045887,
      "learning_rate": 1.388141926584766e-05,
      "loss": 0.6093,
      "step": 1744
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.644452778787221,
      "learning_rate": 1.3874708410206824e-05,
      "loss": 1.1519,
      "step": 1745
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5399370952352298,
      "learning_rate": 1.3867995500676123e-05,
      "loss": 0.5267,
      "step": 1746
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.3852063439628406,
      "learning_rate": 1.3861280540813916e-05,
      "loss": 1.1078,
      "step": 1747
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2745256647605765,
      "learning_rate": 1.385456353417963e-05,
      "loss": 1.1272,
      "step": 1748
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7376661828998752,
      "learning_rate": 1.3847844484333799e-05,
      "loss": 0.6124,
      "step": 1749
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6283413304751424,
      "learning_rate": 1.3841123394838025e-05,
      "loss": 0.5621,
      "step": 1750
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6203049085232897,
      "learning_rate": 1.3834400269255e-05,
      "loss": 0.6006,
      "step": 1751
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7575711423135525,
      "learning_rate": 1.3827675111148486e-05,
      "loss": 0.6563,
      "step": 1752
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.4292932127614164,
      "learning_rate": 1.3820947924083332e-05,
      "loss": 1.0903,
      "step": 1753
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6073897194133002,
      "learning_rate": 1.3814218711625457e-05,
      "loss": 0.5793,
      "step": 1754
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6650492288614875,
      "learning_rate": 1.3807487477341857e-05,
      "loss": 0.6064,
      "step": 1755
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.634652407655739,
      "learning_rate": 1.3800754224800597e-05,
      "loss": 0.6287,
      "step": 1756
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7758842856695347,
      "learning_rate": 1.379401895757081e-05,
      "loss": 0.6049,
      "step": 1757
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7446484994824374,
      "learning_rate": 1.3787281679222697e-05,
      "loss": 0.5969,
      "step": 1758
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.6840123782309573,
      "learning_rate": 1.3780542393327534e-05,
      "loss": 1.1201,
      "step": 1759
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6198870894816149,
      "learning_rate": 1.377380110345765e-05,
      "loss": 0.5953,
      "step": 1760
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8762089527338508,
      "learning_rate": 1.3767057813186444e-05,
      "loss": 0.59,
      "step": 1761
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.657919807978634,
      "learning_rate": 1.376031252608837e-05,
      "loss": 0.5404,
      "step": 1762
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8182535964685456,
      "learning_rate": 1.3753565245738943e-05,
      "loss": 0.5891,
      "step": 1763
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9081825701584172,
      "learning_rate": 1.3746815975714734e-05,
      "loss": 0.6287,
      "step": 1764
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5327820522948394,
      "learning_rate": 1.3740064719593372e-05,
      "loss": 0.6249,
      "step": 1765
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8126522782401508,
      "learning_rate": 1.373331148095353e-05,
      "loss": 0.74,
      "step": 1766
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.6411267531673968,
      "learning_rate": 1.3726556263374944e-05,
      "loss": 1.1718,
      "step": 1767
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6731400184081087,
      "learning_rate": 1.3719799070438388e-05,
      "loss": 0.5305,
      "step": 1768
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.627836061902162,
      "learning_rate": 1.3713039905725689e-05,
      "loss": 0.6084,
      "step": 1769
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5504422134132279,
      "learning_rate": 1.3706278772819716e-05,
      "loss": 0.5619,
      "step": 1770
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6646823399581978,
      "learning_rate": 1.3699515675304387e-05,
      "loss": 0.5209,
      "step": 1771
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7698440817245595,
      "learning_rate": 1.3692750616764654e-05,
      "loss": 0.611,
      "step": 1772
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5616190622832795,
      "learning_rate": 1.3685983600786516e-05,
      "loss": 0.529,
      "step": 1773
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.667698490411001,
      "learning_rate": 1.3679214630957e-05,
      "loss": 0.6248,
      "step": 1774
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6760501179360154,
      "learning_rate": 1.367244371086418e-05,
      "loss": 0.5674,
      "step": 1775
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0266698547396347,
      "learning_rate": 1.3665670844097158e-05,
      "loss": 0.5975,
      "step": 1776
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7794379443755879,
      "learning_rate": 1.3658896034246065e-05,
      "loss": 0.5352,
      "step": 1777
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7436947264035452,
      "learning_rate": 1.3652119284902066e-05,
      "loss": 0.6224,
      "step": 1778
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.3826882540837744,
      "learning_rate": 1.364534059965735e-05,
      "loss": 1.1001,
      "step": 1779
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7890476391240282,
      "learning_rate": 1.3638559982105139e-05,
      "loss": 0.6149,
      "step": 1780
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.5225856758814325,
      "learning_rate": 1.3631777435839672e-05,
      "loss": 1.1486,
      "step": 1781
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6274203062887698,
      "learning_rate": 1.362499296445622e-05,
      "loss": 0.6134,
      "step": 1782
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6596347023769986,
      "learning_rate": 1.3618206571551064e-05,
      "loss": 0.6127,
      "step": 1783
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6414244599792132,
      "learning_rate": 1.3611418260721507e-05,
      "loss": 0.5857,
      "step": 1784
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.644317030273236,
      "learning_rate": 1.3604628035565872e-05,
      "loss": 0.5724,
      "step": 1785
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.079574432042475,
      "learning_rate": 1.3597835899683495e-05,
      "loss": 1.1715,
      "step": 1786
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7089047285898917,
      "learning_rate": 1.3591041856674723e-05,
      "loss": 0.6259,
      "step": 1787
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.441568014964335,
      "learning_rate": 1.3584245910140914e-05,
      "loss": 1.1418,
      "step": 1788
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6200899933604196,
      "learning_rate": 1.3577448063684436e-05,
      "loss": 0.5645,
      "step": 1789
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7190417436777359,
      "learning_rate": 1.3570648320908664e-05,
      "loss": 0.6149,
      "step": 1790
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7628021564208212,
      "learning_rate": 1.3563846685417978e-05,
      "loss": 0.5641,
      "step": 1791
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8858441482101616,
      "learning_rate": 1.3557043160817762e-05,
      "loss": 0.5716,
      "step": 1792
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7551934365150803,
      "learning_rate": 1.3550237750714406e-05,
      "loss": 0.6079,
      "step": 1793
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8158174454993494,
      "learning_rate": 1.3543430458715285e-05,
      "loss": 0.641,
      "step": 1794
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6964600595736883,
      "learning_rate": 1.3536621288428789e-05,
      "loss": 0.5117,
      "step": 1795
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7613175791190259,
      "learning_rate": 1.3529810243464292e-05,
      "loss": 0.6111,
      "step": 1796
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5305154682604694,
      "learning_rate": 1.3522997327432167e-05,
      "loss": 0.5873,
      "step": 1797
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.826521834020669,
      "learning_rate": 1.3516182543943774e-05,
      "loss": 0.5659,
      "step": 1798
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6993905598581474,
      "learning_rate": 1.350936589661147e-05,
      "loss": 0.6165,
      "step": 1799
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8321011299776726,
      "learning_rate": 1.3502547389048594e-05,
      "loss": 0.6493,
      "step": 1800
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6705061720826115,
      "learning_rate": 1.3495727024869471e-05,
      "loss": 0.6175,
      "step": 1801
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.623241288991305,
      "learning_rate": 1.3488904807689419e-05,
      "loss": 1.1852,
      "step": 1802
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.964999361913074,
      "learning_rate": 1.3482080741124723e-05,
      "loss": 0.6006,
      "step": 1803
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2234840916418332,
      "learning_rate": 1.347525482879266e-05,
      "loss": 1.117,
      "step": 1804
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.3755442794151644,
      "learning_rate": 1.3468427074311485e-05,
      "loss": 1.1475,
      "step": 1805
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5757912621217196,
      "learning_rate": 1.3461597481300426e-05,
      "loss": 0.5318,
      "step": 1806
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7073632299583719,
      "learning_rate": 1.3454766053379685e-05,
      "loss": 0.594,
      "step": 1807
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7683345562451673,
      "learning_rate": 1.3447932794170436e-05,
      "loss": 0.6126,
      "step": 1808
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.4638048161747497,
      "learning_rate": 1.3441097707294828e-05,
      "loss": 1.1757,
      "step": 1809
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.801432077309056,
      "learning_rate": 1.3434260796375976e-05,
      "loss": 0.6163,
      "step": 1810
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7407768728191528,
      "learning_rate": 1.3427422065037958e-05,
      "loss": 0.6562,
      "step": 1811
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8788058823190343,
      "learning_rate": 1.3420581516905828e-05,
      "loss": 0.6231,
      "step": 1812
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5713652385851198,
      "learning_rate": 1.3413739155605594e-05,
      "loss": 0.5055,
      "step": 1813
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.648551360581249,
      "learning_rate": 1.3406894984764227e-05,
      "loss": 0.578,
      "step": 1814
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.8722189785753,
      "learning_rate": 1.3400049008009655e-05,
      "loss": 1.1073,
      "step": 1815
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9112232579231068,
      "learning_rate": 1.339320122897077e-05,
      "loss": 0.6895,
      "step": 1816
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7073856740261926,
      "learning_rate": 1.3386351651277415e-05,
      "loss": 0.5974,
      "step": 1817
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.627443979480907,
      "learning_rate": 1.3379500278560383e-05,
      "loss": 0.5068,
      "step": 1818
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6488261021979602,
      "learning_rate": 1.3372647114451422e-05,
      "loss": 0.597,
      "step": 1819
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.854409019064135,
      "learning_rate": 1.3365792162583237e-05,
      "loss": 0.652,
      "step": 1820
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9123167545026711,
      "learning_rate": 1.3358935426589466e-05,
      "loss": 0.6128,
      "step": 1821
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.622794755330186,
      "learning_rate": 1.33520769101047e-05,
      "loss": 1.2479,
      "step": 1822
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7667627212926023,
      "learning_rate": 1.3345216616764478e-05,
      "loss": 0.6274,
      "step": 1823
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7835744565585194,
      "learning_rate": 1.3338354550205274e-05,
      "loss": 0.6306,
      "step": 1824
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.557356995965248,
      "learning_rate": 1.3331490714064506e-05,
      "loss": 1.2031,
      "step": 1825
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.50197428971728,
      "learning_rate": 1.332462511198053e-05,
      "loss": 0.5747,
      "step": 1826
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7064558925272502,
      "learning_rate": 1.331775774759263e-05,
      "loss": 0.6375,
      "step": 1827
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5792486569138129,
      "learning_rate": 1.331088862454104e-05,
      "loss": 0.638,
      "step": 1828
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6633664404224406,
      "learning_rate": 1.3304017746466912e-05,
      "loss": 0.611,
      "step": 1829
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7569244826992139,
      "learning_rate": 1.3297145117012336e-05,
      "loss": 0.5837,
      "step": 1830
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.8233392598473754,
      "learning_rate": 1.3290270739820326e-05,
      "loss": 1.1321,
      "step": 1831
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.457198572911903,
      "learning_rate": 1.3283394618534821e-05,
      "loss": 0.6017,
      "step": 1832
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6414571986606852,
      "learning_rate": 1.3276516756800692e-05,
      "loss": 0.5474,
      "step": 1833
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.409549551890036,
      "learning_rate": 1.326963715826373e-05,
      "loss": 1.1302,
      "step": 1834
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.598789713586044,
      "learning_rate": 1.3262755826570642e-05,
      "loss": 0.61,
      "step": 1835
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6560118041280802,
      "learning_rate": 1.3255872765369055e-05,
      "loss": 0.5663,
      "step": 1836
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6871910461120716,
      "learning_rate": 1.3248987978307517e-05,
      "loss": 0.5863,
      "step": 1837
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.7129103969115897,
      "learning_rate": 1.3242101469035491e-05,
      "loss": 1.182,
      "step": 1838
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.571818771845435,
      "learning_rate": 1.3235213241203344e-05,
      "loss": 1.139,
      "step": 1839
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.3696723183507133,
      "learning_rate": 1.3228323298462362e-05,
      "loss": 1.1511,
      "step": 1840
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5402600323565265,
      "learning_rate": 1.3221431644464743e-05,
      "loss": 0.5908,
      "step": 1841
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.641846986268751,
      "learning_rate": 1.321453828286358e-05,
      "loss": 0.6119,
      "step": 1842
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6738441134866373,
      "learning_rate": 1.3207643217312884e-05,
      "loss": 0.5637,
      "step": 1843
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9664022583616396,
      "learning_rate": 1.3200746451467561e-05,
      "loss": 0.6854,
      "step": 1844
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.5625356515062974,
      "learning_rate": 1.3193847988983422e-05,
      "loss": 1.1,
      "step": 1845
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6018843084009933,
      "learning_rate": 1.3186947833517172e-05,
      "loss": 0.6489,
      "step": 1846
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.475882530034724,
      "learning_rate": 1.3180045988726423e-05,
      "loss": 1.1894,
      "step": 1847
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8012781577804429,
      "learning_rate": 1.3173142458269679e-05,
      "loss": 0.5912,
      "step": 1848
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7887498538575366,
      "learning_rate": 1.3166237245806333e-05,
      "loss": 0.6389,
      "step": 1849
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.814454798674173,
      "learning_rate": 1.3159330354996667e-05,
      "loss": 0.5746,
      "step": 1850
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.375448269548399,
      "learning_rate": 1.3152421789501865e-05,
      "loss": 1.1189,
      "step": 1851
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.611055391208583,
      "learning_rate": 1.3145511552983989e-05,
      "loss": 0.4974,
      "step": 1852
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7840075016273351,
      "learning_rate": 1.3138599649105988e-05,
      "loss": 0.5549,
      "step": 1853
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.745339968848681,
      "learning_rate": 1.3131686081531698e-05,
      "loss": 0.5556,
      "step": 1854
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.748546126843255,
      "learning_rate": 1.3124770853925835e-05,
      "loss": 1.1037,
      "step": 1855
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.674061538140648,
      "learning_rate": 1.3117853969953996e-05,
      "loss": 0.6349,
      "step": 1856
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6344994163459192,
      "learning_rate": 1.3110935433282653e-05,
      "loss": 0.6161,
      "step": 1857
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.558363086539802,
      "learning_rate": 1.3104015247579156e-05,
      "loss": 1.1646,
      "step": 1858
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8344986530466705,
      "learning_rate": 1.3097093416511733e-05,
      "loss": 0.6721,
      "step": 1859
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7153346895495476,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 0.5721,
      "step": 1860
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6535736144569015,
      "learning_rate": 1.3083244832962353e-05,
      "loss": 0.6158,
      "step": 1861
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7275934981140733,
      "learning_rate": 1.3076318087821202e-05,
      "loss": 0.5878,
      "step": 1862
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.821389742951793,
      "learning_rate": 1.3069389711997717e-05,
      "loss": 0.6445,
      "step": 1863
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.7882439294128747,
      "learning_rate": 1.306245970916447e-05,
      "loss": 1.1264,
      "step": 1864
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.4327701356218374,
      "learning_rate": 1.3055528082994889e-05,
      "loss": 1.1716,
      "step": 1865
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.510844271546311,
      "learning_rate": 1.3048594837163258e-05,
      "loss": 1.2118,
      "step": 1866
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7886163194573892,
      "learning_rate": 1.3041659975344727e-05,
      "loss": 0.5705,
      "step": 1867
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6320017188779097,
      "learning_rate": 1.3034723501215301e-05,
      "loss": 0.6297,
      "step": 1868
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.495090463804097,
      "learning_rate": 1.3027785418451836e-05,
      "loss": 1.1622,
      "step": 1869
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5648392638047572,
      "learning_rate": 1.3020845730732043e-05,
      "loss": 0.5482,
      "step": 1870
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7615383099830855,
      "learning_rate": 1.3013904441734486e-05,
      "loss": 0.6147,
      "step": 1871
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7483714834058242,
      "learning_rate": 1.3006961555138576e-05,
      "loss": 0.6155,
      "step": 1872
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.614383697778909,
      "learning_rate": 1.3000017074624564e-05,
      "loss": 0.6357,
      "step": 1873
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7245465734139593,
      "learning_rate": 1.2993071003873561e-05,
      "loss": 0.6166,
      "step": 1874
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.639112683196964,
      "learning_rate": 1.2986123346567505e-05,
      "loss": 0.598,
      "step": 1875
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7052798309076724,
      "learning_rate": 1.2979174106389185e-05,
      "loss": 0.5568,
      "step": 1876
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7126346773934962,
      "learning_rate": 1.2972223287022229e-05,
      "loss": 0.6094,
      "step": 1877
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.755076422606652,
      "learning_rate": 1.2965270892151096e-05,
      "loss": 0.6389,
      "step": 1878
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.48915147112965,
      "learning_rate": 1.2958316925461084e-05,
      "loss": 1.1928,
      "step": 1879
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8841190261971936,
      "learning_rate": 1.2951361390638328e-05,
      "loss": 0.6322,
      "step": 1880
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9646472243881814,
      "learning_rate": 1.2944404291369783e-05,
      "loss": 0.574,
      "step": 1881
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7481522709803057,
      "learning_rate": 1.2937445631343243e-05,
      "loss": 0.572,
      "step": 1882
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4556223839645281,
      "learning_rate": 1.2930485414247331e-05,
      "loss": 0.591,
      "step": 1883
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6193755885550065,
      "learning_rate": 1.2923523643771485e-05,
      "loss": 0.5885,
      "step": 1884
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8287465737263573,
      "learning_rate": 1.2916560323605973e-05,
      "loss": 0.5729,
      "step": 1885
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8141638531586577,
      "learning_rate": 1.2909595457441885e-05,
      "loss": 0.5708,
      "step": 1886
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6223207318688224,
      "learning_rate": 1.2902629048971132e-05,
      "loss": 0.5975,
      "step": 1887
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8039656519869345,
      "learning_rate": 1.2895661101886435e-05,
      "loss": 0.6046,
      "step": 1888
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.8858470263062195,
      "learning_rate": 1.2888691619881338e-05,
      "loss": 1.091,
      "step": 1889
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6713443799438694,
      "learning_rate": 1.2881720606650196e-05,
      "loss": 0.6463,
      "step": 1890
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.3778829821615264,
      "learning_rate": 1.2874748065888174e-05,
      "loss": 1.1441,
      "step": 1891
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7313241588331958,
      "learning_rate": 1.286777400129125e-05,
      "loss": 0.7005,
      "step": 1892
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.650682238271914,
      "learning_rate": 1.2860798416556211e-05,
      "loss": 0.5484,
      "step": 1893
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6742039875362877,
      "learning_rate": 1.285382131538064e-05,
      "loss": 0.5749,
      "step": 1894
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.3217939459458554,
      "learning_rate": 1.2846842701462936e-05,
      "loss": 1.1263,
      "step": 1895
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.658052528989017,
      "learning_rate": 1.2839862578502291e-05,
      "loss": 0.6502,
      "step": 1896
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7337525526423259,
      "learning_rate": 1.2832880950198705e-05,
      "loss": 0.5929,
      "step": 1897
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7249264193577551,
      "learning_rate": 1.2825897820252968e-05,
      "loss": 0.6158,
      "step": 1898
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4509568966110171,
      "learning_rate": 1.2818913192366668e-05,
      "loss": 0.5447,
      "step": 1899
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7282275795858721,
      "learning_rate": 1.2811927070242198e-05,
      "loss": 0.607,
      "step": 1900
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.613505053320466,
      "learning_rate": 1.2804939457582721e-05,
      "loss": 0.5422,
      "step": 1901
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6858417111035477,
      "learning_rate": 1.2797950358092211e-05,
      "loss": 0.5973,
      "step": 1902
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.694241590380876,
      "learning_rate": 1.2790959775475415e-05,
      "loss": 0.6834,
      "step": 1903
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6988623004234435,
      "learning_rate": 1.2783967713437882e-05,
      "loss": 0.5992,
      "step": 1904
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6954923613810926,
      "learning_rate": 1.2776974175685928e-05,
      "loss": 0.5811,
      "step": 1905
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.644393145471712,
      "learning_rate": 1.2769979165926662e-05,
      "loss": 0.5448,
      "step": 1906
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5680897049716667,
      "learning_rate": 1.2762982687867972e-05,
      "loss": 0.626,
      "step": 1907
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4370606695880679,
      "learning_rate": 1.2755984745218522e-05,
      "loss": 0.5608,
      "step": 1908
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.558237935083819,
      "learning_rate": 1.2748985341687755e-05,
      "loss": 0.5566,
      "step": 1909
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7592666909293355,
      "learning_rate": 1.2741984480985883e-05,
      "loss": 0.6276,
      "step": 1910
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.3313032964013005,
      "learning_rate": 1.2734982166823898e-05,
      "loss": 1.195,
      "step": 1911
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.560190295383378,
      "learning_rate": 1.2727978402913555e-05,
      "loss": 1.0969,
      "step": 1912
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6039456973281394,
      "learning_rate": 1.2720973192967384e-05,
      "loss": 0.605,
      "step": 1913
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.728637258187728,
      "learning_rate": 1.2713966540698678e-05,
      "loss": 0.6525,
      "step": 1914
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7210877197207484,
      "learning_rate": 1.2706958449821492e-05,
      "loss": 0.6496,
      "step": 1915
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7560234235497931,
      "learning_rate": 1.2699948924050651e-05,
      "loss": 0.645,
      "step": 1916
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.853517096093763,
      "learning_rate": 1.2692937967101732e-05,
      "loss": 0.5092,
      "step": 1917
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.690217707436799,
      "learning_rate": 1.268592558269108e-05,
      "loss": 1.1738,
      "step": 1918
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6621771581704434,
      "learning_rate": 1.2678911774535787e-05,
      "loss": 0.5637,
      "step": 1919
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6770844557277886,
      "learning_rate": 1.2671896546353705e-05,
      "loss": 0.6483,
      "step": 1920
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7215991040175502,
      "learning_rate": 1.2664879901863442e-05,
      "loss": 0.6026,
      "step": 1921
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6738699880766807,
      "learning_rate": 1.2657861844784347e-05,
      "loss": 0.5846,
      "step": 1922
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.760765756845452,
      "learning_rate": 1.2650842378836528e-05,
      "loss": 0.5706,
      "step": 1923
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6115026972896305,
      "learning_rate": 1.2643821507740832e-05,
      "loss": 0.6409,
      "step": 1924
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.464293320743512,
      "learning_rate": 1.2636799235218857e-05,
      "loss": 1.08,
      "step": 1925
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7361875188948555,
      "learning_rate": 1.2629775564992938e-05,
      "loss": 0.6434,
      "step": 1926
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7173003280115033,
      "learning_rate": 1.2622750500786158e-05,
      "loss": 0.6292,
      "step": 1927
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7036996142121066,
      "learning_rate": 1.2615724046322329e-05,
      "loss": 0.5404,
      "step": 1928
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6513291469501379,
      "learning_rate": 1.260869620532601e-05,
      "loss": 0.6319,
      "step": 1929
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.257020702853397,
      "learning_rate": 1.2601666981522491e-05,
      "loss": 1.1058,
      "step": 1930
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5966369547263592,
      "learning_rate": 1.2594636378637795e-05,
      "loss": 0.5661,
      "step": 1931
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.622840852054772,
      "learning_rate": 1.258760440039867e-05,
      "loss": 0.6115,
      "step": 1932
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4898135877895349,
      "learning_rate": 1.2580571050532607e-05,
      "loss": 0.5932,
      "step": 1933
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5742873965528212,
      "learning_rate": 1.257353633276781e-05,
      "loss": 0.5951,
      "step": 1934
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.836430928435858,
      "learning_rate": 1.2566500250833216e-05,
      "loss": 0.5987,
      "step": 1935
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6676569579524205,
      "learning_rate": 1.255946280845848e-05,
      "loss": 0.53,
      "step": 1936
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.59474040542199,
      "learning_rate": 1.2552424009373988e-05,
      "loss": 0.6001,
      "step": 1937
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5825097199267708,
      "learning_rate": 1.2545383857310828e-05,
      "loss": 0.594,
      "step": 1938
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5300825679766465,
      "learning_rate": 1.2538342356000827e-05,
      "loss": 0.5861,
      "step": 1939
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.389287328091009,
      "learning_rate": 1.253129950917651e-05,
      "loss": 1.126,
      "step": 1940
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7365359203351234,
      "learning_rate": 1.252425532057112e-05,
      "loss": 0.6313,
      "step": 1941
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.651362566526585,
      "learning_rate": 1.2517209793918617e-05,
      "loss": 0.5956,
      "step": 1942
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6961462663624307,
      "learning_rate": 1.2510162932953657e-05,
      "loss": 0.6072,
      "step": 1943
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8443867359502824,
      "learning_rate": 1.2503114741411619e-05,
      "loss": 0.6225,
      "step": 1944
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.680038841124997,
      "learning_rate": 1.2496065223028579e-05,
      "loss": 0.6497,
      "step": 1945
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4576643351294427,
      "learning_rate": 1.2489014381541317e-05,
      "loss": 1.092,
      "step": 1946
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7537838617922685,
      "learning_rate": 1.2481962220687312e-05,
      "loss": 0.6112,
      "step": 1947
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4579507813107413,
      "learning_rate": 1.2474908744204748e-05,
      "loss": 1.1358,
      "step": 1948
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6867699686256512,
      "learning_rate": 1.2467853955832501e-05,
      "loss": 0.5649,
      "step": 1949
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.718857529407968,
      "learning_rate": 1.2460797859310147e-05,
      "loss": 0.5937,
      "step": 1950
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7402459822819303,
      "learning_rate": 1.2453740458377953e-05,
      "loss": 0.6174,
      "step": 1951
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6525036777795559,
      "learning_rate": 1.244668175677688e-05,
      "loss": 0.6147,
      "step": 1952
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8025745388464027,
      "learning_rate": 1.2439621758248568e-05,
      "loss": 0.6271,
      "step": 1953
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.77187984881886,
      "learning_rate": 1.2432560466535362e-05,
      "loss": 1.079,
      "step": 1954
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5835722774347902,
      "learning_rate": 1.2425497885380277e-05,
      "loss": 0.5768,
      "step": 1955
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5302835870649152,
      "learning_rate": 1.2418434018527024e-05,
      "loss": 0.5711,
      "step": 1956
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6441943768890228,
      "learning_rate": 1.2411368869719981e-05,
      "loss": 0.5637,
      "step": 1957
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4887808134671912,
      "learning_rate": 1.2404302442704219e-05,
      "loss": 0.571,
      "step": 1958
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.711804971805729,
      "learning_rate": 1.2397234741225481e-05,
      "loss": 0.6604,
      "step": 1959
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.92908983215272,
      "learning_rate": 1.2390165769030183e-05,
      "loss": 0.6461,
      "step": 1960
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.694773964519694,
      "learning_rate": 1.238309552986542e-05,
      "loss": 0.5736,
      "step": 1961
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.8702888757170246,
      "learning_rate": 1.2376024027478959e-05,
      "loss": 1.1545,
      "step": 1962
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.7307856840105478,
      "learning_rate": 1.2368951265619224e-05,
      "loss": 1.1299,
      "step": 1963
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.385415641464575,
      "learning_rate": 1.2361877248035326e-05,
      "loss": 1.1492,
      "step": 1964
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4761793687830578,
      "learning_rate": 1.2354801978477024e-05,
      "loss": 0.5005,
      "step": 1965
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7292146002059328,
      "learning_rate": 1.2347725460694754e-05,
      "loss": 0.65,
      "step": 1966
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.1381525271707207,
      "learning_rate": 1.234064769843961e-05,
      "loss": 1.0236,
      "step": 1967
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8371491409664815,
      "learning_rate": 1.2333568695463336e-05,
      "loss": 0.5719,
      "step": 1968
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.445538124504697,
      "learning_rate": 1.2326488455518346e-05,
      "loss": 1.1452,
      "step": 1969
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6387418292454827,
      "learning_rate": 1.2319406982357705e-05,
      "loss": 0.6144,
      "step": 1970
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.682202254739706,
      "learning_rate": 1.2312324279735131e-05,
      "loss": 1.098,
      "step": 1971
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4345419103763843,
      "learning_rate": 1.2305240351404994e-05,
      "loss": 1.134,
      "step": 1972
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7023989322274817,
      "learning_rate": 1.2298155201122319e-05,
      "loss": 0.5336,
      "step": 1973
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.915428223403397,
      "learning_rate": 1.2291068832642768e-05,
      "loss": 1.0822,
      "step": 1974
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.1280030733849706,
      "learning_rate": 1.2283981249722656e-05,
      "loss": 1.145,
      "step": 1975
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.51727395578578,
      "learning_rate": 1.2276892456118944e-05,
      "loss": 1.1372,
      "step": 1976
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5702509798774613,
      "learning_rate": 1.2269802455589228e-05,
      "loss": 0.6051,
      "step": 1977
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6536066726300727,
      "learning_rate": 1.2262711251891746e-05,
      "loss": 0.5642,
      "step": 1978
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.743756359133617,
      "learning_rate": 1.2255618848785378e-05,
      "loss": 0.5608,
      "step": 1979
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8408703044556436,
      "learning_rate": 1.2248525250029635e-05,
      "loss": 0.6574,
      "step": 1980
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8753522295731793,
      "learning_rate": 1.2241430459384663e-05,
      "loss": 0.6791,
      "step": 1981
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9733717846040069,
      "learning_rate": 1.2234334480611238e-05,
      "loss": 0.6019,
      "step": 1982
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6437002936330072,
      "learning_rate": 1.2227237317470774e-05,
      "loss": 0.5889,
      "step": 1983
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.4496669051555653,
      "learning_rate": 1.2220138973725304e-05,
      "loss": 1.1573,
      "step": 1984
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6628962640018718,
      "learning_rate": 1.2213039453137485e-05,
      "loss": 0.6184,
      "step": 1985
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9515943657203072,
      "learning_rate": 1.2205938759470608e-05,
      "loss": 0.6073,
      "step": 1986
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6720205698180135,
      "learning_rate": 1.2198836896488578e-05,
      "loss": 0.5005,
      "step": 1987
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0931414538784394,
      "learning_rate": 1.2191733867955922e-05,
      "loss": 0.71,
      "step": 1988
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6706182510354506,
      "learning_rate": 1.2184629677637785e-05,
      "loss": 0.6187,
      "step": 1989
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5248570834950432,
      "learning_rate": 1.2177524329299927e-05,
      "loss": 0.4969,
      "step": 1990
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.815489907708576,
      "learning_rate": 1.2170417826708723e-05,
      "loss": 0.6357,
      "step": 1991
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.4861799707026875,
      "learning_rate": 1.216331017363116e-05,
      "loss": 1.1362,
      "step": 1992
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.716402340215733,
      "learning_rate": 1.2156201373834835e-05,
      "loss": 0.5488,
      "step": 1993
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.648324806266951,
      "learning_rate": 1.214909143108795e-05,
      "loss": 0.5875,
      "step": 1994
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7665183676323957,
      "learning_rate": 1.2141980349159312e-05,
      "loss": 0.6356,
      "step": 1995
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7168644457319562,
      "learning_rate": 1.213486813181834e-05,
      "loss": 0.5521,
      "step": 1996
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6803309445846688,
      "learning_rate": 1.2127754782835048e-05,
      "loss": 0.6256,
      "step": 1997
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5968532776617146,
      "learning_rate": 1.2120640305980056e-05,
      "loss": 0.552,
      "step": 1998
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.348582139927269,
      "learning_rate": 1.2113524705024566e-05,
      "loss": 1.1612,
      "step": 1999
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4087627977347044,
      "learning_rate": 1.2106407983740396e-05,
      "loss": 0.5037,
      "step": 2000
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3021139198792273,
      "learning_rate": 1.2099290145899947e-05,
      "loss": 1.114,
      "step": 2001
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.4388563210885237,
      "learning_rate": 1.2092171195276214e-05,
      "loss": 1.1092,
      "step": 2002
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.769943038610591,
      "learning_rate": 1.208505113564278e-05,
      "loss": 0.6273,
      "step": 2003
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7966134420301108,
      "learning_rate": 1.2077929970773823e-05,
      "loss": 0.6701,
      "step": 2004
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5554221124570908,
      "learning_rate": 1.2070807704444092e-05,
      "loss": 0.5727,
      "step": 2005
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.4727245226185213,
      "learning_rate": 1.2063684340428939e-05,
      "loss": 1.1864,
      "step": 2006
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3470960106527343,
      "learning_rate": 1.205655988250428e-05,
      "loss": 1.0907,
      "step": 2007
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.288001068109873,
      "learning_rate": 1.2049434334446626e-05,
      "loss": 1.1833,
      "step": 2008
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6471806267264721,
      "learning_rate": 1.2042307700033055e-05,
      "loss": 0.562,
      "step": 2009
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.321077081520099,
      "learning_rate": 1.2035179983041225e-05,
      "loss": 1.0817,
      "step": 2010
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6165302018327243,
      "learning_rate": 1.2028051187249367e-05,
      "loss": 0.6394,
      "step": 2011
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.331938104358532,
      "learning_rate": 1.2020921316436289e-05,
      "loss": 1.1196,
      "step": 2012
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7065693883426578,
      "learning_rate": 1.2013790374381357e-05,
      "loss": 0.6008,
      "step": 2013
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6876290241949479,
      "learning_rate": 1.200665836486452e-05,
      "loss": 0.6062,
      "step": 2014
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6981789967801844,
      "learning_rate": 1.1999525291666279e-05,
      "loss": 0.5607,
      "step": 2015
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.014000738656561,
      "learning_rate": 1.1992391158567704e-05,
      "loss": 1.1013,
      "step": 2016
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6430740341337526,
      "learning_rate": 1.1985255969350433e-05,
      "loss": 0.5773,
      "step": 2017
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.763799397823008,
      "learning_rate": 1.1978119727796655e-05,
      "loss": 0.6559,
      "step": 2018
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5570678626042826,
      "learning_rate": 1.1970982437689122e-05,
      "loss": 0.5588,
      "step": 2019
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.625835376414832,
      "learning_rate": 1.1963844102811139e-05,
      "loss": 0.6007,
      "step": 2020
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.619361924151827,
      "learning_rate": 1.1956704726946563e-05,
      "loss": 1.1052,
      "step": 2021
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.733485412249686,
      "learning_rate": 1.1949564313879812e-05,
      "loss": 0.5398,
      "step": 2022
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6533408611525715,
      "learning_rate": 1.1942422867395842e-05,
      "loss": 0.6295,
      "step": 2023
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6892139802163944,
      "learning_rate": 1.1935280391280166e-05,
      "loss": 0.5973,
      "step": 2024
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6073159125058312,
      "learning_rate": 1.1928136889318835e-05,
      "loss": 0.5315,
      "step": 2025
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9098151580288707,
      "learning_rate": 1.1920992365298452e-05,
      "loss": 0.6282,
      "step": 2026
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6125268134497075,
      "learning_rate": 1.1913846823006155e-05,
      "loss": 0.5733,
      "step": 2027
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.472605456769534,
      "learning_rate": 1.1906700266229622e-05,
      "loss": 1.0348,
      "step": 2028
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5827131954368563,
      "learning_rate": 1.1899552698757078e-05,
      "loss": 0.609,
      "step": 2029
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.344215907869108,
      "learning_rate": 1.1892404124377272e-05,
      "loss": 1.1457,
      "step": 2030
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7565311736981724,
      "learning_rate": 1.1885254546879493e-05,
      "loss": 0.598,
      "step": 2031
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9067578009613495,
      "learning_rate": 1.1878103970053558e-05,
      "loss": 0.6035,
      "step": 2032
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9740521831815812,
      "learning_rate": 1.1870952397689818e-05,
      "loss": 0.6527,
      "step": 2033
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6854137310315647,
      "learning_rate": 1.1863799833579152e-05,
      "loss": 0.5819,
      "step": 2034
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7804786106819819,
      "learning_rate": 1.185664628151296e-05,
      "loss": 0.6594,
      "step": 2035
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5921266344316147,
      "learning_rate": 1.1849491745283166e-05,
      "loss": 0.6839,
      "step": 2036
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6673283429115087,
      "learning_rate": 1.1842336228682221e-05,
      "loss": 0.6098,
      "step": 2037
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8768113285107946,
      "learning_rate": 1.183517973550309e-05,
      "loss": 0.6542,
      "step": 2038
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.8235482331380544,
      "learning_rate": 1.1828022269539261e-05,
      "loss": 1.2262,
      "step": 2039
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7453211405077889,
      "learning_rate": 1.1820863834584735e-05,
      "loss": 0.5744,
      "step": 2040
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.235323447941194,
      "learning_rate": 1.1813704434434023e-05,
      "loss": 1.1687,
      "step": 2041
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5121023881231705,
      "learning_rate": 1.1806544072882154e-05,
      "loss": 1.1293,
      "step": 2042
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.788890984957402,
      "learning_rate": 1.1799382753724662e-05,
      "loss": 0.6239,
      "step": 2043
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4550675327597005,
      "learning_rate": 1.1792220480757588e-05,
      "loss": 1.189,
      "step": 2044
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5452907475268167,
      "learning_rate": 1.178505725777749e-05,
      "loss": 0.5314,
      "step": 2045
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.2570764505631993,
      "learning_rate": 1.1777893088581408e-05,
      "loss": 1.1237,
      "step": 2046
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6369154070168561,
      "learning_rate": 1.1770727976966905e-05,
      "loss": 0.6436,
      "step": 2047
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.369265982148791,
      "learning_rate": 1.1763561926732033e-05,
      "loss": 1.1714,
      "step": 2048
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.697197013788082,
      "learning_rate": 1.175639494167534e-05,
      "loss": 0.6055,
      "step": 2049
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5456822992551382,
      "learning_rate": 1.1749227025595873e-05,
      "loss": 1.1434,
      "step": 2050
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7612726490816317,
      "learning_rate": 1.1742058182293178e-05,
      "loss": 0.5628,
      "step": 2051
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6098517311402294,
      "learning_rate": 1.173488841556728e-05,
      "loss": 0.5565,
      "step": 2052
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4723040818801834,
      "learning_rate": 1.1727717729218702e-05,
      "loss": 1.1409,
      "step": 2053
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8094468619402517,
      "learning_rate": 1.1720546127048454e-05,
      "loss": 0.564,
      "step": 2054
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7925500274430588,
      "learning_rate": 1.171337361285803e-05,
      "loss": 0.6364,
      "step": 2055
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6501648195468628,
      "learning_rate": 1.1706200190449403e-05,
      "loss": 0.5798,
      "step": 2056
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.304903271805004,
      "learning_rate": 1.169902586362504e-05,
      "loss": 1.0866,
      "step": 2057
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6090270683970376,
      "learning_rate": 1.1691850636187872e-05,
      "loss": 0.597,
      "step": 2058
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7580599361355098,
      "learning_rate": 1.1684674511941317e-05,
      "loss": 0.6132,
      "step": 2059
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.69018225169346,
      "learning_rate": 1.1677497494689267e-05,
      "loss": 0.5886,
      "step": 2060
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.332769538120496,
      "learning_rate": 1.167031958823609e-05,
      "loss": 1.1229,
      "step": 2061
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6430797008091658,
      "learning_rate": 1.1663140796386614e-05,
      "loss": 0.5861,
      "step": 2062
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.396773922844689,
      "learning_rate": 1.165596112294615e-05,
      "loss": 1.0885,
      "step": 2063
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5339495820111766,
      "learning_rate": 1.1648780571720466e-05,
      "loss": 0.5415,
      "step": 2064
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4259224856012682,
      "learning_rate": 1.1641599146515805e-05,
      "loss": 0.5712,
      "step": 2065
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5966115800283314,
      "learning_rate": 1.1634416851138865e-05,
      "loss": 0.5232,
      "step": 2066
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5549224719391328,
      "learning_rate": 1.1627233689396809e-05,
      "loss": 0.6249,
      "step": 2067
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6639242492786714,
      "learning_rate": 1.1620049665097258e-05,
      "loss": 0.6139,
      "step": 2068
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5976456288157324,
      "learning_rate": 1.1612864782048288e-05,
      "loss": 0.5538,
      "step": 2069
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8396824357454742,
      "learning_rate": 1.1605679044058438e-05,
      "loss": 0.636,
      "step": 2070
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5519940621061479,
      "learning_rate": 1.1598492454936692e-05,
      "loss": 0.5873,
      "step": 2071
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.714548047628342,
      "learning_rate": 1.1591305018492492e-05,
      "loss": 1.1086,
      "step": 2072
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.3545176701668775,
      "learning_rate": 1.158411673853572e-05,
      "loss": 1.1241,
      "step": 2073
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7505100317261992,
      "learning_rate": 1.1576927618876711e-05,
      "loss": 0.554,
      "step": 2074
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.5247086550758167,
      "learning_rate": 1.1569737663326251e-05,
      "loss": 1.1221,
      "step": 2075
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5924867905326288,
      "learning_rate": 1.1562546875695559e-05,
      "loss": 0.563,
      "step": 2076
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6212230050105834,
      "learning_rate": 1.1555355259796296e-05,
      "loss": 0.5806,
      "step": 2077
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7729367942266112,
      "learning_rate": 1.1548162819440568e-05,
      "loss": 0.598,
      "step": 2078
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8935515321946796,
      "learning_rate": 1.1540969558440911e-05,
      "loss": 0.6027,
      "step": 2079
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.764248527013258,
      "learning_rate": 1.1533775480610305e-05,
      "loss": 0.6294,
      "step": 2080
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.4415265337114738,
      "learning_rate": 1.1526580589762157e-05,
      "loss": 1.0878,
      "step": 2081
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6517916347488648,
      "learning_rate": 1.1519384889710306e-05,
      "loss": 0.6189,
      "step": 2082
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.577253232274316,
      "learning_rate": 1.1512188384269016e-05,
      "loss": 0.6146,
      "step": 2083
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8138887232722394,
      "learning_rate": 1.1504991077252986e-05,
      "loss": 0.5808,
      "step": 2084
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7387850104018385,
      "learning_rate": 1.149779297247733e-05,
      "loss": 0.6025,
      "step": 2085
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8204256616065808,
      "learning_rate": 1.1490594073757598e-05,
      "loss": 0.6127,
      "step": 2086
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.6400196919755428,
      "learning_rate": 1.1483394384909748e-05,
      "loss": 1.1444,
      "step": 2087
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6824201521956466,
      "learning_rate": 1.1476193909750161e-05,
      "loss": 0.6114,
      "step": 2088
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.812203134696253,
      "learning_rate": 1.1468992652095643e-05,
      "loss": 0.5532,
      "step": 2089
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5532004751526454,
      "learning_rate": 1.14617906157634e-05,
      "loss": 0.5199,
      "step": 2090
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6178628996426099,
      "learning_rate": 1.1454587804571062e-05,
      "loss": 0.6204,
      "step": 2091
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7249991187037943,
      "learning_rate": 1.1447384222336668e-05,
      "loss": 0.5303,
      "step": 2092
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5162985717300004,
      "learning_rate": 1.1440179872878662e-05,
      "loss": 0.5586,
      "step": 2093
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5931507742070046,
      "learning_rate": 1.1432974760015895e-05,
      "loss": 0.5906,
      "step": 2094
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.722923753428884,
      "learning_rate": 1.1425768887567625e-05,
      "loss": 0.6289,
      "step": 2095
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.752717295322088,
      "learning_rate": 1.1418562259353513e-05,
      "loss": 0.5557,
      "step": 2096
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6563484157327701,
      "learning_rate": 1.141135487919362e-05,
      "loss": 0.5925,
      "step": 2097
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7167573430247678,
      "learning_rate": 1.1404146750908407e-05,
      "loss": 0.5696,
      "step": 2098
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.4119972832996353,
      "learning_rate": 1.1396937878318723e-05,
      "loss": 1.0881,
      "step": 2099
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6979755203741767,
      "learning_rate": 1.1389728265245823e-05,
      "loss": 0.5394,
      "step": 2100
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4784951670105944,
      "learning_rate": 1.138251791551135e-05,
      "loss": 0.5983,
      "step": 2101
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.519790988427037,
      "learning_rate": 1.1375306832937334e-05,
      "loss": 0.6207,
      "step": 2102
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8247039318932445,
      "learning_rate": 1.1368095021346201e-05,
      "loss": 0.6086,
      "step": 2103
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7020450194612704,
      "learning_rate": 1.1360882484560756e-05,
      "loss": 0.6893,
      "step": 2104
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6630677295019327,
      "learning_rate": 1.1353669226404192e-05,
      "loss": 0.6025,
      "step": 2105
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5127485510887584,
      "learning_rate": 1.1346455250700084e-05,
      "loss": 0.548,
      "step": 2106
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.6367374607525997,
      "learning_rate": 1.1339240561272389e-05,
      "loss": 1.1611,
      "step": 2107
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.753257515950397,
      "learning_rate": 1.133202516194544e-05,
      "loss": 0.6232,
      "step": 2108
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.365216502675573,
      "learning_rate": 1.1324809056543942e-05,
      "loss": 1.0882,
      "step": 2109
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7883653495103387,
      "learning_rate": 1.1317592248892986e-05,
      "loss": 0.6722,
      "step": 2110
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5579786405078484,
      "learning_rate": 1.1310374742818025e-05,
      "loss": 0.5923,
      "step": 2111
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7919079086323761,
      "learning_rate": 1.1303156542144885e-05,
      "loss": 0.609,
      "step": 2112
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6783472815127303,
      "learning_rate": 1.1295937650699762e-05,
      "loss": 0.6257,
      "step": 2113
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.73630703563603,
      "learning_rate": 1.1288718072309219e-05,
      "loss": 0.6105,
      "step": 2114
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6673669395781268,
      "learning_rate": 1.128149781080018e-05,
      "loss": 0.5434,
      "step": 2115
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.65455781834748,
      "learning_rate": 1.1274276869999928e-05,
      "loss": 0.6098,
      "step": 2116
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6448110832216416,
      "learning_rate": 1.1267055253736117e-05,
      "loss": 0.6732,
      "step": 2117
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6818519099230806,
      "learning_rate": 1.1259832965836746e-05,
      "loss": 0.6603,
      "step": 2118
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5939454576420256,
      "learning_rate": 1.125261001013018e-05,
      "loss": 0.6349,
      "step": 2119
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.658947710794797,
      "learning_rate": 1.1245386390445132e-05,
      "loss": 0.5946,
      "step": 2120
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.817400128800252,
      "learning_rate": 1.1238162110610673e-05,
      "loss": 0.6047,
      "step": 2121
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7980635175300523,
      "learning_rate": 1.1230937174456216e-05,
      "loss": 0.6235,
      "step": 2122
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4518074649854447,
      "learning_rate": 1.122371158581153e-05,
      "loss": 0.5185,
      "step": 2123
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3785575815575557,
      "learning_rate": 1.1216485348506725e-05,
      "loss": 1.1662,
      "step": 2124
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.343500267777028,
      "learning_rate": 1.1209258466372254e-05,
      "loss": 1.1272,
      "step": 2125
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5893724035307053,
      "learning_rate": 1.1202030943238915e-05,
      "loss": 0.5716,
      "step": 2126
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6143968497764125,
      "learning_rate": 1.1194802782937841e-05,
      "loss": 0.5506,
      "step": 2127
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7563878929764165,
      "learning_rate": 1.1187573989300515e-05,
      "loss": 0.5809,
      "step": 2128
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.711307428149786,
      "learning_rate": 1.1180344566158739e-05,
      "loss": 0.5937,
      "step": 2129
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.519444154028079,
      "learning_rate": 1.1173114517344656e-05,
      "loss": 1.1256,
      "step": 2130
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.389694728823918,
      "learning_rate": 1.1165883846690744e-05,
      "loss": 1.1168,
      "step": 2131
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6426748019197985,
      "learning_rate": 1.115865255802981e-05,
      "loss": 0.5786,
      "step": 2132
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6696621463637646,
      "learning_rate": 1.1151420655194982e-05,
      "loss": 0.5967,
      "step": 2133
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6496679919411088,
      "learning_rate": 1.114418814201972e-05,
      "loss": 0.5558,
      "step": 2134
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8499046352778552,
      "learning_rate": 1.113695502233781e-05,
      "loss": 0.5789,
      "step": 2135
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.88766195344606,
      "learning_rate": 1.1129721299983349e-05,
      "loss": 0.5871,
      "step": 2136
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.7787604160020254,
      "learning_rate": 1.112248697879076e-05,
      "loss": 1.1189,
      "step": 2137
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.529336011073512,
      "learning_rate": 1.1115252062594785e-05,
      "loss": 0.5849,
      "step": 2138
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5567570403595092,
      "learning_rate": 1.1108016555230476e-05,
      "loss": 0.5194,
      "step": 2139
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.84977348143808,
      "learning_rate": 1.1100780460533207e-05,
      "loss": 0.6184,
      "step": 2140
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9722643887313787,
      "learning_rate": 1.1093543782338653e-05,
      "loss": 0.6221,
      "step": 2141
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8578554283847681,
      "learning_rate": 1.1086306524482807e-05,
      "loss": 0.6471,
      "step": 2142
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5395051251488014,
      "learning_rate": 1.1079068690801962e-05,
      "loss": 0.5731,
      "step": 2143
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6416394247564632,
      "learning_rate": 1.1071830285132722e-05,
      "loss": 0.5671,
      "step": 2144
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3061627064701073,
      "learning_rate": 1.1064591311311994e-05,
      "loss": 1.1894,
      "step": 2145
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5384317685184183,
      "learning_rate": 1.1057351773176981e-05,
      "loss": 0.6109,
      "step": 2146
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6341931071066793,
      "learning_rate": 1.1050111674565187e-05,
      "loss": 0.561,
      "step": 2147
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.6691232829941254,
      "learning_rate": 1.104287101931442e-05,
      "loss": 1.0947,
      "step": 2148
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8111811613464501,
      "learning_rate": 1.1035629811262772e-05,
      "loss": 0.6092,
      "step": 2149
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7699535592553917,
      "learning_rate": 1.1028388054248633e-05,
      "loss": 0.5756,
      "step": 2150
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6544501824114806,
      "learning_rate": 1.102114575211069e-05,
      "loss": 0.6333,
      "step": 2151
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4427700040093752,
      "learning_rate": 1.1013902908687908e-05,
      "loss": 0.5923,
      "step": 2152
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6660043884324642,
      "learning_rate": 1.1006659527819547e-05,
      "loss": 0.6343,
      "step": 2153
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6142090008230308,
      "learning_rate": 1.099941561334515e-05,
      "loss": 0.5962,
      "step": 2154
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3852603707458226,
      "learning_rate": 1.0992171169104541e-05,
      "loss": 1.1628,
      "step": 2155
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5089579999852287,
      "learning_rate": 1.0984926198937826e-05,
      "loss": 0.5584,
      "step": 2156
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7672249931163135,
      "learning_rate": 1.097768070668539e-05,
      "loss": 0.592,
      "step": 2157
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7557243361726644,
      "learning_rate": 1.0970434696187893e-05,
      "loss": 0.6886,
      "step": 2158
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6780796481023044,
      "learning_rate": 1.0963188171286275e-05,
      "loss": 0.5929,
      "step": 2159
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.72460899476437,
      "learning_rate": 1.0955941135821737e-05,
      "loss": 0.5464,
      "step": 2160
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6806170207759856,
      "learning_rate": 1.0948693593635763e-05,
      "loss": 0.5987,
      "step": 2161
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7275078972438407,
      "learning_rate": 1.0941445548570104e-05,
      "loss": 0.5744,
      "step": 2162
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.3348063350262263,
      "learning_rate": 1.0934197004466772e-05,
      "loss": 1.1831,
      "step": 2163
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6072352948221715,
      "learning_rate": 1.0926947965168045e-05,
      "loss": 0.5614,
      "step": 2164
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6076625489161709,
      "learning_rate": 1.0919698434516463e-05,
      "loss": 0.5357,
      "step": 2165
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7235905165755847,
      "learning_rate": 1.0912448416354838e-05,
      "loss": 0.6026,
      "step": 2166
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8335319642853969,
      "learning_rate": 1.0905197914526218e-05,
      "loss": 0.6533,
      "step": 2167
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6395425382361286,
      "learning_rate": 1.0897946932873924e-05,
      "loss": 0.5846,
      "step": 2168
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.468518293355415,
      "learning_rate": 1.0890695475241534e-05,
      "loss": 1.0975,
      "step": 2169
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5813984419373546,
      "learning_rate": 1.0883443545472861e-05,
      "loss": 0.5543,
      "step": 2170
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.2486820737171223,
      "learning_rate": 1.0876191147411983e-05,
      "loss": 1.1347,
      "step": 2171
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.680289786196934,
      "learning_rate": 1.0868938284903224e-05,
      "loss": 0.5574,
      "step": 2172
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.730353260780885,
      "learning_rate": 1.086168496179115e-05,
      "loss": 0.6261,
      "step": 2173
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7649280327051213,
      "learning_rate": 1.0854431181920575e-05,
      "loss": 0.6246,
      "step": 2174
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7039341738832197,
      "learning_rate": 1.0847176949136553e-05,
      "loss": 0.6279,
      "step": 2175
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.9790967451437034,
      "learning_rate": 1.0839922267284377e-05,
      "loss": 1.0674,
      "step": 2176
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6051433813817337,
      "learning_rate": 1.0832667140209586e-05,
      "loss": 0.576,
      "step": 2177
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5243482441011282,
      "learning_rate": 1.0825411571757937e-05,
      "loss": 0.5623,
      "step": 2178
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5901915747274902,
      "learning_rate": 1.0818155565775443e-05,
      "loss": 0.5907,
      "step": 2179
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8621692318672833,
      "learning_rate": 1.0810899126108334e-05,
      "loss": 0.5835,
      "step": 2180
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6213843781626447,
      "learning_rate": 1.0803642256603072e-05,
      "loss": 0.5856,
      "step": 2181
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7364194151838952,
      "learning_rate": 1.0796384961106354e-05,
      "loss": 0.6031,
      "step": 2182
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6036586815218155,
      "learning_rate": 1.0789127243465096e-05,
      "loss": 0.607,
      "step": 2183
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.857214866192726,
      "learning_rate": 1.0781869107526441e-05,
      "loss": 0.6044,
      "step": 2184
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6473987908103063,
      "learning_rate": 1.077461055713775e-05,
      "loss": 0.6184,
      "step": 2185
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.611273847849079,
      "learning_rate": 1.0767351596146607e-05,
      "loss": 1.1846,
      "step": 2186
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.701654718402639,
      "learning_rate": 1.0760092228400818e-05,
      "loss": 0.5874,
      "step": 2187
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7241389702588723,
      "learning_rate": 1.075283245774839e-05,
      "loss": 0.624,
      "step": 2188
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.468286037789468,
      "learning_rate": 1.074557228803756e-05,
      "loss": 0.5349,
      "step": 2189
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5715121392898894,
      "learning_rate": 1.073831172311677e-05,
      "loss": 0.6341,
      "step": 2190
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7048598438960303,
      "learning_rate": 1.0731050766834668e-05,
      "loss": 0.5833,
      "step": 2191
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8350112346434526,
      "learning_rate": 1.072378942304011e-05,
      "loss": 0.5952,
      "step": 2192
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.741083110803767,
      "learning_rate": 1.0716527695582164e-05,
      "loss": 0.5785,
      "step": 2193
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.664451161101936,
      "learning_rate": 1.0709265588310098e-05,
      "loss": 0.6653,
      "step": 2194
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8087721788423259,
      "learning_rate": 1.070200310507338e-05,
      "loss": 0.6228,
      "step": 2195
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.831252811460855,
      "learning_rate": 1.0694740249721673e-05,
      "loss": 0.6497,
      "step": 2196
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5293518444616823,
      "learning_rate": 1.0687477026104852e-05,
      "loss": 0.537,
      "step": 2197
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6901979729701877,
      "learning_rate": 1.0680213438072969e-05,
      "loss": 0.6347,
      "step": 2198
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7008997487119568,
      "learning_rate": 1.0672949489476281e-05,
      "loss": 0.6159,
      "step": 2199
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.722189883441894,
      "learning_rate": 1.0665685184165231e-05,
      "loss": 1.1367,
      "step": 2200
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.499928613252261,
      "learning_rate": 1.0658420525990454e-05,
      "loss": 1.0942,
      "step": 2201
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5741245172291856,
      "learning_rate": 1.065115551880277e-05,
      "loss": 0.5913,
      "step": 2202
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4945947500656498,
      "learning_rate": 1.0643890166453186e-05,
      "loss": 0.5301,
      "step": 2203
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.677364341436417,
      "learning_rate": 1.063662447279289e-05,
      "loss": 0.5934,
      "step": 2204
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7474403915527796,
      "learning_rate": 1.062935844167325e-05,
      "loss": 0.6123,
      "step": 2205
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.6855187556623847,
      "learning_rate": 1.0622092076945817e-05,
      "loss": 1.1437,
      "step": 2206
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.3800506773300083,
      "learning_rate": 1.0614825382462317e-05,
      "loss": 1.0465,
      "step": 2207
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6911756441404984,
      "learning_rate": 1.060755836207465e-05,
      "loss": 0.6931,
      "step": 2208
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.398021365671226,
      "learning_rate": 1.0600291019634887e-05,
      "loss": 1.1045,
      "step": 2209
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2879803108922077,
      "learning_rate": 1.0593023358995273e-05,
      "loss": 1.0547,
      "step": 2210
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7962026058038516,
      "learning_rate": 1.058575538400822e-05,
      "loss": 0.5468,
      "step": 2211
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5902076261300186,
      "learning_rate": 1.0578487098526305e-05,
      "loss": 0.6144,
      "step": 2212
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5732160272779852,
      "learning_rate": 1.0571218506402275e-05,
      "loss": 0.585,
      "step": 2213
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5583104512128128,
      "learning_rate": 1.0563949611489032e-05,
      "loss": 0.5484,
      "step": 2214
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.696990002905509,
      "learning_rate": 1.055668041763965e-05,
      "loss": 0.6443,
      "step": 2215
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6378013108264375,
      "learning_rate": 1.0549410928707344e-05,
      "loss": 0.5353,
      "step": 2216
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.705935044857886,
      "learning_rate": 1.0542141148545504e-05,
      "loss": 0.6756,
      "step": 2217
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.5771064064847486,
      "learning_rate": 1.0534871081007663e-05,
      "loss": 1.1899,
      "step": 2218
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.527898575866469,
      "learning_rate": 1.0527600729947509e-05,
      "loss": 0.5939,
      "step": 2219
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.825676651855125,
      "learning_rate": 1.0520330099218878e-05,
      "loss": 0.5967,
      "step": 2220
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.651855143878379,
      "learning_rate": 1.0513059192675762e-05,
      "loss": 0.5527,
      "step": 2221
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7131399198312043,
      "learning_rate": 1.0505788014172288e-05,
      "loss": 0.6119,
      "step": 2222
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.3793877470920566,
      "learning_rate": 1.0498516567562737e-05,
      "loss": 1.1012,
      "step": 2223
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6585105153601392,
      "learning_rate": 1.0491244856701528e-05,
      "loss": 0.5829,
      "step": 2224
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5629570893537912,
      "learning_rate": 1.0483972885443218e-05,
      "loss": 0.5547,
      "step": 2225
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.515994162801748,
      "learning_rate": 1.0476700657642507e-05,
      "loss": 1.101,
      "step": 2226
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.565237175576301,
      "learning_rate": 1.0469428177154225e-05,
      "loss": 1.1964,
      "step": 2227
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5112092654903289,
      "learning_rate": 1.0462155447833344e-05,
      "loss": 0.5723,
      "step": 2228
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.593836367251011,
      "learning_rate": 1.0454882473534961e-05,
      "loss": 0.553,
      "step": 2229
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.457160397500159,
      "learning_rate": 1.0447609258114301e-05,
      "loss": 1.1613,
      "step": 2230
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5405126339138406,
      "learning_rate": 1.044033580542673e-05,
      "loss": 0.6013,
      "step": 2231
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.492459265743082,
      "learning_rate": 1.043306211932772e-05,
      "loss": 1.1971,
      "step": 2232
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.616789079531855,
      "learning_rate": 1.0425788203672882e-05,
      "loss": 0.5754,
      "step": 2233
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2761627318451203,
      "learning_rate": 1.0418514062317944e-05,
      "loss": 1.1027,
      "step": 2234
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.451113147904792,
      "learning_rate": 1.0411239699118756e-05,
      "loss": 0.6323,
      "step": 2235
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.423138829167969,
      "learning_rate": 1.0403965117931279e-05,
      "loss": 1.1246,
      "step": 2236
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5680476404996957,
      "learning_rate": 1.0396690322611596e-05,
      "loss": 0.5795,
      "step": 2237
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5522725907026524,
      "learning_rate": 1.03894153170159e-05,
      "loss": 0.5929,
      "step": 2238
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6282880774458866,
      "learning_rate": 1.03821401050005e-05,
      "loss": 0.6077,
      "step": 2239
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1661979113690033,
      "learning_rate": 1.0374864690421806e-05,
      "loss": 1.1001,
      "step": 2240
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8009639062224263,
      "learning_rate": 1.0367589077136348e-05,
      "loss": 0.5736,
      "step": 2241
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.646861909134479,
      "learning_rate": 1.0360313269000746e-05,
      "loss": 1.1391,
      "step": 2242
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7389235347052745,
      "learning_rate": 1.0353037269871733e-05,
      "loss": 0.589,
      "step": 2243
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2771709828151017,
      "learning_rate": 1.0345761083606145e-05,
      "loss": 1.0938,
      "step": 2244
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.456330782948324,
      "learning_rate": 1.033848471406091e-05,
      "loss": 1.1447,
      "step": 2245
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8676449531956318,
      "learning_rate": 1.0331208165093056e-05,
      "loss": 0.5798,
      "step": 2246
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8822528498618498,
      "learning_rate": 1.0323931440559711e-05,
      "loss": 0.6698,
      "step": 2247
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5661389503277434,
      "learning_rate": 1.031665454431809e-05,
      "loss": 0.5553,
      "step": 2248
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7042383690239897,
      "learning_rate": 1.03093774802255e-05,
      "loss": 0.5663,
      "step": 2249
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.708956023277425,
      "learning_rate": 1.0302100252139343e-05,
      "loss": 0.5802,
      "step": 2250
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7554453902359555,
      "learning_rate": 1.0294822863917094e-05,
      "loss": 0.5933,
      "step": 2251
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.579904067248094,
      "learning_rate": 1.0287545319416328e-05,
      "loss": 0.5539,
      "step": 2252
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5650424057609376,
      "learning_rate": 1.0280267622494691e-05,
      "loss": 0.5702,
      "step": 2253
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.604576544652426,
      "learning_rate": 1.027298977700992e-05,
      "loss": 0.5876,
      "step": 2254
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6902919233169764,
      "learning_rate": 1.0265711786819827e-05,
      "loss": 0.5431,
      "step": 2255
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6784497155419613,
      "learning_rate": 1.0258433655782296e-05,
      "loss": 0.5593,
      "step": 2256
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6299164228955627,
      "learning_rate": 1.0251155387755291e-05,
      "loss": 0.6126,
      "step": 2257
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6574183289292117,
      "learning_rate": 1.024387698659685e-05,
      "loss": 0.5831,
      "step": 2258
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7278411592956304,
      "learning_rate": 1.0236598456165075e-05,
      "loss": 0.6305,
      "step": 2259
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7522935164318385,
      "learning_rate": 1.022931980031814e-05,
      "loss": 0.6529,
      "step": 2260
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.1585538549340058,
      "learning_rate": 1.0222041022914288e-05,
      "loss": 1.1234,
      "step": 2261
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6554340802933232,
      "learning_rate": 1.0214762127811824e-05,
      "loss": 0.5251,
      "step": 2262
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5219963338010953,
      "learning_rate": 1.0207483118869113e-05,
      "loss": 0.5394,
      "step": 2263
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6343042791442763,
      "learning_rate": 1.0200203999944582e-05,
      "loss": 0.6289,
      "step": 2264
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6133742602280776,
      "learning_rate": 1.0192924774896722e-05,
      "loss": 1.16,
      "step": 2265
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6778497988490102,
      "learning_rate": 1.0185645447584072e-05,
      "loss": 0.5887,
      "step": 2266
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.809758882151548,
      "learning_rate": 1.0178366021865226e-05,
      "loss": 0.6308,
      "step": 2267
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.4273838940676997,
      "learning_rate": 1.0171086501598839e-05,
      "loss": 1.1248,
      "step": 2268
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.3809740279681715,
      "learning_rate": 1.0163806890643606e-05,
      "loss": 1.0927,
      "step": 2269
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5070656455425724,
      "learning_rate": 1.015652719285828e-05,
      "loss": 0.5154,
      "step": 2270
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6135622401929088,
      "learning_rate": 1.0149247412101642e-05,
      "loss": 0.5557,
      "step": 2271
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6950842169379952,
      "learning_rate": 1.0141967552232539e-05,
      "loss": 0.5274,
      "step": 2272
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.673778893165447,
      "learning_rate": 1.0134687617109844e-05,
      "loss": 0.6177,
      "step": 2273
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.3405929044938896,
      "learning_rate": 1.0127407610592478e-05,
      "loss": 1.1264,
      "step": 2274
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8232235158107835,
      "learning_rate": 1.0120127536539399e-05,
      "loss": 0.6228,
      "step": 2275
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.286821557102168,
      "learning_rate": 1.0112847398809598e-05,
      "loss": 1.1042,
      "step": 2276
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5564143759541138,
      "learning_rate": 1.0105567201262101e-05,
      "loss": 0.5692,
      "step": 2277
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.630118279347992,
      "learning_rate": 1.0098286947755964e-05,
      "loss": 0.5862,
      "step": 2278
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.240717690638243,
      "learning_rate": 1.009100664215028e-05,
      "loss": 1.0839,
      "step": 2279
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7238945747987728,
      "learning_rate": 1.0083726288304158e-05,
      "loss": 0.6088,
      "step": 2280
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6049871919821652,
      "learning_rate": 1.0076445890076746e-05,
      "loss": 0.5151,
      "step": 2281
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.5166583589758065,
      "learning_rate": 1.00691654513272e-05,
      "loss": 1.1408,
      "step": 2282
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5677908326917658,
      "learning_rate": 1.0061884975914713e-05,
      "loss": 0.6091,
      "step": 2283
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.540546786535842,
      "learning_rate": 1.0054604467698486e-05,
      "loss": 0.5634,
      "step": 2284
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.2724238056570276,
      "learning_rate": 1.004732393053774e-05,
      "loss": 1.1135,
      "step": 2285
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8132412812673002,
      "learning_rate": 1.0040043368291714e-05,
      "loss": 0.5856,
      "step": 2286
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6611120261690309,
      "learning_rate": 1.0032762784819662e-05,
      "loss": 0.5655,
      "step": 2287
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6603728695448896,
      "learning_rate": 1.0025482183980847e-05,
      "loss": 0.5935,
      "step": 2288
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.674614029248983,
      "learning_rate": 1.0018201569634534e-05,
      "loss": 0.6593,
      "step": 2289
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.572270899850164,
      "learning_rate": 1.0010920945640007e-05,
      "loss": 0.6074,
      "step": 2290
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8231571490309253,
      "learning_rate": 1.0003640315856553e-05,
      "loss": 0.6101,
      "step": 2291
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6239125330529682,
      "learning_rate": 9.99635968414345e-06,
      "loss": 0.5789,
      "step": 2292
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.571694839591783,
      "learning_rate": 9.989079054359994e-06,
      "loss": 0.6217,
      "step": 2293
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6459128846727045,
      "learning_rate": 9.98179843036547e-06,
      "loss": 0.5935,
      "step": 2294
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6746625035430394,
      "learning_rate": 9.974517816019155e-06,
      "loss": 0.6008,
      "step": 2295
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7016330870696839,
      "learning_rate": 9.967237215180341e-06,
      "loss": 0.568,
      "step": 2296
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5473816226283847,
      "learning_rate": 9.959956631708291e-06,
      "loss": 0.5895,
      "step": 2297
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7094672644399511,
      "learning_rate": 9.952676069462263e-06,
      "loss": 0.5192,
      "step": 2298
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7312656685605736,
      "learning_rate": 9.94539553230152e-06,
      "loss": 0.6122,
      "step": 2299
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.3740599590309897,
      "learning_rate": 9.938115024085289e-06,
      "loss": 1.134,
      "step": 2300
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.740729380602863,
      "learning_rate": 9.930834548672801e-06,
      "loss": 0.5703,
      "step": 2301
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6554578238663613,
      "learning_rate": 9.923554109923257e-06,
      "loss": 0.5904,
      "step": 2302
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7790378331954,
      "learning_rate": 9.916273711695844e-06,
      "loss": 0.5956,
      "step": 2303
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.539207596950714,
      "learning_rate": 9.908993357849721e-06,
      "loss": 1.127,
      "step": 2304
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5560917640821488,
      "learning_rate": 9.901713052244038e-06,
      "loss": 0.6295,
      "step": 2305
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7026079972343968,
      "learning_rate": 9.8944327987379e-06,
      "loss": 0.5563,
      "step": 2306
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.522065272979815,
      "learning_rate": 9.887152601190406e-06,
      "loss": 0.5413,
      "step": 2307
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.519839539861635,
      "learning_rate": 9.879872463460606e-06,
      "loss": 1.1448,
      "step": 2308
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.670142117877022,
      "learning_rate": 9.872592389407525e-06,
      "loss": 0.6666,
      "step": 2309
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5590189108526706,
      "learning_rate": 9.865312382890161e-06,
      "loss": 0.545,
      "step": 2310
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6284687850568427,
      "learning_rate": 9.858032447767464e-06,
      "loss": 0.606,
      "step": 2311
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6321276127809712,
      "learning_rate": 9.850752587898363e-06,
      "loss": 0.6077,
      "step": 2312
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5820978312343967,
      "learning_rate": 9.843472807141726e-06,
      "loss": 0.6974,
      "step": 2313
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7066015786046553,
      "learning_rate": 9.836193109356397e-06,
      "loss": 0.6113,
      "step": 2314
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.35907837625995,
      "learning_rate": 9.828913498401161e-06,
      "loss": 1.0974,
      "step": 2315
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.572635480166356,
      "learning_rate": 9.821633978134775e-06,
      "loss": 1.1224,
      "step": 2316
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7322961925325724,
      "learning_rate": 9.81435455241593e-06,
      "loss": 0.5935,
      "step": 2317
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7576587855120687,
      "learning_rate": 9.807075225103281e-06,
      "loss": 0.5935,
      "step": 2318
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5763339091578124,
      "learning_rate": 9.799796000055423e-06,
      "loss": 0.5608,
      "step": 2319
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.537122207535947,
      "learning_rate": 9.79251688113089e-06,
      "loss": 1.1644,
      "step": 2320
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6736574317150614,
      "learning_rate": 9.785237872188181e-06,
      "loss": 0.5105,
      "step": 2321
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6764960082713256,
      "learning_rate": 9.777958977085715e-06,
      "loss": 0.5538,
      "step": 2322
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6284995182678839,
      "learning_rate": 9.770680199681863e-06,
      "loss": 0.6614,
      "step": 2323
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6231116811381976,
      "learning_rate": 9.763401543834927e-06,
      "loss": 0.5502,
      "step": 2324
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6353465896331432,
      "learning_rate": 9.756123013403154e-06,
      "loss": 0.596,
      "step": 2325
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.492973082829415,
      "learning_rate": 9.748844612244709e-06,
      "loss": 1.1107,
      "step": 2326
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.359207230746347,
      "learning_rate": 9.741566344217707e-06,
      "loss": 1.0611,
      "step": 2327
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7570563631713867,
      "learning_rate": 9.734288213180175e-06,
      "loss": 0.493,
      "step": 2328
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5399966512248888,
      "learning_rate": 9.727010222990083e-06,
      "loss": 0.5827,
      "step": 2329
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.733295234422507,
      "learning_rate": 9.719732377505314e-06,
      "loss": 0.558,
      "step": 2330
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.932763352977027,
      "learning_rate": 9.712454680583677e-06,
      "loss": 0.6452,
      "step": 2331
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.820358892153058,
      "learning_rate": 9.705177136082911e-06,
      "loss": 1.1674,
      "step": 2332
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.4561050307394656,
      "learning_rate": 9.697899747860663e-06,
      "loss": 1.0931,
      "step": 2333
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7332786078167153,
      "learning_rate": 9.690622519774502e-06,
      "loss": 0.6211,
      "step": 2334
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5780675693307167,
      "learning_rate": 9.68334545568191e-06,
      "loss": 0.5736,
      "step": 2335
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8189851726354356,
      "learning_rate": 9.67606855944029e-06,
      "loss": 0.5038,
      "step": 2336
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7089392399917782,
      "learning_rate": 9.668791834906944e-06,
      "loss": 0.6156,
      "step": 2337
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.84222608205698,
      "learning_rate": 9.661515285939094e-06,
      "loss": 1.0624,
      "step": 2338
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.534775851403156,
      "learning_rate": 9.654238916393862e-06,
      "loss": 0.5549,
      "step": 2339
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4618504486325168,
      "learning_rate": 9.646962730128269e-06,
      "loss": 0.5247,
      "step": 2340
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.4344518189626565,
      "learning_rate": 9.63968673099926e-06,
      "loss": 1.1319,
      "step": 2341
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6312497483733988,
      "learning_rate": 9.632410922863655e-06,
      "loss": 0.5793,
      "step": 2342
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4683945995589451,
      "learning_rate": 9.625135309578195e-06,
      "loss": 0.5541,
      "step": 2343
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8625813159689972,
      "learning_rate": 9.617859894999503e-06,
      "loss": 0.6126,
      "step": 2344
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5601291310313916,
      "learning_rate": 9.610584682984102e-06,
      "loss": 0.5763,
      "step": 2345
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5372387071807505,
      "learning_rate": 9.603309677388406e-06,
      "loss": 0.6454,
      "step": 2346
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6144520387164831,
      "learning_rate": 9.596034882068724e-06,
      "loss": 0.5607,
      "step": 2347
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.324689654352755,
      "learning_rate": 9.588760300881245e-06,
      "loss": 1.0506,
      "step": 2348
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.275813119050215,
      "learning_rate": 9.581485937682058e-06,
      "loss": 1.1503,
      "step": 2349
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5082976803815766,
      "learning_rate": 9.574211796327123e-06,
      "loss": 0.5559,
      "step": 2350
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.547578039895173,
      "learning_rate": 9.566937880672284e-06,
      "loss": 1.1343,
      "step": 2351
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.694479199453764,
      "learning_rate": 9.559664194573276e-06,
      "loss": 0.5801,
      "step": 2352
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5574890171651847,
      "learning_rate": 9.5523907418857e-06,
      "loss": 0.5495,
      "step": 2353
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5920247548053372,
      "learning_rate": 9.545117526465042e-06,
      "loss": 0.617,
      "step": 2354
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6453117391133696,
      "learning_rate": 9.537844552166657e-06,
      "loss": 0.5344,
      "step": 2355
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.3894465214744134,
      "learning_rate": 9.530571822845776e-06,
      "loss": 1.0692,
      "step": 2356
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.3262612792342785,
      "learning_rate": 9.523299342357493e-06,
      "loss": 1.12,
      "step": 2357
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8298600337589457,
      "learning_rate": 9.516027114556783e-06,
      "loss": 0.5871,
      "step": 2358
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8176236791767904,
      "learning_rate": 9.508755143298472e-06,
      "loss": 0.6017,
      "step": 2359
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.644094807324196,
      "learning_rate": 9.501483432437264e-06,
      "loss": 0.6022,
      "step": 2360
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7099954480732047,
      "learning_rate": 9.494211985827717e-06,
      "loss": 0.632,
      "step": 2361
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5810571592530844,
      "learning_rate": 9.48694080732424e-06,
      "loss": 0.5381,
      "step": 2362
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.3315575076940225,
      "learning_rate": 9.479669900781125e-06,
      "loss": 1.0966,
      "step": 2363
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.263879982547681,
      "learning_rate": 9.472399270052494e-06,
      "loss": 1.0564,
      "step": 2364
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5749375164756927,
      "learning_rate": 9.465128918992342e-06,
      "loss": 0.562,
      "step": 2365
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.6548407381595367,
      "learning_rate": 9.457858851454496e-06,
      "loss": 1.1329,
      "step": 2366
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7089926274959515,
      "learning_rate": 9.450589071292658e-06,
      "loss": 0.497,
      "step": 2367
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.430888020875455,
      "learning_rate": 9.443319582360352e-06,
      "loss": 1.1175,
      "step": 2368
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.66660678895986,
      "learning_rate": 9.43605038851097e-06,
      "loss": 0.6345,
      "step": 2369
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6250571764297723,
      "learning_rate": 9.428781493597727e-06,
      "loss": 0.5852,
      "step": 2370
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7336967300590154,
      "learning_rate": 9.421512901473697e-06,
      "loss": 0.7007,
      "step": 2371
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.539840234337824,
      "learning_rate": 9.414244615991786e-06,
      "loss": 1.147,
      "step": 2372
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6961099707958414,
      "learning_rate": 9.40697664100473e-06,
      "loss": 0.5639,
      "step": 2373
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6009239635700714,
      "learning_rate": 9.399708980365116e-06,
      "loss": 0.5903,
      "step": 2374
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5177608004744338,
      "learning_rate": 9.392441637925353e-06,
      "loss": 0.5488,
      "step": 2375
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.7810519217421747,
      "learning_rate": 9.385174617537686e-06,
      "loss": 1.1702,
      "step": 2376
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.5138363814808047,
      "learning_rate": 9.377907923054184e-06,
      "loss": 1.0792,
      "step": 2377
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.6643515615487865,
      "learning_rate": 9.370641558326753e-06,
      "loss": 1.0529,
      "step": 2378
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6785769955375274,
      "learning_rate": 9.363375527207112e-06,
      "loss": 0.5933,
      "step": 2379
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.3658097369814017,
      "learning_rate": 9.356109833546817e-06,
      "loss": 1.1309,
      "step": 2380
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.690685320467535,
      "learning_rate": 9.348844481197235e-06,
      "loss": 0.5627,
      "step": 2381
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7893610856927673,
      "learning_rate": 9.341579474009549e-06,
      "loss": 0.5174,
      "step": 2382
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.6137169166935426,
      "learning_rate": 9.334314815834774e-06,
      "loss": 1.1609,
      "step": 2383
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5362013292403032,
      "learning_rate": 9.327050510523722e-06,
      "loss": 0.5932,
      "step": 2384
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.7426464958908445,
      "learning_rate": 9.319786561927033e-06,
      "loss": 1.0747,
      "step": 2385
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7854353814598527,
      "learning_rate": 9.31252297389515e-06,
      "loss": 0.6256,
      "step": 2386
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8073533230224452,
      "learning_rate": 9.305259750278329e-06,
      "loss": 0.5116,
      "step": 2387
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7071596304715828,
      "learning_rate": 9.297996894926622e-06,
      "loss": 0.5588,
      "step": 2388
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.722467991851377,
      "learning_rate": 9.290734411689905e-06,
      "loss": 0.6371,
      "step": 2389
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.3608262920302137,
      "learning_rate": 9.283472304417836e-06,
      "loss": 1.1205,
      "step": 2390
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.640022276804465,
      "learning_rate": 9.276210576959892e-06,
      "loss": 0.6041,
      "step": 2391
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.2964443206697602,
      "learning_rate": 9.268949233165339e-06,
      "loss": 1.1086,
      "step": 2392
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.713239191219958,
      "learning_rate": 9.261688276883233e-06,
      "loss": 0.541,
      "step": 2393
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.585774280284178,
      "learning_rate": 9.254427711962442e-06,
      "loss": 1.1383,
      "step": 2394
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.7219418766622634,
      "learning_rate": 9.247167542251612e-06,
      "loss": 1.093,
      "step": 2395
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.752585507713525,
      "learning_rate": 9.239907771599188e-06,
      "loss": 0.5841,
      "step": 2396
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9114905711363692,
      "learning_rate": 9.232648403853393e-06,
      "loss": 0.6012,
      "step": 2397
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5529402961672094,
      "learning_rate": 9.225389442862252e-06,
      "loss": 0.582,
      "step": 2398
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8296857743163413,
      "learning_rate": 9.21813089247356e-06,
      "loss": 0.5718,
      "step": 2399
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6020967013716612,
      "learning_rate": 9.210872756534907e-06,
      "loss": 0.6809,
      "step": 2400
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6559540127625743,
      "learning_rate": 9.203615038893646e-06,
      "loss": 0.5665,
      "step": 2401
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.615704528782675,
      "learning_rate": 9.19635774339693e-06,
      "loss": 0.5539,
      "step": 2402
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7722183262838664,
      "learning_rate": 9.18910087389167e-06,
      "loss": 0.5612,
      "step": 2403
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.492961874179623,
      "learning_rate": 9.181844434224559e-06,
      "loss": 0.6448,
      "step": 2404
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5959706851902038,
      "learning_rate": 9.174588428242064e-06,
      "loss": 0.5346,
      "step": 2405
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8142068245640446,
      "learning_rate": 9.167332859790419e-06,
      "loss": 0.6415,
      "step": 2406
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6265729780028608,
      "learning_rate": 9.160077732715625e-06,
      "loss": 0.5583,
      "step": 2407
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5965748054193931,
      "learning_rate": 9.152823050863447e-06,
      "loss": 0.5454,
      "step": 2408
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6451712445943696,
      "learning_rate": 9.145568818079428e-06,
      "loss": 0.5638,
      "step": 2409
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6763465167770881,
      "learning_rate": 9.138315038208851e-06,
      "loss": 0.6227,
      "step": 2410
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6541121031839396,
      "learning_rate": 9.131061715096778e-06,
      "loss": 0.5887,
      "step": 2411
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.712563381573883,
      "learning_rate": 9.123808852588017e-06,
      "loss": 1.0495,
      "step": 2412
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7460295105372583,
      "learning_rate": 9.116556454527142e-06,
      "loss": 0.5994,
      "step": 2413
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5279021264989352,
      "learning_rate": 9.109304524758471e-06,
      "loss": 0.5413,
      "step": 2414
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.28508687491266,
      "learning_rate": 9.102053067126079e-06,
      "loss": 1.0817,
      "step": 2415
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.487699400977762,
      "learning_rate": 9.094802085473787e-06,
      "loss": 1.1244,
      "step": 2416
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.769653634594282,
      "learning_rate": 9.087551583645167e-06,
      "loss": 0.6107,
      "step": 2417
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4322964037948047,
      "learning_rate": 9.080301565483539e-06,
      "loss": 0.5535,
      "step": 2418
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6801746845811032,
      "learning_rate": 9.073052034831956e-06,
      "loss": 0.5644,
      "step": 2419
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0599155925320045,
      "learning_rate": 9.065802995533231e-06,
      "loss": 0.5618,
      "step": 2420
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6806078219543041,
      "learning_rate": 9.058554451429896e-06,
      "loss": 0.6093,
      "step": 2421
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6956231666079165,
      "learning_rate": 9.051306406364238e-06,
      "loss": 0.6148,
      "step": 2422
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.434454692902084,
      "learning_rate": 9.044058864178268e-06,
      "loss": 1.1368,
      "step": 2423
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.470320305129429,
      "learning_rate": 9.03681182871373e-06,
      "loss": 0.5981,
      "step": 2424
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5517860279527735,
      "learning_rate": 9.02956530381211e-06,
      "loss": 0.5037,
      "step": 2425
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6870928945953403,
      "learning_rate": 9.022319293314613e-06,
      "loss": 0.5982,
      "step": 2426
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6614140555868155,
      "learning_rate": 9.015073801062178e-06,
      "loss": 0.6108,
      "step": 2427
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7361769351935865,
      "learning_rate": 9.00782883089546e-06,
      "loss": 0.5466,
      "step": 2428
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7348432676665182,
      "learning_rate": 9.000584386654853e-06,
      "loss": 0.554,
      "step": 2429
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7689678546964587,
      "learning_rate": 8.993340472180453e-06,
      "loss": 0.5754,
      "step": 2430
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7183748580330331,
      "learning_rate": 8.986097091312093e-06,
      "loss": 0.6317,
      "step": 2431
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6063978339029708,
      "learning_rate": 8.978854247889312e-06,
      "loss": 0.668,
      "step": 2432
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7403585461714155,
      "learning_rate": 8.971611945751368e-06,
      "loss": 0.6046,
      "step": 2433
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.443942082593276,
      "learning_rate": 8.964370188737233e-06,
      "loss": 1.0789,
      "step": 2434
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3802093546796397,
      "learning_rate": 8.957128980685582e-06,
      "loss": 1.1055,
      "step": 2435
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.215583933403336,
      "learning_rate": 8.949888325434816e-06,
      "loss": 1.0863,
      "step": 2436
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.892702936833843,
      "learning_rate": 8.942648226823024e-06,
      "loss": 0.5682,
      "step": 2437
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5537310963973008,
      "learning_rate": 8.935408688688011e-06,
      "loss": 0.5933,
      "step": 2438
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7720234055960256,
      "learning_rate": 8.928169714867278e-06,
      "loss": 0.6073,
      "step": 2439
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7102471791499771,
      "learning_rate": 8.920931309198042e-06,
      "loss": 0.5954,
      "step": 2440
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3329390589517387,
      "learning_rate": 8.913693475517195e-06,
      "loss": 1.1476,
      "step": 2441
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6807958040414765,
      "learning_rate": 8.90645621766135e-06,
      "loss": 0.6212,
      "step": 2442
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6391559896195391,
      "learning_rate": 8.899219539466793e-06,
      "loss": 0.613,
      "step": 2443
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7223703706083322,
      "learning_rate": 8.891983444769525e-06,
      "loss": 0.5569,
      "step": 2444
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6761691784308168,
      "learning_rate": 8.88474793740522e-06,
      "loss": 0.5806,
      "step": 2445
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6516456276156126,
      "learning_rate": 8.877513021209244e-06,
      "loss": 0.6346,
      "step": 2446
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7449915859088982,
      "learning_rate": 8.870278700016654e-06,
      "loss": 0.5847,
      "step": 2447
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.542002077745756,
      "learning_rate": 8.863044977662192e-06,
      "loss": 0.4814,
      "step": 2448
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6435146612320324,
      "learning_rate": 8.855811857980282e-06,
      "loss": 0.5997,
      "step": 2449
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7584201239150672,
      "learning_rate": 8.848579344805018e-06,
      "loss": 0.5976,
      "step": 2450
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.4690537837492963,
      "learning_rate": 8.841347441970194e-06,
      "loss": 1.1698,
      "step": 2451
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6222909545505744,
      "learning_rate": 8.834116153309256e-06,
      "loss": 0.5721,
      "step": 2452
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.5278679623518276,
      "learning_rate": 8.826885482655347e-06,
      "loss": 1.1341,
      "step": 2453
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.530415833218068,
      "learning_rate": 8.819655433841263e-06,
      "loss": 0.5965,
      "step": 2454
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.6692642512818163,
      "learning_rate": 8.812426010699488e-06,
      "loss": 1.1013,
      "step": 2455
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6735417691122547,
      "learning_rate": 8.80519721706216e-06,
      "loss": 0.6243,
      "step": 2456
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.683421787644311,
      "learning_rate": 8.797969056761089e-06,
      "loss": 0.6254,
      "step": 2457
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6217164634952737,
      "learning_rate": 8.79074153362775e-06,
      "loss": 0.5651,
      "step": 2458
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7006631637343947,
      "learning_rate": 8.783514651493277e-06,
      "loss": 0.5894,
      "step": 2459
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3748861366047653,
      "learning_rate": 8.776288414188472e-06,
      "loss": 1.0175,
      "step": 2460
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.761956291898036,
      "learning_rate": 8.769062825543784e-06,
      "loss": 0.5858,
      "step": 2461
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3450657941684616,
      "learning_rate": 8.76183788938933e-06,
      "loss": 1.0493,
      "step": 2462
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.4380907942306234,
      "learning_rate": 8.754613609554868e-06,
      "loss": 1.065,
      "step": 2463
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.811622871609793,
      "learning_rate": 8.747389989869823e-06,
      "loss": 0.5522,
      "step": 2464
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3123695399619293,
      "learning_rate": 8.740167034163259e-06,
      "loss": 1.0829,
      "step": 2465
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.253051964425506,
      "learning_rate": 8.732944746263887e-06,
      "loss": 1.0655,
      "step": 2466
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6386425731516432,
      "learning_rate": 8.725723130000074e-06,
      "loss": 0.619,
      "step": 2467
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5701995245788296,
      "learning_rate": 8.718502189199825e-06,
      "loss": 0.571,
      "step": 2468
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7959041618776626,
      "learning_rate": 8.711281927690784e-06,
      "loss": 0.6083,
      "step": 2469
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.379033929521194,
      "learning_rate": 8.70406234930024e-06,
      "loss": 1.0765,
      "step": 2470
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.446224055200841,
      "learning_rate": 8.696843457855119e-06,
      "loss": 1.0966,
      "step": 2471
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.653126980690237,
      "learning_rate": 8.689625257181977e-06,
      "loss": 0.5881,
      "step": 2472
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6941506557179937,
      "learning_rate": 8.682407751107017e-06,
      "loss": 0.6111,
      "step": 2473
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2730424867278254,
      "learning_rate": 8.675190943456058e-06,
      "loss": 1.1242,
      "step": 2474
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6921195231747725,
      "learning_rate": 8.667974838054565e-06,
      "loss": 0.5999,
      "step": 2475
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6286478143241319,
      "learning_rate": 8.660759438727616e-06,
      "loss": 0.6477,
      "step": 2476
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.3372334178070147,
      "learning_rate": 8.653544749299919e-06,
      "loss": 1.1588,
      "step": 2477
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4292246602960517,
      "learning_rate": 8.646330773595812e-06,
      "loss": 1.1005,
      "step": 2478
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4284400664830839,
      "learning_rate": 8.639117515439249e-06,
      "loss": 0.4744,
      "step": 2479
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5683164950897246,
      "learning_rate": 8.631904978653804e-06,
      "loss": 0.6067,
      "step": 2480
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6984163711814741,
      "learning_rate": 8.624693167062668e-06,
      "loss": 0.6294,
      "step": 2481
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7113318232976245,
      "learning_rate": 8.617482084488656e-06,
      "loss": 0.5944,
      "step": 2482
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4869294968526048,
      "learning_rate": 8.610271734754178e-06,
      "loss": 0.5975,
      "step": 2483
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.68134008163073,
      "learning_rate": 8.60306212168128e-06,
      "loss": 0.5481,
      "step": 2484
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5382060674193454,
      "learning_rate": 8.595853249091595e-06,
      "loss": 1.041,
      "step": 2485
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4524794859498709,
      "learning_rate": 8.588645120806381e-06,
      "loss": 0.5476,
      "step": 2486
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6819842867874717,
      "learning_rate": 8.581437740646488e-06,
      "loss": 0.5956,
      "step": 2487
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5339052760395135,
      "learning_rate": 8.574231112432377e-06,
      "loss": 0.6295,
      "step": 2488
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.75515989969332,
      "learning_rate": 8.56702523998411e-06,
      "loss": 0.5918,
      "step": 2489
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4159050939261966,
      "learning_rate": 8.559820127121341e-06,
      "loss": 1.1144,
      "step": 2490
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6616824517399729,
      "learning_rate": 8.552615777663335e-06,
      "loss": 0.5955,
      "step": 2491
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4898497169446943,
      "learning_rate": 8.545412195428936e-06,
      "loss": 0.5803,
      "step": 2492
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4871298584893076,
      "learning_rate": 8.538209384236602e-06,
      "loss": 0.5631,
      "step": 2493
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5635817784101358,
      "learning_rate": 8.531007347904359e-06,
      "loss": 0.5283,
      "step": 2494
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.595890504515322,
      "learning_rate": 8.52380609024984e-06,
      "loss": 0.5599,
      "step": 2495
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4111853136662926,
      "learning_rate": 8.516605615090255e-06,
      "loss": 1.1533,
      "step": 2496
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4915951733611967,
      "learning_rate": 8.509405926242404e-06,
      "loss": 0.548,
      "step": 2497
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.252652367659052,
      "learning_rate": 8.502207027522673e-06,
      "loss": 1.0554,
      "step": 2498
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4819855024440756,
      "learning_rate": 8.49500892274702e-06,
      "loss": 1.0712,
      "step": 2499
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.408923869939118,
      "learning_rate": 8.48781161573099e-06,
      "loss": 0.5454,
      "step": 2500
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.817786807483263,
      "learning_rate": 8.480615110289698e-06,
      "loss": 0.5749,
      "step": 2501
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6052811479709825,
      "learning_rate": 8.473419410237846e-06,
      "loss": 0.5417,
      "step": 2502
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.734476317294055,
      "learning_rate": 8.466224519389694e-06,
      "loss": 0.5476,
      "step": 2503
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6770950971336605,
      "learning_rate": 8.45903044155909e-06,
      "loss": 0.6087,
      "step": 2504
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3369838625153307,
      "learning_rate": 8.451837180559434e-06,
      "loss": 0.5236,
      "step": 2505
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5496487589614036,
      "learning_rate": 8.444644740203706e-06,
      "loss": 0.5609,
      "step": 2506
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.3576862549025175,
      "learning_rate": 8.437453124304446e-06,
      "loss": 1.0772,
      "step": 2507
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4886981150525425,
      "learning_rate": 8.43026233667375e-06,
      "loss": 1.123,
      "step": 2508
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7258975201699407,
      "learning_rate": 8.42307238112329e-06,
      "loss": 0.6157,
      "step": 2509
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2289221743046497,
      "learning_rate": 8.415883261464282e-06,
      "loss": 1.1704,
      "step": 2510
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5558234244001827,
      "learning_rate": 8.408694981507513e-06,
      "loss": 0.5605,
      "step": 2511
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.305200644473646,
      "learning_rate": 8.40150754506331e-06,
      "loss": 1.1323,
      "step": 2512
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.479271086775815,
      "learning_rate": 8.394320955941565e-06,
      "loss": 1.1312,
      "step": 2513
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.5789924861349522,
      "learning_rate": 8.387135217951712e-06,
      "loss": 1.2104,
      "step": 2514
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7581815802064944,
      "learning_rate": 8.379950334902747e-06,
      "loss": 0.6026,
      "step": 2515
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.624478051558774,
      "learning_rate": 8.372766310603191e-06,
      "loss": 0.5365,
      "step": 2516
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7153735527723426,
      "learning_rate": 8.365583148861136e-06,
      "loss": 0.5968,
      "step": 2517
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7528293960757404,
      "learning_rate": 8.358400853484197e-06,
      "loss": 0.6097,
      "step": 2518
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6858536043323518,
      "learning_rate": 8.351219428279536e-06,
      "loss": 0.6314,
      "step": 2519
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.655423174006149,
      "learning_rate": 8.344038877053856e-06,
      "loss": 1.1123,
      "step": 2520
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8624506961083462,
      "learning_rate": 8.336859203613387e-06,
      "loss": 0.5581,
      "step": 2521
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.629998996853792,
      "learning_rate": 8.329680411763915e-06,
      "loss": 0.5565,
      "step": 2522
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.4461994624941217,
      "learning_rate": 8.322502505310733e-06,
      "loss": 1.1044,
      "step": 2523
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4935028502587793,
      "learning_rate": 8.315325488058685e-06,
      "loss": 0.4825,
      "step": 2524
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.352184719301428,
      "learning_rate": 8.30814936381213e-06,
      "loss": 1.1187,
      "step": 2525
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7262913590711928,
      "learning_rate": 8.300974136374964e-06,
      "loss": 0.5963,
      "step": 2526
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.491862328893811,
      "learning_rate": 8.293799809550598e-06,
      "loss": 1.1351,
      "step": 2527
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.880755507872021,
      "learning_rate": 8.286626387141974e-06,
      "loss": 0.5471,
      "step": 2528
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2251396767338014,
      "learning_rate": 8.279453872951551e-06,
      "loss": 1.1581,
      "step": 2529
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5704574971687795,
      "learning_rate": 8.272282270781301e-06,
      "loss": 0.5676,
      "step": 2530
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.425581592961932,
      "learning_rate": 8.265111584432725e-06,
      "loss": 1.1481,
      "step": 2531
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6155807197753678,
      "learning_rate": 8.257941817706825e-06,
      "loss": 0.6144,
      "step": 2532
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6617471067354594,
      "learning_rate": 8.250772974404128e-06,
      "loss": 0.6045,
      "step": 2533
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6576377504177526,
      "learning_rate": 8.243605058324661e-06,
      "loss": 0.55,
      "step": 2534
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.720370186996353,
      "learning_rate": 8.236438073267972e-06,
      "loss": 0.6264,
      "step": 2535
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7825941415183448,
      "learning_rate": 8.229272023033095e-06,
      "loss": 0.5724,
      "step": 2536
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7106739652922816,
      "learning_rate": 8.222106911418593e-06,
      "loss": 0.5657,
      "step": 2537
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6790743302200817,
      "learning_rate": 8.214942742222513e-06,
      "loss": 0.576,
      "step": 2538
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7290138987747858,
      "learning_rate": 8.207779519242414e-06,
      "loss": 0.5814,
      "step": 2539
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.748594278496344,
      "learning_rate": 8.200617246275342e-06,
      "loss": 0.5792,
      "step": 2540
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.666565878858387,
      "learning_rate": 8.19345592711785e-06,
      "loss": 0.5669,
      "step": 2541
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7179396366234616,
      "learning_rate": 8.186295565565982e-06,
      "loss": 0.5921,
      "step": 2542
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.5515229971628526,
      "learning_rate": 8.179136165415266e-06,
      "loss": 1.0647,
      "step": 2543
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.4478964819967444,
      "learning_rate": 8.17197773046074e-06,
      "loss": 1.1639,
      "step": 2544
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2610128092784763,
      "learning_rate": 8.16482026449691e-06,
      "loss": 1.1153,
      "step": 2545
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7449668750580878,
      "learning_rate": 8.157663771317782e-06,
      "loss": 0.5608,
      "step": 2546
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6348369064827803,
      "learning_rate": 8.150508254716834e-06,
      "loss": 0.6033,
      "step": 2547
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9106456117092654,
      "learning_rate": 8.143353718487043e-06,
      "loss": 0.5882,
      "step": 2548
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.7705474496275944,
      "learning_rate": 8.136200166420851e-06,
      "loss": 1.0721,
      "step": 2549
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6255933326966683,
      "learning_rate": 8.129047602310184e-06,
      "loss": 0.5597,
      "step": 2550
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5636336716505388,
      "learning_rate": 8.121896029946447e-06,
      "loss": 0.5549,
      "step": 2551
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6162761155666843,
      "learning_rate": 8.114745453120509e-06,
      "loss": 0.5973,
      "step": 2552
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7470897171442077,
      "learning_rate": 8.107595875622733e-06,
      "loss": 0.6465,
      "step": 2553
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.3670535736383242,
      "learning_rate": 8.100447301242922e-06,
      "loss": 1.1057,
      "step": 2554
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7160577897354798,
      "learning_rate": 8.09329973377038e-06,
      "loss": 0.5775,
      "step": 2555
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6626314652641137,
      "learning_rate": 8.086153176993847e-06,
      "loss": 0.5288,
      "step": 2556
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.694807767660456,
      "learning_rate": 8.079007634701553e-06,
      "loss": 0.6223,
      "step": 2557
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6799153216632716,
      "learning_rate": 8.071863110681167e-06,
      "loss": 0.58,
      "step": 2558
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.6025206785300194,
      "learning_rate": 8.064719608719836e-06,
      "loss": 1.1144,
      "step": 2559
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.8700680721032334,
      "learning_rate": 8.057577132604163e-06,
      "loss": 1.0589,
      "step": 2560
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5015553066462661,
      "learning_rate": 8.050435686120192e-06,
      "loss": 0.5463,
      "step": 2561
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6135803846873429,
      "learning_rate": 8.04329527305344e-06,
      "loss": 0.5563,
      "step": 2562
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6685650192048735,
      "learning_rate": 8.036155897188864e-06,
      "loss": 0.5531,
      "step": 2563
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5951358930968258,
      "learning_rate": 8.029017562310883e-06,
      "loss": 0.572,
      "step": 2564
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8415826384356173,
      "learning_rate": 8.021880272203347e-06,
      "loss": 0.6097,
      "step": 2565
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7995115605130454,
      "learning_rate": 8.014744030649569e-06,
      "loss": 0.5995,
      "step": 2566
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6078924406429904,
      "learning_rate": 8.007608841432296e-06,
      "loss": 0.5091,
      "step": 2567
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6858801483581651,
      "learning_rate": 8.000474708333724e-06,
      "loss": 0.6408,
      "step": 2568
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.855145321419643,
      "learning_rate": 7.993341635135483e-06,
      "loss": 0.6307,
      "step": 2569
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4516482608875878,
      "learning_rate": 7.986209625618644e-06,
      "loss": 0.5265,
      "step": 2570
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8912497593849844,
      "learning_rate": 7.979078683563714e-06,
      "loss": 0.6096,
      "step": 2571
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.5328673817812724,
      "learning_rate": 7.971948812750634e-06,
      "loss": 1.1558,
      "step": 2572
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3860252582514145,
      "learning_rate": 7.96482001695878e-06,
      "loss": 1.1392,
      "step": 2573
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.243395235779239,
      "learning_rate": 7.957692299966947e-06,
      "loss": 1.1044,
      "step": 2574
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.783566209366148,
      "learning_rate": 7.950565665553378e-06,
      "loss": 0.6369,
      "step": 2575
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5904467423020021,
      "learning_rate": 7.94344011749572e-06,
      "loss": 0.6162,
      "step": 2576
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9582231957105813,
      "learning_rate": 7.936315659571066e-06,
      "loss": 0.592,
      "step": 2577
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5602987451395076,
      "learning_rate": 7.929192295555908e-06,
      "loss": 0.5777,
      "step": 2578
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7026153119511065,
      "learning_rate": 7.92207002922618e-06,
      "loss": 0.5628,
      "step": 2579
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6446308374969492,
      "learning_rate": 7.91494886435722e-06,
      "loss": 0.5715,
      "step": 2580
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8452470942726702,
      "learning_rate": 7.90782880472379e-06,
      "loss": 0.61,
      "step": 2581
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5385550058724633,
      "learning_rate": 7.900709854100057e-06,
      "loss": 0.5263,
      "step": 2582
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5147828780459087,
      "learning_rate": 7.893592016259605e-06,
      "loss": 0.5637,
      "step": 2583
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7121331785496055,
      "learning_rate": 7.886475294975437e-06,
      "loss": 0.5554,
      "step": 2584
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.530649426498097,
      "learning_rate": 7.879359694019948e-06,
      "loss": 1.0852,
      "step": 2585
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6653408178009093,
      "learning_rate": 7.872245217164954e-06,
      "loss": 0.6277,
      "step": 2586
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6832344580812624,
      "learning_rate": 7.86513186818166e-06,
      "loss": 0.6811,
      "step": 2587
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.221411229253218,
      "learning_rate": 7.858019650840691e-06,
      "loss": 1.0549,
      "step": 2588
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5186217942558105,
      "learning_rate": 7.850908568912054e-06,
      "loss": 0.5539,
      "step": 2589
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.671573341502806,
      "learning_rate": 7.843798626165168e-06,
      "loss": 0.5948,
      "step": 2590
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6448948970016675,
      "learning_rate": 7.836689826368844e-06,
      "loss": 0.579,
      "step": 2591
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5251701521215817,
      "learning_rate": 7.829582173291279e-06,
      "loss": 0.6387,
      "step": 2592
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5703357173841033,
      "learning_rate": 7.822475670700078e-06,
      "loss": 0.6167,
      "step": 2593
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.706577686492029,
      "learning_rate": 7.815370322362218e-06,
      "loss": 0.5535,
      "step": 2594
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8634532390230052,
      "learning_rate": 7.808266132044083e-06,
      "loss": 0.5647,
      "step": 2595
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6753624468891053,
      "learning_rate": 7.801163103511424e-06,
      "loss": 0.51,
      "step": 2596
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.28401553095287,
      "learning_rate": 7.794061240529395e-06,
      "loss": 1.0865,
      "step": 2597
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8430841146762242,
      "learning_rate": 7.786960546862515e-06,
      "loss": 0.5275,
      "step": 2598
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7315510025824723,
      "learning_rate": 7.7798610262747e-06,
      "loss": 0.6167,
      "step": 2599
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7867089400932061,
      "learning_rate": 7.772762682529227e-06,
      "loss": 0.5184,
      "step": 2600
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6717440100344654,
      "learning_rate": 7.765665519388764e-06,
      "loss": 0.6313,
      "step": 2601
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6700431676831373,
      "learning_rate": 7.75856954061534e-06,
      "loss": 0.6255,
      "step": 2602
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.530740678768807,
      "learning_rate": 7.751474749970369e-06,
      "loss": 1.1057,
      "step": 2603
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6239117913632517,
      "learning_rate": 7.744381151214627e-06,
      "loss": 0.653,
      "step": 2604
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8028076312146566,
      "learning_rate": 7.737288748108256e-06,
      "loss": 0.6286,
      "step": 2605
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6691876016765577,
      "learning_rate": 7.730197544410777e-06,
      "loss": 0.5268,
      "step": 2606
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.277740780386049,
      "learning_rate": 7.723107543881058e-06,
      "loss": 1.1516,
      "step": 2607
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5405302468707853,
      "learning_rate": 7.716018750277347e-06,
      "loss": 0.5497,
      "step": 2608
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6529072853833326,
      "learning_rate": 7.708931167357234e-06,
      "loss": 0.6033,
      "step": 2609
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6758551433017526,
      "learning_rate": 7.701844798877684e-06,
      "loss": 0.6087,
      "step": 2610
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.910924596809108,
      "learning_rate": 7.694759648595007e-06,
      "loss": 0.5898,
      "step": 2611
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.694992431682535,
      "learning_rate": 7.687675720264874e-06,
      "loss": 0.5418,
      "step": 2612
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6780082414497315,
      "learning_rate": 7.680593017642302e-06,
      "loss": 0.6444,
      "step": 2613
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.3729485050693135,
      "learning_rate": 7.673511544481657e-06,
      "loss": 1.1308,
      "step": 2614
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5126835023911196,
      "learning_rate": 7.66643130453667e-06,
      "loss": 0.5426,
      "step": 2615
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5882195589906818,
      "learning_rate": 7.659352301560394e-06,
      "loss": 0.5272,
      "step": 2616
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.314675276546401,
      "learning_rate": 7.652274539305249e-06,
      "loss": 1.11,
      "step": 2617
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7495589421159397,
      "learning_rate": 7.645198021522976e-06,
      "loss": 0.5924,
      "step": 2618
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6149572254226952,
      "learning_rate": 7.638122751964679e-06,
      "loss": 0.5843,
      "step": 2619
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2036270445992794,
      "learning_rate": 7.631048734380779e-06,
      "loss": 1.0871,
      "step": 2620
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.3787884274641495,
      "learning_rate": 7.6239759725210445e-06,
      "loss": 1.123,
      "step": 2621
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.550443493553541,
      "learning_rate": 7.616904470134581e-06,
      "loss": 1.0631,
      "step": 2622
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6180201483061813,
      "learning_rate": 7.609834230969819e-06,
      "loss": 0.5851,
      "step": 2623
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7784705293307204,
      "learning_rate": 7.602765258774525e-06,
      "loss": 0.6518,
      "step": 2624
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7430130939257755,
      "learning_rate": 7.5956975572957825e-06,
      "loss": 0.6016,
      "step": 2625
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5639652415122058,
      "learning_rate": 7.588631130280023e-06,
      "loss": 0.5325,
      "step": 2626
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.555761994752379,
      "learning_rate": 7.5815659814729795e-06,
      "loss": 0.5249,
      "step": 2627
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7359524252279759,
      "learning_rate": 7.574502114619725e-06,
      "loss": 0.6018,
      "step": 2628
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6974503763754714,
      "learning_rate": 7.567439533464639e-06,
      "loss": 0.5639,
      "step": 2629
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.450382875196213,
      "learning_rate": 7.560378241751434e-06,
      "loss": 0.5724,
      "step": 2630
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6586354558303795,
      "learning_rate": 7.553318243223124e-06,
      "loss": 0.5118,
      "step": 2631
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.532760630759603,
      "learning_rate": 7.5462595416220506e-06,
      "loss": 0.539,
      "step": 2632
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.48603252432937,
      "learning_rate": 7.539202140689856e-06,
      "loss": 1.0818,
      "step": 2633
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.58182047286231,
      "learning_rate": 7.5321460441675025e-06,
      "loss": 0.5811,
      "step": 2634
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7266252885518774,
      "learning_rate": 7.5250912557952575e-06,
      "loss": 0.5374,
      "step": 2635
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5455888521189716,
      "learning_rate": 7.5180377793126905e-06,
      "loss": 0.5612,
      "step": 2636
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6858339581060684,
      "learning_rate": 7.510985618458687e-06,
      "loss": 0.5809,
      "step": 2637
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7144932999038218,
      "learning_rate": 7.503934776971421e-06,
      "loss": 0.5757,
      "step": 2638
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8313483053347184,
      "learning_rate": 7.496885258588382e-06,
      "loss": 0.6215,
      "step": 2639
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6106420805216353,
      "learning_rate": 7.489837067046342e-06,
      "loss": 0.5748,
      "step": 2640
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.732367014184862,
      "learning_rate": 7.482790206081386e-06,
      "loss": 0.5709,
      "step": 2641
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.579582560409956,
      "learning_rate": 7.47574467942888e-06,
      "loss": 0.5991,
      "step": 2642
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6339427147112933,
      "learning_rate": 7.468700490823493e-06,
      "loss": 0.5451,
      "step": 2643
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.518579127172627,
      "learning_rate": 7.461657643999177e-06,
      "loss": 1.1517,
      "step": 2644
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8959116044263424,
      "learning_rate": 7.454616142689173e-06,
      "loss": 0.615,
      "step": 2645
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4504278849285728,
      "learning_rate": 7.447575990626019e-06,
      "loss": 0.5275,
      "step": 2646
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5007669827553822,
      "learning_rate": 7.440537191541522e-06,
      "loss": 0.4881,
      "step": 2647
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.753848209688336,
      "learning_rate": 7.43349974916679e-06,
      "loss": 0.5371,
      "step": 2648
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5585197875445846,
      "learning_rate": 7.426463667232193e-06,
      "loss": 0.5663,
      "step": 2649
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5598971638894723,
      "learning_rate": 7.419428949467397e-06,
      "loss": 0.5938,
      "step": 2650
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6299856942924642,
      "learning_rate": 7.412395599601332e-06,
      "loss": 0.5957,
      "step": 2651
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7400239056350932,
      "learning_rate": 7.405363621362209e-06,
      "loss": 0.5765,
      "step": 2652
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6832516050187893,
      "learning_rate": 7.398333018477511e-06,
      "loss": 0.5852,
      "step": 2653
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5649679827721013,
      "learning_rate": 7.391303794673992e-06,
      "loss": 0.643,
      "step": 2654
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.4580139549858506,
      "learning_rate": 7.3842759536776756e-06,
      "loss": 1.039,
      "step": 2655
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.545662059297986,
      "learning_rate": 7.377249499213846e-06,
      "loss": 0.5752,
      "step": 2656
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6202032849437775,
      "learning_rate": 7.370224435007065e-06,
      "loss": 0.5886,
      "step": 2657
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5944312435030352,
      "learning_rate": 7.363200764781145e-06,
      "loss": 0.5399,
      "step": 2658
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6240219461130172,
      "learning_rate": 7.356178492259172e-06,
      "loss": 0.5365,
      "step": 2659
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6593473844330633,
      "learning_rate": 7.349157621163473e-06,
      "loss": 0.5635,
      "step": 2660
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6309196589198365,
      "learning_rate": 7.342138155215655e-06,
      "loss": 0.5319,
      "step": 2661
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5665455467691767,
      "learning_rate": 7.335120098136561e-06,
      "loss": 0.6212,
      "step": 2662
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8136473114087508,
      "learning_rate": 7.328103453646298e-06,
      "loss": 0.6374,
      "step": 2663
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7062935116253197,
      "learning_rate": 7.321088225464215e-06,
      "loss": 0.6406,
      "step": 2664
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7343565279141555,
      "learning_rate": 7.314074417308924e-06,
      "loss": 0.5777,
      "step": 2665
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.238336615419023,
      "learning_rate": 7.307062032898274e-06,
      "loss": 1.0905,
      "step": 2666
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7308821364477394,
      "learning_rate": 7.300051075949352e-06,
      "loss": 0.6877,
      "step": 2667
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.705663640547712,
      "learning_rate": 7.293041550178513e-06,
      "loss": 0.5608,
      "step": 2668
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6859256037666217,
      "learning_rate": 7.286033459301325e-06,
      "loss": 0.5851,
      "step": 2669
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6368998314130325,
      "learning_rate": 7.27902680703262e-06,
      "loss": 0.5256,
      "step": 2670
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8665791249827615,
      "learning_rate": 7.272021597086445e-06,
      "loss": 0.5545,
      "step": 2671
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1760522736286227,
      "learning_rate": 7.265017833176105e-06,
      "loss": 1.0838,
      "step": 2672
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.52148744172886,
      "learning_rate": 7.258015519014119e-06,
      "loss": 0.5575,
      "step": 2673
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.6223292828309392,
      "learning_rate": 7.2510146583122495e-06,
      "loss": 1.0173,
      "step": 2674
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8098403945203125,
      "learning_rate": 7.244015254781482e-06,
      "loss": 0.6125,
      "step": 2675
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.783989476250832,
      "learning_rate": 7.237017312132031e-06,
      "loss": 1.1644,
      "step": 2676
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7181456686207772,
      "learning_rate": 7.230020834073341e-06,
      "loss": 0.5897,
      "step": 2677
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.357838159075139,
      "learning_rate": 7.223025824314074e-06,
      "loss": 1.1828,
      "step": 2678
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.7582586721747653,
      "learning_rate": 7.216032286562122e-06,
      "loss": 1.1979,
      "step": 2679
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6542421022057572,
      "learning_rate": 7.209040224524585e-06,
      "loss": 0.6029,
      "step": 2680
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6828208469152954,
      "learning_rate": 7.202049641907794e-06,
      "loss": 0.5202,
      "step": 2681
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6754786970848796,
      "learning_rate": 7.195060542417282e-06,
      "loss": 0.4817,
      "step": 2682
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7409835665219242,
      "learning_rate": 7.188072929757806e-06,
      "loss": 0.6577,
      "step": 2683
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.602316996421848,
      "learning_rate": 7.181086807633331e-06,
      "loss": 0.5903,
      "step": 2684
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.5589178296704604,
      "learning_rate": 7.174102179747036e-06,
      "loss": 1.0646,
      "step": 2685
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7500072140276712,
      "learning_rate": 7.167119049801301e-06,
      "loss": 0.5729,
      "step": 2686
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5449667071904676,
      "learning_rate": 7.16013742149771e-06,
      "loss": 0.5722,
      "step": 2687
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8684931715080146,
      "learning_rate": 7.15315729853707e-06,
      "loss": 0.6506,
      "step": 2688
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5462142068783837,
      "learning_rate": 7.146178684619363e-06,
      "loss": 0.5583,
      "step": 2689
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.097255539872722,
      "learning_rate": 7.139201583443793e-06,
      "loss": 1.0553,
      "step": 2690
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.623426479409763,
      "learning_rate": 7.13222599870875e-06,
      "loss": 0.5789,
      "step": 2691
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.4117847358728404,
      "learning_rate": 7.1252519341118275e-06,
      "loss": 1.1214,
      "step": 2692
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6425722441005515,
      "learning_rate": 7.118279393349807e-06,
      "loss": 0.5783,
      "step": 2693
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2656209524642676,
      "learning_rate": 7.111308380118666e-06,
      "loss": 1.0977,
      "step": 2694
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.646224408821923,
      "learning_rate": 7.104338898113566e-06,
      "loss": 0.5605,
      "step": 2695
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7033364934604247,
      "learning_rate": 7.097370951028872e-06,
      "loss": 0.4756,
      "step": 2696
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.327648331126226,
      "learning_rate": 7.090404542558119e-06,
      "loss": 1.093,
      "step": 2697
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6532470156773809,
      "learning_rate": 7.08343967639403e-06,
      "loss": 0.562,
      "step": 2698
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5954624182608375,
      "learning_rate": 7.07647635622852e-06,
      "loss": 0.5677,
      "step": 2699
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5416286458828443,
      "learning_rate": 7.069514585752671e-06,
      "loss": 0.5497,
      "step": 2700
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.340163697169957,
      "learning_rate": 7.062554368656758e-06,
      "loss": 1.0899,
      "step": 2701
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.488311061038614,
      "learning_rate": 7.055595708630217e-06,
      "loss": 1.0899,
      "step": 2702
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.712246991456931,
      "learning_rate": 7.048638609361676e-06,
      "loss": 0.6217,
      "step": 2703
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.458757822273892,
      "learning_rate": 7.041683074538916e-06,
      "loss": 0.5686,
      "step": 2704
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5937249824821333,
      "learning_rate": 7.034729107848907e-06,
      "loss": 0.5848,
      "step": 2705
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7874606189435929,
      "learning_rate": 7.027776712977772e-06,
      "loss": 0.5975,
      "step": 2706
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.5643893671039186,
      "learning_rate": 7.020825893610816e-06,
      "loss": 1.0884,
      "step": 2707
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5577790650064531,
      "learning_rate": 7.013876653432501e-06,
      "loss": 0.5853,
      "step": 2708
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.585264045631379,
      "learning_rate": 7.006928996126443e-06,
      "loss": 0.5515,
      "step": 2709
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.276458576534024,
      "learning_rate": 6.999982925375439e-06,
      "loss": 1.1096,
      "step": 2710
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6841793488003007,
      "learning_rate": 6.9930384448614265e-06,
      "loss": 0.5953,
      "step": 2711
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6894202339478304,
      "learning_rate": 6.986095558265516e-06,
      "loss": 0.5927,
      "step": 2712
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5020922423630239,
      "learning_rate": 6.979154269267958e-06,
      "loss": 0.5848,
      "step": 2713
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.425737653616691,
      "learning_rate": 6.972214581548166e-06,
      "loss": 0.5474,
      "step": 2714
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8428745506111308,
      "learning_rate": 6.965276498784701e-06,
      "loss": 0.6502,
      "step": 2715
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3524627221697596,
      "learning_rate": 6.958340024655274e-06,
      "loss": 0.5092,
      "step": 2716
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2883358537023875,
      "learning_rate": 6.9514051628367465e-06,
      "loss": 1.1608,
      "step": 2717
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.588463220776896,
      "learning_rate": 6.944471917005114e-06,
      "loss": 0.5959,
      "step": 2718
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4729296764809106,
      "learning_rate": 6.937540290835534e-06,
      "loss": 0.5615,
      "step": 2719
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.602901473596317,
      "learning_rate": 6.930610288002284e-06,
      "loss": 0.5712,
      "step": 2720
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6634002696851207,
      "learning_rate": 6.9236819121788035e-06,
      "loss": 0.5816,
      "step": 2721
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5734951359065454,
      "learning_rate": 6.916755167037647e-06,
      "loss": 0.6358,
      "step": 2722
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.363935828784092,
      "learning_rate": 6.909830056250527e-06,
      "loss": 0.5196,
      "step": 2723
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5424713482310002,
      "learning_rate": 6.902906583488269e-06,
      "loss": 0.6026,
      "step": 2724
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5508755415656048,
      "learning_rate": 6.895984752420848e-06,
      "loss": 0.5299,
      "step": 2725
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.717351574238104,
      "learning_rate": 6.889064566717349e-06,
      "loss": 0.5585,
      "step": 2726
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1869753248542425,
      "learning_rate": 6.882146030046008e-06,
      "loss": 1.0777,
      "step": 2727
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5033854354054739,
      "learning_rate": 6.875229146074169e-06,
      "loss": 0.6646,
      "step": 2728
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7823633456775576,
      "learning_rate": 6.868313918468305e-06,
      "loss": 0.6434,
      "step": 2729
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5123252015079052,
      "learning_rate": 6.861400350894016e-06,
      "loss": 0.5671,
      "step": 2730
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5290513175781268,
      "learning_rate": 6.854488447016014e-06,
      "loss": 0.5535,
      "step": 2731
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6871864730690322,
      "learning_rate": 6.847578210498139e-06,
      "loss": 0.5522,
      "step": 2732
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6709614871861327,
      "learning_rate": 6.840669645003334e-06,
      "loss": 0.5584,
      "step": 2733
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5929493834907498,
      "learning_rate": 6.8337627541936714e-06,
      "loss": 0.5623,
      "step": 2734
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.433600289010034,
      "learning_rate": 6.826857541730323e-06,
      "loss": 1.0753,
      "step": 2735
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.676280778850989,
      "learning_rate": 6.819954011273577e-06,
      "loss": 0.5718,
      "step": 2736
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.294666771960285,
      "learning_rate": 6.813052166482828e-06,
      "loss": 1.0498,
      "step": 2737
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5328982672770688,
      "learning_rate": 6.806152011016582e-06,
      "loss": 0.5292,
      "step": 2738
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6510431123807745,
      "learning_rate": 6.799253548532445e-06,
      "loss": 0.6107,
      "step": 2739
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5422896392468737,
      "learning_rate": 6.79235678268712e-06,
      "loss": 0.5804,
      "step": 2740
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6766068652808876,
      "learning_rate": 6.7854617171364236e-06,
      "loss": 0.589,
      "step": 2741
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5561661942891023,
      "learning_rate": 6.77856835553526e-06,
      "loss": 0.5618,
      "step": 2742
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5633605238793524,
      "learning_rate": 6.77167670153764e-06,
      "loss": 0.5441,
      "step": 2743
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.307472828901499,
      "learning_rate": 6.7647867587966595e-06,
      "loss": 0.548,
      "step": 2744
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.371806866997689,
      "learning_rate": 6.757898530964513e-06,
      "loss": 1.0712,
      "step": 2745
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.550140207065199,
      "learning_rate": 6.751012021692484e-06,
      "loss": 0.4978,
      "step": 2746
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5514537425144423,
      "learning_rate": 6.744127234630948e-06,
      "loss": 0.5833,
      "step": 2747
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.5628133947027196,
      "learning_rate": 6.737244173429361e-06,
      "loss": 1.1372,
      "step": 2748
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.401824364520341,
      "learning_rate": 6.730362841736273e-06,
      "loss": 1.0897,
      "step": 2749
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7902900398791102,
      "learning_rate": 6.7234832431993115e-06,
      "loss": 0.6144,
      "step": 2750
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7299010491561206,
      "learning_rate": 6.716605381465181e-06,
      "loss": 0.5683,
      "step": 2751
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.475966978684228,
      "learning_rate": 6.70972926017968e-06,
      "loss": 0.5364,
      "step": 2752
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6772703655485903,
      "learning_rate": 6.702854882987666e-06,
      "loss": 0.5649,
      "step": 2753
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8741282031610753,
      "learning_rate": 6.69598225353309e-06,
      "loss": 0.6324,
      "step": 2754
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7281943714538874,
      "learning_rate": 6.689111375458961e-06,
      "loss": 0.6363,
      "step": 2755
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6539981418874918,
      "learning_rate": 6.682242252407372e-06,
      "loss": 0.6034,
      "step": 2756
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.4186512679634222,
      "learning_rate": 6.6753748880194725e-06,
      "loss": 1.1418,
      "step": 2757
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.544097293312286,
      "learning_rate": 6.668509285935496e-06,
      "loss": 0.4602,
      "step": 2758
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6086551342423232,
      "learning_rate": 6.66164544979473e-06,
      "loss": 0.5541,
      "step": 2759
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.678928500523047,
      "learning_rate": 6.654783383235525e-06,
      "loss": 0.5876,
      "step": 2760
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5563885191695328,
      "learning_rate": 6.647923089895304e-06,
      "loss": 0.5629,
      "step": 2761
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5337701799710968,
      "learning_rate": 6.641064573410537e-06,
      "loss": 0.5265,
      "step": 2762
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.548689653823505,
      "learning_rate": 6.634207837416768e-06,
      "loss": 0.5807,
      "step": 2763
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5049177720647013,
      "learning_rate": 6.627352885548577e-06,
      "loss": 0.5928,
      "step": 2764
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.675947712638721,
      "learning_rate": 6.620499721439619e-06,
      "loss": 1.0961,
      "step": 2765
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6388546600476634,
      "learning_rate": 6.613648348722588e-06,
      "loss": 0.5607,
      "step": 2766
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.3360081357703417,
      "learning_rate": 6.606798771029233e-06,
      "loss": 1.1399,
      "step": 2767
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4885056575903948,
      "learning_rate": 6.599950991990345e-06,
      "loss": 0.517,
      "step": 2768
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5380213737169839,
      "learning_rate": 6.593105015235776e-06,
      "loss": 0.5362,
      "step": 2769
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7230772606584,
      "learning_rate": 6.586260844394411e-06,
      "loss": 0.5513,
      "step": 2770
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.555270902896544,
      "learning_rate": 6.579418483094173e-06,
      "loss": 1.097,
      "step": 2771
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5775808929347797,
      "learning_rate": 6.572577934962046e-06,
      "loss": 0.5717,
      "step": 2772
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.537479596793509,
      "learning_rate": 6.565739203624028e-06,
      "loss": 0.5499,
      "step": 2773
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5898350651109436,
      "learning_rate": 6.558902292705176e-06,
      "loss": 0.6344,
      "step": 2774
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.534152604910072,
      "learning_rate": 6.552067205829568e-06,
      "loss": 0.5475,
      "step": 2775
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6768542006390907,
      "learning_rate": 6.545233946620318e-06,
      "loss": 0.6245,
      "step": 2776
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6571117858102438,
      "learning_rate": 6.5384025186995755e-06,
      "loss": 0.5797,
      "step": 2777
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8014366809187836,
      "learning_rate": 6.531572925688516e-06,
      "loss": 0.6187,
      "step": 2778
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.46690226713374,
      "learning_rate": 6.524745171207339e-06,
      "loss": 1.0762,
      "step": 2779
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4675286424404719,
      "learning_rate": 6.517919258875281e-06,
      "loss": 0.5587,
      "step": 2780
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6840325861828476,
      "learning_rate": 6.511095192310587e-06,
      "loss": 0.4971,
      "step": 2781
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.159082747859781,
      "learning_rate": 6.50427297513053e-06,
      "loss": 1.1023,
      "step": 2782
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7597649813336995,
      "learning_rate": 6.49745261095141e-06,
      "loss": 0.5485,
      "step": 2783
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7898416610900787,
      "learning_rate": 6.490634103388531e-06,
      "loss": 0.4778,
      "step": 2784
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7884155934234611,
      "learning_rate": 6.483817456056228e-06,
      "loss": 0.6177,
      "step": 2785
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7548302826790299,
      "learning_rate": 6.477002672567836e-06,
      "loss": 0.5764,
      "step": 2786
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6732215401790131,
      "learning_rate": 6.470189756535712e-06,
      "loss": 0.5692,
      "step": 2787
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7769177206165767,
      "learning_rate": 6.463378711571212e-06,
      "loss": 0.5648,
      "step": 2788
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.428433088395702,
      "learning_rate": 6.456569541284717e-06,
      "loss": 1.1123,
      "step": 2789
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7170744738417814,
      "learning_rate": 6.449762249285596e-06,
      "loss": 0.5586,
      "step": 2790
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.267100444542781,
      "learning_rate": 6.44295683918224e-06,
      "loss": 1.1012,
      "step": 2791
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7685601955975319,
      "learning_rate": 6.436153314582026e-06,
      "loss": 0.6133,
      "step": 2792
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7812731554642456,
      "learning_rate": 6.42935167909134e-06,
      "loss": 0.5715,
      "step": 2793
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8775121460476232,
      "learning_rate": 6.422551936315568e-06,
      "loss": 0.5681,
      "step": 2794
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.447287817102569,
      "learning_rate": 6.415754089859087e-06,
      "loss": 1.085,
      "step": 2795
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7658864715034663,
      "learning_rate": 6.40895814332528e-06,
      "loss": 0.6249,
      "step": 2796
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5444795803685938,
      "learning_rate": 6.402164100316506e-06,
      "loss": 0.5481,
      "step": 2797
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5621764571678256,
      "learning_rate": 6.395371964434129e-06,
      "loss": 0.6066,
      "step": 2798
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.715655668459193,
      "learning_rate": 6.388581739278492e-06,
      "loss": 0.5989,
      "step": 2799
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5283315303979825,
      "learning_rate": 6.381793428448939e-06,
      "loss": 0.5301,
      "step": 2800
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.5835615635991935,
      "learning_rate": 6.375007035543784e-06,
      "loss": 1.1491,
      "step": 2801
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.372853199599837,
      "learning_rate": 6.368222564160329e-06,
      "loss": 1.081,
      "step": 2802
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2753348354073424,
      "learning_rate": 6.361440017894866e-06,
      "loss": 1.0723,
      "step": 2803
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.453042821175819,
      "learning_rate": 6.354659400342653e-06,
      "loss": 0.5154,
      "step": 2804
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7715724267553299,
      "learning_rate": 6.34788071509794e-06,
      "loss": 0.6613,
      "step": 2805
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6162841070325746,
      "learning_rate": 6.341103965753939e-06,
      "loss": 0.528,
      "step": 2806
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.3983555179011455,
      "learning_rate": 6.334329155902843e-06,
      "loss": 1.0787,
      "step": 2807
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.4831660811602037,
      "learning_rate": 6.327556289135819e-06,
      "loss": 1.1889,
      "step": 2808
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7036103800938323,
      "learning_rate": 6.320785369043001e-06,
      "loss": 0.5954,
      "step": 2809
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.923763801671676,
      "learning_rate": 6.314016399213485e-06,
      "loss": 0.5988,
      "step": 2810
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.4257786094006777,
      "learning_rate": 6.307249383235348e-06,
      "loss": 1.0487,
      "step": 2811
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5560482567319984,
      "learning_rate": 6.300484324695617e-06,
      "loss": 0.5855,
      "step": 2812
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8939393101492907,
      "learning_rate": 6.293721227180286e-06,
      "loss": 0.5494,
      "step": 2813
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.368704561330165,
      "learning_rate": 6.286960094274317e-06,
      "loss": 1.0977,
      "step": 2814
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4815536783588923,
      "learning_rate": 6.280200929561615e-06,
      "loss": 0.5648,
      "step": 2815
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7612505547312893,
      "learning_rate": 6.27344373662506e-06,
      "loss": 0.606,
      "step": 2816
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6618995552170575,
      "learning_rate": 6.266688519046471e-06,
      "loss": 0.5902,
      "step": 2817
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6303211667633475,
      "learning_rate": 6.259935280406633e-06,
      "loss": 0.552,
      "step": 2818
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7738838115315119,
      "learning_rate": 6.253184024285267e-06,
      "loss": 0.5709,
      "step": 2819
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.666144361003416,
      "learning_rate": 6.24643475426106e-06,
      "loss": 0.6134,
      "step": 2820
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6456175870050946,
      "learning_rate": 6.2396874739116306e-06,
      "loss": 0.5678,
      "step": 2821
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8313727087475122,
      "learning_rate": 6.232942186813559e-06,
      "loss": 0.567,
      "step": 2822
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.686462096574209,
      "learning_rate": 6.2261988965423535e-06,
      "loss": 0.6506,
      "step": 2823
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6525176552559078,
      "learning_rate": 6.219457606672468e-06,
      "loss": 0.5912,
      "step": 2824
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7760464034867993,
      "learning_rate": 6.212718320777307e-06,
      "loss": 0.5869,
      "step": 2825
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7224125428334467,
      "learning_rate": 6.205981042429193e-06,
      "loss": 0.634,
      "step": 2826
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5890972515324246,
      "learning_rate": 6.199245775199407e-06,
      "loss": 0.5955,
      "step": 2827
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7666263479736164,
      "learning_rate": 6.192512522658144e-06,
      "loss": 0.589,
      "step": 2828
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.695248531552937,
      "learning_rate": 6.185781288374545e-06,
      "loss": 0.5757,
      "step": 2829
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6088982381421093,
      "learning_rate": 6.179052075916668e-06,
      "loss": 0.5546,
      "step": 2830
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5783235569191625,
      "learning_rate": 6.172324888851516e-06,
      "loss": 0.5526,
      "step": 2831
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2205032259669304,
      "learning_rate": 6.165599730745002e-06,
      "loss": 1.1141,
      "step": 2832
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.569510451219489,
      "learning_rate": 6.1588766051619774e-06,
      "loss": 0.5794,
      "step": 2833
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7447107768107557,
      "learning_rate": 6.152155515666206e-06,
      "loss": 0.6399,
      "step": 2834
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9088160548766446,
      "learning_rate": 6.145436465820371e-06,
      "loss": 0.6174,
      "step": 2835
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.74877541641399,
      "learning_rate": 6.13871945918609e-06,
      "loss": 0.5731,
      "step": 2836
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6719470987982348,
      "learning_rate": 6.132004499323878e-06,
      "loss": 0.5936,
      "step": 2837
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.61454337774748,
      "learning_rate": 6.125291589793179e-06,
      "loss": 0.5652,
      "step": 2838
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7082475264389563,
      "learning_rate": 6.1185807341523405e-06,
      "loss": 0.6003,
      "step": 2839
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5462824576240322,
      "learning_rate": 6.111871935958633e-06,
      "loss": 0.5765,
      "step": 2840
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.228475066786984,
      "learning_rate": 6.105165198768219e-06,
      "loss": 1.1072,
      "step": 2841
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.5035650384063306,
      "learning_rate": 6.09846052613619e-06,
      "loss": 1.1091,
      "step": 2842
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.50135364431745,
      "learning_rate": 6.091757921616526e-06,
      "loss": 0.5442,
      "step": 2843
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.3305215603122242,
      "learning_rate": 6.085057388762114e-06,
      "loss": 1.1096,
      "step": 2844
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1826241352935836,
      "learning_rate": 6.078358931124753e-06,
      "loss": 1.0193,
      "step": 2845
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4876795880075306,
      "learning_rate": 6.071662552255125e-06,
      "loss": 0.518,
      "step": 2846
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6376779682352505,
      "learning_rate": 6.06496825570283e-06,
      "loss": 0.5785,
      "step": 2847
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8884170463918515,
      "learning_rate": 6.058276045016347e-06,
      "loss": 0.572,
      "step": 2848
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5833344786033507,
      "learning_rate": 6.051585923743059e-06,
      "loss": 0.6155,
      "step": 2849
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6682584578780262,
      "learning_rate": 6.044897895429234e-06,
      "loss": 0.6048,
      "step": 2850
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6485334315948543,
      "learning_rate": 6.038211963620042e-06,
      "loss": 0.6006,
      "step": 2851
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4784341282874132,
      "learning_rate": 6.031528131859529e-06,
      "loss": 0.5209,
      "step": 2852
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9356945799428793,
      "learning_rate": 6.024846403690639e-06,
      "loss": 0.54,
      "step": 2853
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.408623884646242,
      "learning_rate": 6.018166782655193e-06,
      "loss": 1.0846,
      "step": 2854
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5787905632722108,
      "learning_rate": 6.011489272293894e-06,
      "loss": 0.5589,
      "step": 2855
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6931896398525135,
      "learning_rate": 6.004813876146338e-06,
      "loss": 0.5965,
      "step": 2856
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.169845569560163,
      "learning_rate": 5.998140597750982e-06,
      "loss": 1.1378,
      "step": 2857
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5394300092981146,
      "learning_rate": 5.991469440645181e-06,
      "loss": 0.528,
      "step": 2858
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.3326905624302263,
      "learning_rate": 5.984800408365148e-06,
      "loss": 1.1251,
      "step": 2859
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.429143913114145,
      "learning_rate": 5.978133504445982e-06,
      "loss": 1.0951,
      "step": 2860
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.820884852856305,
      "learning_rate": 5.971468732421641e-06,
      "loss": 0.588,
      "step": 2861
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6996009019029925,
      "learning_rate": 5.964806095824968e-06,
      "loss": 0.599,
      "step": 2862
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6010563308464227,
      "learning_rate": 5.958145598187662e-06,
      "loss": 0.567,
      "step": 2863
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.577370850529895,
      "learning_rate": 5.951487243040299e-06,
      "loss": 1.1934,
      "step": 2864
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.648796447895308,
      "learning_rate": 5.944831033912308e-06,
      "loss": 0.6072,
      "step": 2865
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6373920483882614,
      "learning_rate": 5.938176974331982e-06,
      "loss": 0.5933,
      "step": 2866
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2548495095538468,
      "learning_rate": 5.931525067826488e-06,
      "loss": 1.0721,
      "step": 2867
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.680442881746082,
      "learning_rate": 5.924875317921834e-06,
      "loss": 0.5556,
      "step": 2868
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.7471959984156897,
      "learning_rate": 5.918227728142895e-06,
      "loss": 1.0859,
      "step": 2869
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6130733651514404,
      "learning_rate": 5.911582302013399e-06,
      "loss": 0.5259,
      "step": 2870
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6927797703179952,
      "learning_rate": 5.904939043055928e-06,
      "loss": 0.5928,
      "step": 2871
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7181656658984525,
      "learning_rate": 5.89829795479191e-06,
      "loss": 0.5881,
      "step": 2872
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7211060007318681,
      "learning_rate": 5.891659040741633e-06,
      "loss": 0.6127,
      "step": 2873
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2360698486209394,
      "learning_rate": 5.885022304424217e-06,
      "loss": 1.0787,
      "step": 2874
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.361646223878647,
      "learning_rate": 5.878387749357646e-06,
      "loss": 1.0986,
      "step": 2875
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2945696002744835,
      "learning_rate": 5.871755379058734e-06,
      "loss": 1.1603,
      "step": 2876
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5922974691522835,
      "learning_rate": 5.865125197043137e-06,
      "loss": 0.5855,
      "step": 2877
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5495599074512074,
      "learning_rate": 5.858497206825362e-06,
      "loss": 0.5382,
      "step": 2878
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2905641823612113,
      "learning_rate": 5.851871411918744e-06,
      "loss": 1.1059,
      "step": 2879
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7301195304911032,
      "learning_rate": 5.84524781583546e-06,
      "loss": 0.5803,
      "step": 2880
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7315716761856275,
      "learning_rate": 5.838626422086514e-06,
      "loss": 0.6003,
      "step": 2881
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6858296231700218,
      "learning_rate": 5.832007234181754e-06,
      "loss": 0.6427,
      "step": 2882
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3622208911280977,
      "learning_rate": 5.825390255629848e-06,
      "loss": 1.1214,
      "step": 2883
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6503632993831985,
      "learning_rate": 5.8187754899383046e-06,
      "loss": 0.5355,
      "step": 2884
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6299019794011302,
      "learning_rate": 5.8121629406134414e-06,
      "loss": 0.551,
      "step": 2885
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6383743534831228,
      "learning_rate": 5.805552611160419e-06,
      "loss": 0.5711,
      "step": 2886
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5740873911458546,
      "learning_rate": 5.798944505083213e-06,
      "loss": 0.5842,
      "step": 2887
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.4785062429730256,
      "learning_rate": 5.7923386258846194e-06,
      "loss": 1.0987,
      "step": 2888
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6929562376458729,
      "learning_rate": 5.78573497706626e-06,
      "loss": 0.5701,
      "step": 2889
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.638728100751885,
      "learning_rate": 5.779133562128567e-06,
      "loss": 0.5941,
      "step": 2890
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8896818587639008,
      "learning_rate": 5.772534384570796e-06,
      "loss": 0.599,
      "step": 2891
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7610778571699983,
      "learning_rate": 5.765937447891008e-06,
      "loss": 0.5813,
      "step": 2892
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5595172539187079,
      "learning_rate": 5.7593427555860835e-06,
      "loss": 0.5228,
      "step": 2893
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5614439669244566,
      "learning_rate": 5.752750311151709e-06,
      "loss": 0.5509,
      "step": 2894
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7454363110840423,
      "learning_rate": 5.746160118082386e-06,
      "loss": 0.6574,
      "step": 2895
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.4951673104459324,
      "learning_rate": 5.739572179871409e-06,
      "loss": 1.0844,
      "step": 2896
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.80019361203927,
      "learning_rate": 5.732986500010891e-06,
      "loss": 0.6121,
      "step": 2897
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.721334926546552,
      "learning_rate": 5.7264030819917405e-06,
      "loss": 0.5766,
      "step": 2898
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6107649511973423,
      "learning_rate": 5.719821929303671e-06,
      "loss": 0.5921,
      "step": 2899
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.41820216771309,
      "learning_rate": 5.713243045435193e-06,
      "loss": 1.1059,
      "step": 2900
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2917244885901473,
      "learning_rate": 5.706666433873616e-06,
      "loss": 1.1357,
      "step": 2901
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4805200510663516,
      "learning_rate": 5.700092098105041e-06,
      "loss": 0.5293,
      "step": 2902
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.415317672240945,
      "learning_rate": 5.6935200416143665e-06,
      "loss": 1.0857,
      "step": 2903
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.4268484597798663,
      "learning_rate": 5.686950267885284e-06,
      "loss": 1.1251,
      "step": 2904
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7063994594192566,
      "learning_rate": 5.680382780400271e-06,
      "loss": 0.5817,
      "step": 2905
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2947779042619314,
      "learning_rate": 5.673817582640597e-06,
      "loss": 1.1324,
      "step": 2906
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3450583776642926,
      "learning_rate": 5.667254678086313e-06,
      "loss": 1.0653,
      "step": 2907
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8044213921536183,
      "learning_rate": 5.660694070216255e-06,
      "loss": 0.5496,
      "step": 2908
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5912090152764073,
      "learning_rate": 5.654135762508048e-06,
      "loss": 0.5464,
      "step": 2909
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.478283465606763,
      "learning_rate": 5.647579758438091e-06,
      "loss": 0.5555,
      "step": 2910
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6661510926485104,
      "learning_rate": 5.641026061481565e-06,
      "loss": 0.5362,
      "step": 2911
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5097388690750075,
      "learning_rate": 5.634474675112428e-06,
      "loss": 0.5815,
      "step": 2912
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.553545424892913,
      "learning_rate": 5.627925602803413e-06,
      "loss": 1.1067,
      "step": 2913
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.277627142140489,
      "learning_rate": 5.621378848026022e-06,
      "loss": 1.0974,
      "step": 2914
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.554739407505566,
      "learning_rate": 5.614834414250541e-06,
      "loss": 0.5511,
      "step": 2915
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.3353055985502964,
      "learning_rate": 5.608292304946004e-06,
      "loss": 1.0724,
      "step": 2916
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.706574937215534,
      "learning_rate": 5.601752523580241e-06,
      "loss": 0.6773,
      "step": 2917
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6394370083205974,
      "learning_rate": 5.595215073619822e-06,
      "loss": 0.6216,
      "step": 2918
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.4310230279816842,
      "learning_rate": 5.588679958530095e-06,
      "loss": 1.093,
      "step": 2919
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.497646840305341,
      "learning_rate": 5.582147181775169e-06,
      "loss": 0.545,
      "step": 2920
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.879256256483421,
      "learning_rate": 5.575616746817911e-06,
      "loss": 0.598,
      "step": 2921
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2950430716290784,
      "learning_rate": 5.569088657119949e-06,
      "loss": 1.0735,
      "step": 2922
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6000297611436811,
      "learning_rate": 5.562562916141665e-06,
      "loss": 0.5935,
      "step": 2923
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.674171538517641,
      "learning_rate": 5.5560395273421985e-06,
      "loss": 0.525,
      "step": 2924
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.632185517691478,
      "learning_rate": 5.54951849417944e-06,
      "loss": 0.5484,
      "step": 2925
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5905472321530374,
      "learning_rate": 5.542999820110038e-06,
      "loss": 0.6187,
      "step": 2926
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6175630357131074,
      "learning_rate": 5.536483508589377e-06,
      "loss": 0.5384,
      "step": 2927
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6332039532946074,
      "learning_rate": 5.5299695630716e-06,
      "loss": 0.5771,
      "step": 2928
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3925979561962871,
      "learning_rate": 5.523457987009595e-06,
      "loss": 0.4806,
      "step": 2929
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6490666386319464,
      "learning_rate": 5.516948783854989e-06,
      "loss": 0.5414,
      "step": 2930
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.675964537986228,
      "learning_rate": 5.510441957058158e-06,
      "loss": 0.5442,
      "step": 2931
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.560847842541756,
      "learning_rate": 5.503937510068211e-06,
      "loss": 1.0455,
      "step": 2932
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.4687115613319217,
      "learning_rate": 5.497435446333003e-06,
      "loss": 1.1042,
      "step": 2933
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7659118798769993,
      "learning_rate": 5.4909357692991185e-06,
      "loss": 0.6364,
      "step": 2934
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5210484653544483,
      "learning_rate": 5.484438482411882e-06,
      "loss": 0.5308,
      "step": 2935
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2146864138489857,
      "learning_rate": 5.477943589115347e-06,
      "loss": 1.0973,
      "step": 2936
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.683164890073679,
      "learning_rate": 5.471451092852307e-06,
      "loss": 0.612,
      "step": 2937
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5098172885658343,
      "learning_rate": 5.46496099706427e-06,
      "loss": 0.504,
      "step": 2938
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.723784124216424,
      "learning_rate": 5.458473305191483e-06,
      "loss": 0.6226,
      "step": 2939
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.46137438567381,
      "learning_rate": 5.451988020672916e-06,
      "loss": 0.4988,
      "step": 2940
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6638237496831587,
      "learning_rate": 5.445505146946262e-06,
      "loss": 0.5253,
      "step": 2941
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6559749574908433,
      "learning_rate": 5.439024687447936e-06,
      "loss": 0.5993,
      "step": 2942
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7239886826079958,
      "learning_rate": 5.432546645613077e-06,
      "loss": 0.5535,
      "step": 2943
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6031791946334446,
      "learning_rate": 5.426071024875537e-06,
      "loss": 0.5553,
      "step": 2944
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5380577407900224,
      "learning_rate": 5.419597828667886e-06,
      "loss": 0.5064,
      "step": 2945
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6639033671255132,
      "learning_rate": 5.413127060421416e-06,
      "loss": 0.5411,
      "step": 2946
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.3889460336608366,
      "learning_rate": 5.406658723566114e-06,
      "loss": 1.0961,
      "step": 2947
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7222690854015035,
      "learning_rate": 5.400192821530703e-06,
      "loss": 0.5241,
      "step": 2948
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.4624306307347195,
      "learning_rate": 5.393729357742594e-06,
      "loss": 1.0546,
      "step": 2949
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6836264453821213,
      "learning_rate": 5.387268335627917e-06,
      "loss": 0.6325,
      "step": 2950
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.1935973037188217,
      "learning_rate": 5.380809758611503e-06,
      "loss": 1.1262,
      "step": 2951
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5535519770003792,
      "learning_rate": 5.37435363011689e-06,
      "loss": 0.5529,
      "step": 2952
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5753685152944836,
      "learning_rate": 5.367899953566315e-06,
      "loss": 0.5362,
      "step": 2953
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2777440502054627,
      "learning_rate": 5.3614487323807195e-06,
      "loss": 1.1074,
      "step": 2954
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6141810724764605,
      "learning_rate": 5.35499996997974e-06,
      "loss": 0.572,
      "step": 2955
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.3834814952454715,
      "learning_rate": 5.348553669781708e-06,
      "loss": 1.0493,
      "step": 2956
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.4298917582600117,
      "learning_rate": 5.34210983520366e-06,
      "loss": 1.0725,
      "step": 2957
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.565765540125014,
      "learning_rate": 5.335668469661305e-06,
      "loss": 0.5614,
      "step": 2958
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7765885873982918,
      "learning_rate": 5.32922957656907e-06,
      "loss": 0.6286,
      "step": 2959
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.553959480516341,
      "learning_rate": 5.322793159340049e-06,
      "loss": 0.5979,
      "step": 2960
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.59848921699222,
      "learning_rate": 5.316359221386033e-06,
      "loss": 0.6065,
      "step": 2961
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7543064724940265,
      "learning_rate": 5.3099277661174995e-06,
      "loss": 0.6507,
      "step": 2962
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7515493259947292,
      "learning_rate": 5.3034987969436095e-06,
      "loss": 0.5931,
      "step": 2963
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5678966191331505,
      "learning_rate": 5.2970723172722024e-06,
      "loss": 0.5723,
      "step": 2964
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7184769771293322,
      "learning_rate": 5.2906483305098024e-06,
      "loss": 0.607,
      "step": 2965
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6635911586384333,
      "learning_rate": 5.284226840061611e-06,
      "loss": 0.562,
      "step": 2966
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.44588046709072,
      "learning_rate": 5.2778078493315055e-06,
      "loss": 1.1304,
      "step": 2967
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6072501225645166,
      "learning_rate": 5.271391361722043e-06,
      "loss": 0.5805,
      "step": 2968
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3692982349251883,
      "learning_rate": 5.26497738063444e-06,
      "loss": 1.1188,
      "step": 2969
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5253954794924127,
      "learning_rate": 5.2585659094686005e-06,
      "loss": 0.5882,
      "step": 2970
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.62346756134421,
      "learning_rate": 5.252156951623089e-06,
      "loss": 0.5381,
      "step": 2971
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.4621935516272178,
      "learning_rate": 5.24575051049514e-06,
      "loss": 1.1191,
      "step": 2972
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2994262032112145,
      "learning_rate": 5.239346589480656e-06,
      "loss": 1.0442,
      "step": 2973
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.755895218841248,
      "learning_rate": 5.232945191974201e-06,
      "loss": 1.15,
      "step": 2974
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.693565555922668,
      "learning_rate": 5.226546321369003e-06,
      "loss": 0.5535,
      "step": 2975
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.609889700325155,
      "learning_rate": 5.220149981056947e-06,
      "loss": 0.5415,
      "step": 2976
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.4969917782820823,
      "learning_rate": 5.213756174428585e-06,
      "loss": 1.0698,
      "step": 2977
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.54007748545202,
      "learning_rate": 5.207364904873108e-06,
      "loss": 0.5329,
      "step": 2978
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2988246517929483,
      "learning_rate": 5.200976175778391e-06,
      "loss": 1.1516,
      "step": 2979
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5969090743793504,
      "learning_rate": 5.194589990530933e-06,
      "loss": 0.6312,
      "step": 2980
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6002579616074748,
      "learning_rate": 5.188206352515902e-06,
      "loss": 0.5706,
      "step": 2981
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.727964793207014,
      "learning_rate": 5.1818252651171105e-06,
      "loss": 0.5411,
      "step": 2982
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6202077361387026,
      "learning_rate": 5.175446731717022e-06,
      "loss": 0.5516,
      "step": 2983
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8047547649335884,
      "learning_rate": 5.1690707556967414e-06,
      "loss": 0.5872,
      "step": 2984
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.483809538289634,
      "learning_rate": 5.16269734043602e-06,
      "loss": 0.5687,
      "step": 2985
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.541405351873991,
      "learning_rate": 5.156326489313255e-06,
      "loss": 0.5741,
      "step": 2986
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3600730551248907,
      "learning_rate": 5.149958205705479e-06,
      "loss": 1.1139,
      "step": 2987
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7199552889311103,
      "learning_rate": 5.143592492988369e-06,
      "loss": 0.6125,
      "step": 2988
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.499504042737135,
      "learning_rate": 5.1372293545362295e-06,
      "loss": 0.5961,
      "step": 2989
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3430488953845985,
      "learning_rate": 5.130868793722019e-06,
      "loss": 1.1352,
      "step": 2990
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.646922817279574,
      "learning_rate": 5.124510813917307e-06,
      "loss": 0.5688,
      "step": 2991
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3685256727664283,
      "learning_rate": 5.118155418492312e-06,
      "loss": 1.0985,
      "step": 2992
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5157176093095786,
      "learning_rate": 5.111802610815872e-06,
      "loss": 0.5348,
      "step": 2993
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2626294108010123,
      "learning_rate": 5.105452394255464e-06,
      "loss": 1.0432,
      "step": 2994
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.297590656419406,
      "learning_rate": 5.099104772177179e-06,
      "loss": 1.1131,
      "step": 2995
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.669964237222751,
      "learning_rate": 5.0927597479457415e-06,
      "loss": 0.6284,
      "step": 2996
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6233413363197773,
      "learning_rate": 5.086417324924497e-06,
      "loss": 0.575,
      "step": 2997
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6154330428495238,
      "learning_rate": 5.080077506475412e-06,
      "loss": 0.5638,
      "step": 2998
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.4879976182221486,
      "learning_rate": 5.0737402959590755e-06,
      "loss": 1.1051,
      "step": 2999
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.65441025091617,
      "learning_rate": 5.067405696734678e-06,
      "loss": 0.6107,
      "step": 3000
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5836773144783547,
      "learning_rate": 5.061073712160054e-06,
      "loss": 0.5059,
      "step": 3001
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5708110590283908,
      "learning_rate": 5.054744345591626e-06,
      "loss": 0.4793,
      "step": 3002
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6818148322832331,
      "learning_rate": 5.048417600384442e-06,
      "loss": 0.5548,
      "step": 3003
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7549759454803238,
      "learning_rate": 5.042093479892159e-06,
      "loss": 0.6219,
      "step": 3004
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5541577596196925,
      "learning_rate": 5.035771987467038e-06,
      "loss": 0.5497,
      "step": 3005
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6191191976151016,
      "learning_rate": 5.029453126459955e-06,
      "loss": 0.5831,
      "step": 3006
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7619410060770613,
      "learning_rate": 5.023136900220383e-06,
      "loss": 0.595,
      "step": 3007
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5807096059651693,
      "learning_rate": 5.016823312096408e-06,
      "loss": 0.5335,
      "step": 3008
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6621997742220804,
      "learning_rate": 5.010512365434698e-06,
      "loss": 0.5281,
      "step": 3009
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7131320211618772,
      "learning_rate": 5.00420406358055e-06,
      "loss": 0.5405,
      "step": 3010
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.4109440025045386,
      "learning_rate": 4.9978984098778324e-06,
      "loss": 1.0846,
      "step": 3011
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5787740062298308,
      "learning_rate": 4.9915954076690245e-06,
      "loss": 0.5159,
      "step": 3012
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.195261321099511,
      "learning_rate": 4.985295060295197e-06,
      "loss": 1.07,
      "step": 3013
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.198630260066495,
      "learning_rate": 4.97899737109601e-06,
      "loss": 1.0521,
      "step": 3014
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5891676797886924,
      "learning_rate": 4.972702343409721e-06,
      "loss": 0.6,
      "step": 3015
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.43430866297822,
      "learning_rate": 4.96640998057317e-06,
      "loss": 1.1033,
      "step": 3016
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4839348901390668,
      "learning_rate": 4.96012028592179e-06,
      "loss": 0.5895,
      "step": 3017
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.566013548767842,
      "learning_rate": 4.9538332627895944e-06,
      "loss": 0.5896,
      "step": 3018
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.8930353790045777,
      "learning_rate": 4.947548914509188e-06,
      "loss": 1.0462,
      "step": 3019
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6634580563005719,
      "learning_rate": 4.9412672444117425e-06,
      "loss": 0.558,
      "step": 3020
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8528153011836945,
      "learning_rate": 4.934988255827036e-06,
      "loss": 0.6017,
      "step": 3021
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.29531329116706,
      "learning_rate": 4.928711952083395e-06,
      "loss": 1.0656,
      "step": 3022
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6331269772826296,
      "learning_rate": 4.9224383365077435e-06,
      "loss": 0.5979,
      "step": 3023
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7261855807746642,
      "learning_rate": 4.9161674124255735e-06,
      "loss": 0.5891,
      "step": 3024
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.4874280364918175,
      "learning_rate": 4.909899183160952e-06,
      "loss": 1.1353,
      "step": 3025
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8609103124426818,
      "learning_rate": 4.903633652036517e-06,
      "loss": 0.5959,
      "step": 3026
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5119537487162955,
      "learning_rate": 4.897370822373475e-06,
      "loss": 0.5345,
      "step": 3027
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6225979254046987,
      "learning_rate": 4.891110697491603e-06,
      "loss": 0.5804,
      "step": 3028
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.5838256052049293,
      "learning_rate": 4.884853280709242e-06,
      "loss": 1.0741,
      "step": 3029
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.3461686764813923,
      "learning_rate": 4.8785985753433005e-06,
      "loss": 1.0653,
      "step": 3030
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5944924810132006,
      "learning_rate": 4.872346584709239e-06,
      "loss": 0.6205,
      "step": 3031
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6972654728681151,
      "learning_rate": 4.866097312121102e-06,
      "loss": 0.5593,
      "step": 3032
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7759240762865245,
      "learning_rate": 4.859850760891466e-06,
      "loss": 0.5668,
      "step": 3033
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6415090100774261,
      "learning_rate": 4.8536069343314825e-06,
      "loss": 0.5902,
      "step": 3034
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4658097214793298,
      "learning_rate": 4.847365835750856e-06,
      "loss": 0.5692,
      "step": 3035
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.524925272298223,
      "learning_rate": 4.841127468457839e-06,
      "loss": 0.5134,
      "step": 3036
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7492717031521778,
      "learning_rate": 4.8348918357592424e-06,
      "loss": 0.6358,
      "step": 3037
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.3173633002840592,
      "learning_rate": 4.828658940960426e-06,
      "loss": 1.1797,
      "step": 3038
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8524252178825495,
      "learning_rate": 4.8224287873653e-06,
      "loss": 0.6626,
      "step": 3039
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5296017058795346,
      "learning_rate": 4.8162013782763075e-06,
      "loss": 0.5885,
      "step": 3040
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.577254945008687,
      "learning_rate": 4.809976716994465e-06,
      "loss": 0.5593,
      "step": 3041
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5641082037957525,
      "learning_rate": 4.8037548068193005e-06,
      "loss": 0.5591,
      "step": 3042
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.163596467654239,
      "learning_rate": 4.7975356510489145e-06,
      "loss": 1.0841,
      "step": 3043
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7436894654496582,
      "learning_rate": 4.79131925297992e-06,
      "loss": 0.5965,
      "step": 3044
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7958182515814496,
      "learning_rate": 4.785105615907484e-06,
      "loss": 0.563,
      "step": 3045
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.712468944834193,
      "learning_rate": 4.778894743125306e-06,
      "loss": 0.5563,
      "step": 3046
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7549898949533818,
      "learning_rate": 4.772686637925622e-06,
      "loss": 0.5784,
      "step": 3047
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.166398180362689,
      "learning_rate": 4.766481303599198e-06,
      "loss": 1.0579,
      "step": 3048
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.862019822705897,
      "learning_rate": 4.760278743435335e-06,
      "loss": 1.0338,
      "step": 3049
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5747610642129637,
      "learning_rate": 4.754078960721862e-06,
      "loss": 0.5836,
      "step": 3050
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6170311451853951,
      "learning_rate": 4.747881958745124e-06,
      "loss": 0.5755,
      "step": 3051
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5008424347833034,
      "learning_rate": 4.74168774079002e-06,
      "loss": 0.588,
      "step": 3052
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.3279039136841617,
      "learning_rate": 4.735496310139943e-06,
      "loss": 1.0566,
      "step": 3053
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5662511669854446,
      "learning_rate": 4.729307670076826e-06,
      "loss": 0.5866,
      "step": 3054
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8446948778152747,
      "learning_rate": 4.723121823881119e-06,
      "loss": 0.5257,
      "step": 3055
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.4912425095833233,
      "learning_rate": 4.71693877483179e-06,
      "loss": 1.0916,
      "step": 3056
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.204190252042899,
      "learning_rate": 4.710758526206325e-06,
      "loss": 1.0719,
      "step": 3057
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7829187882597375,
      "learning_rate": 4.704581081280725e-06,
      "loss": 0.5389,
      "step": 3058
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5347942099140943,
      "learning_rate": 4.698406443329504e-06,
      "loss": 0.5096,
      "step": 3059
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6311798827422208,
      "learning_rate": 4.692234615625693e-06,
      "loss": 0.5671,
      "step": 3060
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.344928099848005,
      "learning_rate": 4.686065601440829e-06,
      "loss": 1.0544,
      "step": 3061
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4923426494563699,
      "learning_rate": 4.679899404044949e-06,
      "loss": 0.5471,
      "step": 3062
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4439784825426967,
      "learning_rate": 4.6737360267066235e-06,
      "loss": 0.4941,
      "step": 3063
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.480382036681612,
      "learning_rate": 4.667575472692895e-06,
      "loss": 1.0295,
      "step": 3064
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.414305420336189,
      "learning_rate": 4.661417745269332e-06,
      "loss": 1.0899,
      "step": 3065
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.240243963728451,
      "learning_rate": 4.655262847699997e-06,
      "loss": 1.1197,
      "step": 3066
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4056358736991772,
      "learning_rate": 4.649110783247452e-06,
      "loss": 0.5663,
      "step": 3067
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8179757298317132,
      "learning_rate": 4.642961555172762e-06,
      "loss": 0.592,
      "step": 3068
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8583849778829113,
      "learning_rate": 4.636815166735483e-06,
      "loss": 0.5519,
      "step": 3069
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.734330481165783,
      "learning_rate": 4.6306716211936705e-06,
      "loss": 0.5568,
      "step": 3070
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6072120703401203,
      "learning_rate": 4.6245309218038616e-06,
      "loss": 0.5719,
      "step": 3071
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7133251579349025,
      "learning_rate": 4.618393071821107e-06,
      "loss": 0.5841,
      "step": 3072
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.527129494726055,
      "learning_rate": 4.6122580744989185e-06,
      "loss": 0.5779,
      "step": 3073
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.3390800430321685,
      "learning_rate": 4.606125933089327e-06,
      "loss": 1.0767,
      "step": 3074
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6162901644032082,
      "learning_rate": 4.599996650842822e-06,
      "loss": 0.5534,
      "step": 3075
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5628093133506873,
      "learning_rate": 4.59387023100839e-06,
      "loss": 0.5573,
      "step": 3076
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7420632894864172,
      "learning_rate": 4.587746676833502e-06,
      "loss": 0.5091,
      "step": 3077
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5855281125312901,
      "learning_rate": 4.581625991564107e-06,
      "loss": 0.6484,
      "step": 3078
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2321659817779373,
      "learning_rate": 4.575508178444631e-06,
      "loss": 1.1039,
      "step": 3079
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.490232397133623,
      "learning_rate": 4.5693932407179806e-06,
      "loss": 0.5765,
      "step": 3080
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6064378845728562,
      "learning_rate": 4.5632811816255425e-06,
      "loss": 0.5224,
      "step": 3081
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6277660229146638,
      "learning_rate": 4.55717200440716e-06,
      "loss": 0.5305,
      "step": 3082
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.3539351381206037,
      "learning_rate": 4.551065712301177e-06,
      "loss": 1.035,
      "step": 3083
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2934212885545078,
      "learning_rate": 4.544962308544376e-06,
      "loss": 1.1207,
      "step": 3084
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7310243596422175,
      "learning_rate": 4.53886179637204e-06,
      "loss": 0.5286,
      "step": 3085
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7824398624812887,
      "learning_rate": 4.532764179017894e-06,
      "loss": 0.5437,
      "step": 3086
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5399883941002306,
      "learning_rate": 4.526669459714142e-06,
      "loss": 0.5425,
      "step": 3087
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5489750201333674,
      "learning_rate": 4.5205776416914475e-06,
      "loss": 0.5403,
      "step": 3088
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5423797438543099,
      "learning_rate": 4.514488728178936e-06,
      "loss": 0.5661,
      "step": 3089
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5710950781327746,
      "learning_rate": 4.508402722404196e-06,
      "loss": 0.545,
      "step": 3090
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.3576953273271304,
      "learning_rate": 4.502319627593272e-06,
      "loss": 1.0829,
      "step": 3091
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5795361510104213,
      "learning_rate": 4.496239446970671e-06,
      "loss": 0.5565,
      "step": 3092
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1373910427751266,
      "learning_rate": 4.49016218375934e-06,
      "loss": 1.0659,
      "step": 3093
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.480828024294459,
      "learning_rate": 4.4840878411807045e-06,
      "loss": 0.5326,
      "step": 3094
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1964449290288464,
      "learning_rate": 4.478016422454615e-06,
      "loss": 1.0941,
      "step": 3095
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.306204821016193,
      "learning_rate": 4.471947930799391e-06,
      "loss": 1.0682,
      "step": 3096
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6797212054210182,
      "learning_rate": 4.465882369431792e-06,
      "loss": 0.5432,
      "step": 3097
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.302132890230828,
      "learning_rate": 4.459819741567027e-06,
      "loss": 1.0648,
      "step": 3098
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7198136361840355,
      "learning_rate": 4.453760050418752e-06,
      "loss": 0.5126,
      "step": 3099
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.644867482452352,
      "learning_rate": 4.447703299199059e-06,
      "loss": 0.5717,
      "step": 3100
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5026837947653389,
      "learning_rate": 4.441649491118494e-06,
      "loss": 0.5275,
      "step": 3101
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6239149529450194,
      "learning_rate": 4.435598629386021e-06,
      "loss": 0.5841,
      "step": 3102
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.70650603134013,
      "learning_rate": 4.429550717209075e-06,
      "loss": 0.5872,
      "step": 3103
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6161284187982483,
      "learning_rate": 4.42350575779349e-06,
      "loss": 0.6112,
      "step": 3104
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.492776510609562,
      "learning_rate": 4.417463754343572e-06,
      "loss": 1.1203,
      "step": 3105
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.6543156754379527,
      "learning_rate": 4.411424710062029e-06,
      "loss": 0.9885,
      "step": 3106
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.639147707842652,
      "learning_rate": 4.405388628150018e-06,
      "loss": 0.5487,
      "step": 3107
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.675629370974575,
      "learning_rate": 4.399355511807121e-06,
      "loss": 0.577,
      "step": 3108
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.549834366870419,
      "learning_rate": 4.393325364231347e-06,
      "loss": 0.5817,
      "step": 3109
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6080426308351694,
      "learning_rate": 4.387298188619132e-06,
      "loss": 0.522,
      "step": 3110
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7812932890756195,
      "learning_rate": 4.38127398816534e-06,
      "loss": 0.5894,
      "step": 3111
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6134049571189202,
      "learning_rate": 4.375252766063256e-06,
      "loss": 0.5742,
      "step": 3112
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6077651148484244,
      "learning_rate": 4.369234525504574e-06,
      "loss": 0.5089,
      "step": 3113
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.434006723321301,
      "learning_rate": 4.363219269679435e-06,
      "loss": 0.5695,
      "step": 3114
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.2112338860090626,
      "learning_rate": 4.3572070017763665e-06,
      "loss": 1.0199,
      "step": 3115
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.512341815539975,
      "learning_rate": 4.351197724982342e-06,
      "loss": 0.5423,
      "step": 3116
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5594822383803346,
      "learning_rate": 4.345191442482724e-06,
      "loss": 0.6275,
      "step": 3117
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.690699163596716,
      "learning_rate": 4.3391881574613e-06,
      "loss": 1.0913,
      "step": 3118
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5396457396195549,
      "learning_rate": 4.333187873100271e-06,
      "loss": 0.5469,
      "step": 3119
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6806946591694853,
      "learning_rate": 4.327190592580242e-06,
      "loss": 0.5508,
      "step": 3120
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3773147101388594,
      "learning_rate": 4.321196319080225e-06,
      "loss": 1.0577,
      "step": 3121
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6843998973061391,
      "learning_rate": 4.315205055777644e-06,
      "loss": 0.4993,
      "step": 3122
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6646937789086544,
      "learning_rate": 4.3092168058483256e-06,
      "loss": 0.582,
      "step": 3123
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5597419714645655,
      "learning_rate": 4.303231572466485e-06,
      "loss": 0.5356,
      "step": 3124
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6355575343242217,
      "learning_rate": 4.297249358804769e-06,
      "loss": 0.5899,
      "step": 3125
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.083829636801358,
      "learning_rate": 4.291270168034187e-06,
      "loss": 1.1157,
      "step": 3126
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8615121331917666,
      "learning_rate": 4.285294003324182e-06,
      "loss": 0.6057,
      "step": 3127
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.605204172702178,
      "learning_rate": 4.2793208678425626e-06,
      "loss": 0.5885,
      "step": 3128
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6335738278412661,
      "learning_rate": 4.2733507647555485e-06,
      "loss": 0.5972,
      "step": 3129
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8448997319767282,
      "learning_rate": 4.267383697227747e-06,
      "loss": 0.5784,
      "step": 3130
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.642710271863244,
      "learning_rate": 4.26141966842216e-06,
      "loss": 0.5218,
      "step": 3131
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5135309423915926,
      "learning_rate": 4.255458681500176e-06,
      "loss": 0.6377,
      "step": 3132
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8698770942308536,
      "learning_rate": 4.249500739621564e-06,
      "loss": 0.5957,
      "step": 3133
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.6151405565428285,
      "learning_rate": 4.243545845944497e-06,
      "loss": 1.0271,
      "step": 3134
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6260182346697403,
      "learning_rate": 4.237594003625508e-06,
      "loss": 0.6202,
      "step": 3135
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.2104870935628114,
      "learning_rate": 4.231645215819541e-06,
      "loss": 1.0589,
      "step": 3136
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.310592411640427,
      "learning_rate": 4.225699485679893e-06,
      "loss": 1.0223,
      "step": 3137
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.774589325989087,
      "learning_rate": 4.219756816358257e-06,
      "loss": 0.5554,
      "step": 3138
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.576889715040213,
      "learning_rate": 4.213817211004699e-06,
      "loss": 0.5193,
      "step": 3139
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2279564980036546,
      "learning_rate": 4.207880672767663e-06,
      "loss": 1.0842,
      "step": 3140
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.686028484703405,
      "learning_rate": 4.201947204793964e-06,
      "loss": 0.5875,
      "step": 3141
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.7927069580604753,
      "learning_rate": 4.1960168102287895e-06,
      "loss": 1.1648,
      "step": 3142
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6273682457010068,
      "learning_rate": 4.190089492215704e-06,
      "loss": 0.5281,
      "step": 3143
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4633296244109053,
      "learning_rate": 4.184165253896626e-06,
      "loss": 0.5519,
      "step": 3144
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6754243510165117,
      "learning_rate": 4.178244098411865e-06,
      "loss": 0.5376,
      "step": 3145
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.666328341219734,
      "learning_rate": 4.172326028900072e-06,
      "loss": 1.0542,
      "step": 3146
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4900583870323836,
      "learning_rate": 4.166411048498285e-06,
      "loss": 0.5283,
      "step": 3147
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.78621132978341,
      "learning_rate": 4.160499160341882e-06,
      "loss": 0.653,
      "step": 3148
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5255779101482336,
      "learning_rate": 4.154590367564619e-06,
      "loss": 0.5411,
      "step": 3149
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5944190825125193,
      "learning_rate": 4.148684673298603e-06,
      "loss": 0.566,
      "step": 3150
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7352701826309331,
      "learning_rate": 4.1427820806743015e-06,
      "loss": 0.5972,
      "step": 3151
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3952448074836483,
      "learning_rate": 4.136882592820537e-06,
      "loss": 0.5222,
      "step": 3152
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4543401198319525,
      "learning_rate": 4.130986212864486e-06,
      "loss": 0.5183,
      "step": 3153
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6696887902334747,
      "learning_rate": 4.125092943931682e-06,
      "loss": 0.4982,
      "step": 3154
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6764582113256516,
      "learning_rate": 4.119202789145994e-06,
      "loss": 0.5871,
      "step": 3155
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7175540483003442,
      "learning_rate": 4.1133157516296675e-06,
      "loss": 0.5675,
      "step": 3156
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.679559178378989,
      "learning_rate": 4.107431834503263e-06,
      "loss": 0.5411,
      "step": 3157
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5725963782002361,
      "learning_rate": 4.101551040885719e-06,
      "loss": 0.5247,
      "step": 3158
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6906196674528526,
      "learning_rate": 4.095673373894293e-06,
      "loss": 0.6053,
      "step": 3159
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5073037311247781,
      "learning_rate": 4.089798836644596e-06,
      "loss": 0.5863,
      "step": 3160
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.598206021402074,
      "learning_rate": 4.083927432250583e-06,
      "loss": 0.5562,
      "step": 3161
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6533872622606745,
      "learning_rate": 4.078059163824543e-06,
      "loss": 0.6094,
      "step": 3162
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6926862531609825,
      "learning_rate": 4.072194034477105e-06,
      "loss": 0.5544,
      "step": 3163
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6375989947398601,
      "learning_rate": 4.066332047317226e-06,
      "loss": 0.5164,
      "step": 3164
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5552105854450946,
      "learning_rate": 4.060473205452218e-06,
      "loss": 0.5384,
      "step": 3165
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.649148410435294,
      "learning_rate": 4.054617511987696e-06,
      "loss": 0.5446,
      "step": 3166
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.585906435062024,
      "learning_rate": 4.04876497002764e-06,
      "loss": 0.6449,
      "step": 3167
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.424502968964545,
      "learning_rate": 4.042915582674325e-06,
      "loss": 1.0736,
      "step": 3168
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8705155943827032,
      "learning_rate": 4.0370693530283866e-06,
      "loss": 0.6613,
      "step": 3169
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5217615542502654,
      "learning_rate": 4.031226284188759e-06,
      "loss": 0.5441,
      "step": 3170
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.3982519456150326,
      "learning_rate": 4.0253863792527185e-06,
      "loss": 1.0673,
      "step": 3171
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6429631188109388,
      "learning_rate": 4.019549641315854e-06,
      "loss": 0.6411,
      "step": 3172
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.732875418473094,
      "learning_rate": 4.013716073472084e-06,
      "loss": 0.6012,
      "step": 3173
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5227906106121303,
      "learning_rate": 4.007885678813645e-06,
      "loss": 0.5507,
      "step": 3174
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.668194868200476,
      "learning_rate": 4.002058460431079e-06,
      "loss": 0.5428,
      "step": 3175
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.716920806202553,
      "learning_rate": 3.996234421413267e-06,
      "loss": 0.5528,
      "step": 3176
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.5285276617807524,
      "learning_rate": 3.99041356484738e-06,
      "loss": 1.0419,
      "step": 3177
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6015005606350756,
      "learning_rate": 3.984595893818927e-06,
      "loss": 0.4916,
      "step": 3178
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6743450866941882,
      "learning_rate": 3.978781411411705e-06,
      "loss": 0.582,
      "step": 3179
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6990673753697683,
      "learning_rate": 3.9729701207078366e-06,
      "loss": 0.4965,
      "step": 3180
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6065450168473838,
      "learning_rate": 3.967162024787746e-06,
      "loss": 0.6154,
      "step": 3181
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7498905295581908,
      "learning_rate": 3.9613571267301665e-06,
      "loss": 0.5316,
      "step": 3182
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5633614982013788,
      "learning_rate": 3.9555554296121345e-06,
      "loss": 0.5133,
      "step": 3183
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6069235083188929,
      "learning_rate": 3.949756936508991e-06,
      "loss": 0.5771,
      "step": 3184
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5349139612569926,
      "learning_rate": 3.943961650494379e-06,
      "loss": 0.5957,
      "step": 3185
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6014611777740657,
      "learning_rate": 3.938169574640233e-06,
      "loss": 0.5496,
      "step": 3186
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.692611609705679,
      "learning_rate": 3.9323807120168044e-06,
      "loss": 0.58,
      "step": 3187
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2960469737813076,
      "learning_rate": 3.926595065692619e-06,
      "loss": 1.1053,
      "step": 3188
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.655033097940357,
      "learning_rate": 3.920812638734522e-06,
      "loss": 0.5136,
      "step": 3189
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5954679101630589,
      "learning_rate": 3.915033434207627e-06,
      "loss": 0.5551,
      "step": 3190
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.4619325375912053,
      "learning_rate": 3.909257455175357e-06,
      "loss": 1.1161,
      "step": 3191
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5968806298953229,
      "learning_rate": 3.903484704699418e-06,
      "loss": 0.5454,
      "step": 3192
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6835696312595303,
      "learning_rate": 3.897715185839809e-06,
      "loss": 0.5791,
      "step": 3193
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2650363042676407,
      "learning_rate": 3.891948901654815e-06,
      "loss": 1.0922,
      "step": 3194
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8005959502601354,
      "learning_rate": 3.8861858552009935e-06,
      "loss": 0.5239,
      "step": 3195
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7890599748971519,
      "learning_rate": 3.880426049533212e-06,
      "loss": 0.5048,
      "step": 3196
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.3305245904557164,
      "learning_rate": 3.874669487704592e-06,
      "loss": 1.1207,
      "step": 3197
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7062753046673818,
      "learning_rate": 3.868916172766562e-06,
      "loss": 0.5793,
      "step": 3198
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8311784710630619,
      "learning_rate": 3.863166107768801e-06,
      "loss": 0.5903,
      "step": 3199
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5996335424205121,
      "learning_rate": 3.857419295759296e-06,
      "loss": 0.5651,
      "step": 3200
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7153745789527084,
      "learning_rate": 3.851675739784282e-06,
      "loss": 0.5929,
      "step": 3201
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6730009429801984,
      "learning_rate": 3.8459354428882855e-06,
      "loss": 0.5938,
      "step": 3202
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.3724009624440368,
      "learning_rate": 3.840198408114097e-06,
      "loss": 1.077,
      "step": 3203
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7997984669409157,
      "learning_rate": 3.834464638502784e-06,
      "loss": 0.6348,
      "step": 3204
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.4895672155127566,
      "learning_rate": 3.828734137093683e-06,
      "loss": 1.12,
      "step": 3205
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8084448213383633,
      "learning_rate": 3.823006906924381e-06,
      "loss": 0.6462,
      "step": 3206
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6555822285012851,
      "learning_rate": 3.817282951030764e-06,
      "loss": 0.6227,
      "step": 3207
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.3890288665987094,
      "learning_rate": 3.811562272446948e-06,
      "loss": 1.1478,
      "step": 3208
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5440796368283825,
      "learning_rate": 3.80584487420534e-06,
      "loss": 0.5161,
      "step": 3209
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.593630185348655,
      "learning_rate": 3.8001307593365822e-06,
      "loss": 0.4955,
      "step": 3210
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.744026764954538,
      "learning_rate": 3.7944199308696052e-06,
      "loss": 0.6582,
      "step": 3211
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6222562503734645,
      "learning_rate": 3.7887123918315682e-06,
      "loss": 0.4663,
      "step": 3212
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2620431003043753,
      "learning_rate": 3.7830081452479073e-06,
      "loss": 1.064,
      "step": 3213
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5081771579074565,
      "learning_rate": 3.7773071941423044e-06,
      "loss": 0.5377,
      "step": 3214
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7754684537009557,
      "learning_rate": 3.771609541536698e-06,
      "loss": 0.6073,
      "step": 3215
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2382859107756055,
      "learning_rate": 3.76591519045128e-06,
      "loss": 1.0872,
      "step": 3216
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6943813710067241,
      "learning_rate": 3.7602241439044796e-06,
      "loss": 0.6121,
      "step": 3217
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.691332126446301,
      "learning_rate": 3.754536404912996e-06,
      "loss": 0.5549,
      "step": 3218
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2590140742715423,
      "learning_rate": 3.7488519764917497e-06,
      "loss": 1.074,
      "step": 3219
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.4471248272448762,
      "learning_rate": 3.743170861653934e-06,
      "loss": 1.0442,
      "step": 3220
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5829489645150534,
      "learning_rate": 3.737493063410962e-06,
      "loss": 0.536,
      "step": 3221
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.578825472988789,
      "learning_rate": 3.7318185847724996e-06,
      "loss": 0.5134,
      "step": 3222
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5490918490609265,
      "learning_rate": 3.7261474287464527e-06,
      "loss": 0.564,
      "step": 3223
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6575473524070885,
      "learning_rate": 3.720479598338964e-06,
      "loss": 0.5542,
      "step": 3224
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6174792100268718,
      "learning_rate": 3.7148150965544205e-06,
      "loss": 0.5734,
      "step": 3225
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.333582477628512,
      "learning_rate": 3.7091539263954246e-06,
      "loss": 1.1267,
      "step": 3226
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8004991776690904,
      "learning_rate": 3.7034960908628426e-06,
      "loss": 0.5971,
      "step": 3227
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6093555640755812,
      "learning_rate": 3.697841592955742e-06,
      "loss": 0.5136,
      "step": 3228
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6585541953897485,
      "learning_rate": 3.692190435671452e-06,
      "loss": 0.5336,
      "step": 3229
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6693047348664272,
      "learning_rate": 3.6865426220055002e-06,
      "loss": 0.5708,
      "step": 3230
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.509560606000876,
      "learning_rate": 3.680898154951671e-06,
      "loss": 0.5476,
      "step": 3231
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4727021848490474,
      "learning_rate": 3.67525703750195e-06,
      "loss": 0.4959,
      "step": 3232
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.3291707587606885,
      "learning_rate": 3.669619272646563e-06,
      "loss": 1.0408,
      "step": 3233
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.304566166364097,
      "learning_rate": 3.6639848633739526e-06,
      "loss": 1.0895,
      "step": 3234
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.645619355464505,
      "learning_rate": 3.658353812670784e-06,
      "loss": 0.5743,
      "step": 3235
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6095376395795549,
      "learning_rate": 3.652726123521945e-06,
      "loss": 0.572,
      "step": 3236
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.88163071356798,
      "learning_rate": 3.6471017989105284e-06,
      "loss": 0.5896,
      "step": 3237
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7535114651960555,
      "learning_rate": 3.6414808418178684e-06,
      "loss": 0.587,
      "step": 3238
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.538519749233654,
      "learning_rate": 3.635863255223484e-06,
      "loss": 0.5574,
      "step": 3239
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7045116393750614,
      "learning_rate": 3.6302490421051373e-06,
      "loss": 0.6158,
      "step": 3240
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.3950139836083273,
      "learning_rate": 3.624638205438774e-06,
      "loss": 1.0765,
      "step": 3241
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2406790810854953,
      "learning_rate": 3.6190307481985786e-06,
      "loss": 1.0653,
      "step": 3242
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.3281394069308483,
      "learning_rate": 3.6134266733569177e-06,
      "loss": 1.0578,
      "step": 3243
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7909885969066888,
      "learning_rate": 3.607825983884381e-06,
      "loss": 0.5833,
      "step": 3244
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5740829907256022,
      "learning_rate": 3.602228682749759e-06,
      "loss": 0.5812,
      "step": 3245
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.465758641691556,
      "learning_rate": 3.5966347729200467e-06,
      "loss": 0.5318,
      "step": 3246
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8908113074984008,
      "learning_rate": 3.5910442573604455e-06,
      "loss": 0.5782,
      "step": 3247
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.3872399190928792,
      "learning_rate": 3.585457139034342e-06,
      "loss": 1.0988,
      "step": 3248
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5511335039145866,
      "learning_rate": 3.5798734209033483e-06,
      "loss": 0.6151,
      "step": 3249
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2887194629335963,
      "learning_rate": 3.5742931059272445e-06,
      "loss": 1.0577,
      "step": 3250
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.41440998182126,
      "learning_rate": 3.568716197064036e-06,
      "loss": 0.4785,
      "step": 3251
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5355573978738035,
      "learning_rate": 3.5631426972698948e-06,
      "loss": 0.5511,
      "step": 3252
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7247305325410092,
      "learning_rate": 3.5575726094992112e-06,
      "loss": 0.5886,
      "step": 3253
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5526262711477226,
      "learning_rate": 3.5520059367045467e-06,
      "loss": 0.5329,
      "step": 3254
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6362269434573815,
      "learning_rate": 3.5464426818366638e-06,
      "loss": 0.6143,
      "step": 3255
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5269540743019385,
      "learning_rate": 3.540882847844512e-06,
      "loss": 0.5394,
      "step": 3256
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4293210552524351,
      "learning_rate": 3.535326437675218e-06,
      "loss": 0.5973,
      "step": 3257
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6420818701496902,
      "learning_rate": 3.529773454274115e-06,
      "loss": 0.569,
      "step": 3258
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5502983094559728,
      "learning_rate": 3.524223900584689e-06,
      "loss": 0.5305,
      "step": 3259
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6409460502481181,
      "learning_rate": 3.5186777795486426e-06,
      "loss": 0.5289,
      "step": 3260
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7194488432460615,
      "learning_rate": 3.513135094105826e-06,
      "loss": 0.5815,
      "step": 3261
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9424512572375443,
      "learning_rate": 3.507595847194297e-06,
      "loss": 0.6323,
      "step": 3262
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3503656768263148,
      "learning_rate": 3.5020600417502693e-06,
      "loss": 0.5272,
      "step": 3263
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.345565755911526,
      "learning_rate": 3.4965276807081407e-06,
      "loss": 1.0786,
      "step": 3264
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2395990842889892,
      "learning_rate": 3.4909987670004853e-06,
      "loss": 1.1467,
      "step": 3265
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7967075928258098,
      "learning_rate": 3.485473303558046e-06,
      "loss": 0.5944,
      "step": 3266
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6197571746823207,
      "learning_rate": 3.4799512933097413e-06,
      "loss": 0.5898,
      "step": 3267
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.14629023493625,
      "learning_rate": 3.4744327391826483e-06,
      "loss": 1.0687,
      "step": 3268
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5554877256211819,
      "learning_rate": 3.468917644102031e-06,
      "loss": 0.4881,
      "step": 3269
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.343612303851503,
      "learning_rate": 3.463406010991297e-06,
      "loss": 1.1313,
      "step": 3270
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7010556234834469,
      "learning_rate": 3.4578978427720442e-06,
      "loss": 0.5402,
      "step": 3271
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.775221614355814,
      "learning_rate": 3.4523931423640057e-06,
      "loss": 0.4812,
      "step": 3272
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.481579696995778,
      "learning_rate": 3.446891912685105e-06,
      "loss": 0.5662,
      "step": 3273
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5561745950123147,
      "learning_rate": 3.441394156651402e-06,
      "loss": 0.5649,
      "step": 3274
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5478677875007316,
      "learning_rate": 3.4358998771771267e-06,
      "loss": 0.584,
      "step": 3275
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6745825088464061,
      "learning_rate": 3.430409077174668e-06,
      "loss": 0.5245,
      "step": 3276
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4778097564429344,
      "learning_rate": 3.4249217595545638e-06,
      "loss": 0.5648,
      "step": 3277
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.4797082771241206,
      "learning_rate": 3.419437927225515e-06,
      "loss": 1.1421,
      "step": 3278
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4926895862801777,
      "learning_rate": 3.4139575830943584e-06,
      "loss": 0.5032,
      "step": 3279
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6205138794237068,
      "learning_rate": 3.408480730066106e-06,
      "loss": 0.5643,
      "step": 3280
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5660519959994301,
      "learning_rate": 3.403007371043893e-06,
      "loss": 0.5994,
      "step": 3281
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8742845015424334,
      "learning_rate": 3.3975375089290297e-06,
      "loss": 0.6216,
      "step": 3282
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.645108044331125,
      "learning_rate": 3.392071146620943e-06,
      "loss": 0.5439,
      "step": 3283
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6665146911996107,
      "learning_rate": 3.386608287017236e-06,
      "loss": 0.6035,
      "step": 3284
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6206725608887766,
      "learning_rate": 3.3811489330136284e-06,
      "loss": 0.546,
      "step": 3285
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7169936155769896,
      "learning_rate": 3.375693087503996e-06,
      "loss": 0.5934,
      "step": 3286
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.4356772045130506,
      "learning_rate": 3.3702407533803536e-06,
      "loss": 1.0626,
      "step": 3287
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.705470099079454,
      "learning_rate": 3.3647919335328462e-06,
      "loss": 0.5705,
      "step": 3288
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3895444957902088,
      "learning_rate": 3.359346630849771e-06,
      "loss": 1.0299,
      "step": 3289
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7564543203918213,
      "learning_rate": 3.3539048482175418e-06,
      "loss": 0.588,
      "step": 3290
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.71905856206221,
      "learning_rate": 3.34846658852073e-06,
      "loss": 0.5842,
      "step": 3291
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.804443110774683,
      "learning_rate": 3.3430318546420113e-06,
      "loss": 0.6044,
      "step": 3292
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6629697858831765,
      "learning_rate": 3.337600649462224e-06,
      "loss": 0.6403,
      "step": 3293
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.726341729688377,
      "learning_rate": 3.332172975860306e-06,
      "loss": 0.5832,
      "step": 3294
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6732254110231022,
      "learning_rate": 3.3267488367133427e-06,
      "loss": 0.5524,
      "step": 3295
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7793114366683882,
      "learning_rate": 3.3213282348965382e-06,
      "loss": 0.6327,
      "step": 3296
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5118997606640248,
      "learning_rate": 3.315911173283225e-06,
      "loss": 0.5462,
      "step": 3297
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6513421627728953,
      "learning_rate": 3.3104976547448597e-06,
      "loss": 0.5254,
      "step": 3298
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6921592903180722,
      "learning_rate": 3.305087682151008e-06,
      "loss": 0.5417,
      "step": 3299
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6683576448334145,
      "learning_rate": 3.2996812583693804e-06,
      "loss": 0.545,
      "step": 3300
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5320716212671008,
      "learning_rate": 3.29427838626578e-06,
      "loss": 0.5269,
      "step": 3301
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.350161254926854,
      "learning_rate": 3.2888790687041506e-06,
      "loss": 1.0664,
      "step": 3302
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7008749486277668,
      "learning_rate": 3.283483308546529e-06,
      "loss": 0.5205,
      "step": 3303
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6212030609539299,
      "learning_rate": 3.2780911086530907e-06,
      "loss": 0.5932,
      "step": 3304
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5371513666066072,
      "learning_rate": 3.272702471882102e-06,
      "loss": 0.5569,
      "step": 3305
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.526849194879747,
      "learning_rate": 3.267317401089953e-06,
      "loss": 0.5492,
      "step": 3306
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6997383438940692,
      "learning_rate": 3.26193589913114e-06,
      "loss": 0.503,
      "step": 3307
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.747205248384994,
      "learning_rate": 3.2565579688582683e-06,
      "loss": 0.5693,
      "step": 3308
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7464816075994885,
      "learning_rate": 3.2511836131220533e-06,
      "loss": 0.4927,
      "step": 3309
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5875106626288857,
      "learning_rate": 3.2458128347713002e-06,
      "loss": 0.5147,
      "step": 3310
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6215313330744492,
      "learning_rate": 3.2404456366529445e-06,
      "loss": 0.5693,
      "step": 3311
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.618922614372672,
      "learning_rate": 3.235082021611995e-06,
      "loss": 1.1553,
      "step": 3312
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5012226036147531,
      "learning_rate": 3.229721992491589e-06,
      "loss": 0.5016,
      "step": 3313
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.551321235457095,
      "learning_rate": 3.2243655521329354e-06,
      "loss": 0.5272,
      "step": 3314
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.3080162609084174,
      "learning_rate": 3.219012703375368e-06,
      "loss": 1.0879,
      "step": 3315
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7005234795435833,
      "learning_rate": 3.2136634490562948e-06,
      "loss": 0.5501,
      "step": 3316
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7939065851230993,
      "learning_rate": 3.2083177920112274e-06,
      "loss": 0.5984,
      "step": 3317
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6549517190490373,
      "learning_rate": 3.202975735073777e-06,
      "loss": 0.556,
      "step": 3318
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6727978417445888,
      "learning_rate": 3.197637281075626e-06,
      "loss": 0.5058,
      "step": 3319
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.4402869228609845,
      "learning_rate": 3.1923024328465767e-06,
      "loss": 1.1037,
      "step": 3320
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.329777101158996,
      "learning_rate": 3.1869711932144896e-06,
      "loss": 1.1055,
      "step": 3321
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6520946592423904,
      "learning_rate": 3.181643565005341e-06,
      "loss": 0.6278,
      "step": 3322
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.625631809751002,
      "learning_rate": 3.1763195510431642e-06,
      "loss": 0.5728,
      "step": 3323
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.339467272973785,
      "learning_rate": 3.170999154150105e-06,
      "loss": 1.0577,
      "step": 3324
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4676462697035542,
      "learning_rate": 3.1656823771463672e-06,
      "loss": 0.5501,
      "step": 3325
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6901859905477425,
      "learning_rate": 3.1603692228502515e-06,
      "loss": 0.5948,
      "step": 3326
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7589029008022925,
      "learning_rate": 3.155059694078133e-06,
      "loss": 0.5989,
      "step": 3327
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5102431079798502,
      "learning_rate": 3.1497537936444645e-06,
      "loss": 0.539,
      "step": 3328
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.3259305581798513,
      "learning_rate": 3.144451524361779e-06,
      "loss": 1.0852,
      "step": 3329
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.3310904073728893,
      "learning_rate": 3.1391528890406754e-06,
      "loss": 1.0128,
      "step": 3330
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5930146809725108,
      "learning_rate": 3.1338578904898422e-06,
      "loss": 0.5289,
      "step": 3331
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.33081583228783,
      "learning_rate": 3.1285665315160208e-06,
      "loss": 1.097,
      "step": 3332
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6657333070545388,
      "learning_rate": 3.1232788149240446e-06,
      "loss": 0.4909,
      "step": 3333
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6058558056291996,
      "learning_rate": 3.1179947435167936e-06,
      "loss": 0.5229,
      "step": 3334
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.210056119297554,
      "learning_rate": 3.1127143200952393e-06,
      "loss": 1.1197,
      "step": 3335
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.458418131152111,
      "learning_rate": 3.1074375474583986e-06,
      "loss": 1.1224,
      "step": 3336
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7076367313411789,
      "learning_rate": 3.1021644284033637e-06,
      "loss": 0.6266,
      "step": 3337
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5411504901492312,
      "learning_rate": 3.096894965725289e-06,
      "loss": 0.5406,
      "step": 3338
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7543936271186258,
      "learning_rate": 3.09162916221739e-06,
      "loss": 0.5954,
      "step": 3339
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6591541853991214,
      "learning_rate": 3.0863670206709462e-06,
      "loss": 0.602,
      "step": 3340
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5917154512677265,
      "learning_rate": 3.081108543875283e-06,
      "loss": 0.664,
      "step": 3341
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.670910483211894,
      "learning_rate": 3.0758537346178052e-06,
      "loss": 0.5953,
      "step": 3342
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.3923492008794587,
      "learning_rate": 3.070602595683948e-06,
      "loss": 1.0832,
      "step": 3343
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5038807497945905,
      "learning_rate": 3.065355129857227e-06,
      "loss": 0.5125,
      "step": 3344
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5679193547407806,
      "learning_rate": 3.060111339919184e-06,
      "loss": 0.5329,
      "step": 3345
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5611458731309522,
      "learning_rate": 3.054871228649441e-06,
      "loss": 0.5417,
      "step": 3346
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.308656271337314,
      "learning_rate": 3.0496347988256435e-06,
      "loss": 1.1063,
      "step": 3347
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.543325442376783,
      "learning_rate": 3.044402053223502e-06,
      "loss": 1.0709,
      "step": 3348
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8738522221587677,
      "learning_rate": 3.0391729946167714e-06,
      "loss": 0.5588,
      "step": 3349
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6343476887071153,
      "learning_rate": 3.0339476257772405e-06,
      "loss": 0.5624,
      "step": 3350
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.313579694969755,
      "learning_rate": 3.0287259494747657e-06,
      "loss": 1.0216,
      "step": 3351
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6747601646732555,
      "learning_rate": 3.023507968477217e-06,
      "loss": 0.6157,
      "step": 3352
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6975280766598562,
      "learning_rate": 3.018293685550536e-06,
      "loss": 0.5957,
      "step": 3353
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5205529387577816,
      "learning_rate": 3.013083103458676e-06,
      "loss": 0.5525,
      "step": 3354
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5704375331564642,
      "learning_rate": 3.0078762249636526e-06,
      "loss": 0.5363,
      "step": 3355
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.4072527295301063,
      "learning_rate": 3.002673052825499e-06,
      "loss": 1.0942,
      "step": 3356
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6464691098456032,
      "learning_rate": 2.9974735898022934e-06,
      "loss": 0.5943,
      "step": 3357
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.728416585595539,
      "learning_rate": 2.9922778386501473e-06,
      "loss": 0.5973,
      "step": 3358
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.634558164339204,
      "learning_rate": 2.987085802123203e-06,
      "loss": 0.5677,
      "step": 3359
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6302594003413893,
      "learning_rate": 2.981897482973638e-06,
      "loss": 0.5203,
      "step": 3360
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5604148351802556,
      "learning_rate": 2.9767128839516467e-06,
      "loss": 0.5006,
      "step": 3361
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5885764817285062,
      "learning_rate": 2.971532007805472e-06,
      "loss": 0.6124,
      "step": 3362
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6907489540294018,
      "learning_rate": 2.9663548572813603e-06,
      "loss": 0.6036,
      "step": 3363
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7744235509445445,
      "learning_rate": 2.961181435123607e-06,
      "loss": 0.5161,
      "step": 3364
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5723698189399256,
      "learning_rate": 2.956011744074505e-06,
      "loss": 0.503,
      "step": 3365
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.5208717796515034,
      "learning_rate": 2.9508457868743977e-06,
      "loss": 1.1176,
      "step": 3366
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5403579599335382,
      "learning_rate": 2.9456835662616255e-06,
      "loss": 0.5526,
      "step": 3367
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6669019215587657,
      "learning_rate": 2.94052508497256e-06,
      "loss": 0.6353,
      "step": 3368
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4169232555520375,
      "learning_rate": 2.935370345741588e-06,
      "loss": 0.5807,
      "step": 3369
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5784964813480202,
      "learning_rate": 2.9302193513011133e-06,
      "loss": 0.5888,
      "step": 3370
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6091496777561995,
      "learning_rate": 2.9250721043815576e-06,
      "loss": 0.5317,
      "step": 3371
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7017106918935858,
      "learning_rate": 2.919928607711344e-06,
      "loss": 0.5579,
      "step": 3372
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5864546822506809,
      "learning_rate": 2.9147888640169287e-06,
      "loss": 0.6608,
      "step": 3373
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8379485221523943,
      "learning_rate": 2.9096528760227526e-06,
      "loss": 0.5886,
      "step": 3374
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6754732446189435,
      "learning_rate": 2.9045206464512933e-06,
      "loss": 0.5937,
      "step": 3375
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.516944839805023,
      "learning_rate": 2.8993921780230096e-06,
      "loss": 0.5406,
      "step": 3376
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6675852948138612,
      "learning_rate": 2.8942674734563924e-06,
      "loss": 0.5337,
      "step": 3377
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.671329069231588,
      "learning_rate": 2.889146535467915e-06,
      "loss": 0.6019,
      "step": 3378
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7905836960055417,
      "learning_rate": 2.8840293667720653e-06,
      "loss": 0.577,
      "step": 3379
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.578111094248838,
      "learning_rate": 2.8789159700813374e-06,
      "loss": 0.5674,
      "step": 3380
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5547654847982424,
      "learning_rate": 2.8738063481062083e-06,
      "loss": 0.5014,
      "step": 3381
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4881203211731646,
      "learning_rate": 2.8687005035551797e-06,
      "loss": 0.4978,
      "step": 3382
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.2982399565684934,
      "learning_rate": 2.863598439134724e-06,
      "loss": 1.1188,
      "step": 3383
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6721597471307141,
      "learning_rate": 2.8585001575493355e-06,
      "loss": 0.5629,
      "step": 3384
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.491854056130621,
      "learning_rate": 2.8534056615014784e-06,
      "loss": 0.4775,
      "step": 3385
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8650656376143078,
      "learning_rate": 2.8483149536916353e-06,
      "loss": 0.5965,
      "step": 3386
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6958877510645423,
      "learning_rate": 2.8432280368182594e-06,
      "loss": 0.6141,
      "step": 3387
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.713688696746757,
      "learning_rate": 2.838144913577805e-06,
      "loss": 0.5061,
      "step": 3388
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.2876291396868624,
      "learning_rate": 2.833065586664715e-06,
      "loss": 1.0525,
      "step": 3389
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4360564538183087,
      "learning_rate": 2.827990058771417e-06,
      "loss": 0.5389,
      "step": 3390
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6342287704392018,
      "learning_rate": 2.8229183325883314e-06,
      "loss": 0.5266,
      "step": 3391
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.2243791017692565,
      "learning_rate": 2.8178504108038483e-06,
      "loss": 1.0665,
      "step": 3392
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.718307668588566,
      "learning_rate": 2.8127862961043628e-06,
      "loss": 0.58,
      "step": 3393
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.692344468557161,
      "learning_rate": 2.8077259911742294e-06,
      "loss": 0.5402,
      "step": 3394
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.705003233416139,
      "learning_rate": 2.8026694986958058e-06,
      "loss": 0.6132,
      "step": 3395
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6629431937305166,
      "learning_rate": 2.797616821349405e-06,
      "loss": 0.5498,
      "step": 3396
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.653443233306531,
      "learning_rate": 2.792567961813343e-06,
      "loss": 0.5726,
      "step": 3397
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5543563220708856,
      "learning_rate": 2.7875229227638877e-06,
      "loss": 0.5673,
      "step": 3398
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7707063445908686,
      "learning_rate": 2.7824817068752984e-06,
      "loss": 0.5173,
      "step": 3399
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.259586369191246,
      "learning_rate": 2.777444316819802e-06,
      "loss": 1.0622,
      "step": 3400
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.576406571419652,
      "learning_rate": 2.7724107552675973e-06,
      "loss": 0.5514,
      "step": 3401
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6704981255541314,
      "learning_rate": 2.767381024886858e-06,
      "loss": 0.5815,
      "step": 3402
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4959897715418302,
      "learning_rate": 2.7623551283437144e-06,
      "loss": 0.5938,
      "step": 3403
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.29081352373753,
      "learning_rate": 2.757333068302285e-06,
      "loss": 1.0625,
      "step": 3404
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5618148049735008,
      "learning_rate": 2.752314847424631e-06,
      "loss": 0.5464,
      "step": 3405
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7047461969369082,
      "learning_rate": 2.747300468370804e-06,
      "loss": 0.5533,
      "step": 3406
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.4051970874643436,
      "learning_rate": 2.742289933798793e-06,
      "loss": 1.0438,
      "step": 3407
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5920942816954915,
      "learning_rate": 2.7372832463645736e-06,
      "loss": 0.5716,
      "step": 3408
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.657936205988066,
      "learning_rate": 2.7322804087220623e-06,
      "loss": 0.5688,
      "step": 3409
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4883387898653317,
      "learning_rate": 2.7272814235231472e-06,
      "loss": 0.5505,
      "step": 3410
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1341605419260676,
      "learning_rate": 2.722286293417671e-06,
      "loss": 1.0978,
      "step": 3411
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6836281986164316,
      "learning_rate": 2.7172950210534243e-06,
      "loss": 0.5616,
      "step": 3412
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4689502926297786,
      "learning_rate": 2.7123076090761735e-06,
      "loss": 0.5854,
      "step": 3413
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7445220113054247,
      "learning_rate": 2.7073240601296125e-06,
      "loss": 0.5626,
      "step": 3414
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5925407315365476,
      "learning_rate": 2.7023443768554135e-06,
      "loss": 0.5313,
      "step": 3415
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5542974377848937,
      "learning_rate": 2.6973685618931746e-06,
      "loss": 0.6046,
      "step": 3416
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5960696916157653,
      "learning_rate": 2.6923966178804685e-06,
      "loss": 0.613,
      "step": 3417
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6591719838094874,
      "learning_rate": 2.687428547452794e-06,
      "loss": 0.5421,
      "step": 3418
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5715693096599364,
      "learning_rate": 2.682464353243609e-06,
      "loss": 0.5744,
      "step": 3419
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5112799826512915,
      "learning_rate": 2.6775040378843133e-06,
      "loss": 0.5269,
      "step": 3420
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5314013979933803,
      "learning_rate": 2.672547604004252e-06,
      "loss": 0.5131,
      "step": 3421
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.79458388922593,
      "learning_rate": 2.667595054230713e-06,
      "loss": 0.5649,
      "step": 3422
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.412800780668774,
      "learning_rate": 2.6626463911889154e-06,
      "loss": 1.095,
      "step": 3423
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.495644932619756,
      "learning_rate": 2.6577016175020396e-06,
      "loss": 1.0821,
      "step": 3424
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.561364638829661,
      "learning_rate": 2.652760735791179e-06,
      "loss": 0.5772,
      "step": 3425
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.53457569176031,
      "learning_rate": 2.6478237486753877e-06,
      "loss": 0.5858,
      "step": 3426
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1483428450282465,
      "learning_rate": 2.642890658771633e-06,
      "loss": 1.1282,
      "step": 3427
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.700310325518541,
      "learning_rate": 2.637961468694837e-06,
      "loss": 0.6227,
      "step": 3428
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7581293898984818,
      "learning_rate": 2.63303618105784e-06,
      "loss": 0.6316,
      "step": 3429
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7572417800227023,
      "learning_rate": 2.6281147984714183e-06,
      "loss": 0.6099,
      "step": 3430
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6117991687145883,
      "learning_rate": 2.62319732354428e-06,
      "loss": 0.5806,
      "step": 3431
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5721621021501655,
      "learning_rate": 2.618283758883059e-06,
      "loss": 0.5893,
      "step": 3432
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6541300364045264,
      "learning_rate": 2.6133741070923237e-06,
      "loss": 0.5721,
      "step": 3433
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3747491686264857,
      "learning_rate": 2.6084683707745506e-06,
      "loss": 0.5584,
      "step": 3434
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7396957436345515,
      "learning_rate": 2.603566552530167e-06,
      "loss": 0.5403,
      "step": 3435
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6873333074290233,
      "learning_rate": 2.5986686549574956e-06,
      "loss": 0.5602,
      "step": 3436
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4990642350838463,
      "learning_rate": 2.5937746806528076e-06,
      "loss": 0.566,
      "step": 3437
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.613578970500747,
      "learning_rate": 2.5888846322102667e-06,
      "loss": 0.6211,
      "step": 3438
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5407101865724255,
      "learning_rate": 2.5839985122219847e-06,
      "loss": 0.5091,
      "step": 3439
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5266781663719942,
      "learning_rate": 2.579116323277967e-06,
      "loss": 0.5417,
      "step": 3440
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6378076576255292,
      "learning_rate": 2.574238067966146e-06,
      "loss": 0.5194,
      "step": 3441
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6006668193118703,
      "learning_rate": 2.5693637488723733e-06,
      "loss": 0.4994,
      "step": 3442
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.2900852316836593,
      "learning_rate": 2.5644933685803974e-06,
      "loss": 1.0392,
      "step": 3443
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6707616392513116,
      "learning_rate": 2.5596269296719023e-06,
      "loss": 0.5234,
      "step": 3444
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.623837953156606,
      "learning_rate": 2.5547644347264587e-06,
      "loss": 0.594,
      "step": 3445
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.646646699100077,
      "learning_rate": 2.5499058863215696e-06,
      "loss": 0.5482,
      "step": 3446
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.2016705760873245,
      "learning_rate": 2.5450512870326247e-06,
      "loss": 1.1021,
      "step": 3447
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7144359481452829,
      "learning_rate": 2.5402006394329403e-06,
      "loss": 0.546,
      "step": 3448
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5496395950795636,
      "learning_rate": 2.5353539460937203e-06,
      "loss": 0.5874,
      "step": 3449
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.132142900630312,
      "learning_rate": 2.5305112095840835e-06,
      "loss": 1.1097,
      "step": 3450
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7263548245124651,
      "learning_rate": 2.525672432471048e-06,
      "loss": 0.6165,
      "step": 3451
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3838817213332,
      "learning_rate": 2.5208376173195335e-06,
      "loss": 0.9944,
      "step": 3452
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.296495825496811,
      "learning_rate": 2.5160067666923617e-06,
      "loss": 1.1153,
      "step": 3453
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.168260631203937,
      "learning_rate": 2.5111798831502422e-06,
      "loss": 1.0777,
      "step": 3454
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6214472765520342,
      "learning_rate": 2.506356969251802e-06,
      "loss": 0.525,
      "step": 3455
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6402406432649501,
      "learning_rate": 2.5015380275535395e-06,
      "loss": 0.5815,
      "step": 3456
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7606551116827638,
      "learning_rate": 2.4967230606098726e-06,
      "loss": 0.6538,
      "step": 3457
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.849151497990427,
      "learning_rate": 2.491912070973086e-06,
      "loss": 0.5523,
      "step": 3458
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7890434919745564,
      "learning_rate": 2.4871050611933835e-06,
      "loss": 0.6279,
      "step": 3459
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5963885401734976,
      "learning_rate": 2.4823020338188365e-06,
      "loss": 0.5321,
      "step": 3460
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6511219900501974,
      "learning_rate": 2.477502991395416e-06,
      "loss": 0.5304,
      "step": 3461
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.2834758924347467,
      "learning_rate": 2.4727079364669816e-06,
      "loss": 1.0944,
      "step": 3462
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.497766757054447,
      "learning_rate": 2.4679168715752743e-06,
      "loss": 1.0415,
      "step": 3463
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7902585978714878,
      "learning_rate": 2.4631297992599256e-06,
      "loss": 0.5732,
      "step": 3464
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.673119120421329,
      "learning_rate": 2.458346722058441e-06,
      "loss": 0.5348,
      "step": 3465
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4082400479507338,
      "learning_rate": 2.453567642506224e-06,
      "loss": 0.4615,
      "step": 3466
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4952685613149057,
      "learning_rate": 2.4487925631365396e-06,
      "loss": 0.5185,
      "step": 3467
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.505116504505397,
      "learning_rate": 2.444021486480553e-06,
      "loss": 0.5224,
      "step": 3468
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5544187996990133,
      "learning_rate": 2.4392544150672883e-06,
      "loss": 0.5764,
      "step": 3469
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.323175833798922,
      "learning_rate": 2.4344913514236658e-06,
      "loss": 1.0344,
      "step": 3470
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5475423472292886,
      "learning_rate": 2.429732298074461e-06,
      "loss": 0.5233,
      "step": 3471
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.227963537963749,
      "learning_rate": 2.424977257542338e-06,
      "loss": 1.0643,
      "step": 3472
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5462182262178283,
      "learning_rate": 2.4202262323478332e-06,
      "loss": 0.5493,
      "step": 3473
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6372932293772917,
      "learning_rate": 2.4154792250093406e-06,
      "loss": 0.5672,
      "step": 3474
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.525702583135812,
      "learning_rate": 2.4107362380431474e-06,
      "loss": 1.1054,
      "step": 3475
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4752070855157788,
      "learning_rate": 2.4059972739633853e-06,
      "loss": 0.5345,
      "step": 3476
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.371254012756979,
      "learning_rate": 2.4012623352820773e-06,
      "loss": 1.0901,
      "step": 3477
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.289862304909163,
      "learning_rate": 2.3965314245090887e-06,
      "loss": 1.1196,
      "step": 3478
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6836356860166355,
      "learning_rate": 2.391804544152172e-06,
      "loss": 0.5449,
      "step": 3479
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.590923630851415,
      "learning_rate": 2.3870816967169262e-06,
      "loss": 0.571,
      "step": 3480
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.764793863133477,
      "learning_rate": 2.3823628847068226e-06,
      "loss": 0.6622,
      "step": 3481
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5653886321007857,
      "learning_rate": 2.3776481106231873e-06,
      "loss": 0.5456,
      "step": 3482
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6906816522100017,
      "learning_rate": 2.372937376965211e-06,
      "loss": 0.5926,
      "step": 3483
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.59153260543579,
      "learning_rate": 2.368230686229943e-06,
      "loss": 0.5485,
      "step": 3484
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6988049191532373,
      "learning_rate": 2.3635280409122786e-06,
      "loss": 0.4875,
      "step": 3485
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5799351420850263,
      "learning_rate": 2.3588294435049874e-06,
      "loss": 0.5565,
      "step": 3486
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5823914017067187,
      "learning_rate": 2.3541348964986733e-06,
      "loss": 0.5556,
      "step": 3487
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5897444952650084,
      "learning_rate": 2.349444402381813e-06,
      "loss": 0.5557,
      "step": 3488
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.685960517410604,
      "learning_rate": 2.3447579636407136e-06,
      "loss": 0.4795,
      "step": 3489
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.403564406165609,
      "learning_rate": 2.3400755827595556e-06,
      "loss": 1.1062,
      "step": 3490
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3258570485857533,
      "learning_rate": 2.335397262220347e-06,
      "loss": 1.1029,
      "step": 3491
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.268684272367529,
      "learning_rate": 2.330723004502957e-06,
      "loss": 1.0797,
      "step": 3492
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7470361447823881,
      "learning_rate": 2.326052812085098e-06,
      "loss": 0.5632,
      "step": 3493
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.608148416619344,
      "learning_rate": 2.321386687442326e-06,
      "loss": 0.5993,
      "step": 3494
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6181454592118192,
      "learning_rate": 2.3167246330480442e-06,
      "loss": 0.5378,
      "step": 3495
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.61627000903093,
      "learning_rate": 2.312066651373487e-06,
      "loss": 0.5494,
      "step": 3496
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.585332196030185,
      "learning_rate": 2.3074127448877504e-06,
      "loss": 0.5717,
      "step": 3497
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.475502665582555,
      "learning_rate": 2.302762916057747e-06,
      "loss": 0.5762,
      "step": 3498
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8814214767958901,
      "learning_rate": 2.2981171673482514e-06,
      "loss": 0.59,
      "step": 3499
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6337033087508555,
      "learning_rate": 2.2934755012218503e-06,
      "loss": 0.5946,
      "step": 3500
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6315816141198245,
      "learning_rate": 2.2888379201389934e-06,
      "loss": 0.5376,
      "step": 3501
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6546219926862664,
      "learning_rate": 2.28420442655794e-06,
      "loss": 0.5338,
      "step": 3502
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4333645594988553,
      "learning_rate": 2.279575022934797e-06,
      "loss": 0.5004,
      "step": 3503
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6257759910894192,
      "learning_rate": 2.274949711723502e-06,
      "loss": 0.5234,
      "step": 3504
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.338688677049716,
      "learning_rate": 2.2703284953758186e-06,
      "loss": 1.0869,
      "step": 3505
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8601169198162255,
      "learning_rate": 2.2657113763413487e-06,
      "loss": 0.6654,
      "step": 3506
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6194864496260655,
      "learning_rate": 2.261098357067505e-06,
      "loss": 0.5852,
      "step": 3507
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7283663387484471,
      "learning_rate": 2.2564894399995495e-06,
      "loss": 0.6316,
      "step": 3508
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.625651149428514,
      "learning_rate": 2.2518846275805494e-06,
      "loss": 0.5106,
      "step": 3509
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4996936047614242,
      "learning_rate": 2.247283922251414e-06,
      "loss": 0.5838,
      "step": 3510
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.229235564428087,
      "learning_rate": 2.242687326450861e-06,
      "loss": 1.1179,
      "step": 3511
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7327301045483623,
      "learning_rate": 2.2380948426154346e-06,
      "loss": 0.6414,
      "step": 3512
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7279441981439854,
      "learning_rate": 2.233506473179504e-06,
      "loss": 0.5679,
      "step": 3513
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4903932124063164,
      "learning_rate": 2.2289222205752502e-06,
      "loss": 0.5331,
      "step": 3514
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6952445236275806,
      "learning_rate": 2.2243420872326816e-06,
      "loss": 0.588,
      "step": 3515
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6832577166048694,
      "learning_rate": 2.2197660755796058e-06,
      "loss": 0.5833,
      "step": 3516
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6204606582097876,
      "learning_rate": 2.215194188041668e-06,
      "loss": 0.5299,
      "step": 3517
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.586628585648435,
      "learning_rate": 2.2106264270423062e-06,
      "loss": 0.5025,
      "step": 3518
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5562555777997995,
      "learning_rate": 2.206062795002791e-06,
      "loss": 0.554,
      "step": 3519
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.3916200914294086,
      "learning_rate": 2.2015032943421843e-06,
      "loss": 1.0729,
      "step": 3520
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.292544241128947,
      "learning_rate": 2.1969479274773763e-06,
      "loss": 1.0577,
      "step": 3521
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.174892313157637,
      "learning_rate": 2.1923966968230524e-06,
      "loss": 1.0914,
      "step": 3522
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.63303941417267,
      "learning_rate": 2.187849604791712e-06,
      "loss": 0.5312,
      "step": 3523
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7289990572065816,
      "learning_rate": 2.1833066537936587e-06,
      "loss": 0.5546,
      "step": 3524
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5744546208019272,
      "learning_rate": 2.178767846237002e-06,
      "loss": 0.5388,
      "step": 3525
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6713811651685666,
      "learning_rate": 2.1742331845276574e-06,
      "loss": 0.542,
      "step": 3526
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.50395008907572,
      "learning_rate": 2.1697026710693337e-06,
      "loss": 0.5068,
      "step": 3527
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5607675962181824,
      "learning_rate": 2.165176308263557e-06,
      "loss": 0.5443,
      "step": 3528
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.3401293413753286,
      "learning_rate": 2.1606540985096314e-06,
      "loss": 1.0712,
      "step": 3529
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.3410033267671952,
      "learning_rate": 2.156136044204684e-06,
      "loss": 1.092,
      "step": 3530
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.534775650612242,
      "learning_rate": 2.151622147743616e-06,
      "loss": 0.5729,
      "step": 3531
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6489800464873994,
      "learning_rate": 2.1471124115191456e-06,
      "loss": 0.5441,
      "step": 3532
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.561189559943082,
      "learning_rate": 2.142606837921768e-06,
      "loss": 0.6021,
      "step": 3533
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.608504484416058,
      "learning_rate": 2.138105429339783e-06,
      "loss": 0.5493,
      "step": 3534
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6319012294345816,
      "learning_rate": 2.133608188159277e-06,
      "loss": 0.4845,
      "step": 3535
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7189143993960507,
      "learning_rate": 2.129115116764132e-06,
      "loss": 0.6398,
      "step": 3536
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6231414493033005,
      "learning_rate": 2.1246262175360178e-06,
      "loss": 0.5405,
      "step": 3537
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.3780974114290383,
      "learning_rate": 2.120141492854386e-06,
      "loss": 1.0304,
      "step": 3538
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5196801359538206,
      "learning_rate": 2.1156609450964906e-06,
      "loss": 0.5466,
      "step": 3539
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.236458328466868,
      "learning_rate": 2.111184576637352e-06,
      "loss": 1.0643,
      "step": 3540
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.621217436969229,
      "learning_rate": 2.1067123898497964e-06,
      "loss": 0.4903,
      "step": 3541
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.51312183762788,
      "learning_rate": 2.102244387104415e-06,
      "loss": 0.5676,
      "step": 3542
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5960122387207418,
      "learning_rate": 2.0977805707695918e-06,
      "loss": 0.5674,
      "step": 3543
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5691591922463792,
      "learning_rate": 2.093320943211488e-06,
      "loss": 0.5445,
      "step": 3544
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.575177476438376,
      "learning_rate": 2.0888655067940435e-06,
      "loss": 0.5519,
      "step": 3545
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.639273166146143,
      "learning_rate": 2.084414263878981e-06,
      "loss": 0.5956,
      "step": 3546
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.131585899045124,
      "learning_rate": 2.079967216825796e-06,
      "loss": 1.0899,
      "step": 3547
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6099330922142332,
      "learning_rate": 2.0755243679917623e-06,
      "loss": 0.5596,
      "step": 3548
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5338984615012543,
      "learning_rate": 2.0710857197319224e-06,
      "loss": 0.5534,
      "step": 3549
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6746396044208507,
      "learning_rate": 2.0666512743991072e-06,
      "loss": 0.553,
      "step": 3550
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6489658871052262,
      "learning_rate": 2.0622210343438977e-06,
      "loss": 0.5937,
      "step": 3551
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5922440697763853,
      "learning_rate": 2.0577950019146685e-06,
      "loss": 0.5811,
      "step": 3552
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7150799945721775,
      "learning_rate": 2.0533731794575464e-06,
      "loss": 0.5698,
      "step": 3553
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6460304473386618,
      "learning_rate": 2.0489555693164344e-06,
      "loss": 0.5557,
      "step": 3554
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8277443871868209,
      "learning_rate": 2.0445421738330028e-06,
      "loss": 0.5929,
      "step": 3555
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.234274761536857,
      "learning_rate": 2.0401329953466865e-06,
      "loss": 1.065,
      "step": 3556
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.3620438874348397,
      "learning_rate": 2.035728036194686e-06,
      "loss": 1.0381,
      "step": 3557
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8814115447190034,
      "learning_rate": 2.031327298711959e-06,
      "loss": 0.5741,
      "step": 3558
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4390749349324636,
      "learning_rate": 2.0269307852312394e-06,
      "loss": 0.52,
      "step": 3559
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.3627171843458283,
      "learning_rate": 2.0225384980830044e-06,
      "loss": 1.0748,
      "step": 3560
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6123938423055764,
      "learning_rate": 2.0181504395955077e-06,
      "loss": 0.5583,
      "step": 3561
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4836905610933853,
      "learning_rate": 2.0137666120947463e-06,
      "loss": 0.5077,
      "step": 3562
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.634948003514975,
      "learning_rate": 2.0093870179044906e-06,
      "loss": 0.5651,
      "step": 3563
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9450071583371282,
      "learning_rate": 2.005011659346251e-06,
      "loss": 0.6135,
      "step": 3564
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5657184219960574,
      "learning_rate": 2.0006405387393025e-06,
      "loss": 0.6203,
      "step": 3565
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7434539401342557,
      "learning_rate": 1.9962736584006714e-06,
      "loss": 0.5825,
      "step": 3566
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5075119507671602,
      "learning_rate": 1.9919110206451364e-06,
      "loss": 0.6002,
      "step": 3567
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5811721681831503,
      "learning_rate": 1.9875526277852275e-06,
      "loss": 0.5607,
      "step": 3568
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1898682237071565,
      "learning_rate": 1.983198482131219e-06,
      "loss": 1.1035,
      "step": 3569
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6418188299224568,
      "learning_rate": 1.9788485859911475e-06,
      "loss": 0.5564,
      "step": 3570
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4738067200095204,
      "learning_rate": 1.9745029416707784e-06,
      "loss": 0.5557,
      "step": 3571
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4836669568179652,
      "learning_rate": 1.970161551473645e-06,
      "loss": 0.606,
      "step": 3572
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4626770463131777,
      "learning_rate": 1.9658244177010043e-06,
      "loss": 0.4767,
      "step": 3573
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5989767699916897,
      "learning_rate": 1.961491542651871e-06,
      "loss": 0.579,
      "step": 3574
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6192710232211878,
      "learning_rate": 1.9571629286229977e-06,
      "loss": 0.5417,
      "step": 3575
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5178278218848136,
      "learning_rate": 1.9528385779088776e-06,
      "loss": 0.5351,
      "step": 3576
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6364401580677173,
      "learning_rate": 1.9485184928017453e-06,
      "loss": 0.5748,
      "step": 3577
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6342064650675063,
      "learning_rate": 1.9442026755915767e-06,
      "loss": 0.5593,
      "step": 3578
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7595010345251452,
      "learning_rate": 1.9398911285660817e-06,
      "loss": 0.6186,
      "step": 3579
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.505536130590274,
      "learning_rate": 1.935583854010703e-06,
      "loss": 0.5273,
      "step": 3580
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7920564245068176,
      "learning_rate": 1.9312808542086325e-06,
      "loss": 0.545,
      "step": 3581
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7119752982755403,
      "learning_rate": 1.9269821314407765e-06,
      "loss": 0.5607,
      "step": 3582
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.473792859286223,
      "learning_rate": 1.922687687985795e-06,
      "loss": 0.5568,
      "step": 3583
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4838437924787053,
      "learning_rate": 1.9183975261200625e-06,
      "loss": 0.5412,
      "step": 3584
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3301451322434326,
      "learning_rate": 1.9141116481176926e-06,
      "loss": 0.5128,
      "step": 3585
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.629363681958065,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 0.5821,
      "step": 3586
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7282249110633878,
      "learning_rate": 1.9055527527881334e-06,
      "loss": 0.608,
      "step": 3587
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5762153655264195,
      "learning_rate": 1.9012797399978088e-06,
      "loss": 0.6005,
      "step": 3588
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6455760648634703,
      "learning_rate": 1.8970110201445734e-06,
      "loss": 0.6054,
      "step": 3589
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4387996432649592,
      "learning_rate": 1.8927465954911773e-06,
      "loss": 0.5282,
      "step": 3590
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.630666786232718,
      "learning_rate": 1.8884864682980809e-06,
      "loss": 0.6154,
      "step": 3591
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1910655342424046,
      "learning_rate": 1.8842306408234855e-06,
      "loss": 1.1198,
      "step": 3592
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.283540834764297,
      "learning_rate": 1.8799791153232938e-06,
      "loss": 1.0784,
      "step": 3593
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.265014938705704,
      "learning_rate": 1.875731894051146e-06,
      "loss": 1.0681,
      "step": 3594
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5162520866220157,
      "learning_rate": 1.8714889792583856e-06,
      "loss": 0.5613,
      "step": 3595
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.614316665623063,
      "learning_rate": 1.8672503731940827e-06,
      "loss": 0.6007,
      "step": 3596
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1859939992682325,
      "learning_rate": 1.8630160781050188e-06,
      "loss": 1.0467,
      "step": 3597
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.443798714506547,
      "learning_rate": 1.858786096235693e-06,
      "loss": 0.5267,
      "step": 3598
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4922023417966028,
      "learning_rate": 1.8545604298283204e-06,
      "loss": 0.5794,
      "step": 3599
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3714299312871359,
      "learning_rate": 1.850339081122816e-06,
      "loss": 0.5125,
      "step": 3600
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.3642632355051694,
      "learning_rate": 1.846122052356828e-06,
      "loss": 1.0347,
      "step": 3601
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.2363731303778773,
      "learning_rate": 1.841909345765689e-06,
      "loss": 1.0092,
      "step": 3602
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.172758691464766,
      "learning_rate": 1.837700963582466e-06,
      "loss": 1.1581,
      "step": 3603
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6782570210081211,
      "learning_rate": 1.8334969080379116e-06,
      "loss": 0.5627,
      "step": 3604
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5786740538893889,
      "learning_rate": 1.8292971813604998e-06,
      "loss": 0.4944,
      "step": 3605
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4779327392040158,
      "learning_rate": 1.825101785776403e-06,
      "loss": 0.5564,
      "step": 3606
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5918384674580903,
      "learning_rate": 1.8209107235095003e-06,
      "loss": 0.5606,
      "step": 3607
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6427029061991183,
      "learning_rate": 1.8167239967813721e-06,
      "loss": 0.5406,
      "step": 3608
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6395577851814114,
      "learning_rate": 1.8125416078113035e-06,
      "loss": 0.5646,
      "step": 3609
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7064024810733598,
      "learning_rate": 1.8083635588162796e-06,
      "loss": 0.62,
      "step": 3610
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.3379538825222643,
      "learning_rate": 1.8041898520109757e-06,
      "loss": 1.0581,
      "step": 3611
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.435966577479976,
      "learning_rate": 1.8000204896077855e-06,
      "loss": 0.5534,
      "step": 3612
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7522488598885106,
      "learning_rate": 1.7958554738167778e-06,
      "loss": 0.5803,
      "step": 3613
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4726522650181704,
      "learning_rate": 1.7916948068457352e-06,
      "loss": 0.5781,
      "step": 3614
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5946632042365851,
      "learning_rate": 1.7875384909001215e-06,
      "loss": 0.6104,
      "step": 3615
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6327911481430497,
      "learning_rate": 1.783386528183102e-06,
      "loss": 0.5661,
      "step": 3616
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.391485875603982,
      "learning_rate": 1.779238920895532e-06,
      "loss": 1.116,
      "step": 3617
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.555706080432568,
      "learning_rate": 1.7750956712359601e-06,
      "loss": 0.552,
      "step": 3618
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6551011300153886,
      "learning_rate": 1.7709567814006213e-06,
      "loss": 0.5624,
      "step": 3619
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4904365687943955,
      "learning_rate": 1.7668222535834434e-06,
      "loss": 0.5933,
      "step": 3620
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.4079869004791523,
      "learning_rate": 1.7626920899760425e-06,
      "loss": 1.0893,
      "step": 3621
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6621461648367493,
      "learning_rate": 1.7585662927677105e-06,
      "loss": 0.5638,
      "step": 3622
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6499775428630843,
      "learning_rate": 1.7544448641454436e-06,
      "loss": 0.5856,
      "step": 3623
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.274322427679041,
      "learning_rate": 1.7503278062939044e-06,
      "loss": 0.973,
      "step": 3624
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4812656721914874,
      "learning_rate": 1.7462151213954536e-06,
      "loss": 0.593,
      "step": 3625
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1901922754134584,
      "learning_rate": 1.74210681163012e-06,
      "loss": 1.0976,
      "step": 3626
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6484302074140653,
      "learning_rate": 1.738002879175623e-06,
      "loss": 0.5072,
      "step": 3627
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.27787232378467,
      "learning_rate": 1.733903326207358e-06,
      "loss": 1.0532,
      "step": 3628
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7031758200905394,
      "learning_rate": 1.7298081548984003e-06,
      "loss": 0.55,
      "step": 3629
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5887221277389545,
      "learning_rate": 1.7257173674195016e-06,
      "loss": 0.55,
      "step": 3630
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5933128861570989,
      "learning_rate": 1.7216309659390895e-06,
      "loss": 0.5571,
      "step": 3631
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7446727913618996,
      "learning_rate": 1.7175489526232703e-06,
      "loss": 0.586,
      "step": 3632
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4653429006557477,
      "learning_rate": 1.713471329635813e-06,
      "loss": 0.5315,
      "step": 3633
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5657710639509326,
      "learning_rate": 1.7093980991381787e-06,
      "loss": 0.522,
      "step": 3634
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.4051250234582846,
      "learning_rate": 1.7053292632894813e-06,
      "loss": 1.0668,
      "step": 3635
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.283959811176428,
      "learning_rate": 1.7012648242465145e-06,
      "loss": 1.1195,
      "step": 3636
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.187017139687313,
      "learning_rate": 1.6972047841637419e-06,
      "loss": 1.0913,
      "step": 3637
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9054062452346425,
      "learning_rate": 1.6931491451932902e-06,
      "loss": 0.5599,
      "step": 3638
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8106192604401967,
      "learning_rate": 1.6890979094849592e-06,
      "loss": 0.5571,
      "step": 3639
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2948165423246882,
      "learning_rate": 1.6850510791862097e-06,
      "loss": 1.0692,
      "step": 3640
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.288612119746345,
      "learning_rate": 1.6810086564421723e-06,
      "loss": 1.0853,
      "step": 3641
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6169228484255822,
      "learning_rate": 1.6769706433956302e-06,
      "loss": 0.5803,
      "step": 3642
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.501848408552832,
      "learning_rate": 1.6729370421870484e-06,
      "loss": 0.5169,
      "step": 3643
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.644972275938963,
      "learning_rate": 1.6689078549545313e-06,
      "loss": 0.5886,
      "step": 3644
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2641122782366123,
      "learning_rate": 1.6648830838338636e-06,
      "loss": 1.1068,
      "step": 3645
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5568472410204814,
      "learning_rate": 1.6608627309584734e-06,
      "loss": 0.6022,
      "step": 3646
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.580538388566508,
      "learning_rate": 1.656846798459456e-06,
      "loss": 0.5426,
      "step": 3647
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2750631228349345,
      "learning_rate": 1.6528352884655596e-06,
      "loss": 1.0702,
      "step": 3648
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4643956064742811,
      "learning_rate": 1.6488282031031889e-06,
      "loss": 0.5078,
      "step": 3649
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4645565955126625,
      "learning_rate": 1.6448255444964056e-06,
      "loss": 0.5882,
      "step": 3650
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.577507818281178,
      "learning_rate": 1.6408273147669208e-06,
      "loss": 0.5701,
      "step": 3651
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.403149771616661,
      "learning_rate": 1.6368335160341032e-06,
      "loss": 1.1057,
      "step": 3652
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5675286630934753,
      "learning_rate": 1.632844150414964e-06,
      "loss": 0.5162,
      "step": 3653
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.742636367332345,
      "learning_rate": 1.6288592200241771e-06,
      "loss": 0.5585,
      "step": 3654
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0722908839628476,
      "learning_rate": 1.6248787269740496e-06,
      "loss": 1.0782,
      "step": 3655
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.582532859138536,
      "learning_rate": 1.620902673374556e-06,
      "loss": 0.5077,
      "step": 3656
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2207914962964863,
      "learning_rate": 1.6169310613332988e-06,
      "loss": 1.0782,
      "step": 3657
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6385846103604382,
      "learning_rate": 1.6129638929555357e-06,
      "loss": 0.5707,
      "step": 3658
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5001382949603395,
      "learning_rate": 1.6090011703441687e-06,
      "loss": 0.6613,
      "step": 3659
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.4695213460514833,
      "learning_rate": 1.605042895599742e-06,
      "loss": 1.0878,
      "step": 3660
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6159275914059996,
      "learning_rate": 1.6010890708204407e-06,
      "loss": 0.5799,
      "step": 3661
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6013343908879931,
      "learning_rate": 1.5971396981020937e-06,
      "loss": 0.5174,
      "step": 3662
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6990284463366103,
      "learning_rate": 1.5931947795381696e-06,
      "loss": 0.5871,
      "step": 3663
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6929686534975632,
      "learning_rate": 1.5892543172197694e-06,
      "loss": 0.5727,
      "step": 3664
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7045545549405328,
      "learning_rate": 1.5853183132356453e-06,
      "loss": 0.6646,
      "step": 3665
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.763280185830171,
      "learning_rate": 1.581386769672173e-06,
      "loss": 0.542,
      "step": 3666
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.301808210192265,
      "learning_rate": 1.5774596886133698e-06,
      "loss": 1.0977,
      "step": 3667
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6873639072922937,
      "learning_rate": 1.5735370721408894e-06,
      "loss": 0.5498,
      "step": 3668
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2708994070855013,
      "learning_rate": 1.5696189223340152e-06,
      "loss": 1.075,
      "step": 3669
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6178141939212585,
      "learning_rate": 1.565705241269664e-06,
      "loss": 0.5675,
      "step": 3670
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5229760102557692,
      "learning_rate": 1.5617960310223845e-06,
      "loss": 0.5611,
      "step": 3671
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.378286960456438,
      "learning_rate": 1.5578912936643553e-06,
      "loss": 1.085,
      "step": 3672
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7447591866399748,
      "learning_rate": 1.553991031265385e-06,
      "loss": 0.5542,
      "step": 3673
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6604745153248954,
      "learning_rate": 1.5500952458929087e-06,
      "loss": 0.5401,
      "step": 3674
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6389834225902642,
      "learning_rate": 1.5462039396119844e-06,
      "loss": 0.5426,
      "step": 3675
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6588180916274116,
      "learning_rate": 1.5423171144853078e-06,
      "loss": 0.5333,
      "step": 3676
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.168544370340613,
      "learning_rate": 1.5384347725731841e-06,
      "loss": 1.0406,
      "step": 3677
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5661202040819495,
      "learning_rate": 1.5345569159335527e-06,
      "loss": 0.6023,
      "step": 3678
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.2669580097100144,
      "learning_rate": 1.5306835466219738e-06,
      "loss": 1.055,
      "step": 3679
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6354921876990525,
      "learning_rate": 1.526814666691624e-06,
      "loss": 0.544,
      "step": 3680
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6222786362208896,
      "learning_rate": 1.522950278193307e-06,
      "loss": 0.5222,
      "step": 3681
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5756716880411759,
      "learning_rate": 1.51909038317544e-06,
      "loss": 0.5778,
      "step": 3682
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.37680689044782,
      "learning_rate": 1.515234983684064e-06,
      "loss": 1.0635,
      "step": 3683
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7497825589631308,
      "learning_rate": 1.5113840817628267e-06,
      "loss": 0.5171,
      "step": 3684
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7516420204630294,
      "learning_rate": 1.5075376794530072e-06,
      "loss": 0.605,
      "step": 3685
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3047934204705425,
      "learning_rate": 1.5036957787934825e-06,
      "loss": 1.0775,
      "step": 3686
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1921314440168143,
      "learning_rate": 1.4998583818207613e-06,
      "loss": 1.0149,
      "step": 3687
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6590392962107345,
      "learning_rate": 1.4960254905689475e-06,
      "loss": 0.622,
      "step": 3688
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.780731776478646,
      "learning_rate": 1.4921971070697683e-06,
      "loss": 0.5745,
      "step": 3689
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5749297897159737,
      "learning_rate": 1.488373233352557e-06,
      "loss": 0.5192,
      "step": 3690
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3353078858812595,
      "learning_rate": 1.4845538714442576e-06,
      "loss": 0.409,
      "step": 3691
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6103799495261302,
      "learning_rate": 1.4807390233694218e-06,
      "loss": 0.5532,
      "step": 3692
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4905444610410907,
      "learning_rate": 1.47692869115021e-06,
      "loss": 0.5756,
      "step": 3693
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6223710894519763,
      "learning_rate": 1.4731228768063888e-06,
      "loss": 0.6072,
      "step": 3694
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6616000034765612,
      "learning_rate": 1.469321582355322e-06,
      "loss": 0.5425,
      "step": 3695
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5877792992528634,
      "learning_rate": 1.4655248098119945e-06,
      "loss": 0.528,
      "step": 3696
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7645530177736723,
      "learning_rate": 1.4617325611889766e-06,
      "loss": 0.5941,
      "step": 3697
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.663264524314924,
      "learning_rate": 1.4579448384964522e-06,
      "loss": 0.5491,
      "step": 3698
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5299187578129494,
      "learning_rate": 1.4541616437422002e-06,
      "loss": 0.5649,
      "step": 3699
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.4979729125929535,
      "learning_rate": 1.4503829789316016e-06,
      "loss": 1.0934,
      "step": 3700
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8122801973689455,
      "learning_rate": 1.4466088460676364e-06,
      "loss": 0.5691,
      "step": 3701
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5586731623585837,
      "learning_rate": 1.4428392471508813e-06,
      "loss": 0.4865,
      "step": 3702
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6600671848430433,
      "learning_rate": 1.439074184179511e-06,
      "loss": 0.5451,
      "step": 3703
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.405669574919026,
      "learning_rate": 1.4353136591492933e-06,
      "loss": 1.0698,
      "step": 3704
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1219937668727424,
      "learning_rate": 1.4315576740535964e-06,
      "loss": 1.061,
      "step": 3705
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7433893816763337,
      "learning_rate": 1.4278062308833707e-06,
      "loss": 0.564,
      "step": 3706
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.231205282672189,
      "learning_rate": 1.4240593316271733e-06,
      "loss": 1.0609,
      "step": 3707
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.262931001188077,
      "learning_rate": 1.4203169782711413e-06,
      "loss": 1.0569,
      "step": 3708
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7701869752011907,
      "learning_rate": 1.4165791727990063e-06,
      "loss": 0.6161,
      "step": 3709
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6938882566147913,
      "learning_rate": 1.4128459171920906e-06,
      "loss": 0.555,
      "step": 3710
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.2100199636416065,
      "learning_rate": 1.409117213429303e-06,
      "loss": 1.022,
      "step": 3711
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5140068832322726,
      "learning_rate": 1.4053930634871393e-06,
      "loss": 0.5051,
      "step": 3712
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3392941577516306,
      "learning_rate": 1.4016734693396827e-06,
      "loss": 1.1259,
      "step": 3713
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3155113545787898,
      "learning_rate": 1.3979584329586005e-06,
      "loss": 1.0939,
      "step": 3714
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.524518784495159,
      "learning_rate": 1.3942479563131439e-06,
      "loss": 0.4981,
      "step": 3715
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4771916563455756,
      "learning_rate": 1.3905420413701498e-06,
      "loss": 1.1237,
      "step": 3716
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5386728358799833,
      "learning_rate": 1.3868406900940279e-06,
      "loss": 0.5968,
      "step": 3717
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.2656708322377748,
      "learning_rate": 1.3831439044467853e-06,
      "loss": 1.0812,
      "step": 3718
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5479317609876782,
      "learning_rate": 1.3794516863879925e-06,
      "loss": 0.5729,
      "step": 3719
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7146188795582837,
      "learning_rate": 1.375764037874807e-06,
      "loss": 0.5671,
      "step": 3720
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5843756457424838,
      "learning_rate": 1.3720809608619623e-06,
      "loss": 0.4719,
      "step": 3721
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6156558350149945,
      "learning_rate": 1.3684024573017708e-06,
      "loss": 0.5799,
      "step": 3722
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.530532674524125,
      "learning_rate": 1.364728529144118e-06,
      "loss": 0.5403,
      "step": 3723
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.305307277658546,
      "learning_rate": 1.3610591783364647e-06,
      "loss": 1.099,
      "step": 3724
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1873390301626805,
      "learning_rate": 1.3573944068238477e-06,
      "loss": 1.0223,
      "step": 3725
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5311423282302534,
      "learning_rate": 1.3537342165488688e-06,
      "loss": 0.5383,
      "step": 3726
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.711687924035192,
      "learning_rate": 1.3500786094517148e-06,
      "loss": 0.5589,
      "step": 3727
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7442195242221614,
      "learning_rate": 1.3464275874701282e-06,
      "loss": 0.6305,
      "step": 3728
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5592877049016278,
      "learning_rate": 1.3427811525394318e-06,
      "loss": 0.522,
      "step": 3729
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5721972628018515,
      "learning_rate": 1.3391393065925117e-06,
      "loss": 0.5241,
      "step": 3730
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6376050620053968,
      "learning_rate": 1.3355020515598239e-06,
      "loss": 0.5409,
      "step": 3731
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5601958924866772,
      "learning_rate": 1.3318693893693901e-06,
      "loss": 0.5178,
      "step": 3732
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6734970559743527,
      "learning_rate": 1.328241321946796e-06,
      "loss": 0.5802,
      "step": 3733
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5958476824260228,
      "learning_rate": 1.324617851215194e-06,
      "loss": 0.5464,
      "step": 3734
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6043634505792475,
      "learning_rate": 1.3209989790952992e-06,
      "loss": 0.5176,
      "step": 3735
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6658493159838021,
      "learning_rate": 1.3173847075053902e-06,
      "loss": 0.4718,
      "step": 3736
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6051270411118481,
      "learning_rate": 1.313775038361299e-06,
      "loss": 0.5509,
      "step": 3737
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.668186928849172,
      "learning_rate": 1.3101699735764339e-06,
      "loss": 0.553,
      "step": 3738
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.129071230143133,
      "learning_rate": 1.3065695150617463e-06,
      "loss": 1.0556,
      "step": 3739
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.161576069498166,
      "learning_rate": 1.302973664725754e-06,
      "loss": 1.0936,
      "step": 3740
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.2299450729538375,
      "learning_rate": 1.2993824244745324e-06,
      "loss": 1.095,
      "step": 3741
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6301644888538869,
      "learning_rate": 1.2957957962117107e-06,
      "loss": 0.6432,
      "step": 3742
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5750358634776536,
      "learning_rate": 1.2922137818384762e-06,
      "loss": 0.5582,
      "step": 3743
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.644486253761014,
      "learning_rate": 1.2886363832535652e-06,
      "loss": 0.5711,
      "step": 3744
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6297285724342938,
      "learning_rate": 1.2850636023532747e-06,
      "loss": 0.5423,
      "step": 3745
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.3378552362496317,
      "learning_rate": 1.2814954410314474e-06,
      "loss": 1.0664,
      "step": 3746
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6551127571272881,
      "learning_rate": 1.2779319011794821e-06,
      "loss": 0.5245,
      "step": 3747
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6131379822371874,
      "learning_rate": 1.2743729846863206e-06,
      "loss": 0.543,
      "step": 3748
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5108441435188509,
      "learning_rate": 1.270818693438468e-06,
      "loss": 0.5551,
      "step": 3749
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6736652561508838,
      "learning_rate": 1.2672690293199596e-06,
      "loss": 0.6148,
      "step": 3750
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6039831221419594,
      "learning_rate": 1.2637239942123925e-06,
      "loss": 0.5725,
      "step": 3751
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6920325334099562,
      "learning_rate": 1.2601835899949021e-06,
      "loss": 0.5069,
      "step": 3752
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5867037002956337,
      "learning_rate": 1.2566478185441733e-06,
      "loss": 0.4831,
      "step": 3753
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6555100912168703,
      "learning_rate": 1.253116681734432e-06,
      "loss": 0.5948,
      "step": 3754
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6834078163740391,
      "learning_rate": 1.2495901814374511e-06,
      "loss": 0.5515,
      "step": 3755
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6921741720698478,
      "learning_rate": 1.246068319522542e-06,
      "loss": 0.5196,
      "step": 3756
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4409791782748924,
      "learning_rate": 1.2425510978565592e-06,
      "loss": 0.5265,
      "step": 3757
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7015038674568943,
      "learning_rate": 1.2390385183039011e-06,
      "loss": 0.4947,
      "step": 3758
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7781928320358253,
      "learning_rate": 1.2355305827264951e-06,
      "loss": 0.6085,
      "step": 3759
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6167788589434817,
      "learning_rate": 1.2320272929838183e-06,
      "loss": 0.5779,
      "step": 3760
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.2727263010593437,
      "learning_rate": 1.2285286509328786e-06,
      "loss": 1.042,
      "step": 3761
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6097624949378473,
      "learning_rate": 1.2250346584282235e-06,
      "loss": 0.5848,
      "step": 3762
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5722176985250986,
      "learning_rate": 1.2215453173219327e-06,
      "loss": 0.5905,
      "step": 3763
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5616399388572766,
      "learning_rate": 1.2180606294636244e-06,
      "loss": 0.5853,
      "step": 3764
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5835262133860661,
      "learning_rate": 1.2145805967004453e-06,
      "loss": 0.6072,
      "step": 3765
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5445667322037104,
      "learning_rate": 1.2111052208770791e-06,
      "loss": 0.5319,
      "step": 3766
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5191258243353962,
      "learning_rate": 1.2076345038357407e-06,
      "loss": 0.5084,
      "step": 3767
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7690421767721403,
      "learning_rate": 1.2041684474161675e-06,
      "loss": 0.5547,
      "step": 3768
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.193724362040917,
      "learning_rate": 1.200707053455642e-06,
      "loss": 1.0612,
      "step": 3769
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5627092073037196,
      "learning_rate": 1.1972503237889577e-06,
      "loss": 0.5699,
      "step": 3770
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6349258873086208,
      "learning_rate": 1.1937982602484476e-06,
      "loss": 0.5367,
      "step": 3771
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.691275791043727,
      "learning_rate": 1.1903508646639672e-06,
      "loss": 0.5793,
      "step": 3772
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7142790477730308,
      "learning_rate": 1.1869081388628978e-06,
      "loss": 0.6024,
      "step": 3773
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5278019260569746,
      "learning_rate": 1.1834700846701463e-06,
      "loss": 0.6028,
      "step": 3774
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.2525883876695856,
      "learning_rate": 1.1800367039081427e-06,
      "loss": 1.0643,
      "step": 3775
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.251687027334389,
      "learning_rate": 1.1766079983968382e-06,
      "loss": 1.0631,
      "step": 3776
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1552266011930152,
      "learning_rate": 1.1731839699537096e-06,
      "loss": 1.056,
      "step": 3777
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5809715031627551,
      "learning_rate": 1.169764620393753e-06,
      "loss": 0.5228,
      "step": 3778
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.2549662167360114,
      "learning_rate": 1.1663499515294763e-06,
      "loss": 1.0787,
      "step": 3779
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.556208242951289,
      "learning_rate": 1.162939965170924e-06,
      "loss": 0.5631,
      "step": 3780
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1577642084989868,
      "learning_rate": 1.1595346631256388e-06,
      "loss": 1.0455,
      "step": 3781
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6570409867576446,
      "learning_rate": 1.1561340471986949e-06,
      "loss": 0.5271,
      "step": 3782
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5491307072159946,
      "learning_rate": 1.1527381191926745e-06,
      "loss": 0.5482,
      "step": 3783
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5073293606755622,
      "learning_rate": 1.149346880907678e-06,
      "loss": 0.512,
      "step": 3784
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1609322653434795,
      "learning_rate": 1.1459603341413205e-06,
      "loss": 1.0826,
      "step": 3785
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6680021293713088,
      "learning_rate": 1.1425784806887286e-06,
      "loss": 0.5588,
      "step": 3786
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7801912335065282,
      "learning_rate": 1.1392013223425414e-06,
      "loss": 0.6186,
      "step": 3787
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5265654427765627,
      "learning_rate": 1.1358288608929092e-06,
      "loss": 0.556,
      "step": 3788
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.2498105079030695,
      "learning_rate": 1.1324610981274953e-06,
      "loss": 1.0478,
      "step": 3789
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6425779131266558,
      "learning_rate": 1.1290980358314652e-06,
      "loss": 0.5625,
      "step": 3790
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.6162375873412036,
      "learning_rate": 1.1257396757874993e-06,
      "loss": 0.9827,
      "step": 3791
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3826825673776464,
      "learning_rate": 1.1223860197757842e-06,
      "loss": 0.5294,
      "step": 3792
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4538719732162442,
      "learning_rate": 1.1190370695740127e-06,
      "loss": 0.5333,
      "step": 3793
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6436553813146249,
      "learning_rate": 1.1156928269573831e-06,
      "loss": 0.583,
      "step": 3794
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.3078289130408516,
      "learning_rate": 1.112353293698597e-06,
      "loss": 1.0578,
      "step": 3795
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.585929710762192,
      "learning_rate": 1.109018471567861e-06,
      "loss": 1.0314,
      "step": 3796
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6702745244687813,
      "learning_rate": 1.1056883623328852e-06,
      "loss": 0.5784,
      "step": 3797
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7170666081710948,
      "learning_rate": 1.1023629677588787e-06,
      "loss": 0.5926,
      "step": 3798
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5753648876734399,
      "learning_rate": 1.099042289608555e-06,
      "loss": 0.6102,
      "step": 3799
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.342751481082807,
      "learning_rate": 1.0957263296421273e-06,
      "loss": 1.072,
      "step": 3800
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.731028454304794,
      "learning_rate": 1.0924150896173014e-06,
      "loss": 0.6248,
      "step": 3801
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.730105392464523,
      "learning_rate": 1.08910857128929e-06,
      "loss": 0.5718,
      "step": 3802
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5955694042729978,
      "learning_rate": 1.085806776410797e-06,
      "loss": 0.4897,
      "step": 3803
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6182815551482124,
      "learning_rate": 1.0825097067320267e-06,
      "loss": 0.5284,
      "step": 3804
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.712285455157403,
      "learning_rate": 1.0792173640006743e-06,
      "loss": 0.5262,
      "step": 3805
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5977646591469041,
      "learning_rate": 1.0759297499619325e-06,
      "loss": 0.5228,
      "step": 3806
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6980260006944305,
      "learning_rate": 1.0726468663584877e-06,
      "loss": 0.5343,
      "step": 3807
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.571229215035153,
      "learning_rate": 1.0693687149305166e-06,
      "loss": 0.609,
      "step": 3808
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7422251671876188,
      "learning_rate": 1.0660952974156901e-06,
      "loss": 0.5993,
      "step": 3809
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.617997289883693,
      "learning_rate": 1.062826615549163e-06,
      "loss": 0.5401,
      "step": 3810
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5702990274301023,
      "learning_rate": 1.0595626710635931e-06,
      "loss": 0.5504,
      "step": 3811
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.543298184110017,
      "learning_rate": 1.0563034656891125e-06,
      "loss": 0.5527,
      "step": 3812
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.685995017421294,
      "learning_rate": 1.0530490011533489e-06,
      "loss": 0.5777,
      "step": 3813
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2839954213737386,
      "learning_rate": 1.0497992791814183e-06,
      "loss": 1.0338,
      "step": 3814
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6398088734740548,
      "learning_rate": 1.0465543014959168e-06,
      "loss": 0.6007,
      "step": 3815
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4427710729306162,
      "learning_rate": 1.0433140698169319e-06,
      "loss": 0.5306,
      "step": 3816
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7270783403779124,
      "learning_rate": 1.0400785858620321e-06,
      "loss": 0.5575,
      "step": 3817
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5685107522205037,
      "learning_rate": 1.0368478513462698e-06,
      "loss": 0.5906,
      "step": 3818
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.656577973854962,
      "learning_rate": 1.0336218679821785e-06,
      "loss": 0.5264,
      "step": 3819
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6241731555481644,
      "learning_rate": 1.030400637479777e-06,
      "loss": 0.5454,
      "step": 3820
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6808937166395759,
      "learning_rate": 1.0271841615465583e-06,
      "loss": 0.5541,
      "step": 3821
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.157240716175737,
      "learning_rate": 1.0239724418875018e-06,
      "loss": 1.0106,
      "step": 3822
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6086270244507048,
      "learning_rate": 1.0207654802050626e-06,
      "loss": 0.5436,
      "step": 3823
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.607050229372034,
      "learning_rate": 1.017563278199174e-06,
      "loss": 0.5411,
      "step": 3824
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6900155169105793,
      "learning_rate": 1.0143658375672461e-06,
      "loss": 0.5623,
      "step": 3825
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.292318560406564,
      "learning_rate": 1.0111731600041653e-06,
      "loss": 1.1002,
      "step": 3826
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6162381904655732,
      "learning_rate": 1.0079852472022934e-06,
      "loss": 0.5587,
      "step": 3827
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.234574306525831,
      "learning_rate": 1.0048021008514664e-06,
      "loss": 1.0488,
      "step": 3828
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.163621684028931,
      "learning_rate": 1.0016237226389946e-06,
      "loss": 1.0677,
      "step": 3829
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4689032034950655,
      "learning_rate": 9.984501142496582e-07,
      "loss": 0.5963,
      "step": 3830
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5977523979960429,
      "learning_rate": 9.952812773657139e-07,
      "loss": 0.5799,
      "step": 3831
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.3094473155554027,
      "learning_rate": 9.921172136668822e-07,
      "loss": 0.976,
      "step": 3832
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.582806604737998,
      "learning_rate": 9.889579248303593e-07,
      "loss": 1.0288,
      "step": 3833
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.733042409220257,
      "learning_rate": 9.85803412530808e-07,
      "loss": 0.5676,
      "step": 3834
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5270895950474217,
      "learning_rate": 9.826536784403595e-07,
      "loss": 0.5029,
      "step": 3835
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.284196504291957,
      "learning_rate": 9.795087242286116e-07,
      "loss": 1.036,
      "step": 3836
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6484801369248516,
      "learning_rate": 9.763685515626298e-07,
      "loss": 0.5665,
      "step": 3837
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.670187431832189,
      "learning_rate": 9.732331621069435e-07,
      "loss": 0.6297,
      "step": 3838
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6004654942682717,
      "learning_rate": 9.70102557523549e-07,
      "loss": 0.6306,
      "step": 3839
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5211761759322873,
      "learning_rate": 9.669767394719033e-07,
      "loss": 0.5383,
      "step": 3840
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5715500272640217,
      "learning_rate": 9.638557096089262e-07,
      "loss": 0.6051,
      "step": 3841
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7274524002857035,
      "learning_rate": 9.607394695890048e-07,
      "loss": 0.5704,
      "step": 3842
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.408975474662997,
      "learning_rate": 9.576280210639788e-07,
      "loss": 0.5904,
      "step": 3843
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6028110528120616,
      "learning_rate": 9.545213656831543e-07,
      "loss": 0.527,
      "step": 3844
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.559594462064534,
      "learning_rate": 9.514195050932929e-07,
      "loss": 0.5465,
      "step": 3845
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6611346078252258,
      "learning_rate": 9.483224409386182e-07,
      "loss": 0.5494,
      "step": 3846
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6002427789838638,
      "learning_rate": 9.452301748608095e-07,
      "loss": 0.5535,
      "step": 3847
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3747080796186433,
      "learning_rate": 9.42142708499002e-07,
      "loss": 0.5371,
      "step": 3848
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6442572007607816,
      "learning_rate": 9.39060043489789e-07,
      "loss": 0.6328,
      "step": 3849
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6006220633575707,
      "learning_rate": 9.359821814672143e-07,
      "loss": 0.5485,
      "step": 3850
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.2638071245085287,
      "learning_rate": 9.32909124062783e-07,
      "loss": 1.0828,
      "step": 3851
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4253716320645329,
      "learning_rate": 9.298408729054453e-07,
      "loss": 0.5356,
      "step": 3852
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5208777340933066,
      "learning_rate": 9.26777429621607e-07,
      "loss": 0.557,
      "step": 3853
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.180565763743012,
      "learning_rate": 9.237187958351279e-07,
      "loss": 1.0152,
      "step": 3854
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7290040907924484,
      "learning_rate": 9.206649731673168e-07,
      "loss": 0.5603,
      "step": 3855
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.403434838287273,
      "learning_rate": 9.176159632369297e-07,
      "loss": 0.5038,
      "step": 3856
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.423976769287019,
      "learning_rate": 9.14571767660174e-07,
      "loss": 1.0753,
      "step": 3857
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.2453494675194463,
      "learning_rate": 9.115323880507054e-07,
      "loss": 1.1271,
      "step": 3858
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6972372872325716,
      "learning_rate": 9.084978260196264e-07,
      "loss": 0.5488,
      "step": 3859
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.2002011247639737,
      "learning_rate": 9.054680831754847e-07,
      "loss": 1.0041,
      "step": 3860
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7478583861882495,
      "learning_rate": 9.024431611242746e-07,
      "loss": 0.5485,
      "step": 3861
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4429525499291895,
      "learning_rate": 8.994230614694354e-07,
      "loss": 0.5381,
      "step": 3862
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.63531681223399,
      "learning_rate": 8.96407785811847e-07,
      "loss": 0.5242,
      "step": 3863
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3725152278946573,
      "learning_rate": 8.93397335749836e-07,
      "loss": 0.4561,
      "step": 3864
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5479360161429732,
      "learning_rate": 8.903917128791695e-07,
      "loss": 0.5708,
      "step": 3865
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6759081687005088,
      "learning_rate": 8.873909187930563e-07,
      "loss": 0.62,
      "step": 3866
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4872893410177104,
      "learning_rate": 8.843949550821451e-07,
      "loss": 0.5716,
      "step": 3867
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5943688021650566,
      "learning_rate": 8.814038233345234e-07,
      "loss": 0.5813,
      "step": 3868
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5164003509764574,
      "learning_rate": 8.784175251357208e-07,
      "loss": 0.5392,
      "step": 3869
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6179783813221726,
      "learning_rate": 8.754360620686997e-07,
      "loss": 0.555,
      "step": 3870
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.217737214888175,
      "learning_rate": 8.724594357138627e-07,
      "loss": 1.0559,
      "step": 3871
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5750999457121129,
      "learning_rate": 8.694876476490488e-07,
      "loss": 0.5556,
      "step": 3872
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.303604165491023,
      "learning_rate": 8.665206994495324e-07,
      "loss": 1.0893,
      "step": 3873
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.520659117086607,
      "learning_rate": 8.635585926880185e-07,
      "loss": 0.457,
      "step": 3874
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.650151845711895,
      "learning_rate": 8.606013289346505e-07,
      "loss": 0.5948,
      "step": 3875
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5155802561688956,
      "learning_rate": 8.576489097570017e-07,
      "loss": 0.5824,
      "step": 3876
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5711017107407956,
      "learning_rate": 8.547013367200807e-07,
      "loss": 0.5325,
      "step": 3877
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.54866062327102,
      "learning_rate": 8.51758611386323e-07,
      "loss": 0.5179,
      "step": 3878
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.350295272227548,
      "learning_rate": 8.488207353155986e-07,
      "loss": 0.9977,
      "step": 3879
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5928623087376694,
      "learning_rate": 8.45887710065203e-07,
      "loss": 0.6234,
      "step": 3880
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.4540464630512027,
      "learning_rate": 8.429595371898646e-07,
      "loss": 1.0346,
      "step": 3881
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6177245406653724,
      "learning_rate": 8.400362182417355e-07,
      "loss": 0.5079,
      "step": 3882
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.200828629605776,
      "learning_rate": 8.371177547704001e-07,
      "loss": 1.0333,
      "step": 3883
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7090353195574124,
      "learning_rate": 8.342041483228613e-07,
      "loss": 0.571,
      "step": 3884
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4259187413306456,
      "learning_rate": 8.312954004435536e-07,
      "loss": 0.5053,
      "step": 3885
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.62672323322633,
      "learning_rate": 8.283915126743358e-07,
      "loss": 0.563,
      "step": 3886
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4999092754373484,
      "learning_rate": 8.254924865544878e-07,
      "loss": 0.5556,
      "step": 3887
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7293542361822711,
      "learning_rate": 8.225983236207136e-07,
      "loss": 0.526,
      "step": 3888
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6584030948522748,
      "learning_rate": 8.197090254071405e-07,
      "loss": 0.5651,
      "step": 3889
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7160319087330294,
      "learning_rate": 8.168245934453156e-07,
      "loss": 0.5332,
      "step": 3890
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.71068949031656,
      "learning_rate": 8.139450292642059e-07,
      "loss": 0.6389,
      "step": 3891
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.434538085359495,
      "learning_rate": 8.110703343902005e-07,
      "loss": 0.5285,
      "step": 3892
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.3061478289098494,
      "learning_rate": 8.08200510347108e-07,
      "loss": 1.0739,
      "step": 3893
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6043954968189171,
      "learning_rate": 8.053355586561496e-07,
      "loss": 0.4705,
      "step": 3894
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.30071900515991,
      "learning_rate": 8.024754808359681e-07,
      "loss": 1.0621,
      "step": 3895
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7107498383185455,
      "learning_rate": 7.996202784026219e-07,
      "loss": 0.5776,
      "step": 3896
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4663699950048592,
      "learning_rate": 7.967699528695871e-07,
      "loss": 0.5309,
      "step": 3897
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6992491682418742,
      "learning_rate": 7.939245057477518e-07,
      "loss": 0.5891,
      "step": 3898
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.71634075952433,
      "learning_rate": 7.910839385454183e-07,
      "loss": 0.5396,
      "step": 3899
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5510904542710686,
      "learning_rate": 7.882482527683055e-07,
      "loss": 0.5268,
      "step": 3900
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6054774650942443,
      "learning_rate": 7.854174499195389e-07,
      "loss": 0.527,
      "step": 3901
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5094061644334662,
      "learning_rate": 7.825915314996624e-07,
      "loss": 0.5233,
      "step": 3902
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.69705398935663,
      "learning_rate": 7.797704990066246e-07,
      "loss": 0.571,
      "step": 3903
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4642910215703204,
      "learning_rate": 7.769543539357904e-07,
      "loss": 0.5583,
      "step": 3904
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.701451263337144,
      "learning_rate": 7.74143097779928e-07,
      "loss": 0.6357,
      "step": 3905
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.4184937087597342,
      "learning_rate": 7.713367320292165e-07,
      "loss": 1.0371,
      "step": 3906
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.635313424995491,
      "learning_rate": 7.685352581712447e-07,
      "loss": 0.5187,
      "step": 3907
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5269742904411598,
      "learning_rate": 7.657386776910048e-07,
      "loss": 0.6119,
      "step": 3908
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6086580803064505,
      "learning_rate": 7.629469920708987e-07,
      "loss": 0.5437,
      "step": 3909
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7320046322886236,
      "learning_rate": 7.601602027907318e-07,
      "loss": 0.5063,
      "step": 3910
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7153220640406117,
      "learning_rate": 7.573783113277122e-07,
      "loss": 0.5926,
      "step": 3911
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7294349439169887,
      "learning_rate": 7.54601319156456e-07,
      "loss": 0.6007,
      "step": 3912
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.475059863027769,
      "learning_rate": 7.518292277489791e-07,
      "loss": 1.0279,
      "step": 3913
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4023638185984895,
      "learning_rate": 7.490620385747016e-07,
      "loss": 0.5172,
      "step": 3914
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6256204113546908,
      "learning_rate": 7.462997531004423e-07,
      "loss": 0.5614,
      "step": 3915
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.6443434881997083,
      "learning_rate": 7.435423727904223e-07,
      "loss": 1.1003,
      "step": 3916
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.3251830289054083,
      "learning_rate": 7.407898991062646e-07,
      "loss": 1.0868,
      "step": 3917
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4457466061129487,
      "learning_rate": 7.380423335069875e-07,
      "loss": 0.534,
      "step": 3918
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5952060094939853,
      "learning_rate": 7.352996774490096e-07,
      "loss": 0.556,
      "step": 3919
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.241474522995443,
      "learning_rate": 7.325619323861488e-07,
      "loss": 1.0444,
      "step": 3920
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5038298786866846,
      "learning_rate": 7.298290997696167e-07,
      "loss": 0.605,
      "step": 3921
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6864485626454528,
      "learning_rate": 7.271011810480222e-07,
      "loss": 0.5282,
      "step": 3922
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6495054698037284,
      "learning_rate": 7.243781776673698e-07,
      "loss": 0.6072,
      "step": 3923
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7068105514813292,
      "learning_rate": 7.216600910710581e-07,
      "loss": 0.6249,
      "step": 3924
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7863347325856596,
      "learning_rate": 7.189469226998813e-07,
      "loss": 0.6333,
      "step": 3925
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6373566527143033,
      "learning_rate": 7.162386739920225e-07,
      "loss": 0.4561,
      "step": 3926
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6399802627035722,
      "learning_rate": 7.135353463830585e-07,
      "loss": 0.5351,
      "step": 3927
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5685141490474528,
      "learning_rate": 7.108369413059602e-07,
      "loss": 0.6034,
      "step": 3928
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.791151328014157,
      "learning_rate": 7.081434601910863e-07,
      "loss": 0.5794,
      "step": 3929
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.4063342461396466,
      "learning_rate": 7.054549044661874e-07,
      "loss": 1.1071,
      "step": 3930
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2679140786889347,
      "learning_rate": 7.027712755564009e-07,
      "loss": 1.0965,
      "step": 3931
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7664724863738588,
      "learning_rate": 7.000925748842557e-07,
      "loss": 0.6357,
      "step": 3932
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5633437571042021,
      "learning_rate": 6.974188038696661e-07,
      "loss": 0.5837,
      "step": 3933
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6260183011545841,
      "learning_rate": 6.947499639299337e-07,
      "loss": 0.5276,
      "step": 3934
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.303566434947497,
      "learning_rate": 6.920860564797483e-07,
      "loss": 1.063,
      "step": 3935
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8533443722991605,
      "learning_rate": 6.8942708293118e-07,
      "loss": 0.6084,
      "step": 3936
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6736394652089635,
      "learning_rate": 6.867730446936871e-07,
      "loss": 0.549,
      "step": 3937
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.692640211824006,
      "learning_rate": 6.841239431741143e-07,
      "loss": 0.5277,
      "step": 3938
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5373923256322974,
      "learning_rate": 6.814797797766837e-07,
      "loss": 0.5023,
      "step": 3939
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.329133781240707,
      "learning_rate": 6.788405559030053e-07,
      "loss": 1.0854,
      "step": 3940
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.2580442273992705,
      "learning_rate": 6.762062729520669e-07,
      "loss": 1.1023,
      "step": 3941
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8588150337757334,
      "learning_rate": 6.735769323202379e-07,
      "loss": 0.5662,
      "step": 3942
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.282455611489101,
      "learning_rate": 6.709525354012714e-07,
      "loss": 1.0434,
      "step": 3943
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.412116364022406,
      "learning_rate": 6.683330835862933e-07,
      "loss": 0.5121,
      "step": 3944
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.2250298694619017,
      "learning_rate": 6.657185782638176e-07,
      "loss": 1.017,
      "step": 3945
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6500738275078253,
      "learning_rate": 6.631090208197243e-07,
      "loss": 0.5558,
      "step": 3946
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.3581436678610297,
      "learning_rate": 6.605044126372817e-07,
      "loss": 1.0824,
      "step": 3947
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5821748360566137,
      "learning_rate": 6.579047550971273e-07,
      "loss": 0.5268,
      "step": 3948
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5255234036046665,
      "learning_rate": 6.55310049577279e-07,
      "loss": 0.5358,
      "step": 3949
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4282065636434778,
      "learning_rate": 6.527202974531277e-07,
      "loss": 0.4986,
      "step": 3950
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7506977011426519,
      "learning_rate": 6.501355000974374e-07,
      "loss": 0.5926,
      "step": 3951
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.384771410081724,
      "learning_rate": 6.475556588803477e-07,
      "loss": 1.0622,
      "step": 3952
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.782243466446922,
      "learning_rate": 6.449807751693715e-07,
      "loss": 0.5198,
      "step": 3953
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7906544544662752,
      "learning_rate": 6.424108503293924e-07,
      "loss": 0.5939,
      "step": 3954
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4169709877571517,
      "learning_rate": 6.398458857226642e-07,
      "loss": 0.483,
      "step": 3955
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.159151442402564,
      "learning_rate": 6.372858827088168e-07,
      "loss": 1.1114,
      "step": 3956
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.117326236587264,
      "learning_rate": 6.347308426448429e-07,
      "loss": 1.0563,
      "step": 3957
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7163817314351177,
      "learning_rate": 6.321807668851077e-07,
      "loss": 0.5259,
      "step": 3958
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.2685938022554306,
      "learning_rate": 6.296356567813466e-07,
      "loss": 1.0386,
      "step": 3959
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.603422758500972,
      "learning_rate": 6.270955136826618e-07,
      "loss": 0.569,
      "step": 3960
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3956807266567461,
      "learning_rate": 6.245603389355204e-07,
      "loss": 0.4998,
      "step": 3961
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5224003541258166,
      "learning_rate": 6.220301338837587e-07,
      "loss": 0.5229,
      "step": 3962
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6223476558194587,
      "learning_rate": 6.195048998685771e-07,
      "loss": 0.576,
      "step": 3963
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6562532878857004,
      "learning_rate": 6.169846382285427e-07,
      "loss": 0.5348,
      "step": 3964
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7383093343876121,
      "learning_rate": 6.144693502995846e-07,
      "loss": 0.5831,
      "step": 3965
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6591086509165656,
      "learning_rate": 6.119590374149964e-07,
      "loss": 0.5566,
      "step": 3966
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.417723632466198,
      "learning_rate": 6.094537009054368e-07,
      "loss": 1.0052,
      "step": 3967
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6695794275609055,
      "learning_rate": 6.069533420989215e-07,
      "loss": 0.5882,
      "step": 3968
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7308891232358543,
      "learning_rate": 6.044579623208313e-07,
      "loss": 0.5725,
      "step": 3969
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.549919716984352,
      "learning_rate": 6.019675628939082e-07,
      "loss": 0.5434,
      "step": 3970
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7200947171914287,
      "learning_rate": 5.994821451382527e-07,
      "loss": 0.5424,
      "step": 3971
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8326872558900378,
      "learning_rate": 5.970017103713243e-07,
      "loss": 0.6268,
      "step": 3972
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5838430919682613,
      "learning_rate": 5.945262599079438e-07,
      "loss": 0.5482,
      "step": 3973
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.627573929107591,
      "learning_rate": 5.920557950602856e-07,
      "loss": 0.5895,
      "step": 3974
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5275167455322791,
      "learning_rate": 5.895903171378858e-07,
      "loss": 0.5409,
      "step": 3975
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5626117002831938,
      "learning_rate": 5.871298274476345e-07,
      "loss": 0.5481,
      "step": 3976
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8079558387608723,
      "learning_rate": 5.846743272937771e-07,
      "loss": 0.5244,
      "step": 3977
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5049005935904036,
      "learning_rate": 5.822238179779161e-07,
      "loss": 0.5103,
      "step": 3978
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.60120599097908,
      "learning_rate": 5.797783007990077e-07,
      "loss": 0.5305,
      "step": 3979
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6592084810685042,
      "learning_rate": 5.773377770533617e-07,
      "loss": 0.614,
      "step": 3980
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5771973754521649,
      "learning_rate": 5.7490224803464e-07,
      "loss": 0.5349,
      "step": 3981
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7504528112348965,
      "learning_rate": 5.7247171503386e-07,
      "loss": 0.5284,
      "step": 3982
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.389848811686538,
      "learning_rate": 5.700461793393886e-07,
      "loss": 1.0179,
      "step": 3983
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3951659434235473,
      "learning_rate": 5.676256422369419e-07,
      "loss": 0.4979,
      "step": 3984
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.29316544086117,
      "learning_rate": 5.652101050095915e-07,
      "loss": 1.0753,
      "step": 3985
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7235956447350218,
      "learning_rate": 5.62799568937753e-07,
      "loss": 0.535,
      "step": 3986
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5545160753833445,
      "learning_rate": 5.603940352991977e-07,
      "loss": 0.5722,
      "step": 3987
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6533193738074776,
      "learning_rate": 5.579935053690355e-07,
      "loss": 0.5896,
      "step": 3988
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.630158061108452,
      "learning_rate": 5.555979804197331e-07,
      "loss": 0.5645,
      "step": 3989
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.721264852717106,
      "learning_rate": 5.532074617211003e-07,
      "loss": 0.5155,
      "step": 3990
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6898013213156706,
      "learning_rate": 5.508219505402934e-07,
      "loss": 0.5152,
      "step": 3991
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6356975894315051,
      "learning_rate": 5.484414481418143e-07,
      "loss": 0.5443,
      "step": 3992
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.096563804875982,
      "learning_rate": 5.460659557875104e-07,
      "loss": 1.0132,
      "step": 3993
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.631099460660492,
      "learning_rate": 5.43695474736573e-07,
      "loss": 0.5566,
      "step": 3994
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6558463335632392,
      "learning_rate": 5.413300062455385e-07,
      "loss": 0.5287,
      "step": 3995
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.338078217904769,
      "learning_rate": 5.389695515682835e-07,
      "loss": 1.0664,
      "step": 3996
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5535948588836659,
      "learning_rate": 5.366141119560286e-07,
      "loss": 0.4942,
      "step": 3997
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.308780466804689,
      "learning_rate": 5.342636886573372e-07,
      "loss": 1.0017,
      "step": 3998
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6296427315142832,
      "learning_rate": 5.31918282918109e-07,
      "loss": 0.5799,
      "step": 3999
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.697354295797498,
      "learning_rate": 5.2957789598159e-07,
      "loss": 0.5714,
      "step": 4000
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.4127149345404715,
      "learning_rate": 5.272425290883609e-07,
      "loss": 1.0773,
      "step": 4001
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5240557327015563,
      "learning_rate": 5.24912183476346e-07,
      "loss": 0.5265,
      "step": 4002
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6705624385778723,
      "learning_rate": 5.225868603808049e-07,
      "loss": 0.5159,
      "step": 4003
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5073484828868167,
      "learning_rate": 5.202665610343338e-07,
      "loss": 0.6013,
      "step": 4004
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.209142514360723,
      "learning_rate": 5.179512866668702e-07,
      "loss": 1.0382,
      "step": 4005
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.2482318840549755,
      "learning_rate": 5.156410385056831e-07,
      "loss": 1.0854,
      "step": 4006
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5759590175704385,
      "learning_rate": 5.133358177753823e-07,
      "loss": 0.53,
      "step": 4007
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5698694917686635,
      "learning_rate": 5.110356256979066e-07,
      "loss": 0.6073,
      "step": 4008
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3815539085049278,
      "learning_rate": 5.087404634925353e-07,
      "loss": 0.5631,
      "step": 4009
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.2824797627898348,
      "learning_rate": 5.064503323758774e-07,
      "loss": 0.976,
      "step": 4010
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5052942413754666,
      "learning_rate": 5.041652335618763e-07,
      "loss": 0.5668,
      "step": 4011
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6212901594356128,
      "learning_rate": 5.018851682618087e-07,
      "loss": 0.5244,
      "step": 4012
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5937654475739855,
      "learning_rate": 4.996101376842821e-07,
      "loss": 0.5663,
      "step": 4013
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.125087228613564,
      "learning_rate": 4.973401430352353e-07,
      "loss": 1.0577,
      "step": 4014
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.768074965187177,
      "learning_rate": 4.950751855179381e-07,
      "loss": 0.5567,
      "step": 4015
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.284726361577574,
      "learning_rate": 4.928152663329911e-07,
      "loss": 1.1235,
      "step": 4016
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.476056683137627,
      "learning_rate": 4.905603866783216e-07,
      "loss": 1.097,
      "step": 4017
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7958119115305817,
      "learning_rate": 4.8831054774919e-07,
      "loss": 0.5906,
      "step": 4018
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.413949464992031,
      "learning_rate": 4.860657507381772e-07,
      "loss": 1.0626,
      "step": 4019
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4281459688717564,
      "learning_rate": 4.838259968351999e-07,
      "loss": 0.561,
      "step": 4020
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5129542353567755,
      "learning_rate": 4.815912872274963e-07,
      "loss": 0.5524,
      "step": 4021
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6464899042393726,
      "learning_rate": 4.793616230996323e-07,
      "loss": 0.6312,
      "step": 4022
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.310872431432514,
      "learning_rate": 4.771370056335001e-07,
      "loss": 1.0613,
      "step": 4023
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6538820603952191,
      "learning_rate": 4.749174360083153e-07,
      "loss": 0.5855,
      "step": 4024
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6118020686736902,
      "learning_rate": 4.727029154006191e-07,
      "loss": 0.577,
      "step": 4025
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6761269200171518,
      "learning_rate": 4.7049344498427465e-07,
      "loss": 0.5813,
      "step": 4026
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5784982645531471,
      "learning_rate": 4.6828902593046953e-07,
      "loss": 0.5215,
      "step": 4027
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6912225133684309,
      "learning_rate": 4.6608965940771464e-07,
      "loss": 0.5378,
      "step": 4028
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6032361235020705,
      "learning_rate": 4.6389534658184077e-07,
      "loss": 0.5657,
      "step": 4029
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2155807553004223,
      "learning_rate": 4.6170608861599965e-07,
      "loss": 1.0365,
      "step": 4030
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5516598624956968,
      "learning_rate": 4.59521886670663e-07,
      "loss": 0.5442,
      "step": 4031
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6305009229666672,
      "learning_rate": 4.573427419036269e-07,
      "loss": 0.5659,
      "step": 4032
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8334743994570843,
      "learning_rate": 4.5516865547000186e-07,
      "loss": 0.6214,
      "step": 4033
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5855715349179649,
      "learning_rate": 4.5299962852221934e-07,
      "loss": 0.5499,
      "step": 4034
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7457472293067844,
      "learning_rate": 4.508356622100274e-07,
      "loss": 0.5966,
      "step": 4035
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.256098287676595,
      "learning_rate": 4.486767576804951e-07,
      "loss": 1.0602,
      "step": 4036
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.558538664684484,
      "learning_rate": 4.465229160780027e-07,
      "loss": 0.5339,
      "step": 4037
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5412935898121323,
      "learning_rate": 4.443741385442535e-07,
      "loss": 0.4717,
      "step": 4038
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6963593661526122,
      "learning_rate": 4.422304262182586e-07,
      "loss": 0.5547,
      "step": 4039
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9031112345607344,
      "learning_rate": 4.4009178023635355e-07,
      "loss": 0.5878,
      "step": 4040
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.3409391200394425,
      "learning_rate": 4.3795820173217816e-07,
      "loss": 1.0997,
      "step": 4041
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5538444290446114,
      "learning_rate": 4.3582969183669443e-07,
      "loss": 0.5767,
      "step": 4042
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5752594304815635,
      "learning_rate": 4.3370625167817205e-07,
      "loss": 0.5212,
      "step": 4043
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.248541145070551,
      "learning_rate": 4.3158788238219726e-07,
      "loss": 1.1649,
      "step": 4044
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6700405996059822,
      "learning_rate": 4.294745850716664e-07,
      "loss": 0.5396,
      "step": 4045
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6841624488471023,
      "learning_rate": 4.2736636086678774e-07,
      "loss": 0.5058,
      "step": 4046
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4360098625875923,
      "learning_rate": 4.252632108850796e-07,
      "loss": 0.4801,
      "step": 4047
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6864979713150088,
      "learning_rate": 4.2316513624137245e-07,
      "loss": 0.6262,
      "step": 4048
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.548529689578951,
      "learning_rate": 4.2107213804780556e-07,
      "loss": 1.1256,
      "step": 4049
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6624818046336913,
      "learning_rate": 4.1898421741382367e-07,
      "loss": 0.5483,
      "step": 4050
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4561967912051985,
      "learning_rate": 4.1690137544618815e-07,
      "loss": 0.5621,
      "step": 4051
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8007023725024156,
      "learning_rate": 4.1482361324895916e-07,
      "loss": 0.5713,
      "step": 4052
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6337534001263936,
      "learning_rate": 4.127509319235112e-07,
      "loss": 0.5762,
      "step": 4053
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6304519718144044,
      "learning_rate": 4.106833325685222e-07,
      "loss": 0.5205,
      "step": 4054
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5787798150295502,
      "learning_rate": 4.0862081627997653e-07,
      "loss": 0.5702,
      "step": 4055
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2990877021343343,
      "learning_rate": 4.0656338415116404e-07,
      "loss": 1.0699,
      "step": 4056
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5386120033554516,
      "learning_rate": 4.0451103727268237e-07,
      "loss": 0.5265,
      "step": 4057
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.3717618588498097,
      "learning_rate": 4.024637767324291e-07,
      "loss": 1.0224,
      "step": 4058
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5768219575050118,
      "learning_rate": 4.004216036156083e-07,
      "loss": 0.5394,
      "step": 4059
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6101992472342674,
      "learning_rate": 3.9838451900472865e-07,
      "loss": 0.5496,
      "step": 4060
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6303902591441637,
      "learning_rate": 3.963525239795962e-07,
      "loss": 0.5298,
      "step": 4061
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8214663520477488,
      "learning_rate": 3.943256196173262e-07,
      "loss": 0.6299,
      "step": 4062
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.3030988035022752,
      "learning_rate": 3.9230380699233015e-07,
      "loss": 1.1258,
      "step": 4063
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.435795851225698,
      "learning_rate": 3.9028708717632204e-07,
      "loss": 0.527,
      "step": 4064
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.502129711784549,
      "learning_rate": 3.8827546123831773e-07,
      "loss": 0.5389,
      "step": 4065
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6925358430874764,
      "learning_rate": 3.862689302446321e-07,
      "loss": 0.5599,
      "step": 4066
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.632228157510243,
      "learning_rate": 3.842674952588776e-07,
      "loss": 0.5724,
      "step": 4067
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5488271092858277,
      "learning_rate": 3.8227115734196796e-07,
      "loss": 0.6067,
      "step": 4068
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.439557777676103,
      "learning_rate": 3.802799175521155e-07,
      "loss": 0.547,
      "step": 4069
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6435452224558358,
      "learning_rate": 3.782937769448225e-07,
      "loss": 0.5575,
      "step": 4070
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6896917710951924,
      "learning_rate": 3.7631273657290247e-07,
      "loss": 0.6191,
      "step": 4071
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7189996949873612,
      "learning_rate": 3.74336797486452e-07,
      "loss": 0.5495,
      "step": 4072
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5696464488259505,
      "learning_rate": 3.7236596073287e-07,
      "loss": 0.5195,
      "step": 4073
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.79522003230891,
      "learning_rate": 3.7040022735685076e-07,
      "loss": 0.5304,
      "step": 4074
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5960609678705724,
      "learning_rate": 3.684395984003808e-07,
      "loss": 0.5818,
      "step": 4075
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6155267635813566,
      "learning_rate": 3.664840749027443e-07,
      "loss": 0.5622,
      "step": 4076
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6413636364923567,
      "learning_rate": 3.6453365790051543e-07,
      "loss": 0.5141,
      "step": 4077
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6805794638207314,
      "learning_rate": 3.6258834842756387e-07,
      "loss": 0.5728,
      "step": 4078
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6095533152565424,
      "learning_rate": 3.606481475150536e-07,
      "loss": 0.5541,
      "step": 4079
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.528339146655136,
      "learning_rate": 3.5871305619143647e-07,
      "loss": 0.5199,
      "step": 4080
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7224042543360147,
      "learning_rate": 3.567830754824564e-07,
      "loss": 0.5703,
      "step": 4081
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6301371841512233,
      "learning_rate": 3.5485820641115386e-07,
      "loss": 0.5416,
      "step": 4082
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4562562225774922,
      "learning_rate": 3.5293844999785163e-07,
      "loss": 0.5227,
      "step": 4083
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6732735988764165,
      "learning_rate": 3.5102380726016905e-07,
      "loss": 0.5043,
      "step": 4084
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.614726632008174,
      "learning_rate": 3.49114279213012e-07,
      "loss": 0.5448,
      "step": 4085
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7285536289789545,
      "learning_rate": 3.4720986686857415e-07,
      "loss": 0.602,
      "step": 4086
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.3311904256475824,
      "learning_rate": 3.4531057123634026e-07,
      "loss": 1.0168,
      "step": 4087
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6487700806883911,
      "learning_rate": 3.434163933230794e-07,
      "loss": 0.5922,
      "step": 4088
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6496105422215372,
      "learning_rate": 3.4152733413285175e-07,
      "loss": 0.4866,
      "step": 4089
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6169652185539403,
      "learning_rate": 3.3964339466700194e-07,
      "loss": 0.5277,
      "step": 4090
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4352989200896475,
      "learning_rate": 3.37764575924161e-07,
      "loss": 0.5542,
      "step": 4091
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7373828294398495,
      "learning_rate": 3.3589087890024354e-07,
      "loss": 0.5242,
      "step": 4092
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.3723374596183047,
      "learning_rate": 3.34022304588455e-07,
      "loss": 1.0549,
      "step": 4093
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.796685271591381,
      "learning_rate": 3.321588539792775e-07,
      "loss": 0.5679,
      "step": 4094
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6040203442893846,
      "learning_rate": 3.303005280604854e-07,
      "loss": 0.6042,
      "step": 4095
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4644062345635411,
      "learning_rate": 3.284473278171296e-07,
      "loss": 0.4959,
      "step": 4096
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6482829888342851,
      "learning_rate": 3.265992542315488e-07,
      "loss": 0.5652,
      "step": 4097
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.570306740148652,
      "learning_rate": 3.247563082833627e-07,
      "loss": 0.5482,
      "step": 4098
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5737520035733674,
      "learning_rate": 3.229184909494709e-07,
      "loss": 0.5795,
      "step": 4099
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6197056662424845,
      "learning_rate": 3.210858032040587e-07,
      "loss": 0.5627,
      "step": 4100
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.261219855392241,
      "learning_rate": 3.1925824601858555e-07,
      "loss": 1.0363,
      "step": 4101
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.771729643606383,
      "learning_rate": 3.1743582036180223e-07,
      "loss": 0.6844,
      "step": 4102
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6091243849652934,
      "learning_rate": 3.1561852719972695e-07,
      "loss": 0.5408,
      "step": 4103
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7876078069253987,
      "learning_rate": 3.1380636749566487e-07,
      "loss": 0.5689,
      "step": 4104
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7115231393018608,
      "learning_rate": 3.119993422101997e-07,
      "loss": 0.5953,
      "step": 4105
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.172033479395587,
      "learning_rate": 3.101974523011908e-07,
      "loss": 1.0395,
      "step": 4106
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.288526867985088,
      "learning_rate": 3.0840069872377756e-07,
      "loss": 1.123,
      "step": 4107
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.59486768888426,
      "learning_rate": 3.0660908243037466e-07,
      "loss": 0.6138,
      "step": 4108
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4812537223183668,
      "learning_rate": 3.048226043706759e-07,
      "loss": 0.4892,
      "step": 4109
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5025271930289792,
      "learning_rate": 3.0304126549165035e-07,
      "loss": 0.5662,
      "step": 4110
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7223932024133903,
      "learning_rate": 3.0126506673754387e-07,
      "loss": 0.5402,
      "step": 4111
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6163236977811282,
      "learning_rate": 2.994940090498732e-07,
      "loss": 0.5167,
      "step": 4112
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5817874731406059,
      "learning_rate": 2.977280933674387e-07,
      "loss": 0.5762,
      "step": 4113
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1874845561716385,
      "learning_rate": 2.9596732062630584e-07,
      "loss": 1.0263,
      "step": 4114
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.53653387284499,
      "learning_rate": 2.9421169175981924e-07,
      "loss": 0.5298,
      "step": 4115
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.215382818042579,
      "learning_rate": 2.9246120769859444e-07,
      "loss": 0.9363,
      "step": 4116
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6817263865910967,
      "learning_rate": 2.9071586937052365e-07,
      "loss": 0.5599,
      "step": 4117
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.500493819830851,
      "learning_rate": 2.8897567770076685e-07,
      "loss": 0.563,
      "step": 4118
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5802855703251546,
      "learning_rate": 2.8724063361175725e-07,
      "loss": 0.5921,
      "step": 4119
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.399750698001709,
      "learning_rate": 2.855107380232025e-07,
      "loss": 1.0992,
      "step": 4120
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2415680918932654,
      "learning_rate": 2.837859918520758e-07,
      "loss": 1.0545,
      "step": 4121
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.634951839493458,
      "learning_rate": 2.82066396012628e-07,
      "loss": 0.5949,
      "step": 4122
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5820530062878302,
      "learning_rate": 2.803519514163688e-07,
      "loss": 0.592,
      "step": 4123
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5510504885169882,
      "learning_rate": 2.786426589720914e-07,
      "loss": 0.5684,
      "step": 4124
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6676103010121222,
      "learning_rate": 2.769385195858454e-07,
      "loss": 0.5777,
      "step": 4125
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6474202107179217,
      "learning_rate": 2.752395341609559e-07,
      "loss": 0.5303,
      "step": 4126
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6931277546886982,
      "learning_rate": 2.7354570359801494e-07,
      "loss": 0.5263,
      "step": 4127
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.314906668590132,
      "learning_rate": 2.718570287948807e-07,
      "loss": 1.1141,
      "step": 4128
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6527092111216937,
      "learning_rate": 2.701735106466796e-07,
      "loss": 0.4756,
      "step": 4129
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.328455870315491,
      "learning_rate": 2.684951500458033e-07,
      "loss": 1.1467,
      "step": 4130
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5812139626042345,
      "learning_rate": 2.668219478819134e-07,
      "loss": 0.5773,
      "step": 4131
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8614879335770151,
      "learning_rate": 2.651539050419283e-07,
      "loss": 0.5817,
      "step": 4132
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7547039970144402,
      "learning_rate": 2.6349102241004287e-07,
      "loss": 0.5342,
      "step": 4133
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2947358778386557,
      "learning_rate": 2.6183330086770766e-07,
      "loss": 0.9953,
      "step": 4134
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.728214250064144,
      "learning_rate": 2.6018074129364523e-07,
      "loss": 0.6165,
      "step": 4135
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6374516603998204,
      "learning_rate": 2.585333445638338e-07,
      "loss": 0.5272,
      "step": 4136
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.564459946757823,
      "learning_rate": 2.568911115515205e-07,
      "loss": 0.5911,
      "step": 4137
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6424591659804715,
      "learning_rate": 2.5525404312721325e-07,
      "loss": 0.5152,
      "step": 4138
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2699084207124276,
      "learning_rate": 2.5362214015868246e-07,
      "loss": 1.0995,
      "step": 4139
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.601010247282857,
      "learning_rate": 2.5199540351096064e-07,
      "loss": 0.573,
      "step": 4140
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6640676956558682,
      "learning_rate": 2.503738340463424e-07,
      "loss": 0.5603,
      "step": 4141
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6759596830117116,
      "learning_rate": 2.487574326243847e-07,
      "loss": 0.5574,
      "step": 4142
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.445172039941766,
      "learning_rate": 2.471462001018976e-07,
      "loss": 1.0605,
      "step": 4143
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.3031928146108966,
      "learning_rate": 2.4554013733296244e-07,
      "loss": 1.0352,
      "step": 4144
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5936513919665196,
      "learning_rate": 2.4393924516891157e-07,
      "loss": 0.5248,
      "step": 4145
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5438808150780452,
      "learning_rate": 2.4234352445833944e-07,
      "loss": 0.5802,
      "step": 4146
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5845823899142164,
      "learning_rate": 2.4075297604710057e-07,
      "loss": 0.5704,
      "step": 4147
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2195234499162666,
      "learning_rate": 2.3916760077830503e-07,
      "loss": 1.0846,
      "step": 4148
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.501276439481086,
      "learning_rate": 2.375873994923239e-07,
      "loss": 0.4955,
      "step": 4149
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5295966938316803,
      "learning_rate": 2.360123730267827e-07,
      "loss": 0.5483,
      "step": 4150
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6323015513386177,
      "learning_rate": 2.3444252221656472e-07,
      "loss": 0.5455,
      "step": 4151
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7238542641431032,
      "learning_rate": 2.3287784789381095e-07,
      "loss": 0.5903,
      "step": 4152
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8048783905476187,
      "learning_rate": 2.313183508879191e-07,
      "loss": 0.5962,
      "step": 4153
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.637035069612989,
      "learning_rate": 2.2976403202553789e-07,
      "loss": 0.5772,
      "step": 4154
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7282838376812082,
      "learning_rate": 2.2821489213057824e-07,
      "loss": 0.5401,
      "step": 4155
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6166037306782461,
      "learning_rate": 2.2667093202419777e-07,
      "loss": 0.5357,
      "step": 4156
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.20753615230876,
      "learning_rate": 2.2513215252481513e-07,
      "loss": 1.0318,
      "step": 4157
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4432004601824835,
      "learning_rate": 2.2359855444810008e-07,
      "loss": 0.5102,
      "step": 4158
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6337015069199927,
      "learning_rate": 2.220701386069768e-07,
      "loss": 0.5435,
      "step": 4159
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.661053455356896,
      "learning_rate": 2.2054690581161942e-07,
      "loss": 0.4973,
      "step": 4160
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7632290670952502,
      "learning_rate": 2.1902885686945984e-07,
      "loss": 0.5689,
      "step": 4161
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.301224285412274,
      "learning_rate": 2.1751599258517885e-07,
      "loss": 1.0541,
      "step": 4162
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2468849888542226,
      "learning_rate": 2.160083137607072e-07,
      "loss": 1.0832,
      "step": 4163
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6950118879607108,
      "learning_rate": 2.1450582119523221e-07,
      "loss": 0.6531,
      "step": 4164
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8228708750976055,
      "learning_rate": 2.1300851568518577e-07,
      "loss": 0.5656,
      "step": 4165
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5751113608549492,
      "learning_rate": 2.115163980242574e-07,
      "loss": 0.5341,
      "step": 4166
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6096948795403947,
      "learning_rate": 2.1002946900338105e-07,
      "loss": 0.5028,
      "step": 4167
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6991839498845418,
      "learning_rate": 2.0854772941074075e-07,
      "loss": 0.64,
      "step": 4168
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1712526142024764,
      "learning_rate": 2.0707118003177264e-07,
      "loss": 1.0712,
      "step": 4169
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6322274904141327,
      "learning_rate": 2.055998216491595e-07,
      "loss": 0.6046,
      "step": 4170
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.63555357963855,
      "learning_rate": 2.041336550428341e-07,
      "loss": 0.5061,
      "step": 4171
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.231189516727651,
      "learning_rate": 2.0267268098997485e-07,
      "loss": 1.0374,
      "step": 4172
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5600198419509732,
      "learning_rate": 2.0121690026500996e-07,
      "loss": 0.6293,
      "step": 4173
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6861755237625058,
      "learning_rate": 1.997663136396122e-07,
      "loss": 0.5097,
      "step": 4174
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1314330148758307,
      "learning_rate": 1.9832092188270534e-07,
      "loss": 1.0449,
      "step": 4175
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5136753497478561,
      "learning_rate": 1.9688072576045436e-07,
      "loss": 0.567,
      "step": 4176
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.298117027696984,
      "learning_rate": 1.954457260362741e-07,
      "loss": 1.0797,
      "step": 4177
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2881788819358,
      "learning_rate": 1.9401592347082275e-07,
      "loss": 1.0648,
      "step": 4178
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6106358807993417,
      "learning_rate": 1.925913188220052e-07,
      "loss": 0.5827,
      "step": 4179
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7038280160891928,
      "learning_rate": 1.911719128449685e-07,
      "loss": 0.5177,
      "step": 4180
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5829166091989428,
      "learning_rate": 1.8975770629210632e-07,
      "loss": 0.5463,
      "step": 4181
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4870481641531783,
      "learning_rate": 1.8834869991305682e-07,
      "loss": 0.5107,
      "step": 4182
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8217007985284999,
      "learning_rate": 1.8694489445469922e-07,
      "loss": 0.4927,
      "step": 4183
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.594667796164314,
      "learning_rate": 1.8554629066115827e-07,
      "loss": 0.523,
      "step": 4184
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2829265301451658,
      "learning_rate": 1.841528892737987e-07,
      "loss": 1.0853,
      "step": 4185
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6783874625923403,
      "learning_rate": 1.8276469103122973e-07,
      "loss": 0.5273,
      "step": 4186
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.321337177327477,
      "learning_rate": 1.813816966693027e-07,
      "loss": 1.1526,
      "step": 4187
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6290435864416266,
      "learning_rate": 1.8000390692110792e-07,
      "loss": 0.5746,
      "step": 4188
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4183533898137752,
      "learning_rate": 1.7863132251698e-07,
      "loss": 1.0435,
      "step": 4189
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6928831347418969,
      "learning_rate": 1.7726394418449256e-07,
      "loss": 0.5772,
      "step": 4190
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.591371236194775,
      "learning_rate": 1.759017726484602e-07,
      "loss": 0.5047,
      "step": 4191
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5402965015078836,
      "learning_rate": 1.7454480863093758e-07,
      "loss": 0.5424,
      "step": 4192
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.613223756291702,
      "learning_rate": 1.7319305285121823e-07,
      "loss": 0.5047,
      "step": 4193
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6008199739075684,
      "learning_rate": 1.7184650602583564e-07,
      "loss": 0.6257,
      "step": 4194
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4107474876420967,
      "learning_rate": 1.7050516886856217e-07,
      "loss": 1.095,
      "step": 4195
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2349759461716676,
      "learning_rate": 1.6916904209040686e-07,
      "loss": 1.1026,
      "step": 4196
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7273635891647547,
      "learning_rate": 1.6783812639962094e-07,
      "loss": 0.547,
      "step": 4197
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7786643907915871,
      "learning_rate": 1.665124225016901e-07,
      "loss": 0.5441,
      "step": 4198
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6692254868388379,
      "learning_rate": 1.651919310993366e-07,
      "loss": 0.5505,
      "step": 4199
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2698303051340285,
      "learning_rate": 1.6387665289252175e-07,
      "loss": 1.0776,
      "step": 4200
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.605356790234303,
      "learning_rate": 1.6256658857844443e-07,
      "loss": 0.5696,
      "step": 4201
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6590905152804785,
      "learning_rate": 1.6126173885153695e-07,
      "loss": 0.5891,
      "step": 4202
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6377362311264423,
      "learning_rate": 1.5996210440346937e-07,
      "loss": 0.5906,
      "step": 4203
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5490688029964716,
      "learning_rate": 1.586676859231473e-07,
      "loss": 0.5518,
      "step": 4204
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6218022569105504,
      "learning_rate": 1.5737848409670853e-07,
      "loss": 0.5559,
      "step": 4205
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5700911870768774,
      "learning_rate": 1.5609449960753087e-07,
      "loss": 0.5631,
      "step": 4206
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.197937494722781,
      "learning_rate": 1.5481573313622102e-07,
      "loss": 1.0829,
      "step": 4207
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.3364571132888012,
      "learning_rate": 1.535421853606256e-07,
      "loss": 1.1033,
      "step": 4208
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.674593396375545,
      "learning_rate": 1.5227385695581798e-07,
      "loss": 0.5548,
      "step": 4209
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4763602879586706,
      "learning_rate": 1.5101074859411146e-07,
      "loss": 0.5305,
      "step": 4210
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6135031018309323,
      "learning_rate": 1.4975286094504715e-07,
      "loss": 0.5811,
      "step": 4211
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6341189553567954,
      "learning_rate": 1.4850019467540277e-07,
      "loss": 0.5386,
      "step": 4212
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2924245234273073,
      "learning_rate": 1.4725275044918607e-07,
      "loss": 1.1303,
      "step": 4213
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5945499048792449,
      "learning_rate": 1.4601052892763812e-07,
      "loss": 0.6283,
      "step": 4214
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5074995202068848,
      "learning_rate": 1.4477353076922997e-07,
      "loss": 0.5632,
      "step": 4215
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5858215980430617,
      "learning_rate": 1.4354175662966374e-07,
      "loss": 0.5695,
      "step": 4216
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.528353083420829,
      "learning_rate": 1.4231520716187498e-07,
      "loss": 0.5283,
      "step": 4217
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2912326266077487,
      "learning_rate": 1.410938830160258e-07,
      "loss": 1.0826,
      "step": 4218
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.573795504751742,
      "learning_rate": 1.3987778483951388e-07,
      "loss": 0.5605,
      "step": 4219
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.515070178391331,
      "learning_rate": 1.3866691327696136e-07,
      "loss": 0.5219,
      "step": 4220
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5326605012738628,
      "learning_rate": 1.374612689702226e-07,
      "loss": 0.5147,
      "step": 4221
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7090491428708854,
      "learning_rate": 1.3626085255838196e-07,
      "loss": 0.5764,
      "step": 4222
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6838023780582665,
      "learning_rate": 1.3506566467774928e-07,
      "loss": 0.5156,
      "step": 4223
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5061177016733445,
      "learning_rate": 1.3387570596186673e-07,
      "loss": 0.563,
      "step": 4224
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5902031192556003,
      "learning_rate": 1.3269097704150192e-07,
      "loss": 0.5529,
      "step": 4225
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7827749036943548,
      "learning_rate": 1.315114785446514e-07,
      "loss": 0.596,
      "step": 4226
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5396771523426582,
      "learning_rate": 1.3033721109653842e-07,
      "loss": 0.5489,
      "step": 4227
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4561738499243801,
      "learning_rate": 1.291681753196161e-07,
      "loss": 0.5663,
      "step": 4228
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5939083972284176,
      "learning_rate": 1.2800437183355886e-07,
      "loss": 0.5938,
      "step": 4229
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5597321874657346,
      "learning_rate": 1.2684580125527423e-07,
      "loss": 0.5874,
      "step": 4230
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.481641969479649,
      "learning_rate": 1.2569246419888993e-07,
      "loss": 0.6107,
      "step": 4231
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5887873211923675,
      "learning_rate": 1.2454436127576464e-07,
      "loss": 0.5109,
      "step": 4232
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.240130572372756,
      "learning_rate": 1.2340149309447824e-07,
      "loss": 1.058,
      "step": 4233
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.435715254483859,
      "learning_rate": 1.2226386026083835e-07,
      "loss": 1.018,
      "step": 4234
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4479932129734336,
      "learning_rate": 1.2113146337787817e-07,
      "loss": 0.5486,
      "step": 4235
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7017432409783555,
      "learning_rate": 1.2000430304585088e-07,
      "loss": 0.5499,
      "step": 4236
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7615529626791588,
      "learning_rate": 1.1888237986224071e-07,
      "loss": 0.576,
      "step": 4237
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7207708344738983,
      "learning_rate": 1.1776569442174868e-07,
      "loss": 0.5313,
      "step": 4238
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.4367160020250047,
      "learning_rate": 1.1665424731630681e-07,
      "loss": 1.0221,
      "step": 4239
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.4550610702392746,
      "learning_rate": 1.1554803913506386e-07,
      "loss": 1.0465,
      "step": 4240
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.642729765216436,
      "learning_rate": 1.1444707046439407e-07,
      "loss": 0.534,
      "step": 4241
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5845291606952916,
      "learning_rate": 1.133513418878962e-07,
      "loss": 0.5673,
      "step": 4242
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5900761676138544,
      "learning_rate": 1.1226085398638898e-07,
      "loss": 0.6156,
      "step": 4243
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4786798176218974,
      "learning_rate": 1.1117560733791222e-07,
      "loss": 0.5485,
      "step": 4244
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7505553402253575,
      "learning_rate": 1.1009560251773132e-07,
      "loss": 0.5676,
      "step": 4245
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5745596469193097,
      "learning_rate": 1.0902084009833058e-07,
      "loss": 0.591,
      "step": 4246
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.598580861147918,
      "learning_rate": 1.0795132064941427e-07,
      "loss": 0.5301,
      "step": 4247
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6344992116784198,
      "learning_rate": 1.0688704473791112e-07,
      "loss": 0.5477,
      "step": 4248
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5685319541082792,
      "learning_rate": 1.058280129279654e-07,
      "loss": 0.5278,
      "step": 4249
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3940243579271354,
      "learning_rate": 1.0477422578094698e-07,
      "loss": 0.552,
      "step": 4250
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6419759812807646,
      "learning_rate": 1.0372568385544234e-07,
      "loss": 0.6291,
      "step": 4251
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5283693029778558,
      "learning_rate": 1.0268238770725691e-07,
      "loss": 0.5033,
      "step": 4252
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5085893417093497,
      "learning_rate": 1.0164433788941941e-07,
      "loss": 0.5981,
      "step": 4253
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5500688093193704,
      "learning_rate": 1.0061153495217413e-07,
      "loss": 0.601,
      "step": 4254
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.368431796032403,
      "learning_rate": 9.958397944298426e-08,
      "loss": 1.0293,
      "step": 4255
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6840571157200726,
      "learning_rate": 9.856167190653298e-08,
      "loss": 0.5755,
      "step": 4256
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6130598175548592,
      "learning_rate": 9.754461288472128e-08,
      "loss": 0.542,
      "step": 4257
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6096915536848913,
      "learning_rate": 9.653280291666678e-08,
      "loss": 0.4901,
      "step": 4258
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.715996477930236,
      "learning_rate": 9.552624253870713e-08,
      "loss": 0.5429,
      "step": 4259
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5779397252267782,
      "learning_rate": 9.45249322843933e-08,
      "loss": 0.4761,
      "step": 4260
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3973924340500463,
      "learning_rate": 9.352887268449962e-08,
      "loss": 0.5729,
      "step": 4261
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8671423349524394,
      "learning_rate": 9.253806426700928e-08,
      "loss": 0.5722,
      "step": 4262
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.408186276367237,
      "learning_rate": 9.155250755712885e-08,
      "loss": 1.0601,
      "step": 4263
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6834988429138362,
      "learning_rate": 9.05722030772771e-08,
      "loss": 0.603,
      "step": 4264
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.471583824352764,
      "learning_rate": 8.959715134708946e-08,
      "loss": 0.5297,
      "step": 4265
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6427727497069835,
      "learning_rate": 8.862735288341918e-08,
      "loss": 0.6104,
      "step": 4266
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5690543702273518,
      "learning_rate": 8.766280820033168e-08,
      "loss": 0.5357,
      "step": 4267
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7040411104874247,
      "learning_rate": 8.670351780910913e-08,
      "loss": 0.6,
      "step": 4268
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.161071534963507,
      "learning_rate": 8.574948221824808e-08,
      "loss": 1.0766,
      "step": 4269
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6186098107441222,
      "learning_rate": 8.480070193346069e-08,
      "loss": 0.5542,
      "step": 4270
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5476817278082804,
      "learning_rate": 8.385717745767242e-08,
      "loss": 0.5725,
      "step": 4271
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.271982058456636,
      "learning_rate": 8.291890929102209e-08,
      "loss": 1.0831,
      "step": 4272
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.3292482924559605,
      "learning_rate": 8.198589793086409e-08,
      "loss": 1.0526,
      "step": 4273
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6478852317348929,
      "learning_rate": 8.105814387176503e-08,
      "loss": 0.6133,
      "step": 4274
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5278273841307128,
      "learning_rate": 8.013564760550374e-08,
      "loss": 0.4943,
      "step": 4275
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.4063242466901795,
      "learning_rate": 7.921840962107574e-08,
      "loss": 1.072,
      "step": 4276
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5007772548277851,
      "learning_rate": 7.830643040468655e-08,
      "loss": 0.5864,
      "step": 4277
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5150116334236945,
      "learning_rate": 7.73997104397517e-08,
      "loss": 0.5474,
      "step": 4278
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1869805990169167,
      "learning_rate": 7.64982502069056e-08,
      "loss": 1.0483,
      "step": 4279
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.360736915910127,
      "learning_rate": 7.560205018398714e-08,
      "loss": 1.0574,
      "step": 4280
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6077940077328419,
      "learning_rate": 7.471111084605187e-08,
      "loss": 0.5323,
      "step": 4281
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7258315111451903,
      "learning_rate": 7.382543266536646e-08,
      "loss": 0.6106,
      "step": 4282
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5977778715813995,
      "learning_rate": 7.294501611140536e-08,
      "loss": 0.4937,
      "step": 4283
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5545162092520313,
      "learning_rate": 7.20698616508586e-08,
      "loss": 0.544,
      "step": 4284
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9417715834322937,
      "learning_rate": 7.119996974762178e-08,
      "loss": 0.6346,
      "step": 4285
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7243014159079975,
      "learning_rate": 7.033534086280714e-08,
      "loss": 0.6111,
      "step": 4286
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5619593419699749,
      "learning_rate": 6.947597545473028e-08,
      "loss": 0.5413,
      "step": 4287
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7864551846347936,
      "learning_rate": 6.86218739789235e-08,
      "loss": 0.5415,
      "step": 4288
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7337332126079534,
      "learning_rate": 6.777303688812132e-08,
      "loss": 0.6363,
      "step": 4289
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.255493379032079,
      "learning_rate": 6.692946463227601e-08,
      "loss": 1.0311,
      "step": 4290
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.3465205112192997,
      "learning_rate": 6.609115765854324e-08,
      "loss": 1.0762,
      "step": 4291
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4936153083050028,
      "learning_rate": 6.525811641128865e-08,
      "loss": 0.5783,
      "step": 4292
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7005554932678035,
      "learning_rate": 6.443034133208903e-08,
      "loss": 0.5016,
      "step": 4293
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5306178222574285,
      "learning_rate": 6.360783285972671e-08,
      "loss": 0.5738,
      "step": 4294
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2601954902842305,
      "learning_rate": 6.279059143019295e-08,
      "loss": 1.1063,
      "step": 4295
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5608097665105911,
      "learning_rate": 6.19786174766901e-08,
      "loss": 0.5151,
      "step": 4296
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6093670907401336,
      "learning_rate": 6.117191142962609e-08,
      "loss": 0.5433,
      "step": 4297
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2033955277956285,
      "learning_rate": 6.037047371661331e-08,
      "loss": 1.0608,
      "step": 4298
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.641368291257217,
      "learning_rate": 5.95743047624775e-08,
      "loss": 0.6005,
      "step": 4299
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5079574730270977,
      "learning_rate": 5.878340498924773e-08,
      "loss": 0.5647,
      "step": 4300
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5365658553395636,
      "learning_rate": 5.799777481616087e-08,
      "loss": 0.4939,
      "step": 4301
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2295978633494338,
      "learning_rate": 5.7217414659660466e-08,
      "loss": 1.0289,
      "step": 4302
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5185068368082884,
      "learning_rate": 5.644232493339785e-08,
      "loss": 0.5627,
      "step": 4303
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7818780670739949,
      "learning_rate": 5.5672506048228825e-08,
      "loss": 0.6112,
      "step": 4304
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6261970573023121,
      "learning_rate": 5.490795841221475e-08,
      "loss": 0.5933,
      "step": 4305
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.334697179577481,
      "learning_rate": 5.4148682430625874e-08,
      "loss": 1.0616,
      "step": 4306
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.674565785971677,
      "learning_rate": 5.339467850593472e-08,
      "loss": 0.6045,
      "step": 4307
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6755761481622622,
      "learning_rate": 5.2645947037820445e-08,
      "loss": 0.5691,
      "step": 4308
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7056042137544198,
      "learning_rate": 5.190248842316892e-08,
      "loss": 0.4942,
      "step": 4309
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6891462270374107,
      "learning_rate": 5.116430305606934e-08,
      "loss": 0.5762,
      "step": 4310
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.561667044246919,
      "learning_rate": 5.043139132781538e-08,
      "loss": 0.5234,
      "step": 4311
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6943414349929817,
      "learning_rate": 4.9703753626906273e-08,
      "loss": 0.5858,
      "step": 4312
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4960768777348603,
      "learning_rate": 4.8981390339045695e-08,
      "loss": 0.4963,
      "step": 4313
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5831915053011723,
      "learning_rate": 4.826430184714071e-08,
      "loss": 0.5645,
      "step": 4314
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6333190145080412,
      "learning_rate": 4.755248853130279e-08,
      "loss": 0.56,
      "step": 4315
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5357289799561753,
      "learning_rate": 4.6845950768846796e-08,
      "loss": 0.6121,
      "step": 4316
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6489138438070914,
      "learning_rate": 4.6144688934292025e-08,
      "loss": 0.641,
      "step": 4317
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6549136447111066,
      "learning_rate": 4.544870339936003e-08,
      "loss": 0.5446,
      "step": 4318
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.405116960293893,
      "learning_rate": 4.475799453297569e-08,
      "loss": 1.0883,
      "step": 4319
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4530943955400855,
      "learning_rate": 4.407256270126725e-08,
      "loss": 0.4966,
      "step": 4320
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6695477911283743,
      "learning_rate": 4.339240826756741e-08,
      "loss": 0.6115,
      "step": 4321
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6067094407391282,
      "learning_rate": 4.271753159240777e-08,
      "loss": 0.5473,
      "step": 4322
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4119415450560233,
      "learning_rate": 4.20479330335255e-08,
      "loss": 0.4923,
      "step": 4323
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7080306791027882,
      "learning_rate": 4.138361294585669e-08,
      "loss": 0.6168,
      "step": 4324
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7994555561719283,
      "learning_rate": 4.072457168154298e-08,
      "loss": 0.5916,
      "step": 4325
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.216032089032338,
      "learning_rate": 4.007080958992715e-08,
      "loss": 1.1185,
      "step": 4326
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5855973119716826,
      "learning_rate": 3.942232701755089e-08,
      "loss": 0.5728,
      "step": 4327
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.606959328935578,
      "learning_rate": 3.8779124308161445e-08,
      "loss": 0.5063,
      "step": 4328
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4600611377280683,
      "learning_rate": 3.8141201802701643e-08,
      "loss": 0.563,
      "step": 4329
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.560908766893363,
      "learning_rate": 3.7508559839323224e-08,
      "loss": 0.4941,
      "step": 4330
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.495994068764896,
      "learning_rate": 3.688119875337126e-08,
      "loss": 1.1099,
      "step": 4331
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6141512923114931,
      "learning_rate": 3.6259118877395305e-08,
      "loss": 0.5751,
      "step": 4332
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.106577701939822,
      "learning_rate": 3.564232054114602e-08,
      "loss": 1.0649,
      "step": 4333
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5994910140550285,
      "learning_rate": 3.503080407157411e-08,
      "loss": 0.4843,
      "step": 4334
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5548859314530057,
      "learning_rate": 3.442456979282693e-08,
      "loss": 0.4927,
      "step": 4335
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6021567983959877,
      "learning_rate": 3.382361802625633e-08,
      "loss": 0.5584,
      "step": 4336
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.4147559081068333,
      "learning_rate": 3.322794909041305e-08,
      "loss": 1.0876,
      "step": 4337
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6061007030857029,
      "learning_rate": 3.2637563301045615e-08,
      "loss": 0.5297,
      "step": 4338
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7632180083365119,
      "learning_rate": 3.2052460971104814e-08,
      "loss": 0.542,
      "step": 4339
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4605170986919351,
      "learning_rate": 3.1472642410738106e-08,
      "loss": 0.5281,
      "step": 4340
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.702297210156175,
      "learning_rate": 3.0898107927294084e-08,
      "loss": 0.5342,
      "step": 4341
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.49791546395471,
      "learning_rate": 3.0328857825318024e-08,
      "loss": 0.5511,
      "step": 4342
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7419472570781227,
      "learning_rate": 2.9764892406558554e-08,
      "loss": 0.5384,
      "step": 4343
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6317091791110443,
      "learning_rate": 2.9206211969958764e-08,
      "loss": 0.5799,
      "step": 4344
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5907059575939129,
      "learning_rate": 2.8652816811660656e-08,
      "loss": 0.5438,
      "step": 4345
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6583942058420196,
      "learning_rate": 2.8104707225007356e-08,
      "loss": 0.5747,
      "step": 4346
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6304930623764544,
      "learning_rate": 2.7561883500537568e-08,
      "loss": 0.5651,
      "step": 4347
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5381090540558204,
      "learning_rate": 2.7024345925990015e-08,
      "loss": 0.5157,
      "step": 4348
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5091874136207402,
      "learning_rate": 2.6492094786300103e-08,
      "loss": 0.5319,
      "step": 4349
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.474220781870778,
      "learning_rate": 2.5965130363601042e-08,
      "loss": 1.0282,
      "step": 4350
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1054122317492556,
      "learning_rate": 2.5443452937222723e-08,
      "loss": 1.0905,
      "step": 4351
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7535970037218693,
      "learning_rate": 2.4927062783697275e-08,
      "loss": 0.5916,
      "step": 4352
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6883099185742036,
      "learning_rate": 2.4415960176747965e-08,
      "loss": 0.5347,
      "step": 4353
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6600303964655723,
      "learning_rate": 2.3910145387299188e-08,
      "loss": 0.5699,
      "step": 4354
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2432771464120678,
      "learning_rate": 2.3409618683469803e-08,
      "loss": 1.0147,
      "step": 4355
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.55543665517605,
      "learning_rate": 2.291438033057869e-08,
      "loss": 0.5456,
      "step": 4356
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5419400343457996,
      "learning_rate": 2.242443059113919e-08,
      "loss": 0.5219,
      "step": 4357
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5912171986555959,
      "learning_rate": 2.1939769724861338e-08,
      "loss": 0.5249,
      "step": 4358
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5862432675914007,
      "learning_rate": 2.146039798865407e-08,
      "loss": 0.5616,
      "step": 4359
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6415313116969077,
      "learning_rate": 2.098631563661746e-08,
      "loss": 0.5568,
      "step": 4360
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.52565889698681,
      "learning_rate": 2.0517522920054932e-08,
      "loss": 0.5134,
      "step": 4361
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.4219315689128913,
      "learning_rate": 2.0054020087458824e-08,
      "loss": 1.0431,
      "step": 4362
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.152351275483864,
      "learning_rate": 1.9595807384523713e-08,
      "loss": 1.0419,
      "step": 4363
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.660437022571755,
      "learning_rate": 1.9142885054136418e-08,
      "loss": 0.588,
      "step": 4364
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7223325893969428,
      "learning_rate": 1.8695253336378228e-08,
      "loss": 0.5729,
      "step": 4365
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6363826327604059,
      "learning_rate": 1.8252912468530448e-08,
      "loss": 0.584,
      "step": 4366
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5755525344804824,
      "learning_rate": 1.781586268506663e-08,
      "loss": 0.5797,
      "step": 4367
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5243643610440645,
      "learning_rate": 1.7384104217654796e-08,
      "loss": 0.5734,
      "step": 4368
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5990489057089763,
      "learning_rate": 1.6957637295161867e-08,
      "loss": 0.547,
      "step": 4369
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.589041408542008,
      "learning_rate": 1.6536462143647016e-08,
      "loss": 0.5188,
      "step": 4370
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2119320778224623,
      "learning_rate": 1.6120578986364987e-08,
      "loss": 1.0172,
      "step": 4371
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.48239026402849,
      "learning_rate": 1.5709988043765e-08,
      "loss": 0.5769,
      "step": 4372
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6014569235312577,
      "learning_rate": 1.5304689533491844e-08,
      "loss": 0.5738,
      "step": 4373
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3088006134781858,
      "learning_rate": 1.4904683670384777e-08,
      "loss": 1.1132,
      "step": 4374
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.432450706499111,
      "learning_rate": 1.4509970666477524e-08,
      "loss": 1.0998,
      "step": 4375
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6284994861211646,
      "learning_rate": 1.4120550730997163e-08,
      "loss": 0.5764,
      "step": 4376
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6782805500037192,
      "learning_rate": 1.373642407036635e-08,
      "loss": 0.5793,
      "step": 4377
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6142823510763826,
      "learning_rate": 1.3357590888199989e-08,
      "loss": 0.5776,
      "step": 4378
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4769403526002,
      "learning_rate": 1.2984051385310782e-08,
      "loss": 0.5303,
      "step": 4379
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.4040456475328416,
      "learning_rate": 1.261580575970256e-08,
      "loss": 1.0997,
      "step": 4380
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.318589111855715,
      "learning_rate": 1.2252854206572518e-08,
      "loss": 1.1144,
      "step": 4381
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8118598383707587,
      "learning_rate": 1.1895196918312313e-08,
      "loss": 0.6153,
      "step": 4382
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6936386001509656,
      "learning_rate": 1.154283408450807e-08,
      "loss": 0.5429,
      "step": 4383
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6598196887613899,
      "learning_rate": 1.119576589193927e-08,
      "loss": 0.5494,
      "step": 4384
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5592732675728436,
      "learning_rate": 1.0853992524577639e-08,
      "loss": 0.5854,
      "step": 4385
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5818341144064911,
      "learning_rate": 1.0517514163590482e-08,
      "loss": 0.582,
      "step": 4386
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7117133603100072,
      "learning_rate": 1.0186330987335124e-08,
      "loss": 0.6028,
      "step": 4387
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6201509152767648,
      "learning_rate": 9.860443171364476e-09,
      "loss": 0.5588,
      "step": 4388
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4840926925528688,
      "learning_rate": 9.539850888423685e-09,
      "loss": 0.4222,
      "step": 4389
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5234480681389526,
      "learning_rate": 9.224554308452371e-09,
      "loss": 0.5636,
      "step": 4390
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.670192061115403,
      "learning_rate": 8.914553598580178e-09,
      "loss": 0.5565,
      "step": 4391
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6449647262687004,
      "learning_rate": 8.609848923131215e-09,
      "loss": 0.5762,
      "step": 4392
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6843740853482445,
      "learning_rate": 8.310440443621837e-09,
      "loss": 0.5855,
      "step": 4393
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2351518406782107,
      "learning_rate": 8.016328318761757e-09,
      "loss": 1.1109,
      "step": 4394
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5250892808138807,
      "learning_rate": 7.727512704452933e-09,
      "loss": 0.4997,
      "step": 4395
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6711100896468545,
      "learning_rate": 7.443993753789569e-09,
      "loss": 0.5773,
      "step": 4396
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.587653393984809,
      "learning_rate": 7.165771617058115e-09,
      "loss": 0.5635,
      "step": 4397
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6067286312517477,
      "learning_rate": 6.892846441738376e-09,
      "loss": 0.5357,
      "step": 4398
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5168416774329188,
      "learning_rate": 6.625218372499076e-09,
      "loss": 0.5849,
      "step": 4399
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.3256680786891475,
      "learning_rate": 6.362887551205621e-09,
      "loss": 1.0389,
      "step": 4400
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6867825554714115,
      "learning_rate": 6.1058541169123355e-09,
      "loss": 0.5993,
      "step": 4401
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.294266959327261,
      "learning_rate": 5.854118205865789e-09,
      "loss": 1.0558,
      "step": 4402
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5598085415567884,
      "learning_rate": 5.6076799515070166e-09,
      "loss": 0.5212,
      "step": 4403
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2416110459753145,
      "learning_rate": 5.366539484464861e-09,
      "loss": 1.0573,
      "step": 4404
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7082340801834424,
      "learning_rate": 5.130696932564849e-09,
      "loss": 0.6431,
      "step": 4405
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.688150612628795,
      "learning_rate": 4.900152420818094e-09,
      "loss": 0.5707,
      "step": 4406
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.4206112950058656,
      "learning_rate": 4.674906071433505e-09,
      "loss": 1.075,
      "step": 4407
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4571759806615239,
      "learning_rate": 4.454958003807797e-09,
      "loss": 0.5027,
      "step": 4408
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.4036197831521737,
      "learning_rate": 4.24030833452882e-09,
      "loss": 1.1264,
      "step": 4409
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6647017527855035,
      "learning_rate": 4.030957177380002e-09,
      "loss": 0.59,
      "step": 4410
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.440401821086598,
      "learning_rate": 3.826904643330354e-09,
      "loss": 0.5305,
      "step": 4411
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.574764629748449,
      "learning_rate": 3.6281508405455746e-09,
      "loss": 0.5321,
      "step": 4412
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.3378567227328024,
      "learning_rate": 3.434695874379168e-09,
      "loss": 1.1043,
      "step": 4413
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7292899008687816,
      "learning_rate": 3.246539847376884e-09,
      "loss": 0.4954,
      "step": 4414
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4388775154473985,
      "learning_rate": 3.0636828592756076e-09,
      "loss": 0.5098,
      "step": 4415
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2601258539441473,
      "learning_rate": 2.8861250070044698e-09,
      "loss": 1.0529,
      "step": 4416
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4261623011553302,
      "learning_rate": 2.7138663846826284e-09,
      "loss": 0.5137,
      "step": 4417
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6145574930989564,
      "learning_rate": 2.5469070836181555e-09,
      "loss": 0.5137,
      "step": 4418
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8251873850452527,
      "learning_rate": 2.385247192314699e-09,
      "loss": 0.5848,
      "step": 4419
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5742692383032613,
      "learning_rate": 2.228886796462604e-09,
      "loss": 0.5157,
      "step": 4420
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.3004198253356902,
      "learning_rate": 2.0778259789455692e-09,
      "loss": 1.0152,
      "step": 4421
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7025607516443948,
      "learning_rate": 1.9320648198362104e-09,
      "loss": 0.5949,
      "step": 4422
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7247159003974557,
      "learning_rate": 1.7916033964016088e-09,
      "loss": 0.6058,
      "step": 4423
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4016720166751084,
      "learning_rate": 1.656441783093321e-09,
      "loss": 0.5128,
      "step": 4424
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6639190318991202,
      "learning_rate": 1.5265800515607e-09,
      "loss": 0.5803,
      "step": 4425
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.304443722832669,
      "learning_rate": 1.4020182706386831e-09,
      "loss": 1.0678,
      "step": 4426
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6794033639755892,
      "learning_rate": 1.2827565063533443e-09,
      "loss": 0.5556,
      "step": 4427
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5956049903629512,
      "learning_rate": 1.1687948219252232e-09,
      "loss": 0.5391,
      "step": 4428
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.4038097096664313,
      "learning_rate": 1.0601332777604445e-09,
      "loss": 1.0537,
      "step": 4429
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6547195117817686,
      "learning_rate": 9.567719314595992e-10,
      "loss": 0.4585,
      "step": 4430
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5383734825809001,
      "learning_rate": 8.587108378099729e-10,
      "loss": 0.5158,
      "step": 4431
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.4337932581810797,
      "learning_rate": 7.659500487933181e-10,
      "loss": 1.0158,
      "step": 4432
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.580629142570285,
      "learning_rate": 6.784896135780816e-10,
      "loss": 0.5201,
      "step": 4433
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7373272604937144,
      "learning_rate": 5.963295785271772e-10,
      "loss": 0.5738,
      "step": 4434
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.549995554045313,
      "learning_rate": 5.194699871902132e-10,
      "loss": 0.5027,
      "step": 4435
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7559403820320774,
      "learning_rate": 4.479108803079335e-10,
      "loss": 0.5285,
      "step": 4436
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.207478191278171,
      "learning_rate": 3.8165229581443863e-10,
      "loss": 1.0237,
      "step": 4437
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5431668961637934,
      "learning_rate": 3.2069426882941345e-10,
      "loss": 0.5613,
      "step": 4438
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6348956051878503,
      "learning_rate": 2.650368316658991e-10,
      "loss": 0.5416,
      "step": 4439
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.3276649200128974,
      "learning_rate": 2.1468001382807247e-10,
      "loss": 1.0438,
      "step": 4440
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.2803545877210643,
      "learning_rate": 1.6962384200680525e-10,
      "loss": 1.0815,
      "step": 4441
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.2568912323899295,
      "learning_rate": 1.2986834008632543e-10,
      "loss": 1.0128,
      "step": 4442
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5742606665344632,
      "learning_rate": 9.541352914088642e-11,
      "loss": 0.5744,
      "step": 4443
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.512782931870234,
      "learning_rate": 6.625942743254676e-11,
      "loss": 0.5835,
      "step": 4444
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4964806213402484,
      "learning_rate": 4.2406050415610966e-11,
      "loss": 0.5087,
      "step": 4445
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6961147200456526,
      "learning_rate": 2.3853410734409054e-11,
      "loss": 0.5486,
      "step": 4446
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.600796485118127,
      "learning_rate": 1.060151822329658e-11,
      "loss": 0.5197,
      "step": 4447
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5629085883819842,
      "learning_rate": 2.650379906654621e-12,
      "loss": 0.5182,
      "step": 4448
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9524832130007188,
      "learning_rate": 0.0,
      "loss": 0.6347,
      "step": 4449
    },
    {
      "epoch": 1.0,
      "step": 4449,
      "total_flos": 1.0273217737129984e+16,
      "train_loss": 0.7218901999587825,
      "train_runtime": 105263.8052,
      "train_samples_per_second": 10.82,
      "train_steps_per_second": 0.042
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 4449,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2000,
  "total_flos": 1.0273217737129984e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}