{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999181039424762,
  "eval_steps": 500,
  "global_step": 7631,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 15.184271812438965,
      "learning_rate": 4.366812227074236e-08,
      "loss": 1.3597,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.087559700012207,
      "learning_rate": 8.733624454148472e-08,
      "loss": 1.4088,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.187013626098633,
      "learning_rate": 1.3100436681222707e-07,
      "loss": 1.3813,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.952133178710938,
      "learning_rate": 1.7467248908296944e-07,
      "loss": 1.3992,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.872638702392578,
      "learning_rate": 2.183406113537118e-07,
      "loss": 1.3842,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.776309013366699,
      "learning_rate": 2.6200873362445414e-07,
      "loss": 1.2719,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.666576385498047,
      "learning_rate": 3.056768558951965e-07,
      "loss": 1.3553,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.574089050292969,
      "learning_rate": 3.4934497816593887e-07,
      "loss": 1.4164,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.68317985534668,
      "learning_rate": 3.930131004366813e-07,
      "loss": 1.3999,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.673285484313965,
      "learning_rate": 4.366812227074236e-07,
      "loss": 1.4478,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.704113006591797,
      "learning_rate": 4.80349344978166e-07,
      "loss": 1.3281,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.7592201232910156,
      "learning_rate": 5.240174672489083e-07,
      "loss": 1.287,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.884162187576294,
      "learning_rate": 5.676855895196507e-07,
      "loss": 1.2863,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.222865104675293,
      "learning_rate": 6.11353711790393e-07,
      "loss": 1.356,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.936320304870605,
      "learning_rate": 6.550218340611354e-07,
      "loss": 1.3958,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.170361518859863,
      "learning_rate": 6.986899563318777e-07,
      "loss": 1.3313,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.372904777526855,
      "learning_rate": 7.423580786026202e-07,
      "loss": 1.3952,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.080046653747559,
      "learning_rate": 7.860262008733626e-07,
      "loss": 1.4087,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.848758697509766,
      "learning_rate": 8.296943231441049e-07,
      "loss": 1.3262,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.305721282958984,
      "learning_rate": 8.733624454148472e-07,
      "loss": 1.3136,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.571191787719727,
      "learning_rate": 9.170305676855896e-07,
      "loss": 1.299,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.864514350891113,
      "learning_rate": 9.60698689956332e-07,
      "loss": 1.3208,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.488653182983398,
      "learning_rate": 1.0043668122270742e-06,
      "loss": 1.3087,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.563277244567871,
      "learning_rate": 1.0480349344978166e-06,
      "loss": 1.3088,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.5420446395874023,
      "learning_rate": 1.091703056768559e-06,
      "loss": 1.2566,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.120272159576416,
      "learning_rate": 1.1353711790393014e-06,
      "loss": 1.2916,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.431443929672241,
      "learning_rate": 1.1790393013100437e-06,
      "loss": 1.2789,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.864900588989258,
      "learning_rate": 1.222707423580786e-06,
      "loss": 1.3351,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.267812252044678,
      "learning_rate": 1.2663755458515283e-06,
      "loss": 1.2705,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.4054276943206787,
      "learning_rate": 1.3100436681222709e-06,
      "loss": 1.2983,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.647962808609009,
      "learning_rate": 1.3537117903930134e-06,
      "loss": 1.2676,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.0060627460479736,
      "learning_rate": 1.3973799126637555e-06,
      "loss": 1.1988,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.107463359832764,
      "learning_rate": 1.441048034934498e-06,
      "loss": 1.2391,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.0892086029052734,
      "learning_rate": 1.4847161572052403e-06,
      "loss": 1.1996,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.637197971343994,
      "learning_rate": 1.5283842794759826e-06,
      "loss": 1.2241,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.710655689239502,
      "learning_rate": 1.5720524017467252e-06,
      "loss": 1.2199,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3094117641448975,
      "learning_rate": 1.6157205240174673e-06,
      "loss": 1.1982,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1601836681365967,
      "learning_rate": 1.6593886462882098e-06,
      "loss": 1.2311,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3276045322418213,
      "learning_rate": 1.703056768558952e-06,
      "loss": 1.126,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3068459033966064,
      "learning_rate": 1.7467248908296944e-06,
      "loss": 1.2065,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.76163649559021,
      "learning_rate": 1.790393013100437e-06,
      "loss": 1.2424,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.561727285385132,
      "learning_rate": 1.8340611353711792e-06,
      "loss": 1.1883,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.541086435317993,
      "learning_rate": 1.8777292576419216e-06,
      "loss": 1.1345,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3438589572906494,
      "learning_rate": 1.921397379912664e-06,
      "loss": 1.1724,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2678864002227783,
      "learning_rate": 1.965065502183406e-06,
      "loss": 1.1678,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2679197788238525,
      "learning_rate": 2.0087336244541485e-06,
      "loss": 1.1731,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0144221782684326,
      "learning_rate": 2.052401746724891e-06,
      "loss": 1.1883,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7473580837249756,
      "learning_rate": 2.096069868995633e-06,
      "loss": 1.1014,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5149941444396973,
      "learning_rate": 2.1397379912663756e-06,
      "loss": 1.2587,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3464908599853516,
      "learning_rate": 2.183406113537118e-06,
      "loss": 1.2699,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7884796857833862,
      "learning_rate": 2.2270742358078603e-06,
      "loss": 1.1182,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7958896160125732,
      "learning_rate": 2.270742358078603e-06,
      "loss": 1.079,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9794340133666992,
      "learning_rate": 2.3144104803493453e-06,
      "loss": 1.264,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.089437484741211,
      "learning_rate": 2.3580786026200874e-06,
      "loss": 1.1234,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9000921249389648,
      "learning_rate": 2.40174672489083e-06,
      "loss": 1.1362,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.901570200920105,
      "learning_rate": 2.445414847161572e-06,
      "loss": 1.1168,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.953324556350708,
      "learning_rate": 2.4890829694323146e-06,
      "loss": 1.0725,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9219168424606323,
      "learning_rate": 2.5327510917030567e-06,
      "loss": 1.1458,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6381099224090576,
      "learning_rate": 2.576419213973799e-06,
      "loss": 1.1146,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.722344160079956,
      "learning_rate": 2.6200873362445417e-06,
      "loss": 1.0748,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.685483694076538,
      "learning_rate": 2.6637554585152842e-06,
      "loss": 1.1404,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7123844623565674,
      "learning_rate": 2.7074235807860268e-06,
      "loss": 1.101,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5008476972579956,
      "learning_rate": 2.7510917030567684e-06,
      "loss": 1.257,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5240020751953125,
      "learning_rate": 2.794759825327511e-06,
      "loss": 1.2345,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5795421600341797,
      "learning_rate": 2.8384279475982535e-06,
      "loss": 1.1366,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6865391731262207,
      "learning_rate": 2.882096069868996e-06,
      "loss": 1.0815,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5658175945281982,
      "learning_rate": 2.9257641921397385e-06,
      "loss": 1.04,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7005417346954346,
      "learning_rate": 2.9694323144104806e-06,
      "loss": 1.0613,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7347066402435303,
      "learning_rate": 3.0131004366812227e-06,
      "loss": 1.0875,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.58201003074646,
      "learning_rate": 3.0567685589519653e-06,
      "loss": 1.0801,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4996534585952759,
      "learning_rate": 3.100436681222708e-06,
      "loss": 1.0272,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5953459739685059,
      "learning_rate": 3.1441048034934503e-06,
      "loss": 1.0962,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5690724849700928,
      "learning_rate": 3.1877729257641924e-06,
      "loss": 1.0492,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6362709999084473,
      "learning_rate": 3.2314410480349345e-06,
      "loss": 1.0747,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5067527294158936,
      "learning_rate": 3.275109170305677e-06,
      "loss": 1.0579,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5211260318756104,
      "learning_rate": 3.3187772925764196e-06,
      "loss": 1.1105,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0240404605865479,
      "learning_rate": 3.362445414847162e-06,
      "loss": 1.2333,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.596854329109192,
      "learning_rate": 3.406113537117904e-06,
      "loss": 1.0343,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9191591739654541,
      "learning_rate": 3.4497816593886467e-06,
      "loss": 1.2419,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8874264359474182,
      "learning_rate": 3.493449781659389e-06,
      "loss": 1.23,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.489851951599121,
      "learning_rate": 3.5371179039301313e-06,
      "loss": 1.0947,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5562266111373901,
      "learning_rate": 3.580786026200874e-06,
      "loss": 1.0584,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.458613395690918,
      "learning_rate": 3.624454148471616e-06,
      "loss": 1.0904,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8255189061164856,
      "learning_rate": 3.6681222707423585e-06,
      "loss": 1.2286,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5308469533920288,
      "learning_rate": 3.7117903930131006e-06,
      "loss": 1.0144,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7949082851409912,
      "learning_rate": 3.755458515283843e-06,
      "loss": 1.2136,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6952340602874756,
      "learning_rate": 3.7991266375545856e-06,
      "loss": 1.0732,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8294545412063599,
      "learning_rate": 3.842794759825328e-06,
      "loss": 1.2214,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7787436246871948,
      "learning_rate": 3.88646288209607e-06,
      "loss": 1.2054,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6932886242866516,
      "learning_rate": 3.930131004366812e-06,
      "loss": 1.2202,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6710336208343506,
      "learning_rate": 3.9737991266375545e-06,
      "loss": 1.2104,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6449258327484131,
      "learning_rate": 4.017467248908297e-06,
      "loss": 1.2238,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6785801649093628,
      "learning_rate": 4.0611353711790395e-06,
      "loss": 1.0275,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.554275631904602,
      "learning_rate": 4.104803493449782e-06,
      "loss": 1.0374,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5450009107589722,
      "learning_rate": 4.1484716157205246e-06,
      "loss": 1.0257,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7251917123794556,
      "learning_rate": 4.192139737991266e-06,
      "loss": 1.2133,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6629444360733032,
      "learning_rate": 4.235807860262009e-06,
      "loss": 1.0125,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4590173959732056,
      "learning_rate": 4.279475982532751e-06,
      "loss": 1.0016,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7446169257164001,
      "learning_rate": 4.323144104803494e-06,
      "loss": 1.2125,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4785338640213013,
      "learning_rate": 4.366812227074236e-06,
      "loss": 1.0366,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4842109680175781,
      "learning_rate": 4.410480349344978e-06,
      "loss": 0.9818,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4168837070465088,
      "learning_rate": 4.4541484716157205e-06,
      "loss": 0.9949,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5079370737075806,
      "learning_rate": 4.497816593886463e-06,
      "loss": 1.06,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3976857662200928,
      "learning_rate": 4.541484716157206e-06,
      "loss": 1.0465,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.513974905014038,
      "learning_rate": 4.585152838427948e-06,
      "loss": 1.0024,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8088471293449402,
      "learning_rate": 4.628820960698691e-06,
      "loss": 1.181,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5142829418182373,
      "learning_rate": 4.672489082969432e-06,
      "loss": 0.9735,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6832176446914673,
      "learning_rate": 4.716157205240175e-06,
      "loss": 1.2133,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4363672733306885,
      "learning_rate": 4.759825327510917e-06,
      "loss": 1.0091,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5161020755767822,
      "learning_rate": 4.80349344978166e-06,
      "loss": 1.0322,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5461225509643555,
      "learning_rate": 4.847161572052402e-06,
      "loss": 1.0251,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4133336544036865,
      "learning_rate": 4.890829694323144e-06,
      "loss": 1.0737,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4342045783996582,
      "learning_rate": 4.934497816593887e-06,
      "loss": 1.0255,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9645444750785828,
      "learning_rate": 4.978165938864629e-06,
      "loss": 1.2079,
      "step": 114
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.771979033946991,
      "learning_rate": 5.021834061135371e-06,
      "loss": 1.1903,
      "step": 115
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5585743188858032,
      "learning_rate": 5.065502183406113e-06,
      "loss": 0.9557,
      "step": 116
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3990782499313354,
      "learning_rate": 5.109170305676856e-06,
      "loss": 0.9976,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7741254568099976,
      "learning_rate": 5.152838427947598e-06,
      "loss": 1.2089,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4438834190368652,
      "learning_rate": 5.196506550218341e-06,
      "loss": 1.0428,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4215728044509888,
      "learning_rate": 5.2401746724890834e-06,
      "loss": 1.0115,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7292472124099731,
      "learning_rate": 5.283842794759826e-06,
      "loss": 1.0052,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7085826396942139,
      "learning_rate": 5.3275109170305685e-06,
      "loss": 0.9836,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4814881086349487,
      "learning_rate": 5.371179039301311e-06,
      "loss": 0.9557,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3909896612167358,
      "learning_rate": 5.4148471615720535e-06,
      "loss": 1.0379,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4656494855880737,
      "learning_rate": 5.458515283842796e-06,
      "loss": 1.0256,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9599483013153076,
      "learning_rate": 5.502183406113537e-06,
      "loss": 1.2164,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.363072395324707,
      "learning_rate": 5.545851528384279e-06,
      "loss": 0.9673,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7388890981674194,
      "learning_rate": 5.589519650655022e-06,
      "loss": 1.1907,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5297843217849731,
      "learning_rate": 5.6331877729257645e-06,
      "loss": 1.0676,
      "step": 129
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5226210355758667,
      "learning_rate": 5.676855895196507e-06,
      "loss": 1.0221,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9671232104301453,
      "learning_rate": 5.7205240174672495e-06,
      "loss": 1.1991,
      "step": 131
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9116547107696533,
      "learning_rate": 5.764192139737992e-06,
      "loss": 1.2006,
      "step": 132
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4933277368545532,
      "learning_rate": 5.8078602620087346e-06,
      "loss": 1.0025,
      "step": 133
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6493118405342102,
      "learning_rate": 5.851528384279477e-06,
      "loss": 1.1985,
      "step": 134
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.822118878364563,
      "learning_rate": 5.89519650655022e-06,
      "loss": 1.2019,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.546926498413086,
      "learning_rate": 5.938864628820961e-06,
      "loss": 0.9685,
      "step": 136
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.449734091758728,
      "learning_rate": 5.982532751091703e-06,
      "loss": 1.0188,
      "step": 137
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8797246217727661,
      "learning_rate": 6.0262008733624455e-06,
      "loss": 1.222,
      "step": 138
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4768751859664917,
      "learning_rate": 6.069868995633188e-06,
      "loss": 1.0083,
      "step": 139
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.428026795387268,
      "learning_rate": 6.1135371179039305e-06,
      "loss": 1.0104,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5379831790924072,
      "learning_rate": 6.157205240174673e-06,
      "loss": 1.0164,
      "step": 141
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.387872338294983,
      "learning_rate": 6.200873362445416e-06,
      "loss": 0.9602,
      "step": 142
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8798995614051819,
      "learning_rate": 6.244541484716158e-06,
      "loss": 1.2045,
      "step": 143
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.738410234451294,
      "learning_rate": 6.288209606986901e-06,
      "loss": 1.2027,
      "step": 144
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4975425004959106,
      "learning_rate": 6.331877729257643e-06,
      "loss": 0.9679,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6109999418258667,
      "learning_rate": 6.375545851528385e-06,
      "loss": 0.9499,
      "step": 146
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4859771728515625,
      "learning_rate": 6.419213973799127e-06,
      "loss": 1.0468,
      "step": 147
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4700217247009277,
      "learning_rate": 6.462882096069869e-06,
      "loss": 0.9658,
      "step": 148
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6128220558166504,
      "learning_rate": 6.5065502183406116e-06,
      "loss": 0.9502,
      "step": 149
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4946608543395996,
      "learning_rate": 6.550218340611354e-06,
      "loss": 0.9345,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4591008424758911,
      "learning_rate": 6.593886462882097e-06,
      "loss": 1.0364,
      "step": 151
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3770273923873901,
      "learning_rate": 6.637554585152839e-06,
      "loss": 0.9632,
      "step": 152
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2683231830596924,
      "learning_rate": 6.681222707423582e-06,
      "loss": 1.209,
      "step": 153
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9459937214851379,
      "learning_rate": 6.724890829694324e-06,
      "loss": 1.1757,
      "step": 154
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6540155410766602,
      "learning_rate": 6.768558951965066e-06,
      "loss": 1.008,
      "step": 155
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.608180284500122,
      "learning_rate": 6.812227074235808e-06,
      "loss": 1.0062,
      "step": 156
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5652464628219604,
      "learning_rate": 6.855895196506551e-06,
      "loss": 0.9617,
      "step": 157
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7774461507797241,
      "learning_rate": 6.8995633187772934e-06,
      "loss": 1.1664,
      "step": 158
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5823050737380981,
      "learning_rate": 6.943231441048035e-06,
      "loss": 1.0265,
      "step": 159
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5173331499099731,
      "learning_rate": 6.986899563318778e-06,
      "loss": 0.9706,
      "step": 160
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4883906841278076,
      "learning_rate": 7.03056768558952e-06,
      "loss": 1.0019,
      "step": 161
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1934669017791748,
      "learning_rate": 7.074235807860263e-06,
      "loss": 1.1814,
      "step": 162
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6672334671020508,
      "learning_rate": 7.117903930131005e-06,
      "loss": 0.9585,
      "step": 163
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8747884035110474,
      "learning_rate": 7.161572052401748e-06,
      "loss": 0.964,
      "step": 164
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.577018141746521,
      "learning_rate": 7.205240174672489e-06,
      "loss": 0.9934,
      "step": 165
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4888193607330322,
      "learning_rate": 7.248908296943232e-06,
      "loss": 1.0047,
      "step": 166
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5384252071380615,
      "learning_rate": 7.2925764192139745e-06,
      "loss": 0.9453,
      "step": 167
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.483939290046692,
      "learning_rate": 7.336244541484717e-06,
      "loss": 0.9603,
      "step": 168
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.148611068725586,
      "learning_rate": 7.3799126637554595e-06,
      "loss": 0.9822,
      "step": 169
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6577836275100708,
      "learning_rate": 7.423580786026201e-06,
      "loss": 0.9792,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5029042959213257,
      "learning_rate": 7.467248908296944e-06,
      "loss": 0.9459,
      "step": 171
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.568406343460083,
      "learning_rate": 7.510917030567686e-06,
      "loss": 0.9901,
      "step": 172
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4324817657470703,
      "learning_rate": 7.554585152838429e-06,
      "loss": 1.1852,
      "step": 173
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1369590759277344,
      "learning_rate": 7.598253275109171e-06,
      "loss": 1.21,
      "step": 174
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7289785742759705,
      "learning_rate": 7.641921397379913e-06,
      "loss": 1.2109,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7318261861801147,
      "learning_rate": 7.685589519650655e-06,
      "loss": 1.0093,
      "step": 176
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3872264623641968,
      "learning_rate": 7.729257641921398e-06,
      "loss": 1.1967,
      "step": 177
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5441734790802002,
      "learning_rate": 7.77292576419214e-06,
      "loss": 1.0295,
      "step": 178
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5853267908096313,
      "learning_rate": 7.816593886462883e-06,
      "loss": 0.9854,
      "step": 179
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5659153461456299,
      "learning_rate": 7.860262008733624e-06,
      "loss": 1.1621,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5575003623962402,
      "learning_rate": 7.903930131004366e-06,
      "loss": 0.9918,
      "step": 181
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6159673929214478,
      "learning_rate": 7.947598253275109e-06,
      "loss": 1.0021,
      "step": 182
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4281398057937622,
      "learning_rate": 7.991266375545851e-06,
      "loss": 0.9746,
      "step": 183
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5250587463378906,
      "learning_rate": 8.034934497816594e-06,
      "loss": 0.9335,
      "step": 184
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8764203190803528,
      "learning_rate": 8.078602620087337e-06,
      "loss": 1.183,
      "step": 185
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7672403454780579,
      "learning_rate": 8.122270742358079e-06,
      "loss": 1.1988,
      "step": 186
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.702710747718811,
      "learning_rate": 8.165938864628822e-06,
      "loss": 1.1759,
      "step": 187
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.718480110168457,
      "learning_rate": 8.209606986899564e-06,
      "loss": 1.1651,
      "step": 188
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8376530408859253,
      "learning_rate": 8.253275109170307e-06,
      "loss": 1.1792,
      "step": 189
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9699944257736206,
      "learning_rate": 8.296943231441049e-06,
      "loss": 1.015,
      "step": 190
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7379977703094482,
      "learning_rate": 8.34061135371179e-06,
      "loss": 1.0114,
      "step": 191
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4923131465911865,
      "learning_rate": 8.384279475982532e-06,
      "loss": 0.9593,
      "step": 192
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0008738040924072,
      "learning_rate": 8.427947598253275e-06,
      "loss": 0.9835,
      "step": 193
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.568870186805725,
      "learning_rate": 8.471615720524018e-06,
      "loss": 0.9553,
      "step": 194
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6120963096618652,
      "learning_rate": 8.51528384279476e-06,
      "loss": 0.9647,
      "step": 195
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.550218105316162,
      "learning_rate": 8.558951965065503e-06,
      "loss": 0.8882,
      "step": 196
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5973725318908691,
      "learning_rate": 8.602620087336245e-06,
      "loss": 0.9635,
      "step": 197
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5515429973602295,
      "learning_rate": 8.646288209606988e-06,
      "loss": 0.9952,
      "step": 198
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4325958490371704,
      "learning_rate": 8.68995633187773e-06,
      "loss": 0.9367,
      "step": 199
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4763720035552979,
      "learning_rate": 8.733624454148473e-06,
      "loss": 1.209,
      "step": 200
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4800736904144287,
      "learning_rate": 8.777292576419215e-06,
      "loss": 0.9519,
      "step": 201
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6222997903823853,
      "learning_rate": 8.820960698689956e-06,
      "loss": 0.8872,
      "step": 202
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.497116208076477,
      "learning_rate": 8.864628820960699e-06,
      "loss": 0.9646,
      "step": 203
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8553681969642639,
      "learning_rate": 8.908296943231441e-06,
      "loss": 1.1626,
      "step": 204
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7718945741653442,
      "learning_rate": 8.951965065502184e-06,
      "loss": 0.9696,
      "step": 205
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7568738460540771,
      "learning_rate": 8.995633187772926e-06,
      "loss": 1.1759,
      "step": 206
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6688908338546753,
      "learning_rate": 9.039301310043669e-06,
      "loss": 0.9689,
      "step": 207
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5790385007858276,
      "learning_rate": 9.082969432314411e-06,
      "loss": 0.958,
      "step": 208
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4463136196136475,
      "learning_rate": 9.126637554585154e-06,
      "loss": 0.9841,
      "step": 209
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5564645528793335,
      "learning_rate": 9.170305676855896e-06,
      "loss": 0.9251,
      "step": 210
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.415387749671936,
      "learning_rate": 9.213973799126639e-06,
      "loss": 0.9854,
      "step": 211
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5453453063964844,
      "learning_rate": 9.257641921397381e-06,
      "loss": 0.9746,
      "step": 212
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5361894369125366,
      "learning_rate": 9.301310043668122e-06,
      "loss": 0.9589,
      "step": 213
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.051392912864685,
      "learning_rate": 9.344978165938865e-06,
      "loss": 1.1822,
      "step": 214
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.599447250366211,
      "learning_rate": 9.388646288209607e-06,
      "loss": 1.001,
      "step": 215
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.797721266746521,
      "learning_rate": 9.43231441048035e-06,
      "loss": 1.1706,
      "step": 216
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8003242611885071,
      "learning_rate": 9.475982532751092e-06,
      "loss": 1.1594,
      "step": 217
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5409727096557617,
      "learning_rate": 9.519650655021835e-06,
      "loss": 0.9949,
      "step": 218
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5188815593719482,
      "learning_rate": 9.563318777292577e-06,
      "loss": 0.9077,
      "step": 219
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.48342764377594,
      "learning_rate": 9.60698689956332e-06,
      "loss": 0.9628,
      "step": 220
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4368282556533813,
      "learning_rate": 9.650655021834062e-06,
      "loss": 0.9223,
      "step": 221
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5326075553894043,
      "learning_rate": 9.694323144104805e-06,
      "loss": 0.9435,
      "step": 222
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4978830814361572,
      "learning_rate": 9.737991266375547e-06,
      "loss": 0.9441,
      "step": 223
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1965993642807007,
      "learning_rate": 9.781659388646288e-06,
      "loss": 1.175,
      "step": 224
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4991168975830078,
      "learning_rate": 9.82532751091703e-06,
      "loss": 0.9395,
      "step": 225
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.518691897392273,
      "learning_rate": 9.868995633187773e-06,
      "loss": 0.9874,
      "step": 226
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.516767144203186,
      "learning_rate": 9.912663755458516e-06,
      "loss": 0.9929,
      "step": 227
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0120187997817993,
      "learning_rate": 9.956331877729258e-06,
      "loss": 1.1693,
      "step": 228
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5007449388504028,
      "learning_rate": 1e-05,
      "loss": 0.8615,
      "step": 229
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3592734336853027,
      "learning_rate": 9.999999549658898e-06,
      "loss": 0.9076,
      "step": 230
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5135141611099243,
      "learning_rate": 9.999998198635674e-06,
      "loss": 0.9546,
      "step": 231
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9991651177406311,
      "learning_rate": 9.999995946930572e-06,
      "loss": 1.1749,
      "step": 232
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4234951734542847,
      "learning_rate": 9.999992794543995e-06,
      "loss": 0.9764,
      "step": 233
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5005040168762207,
      "learning_rate": 9.999988741476512e-06,
      "loss": 0.8566,
      "step": 234
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4868618249893188,
      "learning_rate": 9.999983787728855e-06,
      "loss": 1.021,
      "step": 235
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.425580382347107,
      "learning_rate": 9.999977933301912e-06,
      "loss": 0.9473,
      "step": 236
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9950155019760132,
      "learning_rate": 9.999971178196743e-06,
      "loss": 1.1828,
      "step": 237
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4027185440063477,
      "learning_rate": 9.999963522414561e-06,
      "loss": 0.9886,
      "step": 238
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3929423093795776,
      "learning_rate": 9.999954965956747e-06,
      "loss": 0.9005,
      "step": 239
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3437550067901611,
      "learning_rate": 9.999945508824842e-06,
      "loss": 0.9366,
      "step": 240
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7999411821365356,
      "learning_rate": 9.999935151020548e-06,
      "loss": 1.1719,
      "step": 241
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.493462324142456,
      "learning_rate": 9.999923892545734e-06,
      "loss": 0.9751,
      "step": 242
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5382442474365234,
      "learning_rate": 9.999911733402426e-06,
      "loss": 0.9569,
      "step": 243
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.745302677154541,
      "learning_rate": 9.999898673592812e-06,
      "loss": 1.1465,
      "step": 244
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5344090461730957,
      "learning_rate": 9.999884713119248e-06,
      "loss": 0.9478,
      "step": 245
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3987891674041748,
      "learning_rate": 9.99986985198425e-06,
      "loss": 0.8923,
      "step": 246
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.418421745300293,
      "learning_rate": 9.99985409019049e-06,
      "loss": 0.9073,
      "step": 247
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9021090865135193,
      "learning_rate": 9.99983742774081e-06,
      "loss": 1.1603,
      "step": 248
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3897669315338135,
      "learning_rate": 9.999819864638214e-06,
      "loss": 0.992,
      "step": 249
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5116395950317383,
      "learning_rate": 9.999801400885862e-06,
      "loss": 0.955,
      "step": 250
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.454531192779541,
      "learning_rate": 9.999782036487082e-06,
      "loss": 1.0095,
      "step": 251
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.09048330783844,
      "learning_rate": 9.99976177144536e-06,
      "loss": 1.1593,
      "step": 252
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7590124607086182,
      "learning_rate": 9.99974060576435e-06,
      "loss": 1.1647,
      "step": 253
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8064708113670349,
      "learning_rate": 9.99971853944786e-06,
      "loss": 1.1755,
      "step": 254
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9388291835784912,
      "learning_rate": 9.99969557249987e-06,
      "loss": 1.1719,
      "step": 255
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.782835841178894,
      "learning_rate": 9.999671704924514e-06,
      "loss": 0.8983,
      "step": 256
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5426616668701172,
      "learning_rate": 9.999646936726094e-06,
      "loss": 0.8918,
      "step": 257
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.425520658493042,
      "learning_rate": 9.999621267909068e-06,
      "loss": 0.9417,
      "step": 258
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4330857992172241,
      "learning_rate": 9.999594698478063e-06,
      "loss": 0.969,
      "step": 259
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5181360244750977,
      "learning_rate": 9.999567228437864e-06,
      "loss": 0.9226,
      "step": 260
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4494251012802124,
      "learning_rate": 9.999538857793419e-06,
      "loss": 0.9909,
      "step": 261
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5584813356399536,
      "learning_rate": 9.999509586549837e-06,
      "loss": 0.953,
      "step": 262
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6014645099639893,
      "learning_rate": 9.999479414712395e-06,
      "loss": 0.9762,
      "step": 263
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.355467438697815,
      "learning_rate": 9.999448342286526e-06,
      "loss": 0.9444,
      "step": 264
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4598685503005981,
      "learning_rate": 9.999416369277825e-06,
      "loss": 0.9722,
      "step": 265
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6052550077438354,
      "learning_rate": 9.999383495692056e-06,
      "loss": 0.9089,
      "step": 266
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.41050386428833,
      "learning_rate": 9.999349721535136e-06,
      "loss": 0.9215,
      "step": 267
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4923022985458374,
      "learning_rate": 9.999315046813153e-06,
      "loss": 0.9729,
      "step": 268
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.411473035812378,
      "learning_rate": 9.99927947153235e-06,
      "loss": 0.8865,
      "step": 269
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.50513756275177,
      "learning_rate": 9.99924299569914e-06,
      "loss": 0.9534,
      "step": 270
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4415781497955322,
      "learning_rate": 9.999205619320087e-06,
      "loss": 0.9022,
      "step": 271
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2333358526229858,
      "learning_rate": 9.99916734240193e-06,
      "loss": 1.1712,
      "step": 272
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9638535976409912,
      "learning_rate": 9.999128164951559e-06,
      "loss": 1.1827,
      "step": 273
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7556527853012085,
      "learning_rate": 9.999088086976034e-06,
      "loss": 1.1467,
      "step": 274
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9951626658439636,
      "learning_rate": 9.999047108482576e-06,
      "loss": 1.1722,
      "step": 275
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7519001960754395,
      "learning_rate": 9.999005229478561e-06,
      "loss": 0.9628,
      "step": 276
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6137698888778687,
      "learning_rate": 9.998962449971542e-06,
      "loss": 0.9378,
      "step": 277
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0728204250335693,
      "learning_rate": 9.998918769969216e-06,
      "loss": 1.1559,
      "step": 278
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6895015239715576,
      "learning_rate": 9.998874189479457e-06,
      "loss": 0.959,
      "step": 279
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5698795318603516,
      "learning_rate": 9.998828708510295e-06,
      "loss": 0.9152,
      "step": 280
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9829255938529968,
      "learning_rate": 9.99878232706992e-06,
      "loss": 1.1406,
      "step": 281
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6114037036895752,
      "learning_rate": 9.99873504516669e-06,
      "loss": 0.9233,
      "step": 282
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3925020694732666,
      "learning_rate": 9.998686862809119e-06,
      "loss": 0.9368,
      "step": 283
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7990014553070068,
      "learning_rate": 9.998637780005889e-06,
      "loss": 1.1819,
      "step": 284
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.429306983947754,
      "learning_rate": 9.998587796765842e-06,
      "loss": 0.9181,
      "step": 285
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4822051525115967,
      "learning_rate": 9.99853691309798e-06,
      "loss": 0.9888,
      "step": 286
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3861721754074097,
      "learning_rate": 9.998485129011468e-06,
      "loss": 0.9315,
      "step": 287
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4019639492034912,
      "learning_rate": 9.998432444515638e-06,
      "loss": 1.0158,
      "step": 288
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2986009120941162,
      "learning_rate": 9.998378859619978e-06,
      "loss": 0.924,
      "step": 289
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4250671863555908,
      "learning_rate": 9.99832437433414e-06,
      "loss": 0.9301,
      "step": 290
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.342633843421936,
      "learning_rate": 9.99826898866794e-06,
      "loss": 0.9475,
      "step": 291
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.397358775138855,
      "learning_rate": 9.998212702631355e-06,
      "loss": 0.9478,
      "step": 292
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.350216031074524,
      "learning_rate": 9.998155516234523e-06,
      "loss": 0.9135,
      "step": 293
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5043015480041504,
      "learning_rate": 9.998097429487748e-06,
      "loss": 0.8892,
      "step": 294
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3600742816925049,
      "learning_rate": 9.99803844240149e-06,
      "loss": 0.8642,
      "step": 295
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3880120515823364,
      "learning_rate": 9.997978554986376e-06,
      "loss": 0.9318,
      "step": 296
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3473639488220215,
      "learning_rate": 9.997917767253195e-06,
      "loss": 0.9297,
      "step": 297
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0950576066970825,
      "learning_rate": 9.997856079212896e-06,
      "loss": 1.1513,
      "step": 298
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.884467363357544,
      "learning_rate": 9.997793490876593e-06,
      "loss": 1.1602,
      "step": 299
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5468956232070923,
      "learning_rate": 9.997730002255559e-06,
      "loss": 0.9661,
      "step": 300
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.517162799835205,
      "learning_rate": 9.99766561336123e-06,
      "loss": 0.9863,
      "step": 301
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5745347738265991,
      "learning_rate": 9.997600324205204e-06,
      "loss": 0.9436,
      "step": 302
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4865670204162598,
      "learning_rate": 9.997534134799244e-06,
      "loss": 0.9264,
      "step": 303
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7446842193603516,
      "learning_rate": 9.997467045155276e-06,
      "loss": 1.1577,
      "step": 304
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4508453607559204,
      "learning_rate": 9.997399055285379e-06,
      "loss": 1.1745,
      "step": 305
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5860751867294312,
      "learning_rate": 9.997330165201805e-06,
      "loss": 0.9045,
      "step": 306
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5972726345062256,
      "learning_rate": 9.99726037491696e-06,
      "loss": 0.9096,
      "step": 307
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0144840478897095,
      "learning_rate": 9.997189684443418e-06,
      "loss": 1.181,
      "step": 308
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0196070671081543,
      "learning_rate": 9.997118093793912e-06,
      "loss": 1.1738,
      "step": 309
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8360738754272461,
      "learning_rate": 9.99704560298134e-06,
      "loss": 1.1702,
      "step": 310
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8462754487991333,
      "learning_rate": 9.996972212018758e-06,
      "loss": 1.1862,
      "step": 311
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8914956450462341,
      "learning_rate": 9.996897920919388e-06,
      "loss": 1.1595,
      "step": 312
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8713598251342773,
      "learning_rate": 9.996822729696613e-06,
      "loss": 0.8628,
      "step": 313
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4946718215942383,
      "learning_rate": 9.996746638363976e-06,
      "loss": 0.91,
      "step": 314
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5615490674972534,
      "learning_rate": 9.996669646935185e-06,
      "loss": 0.8675,
      "step": 315
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5331690311431885,
      "learning_rate": 9.996591755424105e-06,
      "loss": 1.0021,
      "step": 316
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6537284851074219,
      "learning_rate": 9.996512963844774e-06,
      "loss": 0.9484,
      "step": 317
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5538790225982666,
      "learning_rate": 9.996433272211381e-06,
      "loss": 0.917,
      "step": 318
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4987270832061768,
      "learning_rate": 9.996352680538281e-06,
      "loss": 0.8907,
      "step": 319
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4629443883895874,
      "learning_rate": 9.996271188839994e-06,
      "loss": 0.932,
      "step": 320
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1400537490844727,
      "learning_rate": 9.996188797131197e-06,
      "loss": 1.1805,
      "step": 321
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5498771667480469,
      "learning_rate": 9.996105505426733e-06,
      "loss": 0.9234,
      "step": 322
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5501476526260376,
      "learning_rate": 9.996021313741606e-06,
      "loss": 0.9539,
      "step": 323
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5089913606643677,
      "learning_rate": 9.995936222090982e-06,
      "loss": 0.8937,
      "step": 324
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3915530443191528,
      "learning_rate": 9.995850230490188e-06,
      "loss": 0.9139,
      "step": 325
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4444496631622314,
      "learning_rate": 9.995763338954715e-06,
      "loss": 0.9007,
      "step": 326
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4394372701644897,
      "learning_rate": 9.995675547500216e-06,
      "loss": 0.8876,
      "step": 327
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0779564380645752,
      "learning_rate": 9.995586856142507e-06,
      "loss": 1.1589,
      "step": 328
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.674628496170044,
      "learning_rate": 9.99549726489756e-06,
      "loss": 0.902,
      "step": 329
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5453639030456543,
      "learning_rate": 9.995406773781518e-06,
      "loss": 0.8886,
      "step": 330
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4017726182937622,
      "learning_rate": 9.995315382810677e-06,
      "loss": 0.891,
      "step": 331
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0223113298416138,
      "learning_rate": 9.995223092001505e-06,
      "loss": 1.1594,
      "step": 332
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3800009489059448,
      "learning_rate": 9.995129901370624e-06,
      "loss": 0.8454,
      "step": 333
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.554129719734192,
      "learning_rate": 9.995035810934821e-06,
      "loss": 0.9562,
      "step": 334
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7765074372291565,
      "learning_rate": 9.994940820711045e-06,
      "loss": 1.1374,
      "step": 335
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7265048027038574,
      "learning_rate": 9.99484493071641e-06,
      "loss": 1.1531,
      "step": 336
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4466677904129028,
      "learning_rate": 9.994748140968187e-06,
      "loss": 0.8977,
      "step": 337
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4053211212158203,
      "learning_rate": 9.99465045148381e-06,
      "loss": 0.9059,
      "step": 338
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.425667643547058,
      "learning_rate": 9.994551862280879e-06,
      "loss": 1.0127,
      "step": 339
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2311300039291382,
      "learning_rate": 9.994452373377152e-06,
      "loss": 0.9043,
      "step": 340
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0571866035461426,
      "learning_rate": 9.994351984790552e-06,
      "loss": 1.1503,
      "step": 341
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8495367765426636,
      "learning_rate": 9.994250696539162e-06,
      "loss": 1.1662,
      "step": 342
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4783291816711426,
      "learning_rate": 9.994148508641228e-06,
      "loss": 0.8513,
      "step": 343
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.463494896888733,
      "learning_rate": 9.994045421115155e-06,
      "loss": 0.9286,
      "step": 344
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6140416860580444,
      "learning_rate": 9.993941433979516e-06,
      "loss": 0.9271,
      "step": 345
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0989748239517212,
      "learning_rate": 9.993836547253041e-06,
      "loss": 1.181,
      "step": 346
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0807011127471924,
      "learning_rate": 9.993730760954628e-06,
      "loss": 1.1477,
      "step": 347
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4601367712020874,
      "learning_rate": 9.993624075103328e-06,
      "loss": 0.9788,
      "step": 348
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5152372121810913,
      "learning_rate": 9.99351648971836e-06,
      "loss": 0.9371,
      "step": 349
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0492844581604004,
      "learning_rate": 9.993408004819104e-06,
      "loss": 1.1355,
      "step": 350
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4431166648864746,
      "learning_rate": 9.993298620425106e-06,
      "loss": 0.9533,
      "step": 351
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4595119953155518,
      "learning_rate": 9.993188336556065e-06,
      "loss": 0.8316,
      "step": 352
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9447420835494995,
      "learning_rate": 9.99307715323185e-06,
      "loss": 1.1422,
      "step": 353
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8186748027801514,
      "learning_rate": 9.992965070472487e-06,
      "loss": 0.9135,
      "step": 354
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4124977588653564,
      "learning_rate": 9.992852088298171e-06,
      "loss": 0.8799,
      "step": 355
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4483859539031982,
      "learning_rate": 9.992738206729247e-06,
      "loss": 0.8806,
      "step": 356
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4455571174621582,
      "learning_rate": 9.992623425786235e-06,
      "loss": 0.9097,
      "step": 357
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.024294376373291,
      "learning_rate": 9.992507745489808e-06,
      "loss": 1.1572,
      "step": 358
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3613418340682983,
      "learning_rate": 9.992391165860806e-06,
      "loss": 0.9037,
      "step": 359
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4043768644332886,
      "learning_rate": 9.992273686920229e-06,
      "loss": 0.9524,
      "step": 360
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3769153356552124,
      "learning_rate": 9.992155308689239e-06,
      "loss": 0.854,
      "step": 361
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4317244291305542,
      "learning_rate": 9.992036031189161e-06,
      "loss": 0.8968,
      "step": 362
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3540799617767334,
      "learning_rate": 9.991915854441478e-06,
      "loss": 0.879,
      "step": 363
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4168645143508911,
      "learning_rate": 9.991794778467843e-06,
      "loss": 0.9115,
      "step": 364
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4521801471710205,
      "learning_rate": 9.991672803290061e-06,
      "loss": 0.896,
      "step": 365
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.457749366760254,
      "learning_rate": 9.991549928930111e-06,
      "loss": 1.172,
      "step": 366
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.397606372833252,
      "learning_rate": 9.99142615541012e-06,
      "loss": 0.8818,
      "step": 367
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3981763124465942,
      "learning_rate": 9.99130148275239e-06,
      "loss": 0.9268,
      "step": 368
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3646825551986694,
      "learning_rate": 9.991175910979374e-06,
      "loss": 0.9318,
      "step": 369
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4101550579071045,
      "learning_rate": 9.991049440113693e-06,
      "loss": 0.9055,
      "step": 370
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4720388650894165,
      "learning_rate": 9.990922070178133e-06,
      "loss": 0.9101,
      "step": 371
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4557220935821533,
      "learning_rate": 9.990793801195636e-06,
      "loss": 0.9364,
      "step": 372
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.180110216140747,
      "learning_rate": 9.990664633189306e-06,
      "loss": 1.1389,
      "step": 373
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5308270454406738,
      "learning_rate": 9.99053456618241e-06,
      "loss": 0.8701,
      "step": 374
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5155760049819946,
      "learning_rate": 9.990403600198383e-06,
      "loss": 0.9133,
      "step": 375
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4088834524154663,
      "learning_rate": 9.990271735260813e-06,
      "loss": 0.8789,
      "step": 376
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4935065507888794,
      "learning_rate": 9.990138971393456e-06,
      "loss": 0.848,
      "step": 377
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0805315971374512,
      "learning_rate": 9.990005308620223e-06,
      "loss": 1.1632,
      "step": 378
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.521833896636963,
      "learning_rate": 9.989870746965195e-06,
      "loss": 0.9583,
      "step": 379
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4206234216690063,
      "learning_rate": 9.989735286452613e-06,
      "loss": 0.8943,
      "step": 380
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.497657060623169,
      "learning_rate": 9.989598927106874e-06,
      "loss": 0.9168,
      "step": 381
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4155974388122559,
      "learning_rate": 9.989461668952545e-06,
      "loss": 0.9019,
      "step": 382
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4603191614151,
      "learning_rate": 9.989323512014351e-06,
      "loss": 0.9586,
      "step": 383
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4476569890975952,
      "learning_rate": 9.989184456317176e-06,
      "loss": 0.9204,
      "step": 384
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4836901426315308,
      "learning_rate": 9.989044501886072e-06,
      "loss": 0.8268,
      "step": 385
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4101026058197021,
      "learning_rate": 9.98890364874625e-06,
      "loss": 0.9477,
      "step": 386
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3532098531723022,
      "learning_rate": 9.988761896923078e-06,
      "loss": 0.8897,
      "step": 387
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3791732788085938,
      "learning_rate": 9.988619246442098e-06,
      "loss": 0.9356,
      "step": 388
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4708971977233887,
      "learning_rate": 9.988475697328999e-06,
      "loss": 0.9053,
      "step": 389
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4446396827697754,
      "learning_rate": 9.988331249609646e-06,
      "loss": 0.9158,
      "step": 390
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3244329690933228,
      "learning_rate": 9.988185903310058e-06,
      "loss": 1.1513,
      "step": 391
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.428970217704773,
      "learning_rate": 9.988039658456411e-06,
      "loss": 0.8959,
      "step": 392
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6890389919281006,
      "learning_rate": 9.987892515075058e-06,
      "loss": 0.8993,
      "step": 393
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4718921184539795,
      "learning_rate": 9.9877444731925e-06,
      "loss": 0.939,
      "step": 394
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3758097887039185,
      "learning_rate": 9.987595532835403e-06,
      "loss": 0.8853,
      "step": 395
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3942396640777588,
      "learning_rate": 9.987445694030601e-06,
      "loss": 0.9187,
      "step": 396
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4720768928527832,
      "learning_rate": 9.987294956805082e-06,
      "loss": 0.9587,
      "step": 397
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.069543480873108,
      "learning_rate": 9.987143321186002e-06,
      "loss": 1.1253,
      "step": 398
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4655293226242065,
      "learning_rate": 9.986990787200675e-06,
      "loss": 0.922,
      "step": 399
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4135407209396362,
      "learning_rate": 9.986837354876578e-06,
      "loss": 0.9215,
      "step": 400
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.35868239402771,
      "learning_rate": 9.986683024241348e-06,
      "loss": 0.9096,
      "step": 401
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.512502670288086,
      "learning_rate": 9.986527795322787e-06,
      "loss": 0.896,
      "step": 402
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4110939502716064,
      "learning_rate": 9.986371668148858e-06,
      "loss": 0.9215,
      "step": 403
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0733145475387573,
      "learning_rate": 9.986214642747685e-06,
      "loss": 1.1564,
      "step": 404
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4007169008255005,
      "learning_rate": 9.986056719147553e-06,
      "loss": 0.9532,
      "step": 405
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7566277384757996,
      "learning_rate": 9.98589789737691e-06,
      "loss": 1.1363,
      "step": 406
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4983266592025757,
      "learning_rate": 9.985738177464367e-06,
      "loss": 0.8741,
      "step": 407
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4130704402923584,
      "learning_rate": 9.985577559438694e-06,
      "loss": 0.8624,
      "step": 408
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0458935499191284,
      "learning_rate": 9.985416043328823e-06,
      "loss": 1.156,
      "step": 409
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4352269172668457,
      "learning_rate": 9.985253629163853e-06,
      "loss": 0.9561,
      "step": 410
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4386528730392456,
      "learning_rate": 9.985090316973035e-06,
      "loss": 0.9048,
      "step": 411
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4551750421524048,
      "learning_rate": 9.984926106785795e-06,
      "loss": 0.9586,
      "step": 412
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4629510641098022,
      "learning_rate": 9.984760998631706e-06,
      "loss": 0.9467,
      "step": 413
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4256954193115234,
      "learning_rate": 9.984594992540513e-06,
      "loss": 0.872,
      "step": 414
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4283626079559326,
      "learning_rate": 9.984428088542121e-06,
      "loss": 0.8749,
      "step": 415
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4481284618377686,
      "learning_rate": 9.984260286666592e-06,
      "loss": 0.8962,
      "step": 416
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4917922019958496,
      "learning_rate": 9.984091586944158e-06,
      "loss": 0.8386,
      "step": 417
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.621300220489502,
      "learning_rate": 9.983921989405203e-06,
      "loss": 1.151,
      "step": 418
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4182554483413696,
      "learning_rate": 9.98375149408028e-06,
      "loss": 0.8547,
      "step": 419
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.610759973526001,
      "learning_rate": 9.983580101000103e-06,
      "loss": 0.8829,
      "step": 420
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7343606948852539,
      "learning_rate": 9.983407810195544e-06,
      "loss": 1.1294,
      "step": 421
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4991573095321655,
      "learning_rate": 9.983234621697639e-06,
      "loss": 0.9364,
      "step": 422
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3895467519760132,
      "learning_rate": 9.983060535537587e-06,
      "loss": 0.8646,
      "step": 423
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3933253288269043,
      "learning_rate": 9.982885551746745e-06,
      "loss": 0.8718,
      "step": 424
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3329625129699707,
      "learning_rate": 9.982709670356636e-06,
      "loss": 0.9041,
      "step": 425
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6579738855361938,
      "learning_rate": 9.982532891398941e-06,
      "loss": 0.8983,
      "step": 426
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2048251628875732,
      "learning_rate": 9.982355214905506e-06,
      "loss": 1.1363,
      "step": 427
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4993442296981812,
      "learning_rate": 9.982176640908337e-06,
      "loss": 0.9146,
      "step": 428
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7631571888923645,
      "learning_rate": 9.981997169439599e-06,
      "loss": 1.1473,
      "step": 429
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.434574007987976,
      "learning_rate": 9.981816800531623e-06,
      "loss": 0.9587,
      "step": 430
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.362691879272461,
      "learning_rate": 9.981635534216902e-06,
      "loss": 0.8749,
      "step": 431
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4217510223388672,
      "learning_rate": 9.981453370528085e-06,
      "loss": 0.9726,
      "step": 432
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3622015714645386,
      "learning_rate": 9.98127030949799e-06,
      "loss": 0.8724,
      "step": 433
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4525446891784668,
      "learning_rate": 9.98108635115959e-06,
      "loss": 1.1578,
      "step": 434
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4380748271942139,
      "learning_rate": 9.980901495546025e-06,
      "loss": 0.932,
      "step": 435
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3715463876724243,
      "learning_rate": 9.980715742690592e-06,
      "loss": 0.8845,
      "step": 436
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2973045110702515,
      "learning_rate": 9.980529092626754e-06,
      "loss": 0.8821,
      "step": 437
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3687312602996826,
      "learning_rate": 9.980341545388132e-06,
      "loss": 0.8728,
      "step": 438
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8545926809310913,
      "learning_rate": 9.980153101008509e-06,
      "loss": 1.1571,
      "step": 439
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4665417671203613,
      "learning_rate": 9.979963759521833e-06,
      "loss": 0.9549,
      "step": 440
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4593017101287842,
      "learning_rate": 9.979773520962212e-06,
      "loss": 0.9215,
      "step": 441
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4154950380325317,
      "learning_rate": 9.979582385363912e-06,
      "loss": 0.9206,
      "step": 442
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4294720888137817,
      "learning_rate": 9.979390352761364e-06,
      "loss": 0.8835,
      "step": 443
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.404213786125183,
      "learning_rate": 9.979197423189163e-06,
      "loss": 0.8392,
      "step": 444
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4412208795547485,
      "learning_rate": 9.979003596682058e-06,
      "loss": 0.8446,
      "step": 445
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4696388244628906,
      "learning_rate": 9.978808873274968e-06,
      "loss": 0.8592,
      "step": 446
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5328738689422607,
      "learning_rate": 9.978613253002968e-06,
      "loss": 0.9161,
      "step": 447
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.445266604423523,
      "learning_rate": 9.9784167359013e-06,
      "loss": 0.9329,
      "step": 448
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3907238245010376,
      "learning_rate": 9.978219322005358e-06,
      "loss": 0.8988,
      "step": 449
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.971100389957428,
      "learning_rate": 9.978021011350706e-06,
      "loss": 1.1644,
      "step": 450
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.388284683227539,
      "learning_rate": 9.977821803973067e-06,
      "loss": 0.8952,
      "step": 451
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6878868341445923,
      "learning_rate": 9.977621699908327e-06,
      "loss": 1.1586,
      "step": 452
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.544853925704956,
      "learning_rate": 9.97742069919253e-06,
      "loss": 0.9088,
      "step": 453
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3917094469070435,
      "learning_rate": 9.977218801861884e-06,
      "loss": 0.8937,
      "step": 454
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3360214233398438,
      "learning_rate": 9.97701600795276e-06,
      "loss": 0.9098,
      "step": 455
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4916146993637085,
      "learning_rate": 9.976812317501685e-06,
      "loss": 0.8931,
      "step": 456
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2459080219268799,
      "learning_rate": 9.976607730545354e-06,
      "loss": 1.1553,
      "step": 457
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4430489540100098,
      "learning_rate": 9.97640224712062e-06,
      "loss": 0.8994,
      "step": 458
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.40715491771698,
      "learning_rate": 9.976195867264496e-06,
      "loss": 0.9023,
      "step": 459
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3742717504501343,
      "learning_rate": 9.97598859101416e-06,
      "loss": 0.8899,
      "step": 460
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4290109872817993,
      "learning_rate": 9.975780418406953e-06,
      "loss": 0.9037,
      "step": 461
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3236624002456665,
      "learning_rate": 9.975571349480368e-06,
      "loss": 0.89,
      "step": 462
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3590117692947388,
      "learning_rate": 9.975361384272073e-06,
      "loss": 0.9206,
      "step": 463
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.401292324066162,
      "learning_rate": 9.975150522819886e-06,
      "loss": 0.8776,
      "step": 464
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5784580707550049,
      "learning_rate": 9.97493876516179e-06,
      "loss": 0.8437,
      "step": 465
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4259581565856934,
      "learning_rate": 9.974726111335932e-06,
      "loss": 0.9213,
      "step": 466
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1834633350372314,
      "learning_rate": 9.974512561380622e-06,
      "loss": 1.137,
      "step": 467
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4389088153839111,
      "learning_rate": 9.974298115334322e-06,
      "loss": 0.8804,
      "step": 468
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4307321310043335,
      "learning_rate": 9.974082773235665e-06,
      "loss": 0.9033,
      "step": 469
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8018472194671631,
      "learning_rate": 9.973866535123442e-06,
      "loss": 1.1342,
      "step": 470
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7572311162948608,
      "learning_rate": 9.973649401036605e-06,
      "loss": 1.1443,
      "step": 471
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4835035800933838,
      "learning_rate": 9.973431371014266e-06,
      "loss": 0.9315,
      "step": 472
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7252900004386902,
      "learning_rate": 9.973212445095704e-06,
      "loss": 1.1155,
      "step": 473
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.396541953086853,
      "learning_rate": 9.97299262332035e-06,
      "loss": 0.8398,
      "step": 474
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3422585725784302,
      "learning_rate": 9.972771905727807e-06,
      "loss": 0.8613,
      "step": 475
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8429656028747559,
      "learning_rate": 9.972550292357833e-06,
      "loss": 1.1405,
      "step": 476
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7898710370063782,
      "learning_rate": 9.972327783250348e-06,
      "loss": 1.1412,
      "step": 477
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6817870140075684,
      "learning_rate": 9.972104378445434e-06,
      "loss": 1.1427,
      "step": 478
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.447851300239563,
      "learning_rate": 9.971880077983334e-06,
      "loss": 0.8773,
      "step": 479
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4586381912231445,
      "learning_rate": 9.97165488190445e-06,
      "loss": 0.9022,
      "step": 480
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3647269010543823,
      "learning_rate": 9.971428790249355e-06,
      "loss": 0.9363,
      "step": 481
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.138634443283081,
      "learning_rate": 9.971201803058772e-06,
      "loss": 1.1434,
      "step": 482
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3617175817489624,
      "learning_rate": 9.97097392037359e-06,
      "loss": 0.9792,
      "step": 483
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.337128758430481,
      "learning_rate": 9.970745142234859e-06,
      "loss": 0.9195,
      "step": 484
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4832465648651123,
      "learning_rate": 9.970515468683789e-06,
      "loss": 0.9092,
      "step": 485
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.463374376296997,
      "learning_rate": 9.970284899761756e-06,
      "loss": 0.9153,
      "step": 486
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8973669409751892,
      "learning_rate": 9.97005343551029e-06,
      "loss": 1.1264,
      "step": 487
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7781697511672974,
      "learning_rate": 9.96982107597109e-06,
      "loss": 1.1456,
      "step": 488
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5386170148849487,
      "learning_rate": 9.969587821186008e-06,
      "loss": 0.9254,
      "step": 489
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4629544019699097,
      "learning_rate": 9.969353671197064e-06,
      "loss": 0.8472,
      "step": 490
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3808633089065552,
      "learning_rate": 9.969118626046439e-06,
      "loss": 0.8866,
      "step": 491
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1610136032104492,
      "learning_rate": 9.96888268577647e-06,
      "loss": 1.1541,
      "step": 492
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.366868019104004,
      "learning_rate": 9.968645850429659e-06,
      "loss": 0.8781,
      "step": 493
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3248481750488281,
      "learning_rate": 9.968408120048671e-06,
      "loss": 0.8463,
      "step": 494
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8095569014549255,
      "learning_rate": 9.968169494676327e-06,
      "loss": 1.1379,
      "step": 495
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4439442157745361,
      "learning_rate": 9.967929974355615e-06,
      "loss": 0.8644,
      "step": 496
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6884865760803223,
      "learning_rate": 9.967689559129677e-06,
      "loss": 0.9777,
      "step": 497
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4539486169815063,
      "learning_rate": 9.967448249041825e-06,
      "loss": 0.9126,
      "step": 498
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1375354528427124,
      "learning_rate": 9.967206044135526e-06,
      "loss": 1.1253,
      "step": 499
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4684480428695679,
      "learning_rate": 9.966962944454408e-06,
      "loss": 0.8515,
      "step": 500
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4118716716766357,
      "learning_rate": 9.966718950042267e-06,
      "loss": 0.8487,
      "step": 501
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7854005694389343,
      "learning_rate": 9.96647406094305e-06,
      "loss": 1.1296,
      "step": 502
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7908424735069275,
      "learning_rate": 9.966228277200874e-06,
      "loss": 1.1296,
      "step": 503
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4501421451568604,
      "learning_rate": 9.965981598860012e-06,
      "loss": 0.8997,
      "step": 504
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4971119165420532,
      "learning_rate": 9.965734025964899e-06,
      "loss": 0.9421,
      "step": 505
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9205259084701538,
      "learning_rate": 9.965485558560134e-06,
      "loss": 1.123,
      "step": 506
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8977799415588379,
      "learning_rate": 9.965236196690475e-06,
      "loss": 1.1622,
      "step": 507
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3620481491088867,
      "learning_rate": 9.96498594040084e-06,
      "loss": 0.8797,
      "step": 508
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8201619982719421,
      "learning_rate": 9.964734789736307e-06,
      "loss": 1.1447,
      "step": 509
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3355612754821777,
      "learning_rate": 9.964482744742123e-06,
      "loss": 0.8503,
      "step": 510
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8511717915534973,
      "learning_rate": 9.964229805463685e-06,
      "loss": 1.1409,
      "step": 511
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3619377613067627,
      "learning_rate": 9.96397597194656e-06,
      "loss": 0.9549,
      "step": 512
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3890559673309326,
      "learning_rate": 9.963721244236473e-06,
      "loss": 0.9197,
      "step": 513
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8600515127182007,
      "learning_rate": 9.963465622379308e-06,
      "loss": 1.1212,
      "step": 514
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3459125757217407,
      "learning_rate": 9.963209106421109e-06,
      "loss": 0.8716,
      "step": 515
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3902032375335693,
      "learning_rate": 9.962951696408089e-06,
      "loss": 0.8421,
      "step": 516
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3468577861785889,
      "learning_rate": 9.962693392386615e-06,
      "loss": 0.8964,
      "step": 517
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2846554517745972,
      "learning_rate": 9.962434194403217e-06,
      "loss": 0.9014,
      "step": 518
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3279091119766235,
      "learning_rate": 9.962174102504586e-06,
      "loss": 0.9206,
      "step": 519
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4374936819076538,
      "learning_rate": 9.961913116737573e-06,
      "loss": 0.9355,
      "step": 520
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4977034330368042,
      "learning_rate": 9.961651237149194e-06,
      "loss": 0.9089,
      "step": 521
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.470560908317566,
      "learning_rate": 9.961388463786619e-06,
      "loss": 0.9329,
      "step": 522
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3447781801223755,
      "learning_rate": 9.961124796697185e-06,
      "loss": 0.9452,
      "step": 523
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3796138763427734,
      "learning_rate": 9.96086023592839e-06,
      "loss": 0.8287,
      "step": 524
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3267631530761719,
      "learning_rate": 9.960594781527887e-06,
      "loss": 0.8503,
      "step": 525
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4078810214996338,
      "learning_rate": 9.960328433543499e-06,
      "loss": 0.8848,
      "step": 526
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3252513408660889,
      "learning_rate": 9.9600611920232e-06,
      "loss": 0.8586,
      "step": 527
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8784031867980957,
      "learning_rate": 9.959793057015133e-06,
      "loss": 1.1239,
      "step": 528
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3384171724319458,
      "learning_rate": 9.959524028567598e-06,
      "loss": 0.8943,
      "step": 529
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3979865312576294,
      "learning_rate": 9.959254106729057e-06,
      "loss": 0.8709,
      "step": 530
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6968097686767578,
      "learning_rate": 9.958983291548131e-06,
      "loss": 1.1375,
      "step": 531
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.41080904006958,
      "learning_rate": 9.958711583073608e-06,
      "loss": 0.891,
      "step": 532
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3641427755355835,
      "learning_rate": 9.958438981354427e-06,
      "loss": 0.8384,
      "step": 533
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3597586154937744,
      "learning_rate": 9.9581654864397e-06,
      "loss": 0.8743,
      "step": 534
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3779219388961792,
      "learning_rate": 9.957891098378688e-06,
      "loss": 0.9169,
      "step": 535
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.396963357925415,
      "learning_rate": 9.957615817220818e-06,
      "loss": 0.9061,
      "step": 536
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9208986163139343,
      "learning_rate": 9.957339643015682e-06,
      "loss": 1.1135,
      "step": 537
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6810376048088074,
      "learning_rate": 9.957062575813027e-06,
      "loss": 1.1103,
      "step": 538
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4591164588928223,
      "learning_rate": 9.956784615662764e-06,
      "loss": 0.9671,
      "step": 539
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9406235814094543,
      "learning_rate": 9.956505762614963e-06,
      "loss": 1.1457,
      "step": 540
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4707534313201904,
      "learning_rate": 9.956226016719855e-06,
      "loss": 0.8336,
      "step": 541
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.498218059539795,
      "learning_rate": 9.955945378027832e-06,
      "loss": 0.9083,
      "step": 542
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3331184387207031,
      "learning_rate": 9.955663846589448e-06,
      "loss": 0.9125,
      "step": 543
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.335111379623413,
      "learning_rate": 9.955381422455418e-06,
      "loss": 0.8688,
      "step": 544
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9933720827102661,
      "learning_rate": 9.955098105676617e-06,
      "loss": 1.1404,
      "step": 545
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2994489669799805,
      "learning_rate": 9.954813896304078e-06,
      "loss": 0.8467,
      "step": 546
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3621206283569336,
      "learning_rate": 9.954528794389002e-06,
      "loss": 0.8966,
      "step": 547
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7093163132667542,
      "learning_rate": 9.954242799982742e-06,
      "loss": 1.1484,
      "step": 548
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7090497612953186,
      "learning_rate": 9.953955913136817e-06,
      "loss": 1.1165,
      "step": 549
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3854851722717285,
      "learning_rate": 9.953668133902907e-06,
      "loss": 0.8323,
      "step": 550
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4158225059509277,
      "learning_rate": 9.95337946233285e-06,
      "loss": 0.9133,
      "step": 551
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4152077436447144,
      "learning_rate": 9.95308989847865e-06,
      "loss": 0.8869,
      "step": 552
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2924901247024536,
      "learning_rate": 9.952799442392462e-06,
      "loss": 0.8684,
      "step": 553
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.413560152053833,
      "learning_rate": 9.952508094126611e-06,
      "loss": 0.8518,
      "step": 554
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.135731816291809,
      "learning_rate": 9.952215853733581e-06,
      "loss": 1.1487,
      "step": 555
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4321357011795044,
      "learning_rate": 9.951922721266013e-06,
      "loss": 0.8688,
      "step": 556
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4145641326904297,
      "learning_rate": 9.95162869677671e-06,
      "loss": 0.9331,
      "step": 557
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3561650514602661,
      "learning_rate": 9.95133378031864e-06,
      "loss": 0.93,
      "step": 558
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2764321565628052,
      "learning_rate": 9.951037971944923e-06,
      "loss": 0.8471,
      "step": 559
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4207507371902466,
      "learning_rate": 9.95074127170885e-06,
      "loss": 0.9127,
      "step": 560
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2545958757400513,
      "learning_rate": 9.950443679663867e-06,
      "loss": 0.886,
      "step": 561
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2977739572525024,
      "learning_rate": 9.950145195863577e-06,
      "loss": 0.8749,
      "step": 562
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3941473960876465,
      "learning_rate": 9.94984582036175e-06,
      "loss": 0.8924,
      "step": 563
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2984622716903687,
      "learning_rate": 9.949545553212317e-06,
      "loss": 0.8772,
      "step": 564
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2627735137939453,
      "learning_rate": 9.949244394469362e-06,
      "loss": 0.8528,
      "step": 565
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2966928482055664,
      "learning_rate": 9.948942344187141e-06,
      "loss": 1.1497,
      "step": 566
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.187267303466797,
      "learning_rate": 9.948639402420058e-06,
      "loss": 0.8781,
      "step": 567
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.70731520652771,
      "learning_rate": 9.948335569222687e-06,
      "loss": 1.1269,
      "step": 568
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7971781492233276,
      "learning_rate": 9.94803084464976e-06,
      "loss": 1.1264,
      "step": 569
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3970484733581543,
      "learning_rate": 9.947725228756168e-06,
      "loss": 0.8926,
      "step": 570
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3961576223373413,
      "learning_rate": 9.947418721596963e-06,
      "loss": 0.893,
      "step": 571
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4157097339630127,
      "learning_rate": 9.94711132322736e-06,
      "loss": 0.9153,
      "step": 572
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.474265456199646,
      "learning_rate": 9.94680303370273e-06,
      "loss": 0.9051,
      "step": 573
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.456728219985962,
      "learning_rate": 9.946493853078609e-06,
      "loss": 0.905,
      "step": 574
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4542220830917358,
      "learning_rate": 9.946183781410692e-06,
      "loss": 0.8906,
      "step": 575
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3753303289413452,
      "learning_rate": 9.945872818754831e-06,
      "loss": 0.9061,
      "step": 576
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3030872344970703,
      "learning_rate": 9.945560965167047e-06,
      "loss": 0.8351,
      "step": 577
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.432835578918457,
      "learning_rate": 9.945248220703513e-06,
      "loss": 0.9097,
      "step": 578
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.35128653049469,
      "learning_rate": 9.944934585420565e-06,
      "loss": 0.8557,
      "step": 579
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3301098346710205,
      "learning_rate": 9.944620059374702e-06,
      "loss": 0.8944,
      "step": 580
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4312572479248047,
      "learning_rate": 9.94430464262258e-06,
      "loss": 0.9252,
      "step": 581
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.435789942741394,
      "learning_rate": 9.94398833522102e-06,
      "loss": 0.8661,
      "step": 582
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4297116994857788,
      "learning_rate": 9.943671137226996e-06,
      "loss": 0.8929,
      "step": 583
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7729705572128296,
      "learning_rate": 9.943353048697649e-06,
      "loss": 1.1474,
      "step": 584
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4011746644973755,
      "learning_rate": 9.94303406969028e-06,
      "loss": 0.8927,
      "step": 585
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3489773273468018,
      "learning_rate": 9.942714200262347e-06,
      "loss": 0.9063,
      "step": 586
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3697896003723145,
      "learning_rate": 9.94239344047147e-06,
      "loss": 0.8888,
      "step": 587
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.352391004562378,
      "learning_rate": 9.94207179037543e-06,
      "loss": 0.9067,
      "step": 588
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4554396867752075,
      "learning_rate": 9.941749250032168e-06,
      "loss": 0.8836,
      "step": 589
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4109296798706055,
      "learning_rate": 9.941425819499787e-06,
      "loss": 0.9476,
      "step": 590
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3273769617080688,
      "learning_rate": 9.941101498836544e-06,
      "loss": 0.8337,
      "step": 591
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3268840312957764,
      "learning_rate": 9.940776288100865e-06,
      "loss": 0.9002,
      "step": 592
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4253506660461426,
      "learning_rate": 9.940450187351332e-06,
      "loss": 0.9041,
      "step": 593
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2252858877182007,
      "learning_rate": 9.940123196646686e-06,
      "loss": 1.1517,
      "step": 594
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4468281269073486,
      "learning_rate": 9.93979531604583e-06,
      "loss": 0.8497,
      "step": 595
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3542678356170654,
      "learning_rate": 9.939466545607827e-06,
      "loss": 0.9626,
      "step": 596
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2897785902023315,
      "learning_rate": 9.939136885391903e-06,
      "loss": 0.8686,
      "step": 597
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.29536771774292,
      "learning_rate": 9.938806335457441e-06,
      "loss": 0.8754,
      "step": 598
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7034443616867065,
      "learning_rate": 9.938474895863984e-06,
      "loss": 0.9208,
      "step": 599
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4577059745788574,
      "learning_rate": 9.938142566671234e-06,
      "loss": 0.8897,
      "step": 600
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.349804401397705,
      "learning_rate": 9.93780934793906e-06,
      "loss": 0.8668,
      "step": 601
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3785748481750488,
      "learning_rate": 9.937475239727486e-06,
      "loss": 0.8871,
      "step": 602
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3799060583114624,
      "learning_rate": 9.937140242096697e-06,
      "loss": 0.915,
      "step": 603
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.38051176071167,
      "learning_rate": 9.936804355107035e-06,
      "loss": 0.9192,
      "step": 604
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4106900691986084,
      "learning_rate": 9.93646757881901e-06,
      "loss": 0.9342,
      "step": 605
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2474243640899658,
      "learning_rate": 9.936129913293284e-06,
      "loss": 0.9108,
      "step": 606
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4974886178970337,
      "learning_rate": 9.935791358590684e-06,
      "loss": 0.9131,
      "step": 607
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4571301937103271,
      "learning_rate": 9.935451914772199e-06,
      "loss": 0.879,
      "step": 608
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3307217359542847,
      "learning_rate": 9.93511158189897e-06,
      "loss": 0.8839,
      "step": 609
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5689129829406738,
      "learning_rate": 9.93477036003231e-06,
      "loss": 1.1202,
      "step": 610
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3703906536102295,
      "learning_rate": 9.934428249233676e-06,
      "loss": 0.945,
      "step": 611
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4024999141693115,
      "learning_rate": 9.934085249564705e-06,
      "loss": 0.8533,
      "step": 612
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7160660028457642,
      "learning_rate": 9.933741361087177e-06,
      "loss": 1.1285,
      "step": 613
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7397814393043518,
      "learning_rate": 9.933396583863039e-06,
      "loss": 1.1257,
      "step": 614
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4060548543930054,
      "learning_rate": 9.933050917954404e-06,
      "loss": 0.9254,
      "step": 615
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.825110137462616,
      "learning_rate": 9.932704363423531e-06,
      "loss": 1.1205,
      "step": 616
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7169243097305298,
      "learning_rate": 9.932356920332852e-06,
      "loss": 1.1192,
      "step": 617
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4841409921646118,
      "learning_rate": 9.932008588744955e-06,
      "loss": 0.8525,
      "step": 618
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7938039898872375,
      "learning_rate": 9.931659368722583e-06,
      "loss": 1.1105,
      "step": 619
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3660120964050293,
      "learning_rate": 9.931309260328645e-06,
      "loss": 0.8914,
      "step": 620
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4137117862701416,
      "learning_rate": 9.930958263626212e-06,
      "loss": 0.8516,
      "step": 621
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8915621638298035,
      "learning_rate": 9.930606378678504e-06,
      "loss": 1.1284,
      "step": 622
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5427415370941162,
      "learning_rate": 9.930253605548914e-06,
      "loss": 0.8543,
      "step": 623
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3816478252410889,
      "learning_rate": 9.929899944300988e-06,
      "loss": 0.898,
      "step": 624
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3482506275177002,
      "learning_rate": 9.929545394998433e-06,
      "loss": 0.9059,
      "step": 625
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3177310228347778,
      "learning_rate": 9.929189957705116e-06,
      "loss": 0.9088,
      "step": 626
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9004461169242859,
      "learning_rate": 9.928833632485064e-06,
      "loss": 1.1586,
      "step": 627
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8066871166229248,
      "learning_rate": 9.928476419402466e-06,
      "loss": 1.0933,
      "step": 628
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5269476175308228,
      "learning_rate": 9.928118318521666e-06,
      "loss": 0.8814,
      "step": 629
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4184719324111938,
      "learning_rate": 9.927759329907174e-06,
      "loss": 0.8851,
      "step": 630
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3758310079574585,
      "learning_rate": 9.927399453623655e-06,
      "loss": 0.8775,
      "step": 631
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3582020998001099,
      "learning_rate": 9.927038689735934e-06,
      "loss": 0.901,
      "step": 632
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.119118571281433,
      "learning_rate": 9.926677038309005e-06,
      "loss": 1.1245,
      "step": 633
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8955708146095276,
      "learning_rate": 9.926314499408007e-06,
      "loss": 1.1549,
      "step": 634
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7487639784812927,
      "learning_rate": 9.92595107309825e-06,
      "loss": 1.1044,
      "step": 635
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8026054501533508,
      "learning_rate": 9.925586759445199e-06,
      "loss": 1.1,
      "step": 636
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6434872150421143,
      "learning_rate": 9.92522155851448e-06,
      "loss": 0.9527,
      "step": 637
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9163851737976074,
      "learning_rate": 9.924855470371879e-06,
      "loss": 1.1551,
      "step": 638
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4631417989730835,
      "learning_rate": 9.924488495083345e-06,
      "loss": 0.9215,
      "step": 639
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.383908987045288,
      "learning_rate": 9.92412063271498e-06,
      "loss": 0.8635,
      "step": 640
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4189164638519287,
      "learning_rate": 9.92375188333305e-06,
      "loss": 0.8784,
      "step": 641
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3039149045944214,
      "learning_rate": 9.92338224700398e-06,
      "loss": 0.854,
      "step": 642
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5432255268096924,
      "learning_rate": 9.923011723794358e-06,
      "loss": 1.1298,
      "step": 643
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.385361671447754,
      "learning_rate": 9.922640313770926e-06,
      "loss": 0.8863,
      "step": 644
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5157479047775269,
      "learning_rate": 9.922268017000589e-06,
      "loss": 0.9226,
      "step": 645
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7825354933738708,
      "learning_rate": 9.92189483355041e-06,
      "loss": 1.1384,
      "step": 646
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3088840246200562,
      "learning_rate": 9.921520763487614e-06,
      "loss": 0.8293,
      "step": 647
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3163717985153198,
      "learning_rate": 9.921145806879586e-06,
      "loss": 0.8848,
      "step": 648
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4027422666549683,
      "learning_rate": 9.920769963793868e-06,
      "loss": 0.906,
      "step": 649
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.403896689414978,
      "learning_rate": 9.920393234298163e-06,
      "loss": 0.815,
      "step": 650
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.358139991760254,
      "learning_rate": 9.920015618460333e-06,
      "loss": 1.1145,
      "step": 651
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.026545524597168,
      "learning_rate": 9.919637116348402e-06,
      "loss": 1.124,
      "step": 652
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3533045053482056,
      "learning_rate": 9.91925772803055e-06,
      "loss": 0.8757,
      "step": 653
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4353159666061401,
      "learning_rate": 9.918877453575121e-06,
      "loss": 0.8724,
      "step": 654
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9990230798721313,
      "learning_rate": 9.918496293050615e-06,
      "loss": 1.1334,
      "step": 655
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9189954996109009,
      "learning_rate": 9.918114246525693e-06,
      "loss": 1.1242,
      "step": 656
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3338228464126587,
      "learning_rate": 9.917731314069176e-06,
      "loss": 0.8779,
      "step": 657
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2921979427337646,
      "learning_rate": 9.917347495750043e-06,
      "loss": 0.9382,
      "step": 658
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7902273535728455,
      "learning_rate": 9.916962791637434e-06,
      "loss": 1.1115,
      "step": 659
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3280202150344849,
      "learning_rate": 9.91657720180065e-06,
      "loss": 0.8333,
      "step": 660
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3660304546356201,
      "learning_rate": 9.916190726309147e-06,
      "loss": 0.8414,
      "step": 661
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4421148300170898,
      "learning_rate": 9.915803365232546e-06,
      "loss": 0.9436,
      "step": 662
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0803672075271606,
      "learning_rate": 9.915415118640623e-06,
      "loss": 1.1125,
      "step": 663
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.30853271484375,
      "learning_rate": 9.915025986603315e-06,
      "loss": 0.8362,
      "step": 664
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4874204397201538,
      "learning_rate": 9.91463596919072e-06,
      "loss": 0.8297,
      "step": 665
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7311689853668213,
      "learning_rate": 9.914245066473095e-06,
      "loss": 1.1212,
      "step": 666
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.291114330291748,
      "learning_rate": 9.913853278520855e-06,
      "loss": 0.838,
      "step": 667
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4524532556533813,
      "learning_rate": 9.913460605404574e-06,
      "loss": 0.9066,
      "step": 668
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3321741819381714,
      "learning_rate": 9.913067047194988e-06,
      "loss": 0.8487,
      "step": 669
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3312424421310425,
      "learning_rate": 9.912672603962992e-06,
      "loss": 0.908,
      "step": 670
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3588112592697144,
      "learning_rate": 9.912277275779639e-06,
      "loss": 0.8997,
      "step": 671
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3755024671554565,
      "learning_rate": 9.91188106271614e-06,
      "loss": 0.8834,
      "step": 672
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4533867835998535,
      "learning_rate": 9.911483964843872e-06,
      "loss": 0.8146,
      "step": 673
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5167813301086426,
      "learning_rate": 9.91108598223436e-06,
      "loss": 0.9071,
      "step": 674
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3246586322784424,
      "learning_rate": 9.910687114959302e-06,
      "loss": 0.8808,
      "step": 675
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3824827671051025,
      "learning_rate": 9.910287363090544e-06,
      "loss": 0.8462,
      "step": 676
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3903329372406006,
      "learning_rate": 9.909886726700099e-06,
      "loss": 0.8921,
      "step": 677
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.424190640449524,
      "learning_rate": 9.909485205860134e-06,
      "loss": 0.9018,
      "step": 678
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.364304780960083,
      "learning_rate": 9.909082800642978e-06,
      "loss": 0.8813,
      "step": 679
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4047434329986572,
      "learning_rate": 9.90867951112112e-06,
      "loss": 0.8932,
      "step": 680
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.356815218925476,
      "learning_rate": 9.908275337367205e-06,
      "loss": 0.8838,
      "step": 681
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3360553979873657,
      "learning_rate": 9.907870279454042e-06,
      "loss": 0.9147,
      "step": 682
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.161247730255127,
      "learning_rate": 9.907464337454595e-06,
      "loss": 1.1156,
      "step": 683
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8696446418762207,
      "learning_rate": 9.907057511441988e-06,
      "loss": 1.123,
      "step": 684
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3432575464248657,
      "learning_rate": 9.906649801489507e-06,
      "loss": 0.8619,
      "step": 685
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9346197247505188,
      "learning_rate": 9.906241207670597e-06,
      "loss": 1.1328,
      "step": 686
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4605069160461426,
      "learning_rate": 9.905831730058857e-06,
      "loss": 0.9038,
      "step": 687
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2929143905639648,
      "learning_rate": 9.905421368728049e-06,
      "loss": 0.8776,
      "step": 688
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3503168821334839,
      "learning_rate": 9.905010123752096e-06,
      "loss": 0.8171,
      "step": 689
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4044568538665771,
      "learning_rate": 9.90459799520508e-06,
      "loss": 0.8684,
      "step": 690
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.390357494354248,
      "learning_rate": 9.904184983161236e-06,
      "loss": 0.8437,
      "step": 691
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1524250507354736,
      "learning_rate": 9.903771087694965e-06,
      "loss": 1.1162,
      "step": 692
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.392632246017456,
      "learning_rate": 9.903356308880824e-06,
      "loss": 0.8712,
      "step": 693
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.436040997505188,
      "learning_rate": 9.90294064679353e-06,
      "loss": 0.8986,
      "step": 694
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4587465524673462,
      "learning_rate": 9.902524101507959e-06,
      "loss": 0.8826,
      "step": 695
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7935216426849365,
      "learning_rate": 9.902106673099146e-06,
      "loss": 1.1202,
      "step": 696
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3876769542694092,
      "learning_rate": 9.901688361642283e-06,
      "loss": 0.8387,
      "step": 697
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4072206020355225,
      "learning_rate": 9.901269167212726e-06,
      "loss": 0.9238,
      "step": 698
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3447012901306152,
      "learning_rate": 9.900849089885987e-06,
      "loss": 0.9037,
      "step": 699
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9876176118850708,
      "learning_rate": 9.900428129737737e-06,
      "loss": 1.0796,
      "step": 700
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4688403606414795,
      "learning_rate": 9.900006286843804e-06,
      "loss": 0.8848,
      "step": 701
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.441949486732483,
      "learning_rate": 9.89958356128018e-06,
      "loss": 0.8642,
      "step": 702
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3972554206848145,
      "learning_rate": 9.899159953123012e-06,
      "loss": 0.8781,
      "step": 703
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2994534969329834,
      "learning_rate": 9.898735462448608e-06,
      "loss": 0.8566,
      "step": 704
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2793517112731934,
      "learning_rate": 9.898310089333433e-06,
      "loss": 0.864,
      "step": 705
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4884858131408691,
      "learning_rate": 9.897883833854113e-06,
      "loss": 0.8959,
      "step": 706
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.041508436203003,
      "learning_rate": 9.897456696087433e-06,
      "loss": 1.1085,
      "step": 707
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.771903932094574,
      "learning_rate": 9.897028676110337e-06,
      "loss": 1.1184,
      "step": 708
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3976366519927979,
      "learning_rate": 9.896599773999922e-06,
      "loss": 0.872,
      "step": 709
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3642915487289429,
      "learning_rate": 9.896169989833453e-06,
      "loss": 0.8936,
      "step": 710
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3528268337249756,
      "learning_rate": 9.895739323688348e-06,
      "loss": 0.8982,
      "step": 711
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3927408456802368,
      "learning_rate": 9.895307775642187e-06,
      "loss": 0.8315,
      "step": 712
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.348406195640564,
      "learning_rate": 9.894875345772707e-06,
      "loss": 0.8755,
      "step": 713
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6477687358856201,
      "learning_rate": 9.894442034157806e-06,
      "loss": 1.1243,
      "step": 714
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.347469449043274,
      "learning_rate": 9.894007840875537e-06,
      "loss": 0.8433,
      "step": 715
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0314366817474365,
      "learning_rate": 9.893572766004113e-06,
      "loss": 1.1147,
      "step": 716
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6056795120239258,
      "learning_rate": 9.893136809621911e-06,
      "loss": 0.8795,
      "step": 717
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6615067720413208,
      "learning_rate": 9.89269997180746e-06,
      "loss": 0.873,
      "step": 718
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4128592014312744,
      "learning_rate": 9.892262252639448e-06,
      "loss": 0.8665,
      "step": 719
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3480037450790405,
      "learning_rate": 9.891823652196728e-06,
      "loss": 0.9126,
      "step": 720
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3725343942642212,
      "learning_rate": 9.891384170558306e-06,
      "loss": 0.9012,
      "step": 721
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3169344663619995,
      "learning_rate": 9.890943807803353e-06,
      "loss": 0.8163,
      "step": 722
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.341530203819275,
      "learning_rate": 9.890502564011186e-06,
      "loss": 0.9002,
      "step": 723
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.463136911392212,
      "learning_rate": 9.890060439261298e-06,
      "loss": 0.9058,
      "step": 724
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8610140085220337,
      "learning_rate": 9.889617433633326e-06,
      "loss": 1.1026,
      "step": 725
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.399235725402832,
      "learning_rate": 9.889173547207072e-06,
      "loss": 1.1408,
      "step": 726
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4181815385818481,
      "learning_rate": 9.888728780062496e-06,
      "loss": 0.9119,
      "step": 727
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5453221797943115,
      "learning_rate": 9.888283132279719e-06,
      "loss": 0.9136,
      "step": 728
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4603970050811768,
      "learning_rate": 9.887836603939019e-06,
      "loss": 0.8259,
      "step": 729
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.416600227355957,
      "learning_rate": 9.887389195120829e-06,
      "loss": 0.9156,
      "step": 730
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2915551662445068,
      "learning_rate": 9.886940905905745e-06,
      "loss": 0.8921,
      "step": 731
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8833507299423218,
      "learning_rate": 9.886491736374521e-06,
      "loss": 1.1308,
      "step": 732
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6327617168426514,
      "learning_rate": 9.886041686608067e-06,
      "loss": 1.1151,
      "step": 733
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6696418523788452,
      "learning_rate": 9.885590756687456e-06,
      "loss": 0.8491,
      "step": 734
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8450615406036377,
      "learning_rate": 9.885138946693912e-06,
      "loss": 0.8626,
      "step": 735
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5947970151901245,
      "learning_rate": 9.884686256708829e-06,
      "loss": 0.8666,
      "step": 736
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4487346410751343,
      "learning_rate": 9.884232686813748e-06,
      "loss": 0.8733,
      "step": 737
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4481409788131714,
      "learning_rate": 9.883778237090375e-06,
      "loss": 0.9162,
      "step": 738
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4981938600540161,
      "learning_rate": 9.883322907620575e-06,
      "loss": 0.9642,
      "step": 739
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3989580869674683,
      "learning_rate": 9.882866698486365e-06,
      "loss": 0.9318,
      "step": 740
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2558420896530151,
      "learning_rate": 9.88240960976993e-06,
      "loss": 1.1391,
      "step": 741
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9789220094680786,
      "learning_rate": 9.881951641553604e-06,
      "loss": 1.0992,
      "step": 742
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5420582294464111,
      "learning_rate": 9.881492793919885e-06,
      "loss": 0.8686,
      "step": 743
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5478447675704956,
      "learning_rate": 9.881033066951432e-06,
      "loss": 0.8418,
      "step": 744
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4247372150421143,
      "learning_rate": 9.88057246073105e-06,
      "loss": 0.8561,
      "step": 745
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4121280908584595,
      "learning_rate": 9.880110975341721e-06,
      "loss": 0.9024,
      "step": 746
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4175970554351807,
      "learning_rate": 9.879648610866567e-06,
      "loss": 0.8514,
      "step": 747
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3311649560928345,
      "learning_rate": 9.879185367388883e-06,
      "loss": 0.8472,
      "step": 748
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.406143069267273,
      "learning_rate": 9.878721244992113e-06,
      "loss": 0.8954,
      "step": 749
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4856940507888794,
      "learning_rate": 9.878256243759863e-06,
      "loss": 0.8312,
      "step": 750
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.453726887702942,
      "learning_rate": 9.877790363775894e-06,
      "loss": 0.8568,
      "step": 751
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.558039426803589,
      "learning_rate": 9.877323605124131e-06,
      "loss": 1.1489,
      "step": 752
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4987009763717651,
      "learning_rate": 9.876855967888656e-06,
      "loss": 0.8764,
      "step": 753
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4213765859603882,
      "learning_rate": 9.876387452153702e-06,
      "loss": 0.8698,
      "step": 754
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4000818729400635,
      "learning_rate": 9.87591805800367e-06,
      "loss": 0.8809,
      "step": 755
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3543639183044434,
      "learning_rate": 9.875447785523116e-06,
      "loss": 0.8201,
      "step": 756
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.411035180091858,
      "learning_rate": 9.874976634796747e-06,
      "loss": 0.8785,
      "step": 757
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3726269006729126,
      "learning_rate": 9.874504605909441e-06,
      "loss": 0.804,
      "step": 758
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2741574048995972,
      "learning_rate": 9.874031698946225e-06,
      "loss": 0.7921,
      "step": 759
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9770213961601257,
      "learning_rate": 9.873557913992286e-06,
      "loss": 1.1122,
      "step": 760
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4456791877746582,
      "learning_rate": 9.87308325113297e-06,
      "loss": 0.863,
      "step": 761
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3880348205566406,
      "learning_rate": 9.872607710453784e-06,
      "loss": 0.8864,
      "step": 762
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3548598289489746,
      "learning_rate": 9.872131292040387e-06,
      "loss": 0.8575,
      "step": 763
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.900449275970459,
      "learning_rate": 9.871653995978599e-06,
      "loss": 1.1019,
      "step": 764
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3270487785339355,
      "learning_rate": 9.871175822354401e-06,
      "loss": 0.8801,
      "step": 765
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4705924987792969,
      "learning_rate": 9.87069677125393e-06,
      "loss": 0.9114,
      "step": 766
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7882176041603088,
      "learning_rate": 9.870216842763476e-06,
      "loss": 1.1272,
      "step": 767
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3054530620574951,
      "learning_rate": 9.869736036969495e-06,
      "loss": 0.8514,
      "step": 768
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7152938842773438,
      "learning_rate": 9.869254353958597e-06,
      "loss": 1.1385,
      "step": 769
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2475022077560425,
      "learning_rate": 9.86877179381755e-06,
      "loss": 0.8949,
      "step": 770
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3327149152755737,
      "learning_rate": 9.868288356633284e-06,
      "loss": 0.8744,
      "step": 771
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3714076280593872,
      "learning_rate": 9.867804042492878e-06,
      "loss": 0.8901,
      "step": 772
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.36610746383667,
      "learning_rate": 9.86731885148358e-06,
      "loss": 0.8517,
      "step": 773
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.279373049736023,
      "learning_rate": 9.866832783692788e-06,
      "loss": 0.85,
      "step": 774
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3081536293029785,
      "learning_rate": 9.86634583920806e-06,
      "loss": 0.8174,
      "step": 775
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2119196653366089,
      "learning_rate": 9.865858018117113e-06,
      "loss": 1.119,
      "step": 776
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4182987213134766,
      "learning_rate": 9.865369320507823e-06,
      "loss": 0.8913,
      "step": 777
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3006261587142944,
      "learning_rate": 9.86487974646822e-06,
      "loss": 0.8115,
      "step": 778
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7434101104736328,
      "learning_rate": 9.864389296086494e-06,
      "loss": 1.1358,
      "step": 779
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4370239973068237,
      "learning_rate": 9.863897969450994e-06,
      "loss": 0.876,
      "step": 780
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3577152490615845,
      "learning_rate": 9.863405766650228e-06,
      "loss": 0.8577,
      "step": 781
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8740965127944946,
      "learning_rate": 9.862912687772857e-06,
      "loss": 1.1094,
      "step": 782
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6780912280082703,
      "learning_rate": 9.862418732907702e-06,
      "loss": 1.1308,
      "step": 783
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3635082244873047,
      "learning_rate": 9.861923902143742e-06,
      "loss": 0.8652,
      "step": 784
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4645311832427979,
      "learning_rate": 9.861428195570116e-06,
      "loss": 0.9005,
      "step": 785
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3700798749923706,
      "learning_rate": 9.860931613276118e-06,
      "loss": 0.7688,
      "step": 786
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.025985598564148,
      "learning_rate": 9.860434155351202e-06,
      "loss": 1.1396,
      "step": 787
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2294763326644897,
      "learning_rate": 9.859935821884975e-06,
      "loss": 0.8481,
      "step": 788
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.343949317932129,
      "learning_rate": 9.859436612967208e-06,
      "loss": 0.8284,
      "step": 789
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4458633661270142,
      "learning_rate": 9.858936528687826e-06,
      "loss": 0.8382,
      "step": 790
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3954416513442993,
      "learning_rate": 9.85843556913691e-06,
      "loss": 0.8386,
      "step": 791
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.315195918083191,
      "learning_rate": 9.857933734404705e-06,
      "loss": 0.826,
      "step": 792
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.280226469039917,
      "learning_rate": 9.857431024581607e-06,
      "loss": 0.8172,
      "step": 793
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8147167563438416,
      "learning_rate": 9.856927439758174e-06,
      "loss": 1.1249,
      "step": 794
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4186029434204102,
      "learning_rate": 9.856422980025119e-06,
      "loss": 0.8557,
      "step": 795
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3842164278030396,
      "learning_rate": 9.855917645473314e-06,
      "loss": 0.8124,
      "step": 796
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4329079389572144,
      "learning_rate": 9.855411436193788e-06,
      "loss": 0.851,
      "step": 797
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4865001440048218,
      "learning_rate": 9.854904352277725e-06,
      "loss": 0.877,
      "step": 798
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5147610902786255,
      "learning_rate": 9.854396393816474e-06,
      "loss": 0.8886,
      "step": 799
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.339372992515564,
      "learning_rate": 9.853887560901537e-06,
      "loss": 0.8878,
      "step": 800
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.335883378982544,
      "learning_rate": 9.853377853624568e-06,
      "loss": 0.8594,
      "step": 801
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.255490779876709,
      "learning_rate": 9.852867272077387e-06,
      "loss": 0.897,
      "step": 802
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8733946084976196,
      "learning_rate": 9.85235581635197e-06,
      "loss": 1.1308,
      "step": 803
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.408121943473816,
      "learning_rate": 9.851843486540447e-06,
      "loss": 0.8821,
      "step": 804
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4198157787322998,
      "learning_rate": 9.851330282735106e-06,
      "loss": 0.8449,
      "step": 805
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3500679731369019,
      "learning_rate": 9.850816205028397e-06,
      "loss": 0.9043,
      "step": 806
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.411250114440918,
      "learning_rate": 9.85030125351292e-06,
      "loss": 0.9236,
      "step": 807
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.698094129562378,
      "learning_rate": 9.84978542828144e-06,
      "loss": 0.7983,
      "step": 808
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.019345998764038,
      "learning_rate": 9.849268729426874e-06,
      "loss": 1.1111,
      "step": 809
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.31630277633667,
      "learning_rate": 9.8487511570423e-06,
      "loss": 0.8249,
      "step": 810
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2665348052978516,
      "learning_rate": 9.84823271122095e-06,
      "loss": 0.9044,
      "step": 811
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7357872128486633,
      "learning_rate": 9.847713392056218e-06,
      "loss": 1.1043,
      "step": 812
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7327689528465271,
      "learning_rate": 9.847193199641648e-06,
      "loss": 1.1104,
      "step": 813
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4505388736724854,
      "learning_rate": 9.846672134070947e-06,
      "loss": 0.884,
      "step": 814
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3329482078552246,
      "learning_rate": 9.846150195437978e-06,
      "loss": 0.855,
      "step": 815
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3960882425308228,
      "learning_rate": 9.845627383836763e-06,
      "loss": 0.8515,
      "step": 816
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2992472648620605,
      "learning_rate": 9.845103699361477e-06,
      "loss": 0.8852,
      "step": 817
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2914538383483887,
      "learning_rate": 9.844579142106456e-06,
      "loss": 0.8332,
      "step": 818
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.33333158493042,
      "learning_rate": 9.844053712166192e-06,
      "loss": 0.8705,
      "step": 819
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.402806282043457,
      "learning_rate": 9.843527409635334e-06,
      "loss": 0.8934,
      "step": 820
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3111627101898193,
      "learning_rate": 9.843000234608686e-06,
      "loss": 0.9547,
      "step": 821
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.312011957168579,
      "learning_rate": 9.842472187181214e-06,
      "loss": 0.8833,
      "step": 822
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3761298656463623,
      "learning_rate": 9.84194326744804e-06,
      "loss": 0.9179,
      "step": 823
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2966809272766113,
      "learning_rate": 9.841413475504438e-06,
      "loss": 1.1179,
      "step": 824
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.416750431060791,
      "learning_rate": 9.840882811445844e-06,
      "loss": 0.881,
      "step": 825
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.431952714920044,
      "learning_rate": 9.84035127536785e-06,
      "loss": 0.9109,
      "step": 826
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2804741859436035,
      "learning_rate": 9.839818867366208e-06,
      "loss": 0.8546,
      "step": 827
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3298448324203491,
      "learning_rate": 9.83928558753682e-06,
      "loss": 0.8204,
      "step": 828
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.409306526184082,
      "learning_rate": 9.838751435975751e-06,
      "loss": 0.8673,
      "step": 829
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4077844619750977,
      "learning_rate": 9.83821641277922e-06,
      "loss": 0.8804,
      "step": 830
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3749138116836548,
      "learning_rate": 9.837680518043605e-06,
      "loss": 1.1247,
      "step": 831
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4515223503112793,
      "learning_rate": 9.837143751865442e-06,
      "loss": 0.8638,
      "step": 832
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3676856756210327,
      "learning_rate": 9.836606114341418e-06,
      "loss": 0.8992,
      "step": 833
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3010060787200928,
      "learning_rate": 9.836067605568384e-06,
      "loss": 0.8801,
      "step": 834
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7039147615432739,
      "learning_rate": 9.835528225643346e-06,
      "loss": 1.1272,
      "step": 835
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7361900806427002,
      "learning_rate": 9.834987974663461e-06,
      "loss": 1.0882,
      "step": 836
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4491636753082275,
      "learning_rate": 9.834446852726053e-06,
      "loss": 0.858,
      "step": 837
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3392196893692017,
      "learning_rate": 9.833904859928599e-06,
      "loss": 0.9093,
      "step": 838
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7822921276092529,
      "learning_rate": 9.833361996368725e-06,
      "loss": 1.1065,
      "step": 839
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3192781209945679,
      "learning_rate": 9.832818262144226e-06,
      "loss": 0.9046,
      "step": 840
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6652780771255493,
      "learning_rate": 9.832273657353046e-06,
      "loss": 1.1081,
      "step": 841
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6728348135948181,
      "learning_rate": 9.831728182093289e-06,
      "loss": 1.1225,
      "step": 842
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.401623249053955,
      "learning_rate": 9.831181836463216e-06,
      "loss": 0.8841,
      "step": 843
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3933136463165283,
      "learning_rate": 9.83063462056124e-06,
      "loss": 0.8827,
      "step": 844
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.353148102760315,
      "learning_rate": 9.83008653448594e-06,
      "loss": 0.8119,
      "step": 845
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2895183563232422,
      "learning_rate": 9.82953757833604e-06,
      "loss": 0.8477,
      "step": 846
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3144307136535645,
      "learning_rate": 9.828987752210432e-06,
      "loss": 0.8183,
      "step": 847
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3269503116607666,
      "learning_rate": 9.828437056208158e-06,
      "loss": 0.8913,
      "step": 848
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2633442878723145,
      "learning_rate": 9.827885490428419e-06,
      "loss": 0.8444,
      "step": 849
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3060306310653687,
      "learning_rate": 9.827333054970572e-06,
      "loss": 0.8733,
      "step": 850
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2747738361358643,
      "learning_rate": 9.826779749934128e-06,
      "loss": 0.8337,
      "step": 851
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.393446445465088,
      "learning_rate": 9.826225575418762e-06,
      "loss": 0.9438,
      "step": 852
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3894214630126953,
      "learning_rate": 9.825670531524295e-06,
      "loss": 0.8407,
      "step": 853
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3449562788009644,
      "learning_rate": 9.82511461835072e-06,
      "loss": 0.8307,
      "step": 854
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2698272466659546,
      "learning_rate": 9.824557835998168e-06,
      "loss": 0.8618,
      "step": 855
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3563979864120483,
      "learning_rate": 9.82400018456694e-06,
      "loss": 0.8187,
      "step": 856
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2525721788406372,
      "learning_rate": 9.823441664157489e-06,
      "loss": 0.8025,
      "step": 857
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1612951755523682,
      "learning_rate": 9.822882274870424e-06,
      "loss": 1.1092,
      "step": 858
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4006714820861816,
      "learning_rate": 9.822322016806514e-06,
      "loss": 0.8176,
      "step": 859
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3743163347244263,
      "learning_rate": 9.821760890066677e-06,
      "loss": 0.8519,
      "step": 860
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.295267939567566,
      "learning_rate": 9.821198894751998e-06,
      "loss": 0.8069,
      "step": 861
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8961616158485413,
      "learning_rate": 9.820636030963708e-06,
      "loss": 1.115,
      "step": 862
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3558179140090942,
      "learning_rate": 9.820072298803203e-06,
      "loss": 0.8836,
      "step": 863
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7206955552101135,
      "learning_rate": 9.819507698372028e-06,
      "loss": 1.0919,
      "step": 864
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2601431608200073,
      "learning_rate": 9.818942229771892e-06,
      "loss": 0.8775,
      "step": 865
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3246463537216187,
      "learning_rate": 9.818375893104654e-06,
      "loss": 0.8757,
      "step": 866
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9343482255935669,
      "learning_rate": 9.817808688472331e-06,
      "loss": 1.1121,
      "step": 867
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3151994943618774,
      "learning_rate": 9.817240615977102e-06,
      "loss": 0.8633,
      "step": 868
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7617940902709961,
      "learning_rate": 9.816671675721292e-06,
      "loss": 1.1073,
      "step": 869
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.498123288154602,
      "learning_rate": 9.816101867807393e-06,
      "loss": 0.8792,
      "step": 870
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7352760434150696,
      "learning_rate": 9.815531192338042e-06,
      "loss": 1.0826,
      "step": 871
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.752120852470398,
      "learning_rate": 9.814959649416044e-06,
      "loss": 1.1007,
      "step": 872
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3756773471832275,
      "learning_rate": 9.814387239144352e-06,
      "loss": 0.8738,
      "step": 873
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3443313837051392,
      "learning_rate": 9.813813961626078e-06,
      "loss": 0.8599,
      "step": 874
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6918252110481262,
      "learning_rate": 9.81323981696449e-06,
      "loss": 1.1111,
      "step": 875
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6911874413490295,
      "learning_rate": 9.812664805263015e-06,
      "loss": 1.0898,
      "step": 876
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6437075734138489,
      "learning_rate": 9.812088926625231e-06,
      "loss": 1.096,
      "step": 877
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6396549940109253,
      "learning_rate": 9.811512181154876e-06,
      "loss": 1.1178,
      "step": 878
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3106528520584106,
      "learning_rate": 9.810934568955842e-06,
      "loss": 0.8619,
      "step": 879
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.342394232749939,
      "learning_rate": 9.810356090132176e-06,
      "loss": 0.8821,
      "step": 880
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2810345888137817,
      "learning_rate": 9.809776744788086e-06,
      "loss": 0.8352,
      "step": 881
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3735363483428955,
      "learning_rate": 9.809196533027934e-06,
      "loss": 0.8706,
      "step": 882
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9134303331375122,
      "learning_rate": 9.808615454956235e-06,
      "loss": 1.1079,
      "step": 883
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3441674709320068,
      "learning_rate": 9.808033510677663e-06,
      "loss": 0.8939,
      "step": 884
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.361141562461853,
      "learning_rate": 9.807450700297048e-06,
      "loss": 0.8695,
      "step": 885
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4234682321548462,
      "learning_rate": 9.806867023919376e-06,
      "loss": 0.9061,
      "step": 886
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.815070629119873,
      "learning_rate": 9.806282481649785e-06,
      "loss": 1.1406,
      "step": 887
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3878216743469238,
      "learning_rate": 9.805697073593577e-06,
      "loss": 0.8955,
      "step": 888
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.383426547050476,
      "learning_rate": 9.805110799856201e-06,
      "loss": 0.9462,
      "step": 889
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3262579441070557,
      "learning_rate": 9.804523660543269e-06,
      "loss": 0.8629,
      "step": 890
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3783047199249268,
      "learning_rate": 9.803935655760546e-06,
      "loss": 0.8462,
      "step": 891
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8847288489341736,
      "learning_rate": 9.803346785613952e-06,
      "loss": 1.0877,
      "step": 892
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7303306460380554,
      "learning_rate": 9.802757050209563e-06,
      "loss": 1.095,
      "step": 893
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3569883108139038,
      "learning_rate": 9.802166449653616e-06,
      "loss": 0.8661,
      "step": 894
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3829617500305176,
      "learning_rate": 9.801574984052495e-06,
      "loss": 0.9466,
      "step": 895
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3177803754806519,
      "learning_rate": 9.800982653512747e-06,
      "loss": 0.8781,
      "step": 896
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3591573238372803,
      "learning_rate": 9.800389458141073e-06,
      "loss": 0.8575,
      "step": 897
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.370954155921936,
      "learning_rate": 9.799795398044328e-06,
      "loss": 0.8807,
      "step": 898
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.305951476097107,
      "learning_rate": 9.799200473329522e-06,
      "loss": 0.8882,
      "step": 899
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2871094942092896,
      "learning_rate": 9.798604684103824e-06,
      "loss": 0.9025,
      "step": 900
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3124808073043823,
      "learning_rate": 9.79800803047456e-06,
      "loss": 0.9031,
      "step": 901
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3436634540557861,
      "learning_rate": 9.797410512549207e-06,
      "loss": 0.8666,
      "step": 902
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3776873350143433,
      "learning_rate": 9.796812130435398e-06,
      "loss": 0.863,
      "step": 903
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2805715799331665,
      "learning_rate": 9.796212884240925e-06,
      "loss": 0.8935,
      "step": 904
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5707356929779053,
      "learning_rate": 9.795612774073737e-06,
      "loss": 0.8727,
      "step": 905
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2957172393798828,
      "learning_rate": 9.79501180004193e-06,
      "loss": 0.865,
      "step": 906
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2651047706604004,
      "learning_rate": 9.794409962253764e-06,
      "loss": 0.7931,
      "step": 907
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.25828218460083,
      "learning_rate": 9.793807260817653e-06,
      "loss": 0.8271,
      "step": 908
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.293829321861267,
      "learning_rate": 9.793203695842163e-06,
      "loss": 0.8754,
      "step": 909
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.327667474746704,
      "learning_rate": 9.792599267436022e-06,
      "loss": 0.849,
      "step": 910
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.439543604850769,
      "learning_rate": 9.791993975708107e-06,
      "loss": 0.841,
      "step": 911
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3306312561035156,
      "learning_rate": 9.791387820767454e-06,
      "loss": 0.8964,
      "step": 912
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4951086044311523,
      "learning_rate": 9.79078080272325e-06,
      "loss": 0.8238,
      "step": 913
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4334546327590942,
      "learning_rate": 9.790172921684844e-06,
      "loss": 0.8849,
      "step": 914
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2682017087936401,
      "learning_rate": 9.78956417776174e-06,
      "loss": 0.8295,
      "step": 915
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3272380828857422,
      "learning_rate": 9.78895457106359e-06,
      "loss": 0.8403,
      "step": 916
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2893825769424438,
      "learning_rate": 9.78834410170021e-06,
      "loss": 0.8708,
      "step": 917
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.686290979385376,
      "learning_rate": 9.787732769781566e-06,
      "loss": 1.1003,
      "step": 918
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3365172147750854,
      "learning_rate": 9.787120575417783e-06,
      "loss": 0.8203,
      "step": 919
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3187311887741089,
      "learning_rate": 9.786507518719138e-06,
      "loss": 0.8494,
      "step": 920
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2822719812393188,
      "learning_rate": 9.785893599796064e-06,
      "loss": 0.82,
      "step": 921
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.346379280090332,
      "learning_rate": 9.785278818759153e-06,
      "loss": 0.9149,
      "step": 922
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2561982870101929,
      "learning_rate": 9.784663175719145e-06,
      "loss": 0.8281,
      "step": 923
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3525910377502441,
      "learning_rate": 9.784046670786944e-06,
      "loss": 0.8314,
      "step": 924
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.12542724609375,
      "learning_rate": 9.783429304073604e-06,
      "loss": 1.1165,
      "step": 925
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7953155636787415,
      "learning_rate": 9.782811075690335e-06,
      "loss": 1.1153,
      "step": 926
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3762199878692627,
      "learning_rate": 9.782191985748502e-06,
      "loss": 0.8537,
      "step": 927
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5207123756408691,
      "learning_rate": 9.781572034359624e-06,
      "loss": 0.8582,
      "step": 928
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3326709270477295,
      "learning_rate": 9.780951221635379e-06,
      "loss": 0.8625,
      "step": 929
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.435529112815857,
      "learning_rate": 9.7803295476876e-06,
      "loss": 0.8982,
      "step": 930
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.345436453819275,
      "learning_rate": 9.77970701262827e-06,
      "loss": 0.8923,
      "step": 931
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1075377464294434,
      "learning_rate": 9.77908361656953e-06,
      "loss": 1.1145,
      "step": 932
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3763314485549927,
      "learning_rate": 9.778459359623678e-06,
      "loss": 0.8833,
      "step": 933
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4077486991882324,
      "learning_rate": 9.777834241903166e-06,
      "loss": 0.902,
      "step": 934
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3244688510894775,
      "learning_rate": 9.777208263520597e-06,
      "loss": 0.8858,
      "step": 935
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3369951248168945,
      "learning_rate": 9.776581424588736e-06,
      "loss": 0.8637,
      "step": 936
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.286072015762329,
      "learning_rate": 9.7759537252205e-06,
      "loss": 0.8689,
      "step": 937
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4603615999221802,
      "learning_rate": 9.775325165528957e-06,
      "loss": 0.8709,
      "step": 938
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2868571281433105,
      "learning_rate": 9.774695745627335e-06,
      "loss": 0.8211,
      "step": 939
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3525731563568115,
      "learning_rate": 9.774065465629017e-06,
      "loss": 0.9072,
      "step": 940
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9744030237197876,
      "learning_rate": 9.773434325647537e-06,
      "loss": 1.1336,
      "step": 941
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3087149858474731,
      "learning_rate": 9.77280232579659e-06,
      "loss": 0.8299,
      "step": 942
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3220282793045044,
      "learning_rate": 9.772169466190017e-06,
      "loss": 0.8654,
      "step": 943
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2246981859207153,
      "learning_rate": 9.771535746941822e-06,
      "loss": 0.8124,
      "step": 944
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3113207817077637,
      "learning_rate": 9.770901168166163e-06,
      "loss": 0.8701,
      "step": 945
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.251454472541809,
      "learning_rate": 9.770265729977347e-06,
      "loss": 0.8576,
      "step": 946
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2983964681625366,
      "learning_rate": 9.769629432489842e-06,
      "loss": 1.1132,
      "step": 947
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3208582401275635,
      "learning_rate": 9.768992275818267e-06,
      "loss": 0.8617,
      "step": 948
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.385486364364624,
      "learning_rate": 9.768354260077398e-06,
      "loss": 0.8364,
      "step": 949
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7638668417930603,
      "learning_rate": 9.767715385382163e-06,
      "loss": 1.1056,
      "step": 950
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7150740623474121,
      "learning_rate": 9.76707565184765e-06,
      "loss": 1.1042,
      "step": 951
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3530306816101074,
      "learning_rate": 9.766435059589096e-06,
      "loss": 0.8621,
      "step": 952
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.472536325454712,
      "learning_rate": 9.765793608721894e-06,
      "loss": 0.8332,
      "step": 953
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3460808992385864,
      "learning_rate": 9.765151299361596e-06,
      "loss": 0.8446,
      "step": 954
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1778982877731323,
      "learning_rate": 9.764508131623902e-06,
      "loss": 1.0922,
      "step": 955
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2833527326583862,
      "learning_rate": 9.763864105624673e-06,
      "loss": 0.8031,
      "step": 956
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.294780731201172,
      "learning_rate": 9.763219221479919e-06,
      "loss": 0.8952,
      "step": 957
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3574296236038208,
      "learning_rate": 9.762573479305807e-06,
      "loss": 0.9483,
      "step": 958
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8874380588531494,
      "learning_rate": 9.761926879218661e-06,
      "loss": 1.118,
      "step": 959
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7991656064987183,
      "learning_rate": 9.761279421334957e-06,
      "loss": 1.1243,
      "step": 960
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4197524785995483,
      "learning_rate": 9.760631105771322e-06,
      "loss": 0.806,
      "step": 961
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3532142639160156,
      "learning_rate": 9.759981932644547e-06,
      "loss": 0.8281,
      "step": 962
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3787599802017212,
      "learning_rate": 9.759331902071569e-06,
      "loss": 0.8736,
      "step": 963
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.221160650253296,
      "learning_rate": 9.75868101416948e-06,
      "loss": 1.1169,
      "step": 964
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1381020545959473,
      "learning_rate": 9.758029269055531e-06,
      "loss": 1.1003,
      "step": 965
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3307660818099976,
      "learning_rate": 9.757376666847125e-06,
      "loss": 0.8299,
      "step": 966
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7481122612953186,
      "learning_rate": 9.756723207661818e-06,
      "loss": 1.0986,
      "step": 967
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9532158970832825,
      "learning_rate": 9.756068891617323e-06,
      "loss": 1.1211,
      "step": 968
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7757281064987183,
      "learning_rate": 9.755413718831507e-06,
      "loss": 1.0997,
      "step": 969
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3202097415924072,
      "learning_rate": 9.75475768942239e-06,
      "loss": 0.8158,
      "step": 970
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3714736700057983,
      "learning_rate": 9.754100803508146e-06,
      "loss": 0.8965,
      "step": 971
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3407784700393677,
      "learning_rate": 9.753443061207104e-06,
      "loss": 0.8489,
      "step": 972
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.328841209411621,
      "learning_rate": 9.752784462637748e-06,
      "loss": 0.8838,
      "step": 973
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2986934185028076,
      "learning_rate": 9.752125007918714e-06,
      "loss": 0.8528,
      "step": 974
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3860944509506226,
      "learning_rate": 9.751464697168798e-06,
      "loss": 0.9091,
      "step": 975
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4877029657363892,
      "learning_rate": 9.75080353050694e-06,
      "loss": 1.1058,
      "step": 976
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.357836127281189,
      "learning_rate": 9.750141508052246e-06,
      "loss": 0.8547,
      "step": 977
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4341226816177368,
      "learning_rate": 9.749478629923968e-06,
      "loss": 0.8162,
      "step": 978
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.56149423122406,
      "learning_rate": 9.748814896241512e-06,
      "loss": 0.8767,
      "step": 979
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3813328742980957,
      "learning_rate": 9.748150307124444e-06,
      "loss": 0.8972,
      "step": 980
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3463356494903564,
      "learning_rate": 9.74748486269248e-06,
      "loss": 0.8295,
      "step": 981
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2649152278900146,
      "learning_rate": 9.746818563065491e-06,
      "loss": 0.8369,
      "step": 982
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8242390751838684,
      "learning_rate": 9.7461514083635e-06,
      "loss": 1.1194,
      "step": 983
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4109712839126587,
      "learning_rate": 9.74548339870669e-06,
      "loss": 0.8516,
      "step": 984
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5612703561782837,
      "learning_rate": 9.744814534215388e-06,
      "loss": 0.8311,
      "step": 985
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3504209518432617,
      "learning_rate": 9.744144815010085e-06,
      "loss": 0.8319,
      "step": 986
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3723715543746948,
      "learning_rate": 9.743474241211422e-06,
      "loss": 0.8583,
      "step": 987
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3437469005584717,
      "learning_rate": 9.742802812940191e-06,
      "loss": 0.9125,
      "step": 988
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3320832252502441,
      "learning_rate": 9.742130530317342e-06,
      "loss": 0.8721,
      "step": 989
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3701395988464355,
      "learning_rate": 9.741457393463977e-06,
      "loss": 0.8432,
      "step": 990
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4016684293746948,
      "learning_rate": 9.740783402501356e-06,
      "loss": 0.8948,
      "step": 991
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.074991226196289,
      "learning_rate": 9.740108557550885e-06,
      "loss": 1.1377,
      "step": 992
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3718394041061401,
      "learning_rate": 9.739432858734131e-06,
      "loss": 0.9002,
      "step": 993
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3026753664016724,
      "learning_rate": 9.73875630617281e-06,
      "loss": 0.8456,
      "step": 994
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7883448600769043,
      "learning_rate": 9.738078899988794e-06,
      "loss": 1.1135,
      "step": 995
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7297335863113403,
      "learning_rate": 9.737400640304108e-06,
      "loss": 1.0818,
      "step": 996
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3470708131790161,
      "learning_rate": 9.736721527240935e-06,
      "loss": 0.8624,
      "step": 997
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.34559965133667,
      "learning_rate": 9.736041560921604e-06,
      "loss": 0.8788,
      "step": 998
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3624540567398071,
      "learning_rate": 9.735360741468604e-06,
      "loss": 0.8763,
      "step": 999
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3663926124572754,
      "learning_rate": 9.734679069004575e-06,
      "loss": 0.8222,
      "step": 1000
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9909383654594421,
      "learning_rate": 9.73399654365231e-06,
      "loss": 1.098,
      "step": 1001
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.350122332572937,
      "learning_rate": 9.733313165534757e-06,
      "loss": 0.8367,
      "step": 1002
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.335997462272644,
      "learning_rate": 9.732628934775017e-06,
      "loss": 0.8707,
      "step": 1003
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7652522921562195,
      "learning_rate": 9.731943851496347e-06,
      "loss": 1.1083,
      "step": 1004
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.695787250995636,
      "learning_rate": 9.731257915822152e-06,
      "loss": 1.0908,
      "step": 1005
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4598965644836426,
      "learning_rate": 9.730571127875997e-06,
      "loss": 0.8241,
      "step": 1006
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3486721515655518,
      "learning_rate": 9.729883487781595e-06,
      "loss": 0.8544,
      "step": 1007
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3876491785049438,
      "learning_rate": 9.729194995662818e-06,
      "loss": 0.8808,
      "step": 1008
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3452963829040527,
      "learning_rate": 9.728505651643687e-06,
      "loss": 0.9299,
      "step": 1009
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2493635416030884,
      "learning_rate": 9.727815455848376e-06,
      "loss": 0.7887,
      "step": 1010
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4117122888565063,
      "learning_rate": 9.727124408401221e-06,
      "loss": 0.9358,
      "step": 1011
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3185397386550903,
      "learning_rate": 9.726432509426697e-06,
      "loss": 1.1235,
      "step": 1012
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4095556735992432,
      "learning_rate": 9.725739759049444e-06,
      "loss": 0.8513,
      "step": 1013
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3448107242584229,
      "learning_rate": 9.72504615739425e-06,
      "loss": 0.8162,
      "step": 1014
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2485016584396362,
      "learning_rate": 9.72435170458606e-06,
      "loss": 0.8596,
      "step": 1015
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3754886388778687,
      "learning_rate": 9.72365640074997e-06,
      "loss": 0.84,
      "step": 1016
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4355131387710571,
      "learning_rate": 9.722960246011226e-06,
      "loss": 0.8889,
      "step": 1017
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.460070013999939,
      "learning_rate": 9.722263240495236e-06,
      "loss": 0.8566,
      "step": 1018
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.326243281364441,
      "learning_rate": 9.721565384327554e-06,
      "loss": 0.8569,
      "step": 1019
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2007278203964233,
      "learning_rate": 9.720866677633889e-06,
      "loss": 1.1122,
      "step": 1020
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4064345359802246,
      "learning_rate": 9.720167120540102e-06,
      "loss": 0.8896,
      "step": 1021
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4047629833221436,
      "learning_rate": 9.719466713172212e-06,
      "loss": 0.9016,
      "step": 1022
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4538339376449585,
      "learning_rate": 9.718765455656385e-06,
      "loss": 0.8187,
      "step": 1023
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3435630798339844,
      "learning_rate": 9.718063348118945e-06,
      "loss": 0.7853,
      "step": 1024
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.326456069946289,
      "learning_rate": 9.717360390686366e-06,
      "loss": 0.9007,
      "step": 1025
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2912815809249878,
      "learning_rate": 9.716656583485276e-06,
      "loss": 1.0941,
      "step": 1026
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0064339637756348,
      "learning_rate": 9.715951926642458e-06,
      "loss": 1.1111,
      "step": 1027
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4426138401031494,
      "learning_rate": 9.715246420284845e-06,
      "loss": 0.7825,
      "step": 1028
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4792563915252686,
      "learning_rate": 9.714540064539524e-06,
      "loss": 0.9067,
      "step": 1029
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3334537744522095,
      "learning_rate": 9.713832859533737e-06,
      "loss": 0.8188,
      "step": 1030
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3098528385162354,
      "learning_rate": 9.713124805394877e-06,
      "loss": 0.7565,
      "step": 1031
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3746142387390137,
      "learning_rate": 9.71241590225049e-06,
      "loss": 0.8259,
      "step": 1032
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3921517133712769,
      "learning_rate": 9.711706150228273e-06,
      "loss": 0.8735,
      "step": 1033
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6824891567230225,
      "learning_rate": 9.710995549456082e-06,
      "loss": 1.129,
      "step": 1034
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4065865278244019,
      "learning_rate": 9.71028410006192e-06,
      "loss": 0.842,
      "step": 1035
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4362592697143555,
      "learning_rate": 9.709571802173947e-06,
      "loss": 0.8581,
      "step": 1036
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.530718207359314,
      "learning_rate": 9.70885865592047e-06,
      "loss": 0.8656,
      "step": 1037
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4726663827896118,
      "learning_rate": 9.708144661429956e-06,
      "loss": 1.1119,
      "step": 1038
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8852745890617371,
      "learning_rate": 9.70742981883102e-06,
      "loss": 1.0826,
      "step": 1039
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5418418645858765,
      "learning_rate": 9.706714128252433e-06,
      "loss": 0.8797,
      "step": 1040
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.528317928314209,
      "learning_rate": 9.705997589823112e-06,
      "loss": 0.9164,
      "step": 1041
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4521011114120483,
      "learning_rate": 9.705280203672139e-06,
      "loss": 0.8568,
      "step": 1042
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4913856983184814,
      "learning_rate": 9.704561969928736e-06,
      "loss": 0.8525,
      "step": 1043
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3487982749938965,
      "learning_rate": 9.703842888722284e-06,
      "loss": 0.9038,
      "step": 1044
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4362890720367432,
      "learning_rate": 9.703122960182316e-06,
      "loss": 0.872,
      "step": 1045
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4806180000305176,
      "learning_rate": 9.70240218443852e-06,
      "loss": 0.8786,
      "step": 1046
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4634240865707397,
      "learning_rate": 9.70168056162073e-06,
      "loss": 0.9313,
      "step": 1047
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.384048342704773,
      "learning_rate": 9.700958091858937e-06,
      "loss": 0.8606,
      "step": 1048
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.313143253326416,
      "learning_rate": 9.700234775283287e-06,
      "loss": 0.8745,
      "step": 1049
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.344839096069336,
      "learning_rate": 9.699510612024074e-06,
      "loss": 0.8276,
      "step": 1050
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3478628396987915,
      "learning_rate": 9.698785602211745e-06,
      "loss": 0.8506,
      "step": 1051
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.5423943996429443,
      "learning_rate": 9.698059745976905e-06,
      "loss": 1.1476,
      "step": 1052
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4869111776351929,
      "learning_rate": 9.697333043450301e-06,
      "loss": 0.8898,
      "step": 1053
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.299920916557312,
      "learning_rate": 9.696605494762842e-06,
      "loss": 0.8051,
      "step": 1054
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.406296968460083,
      "learning_rate": 9.695877100045586e-06,
      "loss": 0.8294,
      "step": 1055
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2606079578399658,
      "learning_rate": 9.695147859429744e-06,
      "loss": 0.8368,
      "step": 1056
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3247898817062378,
      "learning_rate": 9.694417773046679e-06,
      "loss": 0.8488,
      "step": 1057
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4160220623016357,
      "learning_rate": 9.693686841027902e-06,
      "loss": 0.9238,
      "step": 1058
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.313095211982727,
      "learning_rate": 9.692955063505084e-06,
      "loss": 0.8289,
      "step": 1059
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2822823524475098,
      "learning_rate": 9.692222440610046e-06,
      "loss": 0.8561,
      "step": 1060
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.118777871131897,
      "learning_rate": 9.691488972474758e-06,
      "loss": 1.1023,
      "step": 1061
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3973467350006104,
      "learning_rate": 9.690754659231346e-06,
      "loss": 0.8524,
      "step": 1062
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3202472925186157,
      "learning_rate": 9.690019501012085e-06,
      "loss": 0.8771,
      "step": 1063
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3528854846954346,
      "learning_rate": 9.689283497949401e-06,
      "loss": 0.8852,
      "step": 1064
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3584227561950684,
      "learning_rate": 9.688546650175883e-06,
      "loss": 0.8466,
      "step": 1065
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.313252329826355,
      "learning_rate": 9.687808957824256e-06,
      "loss": 0.8673,
      "step": 1066
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3541864156723022,
      "learning_rate": 9.68707042102741e-06,
      "loss": 0.8585,
      "step": 1067
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3997725248336792,
      "learning_rate": 9.686331039918381e-06,
      "loss": 1.0875,
      "step": 1068
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3619825839996338,
      "learning_rate": 9.685590814630356e-06,
      "loss": 0.888,
      "step": 1069
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4432567358016968,
      "learning_rate": 9.684849745296682e-06,
      "loss": 0.8929,
      "step": 1070
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.293631911277771,
      "learning_rate": 9.684107832050849e-06,
      "loss": 0.8487,
      "step": 1071
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7651552557945251,
      "learning_rate": 9.683365075026503e-06,
      "loss": 1.117,
      "step": 1072
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2780852317810059,
      "learning_rate": 9.682621474357441e-06,
      "loss": 0.8554,
      "step": 1073
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3576600551605225,
      "learning_rate": 9.681877030177614e-06,
      "loss": 0.9175,
      "step": 1074
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.345676064491272,
      "learning_rate": 9.681131742621123e-06,
      "loss": 0.8026,
      "step": 1075
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3051990270614624,
      "learning_rate": 9.68038561182222e-06,
      "loss": 0.8518,
      "step": 1076
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8056788444519043,
      "learning_rate": 9.679638637915314e-06,
      "loss": 1.0804,
      "step": 1077
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2436094284057617,
      "learning_rate": 9.678890821034959e-06,
      "loss": 0.8124,
      "step": 1078
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6761125922203064,
      "learning_rate": 9.678142161315864e-06,
      "loss": 1.0952,
      "step": 1079
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3807759284973145,
      "learning_rate": 9.677392658892892e-06,
      "loss": 0.874,
      "step": 1080
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4367613792419434,
      "learning_rate": 9.676642313901055e-06,
      "loss": 0.889,
      "step": 1081
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2624220848083496,
      "learning_rate": 9.675891126475514e-06,
      "loss": 0.8238,
      "step": 1082
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3367654085159302,
      "learning_rate": 9.675139096751591e-06,
      "loss": 0.8013,
      "step": 1083
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4017688035964966,
      "learning_rate": 9.674386224864752e-06,
      "loss": 0.827,
      "step": 1084
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0603463649749756,
      "learning_rate": 9.673632510950615e-06,
      "loss": 1.0994,
      "step": 1085
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4333502054214478,
      "learning_rate": 9.672877955144952e-06,
      "loss": 0.8633,
      "step": 1086
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3327155113220215,
      "learning_rate": 9.672122557583685e-06,
      "loss": 0.7961,
      "step": 1087
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4186484813690186,
      "learning_rate": 9.671366318402892e-06,
      "loss": 0.8842,
      "step": 1088
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.865746796131134,
      "learning_rate": 9.670609237738795e-06,
      "loss": 1.1172,
      "step": 1089
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.417777419090271,
      "learning_rate": 9.669851315727777e-06,
      "loss": 0.8728,
      "step": 1090
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3781672716140747,
      "learning_rate": 9.669092552506361e-06,
      "loss": 0.8476,
      "step": 1091
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2996689081192017,
      "learning_rate": 9.668332948211234e-06,
      "loss": 0.868,
      "step": 1092
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3756322860717773,
      "learning_rate": 9.667572502979223e-06,
      "loss": 0.8681,
      "step": 1093
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4981510639190674,
      "learning_rate": 9.666811216947315e-06,
      "loss": 0.8924,
      "step": 1094
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.842001736164093,
      "learning_rate": 9.666049090252647e-06,
      "loss": 1.1029,
      "step": 1095
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8244331479072571,
      "learning_rate": 9.665286123032502e-06,
      "loss": 1.0972,
      "step": 1096
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3263136148452759,
      "learning_rate": 9.664522315424321e-06,
      "loss": 0.8741,
      "step": 1097
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.409070611000061,
      "learning_rate": 9.663757667565691e-06,
      "loss": 0.818,
      "step": 1098
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8011375069618225,
      "learning_rate": 9.662992179594355e-06,
      "loss": 1.1038,
      "step": 1099
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3058700561523438,
      "learning_rate": 9.662225851648205e-06,
      "loss": 0.8168,
      "step": 1100
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3707339763641357,
      "learning_rate": 9.661458683865283e-06,
      "loss": 0.8458,
      "step": 1101
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8453502058982849,
      "learning_rate": 9.660690676383786e-06,
      "loss": 1.1115,
      "step": 1102
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6646011471748352,
      "learning_rate": 9.65992182934206e-06,
      "loss": 1.1021,
      "step": 1103
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.646984338760376,
      "learning_rate": 9.659152142878601e-06,
      "loss": 1.0759,
      "step": 1104
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3984434604644775,
      "learning_rate": 9.658381617132058e-06,
      "loss": 0.8643,
      "step": 1105
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2746307849884033,
      "learning_rate": 9.657610252241232e-06,
      "loss": 0.8335,
      "step": 1106
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2492672204971313,
      "learning_rate": 9.656838048345072e-06,
      "loss": 0.8525,
      "step": 1107
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3479499816894531,
      "learning_rate": 9.65606500558268e-06,
      "loss": 0.7906,
      "step": 1108
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.047597050666809,
      "learning_rate": 9.655291124093311e-06,
      "loss": 1.099,
      "step": 1109
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.305907130241394,
      "learning_rate": 9.65451640401637e-06,
      "loss": 0.8907,
      "step": 1110
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4457427263259888,
      "learning_rate": 9.653740845491408e-06,
      "loss": 0.8672,
      "step": 1111
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7479158043861389,
      "learning_rate": 9.652964448658135e-06,
      "loss": 1.1037,
      "step": 1112
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3147462606430054,
      "learning_rate": 9.652187213656409e-06,
      "loss": 0.8518,
      "step": 1113
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7278882265090942,
      "learning_rate": 9.651409140626236e-06,
      "loss": 1.0773,
      "step": 1114
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7088944315910339,
      "learning_rate": 9.650630229707775e-06,
      "loss": 1.1042,
      "step": 1115
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.354026198387146,
      "learning_rate": 9.649850481041338e-06,
      "loss": 0.8974,
      "step": 1116
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2969934940338135,
      "learning_rate": 9.649069894767386e-06,
      "loss": 0.8185,
      "step": 1117
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3365497589111328,
      "learning_rate": 9.64828847102653e-06,
      "loss": 0.8335,
      "step": 1118
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2403842210769653,
      "learning_rate": 9.647506209959534e-06,
      "loss": 0.7819,
      "step": 1119
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3860759735107422,
      "learning_rate": 9.64672311170731e-06,
      "loss": 0.9002,
      "step": 1120
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3077459335327148,
      "learning_rate": 9.645939176410924e-06,
      "loss": 0.8749,
      "step": 1121
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3032070398330688,
      "learning_rate": 9.64515440421159e-06,
      "loss": 0.9395,
      "step": 1122
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.323256492614746,
      "learning_rate": 9.644368795250675e-06,
      "loss": 0.9024,
      "step": 1123
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3082759380340576,
      "learning_rate": 9.643582349669698e-06,
      "loss": 0.8655,
      "step": 1124
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0045948028564453,
      "learning_rate": 9.642795067610324e-06,
      "loss": 1.0989,
      "step": 1125
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8499698042869568,
      "learning_rate": 9.64200694921437e-06,
      "loss": 1.0927,
      "step": 1126
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3362871408462524,
      "learning_rate": 9.641217994623808e-06,
      "loss": 0.86,
      "step": 1127
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4692864418029785,
      "learning_rate": 9.640428203980755e-06,
      "loss": 0.867,
      "step": 1128
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.376297950744629,
      "learning_rate": 9.639637577427483e-06,
      "loss": 0.864,
      "step": 1129
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3464012145996094,
      "learning_rate": 9.63884611510641e-06,
      "loss": 1.1023,
      "step": 1130
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3691761493682861,
      "learning_rate": 9.63805381716011e-06,
      "loss": 0.8848,
      "step": 1131
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3419827222824097,
      "learning_rate": 9.637260683731303e-06,
      "loss": 0.8821,
      "step": 1132
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3704670667648315,
      "learning_rate": 9.63646671496286e-06,
      "loss": 0.8285,
      "step": 1133
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.250986099243164,
      "learning_rate": 9.635671910997808e-06,
      "loss": 0.7853,
      "step": 1134
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3011780977249146,
      "learning_rate": 9.634876271979317e-06,
      "loss": 0.7876,
      "step": 1135
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3782578706741333,
      "learning_rate": 9.63407979805071e-06,
      "loss": 0.8667,
      "step": 1136
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3367903232574463,
      "learning_rate": 9.633282489355463e-06,
      "loss": 0.8311,
      "step": 1137
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2941375970840454,
      "learning_rate": 9.6324843460372e-06,
      "loss": 0.7956,
      "step": 1138
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.289805293083191,
      "learning_rate": 9.631685368239695e-06,
      "loss": 0.8752,
      "step": 1139
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3420219421386719,
      "learning_rate": 9.630885556106873e-06,
      "loss": 0.8592,
      "step": 1140
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4781378507614136,
      "learning_rate": 9.630084909782809e-06,
      "loss": 0.8658,
      "step": 1141
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9110388159751892,
      "learning_rate": 9.629283429411731e-06,
      "loss": 1.1051,
      "step": 1142
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.342026710510254,
      "learning_rate": 9.628481115138011e-06,
      "loss": 0.8174,
      "step": 1143
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6939809322357178,
      "learning_rate": 9.627677967106178e-06,
      "loss": 1.0878,
      "step": 1144
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3311281204223633,
      "learning_rate": 9.626873985460908e-06,
      "loss": 0.9221,
      "step": 1145
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3180921077728271,
      "learning_rate": 9.626069170347022e-06,
      "loss": 0.8066,
      "step": 1146
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3206984996795654,
      "learning_rate": 9.625263521909508e-06,
      "loss": 0.8649,
      "step": 1147
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4238563776016235,
      "learning_rate": 9.62445704029348e-06,
      "loss": 0.8241,
      "step": 1148
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.247312307357788,
      "learning_rate": 9.623649725644225e-06,
      "loss": 0.8508,
      "step": 1149
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3937351703643799,
      "learning_rate": 9.622841578107163e-06,
      "loss": 0.7633,
      "step": 1150
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.296372890472412,
      "learning_rate": 9.622032597827873e-06,
      "loss": 0.8621,
      "step": 1151
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1043734550476074,
      "learning_rate": 9.621222784952082e-06,
      "loss": 1.1097,
      "step": 1152
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3732316493988037,
      "learning_rate": 9.620412139625666e-06,
      "loss": 0.873,
      "step": 1153
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6776686906814575,
      "learning_rate": 9.619600661994655e-06,
      "loss": 1.089,
      "step": 1154
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4348604679107666,
      "learning_rate": 9.618788352205221e-06,
      "loss": 0.8772,
      "step": 1155
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6733208894729614,
      "learning_rate": 9.617975210403694e-06,
      "loss": 0.8602,
      "step": 1156
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.510589599609375,
      "learning_rate": 9.617161236736549e-06,
      "loss": 0.8133,
      "step": 1157
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4280155897140503,
      "learning_rate": 9.616346431350413e-06,
      "loss": 0.8202,
      "step": 1158
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3200368881225586,
      "learning_rate": 9.61553079439206e-06,
      "loss": 0.8687,
      "step": 1159
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3270020484924316,
      "learning_rate": 9.614714326008419e-06,
      "loss": 0.8705,
      "step": 1160
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3776514530181885,
      "learning_rate": 9.613897026346563e-06,
      "loss": 0.8375,
      "step": 1161
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2709414958953857,
      "learning_rate": 9.61307889555372e-06,
      "loss": 0.796,
      "step": 1162
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3742904663085938,
      "learning_rate": 9.612259933777265e-06,
      "loss": 0.842,
      "step": 1163
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2846381664276123,
      "learning_rate": 9.61144014116472e-06,
      "loss": 0.8607,
      "step": 1164
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2686922550201416,
      "learning_rate": 9.610619517863763e-06,
      "loss": 0.7627,
      "step": 1165
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3759129047393799,
      "learning_rate": 9.609798064022215e-06,
      "loss": 0.9297,
      "step": 1166
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3088573217391968,
      "learning_rate": 9.608975779788054e-06,
      "loss": 0.8742,
      "step": 1167
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3025494813919067,
      "learning_rate": 9.608152665309399e-06,
      "loss": 0.8096,
      "step": 1168
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4734172821044922,
      "learning_rate": 9.607328720734523e-06,
      "loss": 1.093,
      "step": 1169
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3764162063598633,
      "learning_rate": 9.606503946211852e-06,
      "loss": 0.8319,
      "step": 1170
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4670628309249878,
      "learning_rate": 9.605678341889955e-06,
      "loss": 0.8777,
      "step": 1171
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3153493404388428,
      "learning_rate": 9.604851907917555e-06,
      "loss": 0.8406,
      "step": 1172
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8468589782714844,
      "learning_rate": 9.60402464444352e-06,
      "loss": 1.1015,
      "step": 1173
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8504101634025574,
      "learning_rate": 9.603196551616877e-06,
      "loss": 1.0819,
      "step": 1174
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7027091383934021,
      "learning_rate": 9.602367629586789e-06,
      "loss": 1.0928,
      "step": 1175
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.460753083229065,
      "learning_rate": 9.601537878502577e-06,
      "loss": 0.8225,
      "step": 1176
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3811248540878296,
      "learning_rate": 9.60070729851371e-06,
      "loss": 0.9058,
      "step": 1177
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4302347898483276,
      "learning_rate": 9.599875889769806e-06,
      "loss": 0.8663,
      "step": 1178
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2321696281433105,
      "learning_rate": 9.59904365242063e-06,
      "loss": 1.0936,
      "step": 1179
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4175589084625244,
      "learning_rate": 9.598210586616101e-06,
      "loss": 0.8284,
      "step": 1180
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4048457145690918,
      "learning_rate": 9.597376692506282e-06,
      "loss": 0.799,
      "step": 1181
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8415477871894836,
      "learning_rate": 9.596541970241392e-06,
      "loss": 1.1187,
      "step": 1182
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3542717695236206,
      "learning_rate": 9.595706419971788e-06,
      "loss": 0.8879,
      "step": 1183
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3439141511917114,
      "learning_rate": 9.594870041847988e-06,
      "loss": 0.8372,
      "step": 1184
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7675379514694214,
      "learning_rate": 9.594032836020654e-06,
      "loss": 1.1175,
      "step": 1185
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.304986596107483,
      "learning_rate": 9.593194802640596e-06,
      "loss": 0.7958,
      "step": 1186
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.329538345336914,
      "learning_rate": 9.592355941858775e-06,
      "loss": 0.8908,
      "step": 1187
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3093053102493286,
      "learning_rate": 9.591516253826301e-06,
      "loss": 0.838,
      "step": 1188
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2446397542953491,
      "learning_rate": 9.590675738694431e-06,
      "loss": 0.8592,
      "step": 1189
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3298122882843018,
      "learning_rate": 9.58983439661457e-06,
      "loss": 0.8262,
      "step": 1190
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0231144428253174,
      "learning_rate": 9.588992227738281e-06,
      "loss": 1.0808,
      "step": 1191
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.30507230758667,
      "learning_rate": 9.588149232217264e-06,
      "loss": 0.8305,
      "step": 1192
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4013473987579346,
      "learning_rate": 9.587305410203376e-06,
      "loss": 0.8882,
      "step": 1193
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3226655721664429,
      "learning_rate": 9.586460761848617e-06,
      "loss": 0.8677,
      "step": 1194
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3220880031585693,
      "learning_rate": 9.585615287305144e-06,
      "loss": 0.8459,
      "step": 1195
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.550262212753296,
      "learning_rate": 9.58476898672525e-06,
      "loss": 0.8572,
      "step": 1196
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.063392162322998,
      "learning_rate": 9.583921860261393e-06,
      "loss": 1.0804,
      "step": 1197
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3829237222671509,
      "learning_rate": 9.583073908066166e-06,
      "loss": 0.8577,
      "step": 1198
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3293430805206299,
      "learning_rate": 9.582225130292319e-06,
      "loss": 0.8964,
      "step": 1199
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.513918161392212,
      "learning_rate": 9.581375527092744e-06,
      "loss": 0.8484,
      "step": 1200
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3656110763549805,
      "learning_rate": 9.58052509862049e-06,
      "loss": 0.7565,
      "step": 1201
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8094232082366943,
      "learning_rate": 9.579673845028747e-06,
      "loss": 1.0646,
      "step": 1202
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3657726049423218,
      "learning_rate": 9.578821766470859e-06,
      "loss": 0.8116,
      "step": 1203
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.284255862236023,
      "learning_rate": 9.577968863100315e-06,
      "loss": 0.8482,
      "step": 1204
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3633698225021362,
      "learning_rate": 9.577115135070753e-06,
      "loss": 0.8477,
      "step": 1205
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8523616790771484,
      "learning_rate": 9.576260582535963e-06,
      "loss": 1.0947,
      "step": 1206
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3445361852645874,
      "learning_rate": 9.575405205649881e-06,
      "loss": 0.8445,
      "step": 1207
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.364630937576294,
      "learning_rate": 9.574549004566589e-06,
      "loss": 0.8007,
      "step": 1208
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7226793169975281,
      "learning_rate": 9.573691979440321e-06,
      "loss": 1.0808,
      "step": 1209
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7326977252960205,
      "learning_rate": 9.57283413042546e-06,
      "loss": 1.1028,
      "step": 1210
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6823129057884216,
      "learning_rate": 9.571975457676534e-06,
      "loss": 1.1002,
      "step": 1211
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.705535352230072,
      "learning_rate": 9.571115961348222e-06,
      "loss": 1.0917,
      "step": 1212
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3143302202224731,
      "learning_rate": 9.570255641595352e-06,
      "loss": 0.8393,
      "step": 1213
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3889853954315186,
      "learning_rate": 9.569394498572895e-06,
      "loss": 0.8819,
      "step": 1214
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.298348069190979,
      "learning_rate": 9.568532532435979e-06,
      "loss": 0.8367,
      "step": 1215
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.276119351387024,
      "learning_rate": 9.567669743339872e-06,
      "loss": 0.7821,
      "step": 1216
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.312630295753479,
      "learning_rate": 9.566806131439995e-06,
      "loss": 0.7908,
      "step": 1217
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2452200651168823,
      "learning_rate": 9.565941696891917e-06,
      "loss": 1.1036,
      "step": 1218
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.424734115600586,
      "learning_rate": 9.565076439851354e-06,
      "loss": 0.8855,
      "step": 1219
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8839579224586487,
      "learning_rate": 9.564210360474166e-06,
      "loss": 1.0908,
      "step": 1220
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4845165014266968,
      "learning_rate": 9.563343458916372e-06,
      "loss": 0.8671,
      "step": 1221
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4364535808563232,
      "learning_rate": 9.562475735334128e-06,
      "loss": 0.8661,
      "step": 1222
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3614298105239868,
      "learning_rate": 9.561607189883746e-06,
      "loss": 0.8396,
      "step": 1223
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4202100038528442,
      "learning_rate": 9.56073782272168e-06,
      "loss": 0.867,
      "step": 1224
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4447605609893799,
      "learning_rate": 9.559867634004535e-06,
      "loss": 0.8812,
      "step": 1225
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3696284294128418,
      "learning_rate": 9.558996623889065e-06,
      "loss": 0.8872,
      "step": 1226
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3160380125045776,
      "learning_rate": 9.558124792532169e-06,
      "loss": 0.8072,
      "step": 1227
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4051451683044434,
      "learning_rate": 9.557252140090897e-06,
      "loss": 0.8489,
      "step": 1228
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2244306802749634,
      "learning_rate": 9.556378666722445e-06,
      "loss": 1.097,
      "step": 1229
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9009962677955627,
      "learning_rate": 9.555504372584158e-06,
      "loss": 1.0964,
      "step": 1230
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4007644653320312,
      "learning_rate": 9.554629257833528e-06,
      "loss": 0.7906,
      "step": 1231
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6166070699691772,
      "learning_rate": 9.553753322628192e-06,
      "loss": 0.8113,
      "step": 1232
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6417759656906128,
      "learning_rate": 9.552876567125943e-06,
      "loss": 0.8922,
      "step": 1233
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.565451979637146,
      "learning_rate": 9.551998991484715e-06,
      "loss": 0.8566,
      "step": 1234
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6739370822906494,
      "learning_rate": 9.551120595862588e-06,
      "loss": 1.0877,
      "step": 1235
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3996435403823853,
      "learning_rate": 9.550241380417797e-06,
      "loss": 0.8048,
      "step": 1236
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4188367128372192,
      "learning_rate": 9.54936134530872e-06,
      "loss": 0.8273,
      "step": 1237
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1033179759979248,
      "learning_rate": 9.54848049069388e-06,
      "loss": 1.0918,
      "step": 1238
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.531978964805603,
      "learning_rate": 9.547598816731956e-06,
      "loss": 0.8652,
      "step": 1239
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.502541422843933,
      "learning_rate": 9.546716323581766e-06,
      "loss": 0.7921,
      "step": 1240
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9586502313613892,
      "learning_rate": 9.545833011402281e-06,
      "loss": 1.0966,
      "step": 1241
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.359649419784546,
      "learning_rate": 9.544948880352618e-06,
      "loss": 0.8063,
      "step": 1242
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4459589719772339,
      "learning_rate": 9.54406393059204e-06,
      "loss": 0.8511,
      "step": 1243
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3860434293746948,
      "learning_rate": 9.54317816227996e-06,
      "loss": 0.8715,
      "step": 1244
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9017804265022278,
      "learning_rate": 9.542291575575936e-06,
      "loss": 1.0845,
      "step": 1245
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4538642168045044,
      "learning_rate": 9.541404170639672e-06,
      "loss": 0.8067,
      "step": 1246
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3346060514450073,
      "learning_rate": 9.540515947631028e-06,
      "loss": 0.8938,
      "step": 1247
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2944363355636597,
      "learning_rate": 9.539626906710002e-06,
      "loss": 0.8499,
      "step": 1248
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5340720415115356,
      "learning_rate": 9.538737048036741e-06,
      "loss": 0.8478,
      "step": 1249
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4416418075561523,
      "learning_rate": 9.537846371771544e-06,
      "loss": 0.8765,
      "step": 1250
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9851123690605164,
      "learning_rate": 9.536954878074853e-06,
      "loss": 1.0817,
      "step": 1251
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2780073881149292,
      "learning_rate": 9.536062567107258e-06,
      "loss": 0.8265,
      "step": 1252
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2700648307800293,
      "learning_rate": 9.5351694390295e-06,
      "loss": 0.8126,
      "step": 1253
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3250863552093506,
      "learning_rate": 9.534275494002457e-06,
      "loss": 0.892,
      "step": 1254
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7603076100349426,
      "learning_rate": 9.533380732187168e-06,
      "loss": 1.079,
      "step": 1255
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7352347373962402,
      "learning_rate": 9.532485153744808e-06,
      "loss": 1.0905,
      "step": 1256
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4852608442306519,
      "learning_rate": 9.531588758836706e-06,
      "loss": 0.8946,
      "step": 1257
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.361838459968567,
      "learning_rate": 9.530691547624333e-06,
      "loss": 0.786,
      "step": 1258
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3370107412338257,
      "learning_rate": 9.529793520269312e-06,
      "loss": 0.8649,
      "step": 1259
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3394869565963745,
      "learning_rate": 9.528894676933408e-06,
      "loss": 0.8078,
      "step": 1260
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9047752618789673,
      "learning_rate": 9.527995017778538e-06,
      "loss": 1.1023,
      "step": 1261
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3139177560806274,
      "learning_rate": 9.52709454296676e-06,
      "loss": 0.8031,
      "step": 1262
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6108094453811646,
      "learning_rate": 9.526193252660285e-06,
      "loss": 0.8548,
      "step": 1263
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3132632970809937,
      "learning_rate": 9.525291147021467e-06,
      "loss": 0.8637,
      "step": 1264
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7757630348205566,
      "learning_rate": 9.52438822621281e-06,
      "loss": 1.0806,
      "step": 1265
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.357497215270996,
      "learning_rate": 9.523484490396961e-06,
      "loss": 0.857,
      "step": 1266
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3675198554992676,
      "learning_rate": 9.522579939736716e-06,
      "loss": 0.8631,
      "step": 1267
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2886766195297241,
      "learning_rate": 9.521674574395016e-06,
      "loss": 0.8634,
      "step": 1268
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3695002794265747,
      "learning_rate": 9.520768394534955e-06,
      "loss": 0.8655,
      "step": 1269
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3531100749969482,
      "learning_rate": 9.519861400319765e-06,
      "loss": 0.9047,
      "step": 1270
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0868182182312012,
      "learning_rate": 9.51895359191283e-06,
      "loss": 1.1131,
      "step": 1271
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7985073924064636,
      "learning_rate": 9.518044969477678e-06,
      "loss": 1.1003,
      "step": 1272
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.646589994430542,
      "learning_rate": 9.517135533177989e-06,
      "loss": 1.0861,
      "step": 1273
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4518235921859741,
      "learning_rate": 9.516225283177579e-06,
      "loss": 0.8309,
      "step": 1274
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4729489088058472,
      "learning_rate": 9.515314219640424e-06,
      "loss": 0.8678,
      "step": 1275
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4270097017288208,
      "learning_rate": 9.514402342730636e-06,
      "loss": 0.8974,
      "step": 1276
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.377945065498352,
      "learning_rate": 9.513489652612475e-06,
      "loss": 0.8376,
      "step": 1277
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.357333779335022,
      "learning_rate": 9.512576149450356e-06,
      "loss": 0.861,
      "step": 1278
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3402857780456543,
      "learning_rate": 9.511661833408831e-06,
      "loss": 0.8485,
      "step": 1279
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4896376132965088,
      "learning_rate": 9.5107467046526e-06,
      "loss": 1.1196,
      "step": 1280
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2472503185272217,
      "learning_rate": 9.509830763346514e-06,
      "loss": 1.1022,
      "step": 1281
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5922070741653442,
      "learning_rate": 9.508914009655565e-06,
      "loss": 0.8788,
      "step": 1282
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6288601160049438,
      "learning_rate": 9.507996443744896e-06,
      "loss": 0.8581,
      "step": 1283
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5905959606170654,
      "learning_rate": 9.507078065779793e-06,
      "loss": 0.8481,
      "step": 1284
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1764662265777588,
      "learning_rate": 9.506158875925689e-06,
      "loss": 1.0618,
      "step": 1285
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.433725118637085,
      "learning_rate": 9.505238874348161e-06,
      "loss": 0.8844,
      "step": 1286
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4217852354049683,
      "learning_rate": 9.50431806121294e-06,
      "loss": 0.8671,
      "step": 1287
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4560962915420532,
      "learning_rate": 9.503396436685896e-06,
      "loss": 0.8033,
      "step": 1288
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4171996116638184,
      "learning_rate": 9.502474000933046e-06,
      "loss": 0.8231,
      "step": 1289
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4406673908233643,
      "learning_rate": 9.501550754120557e-06,
      "loss": 0.8645,
      "step": 1290
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3404183387756348,
      "learning_rate": 9.500626696414734e-06,
      "loss": 0.8293,
      "step": 1291
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.228440761566162,
      "learning_rate": 9.49970182798204e-06,
      "loss": 0.8786,
      "step": 1292
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7520430684089661,
      "learning_rate": 9.498776148989073e-06,
      "loss": 1.0852,
      "step": 1293
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3946267366409302,
      "learning_rate": 9.497849659602582e-06,
      "loss": 0.8125,
      "step": 1294
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4452165365219116,
      "learning_rate": 9.496922359989464e-06,
      "loss": 0.8488,
      "step": 1295
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.392402172088623,
      "learning_rate": 9.495994250316756e-06,
      "loss": 0.8039,
      "step": 1296
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3784676790237427,
      "learning_rate": 9.495065330751649e-06,
      "loss": 0.8073,
      "step": 1297
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3267258405685425,
      "learning_rate": 9.494135601461469e-06,
      "loss": 0.8547,
      "step": 1298
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2812172174453735,
      "learning_rate": 9.4932050626137e-06,
      "loss": 0.8368,
      "step": 1299
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7517141699790955,
      "learning_rate": 9.492273714375963e-06,
      "loss": 1.0984,
      "step": 1300
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6893753409385681,
      "learning_rate": 9.49134155691603e-06,
      "loss": 1.0939,
      "step": 1301
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.672039806842804,
      "learning_rate": 9.490408590401814e-06,
      "loss": 1.0957,
      "step": 1302
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4504791498184204,
      "learning_rate": 9.489474815001377e-06,
      "loss": 0.8602,
      "step": 1303
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4273673295974731,
      "learning_rate": 9.488540230882925e-06,
      "loss": 0.914,
      "step": 1304
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3424534797668457,
      "learning_rate": 9.487604838214812e-06,
      "loss": 0.8107,
      "step": 1305
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3134636878967285,
      "learning_rate": 9.486668637165536e-06,
      "loss": 0.8625,
      "step": 1306
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.249947190284729,
      "learning_rate": 9.485731627903743e-06,
      "loss": 0.8211,
      "step": 1307
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2365022897720337,
      "learning_rate": 9.484793810598218e-06,
      "loss": 0.7903,
      "step": 1308
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2515043020248413,
      "learning_rate": 9.4838551854179e-06,
      "loss": 0.7785,
      "step": 1309
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5721685886383057,
      "learning_rate": 9.482915752531868e-06,
      "loss": 1.0899,
      "step": 1310
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3527380228042603,
      "learning_rate": 9.481975512109349e-06,
      "loss": 0.8112,
      "step": 1311
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3459982872009277,
      "learning_rate": 9.481034464319713e-06,
      "loss": 0.8053,
      "step": 1312
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.380158543586731,
      "learning_rate": 9.480092609332477e-06,
      "loss": 0.8325,
      "step": 1313
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7222761511802673,
      "learning_rate": 9.479149947317306e-06,
      "loss": 1.0893,
      "step": 1314
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4531276226043701,
      "learning_rate": 9.478206478444006e-06,
      "loss": 0.9031,
      "step": 1315
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7948148846626282,
      "learning_rate": 9.477262202882528e-06,
      "loss": 1.0975,
      "step": 1316
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7087625861167908,
      "learning_rate": 9.476317120802974e-06,
      "loss": 1.1146,
      "step": 1317
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6673977971076965,
      "learning_rate": 9.475371232375588e-06,
      "loss": 1.0804,
      "step": 1318
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3887354135513306,
      "learning_rate": 9.474424537770755e-06,
      "loss": 0.8623,
      "step": 1319
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3968400955200195,
      "learning_rate": 9.473477037159012e-06,
      "loss": 0.8165,
      "step": 1320
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.365622878074646,
      "learning_rate": 9.472528730711039e-06,
      "loss": 0.8876,
      "step": 1321
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2867693901062012,
      "learning_rate": 9.471579618597657e-06,
      "loss": 0.8264,
      "step": 1322
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2688943147659302,
      "learning_rate": 9.47062970098984e-06,
      "loss": 0.812,
      "step": 1323
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3971302509307861,
      "learning_rate": 9.4696789780587e-06,
      "loss": 1.073,
      "step": 1324
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0725568532943726,
      "learning_rate": 9.4687274499755e-06,
      "loss": 1.0657,
      "step": 1325
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3578400611877441,
      "learning_rate": 9.46777511691164e-06,
      "loss": 0.8097,
      "step": 1326
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4870574474334717,
      "learning_rate": 9.466821979038675e-06,
      "loss": 0.8533,
      "step": 1327
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5530592203140259,
      "learning_rate": 9.465868036528295e-06,
      "loss": 0.8444,
      "step": 1328
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.361954927444458,
      "learning_rate": 9.464913289552342e-06,
      "loss": 0.7734,
      "step": 1329
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8530426025390625,
      "learning_rate": 9.463957738282802e-06,
      "loss": 1.0869,
      "step": 1330
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4740731716156006,
      "learning_rate": 9.463001382891805e-06,
      "loss": 0.9011,
      "step": 1331
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3800307512283325,
      "learning_rate": 9.462044223551621e-06,
      "loss": 0.8398,
      "step": 1332
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4854665994644165,
      "learning_rate": 9.461086260434674e-06,
      "loss": 0.8386,
      "step": 1333
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4832950830459595,
      "learning_rate": 9.460127493713526e-06,
      "loss": 0.8282,
      "step": 1334
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.506927728652954,
      "learning_rate": 9.459167923560886e-06,
      "loss": 0.8082,
      "step": 1335
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5116089582443237,
      "learning_rate": 9.458207550149608e-06,
      "loss": 0.8871,
      "step": 1336
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3330212831497192,
      "learning_rate": 9.45724637365269e-06,
      "loss": 0.8706,
      "step": 1337
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.973381519317627,
      "learning_rate": 9.456284394243273e-06,
      "loss": 1.0734,
      "step": 1338
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4354925155639648,
      "learning_rate": 9.455321612094649e-06,
      "loss": 0.8992,
      "step": 1339
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3726251125335693,
      "learning_rate": 9.454358027380246e-06,
      "loss": 0.8516,
      "step": 1340
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4162287712097168,
      "learning_rate": 9.453393640273642e-06,
      "loss": 0.8396,
      "step": 1341
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4941431283950806,
      "learning_rate": 9.452428450948562e-06,
      "loss": 0.8534,
      "step": 1342
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3635848760604858,
      "learning_rate": 9.451462459578864e-06,
      "loss": 0.9012,
      "step": 1343
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1403040885925293,
      "learning_rate": 9.450495666338563e-06,
      "loss": 1.0733,
      "step": 1344
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3520761728286743,
      "learning_rate": 9.449528071401816e-06,
      "loss": 0.8664,
      "step": 1345
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.306984305381775,
      "learning_rate": 9.448559674942916e-06,
      "loss": 0.8653,
      "step": 1346
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3529189825057983,
      "learning_rate": 9.447590477136312e-06,
      "loss": 0.8725,
      "step": 1347
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7388327121734619,
      "learning_rate": 9.446620478156589e-06,
      "loss": 1.0804,
      "step": 1348
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3184235095977783,
      "learning_rate": 9.445649678178482e-06,
      "loss": 0.8428,
      "step": 1349
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7931175827980042,
      "learning_rate": 9.444678077376862e-06,
      "loss": 1.0937,
      "step": 1350
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7553543448448181,
      "learning_rate": 9.443705675926755e-06,
      "loss": 1.0488,
      "step": 1351
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3688057661056519,
      "learning_rate": 9.442732474003323e-06,
      "loss": 0.8903,
      "step": 1352
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3399884700775146,
      "learning_rate": 9.441758471781876e-06,
      "loss": 0.8234,
      "step": 1353
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3390060663223267,
      "learning_rate": 9.440783669437868e-06,
      "loss": 0.8272,
      "step": 1354
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9303665161132812,
      "learning_rate": 9.439808067146895e-06,
      "loss": 1.1196,
      "step": 1355
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7992119193077087,
      "learning_rate": 9.4388316650847e-06,
      "loss": 1.0756,
      "step": 1356
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2842708826065063,
      "learning_rate": 9.437854463427168e-06,
      "loss": 0.8444,
      "step": 1357
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.290238380432129,
      "learning_rate": 9.43687646235033e-06,
      "loss": 0.826,
      "step": 1358
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9324689507484436,
      "learning_rate": 9.435897662030354e-06,
      "loss": 1.0803,
      "step": 1359
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2995412349700928,
      "learning_rate": 9.434918062643564e-06,
      "loss": 0.8334,
      "step": 1360
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7312290072441101,
      "learning_rate": 9.433937664366419e-06,
      "loss": 1.0821,
      "step": 1361
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3977360725402832,
      "learning_rate": 9.432956467375525e-06,
      "loss": 0.8841,
      "step": 1362
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.265762209892273,
      "learning_rate": 9.431974471847633e-06,
      "loss": 0.8377,
      "step": 1363
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.315702199935913,
      "learning_rate": 9.430991677959632e-06,
      "loss": 0.8735,
      "step": 1364
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9992291331291199,
      "learning_rate": 9.430008085888563e-06,
      "loss": 1.1103,
      "step": 1365
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.780348002910614,
      "learning_rate": 9.429023695811603e-06,
      "loss": 1.0873,
      "step": 1366
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4217690229415894,
      "learning_rate": 9.428038507906079e-06,
      "loss": 0.8229,
      "step": 1367
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8966766595840454,
      "learning_rate": 9.42705252234946e-06,
      "loss": 1.0628,
      "step": 1368
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4170005321502686,
      "learning_rate": 9.426065739319356e-06,
      "loss": 0.8412,
      "step": 1369
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.339700698852539,
      "learning_rate": 9.425078158993524e-06,
      "loss": 0.8585,
      "step": 1370
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.323921799659729,
      "learning_rate": 9.424089781549862e-06,
      "loss": 0.8642,
      "step": 1371
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.359611988067627,
      "learning_rate": 9.423100607166413e-06,
      "loss": 0.8606,
      "step": 1372
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3035486936569214,
      "learning_rate": 9.422110636021365e-06,
      "loss": 0.8554,
      "step": 1373
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.307984709739685,
      "learning_rate": 9.421119868293045e-06,
      "loss": 0.8096,
      "step": 1374
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3453319072723389,
      "learning_rate": 9.420128304159927e-06,
      "loss": 0.8415,
      "step": 1375
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3466660976409912,
      "learning_rate": 9.41913594380063e-06,
      "loss": 0.8613,
      "step": 1376
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3745344877243042,
      "learning_rate": 9.418142787393913e-06,
      "loss": 0.8393,
      "step": 1377
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3506063222885132,
      "learning_rate": 9.417148835118681e-06,
      "loss": 0.8478,
      "step": 1378
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3085591793060303,
      "learning_rate": 9.416154087153978e-06,
      "loss": 0.8854,
      "step": 1379
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.272370457649231,
      "learning_rate": 9.415158543678998e-06,
      "loss": 0.8425,
      "step": 1380
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0516295433044434,
      "learning_rate": 9.41416220487307e-06,
      "loss": 1.1131,
      "step": 1381
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8363462686538696,
      "learning_rate": 9.413165070915676e-06,
      "loss": 1.0963,
      "step": 1382
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4301022291183472,
      "learning_rate": 9.412167141986433e-06,
      "loss": 0.8698,
      "step": 1383
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4652000665664673,
      "learning_rate": 9.411168418265105e-06,
      "loss": 0.8723,
      "step": 1384
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3548330068588257,
      "learning_rate": 9.410168899931599e-06,
      "loss": 0.8077,
      "step": 1385
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3682010173797607,
      "learning_rate": 9.409168587165964e-06,
      "loss": 0.7875,
      "step": 1386
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7505738735198975,
      "learning_rate": 9.408167480148394e-06,
      "loss": 1.1067,
      "step": 1387
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3583320379257202,
      "learning_rate": 9.407165579059224e-06,
      "loss": 0.8459,
      "step": 1388
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5074783563613892,
      "learning_rate": 9.406162884078932e-06,
      "loss": 0.8422,
      "step": 1389
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.147648572921753,
      "learning_rate": 9.405159395388141e-06,
      "loss": 1.0998,
      "step": 1390
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3811131715774536,
      "learning_rate": 9.404155113167617e-06,
      "loss": 0.808,
      "step": 1391
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3765759468078613,
      "learning_rate": 9.403150037598266e-06,
      "loss": 0.8734,
      "step": 1392
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.360167145729065,
      "learning_rate": 9.402144168861139e-06,
      "loss": 0.8385,
      "step": 1393
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3884152173995972,
      "learning_rate": 9.401137507137428e-06,
      "loss": 0.8274,
      "step": 1394
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3474150896072388,
      "learning_rate": 9.400130052608473e-06,
      "loss": 0.8968,
      "step": 1395
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3147811889648438,
      "learning_rate": 9.399121805455753e-06,
      "loss": 0.8248,
      "step": 1396
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5646815299987793,
      "learning_rate": 9.398112765860887e-06,
      "loss": 1.1013,
      "step": 1397
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3938018083572388,
      "learning_rate": 9.397102934005642e-06,
      "loss": 0.8678,
      "step": 1398
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.448961853981018,
      "learning_rate": 9.396092310071924e-06,
      "loss": 0.8574,
      "step": 1399
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9481305480003357,
      "learning_rate": 9.395080894241786e-06,
      "loss": 1.0973,
      "step": 1400
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7817569375038147,
      "learning_rate": 9.394068686697418e-06,
      "loss": 1.1159,
      "step": 1401
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7686485052108765,
      "learning_rate": 9.393055687621155e-06,
      "loss": 1.0979,
      "step": 1402
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8624433279037476,
      "learning_rate": 9.392041897195478e-06,
      "loss": 1.0761,
      "step": 1403
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.450152039527893,
      "learning_rate": 9.391027315603007e-06,
      "loss": 0.8242,
      "step": 1404
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4939920902252197,
      "learning_rate": 9.390011943026505e-06,
      "loss": 0.8934,
      "step": 1405
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.365085482597351,
      "learning_rate": 9.388995779648875e-06,
      "loss": 0.8371,
      "step": 1406
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3236066102981567,
      "learning_rate": 9.387978825653169e-06,
      "loss": 1.0781,
      "step": 1407
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3779226541519165,
      "learning_rate": 9.386961081222572e-06,
      "loss": 0.9046,
      "step": 1408
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9663125872612,
      "learning_rate": 9.38594254654042e-06,
      "loss": 1.0534,
      "step": 1409
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7660959362983704,
      "learning_rate": 9.384923221790192e-06,
      "loss": 1.0836,
      "step": 1410
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8398343920707703,
      "learning_rate": 9.3839031071555e-06,
      "loss": 1.064,
      "step": 1411
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4359924793243408,
      "learning_rate": 9.382882202820107e-06,
      "loss": 0.8262,
      "step": 1412
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9235137104988098,
      "learning_rate": 9.381860508967913e-06,
      "loss": 1.1135,
      "step": 1413
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8585918545722961,
      "learning_rate": 9.380838025782964e-06,
      "loss": 1.092,
      "step": 1414
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3760067224502563,
      "learning_rate": 9.379814753449445e-06,
      "loss": 0.8285,
      "step": 1415
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3128650188446045,
      "learning_rate": 9.378790692151685e-06,
      "loss": 0.8551,
      "step": 1416
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2940764427185059,
      "learning_rate": 9.377765842074155e-06,
      "loss": 0.8324,
      "step": 1417
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3178857564926147,
      "learning_rate": 9.376740203401468e-06,
      "loss": 0.8649,
      "step": 1418
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.443345069885254,
      "learning_rate": 9.375713776318381e-06,
      "loss": 0.8414,
      "step": 1419
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.265661358833313,
      "learning_rate": 9.374686561009786e-06,
      "loss": 0.8486,
      "step": 1420
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.428977131843567,
      "learning_rate": 9.373658557660726e-06,
      "loss": 0.8214,
      "step": 1421
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3432332277297974,
      "learning_rate": 9.37262976645638e-06,
      "loss": 0.8531,
      "step": 1422
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8380684852600098,
      "learning_rate": 9.37160018758207e-06,
      "loss": 1.085,
      "step": 1423
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2948821783065796,
      "learning_rate": 9.370569821223264e-06,
      "loss": 0.8435,
      "step": 1424
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.111533284187317,
      "learning_rate": 9.369538667565566e-06,
      "loss": 1.0989,
      "step": 1425
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5948134660720825,
      "learning_rate": 9.368506726794725e-06,
      "loss": 0.8621,
      "step": 1426
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4261435270309448,
      "learning_rate": 9.36747399909663e-06,
      "loss": 0.8212,
      "step": 1427
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3668572902679443,
      "learning_rate": 9.366440484657316e-06,
      "loss": 0.8176,
      "step": 1428
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3179011344909668,
      "learning_rate": 9.365406183662952e-06,
      "loss": 0.8146,
      "step": 1429
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3499373197555542,
      "learning_rate": 9.364371096299855e-06,
      "loss": 0.8155,
      "step": 1430
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2824798822402954,
      "learning_rate": 9.363335222754486e-06,
      "loss": 0.8665,
      "step": 1431
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4032585620880127,
      "learning_rate": 9.36229856321344e-06,
      "loss": 0.83,
      "step": 1432
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3668174743652344,
      "learning_rate": 9.361261117863457e-06,
      "loss": 0.882,
      "step": 1433
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3565223217010498,
      "learning_rate": 9.360222886891417e-06,
      "loss": 0.849,
      "step": 1434
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3566675186157227,
      "learning_rate": 9.359183870484347e-06,
      "loss": 0.8321,
      "step": 1435
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.161623239517212,
      "learning_rate": 9.35814406882941e-06,
      "loss": 1.073,
      "step": 1436
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3060863018035889,
      "learning_rate": 9.357103482113912e-06,
      "loss": 0.8892,
      "step": 1437
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4976246356964111,
      "learning_rate": 9.356062110525302e-06,
      "loss": 0.8202,
      "step": 1438
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.334802508354187,
      "learning_rate": 9.355019954251165e-06,
      "loss": 0.8408,
      "step": 1439
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1697198152542114,
      "learning_rate": 9.353977013479238e-06,
      "loss": 1.0829,
      "step": 1440
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.413004994392395,
      "learning_rate": 9.352933288397387e-06,
      "loss": 0.9262,
      "step": 1441
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5979586839675903,
      "learning_rate": 9.351888779193626e-06,
      "loss": 0.8667,
      "step": 1442
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3513472080230713,
      "learning_rate": 9.35084348605611e-06,
      "loss": 0.8928,
      "step": 1443
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2105923891067505,
      "learning_rate": 9.349797409173136e-06,
      "loss": 0.8311,
      "step": 1444
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2699592113494873,
      "learning_rate": 9.348750548733138e-06,
      "loss": 0.8072,
      "step": 1445
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3437857627868652,
      "learning_rate": 9.347702904924694e-06,
      "loss": 0.8156,
      "step": 1446
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6583819389343262,
      "learning_rate": 9.346654477936523e-06,
      "loss": 1.076,
      "step": 1447
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.313611388206482,
      "learning_rate": 9.345605267957487e-06,
      "loss": 0.838,
      "step": 1448
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6566319465637207,
      "learning_rate": 9.344555275176585e-06,
      "loss": 0.826,
      "step": 1449
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8289582133293152,
      "learning_rate": 9.343504499782958e-06,
      "loss": 1.0813,
      "step": 1450
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3794798851013184,
      "learning_rate": 9.342452941965892e-06,
      "loss": 0.816,
      "step": 1451
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6992301940917969,
      "learning_rate": 9.341400601914807e-06,
      "loss": 1.0774,
      "step": 1452
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4044098854064941,
      "learning_rate": 9.340347479819272e-06,
      "loss": 0.8232,
      "step": 1453
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.801805317401886,
      "learning_rate": 9.339293575868988e-06,
      "loss": 1.0657,
      "step": 1454
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2640876770019531,
      "learning_rate": 9.338238890253805e-06,
      "loss": 0.8745,
      "step": 1455
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2911049127578735,
      "learning_rate": 9.33718342316371e-06,
      "loss": 0.8147,
      "step": 1456
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.854651153087616,
      "learning_rate": 9.33612717478883e-06,
      "loss": 1.0955,
      "step": 1457
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8287932872772217,
      "learning_rate": 9.335070145319433e-06,
      "loss": 1.0799,
      "step": 1458
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6817673444747925,
      "learning_rate": 9.334012334945932e-06,
      "loss": 1.0763,
      "step": 1459
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7506005167961121,
      "learning_rate": 9.332953743858874e-06,
      "loss": 1.0837,
      "step": 1460
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.378578543663025,
      "learning_rate": 9.33189437224895e-06,
      "loss": 0.85,
      "step": 1461
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3268609046936035,
      "learning_rate": 9.330834220306995e-06,
      "loss": 0.8695,
      "step": 1462
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.352238416671753,
      "learning_rate": 9.329773288223976e-06,
      "loss": 0.86,
      "step": 1463
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4584285020828247,
      "learning_rate": 9.328711576191007e-06,
      "loss": 0.8309,
      "step": 1464
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2804322242736816,
      "learning_rate": 9.327649084399341e-06,
      "loss": 0.7703,
      "step": 1465
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.277581810951233,
      "learning_rate": 9.326585813040377e-06,
      "loss": 0.8546,
      "step": 1466
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2665849924087524,
      "learning_rate": 9.325521762305639e-06,
      "loss": 0.8141,
      "step": 1467
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3641159534454346,
      "learning_rate": 9.324456932386809e-06,
      "loss": 0.792,
      "step": 1468
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.352030873298645,
      "learning_rate": 9.323391323475699e-06,
      "loss": 0.9079,
      "step": 1469
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5250591039657593,
      "learning_rate": 9.322324935764264e-06,
      "loss": 0.8703,
      "step": 1470
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.344261884689331,
      "learning_rate": 9.321257769444598e-06,
      "loss": 0.8455,
      "step": 1471
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3150792121887207,
      "learning_rate": 9.32018982470894e-06,
      "loss": 0.7787,
      "step": 1472
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3025643825531006,
      "learning_rate": 9.319121101749664e-06,
      "loss": 0.8784,
      "step": 1473
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3136852979660034,
      "learning_rate": 9.318051600759284e-06,
      "loss": 0.8581,
      "step": 1474
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2911641597747803,
      "learning_rate": 9.316981321930459e-06,
      "loss": 0.8602,
      "step": 1475
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2623460292816162,
      "learning_rate": 9.315910265455983e-06,
      "loss": 0.8373,
      "step": 1476
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3164128065109253,
      "learning_rate": 9.314838431528794e-06,
      "loss": 0.8286,
      "step": 1477
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.388015866279602,
      "learning_rate": 9.313765820341968e-06,
      "loss": 0.7979,
      "step": 1478
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6368601322174072,
      "learning_rate": 9.31269243208872e-06,
      "loss": 1.073,
      "step": 1479
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.34499990940094,
      "learning_rate": 9.311618266962408e-06,
      "loss": 0.8006,
      "step": 1480
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4216598272323608,
      "learning_rate": 9.310543325156529e-06,
      "loss": 0.8615,
      "step": 1481
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.498052716255188,
      "learning_rate": 9.309467606864715e-06,
      "loss": 0.8618,
      "step": 1482
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3018265962600708,
      "learning_rate": 9.308391112280747e-06,
      "loss": 0.8577,
      "step": 1483
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8114219307899475,
      "learning_rate": 9.307313841598538e-06,
      "loss": 1.0858,
      "step": 1484
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7600961327552795,
      "learning_rate": 9.306235795012144e-06,
      "loss": 1.1066,
      "step": 1485
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6589187383651733,
      "learning_rate": 9.305156972715762e-06,
      "loss": 1.0703,
      "step": 1486
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.384537935256958,
      "learning_rate": 9.304077374903726e-06,
      "loss": 0.8674,
      "step": 1487
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9091282486915588,
      "learning_rate": 9.302997001770511e-06,
      "loss": 1.0664,
      "step": 1488
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2807897329330444,
      "learning_rate": 9.301915853510733e-06,
      "loss": 0.8344,
      "step": 1489
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.33499276638031,
      "learning_rate": 9.300833930319142e-06,
      "loss": 0.9091,
      "step": 1490
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2711564302444458,
      "learning_rate": 9.299751232390638e-06,
      "loss": 0.8671,
      "step": 1491
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7413788437843323,
      "learning_rate": 9.298667759920248e-06,
      "loss": 1.104,
      "step": 1492
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3468585014343262,
      "learning_rate": 9.29758351310315e-06,
      "loss": 0.8889,
      "step": 1493
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.284246563911438,
      "learning_rate": 9.296498492134653e-06,
      "loss": 0.8116,
      "step": 1494
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.335875391960144,
      "learning_rate": 9.29541269721021e-06,
      "loss": 0.8426,
      "step": 1495
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.294775366783142,
      "learning_rate": 9.294326128525413e-06,
      "loss": 0.8416,
      "step": 1496
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3336918354034424,
      "learning_rate": 9.29323878627599e-06,
      "loss": 0.8707,
      "step": 1497
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3363111019134521,
      "learning_rate": 9.292150670657817e-06,
      "loss": 0.7817,
      "step": 1498
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2386224269866943,
      "learning_rate": 9.291061781866896e-06,
      "loss": 0.8521,
      "step": 1499
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.96402907371521,
      "learning_rate": 9.28997212009938e-06,
      "loss": 0.8395,
      "step": 1500
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3361625671386719,
      "learning_rate": 9.288881685551554e-06,
      "loss": 0.8445,
      "step": 1501
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2879427671432495,
      "learning_rate": 9.287790478419849e-06,
      "loss": 0.8079,
      "step": 1502
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9230688214302063,
      "learning_rate": 9.286698498900825e-06,
      "loss": 1.0791,
      "step": 1503
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4470514059066772,
      "learning_rate": 9.285605747191193e-06,
      "loss": 0.861,
      "step": 1504
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3106613159179688,
      "learning_rate": 9.284512223487797e-06,
      "loss": 0.7899,
      "step": 1505
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3916840553283691,
      "learning_rate": 9.283417927987616e-06,
      "loss": 0.8736,
      "step": 1506
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2549017667770386,
      "learning_rate": 9.282322860887778e-06,
      "loss": 0.881,
      "step": 1507
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.002413034439087,
      "learning_rate": 9.28122702238554e-06,
      "loss": 1.0959,
      "step": 1508
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4062464237213135,
      "learning_rate": 9.280130412678306e-06,
      "loss": 0.8575,
      "step": 1509
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7213440537452698,
      "learning_rate": 9.27903303196361e-06,
      "loss": 1.1135,
      "step": 1510
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.292047142982483,
      "learning_rate": 9.277934880439138e-06,
      "loss": 0.8701,
      "step": 1511
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3688081502914429,
      "learning_rate": 9.276835958302702e-06,
      "loss": 0.8201,
      "step": 1512
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.295635461807251,
      "learning_rate": 9.275736265752257e-06,
      "loss": 0.7909,
      "step": 1513
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3503159284591675,
      "learning_rate": 9.2746358029859e-06,
      "loss": 0.8968,
      "step": 1514
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3550487756729126,
      "learning_rate": 9.273534570201867e-06,
      "loss": 0.8817,
      "step": 1515
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2036772966384888,
      "learning_rate": 9.272432567598523e-06,
      "loss": 0.846,
      "step": 1516
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3899786472320557,
      "learning_rate": 9.271329795374386e-06,
      "loss": 0.8306,
      "step": 1517
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2401549816131592,
      "learning_rate": 9.270226253728104e-06,
      "loss": 0.8783,
      "step": 1518
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3440431356430054,
      "learning_rate": 9.26912194285846e-06,
      "loss": 0.8008,
      "step": 1519
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2115892171859741,
      "learning_rate": 9.268016862964386e-06,
      "loss": 0.7486,
      "step": 1520
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2541937828063965,
      "learning_rate": 9.266911014244946e-06,
      "loss": 0.7386,
      "step": 1521
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3019882440567017,
      "learning_rate": 9.26580439689934e-06,
      "loss": 0.7817,
      "step": 1522
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6792054176330566,
      "learning_rate": 9.264697011126916e-06,
      "loss": 1.0906,
      "step": 1523
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1956677436828613,
      "learning_rate": 9.263588857127152e-06,
      "loss": 1.1018,
      "step": 1524
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3343505859375,
      "learning_rate": 9.262479935099665e-06,
      "loss": 0.8087,
      "step": 1525
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4420499801635742,
      "learning_rate": 9.261370245244214e-06,
      "loss": 0.7999,
      "step": 1526
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4555829763412476,
      "learning_rate": 9.260259787760695e-06,
      "loss": 0.8261,
      "step": 1527
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3999782800674438,
      "learning_rate": 9.25914856284914e-06,
      "loss": 0.9138,
      "step": 1528
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3798339366912842,
      "learning_rate": 9.258036570709724e-06,
      "loss": 0.8378,
      "step": 1529
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.142245292663574,
      "learning_rate": 9.256923811542755e-06,
      "loss": 1.1172,
      "step": 1530
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2857598066329956,
      "learning_rate": 9.255810285548684e-06,
      "loss": 0.7816,
      "step": 1531
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4648706912994385,
      "learning_rate": 9.254695992928094e-06,
      "loss": 0.7883,
      "step": 1532
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3678169250488281,
      "learning_rate": 9.253580933881711e-06,
      "loss": 0.8192,
      "step": 1533
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.479918360710144,
      "learning_rate": 9.252465108610397e-06,
      "loss": 0.8463,
      "step": 1534
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4155808687210083,
      "learning_rate": 9.251348517315158e-06,
      "loss": 0.8496,
      "step": 1535
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2983982563018799,
      "learning_rate": 9.250231160197127e-06,
      "loss": 0.7876,
      "step": 1536
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.249161958694458,
      "learning_rate": 9.249113037457582e-06,
      "loss": 0.8738,
      "step": 1537
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3105262517929077,
      "learning_rate": 9.247994149297938e-06,
      "loss": 0.839,
      "step": 1538
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3886300325393677,
      "learning_rate": 9.246874495919749e-06,
      "loss": 1.1065,
      "step": 1539
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4158086776733398,
      "learning_rate": 9.245754077524702e-06,
      "loss": 0.8906,
      "step": 1540
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.329932689666748,
      "learning_rate": 9.244632894314628e-06,
      "loss": 0.8478,
      "step": 1541
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.737909197807312,
      "learning_rate": 9.243510946491494e-06,
      "loss": 1.1029,
      "step": 1542
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4213666915893555,
      "learning_rate": 9.242388234257402e-06,
      "loss": 0.8508,
      "step": 1543
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2906756401062012,
      "learning_rate": 9.241264757814591e-06,
      "loss": 0.8215,
      "step": 1544
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4033042192459106,
      "learning_rate": 9.240140517365445e-06,
      "loss": 0.8406,
      "step": 1545
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1651874780654907,
      "learning_rate": 9.239015513112476e-06,
      "loss": 1.0752,
      "step": 1546
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8419615030288696,
      "learning_rate": 9.237889745258342e-06,
      "loss": 1.0632,
      "step": 1547
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6817364692687988,
      "learning_rate": 9.236763214005832e-06,
      "loss": 1.109,
      "step": 1548
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3979108333587646,
      "learning_rate": 9.235635919557876e-06,
      "loss": 0.8189,
      "step": 1549
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.476293921470642,
      "learning_rate": 9.234507862117542e-06,
      "loss": 0.8228,
      "step": 1550
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2856054306030273,
      "learning_rate": 9.233379041888034e-06,
      "loss": 0.8282,
      "step": 1551
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4057316780090332,
      "learning_rate": 9.232249459072693e-06,
      "loss": 0.8774,
      "step": 1552
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6266738176345825,
      "learning_rate": 9.231119113874998e-06,
      "loss": 1.0839,
      "step": 1553
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.395044207572937,
      "learning_rate": 9.229988006498564e-06,
      "loss": 1.0968,
      "step": 1554
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8493461608886719,
      "learning_rate": 9.228856137147149e-06,
      "loss": 1.059,
      "step": 1555
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7055155038833618,
      "learning_rate": 9.22772350602464e-06,
      "loss": 0.8367,
      "step": 1556
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5178554058074951,
      "learning_rate": 9.226590113335065e-06,
      "loss": 0.8429,
      "step": 1557
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5196095705032349,
      "learning_rate": 9.225455959282592e-06,
      "loss": 0.8601,
      "step": 1558
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4135102033615112,
      "learning_rate": 9.224321044071522e-06,
      "loss": 0.8025,
      "step": 1559
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3387326002120972,
      "learning_rate": 9.223185367906293e-06,
      "loss": 0.7768,
      "step": 1560
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.4143495559692383,
      "learning_rate": 9.222048930991485e-06,
      "loss": 1.1109,
      "step": 1561
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3985000848770142,
      "learning_rate": 9.22091173353181e-06,
      "loss": 0.8182,
      "step": 1562
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.417245864868164,
      "learning_rate": 9.21977377573212e-06,
      "loss": 0.781,
      "step": 1563
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5144131183624268,
      "learning_rate": 9.218635057797399e-06,
      "loss": 0.8517,
      "step": 1564
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3292292356491089,
      "learning_rate": 9.217495579932775e-06,
      "loss": 0.7898,
      "step": 1565
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.337173342704773,
      "learning_rate": 9.216355342343508e-06,
      "loss": 0.8536,
      "step": 1566
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3725868463516235,
      "learning_rate": 9.215214345234998e-06,
      "loss": 0.774,
      "step": 1567
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5862005949020386,
      "learning_rate": 9.214072588812779e-06,
      "loss": 1.1013,
      "step": 1568
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.430631637573242,
      "learning_rate": 9.21293007328252e-06,
      "loss": 0.8295,
      "step": 1569
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4040148258209229,
      "learning_rate": 9.211786798850036e-06,
      "loss": 0.8743,
      "step": 1570
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.392773985862732,
      "learning_rate": 9.210642765721267e-06,
      "loss": 0.8117,
      "step": 1571
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8718588352203369,
      "learning_rate": 9.209497974102298e-06,
      "loss": 1.0645,
      "step": 1572
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3504111766815186,
      "learning_rate": 9.208352424199347e-06,
      "loss": 0.8973,
      "step": 1573
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2234197854995728,
      "learning_rate": 9.207206116218769e-06,
      "loss": 0.8518,
      "step": 1574
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8603667616844177,
      "learning_rate": 9.206059050367055e-06,
      "loss": 1.067,
      "step": 1575
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.838808000087738,
      "learning_rate": 9.204911226850835e-06,
      "loss": 1.0632,
      "step": 1576
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3785065412521362,
      "learning_rate": 9.203762645876872e-06,
      "loss": 0.8482,
      "step": 1577
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3343671560287476,
      "learning_rate": 9.202613307652068e-06,
      "loss": 0.8144,
      "step": 1578
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3553016185760498,
      "learning_rate": 9.201463212383463e-06,
      "loss": 0.847,
      "step": 1579
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3182547092437744,
      "learning_rate": 9.200312360278227e-06,
      "loss": 0.8337,
      "step": 1580
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3028277158737183,
      "learning_rate": 9.199160751543674e-06,
      "loss": 0.8186,
      "step": 1581
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3206290006637573,
      "learning_rate": 9.19800838638725e-06,
      "loss": 0.8672,
      "step": 1582
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9969867467880249,
      "learning_rate": 9.196855265016535e-06,
      "loss": 1.0843,
      "step": 1583
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8693169951438904,
      "learning_rate": 9.195701387639253e-06,
      "loss": 1.0718,
      "step": 1584
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6626013517379761,
      "learning_rate": 9.194546754463255e-06,
      "loss": 1.0977,
      "step": 1585
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3658084869384766,
      "learning_rate": 9.193391365696534e-06,
      "loss": 0.8236,
      "step": 1586
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4736109972000122,
      "learning_rate": 9.192235221547218e-06,
      "loss": 0.8702,
      "step": 1587
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.006007432937622,
      "learning_rate": 9.191078322223572e-06,
      "loss": 1.0616,
      "step": 1588
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3091375827789307,
      "learning_rate": 9.189920667933991e-06,
      "loss": 0.857,
      "step": 1589
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2987589836120605,
      "learning_rate": 9.188762258887016e-06,
      "loss": 0.8158,
      "step": 1590
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8392942547798157,
      "learning_rate": 9.187603095291318e-06,
      "loss": 1.083,
      "step": 1591
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.421292781829834,
      "learning_rate": 9.186443177355702e-06,
      "loss": 0.8316,
      "step": 1592
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2648431062698364,
      "learning_rate": 9.185282505289113e-06,
      "loss": 0.8907,
      "step": 1593
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7628222107887268,
      "learning_rate": 9.184121079300629e-06,
      "loss": 1.0857,
      "step": 1594
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3018890619277954,
      "learning_rate": 9.182958899599466e-06,
      "loss": 0.7928,
      "step": 1595
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3708631992340088,
      "learning_rate": 9.181795966394977e-06,
      "loss": 0.8381,
      "step": 1596
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2428743839263916,
      "learning_rate": 9.180632279896646e-06,
      "loss": 0.8354,
      "step": 1597
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3519108295440674,
      "learning_rate": 9.179467840314096e-06,
      "loss": 0.8937,
      "step": 1598
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2739890813827515,
      "learning_rate": 9.178302647857085e-06,
      "loss": 0.8225,
      "step": 1599
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3059591054916382,
      "learning_rate": 9.177136702735507e-06,
      "loss": 0.8053,
      "step": 1600
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2837448120117188,
      "learning_rate": 9.17597000515939e-06,
      "loss": 0.7916,
      "step": 1601
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3311065435409546,
      "learning_rate": 9.174802555338901e-06,
      "loss": 0.8708,
      "step": 1602
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0707241296768188,
      "learning_rate": 9.17363435348434e-06,
      "loss": 1.08,
      "step": 1603
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3218801021575928,
      "learning_rate": 9.172465399806141e-06,
      "loss": 0.9074,
      "step": 1604
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2969094514846802,
      "learning_rate": 9.171295694514875e-06,
      "loss": 0.8397,
      "step": 1605
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.331737995147705,
      "learning_rate": 9.17012523782125e-06,
      "loss": 0.8087,
      "step": 1606
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8058100342750549,
      "learning_rate": 9.168954029936105e-06,
      "loss": 1.1067,
      "step": 1607
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2861045598983765,
      "learning_rate": 9.167782071070423e-06,
      "loss": 0.7835,
      "step": 1608
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.315932273864746,
      "learning_rate": 9.16660936143531e-06,
      "loss": 0.8583,
      "step": 1609
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3216676712036133,
      "learning_rate": 9.165435901242019e-06,
      "loss": 0.8264,
      "step": 1610
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3865629434585571,
      "learning_rate": 9.16426169070193e-06,
      "loss": 0.8678,
      "step": 1611
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8983009457588196,
      "learning_rate": 9.163086730026562e-06,
      "loss": 1.0828,
      "step": 1612
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2357666492462158,
      "learning_rate": 9.16191101942757e-06,
      "loss": 0.8077,
      "step": 1613
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.254167914390564,
      "learning_rate": 9.160734559116738e-06,
      "loss": 0.801,
      "step": 1614
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.31859290599823,
      "learning_rate": 9.159557349305991e-06,
      "loss": 0.8741,
      "step": 1615
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.38735032081604,
      "learning_rate": 9.158379390207389e-06,
      "loss": 0.8346,
      "step": 1616
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.340172529220581,
      "learning_rate": 9.157200682033126e-06,
      "loss": 0.921,
      "step": 1617
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2785874605178833,
      "learning_rate": 9.156021224995526e-06,
      "loss": 0.8089,
      "step": 1618
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2908167839050293,
      "learning_rate": 9.154841019307057e-06,
      "loss": 0.7954,
      "step": 1619
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2360799312591553,
      "learning_rate": 9.153660065180314e-06,
      "loss": 0.7891,
      "step": 1620
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3226290941238403,
      "learning_rate": 9.15247836282803e-06,
      "loss": 0.8112,
      "step": 1621
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.299823522567749,
      "learning_rate": 9.151295912463076e-06,
      "loss": 0.8696,
      "step": 1622
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9990823864936829,
      "learning_rate": 9.150112714298451e-06,
      "loss": 1.0824,
      "step": 1623
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2819373607635498,
      "learning_rate": 9.148928768547292e-06,
      "loss": 0.8439,
      "step": 1624
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2980642318725586,
      "learning_rate": 9.147744075422874e-06,
      "loss": 0.83,
      "step": 1625
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.276577115058899,
      "learning_rate": 9.1465586351386e-06,
      "loss": 0.872,
      "step": 1626
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2838248014450073,
      "learning_rate": 9.145372447908011e-06,
      "loss": 0.7822,
      "step": 1627
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8009050488471985,
      "learning_rate": 9.144185513944785e-06,
      "loss": 1.0713,
      "step": 1628
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4029572010040283,
      "learning_rate": 9.142997833462733e-06,
      "loss": 0.7795,
      "step": 1629
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7254451513290405,
      "learning_rate": 9.141809406675794e-06,
      "loss": 1.0989,
      "step": 1630
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2574540376663208,
      "learning_rate": 9.140620233798053e-06,
      "loss": 0.7956,
      "step": 1631
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.700102686882019,
      "learning_rate": 9.139430315043718e-06,
      "loss": 1.0587,
      "step": 1632
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3034871816635132,
      "learning_rate": 9.13823965062714e-06,
      "loss": 0.8795,
      "step": 1633
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2733728885650635,
      "learning_rate": 9.137048240762802e-06,
      "loss": 0.8261,
      "step": 1634
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8395592570304871,
      "learning_rate": 9.135856085665317e-06,
      "loss": 1.0761,
      "step": 1635
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3518072366714478,
      "learning_rate": 9.13466318554944e-06,
      "loss": 0.8745,
      "step": 1636
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.339910626411438,
      "learning_rate": 9.133469540630051e-06,
      "loss": 0.7878,
      "step": 1637
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7382156252861023,
      "learning_rate": 9.132275151122173e-06,
      "loss": 1.0771,
      "step": 1638
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6888092756271362,
      "learning_rate": 9.131080017240956e-06,
      "loss": 1.0755,
      "step": 1639
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6264774203300476,
      "learning_rate": 9.129884139201688e-06,
      "loss": 1.0698,
      "step": 1640
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.310363531112671,
      "learning_rate": 9.12868751721979e-06,
      "loss": 0.8344,
      "step": 1641
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.346718192100525,
      "learning_rate": 9.127490151510819e-06,
      "loss": 0.8722,
      "step": 1642
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2730985879898071,
      "learning_rate": 9.126292042290464e-06,
      "loss": 0.8624,
      "step": 1643
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.262601375579834,
      "learning_rate": 9.125093189774545e-06,
      "loss": 0.8774,
      "step": 1644
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2227261066436768,
      "learning_rate": 9.123893594179022e-06,
      "loss": 0.7852,
      "step": 1645
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4359707832336426,
      "learning_rate": 9.122693255719986e-06,
      "loss": 0.8628,
      "step": 1646
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2857179641723633,
      "learning_rate": 9.121492174613659e-06,
      "loss": 0.8417,
      "step": 1647
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2717152833938599,
      "learning_rate": 9.120290351076402e-06,
      "loss": 0.8518,
      "step": 1648
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.337329626083374,
      "learning_rate": 9.119087785324708e-06,
      "loss": 0.8677,
      "step": 1649
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0050544738769531,
      "learning_rate": 9.117884477575198e-06,
      "loss": 1.0475,
      "step": 1650
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3908902406692505,
      "learning_rate": 9.116680428044638e-06,
      "loss": 0.8733,
      "step": 1651
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6879110336303711,
      "learning_rate": 9.115475636949916e-06,
      "loss": 1.0798,
      "step": 1652
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3787531852722168,
      "learning_rate": 9.114270104508063e-06,
      "loss": 0.9117,
      "step": 1653
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3510069847106934,
      "learning_rate": 9.113063830936234e-06,
      "loss": 0.7742,
      "step": 1654
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.254549503326416,
      "learning_rate": 9.11185681645173e-06,
      "loss": 0.8088,
      "step": 1655
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3290501832962036,
      "learning_rate": 9.110649061271971e-06,
      "loss": 0.7703,
      "step": 1656
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3232454061508179,
      "learning_rate": 9.109440565614522e-06,
      "loss": 0.835,
      "step": 1657
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0296642780303955,
      "learning_rate": 9.108231329697077e-06,
      "loss": 1.0975,
      "step": 1658
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3471013307571411,
      "learning_rate": 9.107021353737463e-06,
      "loss": 0.853,
      "step": 1659
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4085060358047485,
      "learning_rate": 9.10581063795364e-06,
      "loss": 0.8165,
      "step": 1660
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6634822487831116,
      "learning_rate": 9.104599182563701e-06,
      "loss": 1.0932,
      "step": 1661
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2943519353866577,
      "learning_rate": 9.103386987785876e-06,
      "loss": 0.8701,
      "step": 1662
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3612545728683472,
      "learning_rate": 9.102174053838523e-06,
      "loss": 0.8272,
      "step": 1663
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8419903516769409,
      "learning_rate": 9.100960380940138e-06,
      "loss": 1.0913,
      "step": 1664
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3374271392822266,
      "learning_rate": 9.099745969309346e-06,
      "loss": 0.8764,
      "step": 1665
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2605600357055664,
      "learning_rate": 9.098530819164907e-06,
      "loss": 0.8538,
      "step": 1666
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2952282428741455,
      "learning_rate": 9.097314930725717e-06,
      "loss": 0.8223,
      "step": 1667
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3460195064544678,
      "learning_rate": 9.096098304210794e-06,
      "loss": 0.9212,
      "step": 1668
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2948963642120361,
      "learning_rate": 9.094880939839303e-06,
      "loss": 0.8419,
      "step": 1669
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3245099782943726,
      "learning_rate": 9.093662837830537e-06,
      "loss": 0.9047,
      "step": 1670
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2853890657424927,
      "learning_rate": 9.092443998403915e-06,
      "loss": 0.78,
      "step": 1671
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8150638937950134,
      "learning_rate": 9.091224421779e-06,
      "loss": 1.0669,
      "step": 1672
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2837711572647095,
      "learning_rate": 9.090004108175476e-06,
      "loss": 0.8453,
      "step": 1673
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1872047185897827,
      "learning_rate": 9.088783057813172e-06,
      "loss": 0.8229,
      "step": 1674
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3216646909713745,
      "learning_rate": 9.08756127091204e-06,
      "loss": 0.8383,
      "step": 1675
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2537134885787964,
      "learning_rate": 9.08633874769217e-06,
      "loss": 0.7667,
      "step": 1676
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2662851810455322,
      "learning_rate": 9.085115488373783e-06,
      "loss": 0.7675,
      "step": 1677
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7956417798995972,
      "learning_rate": 9.08389149317723e-06,
      "loss": 1.0755,
      "step": 1678
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.270416498184204,
      "learning_rate": 9.082666762323e-06,
      "loss": 0.8154,
      "step": 1679
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2998617887496948,
      "learning_rate": 9.081441296031713e-06,
      "loss": 0.8073,
      "step": 1680
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2666257619857788,
      "learning_rate": 9.080215094524116e-06,
      "loss": 0.815,
      "step": 1681
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2061070203781128,
      "learning_rate": 9.078988158021094e-06,
      "loss": 0.7855,
      "step": 1682
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3588433265686035,
      "learning_rate": 9.077760486743663e-06,
      "loss": 0.8584,
      "step": 1683
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7692854404449463,
      "learning_rate": 9.076532080912974e-06,
      "loss": 1.0892,
      "step": 1684
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3115777969360352,
      "learning_rate": 9.075302940750304e-06,
      "loss": 0.8493,
      "step": 1685
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1996254920959473,
      "learning_rate": 9.074073066477068e-06,
      "loss": 0.7934,
      "step": 1686
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3353276252746582,
      "learning_rate": 9.07284245831481e-06,
      "loss": 0.8361,
      "step": 1687
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3345648050308228,
      "learning_rate": 9.071611116485209e-06,
      "loss": 0.7994,
      "step": 1688
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2686827182769775,
      "learning_rate": 9.070379041210072e-06,
      "loss": 0.8128,
      "step": 1689
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2637009620666504,
      "learning_rate": 9.069146232711343e-06,
      "loss": 0.7967,
      "step": 1690
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4359453916549683,
      "learning_rate": 9.067912691211095e-06,
      "loss": 0.8282,
      "step": 1691
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3536320924758911,
      "learning_rate": 9.066678416931533e-06,
      "loss": 0.8627,
      "step": 1692
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3026608228683472,
      "learning_rate": 9.065443410094997e-06,
      "loss": 0.8229,
      "step": 1693
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2339284420013428,
      "learning_rate": 9.064207670923953e-06,
      "loss": 0.7838,
      "step": 1694
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3222103118896484,
      "learning_rate": 9.062971199641007e-06,
      "loss": 0.8066,
      "step": 1695
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3282561302185059,
      "learning_rate": 9.061733996468889e-06,
      "loss": 0.8357,
      "step": 1696
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3641444444656372,
      "learning_rate": 9.060496061630466e-06,
      "loss": 0.8332,
      "step": 1697
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9179030656814575,
      "learning_rate": 9.059257395348736e-06,
      "loss": 1.0728,
      "step": 1698
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.32345712184906,
      "learning_rate": 9.058017997846825e-06,
      "loss": 0.8038,
      "step": 1699
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.287960410118103,
      "learning_rate": 9.056777869347995e-06,
      "loss": 0.8301,
      "step": 1700
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1858842372894287,
      "learning_rate": 9.05553701007564e-06,
      "loss": 0.7907,
      "step": 1701
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2855322360992432,
      "learning_rate": 9.054295420253283e-06,
      "loss": 0.8123,
      "step": 1702
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3480548858642578,
      "learning_rate": 9.05305310010458e-06,
      "loss": 0.8873,
      "step": 1703
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0211046934127808,
      "learning_rate": 9.051810049853316e-06,
      "loss": 1.0705,
      "step": 1704
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8556668758392334,
      "learning_rate": 9.050566269723412e-06,
      "loss": 1.0825,
      "step": 1705
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.325930118560791,
      "learning_rate": 9.049321759938918e-06,
      "loss": 0.8584,
      "step": 1706
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.412535309791565,
      "learning_rate": 9.048076520724014e-06,
      "loss": 0.8117,
      "step": 1707
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3565036058425903,
      "learning_rate": 9.046830552303012e-06,
      "loss": 0.8217,
      "step": 1708
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3121122121810913,
      "learning_rate": 9.04558385490036e-06,
      "loss": 0.8286,
      "step": 1709
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5725983381271362,
      "learning_rate": 9.044336428740632e-06,
      "loss": 1.0678,
      "step": 1710
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.431628942489624,
      "learning_rate": 9.043088274048535e-06,
      "loss": 0.8758,
      "step": 1711
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3745734691619873,
      "learning_rate": 9.041839391048905e-06,
      "loss": 0.8405,
      "step": 1712
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4655344486236572,
      "learning_rate": 9.040589779966716e-06,
      "loss": 0.8478,
      "step": 1713
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3494564294815063,
      "learning_rate": 9.039339441027064e-06,
      "loss": 0.8479,
      "step": 1714
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4846928119659424,
      "learning_rate": 9.038088374455181e-06,
      "loss": 0.8596,
      "step": 1715
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.268207311630249,
      "learning_rate": 9.036836580476432e-06,
      "loss": 0.8019,
      "step": 1716
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3779544830322266,
      "learning_rate": 9.03558405931631e-06,
      "loss": 0.8342,
      "step": 1717
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3566651344299316,
      "learning_rate": 9.034330811200438e-06,
      "loss": 0.8592,
      "step": 1718
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3760175704956055,
      "learning_rate": 9.033076836354575e-06,
      "loss": 0.8487,
      "step": 1719
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2966620922088623,
      "learning_rate": 9.031822135004604e-06,
      "loss": 0.8278,
      "step": 1720
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3460150957107544,
      "learning_rate": 9.030566707376545e-06,
      "loss": 0.8586,
      "step": 1721
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8710951209068298,
      "learning_rate": 9.029310553696545e-06,
      "loss": 1.0785,
      "step": 1722
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3235244750976562,
      "learning_rate": 9.028053674190883e-06,
      "loss": 0.8437,
      "step": 1723
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3185927867889404,
      "learning_rate": 9.02679606908597e-06,
      "loss": 0.7824,
      "step": 1724
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4087855815887451,
      "learning_rate": 9.025537738608344e-06,
      "loss": 0.829,
      "step": 1725
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4699827432632446,
      "learning_rate": 9.024278682984679e-06,
      "loss": 0.8215,
      "step": 1726
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.358825922012329,
      "learning_rate": 9.023018902441774e-06,
      "loss": 0.8357,
      "step": 1727
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.263055682182312,
      "learning_rate": 9.021758397206565e-06,
      "loss": 0.8704,
      "step": 1728
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3247886896133423,
      "learning_rate": 9.020497167506113e-06,
      "loss": 0.8386,
      "step": 1729
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2273277044296265,
      "learning_rate": 9.01923521356761e-06,
      "loss": 0.8312,
      "step": 1730
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2800706624984741,
      "learning_rate": 9.017972535618381e-06,
      "loss": 0.8531,
      "step": 1731
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5090426206588745,
      "learning_rate": 9.01670913388588e-06,
      "loss": 0.8171,
      "step": 1732
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3266403675079346,
      "learning_rate": 9.015445008597694e-06,
      "loss": 0.8343,
      "step": 1733
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.28498375415802,
      "learning_rate": 9.014180159981536e-06,
      "loss": 0.8233,
      "step": 1734
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3129730224609375,
      "learning_rate": 9.01291458826525e-06,
      "loss": 0.8969,
      "step": 1735
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2836819887161255,
      "learning_rate": 9.011648293676816e-06,
      "loss": 0.7976,
      "step": 1736
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2378840446472168,
      "learning_rate": 9.010381276444336e-06,
      "loss": 0.8554,
      "step": 1737
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3313651084899902,
      "learning_rate": 9.009113536796046e-06,
      "loss": 0.8701,
      "step": 1738
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2948133945465088,
      "learning_rate": 9.007845074960314e-06,
      "loss": 0.8274,
      "step": 1739
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1136599779129028,
      "learning_rate": 9.006575891165633e-06,
      "loss": 1.0824,
      "step": 1740
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8363925814628601,
      "learning_rate": 9.005305985640633e-06,
      "loss": 1.0826,
      "step": 1741
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6667888760566711,
      "learning_rate": 9.00403535861407e-06,
      "loss": 1.0614,
      "step": 1742
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.058013916015625,
      "learning_rate": 9.002764010314828e-06,
      "loss": 0.8656,
      "step": 1743
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1270629167556763,
      "learning_rate": 9.001491940971922e-06,
      "loss": 1.0746,
      "step": 1744
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4220093488693237,
      "learning_rate": 9.000219150814504e-06,
      "loss": 0.8049,
      "step": 1745
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4682503938674927,
      "learning_rate": 8.998945640071844e-06,
      "loss": 0.857,
      "step": 1746
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3392730951309204,
      "learning_rate": 8.99767140897335e-06,
      "loss": 0.8104,
      "step": 1747
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9308080077171326,
      "learning_rate": 8.996396457748556e-06,
      "loss": 1.0897,
      "step": 1748
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4177066087722778,
      "learning_rate": 8.995120786627129e-06,
      "loss": 0.85,
      "step": 1749
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7723240256309509,
      "learning_rate": 8.993844395838865e-06,
      "loss": 1.0545,
      "step": 1750
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3813790082931519,
      "learning_rate": 8.992567285613686e-06,
      "loss": 0.7992,
      "step": 1751
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2909537553787231,
      "learning_rate": 8.991289456181647e-06,
      "loss": 0.7359,
      "step": 1752
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8249996304512024,
      "learning_rate": 8.99001090777293e-06,
      "loss": 1.0558,
      "step": 1753
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4248733520507812,
      "learning_rate": 8.98873164061785e-06,
      "loss": 0.8101,
      "step": 1754
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.315940260887146,
      "learning_rate": 8.98745165494685e-06,
      "loss": 0.8271,
      "step": 1755
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2560594081878662,
      "learning_rate": 8.986170950990502e-06,
      "loss": 0.7937,
      "step": 1756
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3193650245666504,
      "learning_rate": 8.984889528979505e-06,
      "loss": 0.832,
      "step": 1757
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.278759479522705,
      "learning_rate": 8.983607389144692e-06,
      "loss": 0.7731,
      "step": 1758
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3566937446594238,
      "learning_rate": 8.982324531717024e-06,
      "loss": 0.8933,
      "step": 1759
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3139305114746094,
      "learning_rate": 8.981040956927588e-06,
      "loss": 0.7816,
      "step": 1760
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3401787281036377,
      "learning_rate": 8.979756665007604e-06,
      "loss": 0.848,
      "step": 1761
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3780409097671509,
      "learning_rate": 8.97847165618842e-06,
      "loss": 0.8338,
      "step": 1762
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3569073677062988,
      "learning_rate": 8.977185930701513e-06,
      "loss": 0.8312,
      "step": 1763
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.278976321220398,
      "learning_rate": 8.975899488778487e-06,
      "loss": 0.8256,
      "step": 1764
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4656397104263306,
      "learning_rate": 8.974612330651078e-06,
      "loss": 0.825,
      "step": 1765
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.280103325843811,
      "learning_rate": 8.973324456551152e-06,
      "loss": 0.8689,
      "step": 1766
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3418707847595215,
      "learning_rate": 8.972035866710699e-06,
      "loss": 0.8408,
      "step": 1767
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2647663354873657,
      "learning_rate": 8.970746561361842e-06,
      "loss": 0.8918,
      "step": 1768
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2592475414276123,
      "learning_rate": 8.969456540736835e-06,
      "loss": 0.7953,
      "step": 1769
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4275890588760376,
      "learning_rate": 8.968165805068054e-06,
      "loss": 0.8888,
      "step": 1770
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.055721402168274,
      "learning_rate": 8.966874354588008e-06,
      "loss": 1.0787,
      "step": 1771
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2675570249557495,
      "learning_rate": 8.965582189529336e-06,
      "loss": 0.8538,
      "step": 1772
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2863456010818481,
      "learning_rate": 8.964289310124803e-06,
      "loss": 0.8597,
      "step": 1773
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2627497911453247,
      "learning_rate": 8.962995716607304e-06,
      "loss": 0.8568,
      "step": 1774
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2655525207519531,
      "learning_rate": 8.96170140920986e-06,
      "loss": 0.8804,
      "step": 1775
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2787314653396606,
      "learning_rate": 8.960406388165626e-06,
      "loss": 0.8672,
      "step": 1776
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.224528193473816,
      "learning_rate": 8.959110653707883e-06,
      "loss": 0.8316,
      "step": 1777
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3364564180374146,
      "learning_rate": 8.957814206070037e-06,
      "loss": 0.8427,
      "step": 1778
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1505173444747925,
      "learning_rate": 8.956517045485626e-06,
      "loss": 1.0791,
      "step": 1779
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3472164869308472,
      "learning_rate": 8.955219172188319e-06,
      "loss": 0.9054,
      "step": 1780
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.35947585105896,
      "learning_rate": 8.953920586411906e-06,
      "loss": 0.8132,
      "step": 1781
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7066414952278137,
      "learning_rate": 8.952621288390313e-06,
      "loss": 1.0702,
      "step": 1782
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7370069622993469,
      "learning_rate": 8.951321278357589e-06,
      "loss": 1.0677,
      "step": 1783
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6908316016197205,
      "learning_rate": 8.950020556547913e-06,
      "loss": 1.0533,
      "step": 1784
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.34430730342865,
      "learning_rate": 8.948719123195596e-06,
      "loss": 0.778,
      "step": 1785
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3646339178085327,
      "learning_rate": 8.947416978535069e-06,
      "loss": 0.8135,
      "step": 1786
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6448179483413696,
      "learning_rate": 8.946114122800896e-06,
      "loss": 0.8384,
      "step": 1787
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8671547770500183,
      "learning_rate": 8.944810556227771e-06,
      "loss": 1.0642,
      "step": 1788
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2924929857254028,
      "learning_rate": 8.943506279050515e-06,
      "loss": 0.8186,
      "step": 1789
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2610971927642822,
      "learning_rate": 8.942201291504071e-06,
      "loss": 0.873,
      "step": 1790
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8008129000663757,
      "learning_rate": 8.940895593823519e-06,
      "loss": 1.0685,
      "step": 1791
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.25221848487854,
      "learning_rate": 8.939589186244063e-06,
      "loss": 0.7834,
      "step": 1792
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6913357973098755,
      "learning_rate": 8.938282069001031e-06,
      "loss": 1.0905,
      "step": 1793
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6947531700134277,
      "learning_rate": 8.936974242329887e-06,
      "loss": 1.0658,
      "step": 1794
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2623811960220337,
      "learning_rate": 8.935665706466214e-06,
      "loss": 0.7831,
      "step": 1795
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6775804758071899,
      "learning_rate": 8.93435646164573e-06,
      "loss": 1.0579,
      "step": 1796
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.405082106590271,
      "learning_rate": 8.933046508104275e-06,
      "loss": 0.8845,
      "step": 1797
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2464795112609863,
      "learning_rate": 8.93173584607782e-06,
      "loss": 0.8325,
      "step": 1798
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2720246315002441,
      "learning_rate": 8.930424475802466e-06,
      "loss": 0.8356,
      "step": 1799
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1950538158416748,
      "learning_rate": 8.929112397514436e-06,
      "loss": 0.8591,
      "step": 1800
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2555311918258667,
      "learning_rate": 8.927799611450083e-06,
      "loss": 0.7882,
      "step": 1801
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8036953806877136,
      "learning_rate": 8.926486117845886e-06,
      "loss": 1.0824,
      "step": 1802
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2572015523910522,
      "learning_rate": 8.925171916938459e-06,
      "loss": 0.8396,
      "step": 1803
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3007457256317139,
      "learning_rate": 8.923857008964534e-06,
      "loss": 0.7377,
      "step": 1804
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.211191177368164,
      "learning_rate": 8.92254139416097e-06,
      "loss": 0.7898,
      "step": 1805
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2978363037109375,
      "learning_rate": 8.921225072764764e-06,
      "loss": 0.8362,
      "step": 1806
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2407885789871216,
      "learning_rate": 8.919908045013028e-06,
      "loss": 0.7837,
      "step": 1807
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.23442542552948,
      "learning_rate": 8.91859031114301e-06,
      "loss": 0.8191,
      "step": 1808
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2953159809112549,
      "learning_rate": 8.917271871392079e-06,
      "loss": 0.8358,
      "step": 1809
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7646375894546509,
      "learning_rate": 8.915952725997736e-06,
      "loss": 1.0772,
      "step": 1810
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3052918910980225,
      "learning_rate": 8.914632875197608e-06,
      "loss": 0.8166,
      "step": 1811
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.629289448261261,
      "learning_rate": 8.913312319229448e-06,
      "loss": 1.0788,
      "step": 1812
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6812511086463928,
      "learning_rate": 8.911991058331133e-06,
      "loss": 1.0807,
      "step": 1813
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2895184755325317,
      "learning_rate": 8.910669092740675e-06,
      "loss": 0.8134,
      "step": 1814
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2914799451828003,
      "learning_rate": 8.909346422696204e-06,
      "loss": 0.8147,
      "step": 1815
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.323153018951416,
      "learning_rate": 8.908023048435983e-06,
      "loss": 0.8964,
      "step": 1816
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3081449270248413,
      "learning_rate": 8.906698970198402e-06,
      "loss": 0.8282,
      "step": 1817
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8676469922065735,
      "learning_rate": 8.905374188221971e-06,
      "loss": 1.0794,
      "step": 1818
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7142981886863708,
      "learning_rate": 8.904048702745336e-06,
      "loss": 1.0698,
      "step": 1819
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.669460654258728,
      "learning_rate": 8.902722514007262e-06,
      "loss": 1.0687,
      "step": 1820
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2834211587905884,
      "learning_rate": 8.901395622246645e-06,
      "loss": 0.8486,
      "step": 1821
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8074416518211365,
      "learning_rate": 8.900068027702508e-06,
      "loss": 1.0587,
      "step": 1822
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2293150424957275,
      "learning_rate": 8.898739730613996e-06,
      "loss": 0.872,
      "step": 1823
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3223490715026855,
      "learning_rate": 8.897410731220388e-06,
      "loss": 0.8247,
      "step": 1824
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2900617122650146,
      "learning_rate": 8.896081029761081e-06,
      "loss": 0.795,
      "step": 1825
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3157522678375244,
      "learning_rate": 8.894750626475605e-06,
      "loss": 0.8054,
      "step": 1826
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2358001470565796,
      "learning_rate": 8.893419521603615e-06,
      "loss": 0.7896,
      "step": 1827
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2543447017669678,
      "learning_rate": 8.892087715384888e-06,
      "loss": 0.7979,
      "step": 1828
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.224571704864502,
      "learning_rate": 8.890755208059334e-06,
      "loss": 0.8668,
      "step": 1829
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3232955932617188,
      "learning_rate": 8.889421999866985e-06,
      "loss": 0.8315,
      "step": 1830
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2729510068893433,
      "learning_rate": 8.888088091048001e-06,
      "loss": 0.8825,
      "step": 1831
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3961125612258911,
      "learning_rate": 8.886753481842667e-06,
      "loss": 0.8295,
      "step": 1832
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0144357681274414,
      "learning_rate": 8.885418172491396e-06,
      "loss": 1.0925,
      "step": 1833
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3165770769119263,
      "learning_rate": 8.884082163234722e-06,
      "loss": 0.8082,
      "step": 1834
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3909043073654175,
      "learning_rate": 8.882745454313313e-06,
      "loss": 0.8448,
      "step": 1835
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6949319243431091,
      "learning_rate": 8.88140804596796e-06,
      "loss": 1.0641,
      "step": 1836
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3949224948883057,
      "learning_rate": 8.880069938439573e-06,
      "loss": 0.8357,
      "step": 1837
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3995925188064575,
      "learning_rate": 8.878731131969199e-06,
      "loss": 0.8413,
      "step": 1838
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3563599586486816,
      "learning_rate": 8.877391626798004e-06,
      "loss": 0.8588,
      "step": 1839
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3058009147644043,
      "learning_rate": 8.876051423167281e-06,
      "loss": 0.8655,
      "step": 1840
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7900117635726929,
      "learning_rate": 8.874710521318452e-06,
      "loss": 1.0882,
      "step": 1841
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.338544249534607,
      "learning_rate": 8.87336892149306e-06,
      "loss": 0.8152,
      "step": 1842
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2770298719406128,
      "learning_rate": 8.872026623932779e-06,
      "loss": 0.7759,
      "step": 1843
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.257590413093567,
      "learning_rate": 8.870683628879401e-06,
      "loss": 0.7813,
      "step": 1844
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7401040196418762,
      "learning_rate": 8.869339936574852e-06,
      "loss": 1.0789,
      "step": 1845
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2621629238128662,
      "learning_rate": 8.867995547261179e-06,
      "loss": 0.8402,
      "step": 1846
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2242428064346313,
      "learning_rate": 8.866650461180557e-06,
      "loss": 0.8573,
      "step": 1847
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6852111220359802,
      "learning_rate": 8.865304678575281e-06,
      "loss": 1.0499,
      "step": 1848
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7672492265701294,
      "learning_rate": 8.863958199687779e-06,
      "loss": 0.8697,
      "step": 1849
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4097450971603394,
      "learning_rate": 8.862611024760598e-06,
      "loss": 0.8413,
      "step": 1850
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3103914260864258,
      "learning_rate": 8.861263154036417e-06,
      "loss": 0.8557,
      "step": 1851
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.392441987991333,
      "learning_rate": 8.859914587758035e-06,
      "loss": 0.8306,
      "step": 1852
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7532827258110046,
      "learning_rate": 8.858565326168377e-06,
      "loss": 1.0862,
      "step": 1853
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6912055015563965,
      "learning_rate": 8.857215369510496e-06,
      "loss": 1.0733,
      "step": 1854
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4520113468170166,
      "learning_rate": 8.855864718027565e-06,
      "loss": 0.9174,
      "step": 1855
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3088150024414062,
      "learning_rate": 8.85451337196289e-06,
      "loss": 0.8797,
      "step": 1856
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.206411600112915,
      "learning_rate": 8.853161331559894e-06,
      "loss": 0.8401,
      "step": 1857
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3184003829956055,
      "learning_rate": 8.851808597062131e-06,
      "loss": 0.8535,
      "step": 1858
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2378976345062256,
      "learning_rate": 8.850455168713277e-06,
      "loss": 0.8197,
      "step": 1859
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3691576719284058,
      "learning_rate": 8.849101046757134e-06,
      "loss": 0.8292,
      "step": 1860
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4103931188583374,
      "learning_rate": 8.847746231437629e-06,
      "loss": 0.8888,
      "step": 1861
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.460412859916687,
      "learning_rate": 8.84639072299881e-06,
      "loss": 0.8646,
      "step": 1862
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.203593373298645,
      "learning_rate": 8.845034521684859e-06,
      "loss": 1.0765,
      "step": 1863
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3724534511566162,
      "learning_rate": 8.843677627740075e-06,
      "loss": 0.8668,
      "step": 1864
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2980530261993408,
      "learning_rate": 8.842320041408882e-06,
      "loss": 0.7906,
      "step": 1865
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7067030072212219,
      "learning_rate": 8.840961762935836e-06,
      "loss": 1.0732,
      "step": 1866
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3572189807891846,
      "learning_rate": 8.839602792565606e-06,
      "loss": 0.7857,
      "step": 1867
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3711708784103394,
      "learning_rate": 8.838243130542995e-06,
      "loss": 0.8379,
      "step": 1868
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3372154235839844,
      "learning_rate": 8.83688277711293e-06,
      "loss": 0.7886,
      "step": 1869
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.946338415145874,
      "learning_rate": 8.835521732520459e-06,
      "loss": 1.0962,
      "step": 1870
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7244883179664612,
      "learning_rate": 8.834159997010751e-06,
      "loss": 1.0409,
      "step": 1871
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.376838207244873,
      "learning_rate": 8.83279757082911e-06,
      "loss": 0.8224,
      "step": 1872
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4591606855392456,
      "learning_rate": 8.831434454220957e-06,
      "loss": 0.8428,
      "step": 1873
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4095511436462402,
      "learning_rate": 8.830070647431839e-06,
      "loss": 0.833,
      "step": 1874
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2303322553634644,
      "learning_rate": 8.828706150707424e-06,
      "loss": 0.7623,
      "step": 1875
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4298441410064697,
      "learning_rate": 8.827340964293514e-06,
      "loss": 1.0716,
      "step": 1876
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3159281015396118,
      "learning_rate": 8.825975088436023e-06,
      "loss": 0.8321,
      "step": 1877
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4123843908309937,
      "learning_rate": 8.824608523380998e-06,
      "loss": 0.9064,
      "step": 1878
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.321794033050537,
      "learning_rate": 8.823241269374606e-06,
      "loss": 0.7864,
      "step": 1879
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3610743284225464,
      "learning_rate": 8.82187332666314e-06,
      "loss": 0.8424,
      "step": 1880
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7521881461143494,
      "learning_rate": 8.820504695493018e-06,
      "loss": 1.07,
      "step": 1881
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4275405406951904,
      "learning_rate": 8.819135376110776e-06,
      "loss": 0.8228,
      "step": 1882
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4291152954101562,
      "learning_rate": 8.817765368763081e-06,
      "loss": 0.8657,
      "step": 1883
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6582616567611694,
      "learning_rate": 8.816394673696722e-06,
      "loss": 1.0839,
      "step": 1884
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2855191230773926,
      "learning_rate": 8.815023291158608e-06,
      "loss": 0.8493,
      "step": 1885
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.356127142906189,
      "learning_rate": 8.813651221395779e-06,
      "loss": 0.8037,
      "step": 1886
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7313129305839539,
      "learning_rate": 8.812278464655394e-06,
      "loss": 1.0734,
      "step": 1887
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2300574779510498,
      "learning_rate": 8.810905021184732e-06,
      "loss": 0.8036,
      "step": 1888
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.283889889717102,
      "learning_rate": 8.809530891231206e-06,
      "loss": 0.8749,
      "step": 1889
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2766705751419067,
      "learning_rate": 8.808156075042343e-06,
      "loss": 0.8679,
      "step": 1890
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3103538751602173,
      "learning_rate": 8.806780572865801e-06,
      "loss": 0.8227,
      "step": 1891
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3549275398254395,
      "learning_rate": 8.805404384949354e-06,
      "loss": 0.7771,
      "step": 1892
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.834047257900238,
      "learning_rate": 8.804027511540906e-06,
      "loss": 1.0541,
      "step": 1893
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2969368696212769,
      "learning_rate": 8.802649952888483e-06,
      "loss": 0.819,
      "step": 1894
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3755488395690918,
      "learning_rate": 8.80127170924023e-06,
      "loss": 0.8747,
      "step": 1895
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6993736624717712,
      "learning_rate": 8.799892780844424e-06,
      "loss": 1.0686,
      "step": 1896
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6889926791191101,
      "learning_rate": 8.798513167949453e-06,
      "loss": 1.064,
      "step": 1897
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3558752536773682,
      "learning_rate": 8.797132870803844e-06,
      "loss": 0.834,
      "step": 1898
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2144259214401245,
      "learning_rate": 8.795751889656232e-06,
      "loss": 0.7897,
      "step": 1899
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3614393472671509,
      "learning_rate": 8.794370224755386e-06,
      "loss": 0.8315,
      "step": 1900
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2464807033538818,
      "learning_rate": 8.792987876350194e-06,
      "loss": 0.8439,
      "step": 1901
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3231302499771118,
      "learning_rate": 8.791604844689664e-06,
      "loss": 0.793,
      "step": 1902
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2510350942611694,
      "learning_rate": 8.790221130022934e-06,
      "loss": 0.8112,
      "step": 1903
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.331664800643921,
      "learning_rate": 8.78883673259926e-06,
      "loss": 0.8463,
      "step": 1904
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2944097518920898,
      "learning_rate": 8.787451652668023e-06,
      "loss": 0.857,
      "step": 1905
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1620721817016602,
      "learning_rate": 8.786065890478725e-06,
      "loss": 1.094,
      "step": 1906
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.243073582649231,
      "learning_rate": 8.784679446280994e-06,
      "loss": 0.8165,
      "step": 1907
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3834375143051147,
      "learning_rate": 8.783292320324578e-06,
      "loss": 0.8436,
      "step": 1908
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3872838020324707,
      "learning_rate": 8.781904512859348e-06,
      "loss": 0.7621,
      "step": 1909
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8414198160171509,
      "learning_rate": 8.780516024135301e-06,
      "loss": 1.0722,
      "step": 1910
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2474851608276367,
      "learning_rate": 8.779126854402554e-06,
      "loss": 0.7443,
      "step": 1911
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2861427068710327,
      "learning_rate": 8.777737003911346e-06,
      "loss": 0.8242,
      "step": 1912
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7364066243171692,
      "learning_rate": 8.776346472912039e-06,
      "loss": 1.065,
      "step": 1913
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2502706050872803,
      "learning_rate": 8.77495526165512e-06,
      "loss": 0.7873,
      "step": 1914
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6553205251693726,
      "learning_rate": 8.773563370391198e-06,
      "loss": 1.0924,
      "step": 1915
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7275654673576355,
      "learning_rate": 8.772170799371e-06,
      "loss": 1.0902,
      "step": 1916
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3121286630630493,
      "learning_rate": 8.77077754884538e-06,
      "loss": 0.8146,
      "step": 1917
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3960120677947998,
      "learning_rate": 8.769383619065315e-06,
      "loss": 0.8128,
      "step": 1918
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2589919567108154,
      "learning_rate": 8.767989010281899e-06,
      "loss": 0.7758,
      "step": 1919
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2385464906692505,
      "learning_rate": 8.766593722746355e-06,
      "loss": 0.7913,
      "step": 1920
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3459413051605225,
      "learning_rate": 8.765197756710025e-06,
      "loss": 0.7766,
      "step": 1921
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.001219630241394,
      "learning_rate": 8.763801112424372e-06,
      "loss": 1.064,
      "step": 1922
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2978930473327637,
      "learning_rate": 8.762403790140982e-06,
      "loss": 0.785,
      "step": 1923
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.300148606300354,
      "learning_rate": 8.761005790111567e-06,
      "loss": 0.8321,
      "step": 1924
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3221545219421387,
      "learning_rate": 8.759607112587956e-06,
      "loss": 0.8325,
      "step": 1925
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2772446870803833,
      "learning_rate": 8.7582077578221e-06,
      "loss": 0.7166,
      "step": 1926
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7541691660881042,
      "learning_rate": 8.756807726066073e-06,
      "loss": 1.0559,
      "step": 1927
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7653598189353943,
      "learning_rate": 8.755407017572075e-06,
      "loss": 1.0664,
      "step": 1928
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6385682225227356,
      "learning_rate": 8.754005632592425e-06,
      "loss": 1.0653,
      "step": 1929
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3316268920898438,
      "learning_rate": 8.752603571379559e-06,
      "loss": 0.7805,
      "step": 1930
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.302912950515747,
      "learning_rate": 8.751200834186044e-06,
      "loss": 0.7954,
      "step": 1931
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3635990619659424,
      "learning_rate": 8.749797421264563e-06,
      "loss": 0.8489,
      "step": 1932
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2121429443359375,
      "learning_rate": 8.748393332867919e-06,
      "loss": 0.8287,
      "step": 1933
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1429482698440552,
      "learning_rate": 8.746988569249044e-06,
      "loss": 1.0471,
      "step": 1934
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3648327589035034,
      "learning_rate": 8.745583130660983e-06,
      "loss": 0.8112,
      "step": 1935
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.307407021522522,
      "learning_rate": 8.74417701735691e-06,
      "loss": 0.8249,
      "step": 1936
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.311671495437622,
      "learning_rate": 8.742770229590113e-06,
      "loss": 0.82,
      "step": 1937
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3225477933883667,
      "learning_rate": 8.741362767614012e-06,
      "loss": 0.8327,
      "step": 1938
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8101451992988586,
      "learning_rate": 8.739954631682136e-06,
      "loss": 1.0768,
      "step": 1939
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3318837881088257,
      "learning_rate": 8.738545822048144e-06,
      "loss": 0.8393,
      "step": 1940
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6810984015464783,
      "learning_rate": 8.737136338965816e-06,
      "loss": 1.08,
      "step": 1941
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2531096935272217,
      "learning_rate": 8.735726182689048e-06,
      "loss": 0.7701,
      "step": 1942
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6861674785614014,
      "learning_rate": 8.734315353471863e-06,
      "loss": 1.0531,
      "step": 1943
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2720003128051758,
      "learning_rate": 8.732903851568401e-06,
      "loss": 0.8318,
      "step": 1944
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.331344485282898,
      "learning_rate": 8.731491677232926e-06,
      "loss": 0.7942,
      "step": 1945
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7447697520256042,
      "learning_rate": 8.73007883071982e-06,
      "loss": 1.0606,
      "step": 1946
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3585680723190308,
      "learning_rate": 8.728665312283593e-06,
      "loss": 0.8611,
      "step": 1947
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6275368332862854,
      "learning_rate": 8.727251122178866e-06,
      "loss": 1.067,
      "step": 1948
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2930104732513428,
      "learning_rate": 8.725836260660387e-06,
      "loss": 0.9008,
      "step": 1949
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2666095495224,
      "learning_rate": 8.724420727983028e-06,
      "loss": 0.8796,
      "step": 1950
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3847695589065552,
      "learning_rate": 8.723004524401774e-06,
      "loss": 0.8434,
      "step": 1951
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3032585382461548,
      "learning_rate": 8.721587650171737e-06,
      "loss": 0.8295,
      "step": 1952
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.283736228942871,
      "learning_rate": 8.720170105548145e-06,
      "loss": 0.8455,
      "step": 1953
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9150474667549133,
      "learning_rate": 8.718751890786354e-06,
      "loss": 1.0484,
      "step": 1954
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7129516005516052,
      "learning_rate": 8.717333006141831e-06,
      "loss": 1.0451,
      "step": 1955
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.640550434589386,
      "learning_rate": 8.715913451870173e-06,
      "loss": 1.0715,
      "step": 1956
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2692164182662964,
      "learning_rate": 8.71449322822709e-06,
      "loss": 0.8701,
      "step": 1957
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8572101593017578,
      "learning_rate": 8.71307233546842e-06,
      "loss": 1.0601,
      "step": 1958
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8301312923431396,
      "learning_rate": 8.711650773850114e-06,
      "loss": 1.0697,
      "step": 1959
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2234724760055542,
      "learning_rate": 8.710228543628248e-06,
      "loss": 0.756,
      "step": 1960
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3458102941513062,
      "learning_rate": 8.708805645059019e-06,
      "loss": 0.8703,
      "step": 1961
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7977344989776611,
      "learning_rate": 8.70738207839874e-06,
      "loss": 1.0612,
      "step": 1962
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.246549367904663,
      "learning_rate": 8.705957843903852e-06,
      "loss": 0.8627,
      "step": 1963
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7873830199241638,
      "learning_rate": 8.704532941830905e-06,
      "loss": 1.06,
      "step": 1964
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2131527662277222,
      "learning_rate": 8.703107372436583e-06,
      "loss": 0.8161,
      "step": 1965
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3184653520584106,
      "learning_rate": 8.701681135977678e-06,
      "loss": 0.8234,
      "step": 1966
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7708240747451782,
      "learning_rate": 8.700254232711107e-06,
      "loss": 1.0822,
      "step": 1967
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2122009992599487,
      "learning_rate": 8.69882666289391e-06,
      "loss": 0.8442,
      "step": 1968
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6856926083564758,
      "learning_rate": 8.697398426783243e-06,
      "loss": 1.0388,
      "step": 1969
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6147701740264893,
      "learning_rate": 8.695969524636383e-06,
      "loss": 1.0854,
      "step": 1970
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2734159231185913,
      "learning_rate": 8.694539956710728e-06,
      "loss": 0.7701,
      "step": 1971
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2640506029129028,
      "learning_rate": 8.693109723263794e-06,
      "loss": 0.7933,
      "step": 1972
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4286226034164429,
      "learning_rate": 8.691678824553223e-06,
      "loss": 0.8523,
      "step": 1973
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7404916882514954,
      "learning_rate": 8.690247260836765e-06,
      "loss": 1.0596,
      "step": 1974
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3080003261566162,
      "learning_rate": 8.688815032372302e-06,
      "loss": 0.8694,
      "step": 1975
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7124172449111938,
      "learning_rate": 8.687382139417827e-06,
      "loss": 1.1014,
      "step": 1976
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2998406887054443,
      "learning_rate": 8.685948582231458e-06,
      "loss": 0.8039,
      "step": 1977
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3198556900024414,
      "learning_rate": 8.68451436107143e-06,
      "loss": 0.8842,
      "step": 1978
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7443696856498718,
      "learning_rate": 8.683079476196101e-06,
      "loss": 1.0722,
      "step": 1979
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.316940188407898,
      "learning_rate": 8.681643927863944e-06,
      "loss": 0.8025,
      "step": 1980
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3400176763534546,
      "learning_rate": 8.680207716333552e-06,
      "loss": 0.8107,
      "step": 1981
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7319962382316589,
      "learning_rate": 8.678770841863644e-06,
      "loss": 1.0787,
      "step": 1982
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2734285593032837,
      "learning_rate": 8.677333304713048e-06,
      "loss": 0.8046,
      "step": 1983
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6698088049888611,
      "learning_rate": 8.67589510514072e-06,
      "loss": 1.0542,
      "step": 1984
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3064395189285278,
      "learning_rate": 8.674456243405732e-06,
      "loss": 0.796,
      "step": 1985
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6426354646682739,
      "learning_rate": 8.673016719767274e-06,
      "loss": 1.0766,
      "step": 1986
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1980904340744019,
      "learning_rate": 8.671576534484655e-06,
      "loss": 0.8009,
      "step": 1987
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2421566247940063,
      "learning_rate": 8.670135687817312e-06,
      "loss": 0.7694,
      "step": 1988
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3358467817306519,
      "learning_rate": 8.668694180024785e-06,
      "loss": 0.8294,
      "step": 1989
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8346717953681946,
      "learning_rate": 8.66725201136675e-06,
      "loss": 1.0517,
      "step": 1990
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2427446842193604,
      "learning_rate": 8.665809182102987e-06,
      "loss": 0.8535,
      "step": 1991
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2377556562423706,
      "learning_rate": 8.664365692493408e-06,
      "loss": 0.8106,
      "step": 1992
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3364076614379883,
      "learning_rate": 8.662921542798032e-06,
      "loss": 0.7929,
      "step": 1993
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.321373462677002,
      "learning_rate": 8.661476733277011e-06,
      "loss": 0.7856,
      "step": 1994
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3164507150650024,
      "learning_rate": 8.660031264190603e-06,
      "loss": 0.8532,
      "step": 1995
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2518762350082397,
      "learning_rate": 8.658585135799188e-06,
      "loss": 0.839,
      "step": 1996
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.313812017440796,
      "learning_rate": 8.65713834836327e-06,
      "loss": 0.8194,
      "step": 1997
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.358335018157959,
      "learning_rate": 8.655690902143466e-06,
      "loss": 0.833,
      "step": 1998
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2768863439559937,
      "learning_rate": 8.654242797400516e-06,
      "loss": 0.8367,
      "step": 1999
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.281152367591858,
      "learning_rate": 8.652794034395274e-06,
      "loss": 0.8333,
      "step": 2000
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3053454160690308,
      "learning_rate": 8.651344613388714e-06,
      "loss": 0.8195,
      "step": 2001
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2904586791992188,
      "learning_rate": 8.649894534641933e-06,
      "loss": 0.7689,
      "step": 2002
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8353250622749329,
      "learning_rate": 8.648443798416142e-06,
      "loss": 1.0624,
      "step": 2003
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7321689128875732,
      "learning_rate": 8.646992404972671e-06,
      "loss": 1.0845,
      "step": 2004
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2996009588241577,
      "learning_rate": 8.645540354572969e-06,
      "loss": 0.8454,
      "step": 2005
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4027513265609741,
      "learning_rate": 8.644087647478602e-06,
      "loss": 0.8713,
      "step": 2006
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8963570594787598,
      "learning_rate": 8.642634283951257e-06,
      "loss": 1.0605,
      "step": 2007
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3231281042099,
      "learning_rate": 8.641180264252738e-06,
      "loss": 0.856,
      "step": 2008
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3284748792648315,
      "learning_rate": 8.639725588644965e-06,
      "loss": 0.8497,
      "step": 2009
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2296876907348633,
      "learning_rate": 8.63827025738998e-06,
      "loss": 0.7868,
      "step": 2010
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3131362199783325,
      "learning_rate": 8.63681427074994e-06,
      "loss": 0.8688,
      "step": 2011
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3443299531936646,
      "learning_rate": 8.635357628987122e-06,
      "loss": 0.8209,
      "step": 2012
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3269659280776978,
      "learning_rate": 8.63390033236392e-06,
      "loss": 0.8249,
      "step": 2013
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1724389791488647,
      "learning_rate": 8.632442381142846e-06,
      "loss": 0.7466,
      "step": 2014
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.235965609550476,
      "learning_rate": 8.630983775586532e-06,
      "loss": 0.8063,
      "step": 2015
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.777687132358551,
      "learning_rate": 8.629524515957724e-06,
      "loss": 1.0289,
      "step": 2016
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2759253978729248,
      "learning_rate": 8.628064602519287e-06,
      "loss": 0.8067,
      "step": 2017
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4307887554168701,
      "learning_rate": 8.626604035534205e-06,
      "loss": 0.8323,
      "step": 2018
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2453064918518066,
      "learning_rate": 8.625142815265581e-06,
      "loss": 0.7775,
      "step": 2019
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.369746208190918,
      "learning_rate": 8.623680941976634e-06,
      "loss": 0.832,
      "step": 2020
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3217054605484009,
      "learning_rate": 8.622218415930698e-06,
      "loss": 0.8553,
      "step": 2021
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3739166259765625,
      "learning_rate": 8.62075523739123e-06,
      "loss": 0.8722,
      "step": 2022
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8660325407981873,
      "learning_rate": 8.619291406621801e-06,
      "loss": 1.0581,
      "step": 2023
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3720946311950684,
      "learning_rate": 8.617826923886102e-06,
      "loss": 0.8303,
      "step": 2024
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3625508546829224,
      "learning_rate": 8.616361789447935e-06,
      "loss": 0.8839,
      "step": 2025
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2181113958358765,
      "learning_rate": 8.614896003571227e-06,
      "loss": 0.7901,
      "step": 2026
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3538272380828857,
      "learning_rate": 8.61342956652002e-06,
      "loss": 0.8504,
      "step": 2027
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3165595531463623,
      "learning_rate": 8.611962478558473e-06,
      "loss": 0.882,
      "step": 2028
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0070933103561401,
      "learning_rate": 8.61049473995086e-06,
      "loss": 1.0765,
      "step": 2029
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7531014084815979,
      "learning_rate": 8.609026350961576e-06,
      "loss": 1.0642,
      "step": 2030
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3275821208953857,
      "learning_rate": 8.60755731185513e-06,
      "loss": 0.7824,
      "step": 2031
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.359161138534546,
      "learning_rate": 8.60608762289615e-06,
      "loss": 0.8348,
      "step": 2032
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3163193464279175,
      "learning_rate": 8.604617284349381e-06,
      "loss": 0.8237,
      "step": 2033
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2834011316299438,
      "learning_rate": 8.603146296479683e-06,
      "loss": 0.8385,
      "step": 2034
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2916423082351685,
      "learning_rate": 8.601674659552038e-06,
      "loss": 0.8142,
      "step": 2035
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.257223129272461,
      "learning_rate": 8.600202373831538e-06,
      "loss": 0.8067,
      "step": 2036
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4997073411941528,
      "learning_rate": 8.598729439583398e-06,
      "loss": 0.8354,
      "step": 2037
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4195894002914429,
      "learning_rate": 8.597255857072944e-06,
      "loss": 0.8208,
      "step": 2038
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3646146059036255,
      "learning_rate": 8.595781626565624e-06,
      "loss": 0.8047,
      "step": 2039
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3627679347991943,
      "learning_rate": 8.594306748327e-06,
      "loss": 0.8117,
      "step": 2040
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.325811505317688,
      "learning_rate": 8.592831222622752e-06,
      "loss": 0.8574,
      "step": 2041
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.151857852935791,
      "learning_rate": 8.591355049718675e-06,
      "loss": 1.0677,
      "step": 2042
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7800889015197754,
      "learning_rate": 8.589878229880682e-06,
      "loss": 1.0701,
      "step": 2043
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3625085353851318,
      "learning_rate": 8.588400763374802e-06,
      "loss": 0.7871,
      "step": 2044
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7267210483551025,
      "learning_rate": 8.586922650467182e-06,
      "loss": 1.0331,
      "step": 2045
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8154907822608948,
      "learning_rate": 8.58544389142408e-06,
      "loss": 1.0691,
      "step": 2046
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6447861194610596,
      "learning_rate": 8.58396448651188e-06,
      "loss": 0.823,
      "step": 2047
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5283424854278564,
      "learning_rate": 8.582484435997071e-06,
      "loss": 0.8301,
      "step": 2048
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4347697496414185,
      "learning_rate": 8.581003740146269e-06,
      "loss": 0.8118,
      "step": 2049
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.29725980758667,
      "learning_rate": 8.579522399226196e-06,
      "loss": 0.8674,
      "step": 2050
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3518065214157104,
      "learning_rate": 8.578040413503702e-06,
      "loss": 0.8349,
      "step": 2051
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.370468020439148,
      "learning_rate": 8.576557783245743e-06,
      "loss": 0.8227,
      "step": 2052
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3315178155899048,
      "learning_rate": 8.575074508719392e-06,
      "loss": 0.8293,
      "step": 2053
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4041706323623657,
      "learning_rate": 8.573590590191844e-06,
      "loss": 0.8135,
      "step": 2054
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3435187339782715,
      "learning_rate": 8.572106027930409e-06,
      "loss": 0.8418,
      "step": 2055
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8485313653945923,
      "learning_rate": 8.570620822202507e-06,
      "loss": 1.0853,
      "step": 2056
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3430956602096558,
      "learning_rate": 8.569134973275678e-06,
      "loss": 0.8416,
      "step": 2057
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5095409154891968,
      "learning_rate": 8.567648481417581e-06,
      "loss": 0.8429,
      "step": 2058
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9803746938705444,
      "learning_rate": 8.566161346895983e-06,
      "loss": 1.057,
      "step": 2059
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3448458909988403,
      "learning_rate": 8.564673569978774e-06,
      "loss": 0.8566,
      "step": 2060
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.357819676399231,
      "learning_rate": 8.563185150933955e-06,
      "loss": 0.7776,
      "step": 2061
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.054176926612854,
      "learning_rate": 8.561696090029645e-06,
      "loss": 1.0614,
      "step": 2062
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0630730390548706,
      "learning_rate": 8.560206387534078e-06,
      "loss": 1.0419,
      "step": 2063
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8362374305725098,
      "learning_rate": 8.558716043715606e-06,
      "loss": 1.0646,
      "step": 2064
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.392660140991211,
      "learning_rate": 8.557225058842691e-06,
      "loss": 0.8443,
      "step": 2065
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.453412413597107,
      "learning_rate": 8.555733433183915e-06,
      "loss": 0.7922,
      "step": 2066
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5513023138046265,
      "learning_rate": 8.554241167007976e-06,
      "loss": 0.8558,
      "step": 2067
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2227396965026855,
      "learning_rate": 8.55274826058368e-06,
      "loss": 1.0532,
      "step": 2068
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.277818202972412,
      "learning_rate": 8.551254714179961e-06,
      "loss": 0.8216,
      "step": 2069
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8948043584823608,
      "learning_rate": 8.549760528065855e-06,
      "loss": 1.0386,
      "step": 2070
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6867256164550781,
      "learning_rate": 8.548265702510524e-06,
      "loss": 1.0626,
      "step": 2071
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7537946105003357,
      "learning_rate": 8.54677023778324e-06,
      "loss": 1.0813,
      "step": 2072
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4945894479751587,
      "learning_rate": 8.545274134153386e-06,
      "loss": 0.8341,
      "step": 2073
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0052549839019775,
      "learning_rate": 8.54377739189047e-06,
      "loss": 1.0661,
      "step": 2074
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8262003064155579,
      "learning_rate": 8.542280011264108e-06,
      "loss": 1.0603,
      "step": 2075
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7154897451400757,
      "learning_rate": 8.540781992544032e-06,
      "loss": 1.0724,
      "step": 2076
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3548600673675537,
      "learning_rate": 8.539283336000091e-06,
      "loss": 0.8239,
      "step": 2077
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8879759311676025,
      "learning_rate": 8.537784041902247e-06,
      "loss": 1.0523,
      "step": 2078
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2743704319000244,
      "learning_rate": 8.536284110520577e-06,
      "loss": 0.8726,
      "step": 2079
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.234262466430664,
      "learning_rate": 8.534783542125276e-06,
      "loss": 0.8005,
      "step": 2080
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.23948335647583,
      "learning_rate": 8.533282336986646e-06,
      "loss": 0.8193,
      "step": 2081
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.227992057800293,
      "learning_rate": 8.531780495375112e-06,
      "loss": 0.7796,
      "step": 2082
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2979785203933716,
      "learning_rate": 8.530278017561214e-06,
      "loss": 0.7912,
      "step": 2083
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2911186218261719,
      "learning_rate": 8.528774903815596e-06,
      "loss": 0.7722,
      "step": 2084
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3532379865646362,
      "learning_rate": 8.527271154409027e-06,
      "loss": 0.8191,
      "step": 2085
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1418355703353882,
      "learning_rate": 8.525766769612387e-06,
      "loss": 1.061,
      "step": 2086
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3698991537094116,
      "learning_rate": 8.52426174969667e-06,
      "loss": 0.8497,
      "step": 2087
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.309329867362976,
      "learning_rate": 8.522756094932987e-06,
      "loss": 0.7645,
      "step": 2088
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7531285285949707,
      "learning_rate": 8.521249805592558e-06,
      "loss": 1.0818,
      "step": 2089
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3222055435180664,
      "learning_rate": 8.519742881946723e-06,
      "loss": 0.7973,
      "step": 2090
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2364778518676758,
      "learning_rate": 8.518235324266931e-06,
      "loss": 0.7497,
      "step": 2091
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3665462732315063,
      "learning_rate": 8.51672713282475e-06,
      "loss": 0.7891,
      "step": 2092
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2731465101242065,
      "learning_rate": 8.515218307891864e-06,
      "loss": 0.8094,
      "step": 2093
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.367310643196106,
      "learning_rate": 8.513708849740061e-06,
      "loss": 0.8469,
      "step": 2094
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2527143955230713,
      "learning_rate": 8.512198758641252e-06,
      "loss": 0.7565,
      "step": 2095
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2415963411331177,
      "learning_rate": 8.510688034867461e-06,
      "loss": 0.791,
      "step": 2096
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.018505334854126,
      "learning_rate": 8.509176678690823e-06,
      "loss": 1.0563,
      "step": 2097
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3205889463424683,
      "learning_rate": 8.507664690383585e-06,
      "loss": 0.8279,
      "step": 2098
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7186800241470337,
      "learning_rate": 8.506152070218119e-06,
      "loss": 1.0708,
      "step": 2099
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3977471590042114,
      "learning_rate": 8.504638818466895e-06,
      "loss": 0.7849,
      "step": 2100
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.437741756439209,
      "learning_rate": 8.50312493540251e-06,
      "loss": 0.8014,
      "step": 2101
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.348262071609497,
      "learning_rate": 8.501610421297666e-06,
      "loss": 0.7638,
      "step": 2102
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3127715587615967,
      "learning_rate": 8.500095276425187e-06,
      "loss": 0.792,
      "step": 2103
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2769877910614014,
      "learning_rate": 8.498579501058e-06,
      "loss": 0.7787,
      "step": 2104
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2197740077972412,
      "learning_rate": 8.497063095469155e-06,
      "loss": 0.84,
      "step": 2105
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.304286003112793,
      "learning_rate": 8.49554605993181e-06,
      "loss": 0.8265,
      "step": 2106
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3175288438796997,
      "learning_rate": 8.494028394719241e-06,
      "loss": 0.8705,
      "step": 2107
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2569680213928223,
      "learning_rate": 8.492510100104833e-06,
      "loss": 0.7943,
      "step": 2108
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1883686780929565,
      "learning_rate": 8.490991176362086e-06,
      "loss": 1.0566,
      "step": 2109
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2929168939590454,
      "learning_rate": 8.489471623764613e-06,
      "loss": 0.8198,
      "step": 2110
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3104274272918701,
      "learning_rate": 8.487951442586141e-06,
      "loss": 0.8393,
      "step": 2111
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2963687181472778,
      "learning_rate": 8.486430633100513e-06,
      "loss": 0.8271,
      "step": 2112
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7623801827430725,
      "learning_rate": 8.484909195581682e-06,
      "loss": 1.0437,
      "step": 2113
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3181819915771484,
      "learning_rate": 8.483387130303708e-06,
      "loss": 0.7641,
      "step": 2114
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5087476968765259,
      "learning_rate": 8.481864437540777e-06,
      "loss": 0.896,
      "step": 2115
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7746613621711731,
      "learning_rate": 8.480341117567177e-06,
      "loss": 1.0709,
      "step": 2116
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3439350128173828,
      "learning_rate": 8.478817170657319e-06,
      "loss": 0.8626,
      "step": 2117
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3563779592514038,
      "learning_rate": 8.477292597085714e-06,
      "loss": 0.8166,
      "step": 2118
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3597296476364136,
      "learning_rate": 8.475767397126999e-06,
      "loss": 0.8481,
      "step": 2119
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2916091680526733,
      "learning_rate": 8.474241571055916e-06,
      "loss": 0.8682,
      "step": 2120
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5046112537384033,
      "learning_rate": 8.472715119147324e-06,
      "loss": 0.8394,
      "step": 2121
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2579549551010132,
      "learning_rate": 8.47118804167619e-06,
      "loss": 0.8656,
      "step": 2122
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9323723316192627,
      "learning_rate": 8.469660338917596e-06,
      "loss": 1.0699,
      "step": 2123
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2952396869659424,
      "learning_rate": 8.468132011146739e-06,
      "loss": 0.8699,
      "step": 2124
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.186766266822815,
      "learning_rate": 8.466603058638925e-06,
      "loss": 0.8728,
      "step": 2125
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.354356288909912,
      "learning_rate": 8.465073481669574e-06,
      "loss": 0.8514,
      "step": 2126
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3758699893951416,
      "learning_rate": 8.46354328051422e-06,
      "loss": 0.845,
      "step": 2127
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.233117699623108,
      "learning_rate": 8.462012455448507e-06,
      "loss": 0.8721,
      "step": 2128
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3756985664367676,
      "learning_rate": 8.460481006748194e-06,
      "loss": 0.8558,
      "step": 2129
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.343338966369629,
      "learning_rate": 8.458948934689145e-06,
      "loss": 0.8049,
      "step": 2130
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3250800371170044,
      "learning_rate": 8.457416239547352e-06,
      "loss": 0.8088,
      "step": 2131
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2726351022720337,
      "learning_rate": 8.4558829215989e-06,
      "loss": 0.8654,
      "step": 2132
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9079182744026184,
      "learning_rate": 8.45434898112e-06,
      "loss": 1.0771,
      "step": 2133
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3360838890075684,
      "learning_rate": 8.45281441838697e-06,
      "loss": 0.861,
      "step": 2134
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3649133443832397,
      "learning_rate": 8.45127923367624e-06,
      "loss": 0.8113,
      "step": 2135
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3553922176361084,
      "learning_rate": 8.449743427264351e-06,
      "loss": 0.8132,
      "step": 2136
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2094424962997437,
      "learning_rate": 8.448206999427962e-06,
      "loss": 0.7818,
      "step": 2137
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2930536270141602,
      "learning_rate": 8.446669950443836e-06,
      "loss": 0.7719,
      "step": 2138
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2626694440841675,
      "learning_rate": 8.445132280588853e-06,
      "loss": 0.7913,
      "step": 2139
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0241338014602661,
      "learning_rate": 8.443593990140001e-06,
      "loss": 1.0588,
      "step": 2140
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2478176355361938,
      "learning_rate": 8.442055079374389e-06,
      "loss": 0.8217,
      "step": 2141
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.298292636871338,
      "learning_rate": 8.440515548569224e-06,
      "loss": 0.7437,
      "step": 2142
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.345179796218872,
      "learning_rate": 8.438975398001834e-06,
      "loss": 0.8103,
      "step": 2143
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3173339366912842,
      "learning_rate": 8.437434627949655e-06,
      "loss": 0.8121,
      "step": 2144
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8525071740150452,
      "learning_rate": 8.435893238690238e-06,
      "loss": 1.0771,
      "step": 2145
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4838961362838745,
      "learning_rate": 8.434351230501242e-06,
      "loss": 0.8422,
      "step": 2146
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6677963733673096,
      "learning_rate": 8.432808603660439e-06,
      "loss": 1.0744,
      "step": 2147
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2671858072280884,
      "learning_rate": 8.431265358445713e-06,
      "loss": 0.7853,
      "step": 2148
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3707600831985474,
      "learning_rate": 8.429721495135057e-06,
      "loss": 0.8353,
      "step": 2149
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.274660587310791,
      "learning_rate": 8.42817701400658e-06,
      "loss": 0.8335,
      "step": 2150
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7898387908935547,
      "learning_rate": 8.426631915338497e-06,
      "loss": 1.0854,
      "step": 2151
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7051751017570496,
      "learning_rate": 8.425086199409135e-06,
      "loss": 1.0778,
      "step": 2152
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.454485297203064,
      "learning_rate": 8.423539866496938e-06,
      "loss": 0.8094,
      "step": 2153
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3405841588974,
      "learning_rate": 8.421992916880454e-06,
      "loss": 0.8195,
      "step": 2154
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2452130317687988,
      "learning_rate": 8.420445350838346e-06,
      "loss": 0.7782,
      "step": 2155
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3161585330963135,
      "learning_rate": 8.418897168649388e-06,
      "loss": 0.8177,
      "step": 2156
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.256535291671753,
      "learning_rate": 8.417348370592462e-06,
      "loss": 0.7864,
      "step": 2157
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4136390686035156,
      "learning_rate": 8.415798956946564e-06,
      "loss": 0.8114,
      "step": 2158
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2633192539215088,
      "learning_rate": 8.414248927990797e-06,
      "loss": 0.8097,
      "step": 2159
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2200617790222168,
      "learning_rate": 8.412698284004385e-06,
      "loss": 1.0633,
      "step": 2160
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3328580856323242,
      "learning_rate": 8.411147025266649e-06,
      "loss": 0.816,
      "step": 2161
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2899833917617798,
      "learning_rate": 8.409595152057028e-06,
      "loss": 0.8184,
      "step": 2162
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7929823994636536,
      "learning_rate": 8.408042664655071e-06,
      "loss": 1.0614,
      "step": 2163
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3304287195205688,
      "learning_rate": 8.406489563340442e-06,
      "loss": 0.8175,
      "step": 2164
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.367633581161499,
      "learning_rate": 8.404935848392905e-06,
      "loss": 0.8186,
      "step": 2165
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8778930306434631,
      "learning_rate": 8.403381520092346e-06,
      "loss": 1.0577,
      "step": 2166
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7910637855529785,
      "learning_rate": 8.401826578718752e-06,
      "loss": 1.0657,
      "step": 2167
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3406966924667358,
      "learning_rate": 8.400271024552225e-06,
      "loss": 0.8289,
      "step": 2168
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3196682929992676,
      "learning_rate": 8.39871485787298e-06,
      "loss": 0.8255,
      "step": 2169
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.329361915588379,
      "learning_rate": 8.397158078961336e-06,
      "loss": 0.8203,
      "step": 2170
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3565034866333008,
      "learning_rate": 8.39560068809773e-06,
      "loss": 0.8234,
      "step": 2171
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4368844032287598,
      "learning_rate": 8.394042685562699e-06,
      "loss": 1.0674,
      "step": 2172
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2711973190307617,
      "learning_rate": 8.3924840716369e-06,
      "loss": 0.8396,
      "step": 2173
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2760454416275024,
      "learning_rate": 8.390924846601095e-06,
      "loss": 0.8142,
      "step": 2174
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3199987411499023,
      "learning_rate": 8.389365010736157e-06,
      "loss": 0.8128,
      "step": 2175
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7459649443626404,
      "learning_rate": 8.38780456432307e-06,
      "loss": 1.054,
      "step": 2176
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7220907807350159,
      "learning_rate": 8.386243507642926e-06,
      "loss": 1.0389,
      "step": 2177
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.334349274635315,
      "learning_rate": 8.384681840976932e-06,
      "loss": 0.8055,
      "step": 2178
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2760862112045288,
      "learning_rate": 8.383119564606394e-06,
      "loss": 0.7921,
      "step": 2179
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9175894260406494,
      "learning_rate": 8.381556678812742e-06,
      "loss": 1.0646,
      "step": 2180
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2340179681777954,
      "learning_rate": 8.379993183877506e-06,
      "loss": 0.7865,
      "step": 2181
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3277244567871094,
      "learning_rate": 8.378429080082327e-06,
      "loss": 0.8622,
      "step": 2182
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.332344889640808,
      "learning_rate": 8.376864367708958e-06,
      "loss": 0.831,
      "step": 2183
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2653818130493164,
      "learning_rate": 8.375299047039263e-06,
      "loss": 0.7868,
      "step": 2184
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3013697862625122,
      "learning_rate": 8.37373311835521e-06,
      "loss": 0.8212,
      "step": 2185
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3225735425949097,
      "learning_rate": 8.372166581938882e-06,
      "loss": 0.9053,
      "step": 2186
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9077569246292114,
      "learning_rate": 8.370599438072465e-06,
      "loss": 1.0485,
      "step": 2187
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3015589714050293,
      "learning_rate": 8.369031687038267e-06,
      "loss": 0.8118,
      "step": 2188
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4291507005691528,
      "learning_rate": 8.367463329118688e-06,
      "loss": 0.8768,
      "step": 2189
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6455700397491455,
      "learning_rate": 8.365894364596253e-06,
      "loss": 0.8357,
      "step": 2190
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7309495210647583,
      "learning_rate": 8.364324793753587e-06,
      "loss": 1.0833,
      "step": 2191
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3228507041931152,
      "learning_rate": 8.362754616873428e-06,
      "loss": 0.8066,
      "step": 2192
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.334205150604248,
      "learning_rate": 8.36118383423862e-06,
      "loss": 0.831,
      "step": 2193
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2438501119613647,
      "learning_rate": 8.35961244613212e-06,
      "loss": 0.8296,
      "step": 2194
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2286453247070312,
      "learning_rate": 8.358040452836991e-06,
      "loss": 0.8268,
      "step": 2195
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6943143606185913,
      "learning_rate": 8.356467854636408e-06,
      "loss": 1.0571,
      "step": 2196
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2963814735412598,
      "learning_rate": 8.354894651813653e-06,
      "loss": 0.8506,
      "step": 2197
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6729733943939209,
      "learning_rate": 8.353320844652114e-06,
      "loss": 1.0338,
      "step": 2198
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3505526781082153,
      "learning_rate": 8.351746433435295e-06,
      "loss": 0.8535,
      "step": 2199
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2567750215530396,
      "learning_rate": 8.350171418446803e-06,
      "loss": 0.8911,
      "step": 2200
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3300085067749023,
      "learning_rate": 8.348595799970356e-06,
      "loss": 0.843,
      "step": 2201
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7505722045898438,
      "learning_rate": 8.347019578289781e-06,
      "loss": 1.062,
      "step": 2202
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3167468309402466,
      "learning_rate": 8.34544275368901e-06,
      "loss": 0.8383,
      "step": 2203
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6585839986801147,
      "learning_rate": 8.34386532645209e-06,
      "loss": 1.0683,
      "step": 2204
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6508877873420715,
      "learning_rate": 8.342287296863173e-06,
      "loss": 1.0725,
      "step": 2205
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2693681716918945,
      "learning_rate": 8.340708665206517e-06,
      "loss": 0.7993,
      "step": 2206
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3406904935836792,
      "learning_rate": 8.339129431766492e-06,
      "loss": 0.8185,
      "step": 2207
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8302098512649536,
      "learning_rate": 8.337549596827578e-06,
      "loss": 1.0534,
      "step": 2208
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.307638168334961,
      "learning_rate": 8.335969160674357e-06,
      "loss": 0.8189,
      "step": 2209
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2387748956680298,
      "learning_rate": 8.334388123591528e-06,
      "loss": 0.8511,
      "step": 2210
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3132387399673462,
      "learning_rate": 8.332806485863888e-06,
      "loss": 0.8314,
      "step": 2211
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8538398742675781,
      "learning_rate": 8.331224247776351e-06,
      "loss": 0.8051,
      "step": 2212
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.240580439567566,
      "learning_rate": 8.329641409613933e-06,
      "loss": 0.7899,
      "step": 2213
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6991777420043945,
      "learning_rate": 8.328057971661765e-06,
      "loss": 1.0482,
      "step": 2214
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3388501405715942,
      "learning_rate": 8.326473934205077e-06,
      "loss": 0.8145,
      "step": 2215
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3156468868255615,
      "learning_rate": 8.324889297529217e-06,
      "loss": 0.8412,
      "step": 2216
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2365913391113281,
      "learning_rate": 8.323304061919633e-06,
      "loss": 0.8106,
      "step": 2217
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6684228181838989,
      "learning_rate": 8.321718227661881e-06,
      "loss": 1.0869,
      "step": 2218
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2392035722732544,
      "learning_rate": 8.320131795041633e-06,
      "loss": 0.7937,
      "step": 2219
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3067657947540283,
      "learning_rate": 8.31854476434466e-06,
      "loss": 0.7763,
      "step": 2220
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2275733947753906,
      "learning_rate": 8.316957135856845e-06,
      "loss": 0.715,
      "step": 2221
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2655072212219238,
      "learning_rate": 8.315368909864177e-06,
      "loss": 0.8288,
      "step": 2222
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2566604614257812,
      "learning_rate": 8.313780086652753e-06,
      "loss": 0.7927,
      "step": 2223
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3445963859558105,
      "learning_rate": 8.31219066650878e-06,
      "loss": 0.801,
      "step": 2224
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7316779494285583,
      "learning_rate": 8.31060064971857e-06,
      "loss": 1.0614,
      "step": 2225
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2321075201034546,
      "learning_rate": 8.30901003656854e-06,
      "loss": 0.769,
      "step": 2226
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2649712562561035,
      "learning_rate": 8.307418827345224e-06,
      "loss": 0.8138,
      "step": 2227
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6596532464027405,
      "learning_rate": 8.305827022335249e-06,
      "loss": 1.0467,
      "step": 2228
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.315221905708313,
      "learning_rate": 8.30423462182536e-06,
      "loss": 0.8232,
      "step": 2229
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3013619184494019,
      "learning_rate": 8.30264162610241e-06,
      "loss": 0.842,
      "step": 2230
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2323247194290161,
      "learning_rate": 8.30104803545335e-06,
      "loss": 0.7965,
      "step": 2231
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7529309391975403,
      "learning_rate": 8.299453850165247e-06,
      "loss": 1.0715,
      "step": 2232
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3505305051803589,
      "learning_rate": 8.297859070525271e-06,
      "loss": 0.7978,
      "step": 2233
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3158682584762573,
      "learning_rate": 8.296263696820701e-06,
      "loss": 0.8458,
      "step": 2234
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3491370677947998,
      "learning_rate": 8.29466772933892e-06,
      "loss": 0.831,
      "step": 2235
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3090577125549316,
      "learning_rate": 8.293071168367422e-06,
      "loss": 0.8476,
      "step": 2236
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3855159282684326,
      "learning_rate": 8.291474014193803e-06,
      "loss": 0.8599,
      "step": 2237
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2360128164291382,
      "learning_rate": 8.289876267105771e-06,
      "loss": 0.8318,
      "step": 2238
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.298125982284546,
      "learning_rate": 8.288277927391138e-06,
      "loss": 0.8436,
      "step": 2239
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.554591178894043,
      "learning_rate": 8.286678995337823e-06,
      "loss": 0.8586,
      "step": 2240
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3360745906829834,
      "learning_rate": 8.285079471233852e-06,
      "loss": 0.8069,
      "step": 2241
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2689207792282104,
      "learning_rate": 8.283479355367357e-06,
      "loss": 0.8056,
      "step": 2242
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3015549182891846,
      "learning_rate": 8.281878648026577e-06,
      "loss": 0.8087,
      "step": 2243
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2085243463516235,
      "learning_rate": 8.280277349499863e-06,
      "loss": 0.8133,
      "step": 2244
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.349135160446167,
      "learning_rate": 8.278675460075658e-06,
      "loss": 0.8483,
      "step": 2245
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2536485195159912,
      "learning_rate": 8.277072980042528e-06,
      "loss": 0.7936,
      "step": 2246
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.967313826084137,
      "learning_rate": 8.275469909689134e-06,
      "loss": 1.0522,
      "step": 2247
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.233742117881775,
      "learning_rate": 8.27386624930425e-06,
      "loss": 0.8066,
      "step": 2248
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6478769779205322,
      "learning_rate": 8.27226199917675e-06,
      "loss": 1.062,
      "step": 2249
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.689708948135376,
      "learning_rate": 8.270657159595623e-06,
      "loss": 1.0359,
      "step": 2250
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2247650623321533,
      "learning_rate": 8.269051730849955e-06,
      "loss": 0.755,
      "step": 2251
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7044805884361267,
      "learning_rate": 8.267445713228943e-06,
      "loss": 1.0503,
      "step": 2252
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2581828832626343,
      "learning_rate": 8.265839107021893e-06,
      "loss": 0.8508,
      "step": 2253
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3046001195907593,
      "learning_rate": 8.264231912518207e-06,
      "loss": 0.8495,
      "step": 2254
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6816251277923584,
      "learning_rate": 8.262624130007404e-06,
      "loss": 1.0723,
      "step": 2255
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7163892388343811,
      "learning_rate": 8.261015759779103e-06,
      "loss": 1.0549,
      "step": 2256
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3464144468307495,
      "learning_rate": 8.259406802123029e-06,
      "loss": 0.7835,
      "step": 2257
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2317105531692505,
      "learning_rate": 8.257797257329017e-06,
      "loss": 0.8126,
      "step": 2258
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6607652902603149,
      "learning_rate": 8.256187125687001e-06,
      "loss": 1.0562,
      "step": 2259
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2541719675064087,
      "learning_rate": 8.254576407487026e-06,
      "loss": 0.7874,
      "step": 2260
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3454391956329346,
      "learning_rate": 8.25296510301924e-06,
      "loss": 0.7891,
      "step": 2261
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2383118867874146,
      "learning_rate": 8.251353212573899e-06,
      "loss": 0.7375,
      "step": 2262
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2296299934387207,
      "learning_rate": 8.249740736441362e-06,
      "loss": 0.7857,
      "step": 2263
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8320280909538269,
      "learning_rate": 8.248127674912097e-06,
      "loss": 1.0848,
      "step": 2264
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2480639219284058,
      "learning_rate": 8.246514028276675e-06,
      "loss": 0.8764,
      "step": 2265
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2436304092407227,
      "learning_rate": 8.244899796825767e-06,
      "loss": 0.8256,
      "step": 2266
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2819372415542603,
      "learning_rate": 8.243284980850164e-06,
      "loss": 0.7879,
      "step": 2267
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3385255336761475,
      "learning_rate": 8.241669580640746e-06,
      "loss": 0.8217,
      "step": 2268
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2307171821594238,
      "learning_rate": 8.240053596488507e-06,
      "loss": 0.8209,
      "step": 2269
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.353745460510254,
      "learning_rate": 8.238437028684546e-06,
      "loss": 0.8486,
      "step": 2270
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.246900200843811,
      "learning_rate": 8.236819877520064e-06,
      "loss": 0.7817,
      "step": 2271
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2696727514266968,
      "learning_rate": 8.235202143286371e-06,
      "loss": 0.8353,
      "step": 2272
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2882057428359985,
      "learning_rate": 8.23358382627488e-06,
      "loss": 0.8726,
      "step": 2273
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8346015810966492,
      "learning_rate": 8.231964926777106e-06,
      "loss": 1.0342,
      "step": 2274
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2684465646743774,
      "learning_rate": 8.230345445084675e-06,
      "loss": 0.7922,
      "step": 2275
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.644168734550476,
      "learning_rate": 8.228725381489312e-06,
      "loss": 0.8181,
      "step": 2276
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6519126892089844,
      "learning_rate": 8.227104736282851e-06,
      "loss": 1.0703,
      "step": 2277
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3948687314987183,
      "learning_rate": 8.225483509757231e-06,
      "loss": 0.7824,
      "step": 2278
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3471593856811523,
      "learning_rate": 8.22386170220449e-06,
      "loss": 0.8344,
      "step": 2279
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2382962703704834,
      "learning_rate": 8.222239313916776e-06,
      "loss": 0.8244,
      "step": 2280
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.381190538406372,
      "learning_rate": 8.220616345186342e-06,
      "loss": 0.8143,
      "step": 2281
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3538668155670166,
      "learning_rate": 8.218992796305544e-06,
      "loss": 0.8463,
      "step": 2282
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0036927461624146,
      "learning_rate": 8.217368667566838e-06,
      "loss": 1.0561,
      "step": 2283
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3198798894882202,
      "learning_rate": 8.215743959262793e-06,
      "loss": 0.8492,
      "step": 2284
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2627427577972412,
      "learning_rate": 8.214118671686078e-06,
      "loss": 0.85,
      "step": 2285
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.713156521320343,
      "learning_rate": 8.212492805129465e-06,
      "loss": 1.0647,
      "step": 2286
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3753211498260498,
      "learning_rate": 8.21086635988583e-06,
      "loss": 0.8308,
      "step": 2287
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.781042754650116,
      "learning_rate": 8.209239336248159e-06,
      "loss": 1.0482,
      "step": 2288
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6832180619239807,
      "learning_rate": 8.207611734509537e-06,
      "loss": 1.0571,
      "step": 2289
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6540156006813049,
      "learning_rate": 8.205983554963152e-06,
      "loss": 1.0564,
      "step": 2290
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2868304252624512,
      "learning_rate": 8.204354797902304e-06,
      "loss": 0.8112,
      "step": 2291
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.206943154335022,
      "learning_rate": 8.202725463620384e-06,
      "loss": 0.8333,
      "step": 2292
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0008858442306519,
      "learning_rate": 8.2010955524109e-06,
      "loss": 1.0558,
      "step": 2293
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3120478391647339,
      "learning_rate": 8.199465064567457e-06,
      "loss": 0.8046,
      "step": 2294
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2049891948699951,
      "learning_rate": 8.197834000383766e-06,
      "loss": 0.771,
      "step": 2295
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3536620140075684,
      "learning_rate": 8.19620236015364e-06,
      "loss": 0.8843,
      "step": 2296
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.315049171447754,
      "learning_rate": 8.194570144170997e-06,
      "loss": 0.8834,
      "step": 2297
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.273278832435608,
      "learning_rate": 8.192937352729858e-06,
      "loss": 0.8529,
      "step": 2298
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8661036491394043,
      "learning_rate": 8.191303986124347e-06,
      "loss": 1.0339,
      "step": 2299
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.294729471206665,
      "learning_rate": 8.189670044648696e-06,
      "loss": 0.8016,
      "step": 2300
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7026841640472412,
      "learning_rate": 8.188035528597237e-06,
      "loss": 1.0624,
      "step": 2301
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.274107575416565,
      "learning_rate": 8.186400438264405e-06,
      "loss": 0.8343,
      "step": 2302
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3432408571243286,
      "learning_rate": 8.184764773944737e-06,
      "loss": 0.7599,
      "step": 2303
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3434797525405884,
      "learning_rate": 8.183128535932882e-06,
      "loss": 0.8276,
      "step": 2304
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4355965852737427,
      "learning_rate": 8.18149172452358e-06,
      "loss": 0.8242,
      "step": 2305
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2860522270202637,
      "learning_rate": 8.179854340011683e-06,
      "loss": 0.8781,
      "step": 2306
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2322996854782104,
      "learning_rate": 8.178216382692143e-06,
      "loss": 0.8236,
      "step": 2307
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3248950242996216,
      "learning_rate": 8.176577852860015e-06,
      "loss": 0.8102,
      "step": 2308
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1406500339508057,
      "learning_rate": 8.174938750810461e-06,
      "loss": 1.0582,
      "step": 2309
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9193026423454285,
      "learning_rate": 8.173299076838739e-06,
      "loss": 1.0465,
      "step": 2310
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7610611915588379,
      "learning_rate": 8.171658831240216e-06,
      "loss": 1.0469,
      "step": 2311
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.747649073600769,
      "learning_rate": 8.170018014310361e-06,
      "loss": 1.0414,
      "step": 2312
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4794846773147583,
      "learning_rate": 8.168376626344743e-06,
      "loss": 0.8717,
      "step": 2313
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3833250999450684,
      "learning_rate": 8.166734667639039e-06,
      "loss": 0.8406,
      "step": 2314
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2221722602844238,
      "learning_rate": 8.16509213848902e-06,
      "loss": 0.7501,
      "step": 2315
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4319308996200562,
      "learning_rate": 8.163449039190569e-06,
      "loss": 1.0734,
      "step": 2316
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2726119756698608,
      "learning_rate": 8.161805370039671e-06,
      "loss": 0.7692,
      "step": 2317
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2777305841445923,
      "learning_rate": 8.160161131332404e-06,
      "loss": 0.7639,
      "step": 2318
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2404143810272217,
      "learning_rate": 8.15851632336496e-06,
      "loss": 0.8022,
      "step": 2319
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2857813835144043,
      "learning_rate": 8.156870946433628e-06,
      "loss": 0.8105,
      "step": 2320
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.263426661491394,
      "learning_rate": 8.1552250008348e-06,
      "loss": 0.8159,
      "step": 2321
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3278601169586182,
      "learning_rate": 8.153578486864967e-06,
      "loss": 0.8233,
      "step": 2322
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2499425411224365,
      "learning_rate": 8.151931404820734e-06,
      "loss": 0.833,
      "step": 2323
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1963523626327515,
      "learning_rate": 8.150283754998795e-06,
      "loss": 1.0578,
      "step": 2324
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9049594402313232,
      "learning_rate": 8.148635537695955e-06,
      "loss": 1.055,
      "step": 2325
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6996373534202576,
      "learning_rate": 8.146986753209114e-06,
      "loss": 1.0428,
      "step": 2326
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.336546778678894,
      "learning_rate": 8.14533740183528e-06,
      "loss": 0.8483,
      "step": 2327
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3412518501281738,
      "learning_rate": 8.143687483871563e-06,
      "loss": 0.8057,
      "step": 2328
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4474799633026123,
      "learning_rate": 8.142036999615171e-06,
      "loss": 0.7894,
      "step": 2329
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.322127342224121,
      "learning_rate": 8.14038594936342e-06,
      "loss": 0.7782,
      "step": 2330
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.336795449256897,
      "learning_rate": 8.138734333413715e-06,
      "loss": 0.8195,
      "step": 2331
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2548794746398926,
      "learning_rate": 8.137082152063584e-06,
      "loss": 0.8483,
      "step": 2332
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3003495931625366,
      "learning_rate": 8.13542940561064e-06,
      "loss": 0.8531,
      "step": 2333
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4558745622634888,
      "learning_rate": 8.133776094352602e-06,
      "loss": 0.8319,
      "step": 2334
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3912816047668457,
      "learning_rate": 8.13212221858729e-06,
      "loss": 0.8469,
      "step": 2335
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4121745824813843,
      "learning_rate": 8.130467778612634e-06,
      "loss": 0.8339,
      "step": 2336
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4149901866912842,
      "learning_rate": 8.128812774726651e-06,
      "loss": 0.8324,
      "step": 2337
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3546104431152344,
      "learning_rate": 8.127157207227473e-06,
      "loss": 0.8271,
      "step": 2338
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1870675086975098,
      "learning_rate": 8.125501076413325e-06,
      "loss": 0.8149,
      "step": 2339
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.4117376804351807,
      "learning_rate": 8.123844382582538e-06,
      "loss": 1.0573,
      "step": 2340
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.087226629257202,
      "learning_rate": 8.122187126033542e-06,
      "loss": 1.0887,
      "step": 2341
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2955412864685059,
      "learning_rate": 8.12052930706487e-06,
      "loss": 0.807,
      "step": 2342
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2822808027267456,
      "learning_rate": 8.118870925975158e-06,
      "loss": 0.8173,
      "step": 2343
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3568933010101318,
      "learning_rate": 8.117211983063134e-06,
      "loss": 0.8523,
      "step": 2344
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3122979402542114,
      "learning_rate": 8.115552478627642e-06,
      "loss": 0.8559,
      "step": 2345
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2591246366500854,
      "learning_rate": 8.113892412967614e-06,
      "loss": 0.7347,
      "step": 2346
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2768720388412476,
      "learning_rate": 8.11223178638209e-06,
      "loss": 0.8104,
      "step": 2347
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.299814224243164,
      "learning_rate": 8.11057059917021e-06,
      "loss": 0.8544,
      "step": 2348
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2987196445465088,
      "learning_rate": 8.108908851631213e-06,
      "loss": 0.8073,
      "step": 2349
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2534143924713135,
      "learning_rate": 8.10724654406444e-06,
      "loss": 0.802,
      "step": 2350
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3340709209442139,
      "learning_rate": 8.105583676769336e-06,
      "loss": 0.8857,
      "step": 2351
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3166790008544922,
      "learning_rate": 8.103920250045442e-06,
      "loss": 0.8332,
      "step": 2352
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1440348625183105,
      "learning_rate": 8.1022562641924e-06,
      "loss": 1.0732,
      "step": 2353
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3210127353668213,
      "learning_rate": 8.100591719509959e-06,
      "loss": 0.8429,
      "step": 2354
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.284909725189209,
      "learning_rate": 8.09892661629796e-06,
      "loss": 0.86,
      "step": 2355
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2983791828155518,
      "learning_rate": 8.09726095485635e-06,
      "loss": 1.0518,
      "step": 2356
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8168741464614868,
      "learning_rate": 8.095594735485177e-06,
      "loss": 1.0649,
      "step": 2357
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8085513710975647,
      "learning_rate": 8.093927958484585e-06,
      "loss": 1.0455,
      "step": 2358
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5008617639541626,
      "learning_rate": 8.092260624154822e-06,
      "loss": 0.8617,
      "step": 2359
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4207271337509155,
      "learning_rate": 8.09059273279624e-06,
      "loss": 0.8654,
      "step": 2360
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5426335334777832,
      "learning_rate": 8.08892428470928e-06,
      "loss": 0.8757,
      "step": 2361
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9956741333007812,
      "learning_rate": 8.087255280194494e-06,
      "loss": 1.0515,
      "step": 2362
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3428007364273071,
      "learning_rate": 8.08558571955253e-06,
      "loss": 0.794,
      "step": 2363
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4991925954818726,
      "learning_rate": 8.083915603084138e-06,
      "loss": 0.8889,
      "step": 2364
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.480966329574585,
      "learning_rate": 8.082244931090167e-06,
      "loss": 0.7879,
      "step": 2365
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.347727656364441,
      "learning_rate": 8.080573703871563e-06,
      "loss": 0.8146,
      "step": 2366
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.361616849899292,
      "learning_rate": 8.078901921729377e-06,
      "loss": 0.848,
      "step": 2367
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2988739013671875,
      "learning_rate": 8.077229584964757e-06,
      "loss": 1.0586,
      "step": 2368
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9493000507354736,
      "learning_rate": 8.075556693878954e-06,
      "loss": 1.0515,
      "step": 2369
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7267917394638062,
      "learning_rate": 8.073883248773313e-06,
      "loss": 1.0596,
      "step": 2370
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4041746854782104,
      "learning_rate": 8.072209249949285e-06,
      "loss": 0.8396,
      "step": 2371
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4053585529327393,
      "learning_rate": 8.07053469770842e-06,
      "loss": 0.8027,
      "step": 2372
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5063879489898682,
      "learning_rate": 8.068859592352359e-06,
      "loss": 1.0686,
      "step": 2373
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.419188141822815,
      "learning_rate": 8.067183934182856e-06,
      "loss": 0.8443,
      "step": 2374
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2797415256500244,
      "learning_rate": 8.065507723501757e-06,
      "loss": 0.8225,
      "step": 2375
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2801581621170044,
      "learning_rate": 8.063830960611005e-06,
      "loss": 0.8499,
      "step": 2376
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.293140172958374,
      "learning_rate": 8.062153645812651e-06,
      "loss": 0.8035,
      "step": 2377
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.344037413597107,
      "learning_rate": 8.060475779408839e-06,
      "loss": 0.8288,
      "step": 2378
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.297466516494751,
      "learning_rate": 8.058797361701812e-06,
      "loss": 0.7927,
      "step": 2379
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.38974928855896,
      "learning_rate": 8.057118392993916e-06,
      "loss": 0.7824,
      "step": 2380
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3436881303787231,
      "learning_rate": 8.055438873587593e-06,
      "loss": 0.7661,
      "step": 2381
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2536983489990234,
      "learning_rate": 8.053758803785387e-06,
      "loss": 0.7708,
      "step": 2382
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3313558101654053,
      "learning_rate": 8.052078183889938e-06,
      "loss": 1.0685,
      "step": 2383
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3061329126358032,
      "learning_rate": 8.05039701420399e-06,
      "loss": 0.7361,
      "step": 2384
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9409991502761841,
      "learning_rate": 8.04871529503038e-06,
      "loss": 1.0411,
      "step": 2385
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.747991144657135,
      "learning_rate": 8.047033026672049e-06,
      "loss": 1.0308,
      "step": 2386
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.297468900680542,
      "learning_rate": 8.045350209432033e-06,
      "loss": 0.838,
      "step": 2387
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3288829326629639,
      "learning_rate": 8.04366684361347e-06,
      "loss": 0.7874,
      "step": 2388
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.020801067352295,
      "learning_rate": 8.041982929519595e-06,
      "loss": 1.091,
      "step": 2389
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2820755243301392,
      "learning_rate": 8.040298467453742e-06,
      "loss": 0.7518,
      "step": 2390
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0250736474990845,
      "learning_rate": 8.038613457719344e-06,
      "loss": 1.071,
      "step": 2391
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2946605682373047,
      "learning_rate": 8.036927900619932e-06,
      "loss": 0.7929,
      "step": 2392
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3285164833068848,
      "learning_rate": 8.035241796459142e-06,
      "loss": 0.8428,
      "step": 2393
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.252622365951538,
      "learning_rate": 8.033555145540693e-06,
      "loss": 0.849,
      "step": 2394
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3994954824447632,
      "learning_rate": 8.031867948168418e-06,
      "loss": 0.8293,
      "step": 2395
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2788715362548828,
      "learning_rate": 8.030180204646245e-06,
      "loss": 0.8058,
      "step": 2396
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.290447473526001,
      "learning_rate": 8.028491915278193e-06,
      "loss": 0.797,
      "step": 2397
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2737709283828735,
      "learning_rate": 8.026803080368387e-06,
      "loss": 0.8011,
      "step": 2398
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9675251245498657,
      "learning_rate": 8.025113700221047e-06,
      "loss": 1.0632,
      "step": 2399
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3173128366470337,
      "learning_rate": 8.023423775140494e-06,
      "loss": 0.8218,
      "step": 2400
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2689181566238403,
      "learning_rate": 8.021733305431141e-06,
      "loss": 0.8257,
      "step": 2401
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3816958665847778,
      "learning_rate": 8.020042291397507e-06,
      "loss": 0.8724,
      "step": 2402
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2911607027053833,
      "learning_rate": 8.018350733344203e-06,
      "loss": 0.8591,
      "step": 2403
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7272858619689941,
      "learning_rate": 8.016658631575942e-06,
      "loss": 1.027,
      "step": 2404
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7035839557647705,
      "learning_rate": 8.01496598639753e-06,
      "loss": 1.055,
      "step": 2405
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2960233688354492,
      "learning_rate": 8.01327279811388e-06,
      "loss": 0.788,
      "step": 2406
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2901185750961304,
      "learning_rate": 8.01157906702999e-06,
      "loss": 0.7952,
      "step": 2407
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7344605922698975,
      "learning_rate": 8.009884793450968e-06,
      "loss": 0.8056,
      "step": 2408
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7378737330436707,
      "learning_rate": 8.008189977682013e-06,
      "loss": 1.0588,
      "step": 2409
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.339284896850586,
      "learning_rate": 8.00649462002842e-06,
      "loss": 0.7626,
      "step": 2410
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2407617568969727,
      "learning_rate": 8.004798720795588e-06,
      "loss": 0.8558,
      "step": 2411
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2602213621139526,
      "learning_rate": 8.00310228028901e-06,
      "loss": 0.8476,
      "step": 2412
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7805955410003662,
      "learning_rate": 8.001405298814275e-06,
      "loss": 1.0583,
      "step": 2413
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2788337469100952,
      "learning_rate": 7.999707776677072e-06,
      "loss": 0.8293,
      "step": 2414
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2676409482955933,
      "learning_rate": 7.998009714183188e-06,
      "loss": 0.8089,
      "step": 2415
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.299561619758606,
      "learning_rate": 7.996311111638505e-06,
      "loss": 0.7897,
      "step": 2416
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3272207975387573,
      "learning_rate": 7.994611969349004e-06,
      "loss": 0.8973,
      "step": 2417
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6860896944999695,
      "learning_rate": 7.99291228762076e-06,
      "loss": 1.0392,
      "step": 2418
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.33208167552948,
      "learning_rate": 7.991212066759949e-06,
      "loss": 0.8574,
      "step": 2419
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.268469214439392,
      "learning_rate": 7.989511307072842e-06,
      "loss": 0.8084,
      "step": 2420
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.269796371459961,
      "learning_rate": 7.98781000886581e-06,
      "loss": 0.8513,
      "step": 2421
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3539026975631714,
      "learning_rate": 7.986108172445318e-06,
      "loss": 0.8227,
      "step": 2422
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7556726336479187,
      "learning_rate": 7.984405798117928e-06,
      "loss": 1.0427,
      "step": 2423
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.245356798171997,
      "learning_rate": 7.982702886190299e-06,
      "loss": 0.7953,
      "step": 2424
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6648096442222595,
      "learning_rate": 7.980999436969187e-06,
      "loss": 1.0527,
      "step": 2425
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2924011945724487,
      "learning_rate": 7.97929545076145e-06,
      "loss": 0.8448,
      "step": 2426
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.209688425064087,
      "learning_rate": 7.977590927874033e-06,
      "loss": 0.7962,
      "step": 2427
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6507160067558289,
      "learning_rate": 7.975885868613985e-06,
      "loss": 1.0814,
      "step": 2428
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2548350095748901,
      "learning_rate": 7.974180273288447e-06,
      "loss": 0.8331,
      "step": 2429
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3888601064682007,
      "learning_rate": 7.972474142204663e-06,
      "loss": 0.797,
      "step": 2430
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1890324354171753,
      "learning_rate": 7.970767475669966e-06,
      "loss": 0.7808,
      "step": 2431
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1830003261566162,
      "learning_rate": 7.969060273991787e-06,
      "loss": 0.7431,
      "step": 2432
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.224105954170227,
      "learning_rate": 7.96735253747766e-06,
      "loss": 0.7935,
      "step": 2433
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7236650586128235,
      "learning_rate": 7.965644266435209e-06,
      "loss": 1.0416,
      "step": 2434
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7124893069267273,
      "learning_rate": 7.963935461172156e-06,
      "loss": 1.0579,
      "step": 2435
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.264885663986206,
      "learning_rate": 7.962226121996317e-06,
      "loss": 0.8114,
      "step": 2436
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2466707229614258,
      "learning_rate": 7.960516249215609e-06,
      "loss": 0.8698,
      "step": 2437
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3209741115570068,
      "learning_rate": 7.95880584313804e-06,
      "loss": 0.8609,
      "step": 2438
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2441574335098267,
      "learning_rate": 7.957094904071718e-06,
      "loss": 0.832,
      "step": 2439
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3141212463378906,
      "learning_rate": 7.955383432324846e-06,
      "loss": 0.8428,
      "step": 2440
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3295878171920776,
      "learning_rate": 7.95367142820572e-06,
      "loss": 0.8214,
      "step": 2441
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9409511685371399,
      "learning_rate": 7.951958892022738e-06,
      "loss": 1.0618,
      "step": 2442
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8525080680847168,
      "learning_rate": 7.950245824084386e-06,
      "loss": 1.0569,
      "step": 2443
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3812545537948608,
      "learning_rate": 7.948532224699253e-06,
      "loss": 0.8353,
      "step": 2444
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3041305541992188,
      "learning_rate": 7.946818094176019e-06,
      "loss": 0.822,
      "step": 2445
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2650188207626343,
      "learning_rate": 7.945103432823463e-06,
      "loss": 0.7419,
      "step": 2446
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3111857175827026,
      "learning_rate": 7.943388240950456e-06,
      "loss": 0.8495,
      "step": 2447
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3768796920776367,
      "learning_rate": 7.941672518865969e-06,
      "loss": 1.0664,
      "step": 2448
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1530554294586182,
      "learning_rate": 7.939956266879063e-06,
      "loss": 1.0756,
      "step": 2449
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3755894899368286,
      "learning_rate": 7.9382394852989e-06,
      "loss": 0.7623,
      "step": 2450
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4236427545547485,
      "learning_rate": 7.936522174434736e-06,
      "loss": 0.8797,
      "step": 2451
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9623661041259766,
      "learning_rate": 7.934804334595915e-06,
      "loss": 1.0415,
      "step": 2452
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.344686508178711,
      "learning_rate": 7.93308596609189e-06,
      "loss": 0.8541,
      "step": 2453
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2476487159729004,
      "learning_rate": 7.9313670692322e-06,
      "loss": 0.8177,
      "step": 2454
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2801402807235718,
      "learning_rate": 7.929647644326477e-06,
      "loss": 0.8553,
      "step": 2455
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3015811443328857,
      "learning_rate": 7.927927691684457e-06,
      "loss": 0.7966,
      "step": 2456
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3455320596694946,
      "learning_rate": 7.926207211615962e-06,
      "loss": 0.8419,
      "step": 2457
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3227157592773438,
      "learning_rate": 7.924486204430917e-06,
      "loss": 0.8013,
      "step": 2458
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2085819244384766,
      "learning_rate": 7.922764670439335e-06,
      "loss": 0.8116,
      "step": 2459
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.306480884552002,
      "learning_rate": 7.921042609951328e-06,
      "loss": 0.8437,
      "step": 2460
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2559244632720947,
      "learning_rate": 7.919320023277101e-06,
      "loss": 0.8105,
      "step": 2461
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2470364570617676,
      "learning_rate": 7.917596910726958e-06,
      "loss": 0.8138,
      "step": 2462
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.340578317642212,
      "learning_rate": 7.91587327261129e-06,
      "loss": 0.8176,
      "step": 2463
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3064417839050293,
      "learning_rate": 7.914149109240592e-06,
      "loss": 0.8617,
      "step": 2464
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3108114004135132,
      "learning_rate": 7.912424420925444e-06,
      "loss": 0.793,
      "step": 2465
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2520644664764404,
      "learning_rate": 7.910699207976527e-06,
      "loss": 1.0546,
      "step": 2466
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2845418453216553,
      "learning_rate": 7.908973470704614e-06,
      "loss": 0.8081,
      "step": 2467
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.270412564277649,
      "learning_rate": 7.907247209420575e-06,
      "loss": 0.78,
      "step": 2468
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7631124258041382,
      "learning_rate": 7.905520424435371e-06,
      "loss": 1.0655,
      "step": 2469
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2135452032089233,
      "learning_rate": 7.90379311606006e-06,
      "loss": 0.7203,
      "step": 2470
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.684046745300293,
      "learning_rate": 7.90206528460579e-06,
      "loss": 1.0637,
      "step": 2471
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7015635967254639,
      "learning_rate": 7.90033693038381e-06,
      "loss": 1.0492,
      "step": 2472
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3352164030075073,
      "learning_rate": 7.89860805370546e-06,
      "loss": 0.8363,
      "step": 2473
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2058799266815186,
      "learning_rate": 7.89687865488217e-06,
      "loss": 0.7573,
      "step": 2474
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3414084911346436,
      "learning_rate": 7.895148734225472e-06,
      "loss": 0.8021,
      "step": 2475
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2351704835891724,
      "learning_rate": 7.893418292046984e-06,
      "loss": 0.7817,
      "step": 2476
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2890262603759766,
      "learning_rate": 7.891687328658423e-06,
      "loss": 0.8316,
      "step": 2477
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2891701459884644,
      "learning_rate": 7.889955844371601e-06,
      "loss": 0.8358,
      "step": 2478
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.293441653251648,
      "learning_rate": 7.88822383949842e-06,
      "loss": 0.8399,
      "step": 2479
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0039305686950684,
      "learning_rate": 7.886491314350874e-06,
      "loss": 1.051,
      "step": 2480
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2761539220809937,
      "learning_rate": 7.884758269241057e-06,
      "loss": 0.7985,
      "step": 2481
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7134110927581787,
      "learning_rate": 7.883024704481154e-06,
      "loss": 1.0337,
      "step": 2482
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3148313760757446,
      "learning_rate": 7.881290620383442e-06,
      "loss": 0.8337,
      "step": 2483
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2962195873260498,
      "learning_rate": 7.879556017260293e-06,
      "loss": 0.8241,
      "step": 2484
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7692160606384277,
      "learning_rate": 7.87782089542417e-06,
      "loss": 1.0406,
      "step": 2485
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.261567234992981,
      "learning_rate": 7.876085255187638e-06,
      "loss": 0.8207,
      "step": 2486
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2714203596115112,
      "learning_rate": 7.874349096863341e-06,
      "loss": 0.847,
      "step": 2487
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.72441166639328,
      "learning_rate": 7.872612420764031e-06,
      "loss": 1.0542,
      "step": 2488
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7152829766273499,
      "learning_rate": 7.870875227202544e-06,
      "loss": 1.0566,
      "step": 2489
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6425485014915466,
      "learning_rate": 7.86913751649181e-06,
      "loss": 1.0329,
      "step": 2490
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.114323854446411,
      "learning_rate": 7.867399288944855e-06,
      "loss": 0.7971,
      "step": 2491
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2756853103637695,
      "learning_rate": 7.8656605448748e-06,
      "loss": 0.72,
      "step": 2492
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3420312404632568,
      "learning_rate": 7.863921284594856e-06,
      "loss": 0.7915,
      "step": 2493
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2921172380447388,
      "learning_rate": 7.86218150841832e-06,
      "loss": 0.8505,
      "step": 2494
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1865617036819458,
      "learning_rate": 7.8604412166586e-06,
      "loss": 0.7548,
      "step": 2495
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2717751264572144,
      "learning_rate": 7.858700409629177e-06,
      "loss": 0.8055,
      "step": 2496
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1140557527542114,
      "learning_rate": 7.856959087643638e-06,
      "loss": 1.0299,
      "step": 2497
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.27971351146698,
      "learning_rate": 7.855217251015658e-06,
      "loss": 0.8394,
      "step": 2498
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.201096534729004,
      "learning_rate": 7.853474900059005e-06,
      "loss": 0.7427,
      "step": 2499
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2111622095108032,
      "learning_rate": 7.851732035087537e-06,
      "loss": 0.7934,
      "step": 2500
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3654934167861938,
      "learning_rate": 7.849988656415214e-06,
      "loss": 0.804,
      "step": 2501
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7458546161651611,
      "learning_rate": 7.848244764356076e-06,
      "loss": 1.059,
      "step": 2502
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.242089867591858,
      "learning_rate": 7.846500359224266e-06,
      "loss": 0.7692,
      "step": 2503
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.326494812965393,
      "learning_rate": 7.844755441334012e-06,
      "loss": 0.8406,
      "step": 2504
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2954515218734741,
      "learning_rate": 7.843010010999636e-06,
      "loss": 0.7722,
      "step": 2505
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2058829069137573,
      "learning_rate": 7.841264068535559e-06,
      "loss": 0.8089,
      "step": 2506
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2790286540985107,
      "learning_rate": 7.839517614256282e-06,
      "loss": 0.8387,
      "step": 2507
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.244350552558899,
      "learning_rate": 7.837770648476411e-06,
      "loss": 0.8137,
      "step": 2508
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.305869698524475,
      "learning_rate": 7.836023171510637e-06,
      "loss": 0.8336,
      "step": 2509
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3878169059753418,
      "learning_rate": 7.83427518367374e-06,
      "loss": 0.815,
      "step": 2510
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.330985426902771,
      "learning_rate": 7.832526685280601e-06,
      "loss": 0.8018,
      "step": 2511
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2867670059204102,
      "learning_rate": 7.830777676646187e-06,
      "loss": 0.8285,
      "step": 2512
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.274138331413269,
      "learning_rate": 7.829028158085556e-06,
      "loss": 0.7965,
      "step": 2513
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3372647762298584,
      "learning_rate": 7.827278129913865e-06,
      "loss": 0.7581,
      "step": 2514
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2668951749801636,
      "learning_rate": 7.825527592446353e-06,
      "loss": 0.7737,
      "step": 2515
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.230137825012207,
      "learning_rate": 7.823776545998357e-06,
      "loss": 0.8642,
      "step": 2516
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.29581880569458,
      "learning_rate": 7.822024990885304e-06,
      "loss": 0.7859,
      "step": 2517
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8982836604118347,
      "learning_rate": 7.820272927422717e-06,
      "loss": 1.0901,
      "step": 2518
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7081966996192932,
      "learning_rate": 7.818520355926198e-06,
      "loss": 1.0613,
      "step": 2519
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.373171329498291,
      "learning_rate": 7.816767276711458e-06,
      "loss": 0.8281,
      "step": 2520
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3730642795562744,
      "learning_rate": 7.815013690094284e-06,
      "loss": 0.7785,
      "step": 2521
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9014841318130493,
      "learning_rate": 7.813259596390565e-06,
      "loss": 1.0711,
      "step": 2522
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7678859233856201,
      "learning_rate": 7.811504995916275e-06,
      "loss": 1.0386,
      "step": 2523
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2440314292907715,
      "learning_rate": 7.809749888987482e-06,
      "loss": 0.847,
      "step": 2524
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.404565334320068,
      "learning_rate": 7.807994275920345e-06,
      "loss": 1.0487,
      "step": 2525
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3656511306762695,
      "learning_rate": 7.806238157031112e-06,
      "loss": 0.8219,
      "step": 2526
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2604308128356934,
      "learning_rate": 7.804481532636128e-06,
      "loss": 0.7408,
      "step": 2527
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3183187246322632,
      "learning_rate": 7.80272440305182e-06,
      "loss": 0.7452,
      "step": 2528
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2462689876556396,
      "learning_rate": 7.800966768594715e-06,
      "loss": 0.7875,
      "step": 2529
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2293835878372192,
      "learning_rate": 7.799208629581426e-06,
      "loss": 0.8416,
      "step": 2530
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.34609854221344,
      "learning_rate": 7.797449986328657e-06,
      "loss": 0.844,
      "step": 2531
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.174659013748169,
      "learning_rate": 7.795690839153205e-06,
      "loss": 1.0632,
      "step": 2532
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.322798728942871,
      "learning_rate": 7.793931188371955e-06,
      "loss": 0.8078,
      "step": 2533
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6920681595802307,
      "learning_rate": 7.792171034301885e-06,
      "loss": 1.0783,
      "step": 2534
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6444718837738037,
      "learning_rate": 7.790410377260065e-06,
      "loss": 0.8798,
      "step": 2535
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.37235426902771,
      "learning_rate": 7.78864921756365e-06,
      "loss": 0.8379,
      "step": 2536
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2630258798599243,
      "learning_rate": 7.78688755552989e-06,
      "loss": 0.8618,
      "step": 2537
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1913349628448486,
      "learning_rate": 7.785125391476125e-06,
      "loss": 0.756,
      "step": 2538
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.367850422859192,
      "learning_rate": 7.783362725719788e-06,
      "loss": 0.8446,
      "step": 2539
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1661646366119385,
      "learning_rate": 7.781599558578394e-06,
      "loss": 1.0613,
      "step": 2540
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4000574350357056,
      "learning_rate": 7.779835890369555e-06,
      "loss": 0.8525,
      "step": 2541
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3961046934127808,
      "learning_rate": 7.778071721410973e-06,
      "loss": 0.7712,
      "step": 2542
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3164236545562744,
      "learning_rate": 7.776307052020442e-06,
      "loss": 0.8086,
      "step": 2543
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3607959747314453,
      "learning_rate": 7.774541882515839e-06,
      "loss": 0.8252,
      "step": 2544
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3954854011535645,
      "learning_rate": 7.772776213215136e-06,
      "loss": 0.8133,
      "step": 2545
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3988208770751953,
      "learning_rate": 7.771010044436393e-06,
      "loss": 0.8434,
      "step": 2546
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2888041734695435,
      "learning_rate": 7.769243376497766e-06,
      "loss": 0.8099,
      "step": 2547
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3678622245788574,
      "learning_rate": 7.767476209717495e-06,
      "loss": 0.8312,
      "step": 2548
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8749668002128601,
      "learning_rate": 7.765708544413907e-06,
      "loss": 1.0642,
      "step": 2549
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2756805419921875,
      "learning_rate": 7.763940380905427e-06,
      "loss": 0.7894,
      "step": 2550
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4594438076019287,
      "learning_rate": 7.762171719510564e-06,
      "loss": 0.7936,
      "step": 2551
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2465521097183228,
      "learning_rate": 7.760402560547919e-06,
      "loss": 0.7673,
      "step": 2552
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3271934986114502,
      "learning_rate": 7.758632904336183e-06,
      "loss": 0.7921,
      "step": 2553
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.367285966873169,
      "learning_rate": 7.756862751194132e-06,
      "loss": 0.7869,
      "step": 2554
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7380182147026062,
      "learning_rate": 7.755092101440638e-06,
      "loss": 1.0411,
      "step": 2555
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3747971057891846,
      "learning_rate": 7.75332095539466e-06,
      "loss": 0.8192,
      "step": 2556
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.234017014503479,
      "learning_rate": 7.751549313375246e-06,
      "loss": 0.8057,
      "step": 2557
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4690247774124146,
      "learning_rate": 7.74977717570153e-06,
      "loss": 0.8583,
      "step": 2558
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6625761389732361,
      "learning_rate": 7.748004542692742e-06,
      "loss": 1.0379,
      "step": 2559
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.5717265605926514,
      "learning_rate": 7.746231414668195e-06,
      "loss": 0.8404,
      "step": 2560
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.273646354675293,
      "learning_rate": 7.744457791947297e-06,
      "loss": 0.8654,
      "step": 2561
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7211491465568542,
      "learning_rate": 7.742683674849541e-06,
      "loss": 1.0453,
      "step": 2562
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2927120923995972,
      "learning_rate": 7.74090906369451e-06,
      "loss": 0.8201,
      "step": 2563
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6707705855369568,
      "learning_rate": 7.739133958801874e-06,
      "loss": 1.0464,
      "step": 2564
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3465341329574585,
      "learning_rate": 7.737358360491396e-06,
      "loss": 0.8373,
      "step": 2565
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.20065176486969,
      "learning_rate": 7.735582269082928e-06,
      "loss": 0.803,
      "step": 2566
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.30540931224823,
      "learning_rate": 7.733805684896407e-06,
      "loss": 0.7698,
      "step": 2567
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7177996635437012,
      "learning_rate": 7.732028608251859e-06,
      "loss": 1.0378,
      "step": 2568
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2249395847320557,
      "learning_rate": 7.730251039469401e-06,
      "loss": 0.7517,
      "step": 2569
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3050040006637573,
      "learning_rate": 7.728472978869239e-06,
      "loss": 0.8432,
      "step": 2570
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.246069312095642,
      "learning_rate": 7.726694426771667e-06,
      "loss": 0.8152,
      "step": 2571
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.260708212852478,
      "learning_rate": 7.724915383497064e-06,
      "loss": 0.7995,
      "step": 2572
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3880236148834229,
      "learning_rate": 7.723135849365906e-06,
      "loss": 0.7923,
      "step": 2573
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.246228814125061,
      "learning_rate": 7.721355824698746e-06,
      "loss": 0.7648,
      "step": 2574
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4192278385162354,
      "learning_rate": 7.719575309816232e-06,
      "loss": 0.8458,
      "step": 2575
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.285375952720642,
      "learning_rate": 7.717794305039105e-06,
      "loss": 0.8758,
      "step": 2576
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.276970386505127,
      "learning_rate": 7.716012810688183e-06,
      "loss": 0.8101,
      "step": 2577
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.364209532737732,
      "learning_rate": 7.714230827084383e-06,
      "loss": 0.855,
      "step": 2578
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3738677501678467,
      "learning_rate": 7.7124483545487e-06,
      "loss": 0.8493,
      "step": 2579
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3308827877044678,
      "learning_rate": 7.710665393402226e-06,
      "loss": 0.8322,
      "step": 2580
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8890994191169739,
      "learning_rate": 7.708881943966135e-06,
      "loss": 1.0664,
      "step": 2581
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3023958206176758,
      "learning_rate": 7.707098006561692e-06,
      "loss": 0.8138,
      "step": 2582
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3204020261764526,
      "learning_rate": 7.70531358151025e-06,
      "loss": 0.7712,
      "step": 2583
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2634247541427612,
      "learning_rate": 7.703528669133248e-06,
      "loss": 0.7928,
      "step": 2584
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2665469646453857,
      "learning_rate": 7.701743269752217e-06,
      "loss": 0.794,
      "step": 2585
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.28340482711792,
      "learning_rate": 7.699957383688767e-06,
      "loss": 0.7897,
      "step": 2586
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2726073265075684,
      "learning_rate": 7.698171011264604e-06,
      "loss": 0.7384,
      "step": 2587
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3244789838790894,
      "learning_rate": 7.696384152801519e-06,
      "loss": 0.8349,
      "step": 2588
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3723942041397095,
      "learning_rate": 7.694596808621389e-06,
      "loss": 0.844,
      "step": 2589
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8800567984580994,
      "learning_rate": 7.692808979046182e-06,
      "loss": 1.0449,
      "step": 2590
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.219887375831604,
      "learning_rate": 7.69102066439795e-06,
      "loss": 0.8008,
      "step": 2591
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3395707607269287,
      "learning_rate": 7.689231864998832e-06,
      "loss": 0.8204,
      "step": 2592
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6177923083305359,
      "learning_rate": 7.687442581171059e-06,
      "loss": 1.0373,
      "step": 2593
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6265087723731995,
      "learning_rate": 7.685652813236946e-06,
      "loss": 1.0371,
      "step": 2594
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3123242855072021,
      "learning_rate": 7.683862561518892e-06,
      "loss": 0.8413,
      "step": 2595
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6907925605773926,
      "learning_rate": 7.68207182633939e-06,
      "loss": 1.0719,
      "step": 2596
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6417236924171448,
      "learning_rate": 7.680280608021014e-06,
      "loss": 1.0419,
      "step": 2597
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3582171201705933,
      "learning_rate": 7.67848890688643e-06,
      "loss": 0.8102,
      "step": 2598
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2833127975463867,
      "learning_rate": 7.676696723258387e-06,
      "loss": 0.7632,
      "step": 2599
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3222787380218506,
      "learning_rate": 7.674904057459726e-06,
      "loss": 0.8595,
      "step": 2600
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.332717776298523,
      "learning_rate": 7.673110909813366e-06,
      "loss": 0.7975,
      "step": 2601
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8947699069976807,
      "learning_rate": 7.671317280642323e-06,
      "loss": 1.0433,
      "step": 2602
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2193801403045654,
      "learning_rate": 7.66952317026969e-06,
      "loss": 0.8237,
      "step": 2603
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2433263063430786,
      "learning_rate": 7.66772857901866e-06,
      "loss": 0.7571,
      "step": 2604
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3016198873519897,
      "learning_rate": 7.665933507212495e-06,
      "loss": 0.782,
      "step": 2605
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.24861741065979,
      "learning_rate": 7.664137955174558e-06,
      "loss": 0.8437,
      "step": 2606
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2553343772888184,
      "learning_rate": 7.66234192322829e-06,
      "loss": 0.8061,
      "step": 2607
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2996785640716553,
      "learning_rate": 7.660545411697227e-06,
      "loss": 0.7957,
      "step": 2608
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2917416095733643,
      "learning_rate": 7.658748420904983e-06,
      "loss": 0.8921,
      "step": 2609
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2558135986328125,
      "learning_rate": 7.65695095117526e-06,
      "loss": 0.7954,
      "step": 2610
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2286337614059448,
      "learning_rate": 7.655153002831849e-06,
      "loss": 0.8317,
      "step": 2611
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2859591245651245,
      "learning_rate": 7.653354576198629e-06,
      "loss": 0.8241,
      "step": 2612
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3496003150939941,
      "learning_rate": 7.651555671599557e-06,
      "loss": 0.8165,
      "step": 2613
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2844172716140747,
      "learning_rate": 7.649756289358686e-06,
      "loss": 0.8248,
      "step": 2614
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2940306663513184,
      "learning_rate": 7.647956429800147e-06,
      "loss": 0.8576,
      "step": 2615
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.255055546760559,
      "learning_rate": 7.646156093248162e-06,
      "loss": 0.8083,
      "step": 2616
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2337946891784668,
      "learning_rate": 7.644355280027037e-06,
      "loss": 0.7927,
      "step": 2617
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2845919132232666,
      "learning_rate": 7.642553990461165e-06,
      "loss": 0.7737,
      "step": 2618
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3135586977005005,
      "learning_rate": 7.640752224875022e-06,
      "loss": 0.8855,
      "step": 2619
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2901058197021484,
      "learning_rate": 7.63894998359317e-06,
      "loss": 0.8779,
      "step": 2620
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9185654520988464,
      "learning_rate": 7.637147266940264e-06,
      "loss": 1.0476,
      "step": 2621
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2360529899597168,
      "learning_rate": 7.635344075241034e-06,
      "loss": 0.8379,
      "step": 2622
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.378575325012207,
      "learning_rate": 7.633540408820301e-06,
      "loss": 0.7636,
      "step": 2623
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2629324197769165,
      "learning_rate": 7.631736268002975e-06,
      "loss": 0.8452,
      "step": 2624
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7556043267250061,
      "learning_rate": 7.629931653114044e-06,
      "loss": 1.0657,
      "step": 2625
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2106883525848389,
      "learning_rate": 7.628126564478584e-06,
      "loss": 0.7854,
      "step": 2626
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3212138414382935,
      "learning_rate": 7.626321002421761e-06,
      "loss": 0.836,
      "step": 2627
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3041436672210693,
      "learning_rate": 7.624514967268819e-06,
      "loss": 0.8302,
      "step": 2628
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2709201574325562,
      "learning_rate": 7.622708459345093e-06,
      "loss": 0.7551,
      "step": 2629
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2291224002838135,
      "learning_rate": 7.6209014789759995e-06,
      "loss": 0.7775,
      "step": 2630
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2419832944869995,
      "learning_rate": 7.619094026487042e-06,
      "loss": 0.777,
      "step": 2631
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.288009762763977,
      "learning_rate": 7.617286102203809e-06,
      "loss": 0.7849,
      "step": 2632
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2771214246749878,
      "learning_rate": 7.615477706451975e-06,
      "loss": 0.84,
      "step": 2633
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6931532621383667,
      "learning_rate": 7.613668839557294e-06,
      "loss": 1.0265,
      "step": 2634
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2974807024002075,
      "learning_rate": 7.611859501845611e-06,
      "loss": 0.854,
      "step": 2635
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2662826776504517,
      "learning_rate": 7.6100496936428545e-06,
      "loss": 0.8302,
      "step": 2636
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3533098697662354,
      "learning_rate": 7.6082394152750365e-06,
      "loss": 0.8861,
      "step": 2637
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2705013751983643,
      "learning_rate": 7.606428667068253e-06,
      "loss": 0.823,
      "step": 2638
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3749542236328125,
      "learning_rate": 7.6046174493486854e-06,
      "loss": 0.8091,
      "step": 2639
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3379907608032227,
      "learning_rate": 7.6028057624426025e-06,
      "loss": 0.7518,
      "step": 2640
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2582687139511108,
      "learning_rate": 7.600993606676355e-06,
      "loss": 0.8049,
      "step": 2641
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2314637899398804,
      "learning_rate": 7.599180982376373e-06,
      "loss": 0.7589,
      "step": 2642
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7365613579750061,
      "learning_rate": 7.597367889869183e-06,
      "loss": 1.0454,
      "step": 2643
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.264845848083496,
      "learning_rate": 7.595554329481385e-06,
      "loss": 0.7433,
      "step": 2644
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2576850652694702,
      "learning_rate": 7.593740301539667e-06,
      "loss": 0.8443,
      "step": 2645
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2479294538497925,
      "learning_rate": 7.5919258063708055e-06,
      "loss": 0.7845,
      "step": 2646
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2920308113098145,
      "learning_rate": 7.590110844301655e-06,
      "loss": 0.8684,
      "step": 2647
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7477682828903198,
      "learning_rate": 7.588295415659154e-06,
      "loss": 1.0345,
      "step": 2648
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2657362222671509,
      "learning_rate": 7.586479520770329e-06,
      "loss": 0.7954,
      "step": 2649
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4082988500595093,
      "learning_rate": 7.58466315996229e-06,
      "loss": 0.8326,
      "step": 2650
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2541872262954712,
      "learning_rate": 7.582846333562231e-06,
      "loss": 0.8323,
      "step": 2651
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2598179578781128,
      "learning_rate": 7.581029041897423e-06,
      "loss": 0.7757,
      "step": 2652
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2624545097351074,
      "learning_rate": 7.579211285295231e-06,
      "loss": 0.7734,
      "step": 2653
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8126119375228882,
      "learning_rate": 7.577393064083101e-06,
      "loss": 1.0247,
      "step": 2654
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6778536438941956,
      "learning_rate": 7.575574378588555e-06,
      "loss": 1.0376,
      "step": 2655
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.282246470451355,
      "learning_rate": 7.573755229139208e-06,
      "loss": 0.7831,
      "step": 2656
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2904164791107178,
      "learning_rate": 7.571935616062756e-06,
      "loss": 0.8554,
      "step": 2657
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7246586680412292,
      "learning_rate": 7.570115539686975e-06,
      "loss": 1.0735,
      "step": 2658
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.257712960243225,
      "learning_rate": 7.568295000339729e-06,
      "loss": 0.8255,
      "step": 2659
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2384448051452637,
      "learning_rate": 7.566473998348963e-06,
      "loss": 0.8147,
      "step": 2660
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6792202591896057,
      "learning_rate": 7.5646525340427054e-06,
      "loss": 1.0604,
      "step": 2661
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2886422872543335,
      "learning_rate": 7.56283060774907e-06,
      "loss": 0.8056,
      "step": 2662
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6499631404876709,
      "learning_rate": 7.561008219796249e-06,
      "loss": 1.0467,
      "step": 2663
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.258326768875122,
      "learning_rate": 7.559185370512523e-06,
      "loss": 0.8122,
      "step": 2664
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4665894508361816,
      "learning_rate": 7.557362060226254e-06,
      "loss": 0.8464,
      "step": 2665
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2900640964508057,
      "learning_rate": 7.555538289265885e-06,
      "loss": 0.8253,
      "step": 2666
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6922728419303894,
      "learning_rate": 7.553714057959944e-06,
      "loss": 1.0597,
      "step": 2667
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2738957405090332,
      "learning_rate": 7.551889366637043e-06,
      "loss": 0.8085,
      "step": 2668
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2102818489074707,
      "learning_rate": 7.550064215625874e-06,
      "loss": 0.8589,
      "step": 2669
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2610357999801636,
      "learning_rate": 7.548238605255215e-06,
      "loss": 0.8233,
      "step": 2670
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1775219440460205,
      "learning_rate": 7.5464125358539216e-06,
      "loss": 0.7987,
      "step": 2671
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3311848640441895,
      "learning_rate": 7.544586007750938e-06,
      "loss": 0.7991,
      "step": 2672
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3056824207305908,
      "learning_rate": 7.542759021275289e-06,
      "loss": 0.8143,
      "step": 2673
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2831999063491821,
      "learning_rate": 7.540931576756079e-06,
      "loss": 0.8262,
      "step": 2674
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3060388565063477,
      "learning_rate": 7.539103674522499e-06,
      "loss": 0.8492,
      "step": 2675
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7480466365814209,
      "learning_rate": 7.53727531490382e-06,
      "loss": 1.0379,
      "step": 2676
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6797860264778137,
      "learning_rate": 7.5354464982293975e-06,
      "loss": 1.0573,
      "step": 2677
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2837755680084229,
      "learning_rate": 7.533617224828667e-06,
      "loss": 0.83,
      "step": 2678
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3118484020233154,
      "learning_rate": 7.531787495031148e-06,
      "loss": 0.8089,
      "step": 2679
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3086966276168823,
      "learning_rate": 7.52995730916644e-06,
      "loss": 0.8312,
      "step": 2680
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2710833549499512,
      "learning_rate": 7.528126667564226e-06,
      "loss": 0.8122,
      "step": 2681
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.363019347190857,
      "learning_rate": 7.526295570554274e-06,
      "loss": 0.7647,
      "step": 2682
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8557077646255493,
      "learning_rate": 7.524464018466428e-06,
      "loss": 1.0707,
      "step": 2683
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.319028615951538,
      "learning_rate": 7.522632011630622e-06,
      "loss": 0.7886,
      "step": 2684
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7032859325408936,
      "learning_rate": 7.520799550376863e-06,
      "loss": 1.0232,
      "step": 2685
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3200355768203735,
      "learning_rate": 7.518966635035246e-06,
      "loss": 0.7773,
      "step": 2686
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3378888368606567,
      "learning_rate": 7.517133265935943e-06,
      "loss": 0.7954,
      "step": 2687
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2524116039276123,
      "learning_rate": 7.515299443409214e-06,
      "loss": 0.7616,
      "step": 2688
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2764149904251099,
      "learning_rate": 7.513465167785394e-06,
      "loss": 0.8452,
      "step": 2689
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9811126589775085,
      "learning_rate": 7.511630439394908e-06,
      "loss": 1.0525,
      "step": 2690
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2976455688476562,
      "learning_rate": 7.509795258568251e-06,
      "loss": 0.8401,
      "step": 2691
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7399026155471802,
      "learning_rate": 7.507959625636012e-06,
      "loss": 1.0545,
      "step": 2692
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3024183511734009,
      "learning_rate": 7.5061235409288515e-06,
      "loss": 0.8189,
      "step": 2693
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3666070699691772,
      "learning_rate": 7.504287004777517e-06,
      "loss": 0.8551,
      "step": 2694
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8558771014213562,
      "learning_rate": 7.5024500175128336e-06,
      "loss": 1.0632,
      "step": 2695
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2387150526046753,
      "learning_rate": 7.500612579465711e-06,
      "loss": 0.7759,
      "step": 2696
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2155805826187134,
      "learning_rate": 7.498774690967139e-06,
      "loss": 0.8447,
      "step": 2697
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2617101669311523,
      "learning_rate": 7.49693635234819e-06,
      "loss": 0.7657,
      "step": 2698
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3764618635177612,
      "learning_rate": 7.49509756394001e-06,
      "loss": 0.8454,
      "step": 2699
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7886989712715149,
      "learning_rate": 7.493258326073838e-06,
      "loss": 1.0419,
      "step": 2700
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6857669949531555,
      "learning_rate": 7.491418639080985e-06,
      "loss": 1.0658,
      "step": 2701
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.291157841682434,
      "learning_rate": 7.489578503292847e-06,
      "loss": 0.8223,
      "step": 2702
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3041760921478271,
      "learning_rate": 7.487737919040897e-06,
      "loss": 0.8586,
      "step": 2703
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3204021453857422,
      "learning_rate": 7.4858968866566925e-06,
      "loss": 0.8189,
      "step": 2704
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.27237069606781,
      "learning_rate": 7.484055406471872e-06,
      "loss": 0.8536,
      "step": 2705
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2767186164855957,
      "learning_rate": 7.482213478818151e-06,
      "loss": 0.8138,
      "step": 2706
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2069545984268188,
      "learning_rate": 7.480371104027331e-06,
      "loss": 0.7876,
      "step": 2707
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9799894094467163,
      "learning_rate": 7.478528282431287e-06,
      "loss": 1.0556,
      "step": 2708
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3041801452636719,
      "learning_rate": 7.4766850143619795e-06,
      "loss": 0.8181,
      "step": 2709
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3103934526443481,
      "learning_rate": 7.47484130015145e-06,
      "loss": 0.7864,
      "step": 2710
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2394318580627441,
      "learning_rate": 7.472997140131816e-06,
      "loss": 0.7803,
      "step": 2711
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.294140100479126,
      "learning_rate": 7.471152534635281e-06,
      "loss": 0.7597,
      "step": 2712
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3606172800064087,
      "learning_rate": 7.469307483994122e-06,
      "loss": 0.8575,
      "step": 2713
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2977555990219116,
      "learning_rate": 7.467461988540704e-06,
      "loss": 0.823,
      "step": 2714
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7815783023834229,
      "learning_rate": 7.465616048607464e-06,
      "loss": 1.0441,
      "step": 2715
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3151220083236694,
      "learning_rate": 7.463769664526926e-06,
      "loss": 0.8435,
      "step": 2716
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6663060784339905,
      "learning_rate": 7.461922836631688e-06,
      "loss": 1.0615,
      "step": 2717
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3200733661651611,
      "learning_rate": 7.4600755652544344e-06,
      "loss": 0.8011,
      "step": 2718
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3083603382110596,
      "learning_rate": 7.458227850727925e-06,
      "loss": 0.8352,
      "step": 2719
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3221420049667358,
      "learning_rate": 7.456379693384999e-06,
      "loss": 0.8213,
      "step": 2720
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7308480143547058,
      "learning_rate": 7.454531093558579e-06,
      "loss": 1.0612,
      "step": 2721
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2912379503250122,
      "learning_rate": 7.4526820515816635e-06,
      "loss": 0.79,
      "step": 2722
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2558602094650269,
      "learning_rate": 7.450832567787333e-06,
      "loss": 0.7727,
      "step": 2723
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3697075843811035,
      "learning_rate": 7.448982642508746e-06,
      "loss": 0.8097,
      "step": 2724
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6447073221206665,
      "learning_rate": 7.447132276079143e-06,
      "loss": 1.034,
      "step": 2725
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3485716581344604,
      "learning_rate": 7.445281468831843e-06,
      "loss": 0.8719,
      "step": 2726
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2983626127243042,
      "learning_rate": 7.443430221100242e-06,
      "loss": 0.7475,
      "step": 2727
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6188234686851501,
      "learning_rate": 7.441578533217817e-06,
      "loss": 1.0348,
      "step": 2728
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2790700197219849,
      "learning_rate": 7.439726405518126e-06,
      "loss": 0.8204,
      "step": 2729
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6127369999885559,
      "learning_rate": 7.437873838334805e-06,
      "loss": 1.0524,
      "step": 2730
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2403777837753296,
      "learning_rate": 7.436020832001567e-06,
      "loss": 0.7811,
      "step": 2731
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2826899290084839,
      "learning_rate": 7.434167386852206e-06,
      "loss": 0.8544,
      "step": 2732
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.276228666305542,
      "learning_rate": 7.432313503220597e-06,
      "loss": 0.7926,
      "step": 2733
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6789638996124268,
      "learning_rate": 7.43045918144069e-06,
      "loss": 1.0183,
      "step": 2734
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2214628458023071,
      "learning_rate": 7.428604421846517e-06,
      "loss": 0.8386,
      "step": 2735
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6755499839782715,
      "learning_rate": 7.426749224772188e-06,
      "loss": 1.0385,
      "step": 2736
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2614643573760986,
      "learning_rate": 7.424893590551891e-06,
      "loss": 0.771,
      "step": 2737
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2126836776733398,
      "learning_rate": 7.423037519519892e-06,
      "loss": 0.7933,
      "step": 2738
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2453360557556152,
      "learning_rate": 7.42118101201054e-06,
      "loss": 0.8454,
      "step": 2739
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2275031805038452,
      "learning_rate": 7.4193240683582576e-06,
      "loss": 0.7342,
      "step": 2740
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6885918378829956,
      "learning_rate": 7.417466688897549e-06,
      "loss": 1.0358,
      "step": 2741
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6355310082435608,
      "learning_rate": 7.415608873962995e-06,
      "loss": 1.0265,
      "step": 2742
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6311420798301697,
      "learning_rate": 7.413750623889257e-06,
      "loss": 1.0574,
      "step": 2743
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.315608024597168,
      "learning_rate": 7.411891939011073e-06,
      "loss": 0.8458,
      "step": 2744
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3102211952209473,
      "learning_rate": 7.410032819663259e-06,
      "loss": 0.7957,
      "step": 2745
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2945636510849,
      "learning_rate": 7.40817326618071e-06,
      "loss": 0.8145,
      "step": 2746
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7820457816123962,
      "learning_rate": 7.406313278898402e-06,
      "loss": 1.0388,
      "step": 2747
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2705206871032715,
      "learning_rate": 7.404452858151383e-06,
      "loss": 0.7794,
      "step": 2748
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.302814245223999,
      "learning_rate": 7.402592004274785e-06,
      "loss": 0.8233,
      "step": 2749
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2435963153839111,
      "learning_rate": 7.4007307176038156e-06,
      "loss": 0.8569,
      "step": 2750
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6517595648765564,
      "learning_rate": 7.398868998473758e-06,
      "loss": 1.0311,
      "step": 2751
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6687299609184265,
      "learning_rate": 7.397006847219978e-06,
      "loss": 1.0676,
      "step": 2752
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6491295695304871,
      "learning_rate": 7.395144264177918e-06,
      "loss": 1.0572,
      "step": 2753
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4603912830352783,
      "learning_rate": 7.393281249683093e-06,
      "loss": 0.8523,
      "step": 2754
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.322368860244751,
      "learning_rate": 7.391417804071104e-06,
      "loss": 0.857,
      "step": 2755
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.288838505744934,
      "learning_rate": 7.389553927677622e-06,
      "loss": 0.8,
      "step": 2756
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.294402837753296,
      "learning_rate": 7.3876896208384e-06,
      "loss": 0.7897,
      "step": 2757
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2559548616409302,
      "learning_rate": 7.38582488388927e-06,
      "loss": 0.7847,
      "step": 2758
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3433001041412354,
      "learning_rate": 7.383959717166136e-06,
      "loss": 0.8157,
      "step": 2759
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3709932565689087,
      "learning_rate": 7.382094121004983e-06,
      "loss": 0.8408,
      "step": 2760
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3129570484161377,
      "learning_rate": 7.380228095741875e-06,
      "loss": 0.7967,
      "step": 2761
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.233628273010254,
      "learning_rate": 7.378361641712948e-06,
      "loss": 0.8122,
      "step": 2762
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2985341548919678,
      "learning_rate": 7.376494759254422e-06,
      "loss": 0.7685,
      "step": 2763
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2544775009155273,
      "learning_rate": 7.3746274487025885e-06,
      "loss": 0.8101,
      "step": 2764
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3051704168319702,
      "learning_rate": 7.3727597103938175e-06,
      "loss": 0.8204,
      "step": 2765
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1899112462997437,
      "learning_rate": 7.370891544664558e-06,
      "loss": 0.7932,
      "step": 2766
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.285726547241211,
      "learning_rate": 7.369022951851333e-06,
      "loss": 0.792,
      "step": 2767
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.267879843711853,
      "learning_rate": 7.367153932290748e-06,
      "loss": 0.7613,
      "step": 2768
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3287932872772217,
      "learning_rate": 7.365284486319477e-06,
      "loss": 0.9291,
      "step": 2769
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8326017260551453,
      "learning_rate": 7.363414614274278e-06,
      "loss": 1.0323,
      "step": 2770
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2360173463821411,
      "learning_rate": 7.361544316491982e-06,
      "loss": 0.7916,
      "step": 2771
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6742331981658936,
      "learning_rate": 7.3596735933095e-06,
      "loss": 1.0523,
      "step": 2772
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.237335443496704,
      "learning_rate": 7.357802445063814e-06,
      "loss": 0.7664,
      "step": 2773
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.274049162864685,
      "learning_rate": 7.35593087209199e-06,
      "loss": 0.7575,
      "step": 2774
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2318555116653442,
      "learning_rate": 7.354058874731161e-06,
      "loss": 0.7748,
      "step": 2775
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2217954397201538,
      "learning_rate": 7.352186453318546e-06,
      "loss": 0.796,
      "step": 2776
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8335350751876831,
      "learning_rate": 7.350313608191436e-06,
      "loss": 1.0562,
      "step": 2777
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2413907051086426,
      "learning_rate": 7.3484403396871975e-06,
      "loss": 0.8065,
      "step": 2778
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3483872413635254,
      "learning_rate": 7.346566648143275e-06,
      "loss": 0.8232,
      "step": 2779
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.258041501045227,
      "learning_rate": 7.344692533897188e-06,
      "loss": 0.7953,
      "step": 2780
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2445213794708252,
      "learning_rate": 7.3428179972865345e-06,
      "loss": 0.8073,
      "step": 2781
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2722952365875244,
      "learning_rate": 7.340943038648986e-06,
      "loss": 0.8053,
      "step": 2782
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2741235494613647,
      "learning_rate": 7.33906765832229e-06,
      "loss": 0.8177,
      "step": 2783
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2862735986709595,
      "learning_rate": 7.33719185664427e-06,
      "loss": 0.8275,
      "step": 2784
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9355446696281433,
      "learning_rate": 7.33531563395283e-06,
      "loss": 1.0557,
      "step": 2785
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2710418701171875,
      "learning_rate": 7.333438990585942e-06,
      "loss": 0.8263,
      "step": 2786
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.472899317741394,
      "learning_rate": 7.331561926881661e-06,
      "loss": 0.7858,
      "step": 2787
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2601771354675293,
      "learning_rate": 7.329684443178112e-06,
      "loss": 0.8203,
      "step": 2788
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3640902042388916,
      "learning_rate": 7.3278065398135e-06,
      "loss": 0.8493,
      "step": 2789
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2328191995620728,
      "learning_rate": 7.3259282171261035e-06,
      "loss": 0.7242,
      "step": 2790
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8114833831787109,
      "learning_rate": 7.324049475454277e-06,
      "loss": 1.048,
      "step": 2791
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2465085983276367,
      "learning_rate": 7.3221703151364485e-06,
      "loss": 0.8092,
      "step": 2792
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3093421459197998,
      "learning_rate": 7.320290736511125e-06,
      "loss": 0.7601,
      "step": 2793
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2540929317474365,
      "learning_rate": 7.318410739916888e-06,
      "loss": 0.7673,
      "step": 2794
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2435048818588257,
      "learning_rate": 7.316530325692392e-06,
      "loss": 0.8471,
      "step": 2795
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4347034692764282,
      "learning_rate": 7.314649494176368e-06,
      "loss": 0.8211,
      "step": 2796
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7501163482666016,
      "learning_rate": 7.312768245707622e-06,
      "loss": 1.0405,
      "step": 2797
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2286430597305298,
      "learning_rate": 7.310886580625036e-06,
      "loss": 0.7863,
      "step": 2798
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.413932204246521,
      "learning_rate": 7.309004499267567e-06,
      "loss": 0.8114,
      "step": 2799
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.302642583847046,
      "learning_rate": 7.307122001974245e-06,
      "loss": 0.7092,
      "step": 2800
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.298126459121704,
      "learning_rate": 7.305239089084179e-06,
      "loss": 0.8066,
      "step": 2801
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.671734631061554,
      "learning_rate": 7.303355760936547e-06,
      "loss": 1.0708,
      "step": 2802
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6369605660438538,
      "learning_rate": 7.301472017870607e-06,
      "loss": 1.0424,
      "step": 2803
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2623143196105957,
      "learning_rate": 7.2995878602256885e-06,
      "loss": 0.7431,
      "step": 2804
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3337280750274658,
      "learning_rate": 7.297703288341199e-06,
      "loss": 0.7395,
      "step": 2805
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6127558350563049,
      "learning_rate": 7.295818302556617e-06,
      "loss": 1.0635,
      "step": 2806
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2759839296340942,
      "learning_rate": 7.293932903211497e-06,
      "loss": 0.7787,
      "step": 2807
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3310595750808716,
      "learning_rate": 7.292047090645469e-06,
      "loss": 0.7987,
      "step": 2808
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.805177092552185,
      "learning_rate": 7.290160865198234e-06,
      "loss": 0.7613,
      "step": 2809
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6932287812232971,
      "learning_rate": 7.288274227209575e-06,
      "loss": 1.0393,
      "step": 2810
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6416674256324768,
      "learning_rate": 7.286387177019341e-06,
      "loss": 1.0603,
      "step": 2811
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.212026596069336,
      "learning_rate": 7.284499714967458e-06,
      "loss": 0.8467,
      "step": 2812
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.332287311553955,
      "learning_rate": 7.2826118413939275e-06,
      "loss": 0.8023,
      "step": 2813
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3291869163513184,
      "learning_rate": 7.2807235566388245e-06,
      "loss": 0.8348,
      "step": 2814
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7149257063865662,
      "learning_rate": 7.278834861042299e-06,
      "loss": 1.0499,
      "step": 2815
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2019628286361694,
      "learning_rate": 7.276945754944571e-06,
      "loss": 0.8,
      "step": 2816
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3334742784500122,
      "learning_rate": 7.275056238685939e-06,
      "loss": 0.8496,
      "step": 2817
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2100878953933716,
      "learning_rate": 7.2731663126067745e-06,
      "loss": 0.7975,
      "step": 2818
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.272728681564331,
      "learning_rate": 7.271275977047521e-06,
      "loss": 0.8025,
      "step": 2819
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2340726852416992,
      "learning_rate": 7.269385232348697e-06,
      "loss": 0.8227,
      "step": 2820
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2655013799667358,
      "learning_rate": 7.267494078850894e-06,
      "loss": 0.8565,
      "step": 2821
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2044591903686523,
      "learning_rate": 7.265602516894778e-06,
      "loss": 0.8167,
      "step": 2822
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3189336061477661,
      "learning_rate": 7.26371054682109e-06,
      "loss": 0.7843,
      "step": 2823
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6755186319351196,
      "learning_rate": 7.261818168970642e-06,
      "loss": 1.0399,
      "step": 2824
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6335789561271667,
      "learning_rate": 7.259925383684318e-06,
      "loss": 1.0305,
      "step": 2825
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3072432279586792,
      "learning_rate": 7.25803219130308e-06,
      "loss": 0.789,
      "step": 2826
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6262878775596619,
      "learning_rate": 7.256138592167958e-06,
      "loss": 1.0328,
      "step": 2827
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2597988843917847,
      "learning_rate": 7.254244586620061e-06,
      "loss": 0.7512,
      "step": 2828
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2815335988998413,
      "learning_rate": 7.252350175000569e-06,
      "loss": 0.8158,
      "step": 2829
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.316455364227295,
      "learning_rate": 7.250455357650732e-06,
      "loss": 0.8163,
      "step": 2830
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2655013799667358,
      "learning_rate": 7.2485601349118775e-06,
      "loss": 0.7876,
      "step": 2831
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2898361682891846,
      "learning_rate": 7.246664507125403e-06,
      "loss": 0.8085,
      "step": 2832
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3070019483566284,
      "learning_rate": 7.244768474632782e-06,
      "loss": 0.7973,
      "step": 2833
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.228796362876892,
      "learning_rate": 7.242872037775557e-06,
      "loss": 0.7828,
      "step": 2834
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.892650842666626,
      "learning_rate": 7.240975196895346e-06,
      "loss": 0.775,
      "step": 2835
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3369938135147095,
      "learning_rate": 7.23907795233384e-06,
      "loss": 0.8472,
      "step": 2836
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.364446759223938,
      "learning_rate": 7.2371803044328e-06,
      "loss": 0.9008,
      "step": 2837
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2651785612106323,
      "learning_rate": 7.2352822535340635e-06,
      "loss": 0.7891,
      "step": 2838
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8147613406181335,
      "learning_rate": 7.233383799979537e-06,
      "loss": 1.0382,
      "step": 2839
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7481102347373962,
      "learning_rate": 7.2314849441112024e-06,
      "loss": 1.035,
      "step": 2840
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1638827323913574,
      "learning_rate": 7.2295856862711125e-06,
      "loss": 0.7881,
      "step": 2841
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.233296513557434,
      "learning_rate": 7.2276860268013925e-06,
      "loss": 0.764,
      "step": 2842
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7168582677841187,
      "learning_rate": 7.225785966044242e-06,
      "loss": 1.0318,
      "step": 2843
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.24905526638031,
      "learning_rate": 7.223885504341927e-06,
      "loss": 0.7283,
      "step": 2844
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7545636892318726,
      "learning_rate": 7.2219846420367956e-06,
      "loss": 1.0362,
      "step": 2845
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6969816088676453,
      "learning_rate": 7.2200833794712574e-06,
      "loss": 1.0389,
      "step": 2846
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6280759572982788,
      "learning_rate": 7.218181716987803e-06,
      "loss": 1.0627,
      "step": 2847
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3772320747375488,
      "learning_rate": 7.216279654928988e-06,
      "loss": 0.852,
      "step": 2848
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.706343412399292,
      "learning_rate": 7.2143771936374445e-06,
      "loss": 1.0484,
      "step": 2849
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3212953805923462,
      "learning_rate": 7.212474333455875e-06,
      "loss": 0.8461,
      "step": 2850
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2899967432022095,
      "learning_rate": 7.210571074727053e-06,
      "loss": 0.8413,
      "step": 2851
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3111859560012817,
      "learning_rate": 7.208667417793827e-06,
      "loss": 0.7825,
      "step": 2852
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2504732608795166,
      "learning_rate": 7.206763362999114e-06,
      "loss": 0.7549,
      "step": 2853
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8233475089073181,
      "learning_rate": 7.204858910685902e-06,
      "loss": 1.031,
      "step": 2854
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3163201808929443,
      "learning_rate": 7.202954061197254e-06,
      "loss": 0.8312,
      "step": 2855
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7262981534004211,
      "learning_rate": 7.201048814876303e-06,
      "loss": 1.0246,
      "step": 2856
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6573958992958069,
      "learning_rate": 7.199143172066252e-06,
      "loss": 1.0638,
      "step": 2857
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2146592140197754,
      "learning_rate": 7.197237133110377e-06,
      "loss": 0.7358,
      "step": 2858
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.358587622642517,
      "learning_rate": 7.195330698352026e-06,
      "loss": 0.838,
      "step": 2859
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2878419160842896,
      "learning_rate": 7.193423868134616e-06,
      "loss": 0.7739,
      "step": 2860
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8561172485351562,
      "learning_rate": 7.191516642801639e-06,
      "loss": 1.0639,
      "step": 2861
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.251637578010559,
      "learning_rate": 7.189609022696653e-06,
      "loss": 0.7694,
      "step": 2862
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.262097954750061,
      "learning_rate": 7.18770100816329e-06,
      "loss": 0.7959,
      "step": 2863
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2788474559783936,
      "learning_rate": 7.185792599545256e-06,
      "loss": 0.8587,
      "step": 2864
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7326523661613464,
      "learning_rate": 7.183883797186322e-06,
      "loss": 1.0591,
      "step": 2865
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3038229942321777,
      "learning_rate": 7.181974601430335e-06,
      "loss": 0.7874,
      "step": 2866
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5329389572143555,
      "learning_rate": 7.180065012621208e-06,
      "loss": 0.7988,
      "step": 2867
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6624301075935364,
      "learning_rate": 7.178155031102929e-06,
      "loss": 1.0438,
      "step": 2868
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2483928203582764,
      "learning_rate": 7.176244657219556e-06,
      "loss": 0.8503,
      "step": 2869
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2916971445083618,
      "learning_rate": 7.174333891315216e-06,
      "loss": 0.8077,
      "step": 2870
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6808496117591858,
      "learning_rate": 7.172422733734109e-06,
      "loss": 1.0293,
      "step": 2871
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2517554759979248,
      "learning_rate": 7.170511184820502e-06,
      "loss": 0.7986,
      "step": 2872
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6484936475753784,
      "learning_rate": 7.168599244918736e-06,
      "loss": 1.0442,
      "step": 2873
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.395126223564148,
      "learning_rate": 7.166686914373221e-06,
      "loss": 0.8133,
      "step": 2874
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.278907060623169,
      "learning_rate": 7.164774193528436e-06,
      "loss": 0.8679,
      "step": 2875
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6800207495689392,
      "learning_rate": 7.162861082728934e-06,
      "loss": 1.0426,
      "step": 2876
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2990829944610596,
      "learning_rate": 7.160947582319335e-06,
      "loss": 0.8441,
      "step": 2877
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2275421619415283,
      "learning_rate": 7.15903369264433e-06,
      "loss": 0.7383,
      "step": 2878
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.28658926486969,
      "learning_rate": 7.157119414048679e-06,
      "loss": 0.8172,
      "step": 2879
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6987205147743225,
      "learning_rate": 7.155204746877214e-06,
      "loss": 1.0423,
      "step": 2880
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6707462072372437,
      "learning_rate": 7.153289691474839e-06,
      "loss": 1.0322,
      "step": 2881
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2703933715820312,
      "learning_rate": 7.151374248186521e-06,
      "loss": 0.7983,
      "step": 2882
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6457143425941467,
      "learning_rate": 7.149458417357305e-06,
      "loss": 1.0465,
      "step": 2883
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.272589921951294,
      "learning_rate": 7.147542199332298e-06,
      "loss": 0.8942,
      "step": 2884
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2190887928009033,
      "learning_rate": 7.1456255944566856e-06,
      "loss": 0.8517,
      "step": 2885
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2863906621932983,
      "learning_rate": 7.1437086030757144e-06,
      "loss": 0.8543,
      "step": 2886
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7511225342750549,
      "learning_rate": 7.1417912255347054e-06,
      "loss": 1.0346,
      "step": 2887
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2447174787521362,
      "learning_rate": 7.139873462179048e-06,
      "loss": 0.8364,
      "step": 2888
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4096838235855103,
      "learning_rate": 7.137955313354202e-06,
      "loss": 0.8468,
      "step": 2889
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2546823024749756,
      "learning_rate": 7.136036779405695e-06,
      "loss": 0.7709,
      "step": 2890
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2373567819595337,
      "learning_rate": 7.1341178606791265e-06,
      "loss": 0.7898,
      "step": 2891
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.278057336807251,
      "learning_rate": 7.13219855752016e-06,
      "loss": 0.7303,
      "step": 2892
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2399970293045044,
      "learning_rate": 7.130278870274538e-06,
      "loss": 0.8052,
      "step": 2893
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7270461320877075,
      "learning_rate": 7.12835879928806e-06,
      "loss": 1.0397,
      "step": 2894
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6732439994812012,
      "learning_rate": 7.126438344906603e-06,
      "loss": 1.0553,
      "step": 2895
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2281203269958496,
      "learning_rate": 7.124517507476113e-06,
      "loss": 0.7622,
      "step": 2896
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6377540826797485,
      "learning_rate": 7.1225962873426e-06,
      "loss": 1.0332,
      "step": 2897
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.395918846130371,
      "learning_rate": 7.120674684852147e-06,
      "loss": 0.8719,
      "step": 2898
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6993288993835449,
      "learning_rate": 7.118752700350904e-06,
      "loss": 1.0331,
      "step": 2899
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6783736944198608,
      "learning_rate": 7.116830334185091e-06,
      "loss": 1.0354,
      "step": 2900
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.280563473701477,
      "learning_rate": 7.1149075867009965e-06,
      "loss": 0.8025,
      "step": 2901
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.201134204864502,
      "learning_rate": 7.112984458244976e-06,
      "loss": 0.7933,
      "step": 2902
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2782403230667114,
      "learning_rate": 7.111060949163456e-06,
      "loss": 0.7885,
      "step": 2903
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8726444244384766,
      "learning_rate": 7.109137059802931e-06,
      "loss": 1.0341,
      "step": 2904
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2689614295959473,
      "learning_rate": 7.107212790509962e-06,
      "loss": 0.7727,
      "step": 2905
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2571280002593994,
      "learning_rate": 7.1052881416311815e-06,
      "loss": 0.8151,
      "step": 2906
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2657428979873657,
      "learning_rate": 7.103363113513288e-06,
      "loss": 0.7362,
      "step": 2907
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7055531740188599,
      "learning_rate": 7.101437706503052e-06,
      "loss": 1.0503,
      "step": 2908
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.249312400817871,
      "learning_rate": 7.099511920947305e-06,
      "loss": 0.782,
      "step": 2909
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7112765312194824,
      "learning_rate": 7.0975857571929545e-06,
      "loss": 1.0425,
      "step": 2910
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2699865102767944,
      "learning_rate": 7.09565921558697e-06,
      "loss": 0.8229,
      "step": 2911
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2708817720413208,
      "learning_rate": 7.093732296476394e-06,
      "loss": 0.779,
      "step": 2912
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7032585740089417,
      "learning_rate": 7.091805000208336e-06,
      "loss": 1.051,
      "step": 2913
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2021958827972412,
      "learning_rate": 7.089877327129969e-06,
      "loss": 0.7792,
      "step": 2914
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.339658260345459,
      "learning_rate": 7.087949277588539e-06,
      "loss": 0.7827,
      "step": 2915
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4740779399871826,
      "learning_rate": 7.086020851931359e-06,
      "loss": 0.7459,
      "step": 2916
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3690189123153687,
      "learning_rate": 7.084092050505807e-06,
      "loss": 0.8035,
      "step": 2917
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2887986898422241,
      "learning_rate": 7.08216287365933e-06,
      "loss": 0.8247,
      "step": 2918
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.203320026397705,
      "learning_rate": 7.080233321739443e-06,
      "loss": 0.778,
      "step": 2919
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8038262724876404,
      "learning_rate": 7.078303395093731e-06,
      "loss": 1.0629,
      "step": 2920
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2425655126571655,
      "learning_rate": 7.076373094069842e-06,
      "loss": 0.798,
      "step": 2921
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6712220311164856,
      "learning_rate": 7.074442419015496e-06,
      "loss": 1.0225,
      "step": 2922
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3537408113479614,
      "learning_rate": 7.072511370278474e-06,
      "loss": 0.8278,
      "step": 2923
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2730181217193604,
      "learning_rate": 7.0705799482066305e-06,
      "loss": 0.7938,
      "step": 2924
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.29906165599823,
      "learning_rate": 7.068648153147886e-06,
      "loss": 0.8226,
      "step": 2925
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2719075679779053,
      "learning_rate": 7.0667159854502254e-06,
      "loss": 0.7954,
      "step": 2926
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.921220064163208,
      "learning_rate": 7.064783445461705e-06,
      "loss": 1.0343,
      "step": 2927
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2626194953918457,
      "learning_rate": 7.062850533530441e-06,
      "loss": 0.8129,
      "step": 2928
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.235610008239746,
      "learning_rate": 7.060917250004627e-06,
      "loss": 0.7692,
      "step": 2929
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3516318798065186,
      "learning_rate": 7.058983595232512e-06,
      "loss": 0.8577,
      "step": 2930
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7463011741638184,
      "learning_rate": 7.057049569562422e-06,
      "loss": 1.0618,
      "step": 2931
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2729295492172241,
      "learning_rate": 7.0551151733427446e-06,
      "loss": 0.7313,
      "step": 2932
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2721961736679077,
      "learning_rate": 7.053180406921934e-06,
      "loss": 0.8212,
      "step": 2933
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1774930953979492,
      "learning_rate": 7.051245270648514e-06,
      "loss": 0.755,
      "step": 2934
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6507377028465271,
      "learning_rate": 7.049309764871071e-06,
      "loss": 1.0311,
      "step": 2935
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.269537329673767,
      "learning_rate": 7.04737388993826e-06,
      "loss": 0.7837,
      "step": 2936
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3089678287506104,
      "learning_rate": 7.045437646198806e-06,
      "loss": 0.8698,
      "step": 2937
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3504327535629272,
      "learning_rate": 7.043501034001493e-06,
      "loss": 0.8033,
      "step": 2938
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6563245058059692,
      "learning_rate": 7.041564053695178e-06,
      "loss": 1.0398,
      "step": 2939
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6650951504707336,
      "learning_rate": 7.039626705628779e-06,
      "loss": 1.0479,
      "step": 2940
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.305060863494873,
      "learning_rate": 7.037688990151288e-06,
      "loss": 0.7928,
      "step": 2941
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2188221216201782,
      "learning_rate": 7.035750907611752e-06,
      "loss": 0.8106,
      "step": 2942
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6313353776931763,
      "learning_rate": 7.033812458359295e-06,
      "loss": 1.0528,
      "step": 2943
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.294593334197998,
      "learning_rate": 7.0318736427431e-06,
      "loss": 0.7997,
      "step": 2944
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6968923211097717,
      "learning_rate": 7.029934461112419e-06,
      "loss": 1.0171,
      "step": 2945
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2661726474761963,
      "learning_rate": 7.02799491381657e-06,
      "loss": 0.8634,
      "step": 2946
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2737492322921753,
      "learning_rate": 7.026055001204935e-06,
      "loss": 0.8401,
      "step": 2947
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2030562162399292,
      "learning_rate": 7.024114723626962e-06,
      "loss": 0.7733,
      "step": 2948
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.273163080215454,
      "learning_rate": 7.0221740814321695e-06,
      "loss": 0.7553,
      "step": 2949
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6441154479980469,
      "learning_rate": 7.020233074970134e-06,
      "loss": 1.0331,
      "step": 2950
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6825714707374573,
      "learning_rate": 7.018291704590503e-06,
      "loss": 1.0407,
      "step": 2951
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2561602592468262,
      "learning_rate": 7.0163499706429885e-06,
      "loss": 0.7289,
      "step": 2952
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3262830972671509,
      "learning_rate": 7.014407873477367e-06,
      "loss": 0.8127,
      "step": 2953
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3055301904678345,
      "learning_rate": 7.012465413443481e-06,
      "loss": 0.7947,
      "step": 2954
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.659146249294281,
      "learning_rate": 7.010522590891238e-06,
      "loss": 1.0287,
      "step": 2955
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6680479645729065,
      "learning_rate": 7.008579406170612e-06,
      "loss": 1.0372,
      "step": 2956
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6259925365447998,
      "learning_rate": 7.006635859631641e-06,
      "loss": 1.0439,
      "step": 2957
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2666460275650024,
      "learning_rate": 7.004691951624429e-06,
      "loss": 0.7912,
      "step": 2958
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6029664874076843,
      "learning_rate": 7.002747682499141e-06,
      "loss": 1.0385,
      "step": 2959
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2829580307006836,
      "learning_rate": 7.000803052606018e-06,
      "loss": 0.7501,
      "step": 2960
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2712130546569824,
      "learning_rate": 6.998858062295352e-06,
      "loss": 0.7608,
      "step": 2961
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6793698668479919,
      "learning_rate": 6.996912711917509e-06,
      "loss": 1.0423,
      "step": 2962
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.285145878791809,
      "learning_rate": 6.994967001822918e-06,
      "loss": 0.8547,
      "step": 2963
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2689495086669922,
      "learning_rate": 6.9930209323620735e-06,
      "loss": 0.8521,
      "step": 2964
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6258223652839661,
      "learning_rate": 6.99107450388553e-06,
      "loss": 1.0444,
      "step": 2965
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2638086080551147,
      "learning_rate": 6.989127716743913e-06,
      "loss": 0.8147,
      "step": 2966
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5012189149856567,
      "learning_rate": 6.987180571287908e-06,
      "loss": 0.7747,
      "step": 2967
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2371574640274048,
      "learning_rate": 6.985233067868269e-06,
      "loss": 0.8733,
      "step": 2968
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6582838296890259,
      "learning_rate": 6.98328520683581e-06,
      "loss": 1.0363,
      "step": 2969
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3467963933944702,
      "learning_rate": 6.981336988541414e-06,
      "loss": 0.8015,
      "step": 2970
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6363788843154907,
      "learning_rate": 6.979388413336025e-06,
      "loss": 1.0365,
      "step": 2971
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.274036169052124,
      "learning_rate": 6.977439481570651e-06,
      "loss": 0.7699,
      "step": 2972
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.483742594718933,
      "learning_rate": 6.9754901935963695e-06,
      "loss": 0.8166,
      "step": 2973
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6619237661361694,
      "learning_rate": 6.973540549764315e-06,
      "loss": 1.0279,
      "step": 2974
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6566389799118042,
      "learning_rate": 6.971590550425688e-06,
      "loss": 1.0428,
      "step": 2975
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2588512897491455,
      "learning_rate": 6.96964019593176e-06,
      "loss": 0.7717,
      "step": 2976
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2605153322219849,
      "learning_rate": 6.967689486633856e-06,
      "loss": 0.8307,
      "step": 2977
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6376224160194397,
      "learning_rate": 6.965738422883372e-06,
      "loss": 1.0351,
      "step": 2978
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1860902309417725,
      "learning_rate": 6.963787005031765e-06,
      "loss": 0.7892,
      "step": 2979
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2764273881912231,
      "learning_rate": 6.961835233430556e-06,
      "loss": 0.7561,
      "step": 2980
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6406945586204529,
      "learning_rate": 6.959883108431331e-06,
      "loss": 1.0215,
      "step": 2981
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2607821226119995,
      "learning_rate": 6.957930630385738e-06,
      "loss": 0.8576,
      "step": 2982
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2291147708892822,
      "learning_rate": 6.955977799645491e-06,
      "loss": 0.8276,
      "step": 2983
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2901140451431274,
      "learning_rate": 6.954024616562365e-06,
      "loss": 0.7873,
      "step": 2984
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.725656270980835,
      "learning_rate": 6.952071081488199e-06,
      "loss": 1.0461,
      "step": 2985
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6413031220436096,
      "learning_rate": 6.9501171947748955e-06,
      "loss": 1.0378,
      "step": 2986
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2119171619415283,
      "learning_rate": 6.9481629567744224e-06,
      "loss": 0.7717,
      "step": 2987
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2422460317611694,
      "learning_rate": 6.94620836783881e-06,
      "loss": 0.7756,
      "step": 2988
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.295392632484436,
      "learning_rate": 6.944253428320148e-06,
      "loss": 0.7636,
      "step": 2989
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7211912274360657,
      "learning_rate": 6.942298138570594e-06,
      "loss": 1.0286,
      "step": 2990
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6421566605567932,
      "learning_rate": 6.940342498942365e-06,
      "loss": 1.0382,
      "step": 2991
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4483098983764648,
      "learning_rate": 6.938386509787745e-06,
      "loss": 0.8299,
      "step": 2992
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3255491256713867,
      "learning_rate": 6.936430171459079e-06,
      "loss": 0.8083,
      "step": 2993
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6821433901786804,
      "learning_rate": 6.934473484308775e-06,
      "loss": 1.0222,
      "step": 2994
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3189318180084229,
      "learning_rate": 6.9325164486893015e-06,
      "loss": 0.8239,
      "step": 2995
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6739751100540161,
      "learning_rate": 6.930559064953194e-06,
      "loss": 1.0219,
      "step": 2996
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2131903171539307,
      "learning_rate": 6.928601333453049e-06,
      "loss": 0.8474,
      "step": 2997
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2766222953796387,
      "learning_rate": 6.926643254541522e-06,
      "loss": 0.7891,
      "step": 2998
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3135693073272705,
      "learning_rate": 6.924684828571338e-06,
      "loss": 0.8069,
      "step": 2999
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2510842084884644,
      "learning_rate": 6.922726055895278e-06,
      "loss": 0.8708,
      "step": 3000
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3722354173660278,
      "learning_rate": 6.92076693686619e-06,
      "loss": 0.8215,
      "step": 3001
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.261389970779419,
      "learning_rate": 6.918807471836985e-06,
      "loss": 0.8331,
      "step": 3002
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2709879875183105,
      "learning_rate": 6.916847661160628e-06,
      "loss": 0.8396,
      "step": 3003
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2455782890319824,
      "learning_rate": 6.9148875051901584e-06,
      "loss": 0.7684,
      "step": 3004
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3036270141601562,
      "learning_rate": 6.912927004278669e-06,
      "loss": 0.8257,
      "step": 3005
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2117453813552856,
      "learning_rate": 6.9109661587793176e-06,
      "loss": 0.8629,
      "step": 3006
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2767579555511475,
      "learning_rate": 6.909004969045323e-06,
      "loss": 0.7893,
      "step": 3007
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2999799251556396,
      "learning_rate": 6.9070434354299685e-06,
      "loss": 0.8224,
      "step": 3008
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2423070669174194,
      "learning_rate": 6.905081558286598e-06,
      "loss": 0.8119,
      "step": 3009
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.223305344581604,
      "learning_rate": 6.903119337968615e-06,
      "loss": 0.7912,
      "step": 3010
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2629191875457764,
      "learning_rate": 6.90115677482949e-06,
      "loss": 0.8049,
      "step": 3011
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2879396677017212,
      "learning_rate": 6.899193869222749e-06,
      "loss": 0.7846,
      "step": 3012
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8264934420585632,
      "learning_rate": 6.8972306215019846e-06,
      "loss": 1.0381,
      "step": 3013
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2444674968719482,
      "learning_rate": 6.895267032020849e-06,
      "loss": 0.8138,
      "step": 3014
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2016420364379883,
      "learning_rate": 6.893303101133055e-06,
      "loss": 0.8037,
      "step": 3015
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2382227182388306,
      "learning_rate": 6.89133882919238e-06,
      "loss": 0.8107,
      "step": 3016
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6618221402168274,
      "learning_rate": 6.889374216552661e-06,
      "loss": 1.0595,
      "step": 3017
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2552932500839233,
      "learning_rate": 6.887409263567795e-06,
      "loss": 0.8336,
      "step": 3018
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6080617904663086,
      "learning_rate": 6.885443970591741e-06,
      "loss": 1.0265,
      "step": 3019
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2942190170288086,
      "learning_rate": 6.883478337978522e-06,
      "loss": 0.8295,
      "step": 3020
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2747360467910767,
      "learning_rate": 6.881512366082219e-06,
      "loss": 0.8161,
      "step": 3021
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1785200834274292,
      "learning_rate": 6.8795460552569735e-06,
      "loss": 0.7742,
      "step": 3022
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6456913948059082,
      "learning_rate": 6.877579405856993e-06,
      "loss": 1.0225,
      "step": 3023
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2324023246765137,
      "learning_rate": 6.8756124182365405e-06,
      "loss": 0.7375,
      "step": 3024
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2907130718231201,
      "learning_rate": 6.8736450927499446e-06,
      "loss": 0.8009,
      "step": 3025
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2957773208618164,
      "learning_rate": 6.871677429751588e-06,
      "loss": 0.8162,
      "step": 3026
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1973052024841309,
      "learning_rate": 6.869709429595922e-06,
      "loss": 0.7868,
      "step": 3027
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2980901002883911,
      "learning_rate": 6.867741092637454e-06,
      "loss": 0.8135,
      "step": 3028
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3746294975280762,
      "learning_rate": 6.865772419230754e-06,
      "loss": 0.8191,
      "step": 3029
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2429563999176025,
      "learning_rate": 6.863803409730452e-06,
      "loss": 0.7987,
      "step": 3030
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2863705158233643,
      "learning_rate": 6.861834064491235e-06,
      "loss": 0.8226,
      "step": 3031
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6995349526405334,
      "learning_rate": 6.859864383867858e-06,
      "loss": 1.0408,
      "step": 3032
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2103617191314697,
      "learning_rate": 6.85789436821513e-06,
      "loss": 0.7792,
      "step": 3033
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2236779928207397,
      "learning_rate": 6.855924017887925e-06,
      "loss": 0.7966,
      "step": 3034
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2460095882415771,
      "learning_rate": 6.853953333241172e-06,
      "loss": 0.8742,
      "step": 3035
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6565505862236023,
      "learning_rate": 6.8519823146298645e-06,
      "loss": 1.0375,
      "step": 3036
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2102798223495483,
      "learning_rate": 6.850010962409055e-06,
      "loss": 0.7765,
      "step": 3037
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2209279537200928,
      "learning_rate": 6.848039276933855e-06,
      "loss": 0.7969,
      "step": 3038
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.289829134941101,
      "learning_rate": 6.8460672585594365e-06,
      "loss": 0.7811,
      "step": 3039
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2278474569320679,
      "learning_rate": 6.844094907641033e-06,
      "loss": 0.7921,
      "step": 3040
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6473862528800964,
      "learning_rate": 6.842122224533937e-06,
      "loss": 1.0408,
      "step": 3041
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2864595651626587,
      "learning_rate": 6.8401492095935e-06,
      "loss": 0.8706,
      "step": 3042
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3372489213943481,
      "learning_rate": 6.838175863175134e-06,
      "loss": 0.8794,
      "step": 3043
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3146395683288574,
      "learning_rate": 6.8362021856343104e-06,
      "loss": 0.852,
      "step": 3044
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2877943515777588,
      "learning_rate": 6.834228177326561e-06,
      "loss": 0.8388,
      "step": 3045
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.312042474746704,
      "learning_rate": 6.832253838607475e-06,
      "loss": 0.8146,
      "step": 3046
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.28412663936615,
      "learning_rate": 6.830279169832705e-06,
      "loss": 0.8155,
      "step": 3047
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2792332172393799,
      "learning_rate": 6.82830417135796e-06,
      "loss": 0.8465,
      "step": 3048
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2640810012817383,
      "learning_rate": 6.826328843539009e-06,
      "loss": 0.758,
      "step": 3049
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2998749017715454,
      "learning_rate": 6.824353186731681e-06,
      "loss": 0.7713,
      "step": 3050
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6792919039726257,
      "learning_rate": 6.822377201291863e-06,
      "loss": 1.032,
      "step": 3051
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2298908233642578,
      "learning_rate": 6.8204008875755024e-06,
      "loss": 0.8137,
      "step": 3052
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2080867290496826,
      "learning_rate": 6.818424245938605e-06,
      "loss": 0.8069,
      "step": 3053
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.310675859451294,
      "learning_rate": 6.816447276737236e-06,
      "loss": 0.8525,
      "step": 3054
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2166271209716797,
      "learning_rate": 6.814469980327519e-06,
      "loss": 0.8037,
      "step": 3055
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.275963306427002,
      "learning_rate": 6.812492357065641e-06,
      "loss": 0.7823,
      "step": 3056
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2036985158920288,
      "learning_rate": 6.810514407307838e-06,
      "loss": 0.8116,
      "step": 3057
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2537227869033813,
      "learning_rate": 6.808536131410416e-06,
      "loss": 0.7566,
      "step": 3058
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2405595779418945,
      "learning_rate": 6.8065575297297305e-06,
      "loss": 0.8,
      "step": 3059
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.277983546257019,
      "learning_rate": 6.804578602622202e-06,
      "loss": 0.7947,
      "step": 3060
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.309846043586731,
      "learning_rate": 6.802599350444306e-06,
      "loss": 0.8571,
      "step": 3061
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.686067521572113,
      "learning_rate": 6.800619773552582e-06,
      "loss": 1.0179,
      "step": 3062
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2584058046340942,
      "learning_rate": 6.798639872303618e-06,
      "loss": 0.7867,
      "step": 3063
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3397070169448853,
      "learning_rate": 6.7966596470540715e-06,
      "loss": 0.829,
      "step": 3064
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2090249061584473,
      "learning_rate": 6.794679098160649e-06,
      "loss": 0.8156,
      "step": 3065
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6613460183143616,
      "learning_rate": 6.792698225980122e-06,
      "loss": 1.0626,
      "step": 3066
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1932437419891357,
      "learning_rate": 6.790717030869319e-06,
      "loss": 0.7826,
      "step": 3067
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6111549139022827,
      "learning_rate": 6.788735513185122e-06,
      "loss": 1.0369,
      "step": 3068
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6286967396736145,
      "learning_rate": 6.786753673284476e-06,
      "loss": 1.0369,
      "step": 3069
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.625505805015564,
      "learning_rate": 6.784771511524384e-06,
      "loss": 1.0214,
      "step": 3070
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3232123851776123,
      "learning_rate": 6.7827890282619035e-06,
      "loss": 0.7984,
      "step": 3071
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2326470613479614,
      "learning_rate": 6.780806223854154e-06,
      "loss": 0.8192,
      "step": 3072
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2751215696334839,
      "learning_rate": 6.778823098658309e-06,
      "loss": 0.7834,
      "step": 3073
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6648225784301758,
      "learning_rate": 6.776839653031602e-06,
      "loss": 1.0415,
      "step": 3074
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.219928503036499,
      "learning_rate": 6.774855887331325e-06,
      "loss": 0.7637,
      "step": 3075
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3105446100234985,
      "learning_rate": 6.772871801914826e-06,
      "loss": 0.8155,
      "step": 3076
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2070902585983276,
      "learning_rate": 6.7708873971395095e-06,
      "loss": 0.7623,
      "step": 3077
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3026690483093262,
      "learning_rate": 6.768902673362841e-06,
      "loss": 0.8566,
      "step": 3078
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.260720133781433,
      "learning_rate": 6.766917630942341e-06,
      "loss": 0.8176,
      "step": 3079
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.229443907737732,
      "learning_rate": 6.764932270235587e-06,
      "loss": 0.7726,
      "step": 3080
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2643990516662598,
      "learning_rate": 6.762946591600218e-06,
      "loss": 0.7902,
      "step": 3081
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2345991134643555,
      "learning_rate": 6.7609605953939225e-06,
      "loss": 0.787,
      "step": 3082
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.234443187713623,
      "learning_rate": 6.758974281974454e-06,
      "loss": 0.8339,
      "step": 3083
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1957639455795288,
      "learning_rate": 6.756987651699618e-06,
      "loss": 0.7914,
      "step": 3084
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3276734352111816,
      "learning_rate": 6.755000704927281e-06,
      "loss": 0.8142,
      "step": 3085
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3385305404663086,
      "learning_rate": 6.753013442015365e-06,
      "loss": 0.7806,
      "step": 3086
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.273766279220581,
      "learning_rate": 6.751025863321845e-06,
      "loss": 0.8195,
      "step": 3087
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3958218097686768,
      "learning_rate": 6.749037969204759e-06,
      "loss": 0.8143,
      "step": 3088
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2924946546554565,
      "learning_rate": 6.747049760022197e-06,
      "loss": 0.7806,
      "step": 3089
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2119849920272827,
      "learning_rate": 6.745061236132311e-06,
      "loss": 0.757,
      "step": 3090
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7673492431640625,
      "learning_rate": 6.743072397893307e-06,
      "loss": 1.0408,
      "step": 3091
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.719059407711029,
      "learning_rate": 6.741083245663443e-06,
      "loss": 1.052,
      "step": 3092
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2910935878753662,
      "learning_rate": 6.739093779801039e-06,
      "loss": 0.7798,
      "step": 3093
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.283164381980896,
      "learning_rate": 6.737104000664474e-06,
      "loss": 0.7858,
      "step": 3094
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6957629323005676,
      "learning_rate": 6.735113908612176e-06,
      "loss": 1.0404,
      "step": 3095
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2210137844085693,
      "learning_rate": 6.733123504002635e-06,
      "loss": 0.8317,
      "step": 3096
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2576048374176025,
      "learning_rate": 6.731132787194394e-06,
      "loss": 0.772,
      "step": 3097
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7066184878349304,
      "learning_rate": 6.729141758546054e-06,
      "loss": 1.0377,
      "step": 3098
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.222085952758789,
      "learning_rate": 6.727150418416273e-06,
      "loss": 0.7523,
      "step": 3099
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2757476568222046,
      "learning_rate": 6.7251587671637616e-06,
      "loss": 0.7848,
      "step": 3100
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.237932562828064,
      "learning_rate": 6.723166805147292e-06,
      "loss": 0.7796,
      "step": 3101
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2072426080703735,
      "learning_rate": 6.721174532725686e-06,
      "loss": 0.8356,
      "step": 3102
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.343312382698059,
      "learning_rate": 6.719181950257826e-06,
      "loss": 0.8148,
      "step": 3103
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7215041518211365,
      "learning_rate": 6.717189058102649e-06,
      "loss": 1.0394,
      "step": 3104
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6871816515922546,
      "learning_rate": 6.715195856619146e-06,
      "loss": 1.0364,
      "step": 3105
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6066714525222778,
      "learning_rate": 6.713202346166368e-06,
      "loss": 1.0147,
      "step": 3106
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.282860517501831,
      "learning_rate": 6.7112085271034154e-06,
      "loss": 0.85,
      "step": 3107
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6682892441749573,
      "learning_rate": 6.70921439978945e-06,
      "loss": 1.0313,
      "step": 3108
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3403455018997192,
      "learning_rate": 6.707219964583688e-06,
      "loss": 0.8406,
      "step": 3109
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6974784135818481,
      "learning_rate": 6.705225221845395e-06,
      "loss": 1.0318,
      "step": 3110
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2629534006118774,
      "learning_rate": 6.703230171933902e-06,
      "loss": 0.7693,
      "step": 3111
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5550833940505981,
      "learning_rate": 6.701234815208587e-06,
      "loss": 0.762,
      "step": 3112
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3460700511932373,
      "learning_rate": 6.699239152028888e-06,
      "loss": 0.8405,
      "step": 3113
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.258400797843933,
      "learning_rate": 6.697243182754297e-06,
      "loss": 0.81,
      "step": 3114
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2016010284423828,
      "learning_rate": 6.69524690774436e-06,
      "loss": 0.7309,
      "step": 3115
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3213250637054443,
      "learning_rate": 6.693250327358678e-06,
      "loss": 0.8292,
      "step": 3116
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2663828134536743,
      "learning_rate": 6.691253441956909e-06,
      "loss": 0.7695,
      "step": 3117
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2077994346618652,
      "learning_rate": 6.689256251898763e-06,
      "loss": 0.8221,
      "step": 3118
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3388749361038208,
      "learning_rate": 6.687258757544012e-06,
      "loss": 0.8222,
      "step": 3119
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2886627912521362,
      "learning_rate": 6.6852609592524706e-06,
      "loss": 0.7951,
      "step": 3120
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3129054307937622,
      "learning_rate": 6.68326285738402e-06,
      "loss": 0.8544,
      "step": 3121
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3711023330688477,
      "learning_rate": 6.6812644522985885e-06,
      "loss": 0.7985,
      "step": 3122
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.306657314300537,
      "learning_rate": 6.679265744356164e-06,
      "loss": 0.695,
      "step": 3123
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3011280298233032,
      "learning_rate": 6.677266733916784e-06,
      "loss": 0.7906,
      "step": 3124
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.304780125617981,
      "learning_rate": 6.675267421340544e-06,
      "loss": 0.8432,
      "step": 3125
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2888740301132202,
      "learning_rate": 6.673267806987594e-06,
      "loss": 0.8134,
      "step": 3126
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.18796706199646,
      "learning_rate": 6.671267891218137e-06,
      "loss": 0.7539,
      "step": 3127
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.177770972251892,
      "learning_rate": 6.669267674392432e-06,
      "loss": 0.8054,
      "step": 3128
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2144043445587158,
      "learning_rate": 6.667267156870788e-06,
      "loss": 0.8192,
      "step": 3129
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2419956922531128,
      "learning_rate": 6.665266339013573e-06,
      "loss": 0.8099,
      "step": 3130
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8398460149765015,
      "learning_rate": 6.663265221181205e-06,
      "loss": 1.0481,
      "step": 3131
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2838134765625,
      "learning_rate": 6.661263803734163e-06,
      "loss": 0.7703,
      "step": 3132
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2848505973815918,
      "learning_rate": 6.659262087032972e-06,
      "loss": 0.8568,
      "step": 3133
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.281233310699463,
      "learning_rate": 6.657260071438213e-06,
      "loss": 0.7896,
      "step": 3134
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2199010848999023,
      "learning_rate": 6.6552577573105236e-06,
      "loss": 0.8499,
      "step": 3135
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.213002324104309,
      "learning_rate": 6.653255145010594e-06,
      "loss": 0.8127,
      "step": 3136
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2472434043884277,
      "learning_rate": 6.651252234899167e-06,
      "loss": 0.7573,
      "step": 3137
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.664616584777832,
      "learning_rate": 6.64924902733704e-06,
      "loss": 1.057,
      "step": 3138
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2084764242172241,
      "learning_rate": 6.647245522685062e-06,
      "loss": 0.8094,
      "step": 3139
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2984660863876343,
      "learning_rate": 6.6452417213041394e-06,
      "loss": 0.77,
      "step": 3140
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6580636501312256,
      "learning_rate": 6.643237623555228e-06,
      "loss": 1.0588,
      "step": 3141
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3445175886154175,
      "learning_rate": 6.641233229799342e-06,
      "loss": 0.7536,
      "step": 3142
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6823092103004456,
      "learning_rate": 6.639228540397542e-06,
      "loss": 1.0293,
      "step": 3143
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2250745296478271,
      "learning_rate": 6.6372235557109475e-06,
      "loss": 0.8396,
      "step": 3144
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6876745223999023,
      "learning_rate": 6.6352182761007286e-06,
      "loss": 1.0437,
      "step": 3145
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3081833124160767,
      "learning_rate": 6.63321270192811e-06,
      "loss": 0.79,
      "step": 3146
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.267523169517517,
      "learning_rate": 6.6312068335543685e-06,
      "loss": 0.7873,
      "step": 3147
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3817673921585083,
      "learning_rate": 6.629200671340834e-06,
      "loss": 0.8472,
      "step": 3148
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2394827604293823,
      "learning_rate": 6.627194215648889e-06,
      "loss": 0.7041,
      "step": 3149
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3052968978881836,
      "learning_rate": 6.625187466839969e-06,
      "loss": 0.8329,
      "step": 3150
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2597897052764893,
      "learning_rate": 6.623180425275565e-06,
      "loss": 0.8601,
      "step": 3151
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2288211584091187,
      "learning_rate": 6.621173091317216e-06,
      "loss": 0.7578,
      "step": 3152
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3129562139511108,
      "learning_rate": 6.619165465326516e-06,
      "loss": 0.8079,
      "step": 3153
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9645841121673584,
      "learning_rate": 6.617157547665112e-06,
      "loss": 1.0226,
      "step": 3154
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.749131202697754,
      "learning_rate": 6.615149338694704e-06,
      "loss": 0.7648,
      "step": 3155
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2471275329589844,
      "learning_rate": 6.613140838777041e-06,
      "loss": 0.8151,
      "step": 3156
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3300126791000366,
      "learning_rate": 6.611132048273932e-06,
      "loss": 0.7919,
      "step": 3157
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3162893056869507,
      "learning_rate": 6.609122967547228e-06,
      "loss": 0.8135,
      "step": 3158
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2786118984222412,
      "learning_rate": 6.607113596958841e-06,
      "loss": 0.7938,
      "step": 3159
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4264636039733887,
      "learning_rate": 6.605103936870731e-06,
      "loss": 0.8428,
      "step": 3160
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3694344758987427,
      "learning_rate": 6.60309398764491e-06,
      "loss": 0.8429,
      "step": 3161
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.724994957447052,
      "learning_rate": 6.601083749643444e-06,
      "loss": 1.027,
      "step": 3162
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.324989676475525,
      "learning_rate": 6.599073223228449e-06,
      "loss": 0.8548,
      "step": 3163
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6545519232749939,
      "learning_rate": 6.5970624087620975e-06,
      "loss": 1.0717,
      "step": 3164
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3482511043548584,
      "learning_rate": 6.595051306606606e-06,
      "loss": 0.8167,
      "step": 3165
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2772353887557983,
      "learning_rate": 6.593039917124251e-06,
      "loss": 0.7903,
      "step": 3166
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.269319772720337,
      "learning_rate": 6.591028240677355e-06,
      "loss": 0.7603,
      "step": 3167
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2134469747543335,
      "learning_rate": 6.589016277628294e-06,
      "loss": 0.7362,
      "step": 3168
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7312302589416504,
      "learning_rate": 6.587004028339496e-06,
      "loss": 1.0506,
      "step": 3169
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3653862476348877,
      "learning_rate": 6.584991493173443e-06,
      "loss": 0.8022,
      "step": 3170
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2580318450927734,
      "learning_rate": 6.582978672492662e-06,
      "loss": 0.8169,
      "step": 3171
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3540300130844116,
      "learning_rate": 6.580965566659739e-06,
      "loss": 0.8708,
      "step": 3172
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6914825439453125,
      "learning_rate": 6.5789521760373035e-06,
      "loss": 1.0113,
      "step": 3173
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2602736949920654,
      "learning_rate": 6.576938500988045e-06,
      "loss": 0.8407,
      "step": 3174
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6579210758209229,
      "learning_rate": 6.5749245418746965e-06,
      "loss": 1.0569,
      "step": 3175
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3390798568725586,
      "learning_rate": 6.5729102990600466e-06,
      "loss": 0.8008,
      "step": 3176
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6806230545043945,
      "learning_rate": 6.5708957729069355e-06,
      "loss": 1.0568,
      "step": 3177
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6574374437332153,
      "learning_rate": 6.568880963778249e-06,
      "loss": 1.023,
      "step": 3178
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.271749258041382,
      "learning_rate": 6.56686587203693e-06,
      "loss": 0.7847,
      "step": 3179
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3595277070999146,
      "learning_rate": 6.564850498045971e-06,
      "loss": 0.8297,
      "step": 3180
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3582509756088257,
      "learning_rate": 6.562834842168413e-06,
      "loss": 0.8362,
      "step": 3181
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.240303635597229,
      "learning_rate": 6.560818904767349e-06,
      "loss": 0.7847,
      "step": 3182
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2677065134048462,
      "learning_rate": 6.558802686205922e-06,
      "loss": 0.8352,
      "step": 3183
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6636248826980591,
      "learning_rate": 6.556786186847328e-06,
      "loss": 1.0423,
      "step": 3184
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2368227243423462,
      "learning_rate": 6.554769407054811e-06,
      "loss": 0.8097,
      "step": 3185
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2415138483047485,
      "learning_rate": 6.552752347191667e-06,
      "loss": 0.8316,
      "step": 3186
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.25095796585083,
      "learning_rate": 6.5507350076212405e-06,
      "loss": 0.7833,
      "step": 3187
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2606003284454346,
      "learning_rate": 6.548717388706931e-06,
      "loss": 0.8071,
      "step": 3188
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3161183595657349,
      "learning_rate": 6.546699490812184e-06,
      "loss": 0.8378,
      "step": 3189
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.726943850517273,
      "learning_rate": 6.544681314300494e-06,
      "loss": 1.0455,
      "step": 3190
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3213789463043213,
      "learning_rate": 6.542662859535411e-06,
      "loss": 0.8139,
      "step": 3191
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2387535572052002,
      "learning_rate": 6.540644126880531e-06,
      "loss": 0.8084,
      "step": 3192
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1623170375823975,
      "learning_rate": 6.5386251166995006e-06,
      "loss": 0.7383,
      "step": 3193
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2690472602844238,
      "learning_rate": 6.53660582935602e-06,
      "loss": 0.761,
      "step": 3194
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2702816724777222,
      "learning_rate": 6.534586265213834e-06,
      "loss": 0.8377,
      "step": 3195
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6578128933906555,
      "learning_rate": 6.53256642463674e-06,
      "loss": 1.0451,
      "step": 3196
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2632137537002563,
      "learning_rate": 6.530546307988586e-06,
      "loss": 0.8113,
      "step": 3197
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.272098183631897,
      "learning_rate": 6.528525915633267e-06,
      "loss": 0.7883,
      "step": 3198
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1938425302505493,
      "learning_rate": 6.526505247934729e-06,
      "loss": 0.7862,
      "step": 3199
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3401010036468506,
      "learning_rate": 6.5244843052569696e-06,
      "loss": 0.8226,
      "step": 3200
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2502731084823608,
      "learning_rate": 6.522463087964033e-06,
      "loss": 0.7554,
      "step": 3201
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.706699788570404,
      "learning_rate": 6.520441596420015e-06,
      "loss": 1.0352,
      "step": 3202
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2334201335906982,
      "learning_rate": 6.51841983098906e-06,
      "loss": 0.7716,
      "step": 3203
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2969826459884644,
      "learning_rate": 6.516397792035361e-06,
      "loss": 0.7822,
      "step": 3204
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.320380449295044,
      "learning_rate": 6.51437547992316e-06,
      "loss": 0.8269,
      "step": 3205
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3197284936904907,
      "learning_rate": 6.512352895016751e-06,
      "loss": 0.7413,
      "step": 3206
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6694859862327576,
      "learning_rate": 6.510330037680475e-06,
      "loss": 1.0609,
      "step": 3207
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2874412536621094,
      "learning_rate": 6.5083069082787205e-06,
      "loss": 0.8309,
      "step": 3208
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2557868957519531,
      "learning_rate": 6.506283507175929e-06,
      "loss": 0.7728,
      "step": 3209
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.220211148262024,
      "learning_rate": 6.504259834736587e-06,
      "loss": 0.8425,
      "step": 3210
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6540219187736511,
      "learning_rate": 6.502235891325233e-06,
      "loss": 1.028,
      "step": 3211
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6506996750831604,
      "learning_rate": 6.500211677306451e-06,
      "loss": 1.0233,
      "step": 3212
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2928248643875122,
      "learning_rate": 6.4981871930448795e-06,
      "loss": 0.8081,
      "step": 3213
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6097484230995178,
      "learning_rate": 6.496162438905198e-06,
      "loss": 1.035,
      "step": 3214
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.278699278831482,
      "learning_rate": 6.49413741525214e-06,
      "loss": 0.8109,
      "step": 3215
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2124871015548706,
      "learning_rate": 6.492112122450487e-06,
      "loss": 0.791,
      "step": 3216
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4186934232711792,
      "learning_rate": 6.4900865608650666e-06,
      "loss": 0.8478,
      "step": 3217
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.273672103881836,
      "learning_rate": 6.488060730860757e-06,
      "loss": 0.7472,
      "step": 3218
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8144898414611816,
      "learning_rate": 6.486034632802484e-06,
      "loss": 1.0314,
      "step": 3219
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7920759320259094,
      "learning_rate": 6.48400826705522e-06,
      "loss": 1.0333,
      "step": 3220
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2282843589782715,
      "learning_rate": 6.481981633983989e-06,
      "loss": 0.7794,
      "step": 3221
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2887787818908691,
      "learning_rate": 6.479954733953864e-06,
      "loss": 0.8386,
      "step": 3222
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3221962451934814,
      "learning_rate": 6.477927567329958e-06,
      "loss": 0.8108,
      "step": 3223
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3527534008026123,
      "learning_rate": 6.475900134477441e-06,
      "loss": 0.7529,
      "step": 3224
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8097109794616699,
      "learning_rate": 6.473872435761527e-06,
      "loss": 1.0204,
      "step": 3225
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2529935836791992,
      "learning_rate": 6.471844471547479e-06,
      "loss": 0.7895,
      "step": 3226
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2513978481292725,
      "learning_rate": 6.469816242200606e-06,
      "loss": 0.8262,
      "step": 3227
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2719911336898804,
      "learning_rate": 6.467787748086266e-06,
      "loss": 0.7843,
      "step": 3228
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3003512620925903,
      "learning_rate": 6.465758989569866e-06,
      "loss": 0.8782,
      "step": 3229
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7109234929084778,
      "learning_rate": 6.463729967016858e-06,
      "loss": 1.0586,
      "step": 3230
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6469705104827881,
      "learning_rate": 6.461700680792745e-06,
      "loss": 1.0596,
      "step": 3231
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2416937351226807,
      "learning_rate": 6.459671131263073e-06,
      "loss": 0.7914,
      "step": 3232
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2933175563812256,
      "learning_rate": 6.4576413187934394e-06,
      "loss": 0.8608,
      "step": 3233
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2727285623550415,
      "learning_rate": 6.455611243749486e-06,
      "loss": 0.7893,
      "step": 3234
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3284027576446533,
      "learning_rate": 6.453580906496905e-06,
      "loss": 0.7782,
      "step": 3235
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.27144193649292,
      "learning_rate": 6.451550307401432e-06,
      "loss": 0.8317,
      "step": 3236
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.245699405670166,
      "learning_rate": 6.449519446828855e-06,
      "loss": 0.8005,
      "step": 3237
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8403286337852478,
      "learning_rate": 6.4474883251450015e-06,
      "loss": 1.0382,
      "step": 3238
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1758794784545898,
      "learning_rate": 6.445456942715756e-06,
      "loss": 0.7856,
      "step": 3239
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2422194480895996,
      "learning_rate": 6.4434252999070406e-06,
      "loss": 0.7354,
      "step": 3240
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.351044774055481,
      "learning_rate": 6.441393397084827e-06,
      "loss": 0.7704,
      "step": 3241
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7038664817810059,
      "learning_rate": 6.439361234615139e-06,
      "loss": 1.023,
      "step": 3242
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.285452127456665,
      "learning_rate": 6.437328812864041e-06,
      "loss": 0.8182,
      "step": 3243
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2906708717346191,
      "learning_rate": 6.435296132197647e-06,
      "loss": 0.8489,
      "step": 3244
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2374320030212402,
      "learning_rate": 6.433263192982116e-06,
      "loss": 0.7945,
      "step": 3245
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7030126452445984,
      "learning_rate": 6.431229995583656e-06,
      "loss": 1.0456,
      "step": 3246
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2357062101364136,
      "learning_rate": 6.429196540368516e-06,
      "loss": 0.7927,
      "step": 3247
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2219781875610352,
      "learning_rate": 6.427162827702999e-06,
      "loss": 0.7928,
      "step": 3248
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.263859748840332,
      "learning_rate": 6.425128857953451e-06,
      "loss": 0.8047,
      "step": 3249
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2390060424804688,
      "learning_rate": 6.423094631486264e-06,
      "loss": 0.8241,
      "step": 3250
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6943132877349854,
      "learning_rate": 6.421060148667874e-06,
      "loss": 1.0367,
      "step": 3251
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.278285026550293,
      "learning_rate": 6.419025409864766e-06,
      "loss": 0.8472,
      "step": 3252
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4094160795211792,
      "learning_rate": 6.416990415443471e-06,
      "loss": 0.8598,
      "step": 3253
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2265691757202148,
      "learning_rate": 6.414955165770569e-06,
      "loss": 0.773,
      "step": 3254
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2281330823898315,
      "learning_rate": 6.4129196612126765e-06,
      "loss": 0.7558,
      "step": 3255
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6095368266105652,
      "learning_rate": 6.4108839021364665e-06,
      "loss": 1.0148,
      "step": 3256
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.255950689315796,
      "learning_rate": 6.4088478889086525e-06,
      "loss": 0.7709,
      "step": 3257
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2895550727844238,
      "learning_rate": 6.406811621895994e-06,
      "loss": 0.7627,
      "step": 3258
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.306342363357544,
      "learning_rate": 6.404775101465298e-06,
      "loss": 0.79,
      "step": 3259
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2568204402923584,
      "learning_rate": 6.402738327983413e-06,
      "loss": 0.7735,
      "step": 3260
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6480817198753357,
      "learning_rate": 6.40070130181724e-06,
      "loss": 1.0228,
      "step": 3261
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2717936038970947,
      "learning_rate": 6.398664023333719e-06,
      "loss": 0.7391,
      "step": 3262
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3422527313232422,
      "learning_rate": 6.39662649289984e-06,
      "loss": 0.8743,
      "step": 3263
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.34555983543396,
      "learning_rate": 6.394588710882635e-06,
      "loss": 0.7846,
      "step": 3264
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6539196372032166,
      "learning_rate": 6.392550677649183e-06,
      "loss": 1.05,
      "step": 3265
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6264281868934631,
      "learning_rate": 6.390512393566609e-06,
      "loss": 1.0173,
      "step": 3266
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5920825004577637,
      "learning_rate": 6.388473859002082e-06,
      "loss": 1.0268,
      "step": 3267
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3168489933013916,
      "learning_rate": 6.3864350743228165e-06,
      "loss": 0.8043,
      "step": 3268
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6218502521514893,
      "learning_rate": 6.38439603989607e-06,
      "loss": 1.0531,
      "step": 3269
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2571439743041992,
      "learning_rate": 6.382356756089149e-06,
      "loss": 0.7759,
      "step": 3270
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6279118657112122,
      "learning_rate": 6.380317223269401e-06,
      "loss": 1.0504,
      "step": 3271
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3194977045059204,
      "learning_rate": 6.3782774418042216e-06,
      "loss": 0.854,
      "step": 3272
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2148983478546143,
      "learning_rate": 6.376237412061051e-06,
      "loss": 0.7892,
      "step": 3273
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2016135454177856,
      "learning_rate": 6.37419713440737e-06,
      "loss": 0.7663,
      "step": 3274
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2229334115982056,
      "learning_rate": 6.372156609210709e-06,
      "loss": 0.786,
      "step": 3275
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2355068922042847,
      "learning_rate": 6.370115836838642e-06,
      "loss": 0.7825,
      "step": 3276
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8798609375953674,
      "learning_rate": 6.368074817658783e-06,
      "loss": 1.029,
      "step": 3277
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3079439401626587,
      "learning_rate": 6.366033552038796e-06,
      "loss": 0.8635,
      "step": 3278
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2999154329299927,
      "learning_rate": 6.363992040346388e-06,
      "loss": 0.8712,
      "step": 3279
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3076856136322021,
      "learning_rate": 6.361950282949306e-06,
      "loss": 0.8031,
      "step": 3280
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2671738862991333,
      "learning_rate": 6.359908280215349e-06,
      "loss": 0.8303,
      "step": 3281
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6922428011894226,
      "learning_rate": 6.357866032512355e-06,
      "loss": 1.0335,
      "step": 3282
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1918790340423584,
      "learning_rate": 6.355823540208207e-06,
      "loss": 0.8064,
      "step": 3283
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2342265844345093,
      "learning_rate": 6.3537808036708314e-06,
      "loss": 0.8331,
      "step": 3284
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2507531642913818,
      "learning_rate": 6.3517378232682005e-06,
      "loss": 0.8004,
      "step": 3285
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3462183475494385,
      "learning_rate": 6.34969459936833e-06,
      "loss": 0.795,
      "step": 3286
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2294597625732422,
      "learning_rate": 6.347651132339279e-06,
      "loss": 0.7542,
      "step": 3287
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6969483494758606,
      "learning_rate": 6.345607422549148e-06,
      "loss": 1.0309,
      "step": 3288
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1934531927108765,
      "learning_rate": 6.3435634703660855e-06,
      "loss": 0.7719,
      "step": 3289
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2664304971694946,
      "learning_rate": 6.341519276158282e-06,
      "loss": 0.804,
      "step": 3290
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3136844635009766,
      "learning_rate": 6.339474840293971e-06,
      "loss": 0.7832,
      "step": 3291
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3345879316329956,
      "learning_rate": 6.337430163141431e-06,
      "loss": 0.8415,
      "step": 3292
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2697932720184326,
      "learning_rate": 6.335385245068979e-06,
      "loss": 0.7789,
      "step": 3293
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3372738361358643,
      "learning_rate": 6.333340086444984e-06,
      "loss": 0.8301,
      "step": 3294
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7299793362617493,
      "learning_rate": 6.33129468763785e-06,
      "loss": 1.0586,
      "step": 3295
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3484272956848145,
      "learning_rate": 6.329249049016032e-06,
      "loss": 0.8623,
      "step": 3296
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6669674515724182,
      "learning_rate": 6.32720317094802e-06,
      "loss": 1.0558,
      "step": 3297
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6287546157836914,
      "learning_rate": 6.325157053802354e-06,
      "loss": 1.0085,
      "step": 3298
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2137460708618164,
      "learning_rate": 6.323110697947611e-06,
      "loss": 0.7515,
      "step": 3299
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.257572889328003,
      "learning_rate": 6.321064103752415e-06,
      "loss": 0.7435,
      "step": 3300
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.268349051475525,
      "learning_rate": 6.319017271585437e-06,
      "loss": 0.8624,
      "step": 3301
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2532165050506592,
      "learning_rate": 6.31697020181538e-06,
      "loss": 0.7955,
      "step": 3302
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.388105034828186,
      "learning_rate": 6.314922894810999e-06,
      "loss": 0.8349,
      "step": 3303
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.99912428855896,
      "learning_rate": 6.312875350941088e-06,
      "loss": 1.0287,
      "step": 3304
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8152183294296265,
      "learning_rate": 6.310827570574484e-06,
      "loss": 1.0316,
      "step": 3305
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6450175046920776,
      "learning_rate": 6.3087795540800664e-06,
      "loss": 1.0575,
      "step": 3306
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4192891120910645,
      "learning_rate": 6.306731301826759e-06,
      "loss": 0.7829,
      "step": 3307
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8579431176185608,
      "learning_rate": 6.304682814183526e-06,
      "loss": 1.0342,
      "step": 3308
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8754482269287109,
      "learning_rate": 6.302634091519374e-06,
      "loss": 1.0484,
      "step": 3309
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.272590160369873,
      "learning_rate": 6.300585134203354e-06,
      "loss": 0.7528,
      "step": 3310
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2649457454681396,
      "learning_rate": 6.298535942604556e-06,
      "loss": 0.7734,
      "step": 3311
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2167613506317139,
      "learning_rate": 6.296486517092116e-06,
      "loss": 0.7908,
      "step": 3312
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2518635988235474,
      "learning_rate": 6.2944368580352085e-06,
      "loss": 0.7487,
      "step": 3313
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7532116174697876,
      "learning_rate": 6.292386965803054e-06,
      "loss": 1.0483,
      "step": 3314
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7220349311828613,
      "learning_rate": 6.290336840764913e-06,
      "loss": 1.0178,
      "step": 3315
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2664216756820679,
      "learning_rate": 6.288286483290082e-06,
      "loss": 0.7715,
      "step": 3316
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2690086364746094,
      "learning_rate": 6.286235893747913e-06,
      "loss": 0.7958,
      "step": 3317
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2576954364776611,
      "learning_rate": 6.284185072507786e-06,
      "loss": 0.8236,
      "step": 3318
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2900464534759521,
      "learning_rate": 6.28213401993913e-06,
      "loss": 0.8142,
      "step": 3319
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2256873846054077,
      "learning_rate": 6.280082736411418e-06,
      "loss": 0.7934,
      "step": 3320
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2719151973724365,
      "learning_rate": 6.278031222294157e-06,
      "loss": 0.7828,
      "step": 3321
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4428848028182983,
      "learning_rate": 6.275979477956899e-06,
      "loss": 0.7894,
      "step": 3322
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3023536205291748,
      "learning_rate": 6.2739275037692394e-06,
      "loss": 0.8204,
      "step": 3323
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3042066097259521,
      "learning_rate": 6.2718753001008145e-06,
      "loss": 0.7971,
      "step": 3324
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.302716612815857,
      "learning_rate": 6.2698228673213005e-06,
      "loss": 0.7532,
      "step": 3325
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.349428415298462,
      "learning_rate": 6.2677702058004145e-06,
      "loss": 0.8358,
      "step": 3326
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2808505296707153,
      "learning_rate": 6.265717315907915e-06,
      "loss": 0.7633,
      "step": 3327
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7773606777191162,
      "learning_rate": 6.263664198013604e-06,
      "loss": 1.0466,
      "step": 3328
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2529302835464478,
      "learning_rate": 6.261610852487323e-06,
      "loss": 0.7722,
      "step": 3329
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3057674169540405,
      "learning_rate": 6.259557279698952e-06,
      "loss": 0.8154,
      "step": 3330
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2904725074768066,
      "learning_rate": 6.257503480018417e-06,
      "loss": 0.7611,
      "step": 3331
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3948614597320557,
      "learning_rate": 6.255449453815679e-06,
      "loss": 0.7814,
      "step": 3332
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3008207082748413,
      "learning_rate": 6.253395201460746e-06,
      "loss": 0.7273,
      "step": 3333
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3719993829727173,
      "learning_rate": 6.251340723323663e-06,
      "loss": 0.8688,
      "step": 3334
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7720692157745361,
      "learning_rate": 6.249286019774517e-06,
      "loss": 1.0161,
      "step": 3335
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2283843755722046,
      "learning_rate": 6.247231091183432e-06,
      "loss": 0.829,
      "step": 3336
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2398498058319092,
      "learning_rate": 6.245175937920579e-06,
      "loss": 0.8367,
      "step": 3337
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2658958435058594,
      "learning_rate": 6.2431205603561626e-06,
      "loss": 0.8121,
      "step": 3338
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2480111122131348,
      "learning_rate": 6.241064958860435e-06,
      "loss": 0.864,
      "step": 3339
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2439261674880981,
      "learning_rate": 6.239009133803682e-06,
      "loss": 0.782,
      "step": 3340
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2019920349121094,
      "learning_rate": 6.2369530855562335e-06,
      "loss": 0.7389,
      "step": 3341
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7928239703178406,
      "learning_rate": 6.234896814488459e-06,
      "loss": 1.0441,
      "step": 3342
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.210228681564331,
      "learning_rate": 6.232840320970769e-06,
      "loss": 0.8353,
      "step": 3343
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6598984599113464,
      "learning_rate": 6.2307836053736096e-06,
      "loss": 1.0245,
      "step": 3344
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2703683376312256,
      "learning_rate": 6.228726668067473e-06,
      "loss": 0.7816,
      "step": 3345
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2603676319122314,
      "learning_rate": 6.2266695094228866e-06,
      "loss": 0.7821,
      "step": 3346
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3751237392425537,
      "learning_rate": 6.224612129810422e-06,
      "loss": 0.8047,
      "step": 3347
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.747105598449707,
      "learning_rate": 6.222554529600689e-06,
      "loss": 1.0362,
      "step": 3348
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3374944925308228,
      "learning_rate": 6.220496709164332e-06,
      "loss": 0.8308,
      "step": 3349
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2984471321105957,
      "learning_rate": 6.218438668872042e-06,
      "loss": 0.8259,
      "step": 3350
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.371522068977356,
      "learning_rate": 6.2163804090945465e-06,
      "loss": 0.8086,
      "step": 3351
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2277151346206665,
      "learning_rate": 6.214321930202615e-06,
      "loss": 0.8328,
      "step": 3352
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2530419826507568,
      "learning_rate": 6.212263232567053e-06,
      "loss": 0.8117,
      "step": 3353
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2976771593093872,
      "learning_rate": 6.210204316558707e-06,
      "loss": 0.8339,
      "step": 3354
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2032679319381714,
      "learning_rate": 6.208145182548462e-06,
      "loss": 0.8081,
      "step": 3355
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2155694961547852,
      "learning_rate": 6.2060858309072445e-06,
      "loss": 0.764,
      "step": 3356
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2220360040664673,
      "learning_rate": 6.204026262006017e-06,
      "loss": 0.7952,
      "step": 3357
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7348740696907043,
      "learning_rate": 6.201966476215788e-06,
      "loss": 1.0543,
      "step": 3358
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.691826581954956,
      "learning_rate": 6.199906473907593e-06,
      "loss": 1.0462,
      "step": 3359
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6162759065628052,
      "learning_rate": 6.197846255452516e-06,
      "loss": 1.0312,
      "step": 3360
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.285890817642212,
      "learning_rate": 6.195785821221679e-06,
      "loss": 0.8159,
      "step": 3361
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6504791975021362,
      "learning_rate": 6.193725171586241e-06,
      "loss": 1.0507,
      "step": 3362
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2750651836395264,
      "learning_rate": 6.191664306917399e-06,
      "loss": 0.8566,
      "step": 3363
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4217071533203125,
      "learning_rate": 6.189603227586389e-06,
      "loss": 0.8365,
      "step": 3364
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2249138355255127,
      "learning_rate": 6.187541933964489e-06,
      "loss": 0.8,
      "step": 3365
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2768375873565674,
      "learning_rate": 6.18548042642301e-06,
      "loss": 0.7586,
      "step": 3366
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.313559651374817,
      "learning_rate": 6.1834187053333095e-06,
      "loss": 0.8199,
      "step": 3367
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.259647011756897,
      "learning_rate": 6.181356771066773e-06,
      "loss": 0.7551,
      "step": 3368
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2728840112686157,
      "learning_rate": 6.179294623994833e-06,
      "loss": 0.8002,
      "step": 3369
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.260987401008606,
      "learning_rate": 6.177232264488958e-06,
      "loss": 0.7631,
      "step": 3370
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3557298183441162,
      "learning_rate": 6.175169692920653e-06,
      "loss": 0.8691,
      "step": 3371
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2618305683135986,
      "learning_rate": 6.173106909661461e-06,
      "loss": 0.8589,
      "step": 3372
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1635563373565674,
      "learning_rate": 6.171043915082967e-06,
      "loss": 0.7891,
      "step": 3373
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3459616899490356,
      "learning_rate": 6.168980709556789e-06,
      "loss": 0.8011,
      "step": 3374
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2873926162719727,
      "learning_rate": 6.1669172934545875e-06,
      "loss": 0.7521,
      "step": 3375
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3447515964508057,
      "learning_rate": 6.164853667148059e-06,
      "loss": 0.8386,
      "step": 3376
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.322892665863037,
      "learning_rate": 6.162789831008935e-06,
      "loss": 0.8191,
      "step": 3377
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2995227575302124,
      "learning_rate": 6.160725785408991e-06,
      "loss": 0.7864,
      "step": 3378
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3553348779678345,
      "learning_rate": 6.158661530720035e-06,
      "loss": 0.8387,
      "step": 3379
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9102464318275452,
      "learning_rate": 6.156597067313914e-06,
      "loss": 1.0385,
      "step": 3380
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.302348256111145,
      "learning_rate": 6.154532395562515e-06,
      "loss": 0.7803,
      "step": 3381
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7089121341705322,
      "learning_rate": 6.152467515837759e-06,
      "loss": 1.042,
      "step": 3382
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3002712726593018,
      "learning_rate": 6.150402428511607e-06,
      "loss": 0.7892,
      "step": 3383
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2656620740890503,
      "learning_rate": 6.1483371339560545e-06,
      "loss": 0.7649,
      "step": 3384
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.710185170173645,
      "learning_rate": 6.146271632543142e-06,
      "loss": 1.0173,
      "step": 3385
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6984202861785889,
      "learning_rate": 6.144205924644934e-06,
      "loss": 1.0247,
      "step": 3386
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3199387788772583,
      "learning_rate": 6.1421400106335435e-06,
      "loss": 0.7445,
      "step": 3387
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6579257845878601,
      "learning_rate": 6.140073890881117e-06,
      "loss": 1.0177,
      "step": 3388
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3330281972885132,
      "learning_rate": 6.138007565759838e-06,
      "loss": 0.8023,
      "step": 3389
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.291419267654419,
      "learning_rate": 6.135941035641927e-06,
      "loss": 0.7535,
      "step": 3390
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8234078884124756,
      "learning_rate": 6.133874300899641e-06,
      "loss": 1.0331,
      "step": 3391
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3416975736618042,
      "learning_rate": 6.1318073619052735e-06,
      "loss": 0.7844,
      "step": 3392
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.313602089881897,
      "learning_rate": 6.129740219031156e-06,
      "loss": 0.8045,
      "step": 3393
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2534157037734985,
      "learning_rate": 6.127672872649657e-06,
      "loss": 0.7785,
      "step": 3394
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7402162551879883,
      "learning_rate": 6.12560532313318e-06,
      "loss": 1.0474,
      "step": 3395
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.730133056640625,
      "learning_rate": 6.123537570854168e-06,
      "loss": 1.0418,
      "step": 3396
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.201479196548462,
      "learning_rate": 6.121469616185096e-06,
      "loss": 0.8143,
      "step": 3397
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2655866146087646,
      "learning_rate": 6.119401459498479e-06,
      "loss": 0.8159,
      "step": 3398
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.237989902496338,
      "learning_rate": 6.117333101166867e-06,
      "loss": 0.782,
      "step": 3399
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6814628839492798,
      "learning_rate": 6.115264541562847e-06,
      "loss": 1.04,
      "step": 3400
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3031883239746094,
      "learning_rate": 6.1131957810590425e-06,
      "loss": 0.815,
      "step": 3401
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2150652408599854,
      "learning_rate": 6.1111268200281115e-06,
      "loss": 0.7831,
      "step": 3402
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.36676824092865,
      "learning_rate": 6.1090576588427495e-06,
      "loss": 0.7634,
      "step": 3403
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2485196590423584,
      "learning_rate": 6.106988297875691e-06,
      "loss": 0.7791,
      "step": 3404
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2320643663406372,
      "learning_rate": 6.104918737499699e-06,
      "loss": 0.7516,
      "step": 3405
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2477840185165405,
      "learning_rate": 6.102848978087576e-06,
      "loss": 0.7685,
      "step": 3406
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.224299430847168,
      "learning_rate": 6.100779020012167e-06,
      "loss": 0.8004,
      "step": 3407
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7354417443275452,
      "learning_rate": 6.0987088636463406e-06,
      "loss": 1.0273,
      "step": 3408
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.372863531112671,
      "learning_rate": 6.096638509363013e-06,
      "loss": 0.8002,
      "step": 3409
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2926284074783325,
      "learning_rate": 6.0945679575351255e-06,
      "loss": 0.8149,
      "step": 3410
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3141741752624512,
      "learning_rate": 6.092497208535662e-06,
      "loss": 0.7987,
      "step": 3411
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.401727557182312,
      "learning_rate": 6.090426262737639e-06,
      "loss": 0.8799,
      "step": 3412
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2418791055679321,
      "learning_rate": 6.088355120514113e-06,
      "loss": 0.8,
      "step": 3413
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3231621980667114,
      "learning_rate": 6.086283782238168e-06,
      "loss": 0.8522,
      "step": 3414
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2262018918991089,
      "learning_rate": 6.084212248282928e-06,
      "loss": 0.7591,
      "step": 3415
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2337464094161987,
      "learning_rate": 6.0821405190215535e-06,
      "loss": 0.7935,
      "step": 3416
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3476173877716064,
      "learning_rate": 6.080068594827238e-06,
      "loss": 0.8224,
      "step": 3417
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2773865461349487,
      "learning_rate": 6.0779964760732095e-06,
      "loss": 0.7778,
      "step": 3418
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6974600553512573,
      "learning_rate": 6.075924163132734e-06,
      "loss": 0.7467,
      "step": 3419
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7964766621589661,
      "learning_rate": 6.073851656379107e-06,
      "loss": 1.0366,
      "step": 3420
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.238128900527954,
      "learning_rate": 6.071778956185666e-06,
      "loss": 0.7749,
      "step": 3421
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2156047821044922,
      "learning_rate": 6.069706062925777e-06,
      "loss": 0.8152,
      "step": 3422
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3306854963302612,
      "learning_rate": 6.067632976972848e-06,
      "loss": 0.7871,
      "step": 3423
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.252988576889038,
      "learning_rate": 6.065559698700312e-06,
      "loss": 0.7753,
      "step": 3424
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6273494958877563,
      "learning_rate": 6.063486228481645e-06,
      "loss": 1.019,
      "step": 3425
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.627089262008667,
      "learning_rate": 6.061412566690355e-06,
      "loss": 1.0328,
      "step": 3426
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.335551142692566,
      "learning_rate": 6.059338713699983e-06,
      "loss": 0.7706,
      "step": 3427
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3169434070587158,
      "learning_rate": 6.057264669884105e-06,
      "loss": 0.8208,
      "step": 3428
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6439719796180725,
      "learning_rate": 6.055190435616333e-06,
      "loss": 1.0635,
      "step": 3429
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2298423051834106,
      "learning_rate": 6.053116011270311e-06,
      "loss": 0.833,
      "step": 3430
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6344558596611023,
      "learning_rate": 6.051041397219719e-06,
      "loss": 1.0437,
      "step": 3431
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2381891012191772,
      "learning_rate": 6.048966593838272e-06,
      "loss": 0.7883,
      "step": 3432
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2315516471862793,
      "learning_rate": 6.0468916014997156e-06,
      "loss": 0.7683,
      "step": 3433
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2786287069320679,
      "learning_rate": 6.0448164205778315e-06,
      "loss": 0.8275,
      "step": 3434
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2310160398483276,
      "learning_rate": 6.042741051446437e-06,
      "loss": 0.8405,
      "step": 3435
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.223327875137329,
      "learning_rate": 6.040665494479381e-06,
      "loss": 0.8889,
      "step": 3436
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2118993997573853,
      "learning_rate": 6.038589750050546e-06,
      "loss": 0.7844,
      "step": 3437
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.710473895072937,
      "learning_rate": 6.036513818533852e-06,
      "loss": 1.0145,
      "step": 3438
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.279862880706787,
      "learning_rate": 6.034437700303246e-06,
      "loss": 0.7309,
      "step": 3439
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6608559489250183,
      "learning_rate": 6.0323613957327156e-06,
      "loss": 1.0318,
      "step": 3440
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2931371927261353,
      "learning_rate": 6.0302849051962765e-06,
      "loss": 0.784,
      "step": 3441
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3072267770767212,
      "learning_rate": 6.028208229067983e-06,
      "loss": 0.8236,
      "step": 3442
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2545052766799927,
      "learning_rate": 6.026131367721919e-06,
      "loss": 0.7721,
      "step": 3443
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2557082176208496,
      "learning_rate": 6.024054321532201e-06,
      "loss": 0.8294,
      "step": 3444
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2484010457992554,
      "learning_rate": 6.021977090872983e-06,
      "loss": 0.7848,
      "step": 3445
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.233152151107788,
      "learning_rate": 6.0198996761184515e-06,
      "loss": 0.7626,
      "step": 3446
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.221070408821106,
      "learning_rate": 6.017822077642821e-06,
      "loss": 0.854,
      "step": 3447
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.226967453956604,
      "learning_rate": 6.015744295820343e-06,
      "loss": 0.8025,
      "step": 3448
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2086797952651978,
      "learning_rate": 6.013666331025303e-06,
      "loss": 0.7422,
      "step": 3449
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.282947063446045,
      "learning_rate": 6.011588183632019e-06,
      "loss": 0.7946,
      "step": 3450
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2504196166992188,
      "learning_rate": 6.009509854014841e-06,
      "loss": 0.8352,
      "step": 3451
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2593642473220825,
      "learning_rate": 6.0074313425481504e-06,
      "loss": 0.8079,
      "step": 3452
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2744452953338623,
      "learning_rate": 6.005352649606362e-06,
      "loss": 0.7941,
      "step": 3453
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3871760368347168,
      "learning_rate": 6.003273775563927e-06,
      "loss": 0.7863,
      "step": 3454
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2963820695877075,
      "learning_rate": 6.001194720795324e-06,
      "loss": 0.8801,
      "step": 3455
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2341135740280151,
      "learning_rate": 5.999115485675069e-06,
      "loss": 0.7741,
      "step": 3456
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3108747005462646,
      "learning_rate": 5.997036070577705e-06,
      "loss": 0.7785,
      "step": 3457
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.266507863998413,
      "learning_rate": 5.994956475877814e-06,
      "loss": 0.8026,
      "step": 3458
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2467762231826782,
      "learning_rate": 5.992876701950003e-06,
      "loss": 0.7825,
      "step": 3459
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2291215658187866,
      "learning_rate": 5.990796749168919e-06,
      "loss": 0.7766,
      "step": 3460
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2319931983947754,
      "learning_rate": 5.988716617909233e-06,
      "loss": 0.8117,
      "step": 3461
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2947547435760498,
      "learning_rate": 5.986636308545655e-06,
      "loss": 0.8746,
      "step": 3462
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3665294647216797,
      "learning_rate": 5.984555821452923e-06,
      "loss": 0.8097,
      "step": 3463
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8128418326377869,
      "learning_rate": 5.982475157005812e-06,
      "loss": 1.0387,
      "step": 3464
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7218847274780273,
      "learning_rate": 5.9803943155791225e-06,
      "loss": 1.0384,
      "step": 3465
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6495229005813599,
      "learning_rate": 5.978313297547688e-06,
      "loss": 1.0129,
      "step": 3466
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3723807334899902,
      "learning_rate": 5.9762321032863816e-06,
      "loss": 0.8185,
      "step": 3467
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3550467491149902,
      "learning_rate": 5.974150733170096e-06,
      "loss": 0.7994,
      "step": 3468
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2581965923309326,
      "learning_rate": 5.972069187573766e-06,
      "loss": 0.7668,
      "step": 3469
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2693536281585693,
      "learning_rate": 5.969987466872354e-06,
      "loss": 0.8156,
      "step": 3470
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2126158475875854,
      "learning_rate": 5.967905571440849e-06,
      "loss": 0.8255,
      "step": 3471
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2836854457855225,
      "learning_rate": 5.9658235016542806e-06,
      "loss": 0.7904,
      "step": 3472
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2690929174423218,
      "learning_rate": 5.963741257887704e-06,
      "loss": 0.7711,
      "step": 3473
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.327437162399292,
      "learning_rate": 5.961658840516209e-06,
      "loss": 0.8412,
      "step": 3474
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2367446422576904,
      "learning_rate": 5.959576249914913e-06,
      "loss": 0.7255,
      "step": 3475
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.373458743095398,
      "learning_rate": 5.957493486458965e-06,
      "loss": 1.0505,
      "step": 3476
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3083288669586182,
      "learning_rate": 5.95541055052355e-06,
      "loss": 0.8384,
      "step": 3477
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3564646244049072,
      "learning_rate": 5.953327442483878e-06,
      "loss": 0.7751,
      "step": 3478
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3539491891860962,
      "learning_rate": 5.951244162715194e-06,
      "loss": 0.7883,
      "step": 3479
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7257155776023865,
      "learning_rate": 5.949160711592773e-06,
      "loss": 1.0292,
      "step": 3480
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3239428997039795,
      "learning_rate": 5.947077089491919e-06,
      "loss": 0.7582,
      "step": 3481
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3277117013931274,
      "learning_rate": 5.9449932967879686e-06,
      "loss": 0.8398,
      "step": 3482
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7369264960289001,
      "learning_rate": 5.9429093338562895e-06,
      "loss": 1.0439,
      "step": 3483
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2788331508636475,
      "learning_rate": 5.94082520107228e-06,
      "loss": 0.7433,
      "step": 3484
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6314229369163513,
      "learning_rate": 5.9387408988113656e-06,
      "loss": 1.0419,
      "step": 3485
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6185646057128906,
      "learning_rate": 5.936656427449009e-06,
      "loss": 1.015,
      "step": 3486
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2640678882598877,
      "learning_rate": 5.934571787360696e-06,
      "loss": 0.795,
      "step": 3487
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3258198499679565,
      "learning_rate": 5.932486978921948e-06,
      "loss": 0.8003,
      "step": 3488
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3256105184555054,
      "learning_rate": 5.930402002508315e-06,
      "loss": 0.7711,
      "step": 3489
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2449619770050049,
      "learning_rate": 5.928316858495377e-06,
      "loss": 0.8253,
      "step": 3490
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.241935133934021,
      "learning_rate": 5.926231547258743e-06,
      "loss": 0.7223,
      "step": 3491
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9366984963417053,
      "learning_rate": 5.924146069174055e-06,
      "loss": 1.0374,
      "step": 3492
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.30239737033844,
      "learning_rate": 5.922060424616985e-06,
      "loss": 0.7712,
      "step": 3493
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7135071158409119,
      "learning_rate": 5.91997461396323e-06,
      "loss": 1.0253,
      "step": 3494
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3849388360977173,
      "learning_rate": 5.9178886375885235e-06,
      "loss": 0.8352,
      "step": 3495
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.705193042755127,
      "learning_rate": 5.915802495868624e-06,
      "loss": 1.0428,
      "step": 3496
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.180841326713562,
      "learning_rate": 5.913716189179324e-06,
      "loss": 0.7991,
      "step": 3497
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8035397529602051,
      "learning_rate": 5.911629717896441e-06,
      "loss": 1.0479,
      "step": 3498
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.228406548500061,
      "learning_rate": 5.909543082395826e-06,
      "loss": 0.7763,
      "step": 3499
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2595696449279785,
      "learning_rate": 5.907456283053357e-06,
      "loss": 0.7955,
      "step": 3500
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.304161787033081,
      "learning_rate": 5.905369320244942e-06,
      "loss": 0.8254,
      "step": 3501
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3358783721923828,
      "learning_rate": 5.903282194346522e-06,
      "loss": 0.8001,
      "step": 3502
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2588107585906982,
      "learning_rate": 5.9011949057340624e-06,
      "loss": 0.7741,
      "step": 3503
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.217016577720642,
      "learning_rate": 5.899107454783558e-06,
      "loss": 0.7757,
      "step": 3504
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7688717842102051,
      "learning_rate": 5.897019841871039e-06,
      "loss": 1.0246,
      "step": 3505
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2766872644424438,
      "learning_rate": 5.894932067372556e-06,
      "loss": 0.7819,
      "step": 3506
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7199055552482605,
      "learning_rate": 5.892844131664199e-06,
      "loss": 1.0203,
      "step": 3507
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2592405080795288,
      "learning_rate": 5.890756035122076e-06,
      "loss": 0.8656,
      "step": 3508
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6636464595794678,
      "learning_rate": 5.888667778122331e-06,
      "loss": 1.0593,
      "step": 3509
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.644812822341919,
      "learning_rate": 5.886579361041136e-06,
      "loss": 1.0258,
      "step": 3510
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3056974411010742,
      "learning_rate": 5.884490784254691e-06,
      "loss": 0.791,
      "step": 3511
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6644850373268127,
      "learning_rate": 5.882402048139224e-06,
      "loss": 1.0345,
      "step": 3512
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3173491954803467,
      "learning_rate": 5.880313153070992e-06,
      "loss": 0.8446,
      "step": 3513
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.278624415397644,
      "learning_rate": 5.878224099426283e-06,
      "loss": 0.7513,
      "step": 3514
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4196677207946777,
      "learning_rate": 5.87613488758141e-06,
      "loss": 0.7304,
      "step": 3515
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.241102933883667,
      "learning_rate": 5.874045517912717e-06,
      "loss": 0.7528,
      "step": 3516
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3503435850143433,
      "learning_rate": 5.871955990796575e-06,
      "loss": 0.8173,
      "step": 3517
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2672683000564575,
      "learning_rate": 5.869866306609385e-06,
      "loss": 0.7649,
      "step": 3518
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2173774242401123,
      "learning_rate": 5.867776465727574e-06,
      "loss": 0.8081,
      "step": 3519
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2644431591033936,
      "learning_rate": 5.865686468527599e-06,
      "loss": 0.792,
      "step": 3520
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2485631704330444,
      "learning_rate": 5.863596315385947e-06,
      "loss": 0.827,
      "step": 3521
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7921561598777771,
      "learning_rate": 5.861506006679125e-06,
      "loss": 1.0344,
      "step": 3522
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7135121822357178,
      "learning_rate": 5.859415542783678e-06,
      "loss": 1.0224,
      "step": 3523
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.255558967590332,
      "learning_rate": 5.857324924076175e-06,
      "loss": 0.7687,
      "step": 3524
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3525739908218384,
      "learning_rate": 5.8552341509332125e-06,
      "loss": 0.7948,
      "step": 3525
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2800489664077759,
      "learning_rate": 5.8531432237314135e-06,
      "loss": 0.7576,
      "step": 3526
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2976096868515015,
      "learning_rate": 5.85105214284743e-06,
      "loss": 0.7882,
      "step": 3527
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7403242588043213,
      "learning_rate": 5.8489609086579425e-06,
      "loss": 1.0349,
      "step": 3528
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.236203908920288,
      "learning_rate": 5.846869521539659e-06,
      "loss": 0.8171,
      "step": 3529
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.645074188709259,
      "learning_rate": 5.844777981869315e-06,
      "loss": 1.0465,
      "step": 3530
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2328791618347168,
      "learning_rate": 5.842686290023671e-06,
      "loss": 0.7792,
      "step": 3531
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3612656593322754,
      "learning_rate": 5.840594446379518e-06,
      "loss": 0.8415,
      "step": 3532
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2742613554000854,
      "learning_rate": 5.838502451313675e-06,
      "loss": 0.8065,
      "step": 3533
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3375805616378784,
      "learning_rate": 5.836410305202983e-06,
      "loss": 0.8291,
      "step": 3534
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2233620882034302,
      "learning_rate": 5.834318008424316e-06,
      "loss": 0.7692,
      "step": 3535
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.296444058418274,
      "learning_rate": 5.832225561354573e-06,
      "loss": 0.7853,
      "step": 3536
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3012555837631226,
      "learning_rate": 5.8301329643706794e-06,
      "loss": 0.7893,
      "step": 3537
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3002946376800537,
      "learning_rate": 5.828040217849589e-06,
      "loss": 0.8382,
      "step": 3538
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.733623743057251,
      "learning_rate": 5.825947322168281e-06,
      "loss": 1.0246,
      "step": 3539
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.239983320236206,
      "learning_rate": 5.823854277703761e-06,
      "loss": 0.783,
      "step": 3540
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.240835428237915,
      "learning_rate": 5.821761084833064e-06,
      "loss": 0.7433,
      "step": 3541
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7002512812614441,
      "learning_rate": 5.819667743933251e-06,
      "loss": 1.0366,
      "step": 3542
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3481582403182983,
      "learning_rate": 5.8175742553814065e-06,
      "loss": 0.7548,
      "step": 3543
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2270174026489258,
      "learning_rate": 5.815480619554646e-06,
      "loss": 0.7943,
      "step": 3544
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2870142459869385,
      "learning_rate": 5.81338683683011e-06,
      "loss": 0.7811,
      "step": 3545
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3206472396850586,
      "learning_rate": 5.811292907584963e-06,
      "loss": 0.8286,
      "step": 3546
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3997230529785156,
      "learning_rate": 5.809198832196401e-06,
      "loss": 0.8033,
      "step": 3547
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3987468481063843,
      "learning_rate": 5.80710461104164e-06,
      "loss": 0.8507,
      "step": 3548
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3314509391784668,
      "learning_rate": 5.805010244497928e-06,
      "loss": 0.8096,
      "step": 3549
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.268099069595337,
      "learning_rate": 5.802915732942536e-06,
      "loss": 0.7889,
      "step": 3550
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2916936874389648,
      "learning_rate": 5.800821076752761e-06,
      "loss": 0.7707,
      "step": 3551
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8361548781394958,
      "learning_rate": 5.798726276305927e-06,
      "loss": 1.041,
      "step": 3552
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3970431089401245,
      "learning_rate": 5.796631331979385e-06,
      "loss": 0.8325,
      "step": 3553
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3105114698410034,
      "learning_rate": 5.794536244150512e-06,
      "loss": 0.7727,
      "step": 3554
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.216665267944336,
      "learning_rate": 5.792441013196706e-06,
      "loss": 0.8238,
      "step": 3555
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2820628881454468,
      "learning_rate": 5.790345639495396e-06,
      "loss": 0.8147,
      "step": 3556
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6779828071594238,
      "learning_rate": 5.788250123424037e-06,
      "loss": 1.0483,
      "step": 3557
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.718799352645874,
      "learning_rate": 5.786154465360107e-06,
      "loss": 1.0302,
      "step": 3558
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3020066022872925,
      "learning_rate": 5.784058665681108e-06,
      "loss": 0.7854,
      "step": 3559
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2817944288253784,
      "learning_rate": 5.781962724764572e-06,
      "loss": 0.7585,
      "step": 3560
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.320773959159851,
      "learning_rate": 5.779866642988055e-06,
      "loss": 0.7577,
      "step": 3561
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2883119583129883,
      "learning_rate": 5.777770420729135e-06,
      "loss": 0.8,
      "step": 3562
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6846409440040588,
      "learning_rate": 5.775674058365423e-06,
      "loss": 1.0188,
      "step": 3563
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2700852155685425,
      "learning_rate": 5.773577556274543e-06,
      "loss": 0.8258,
      "step": 3564
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6808598637580872,
      "learning_rate": 5.771480914834158e-06,
      "loss": 1.0505,
      "step": 3565
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3673462867736816,
      "learning_rate": 5.769384134421945e-06,
      "loss": 0.7969,
      "step": 3566
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.297351598739624,
      "learning_rate": 5.767287215415613e-06,
      "loss": 0.7727,
      "step": 3567
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6512701511383057,
      "learning_rate": 5.765190158192895e-06,
      "loss": 1.0387,
      "step": 3568
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6141085028648376,
      "learning_rate": 5.7630929631315424e-06,
      "loss": 1.0198,
      "step": 3569
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6210313439369202,
      "learning_rate": 5.76099563060934e-06,
      "loss": 1.0164,
      "step": 3570
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2290221452713013,
      "learning_rate": 5.758898161004095e-06,
      "loss": 0.7582,
      "step": 3571
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2399038076400757,
      "learning_rate": 5.756800554693636e-06,
      "loss": 0.7976,
      "step": 3572
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.647184431552887,
      "learning_rate": 5.754702812055819e-06,
      "loss": 1.0211,
      "step": 3573
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3206439018249512,
      "learning_rate": 5.752604933468523e-06,
      "loss": 0.774,
      "step": 3574
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6628044247627258,
      "learning_rate": 5.750506919309652e-06,
      "loss": 1.0255,
      "step": 3575
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.201798915863037,
      "learning_rate": 5.748408769957137e-06,
      "loss": 0.7687,
      "step": 3576
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2565959692001343,
      "learning_rate": 5.746310485788931e-06,
      "loss": 0.8065,
      "step": 3577
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.306565523147583,
      "learning_rate": 5.744212067183008e-06,
      "loss": 0.8345,
      "step": 3578
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2765146493911743,
      "learning_rate": 5.742113514517375e-06,
      "loss": 0.8171,
      "step": 3579
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6632492542266846,
      "learning_rate": 5.740014828170053e-06,
      "loss": 1.0438,
      "step": 3580
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2487698793411255,
      "learning_rate": 5.737916008519093e-06,
      "loss": 0.7199,
      "step": 3581
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1944934129714966,
      "learning_rate": 5.735817055942572e-06,
      "loss": 0.8064,
      "step": 3582
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4005860090255737,
      "learning_rate": 5.733717970818584e-06,
      "loss": 0.8597,
      "step": 3583
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2215381860733032,
      "learning_rate": 5.731618753525252e-06,
      "loss": 0.8035,
      "step": 3584
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2571256160736084,
      "learning_rate": 5.729519404440722e-06,
      "loss": 0.7847,
      "step": 3585
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6156482100486755,
      "learning_rate": 5.727419923943163e-06,
      "loss": 1.0418,
      "step": 3586
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2609963417053223,
      "learning_rate": 5.7253203124107695e-06,
      "loss": 0.8412,
      "step": 3587
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.608493447303772,
      "learning_rate": 5.7232205702217545e-06,
      "loss": 1.0268,
      "step": 3588
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6265937685966492,
      "learning_rate": 5.721120697754361e-06,
      "loss": 1.0405,
      "step": 3589
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2142817974090576,
      "learning_rate": 5.719020695386851e-06,
      "loss": 0.7488,
      "step": 3590
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.264206051826477,
      "learning_rate": 5.716920563497514e-06,
      "loss": 0.8041,
      "step": 3591
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2801730632781982,
      "learning_rate": 5.714820302464656e-06,
      "loss": 0.805,
      "step": 3592
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3522926568984985,
      "learning_rate": 5.712719912666615e-06,
      "loss": 0.8777,
      "step": 3593
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1303608417510986,
      "learning_rate": 5.710619394481745e-06,
      "loss": 0.7586,
      "step": 3594
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.228682279586792,
      "learning_rate": 5.708518748288425e-06,
      "loss": 0.7924,
      "step": 3595
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2779197692871094,
      "learning_rate": 5.706417974465061e-06,
      "loss": 0.7765,
      "step": 3596
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7575573325157166,
      "learning_rate": 5.7043170733900775e-06,
      "loss": 1.0436,
      "step": 3597
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7217233180999756,
      "learning_rate": 5.702216045441924e-06,
      "loss": 1.0565,
      "step": 3598
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3374043703079224,
      "learning_rate": 5.700114890999071e-06,
      "loss": 0.781,
      "step": 3599
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.616685688495636,
      "learning_rate": 5.698013610440012e-06,
      "loss": 1.0346,
      "step": 3600
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3070385456085205,
      "learning_rate": 5.695912204143268e-06,
      "loss": 0.804,
      "step": 3601
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4049899578094482,
      "learning_rate": 5.6938106724873745e-06,
      "loss": 0.7577,
      "step": 3602
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.264022946357727,
      "learning_rate": 5.691709015850897e-06,
      "loss": 0.7545,
      "step": 3603
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.241397500038147,
      "learning_rate": 5.689607234612418e-06,
      "loss": 0.8072,
      "step": 3604
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9325631260871887,
      "learning_rate": 5.687505329150548e-06,
      "loss": 1.0232,
      "step": 3605
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.333619236946106,
      "learning_rate": 5.685403299843914e-06,
      "loss": 0.8021,
      "step": 3606
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2459945678710938,
      "learning_rate": 5.68330114707117e-06,
      "loss": 0.8203,
      "step": 3607
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3694214820861816,
      "learning_rate": 5.681198871210988e-06,
      "loss": 0.8075,
      "step": 3608
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.68352872133255,
      "learning_rate": 5.679096472642066e-06,
      "loss": 1.0452,
      "step": 3609
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2311230897903442,
      "learning_rate": 5.676993951743124e-06,
      "loss": 0.7526,
      "step": 3610
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1650457382202148,
      "learning_rate": 5.6748913088929005e-06,
      "loss": 0.7641,
      "step": 3611
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2223988771438599,
      "learning_rate": 5.6727885444701584e-06,
      "loss": 0.7354,
      "step": 3612
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2729475498199463,
      "learning_rate": 5.670685658853682e-06,
      "loss": 0.8322,
      "step": 3613
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2817542552947998,
      "learning_rate": 5.6685826524222785e-06,
      "loss": 0.7654,
      "step": 3614
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2345688343048096,
      "learning_rate": 5.666479525554777e-06,
      "loss": 0.8005,
      "step": 3615
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2466816902160645,
      "learning_rate": 5.664376278630025e-06,
      "loss": 0.7907,
      "step": 3616
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3106924295425415,
      "learning_rate": 5.662272912026894e-06,
      "loss": 0.7921,
      "step": 3617
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2371209859848022,
      "learning_rate": 5.66016942612428e-06,
      "loss": 0.7777,
      "step": 3618
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.814258873462677,
      "learning_rate": 5.658065821301093e-06,
      "loss": 1.0448,
      "step": 3619
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.210289716720581,
      "learning_rate": 5.6559620979362736e-06,
      "loss": 0.756,
      "step": 3620
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2294976711273193,
      "learning_rate": 5.653858256408775e-06,
      "loss": 0.7706,
      "step": 3621
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.298014521598816,
      "learning_rate": 5.651754297097577e-06,
      "loss": 0.8609,
      "step": 3622
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2636665105819702,
      "learning_rate": 5.64965022038168e-06,
      "loss": 0.7555,
      "step": 3623
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.282814383506775,
      "learning_rate": 5.647546026640106e-06,
      "loss": 0.7368,
      "step": 3624
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.641182005405426,
      "learning_rate": 5.6454417162518935e-06,
      "loss": 1.0119,
      "step": 3625
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6247393488883972,
      "learning_rate": 5.6433372895961095e-06,
      "loss": 1.052,
      "step": 3626
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6060842275619507,
      "learning_rate": 5.641232747051834e-06,
      "loss": 1.0204,
      "step": 3627
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2750885486602783,
      "learning_rate": 5.639128088998175e-06,
      "loss": 0.7627,
      "step": 3628
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6156589388847351,
      "learning_rate": 5.637023315814257e-06,
      "loss": 1.0003,
      "step": 3629
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2756575345993042,
      "learning_rate": 5.634918427879225e-06,
      "loss": 0.8459,
      "step": 3630
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6569673418998718,
      "learning_rate": 5.632813425572247e-06,
      "loss": 1.0253,
      "step": 3631
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.289048433303833,
      "learning_rate": 5.630708309272513e-06,
      "loss": 0.8238,
      "step": 3632
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.249577283859253,
      "learning_rate": 5.62860307935923e-06,
      "loss": 0.7711,
      "step": 3633
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2307593822479248,
      "learning_rate": 5.626497736211622e-06,
      "loss": 0.8107,
      "step": 3634
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.186981201171875,
      "learning_rate": 5.624392280208943e-06,
      "loss": 0.7764,
      "step": 3635
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3225526809692383,
      "learning_rate": 5.622286711730461e-06,
      "loss": 0.8003,
      "step": 3636
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2143385410308838,
      "learning_rate": 5.620181031155467e-06,
      "loss": 0.7946,
      "step": 3637
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6549328565597534,
      "learning_rate": 5.618075238863269e-06,
      "loss": 1.0442,
      "step": 3638
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.26165771484375,
      "learning_rate": 5.615969335233197e-06,
      "loss": 0.8088,
      "step": 3639
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.276756763458252,
      "learning_rate": 5.613863320644601e-06,
      "loss": 0.8056,
      "step": 3640
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2695521116256714,
      "learning_rate": 5.611757195476852e-06,
      "loss": 0.7763,
      "step": 3641
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2919572591781616,
      "learning_rate": 5.609650960109338e-06,
      "loss": 0.8483,
      "step": 3642
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.282287359237671,
      "learning_rate": 5.607544614921472e-06,
      "loss": 0.8465,
      "step": 3643
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3017696142196655,
      "learning_rate": 5.60543816029268e-06,
      "loss": 0.7967,
      "step": 3644
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6548089981079102,
      "learning_rate": 5.603331596602412e-06,
      "loss": 1.0275,
      "step": 3645
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2277913093566895,
      "learning_rate": 5.601224924230139e-06,
      "loss": 0.7996,
      "step": 3646
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2099913358688354,
      "learning_rate": 5.5991181435553476e-06,
      "loss": 0.8197,
      "step": 3647
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2081544399261475,
      "learning_rate": 5.5970112549575465e-06,
      "loss": 0.8536,
      "step": 3648
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1687785387039185,
      "learning_rate": 5.594904258816262e-06,
      "loss": 0.7769,
      "step": 3649
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.28684663772583,
      "learning_rate": 5.592797155511043e-06,
      "loss": 0.7871,
      "step": 3650
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.27452552318573,
      "learning_rate": 5.590689945421452e-06,
      "loss": 0.7987,
      "step": 3651
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2236433029174805,
      "learning_rate": 5.58858262892708e-06,
      "loss": 0.7936,
      "step": 3652
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2104272842407227,
      "learning_rate": 5.586475206407527e-06,
      "loss": 0.8026,
      "step": 3653
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6999381184577942,
      "learning_rate": 5.5843676782424176e-06,
      "loss": 1.0546,
      "step": 3654
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2578978538513184,
      "learning_rate": 5.5822600448113926e-06,
      "loss": 0.8384,
      "step": 3655
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.260568380355835,
      "learning_rate": 5.5801523064941175e-06,
      "loss": 0.7686,
      "step": 3656
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6854953169822693,
      "learning_rate": 5.578044463670271e-06,
      "loss": 1.0325,
      "step": 3657
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6419903039932251,
      "learning_rate": 5.575936516719552e-06,
      "loss": 1.0521,
      "step": 3658
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2933344841003418,
      "learning_rate": 5.573828466021679e-06,
      "loss": 0.8549,
      "step": 3659
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.333526611328125,
      "learning_rate": 5.5717203119563875e-06,
      "loss": 0.798,
      "step": 3660
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6754025816917419,
      "learning_rate": 5.569612054903436e-06,
      "loss": 1.0171,
      "step": 3661
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2445743083953857,
      "learning_rate": 5.567503695242595e-06,
      "loss": 0.7861,
      "step": 3662
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7105279564857483,
      "learning_rate": 5.5653952333536575e-06,
      "loss": 1.0472,
      "step": 3663
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6428475975990295,
      "learning_rate": 5.563286669616435e-06,
      "loss": 1.0159,
      "step": 3664
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6009505391120911,
      "learning_rate": 5.561178004410758e-06,
      "loss": 1.0274,
      "step": 3665
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6322678327560425,
      "learning_rate": 5.559069238116472e-06,
      "loss": 1.0377,
      "step": 3666
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2406800985336304,
      "learning_rate": 5.556960371113445e-06,
      "loss": 0.6891,
      "step": 3667
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2664974927902222,
      "learning_rate": 5.5548514037815556e-06,
      "loss": 0.7997,
      "step": 3668
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2185169458389282,
      "learning_rate": 5.552742336500711e-06,
      "loss": 0.8316,
      "step": 3669
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.360322117805481,
      "learning_rate": 5.55063316965083e-06,
      "loss": 0.7826,
      "step": 3670
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.220720887184143,
      "learning_rate": 5.5485239036118495e-06,
      "loss": 0.7583,
      "step": 3671
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2842235565185547,
      "learning_rate": 5.546414538763725e-06,
      "loss": 0.8157,
      "step": 3672
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.401082158088684,
      "learning_rate": 5.54430507548643e-06,
      "loss": 0.7895,
      "step": 3673
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.148834228515625,
      "learning_rate": 5.542195514159956e-06,
      "loss": 0.762,
      "step": 3674
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1621869802474976,
      "learning_rate": 5.540085855164312e-06,
      "loss": 0.786,
      "step": 3675
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9340975880622864,
      "learning_rate": 5.5379760988795265e-06,
      "loss": 1.0106,
      "step": 3676
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.256158471107483,
      "learning_rate": 5.5358662456856385e-06,
      "loss": 0.7895,
      "step": 3677
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.219059705734253,
      "learning_rate": 5.533756295962715e-06,
      "loss": 0.7708,
      "step": 3678
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2797869443893433,
      "learning_rate": 5.531646250090829e-06,
      "loss": 0.8167,
      "step": 3679
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2192506790161133,
      "learning_rate": 5.5295361084500806e-06,
      "loss": 0.7674,
      "step": 3680
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2484500408172607,
      "learning_rate": 5.527425871420584e-06,
      "loss": 0.7768,
      "step": 3681
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7520677447319031,
      "learning_rate": 5.5253155393824676e-06,
      "loss": 1.0365,
      "step": 3682
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3096075057983398,
      "learning_rate": 5.523205112715878e-06,
      "loss": 0.8727,
      "step": 3683
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1953353881835938,
      "learning_rate": 5.521094591800983e-06,
      "loss": 0.7643,
      "step": 3684
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2022067308425903,
      "learning_rate": 5.5189839770179645e-06,
      "loss": 0.7835,
      "step": 3685
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2796363830566406,
      "learning_rate": 5.5168732687470175e-06,
      "loss": 0.7731,
      "step": 3686
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2680302858352661,
      "learning_rate": 5.51476246736836e-06,
      "loss": 0.7981,
      "step": 3687
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1891645193099976,
      "learning_rate": 5.512651573262223e-06,
      "loss": 0.8186,
      "step": 3688
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2580169439315796,
      "learning_rate": 5.510540586808857e-06,
      "loss": 0.7665,
      "step": 3689
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4434142112731934,
      "learning_rate": 5.508429508388528e-06,
      "loss": 0.83,
      "step": 3690
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2221333980560303,
      "learning_rate": 5.506318338381516e-06,
      "loss": 0.7775,
      "step": 3691
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2834217548370361,
      "learning_rate": 5.50420707716812e-06,
      "loss": 0.8453,
      "step": 3692
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7173978090286255,
      "learning_rate": 5.502095725128656e-06,
      "loss": 1.0346,
      "step": 3693
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3032547235488892,
      "learning_rate": 5.499984282643456e-06,
      "loss": 0.7613,
      "step": 3694
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1934860944747925,
      "learning_rate": 5.497872750092866e-06,
      "loss": 0.7869,
      "step": 3695
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2525790929794312,
      "learning_rate": 5.495761127857251e-06,
      "loss": 0.7751,
      "step": 3696
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2640446424484253,
      "learning_rate": 5.493649416316991e-06,
      "loss": 0.8085,
      "step": 3697
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6850883364677429,
      "learning_rate": 5.491537615852483e-06,
      "loss": 1.0221,
      "step": 3698
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2665538787841797,
      "learning_rate": 5.489425726844138e-06,
      "loss": 0.8061,
      "step": 3699
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.23735511302948,
      "learning_rate": 5.487313749672385e-06,
      "loss": 0.7733,
      "step": 3700
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6235429644584656,
      "learning_rate": 5.485201684717667e-06,
      "loss": 1.0348,
      "step": 3701
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5903208255767822,
      "learning_rate": 5.483089532360444e-06,
      "loss": 0.8442,
      "step": 3702
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2829039096832275,
      "learning_rate": 5.480977292981192e-06,
      "loss": 0.7574,
      "step": 3703
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.237320899963379,
      "learning_rate": 5.4788649669604044e-06,
      "loss": 0.767,
      "step": 3704
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2963125705718994,
      "learning_rate": 5.476752554678584e-06,
      "loss": 0.8188,
      "step": 3705
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2823387384414673,
      "learning_rate": 5.474640056516256e-06,
      "loss": 0.754,
      "step": 3706
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.178343415260315,
      "learning_rate": 5.472527472853957e-06,
      "loss": 0.8343,
      "step": 3707
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.356539249420166,
      "learning_rate": 5.470414804072243e-06,
      "loss": 0.7736,
      "step": 3708
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2051864862442017,
      "learning_rate": 5.468302050551678e-06,
      "loss": 0.7971,
      "step": 3709
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.270196557044983,
      "learning_rate": 5.466189212672849e-06,
      "loss": 0.8047,
      "step": 3710
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7557721734046936,
      "learning_rate": 5.464076290816354e-06,
      "loss": 1.0415,
      "step": 3711
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3114548921585083,
      "learning_rate": 5.461963285362809e-06,
      "loss": 0.8095,
      "step": 3712
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6420294046401978,
      "learning_rate": 5.459850196692843e-06,
      "loss": 1.0326,
      "step": 3713
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6264178156852722,
      "learning_rate": 5.457737025187098e-06,
      "loss": 1.031,
      "step": 3714
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6447012424468994,
      "learning_rate": 5.455623771226234e-06,
      "loss": 1.0118,
      "step": 3715
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.370477557182312,
      "learning_rate": 5.453510435190925e-06,
      "loss": 0.7747,
      "step": 3716
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3408634662628174,
      "learning_rate": 5.451397017461862e-06,
      "loss": 0.7892,
      "step": 3717
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2667006254196167,
      "learning_rate": 5.4492835184197465e-06,
      "loss": 0.842,
      "step": 3718
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2468342781066895,
      "learning_rate": 5.447169938445297e-06,
      "loss": 0.75,
      "step": 3719
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2232095003128052,
      "learning_rate": 5.445056277919247e-06,
      "loss": 0.8195,
      "step": 3720
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.323395013809204,
      "learning_rate": 5.442942537222342e-06,
      "loss": 0.8275,
      "step": 3721
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8186054229736328,
      "learning_rate": 5.440828716735347e-06,
      "loss": 1.0471,
      "step": 3722
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8159579038619995,
      "learning_rate": 5.438714816839035e-06,
      "loss": 0.9845,
      "step": 3723
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2409570217132568,
      "learning_rate": 5.436600837914195e-06,
      "loss": 0.8524,
      "step": 3724
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.282983660697937,
      "learning_rate": 5.4344867803416365e-06,
      "loss": 0.7698,
      "step": 3725
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6620689630508423,
      "learning_rate": 5.4323726445021755e-06,
      "loss": 1.0291,
      "step": 3726
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2600305080413818,
      "learning_rate": 5.430258430776645e-06,
      "loss": 0.7446,
      "step": 3727
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2942733764648438,
      "learning_rate": 5.428144139545892e-06,
      "loss": 0.8104,
      "step": 3728
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3262182474136353,
      "learning_rate": 5.426029771190777e-06,
      "loss": 0.8027,
      "step": 3729
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.306847333908081,
      "learning_rate": 5.4239153260921764e-06,
      "loss": 0.8299,
      "step": 3730
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2400145530700684,
      "learning_rate": 5.421800804630978e-06,
      "loss": 0.8244,
      "step": 3731
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2771549224853516,
      "learning_rate": 5.419686207188084e-06,
      "loss": 0.7933,
      "step": 3732
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.220327377319336,
      "learning_rate": 5.417571534144409e-06,
      "loss": 0.7869,
      "step": 3733
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.327940583229065,
      "learning_rate": 5.415456785880885e-06,
      "loss": 0.7538,
      "step": 3734
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2741843461990356,
      "learning_rate": 5.4133419627784545e-06,
      "loss": 0.8014,
      "step": 3735
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8366485834121704,
      "learning_rate": 5.411227065218074e-06,
      "loss": 1.0083,
      "step": 3736
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2860338687896729,
      "learning_rate": 5.409112093580714e-06,
      "loss": 0.7597,
      "step": 3737
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2310233116149902,
      "learning_rate": 5.406997048247357e-06,
      "loss": 0.7215,
      "step": 3738
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2357501983642578,
      "learning_rate": 5.404881929599001e-06,
      "loss": 0.7678,
      "step": 3739
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6899245381355286,
      "learning_rate": 5.402766738016654e-06,
      "loss": 1.0411,
      "step": 3740
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2622743844985962,
      "learning_rate": 5.400651473881342e-06,
      "loss": 0.7697,
      "step": 3741
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2585879564285278,
      "learning_rate": 5.398536137574099e-06,
      "loss": 0.7673,
      "step": 3742
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1974668502807617,
      "learning_rate": 5.396420729475974e-06,
      "loss": 0.7987,
      "step": 3743
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6379837393760681,
      "learning_rate": 5.394305249968029e-06,
      "loss": 1.0223,
      "step": 3744
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.649960458278656,
      "learning_rate": 5.39218969943134e-06,
      "loss": 1.0331,
      "step": 3745
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3190327882766724,
      "learning_rate": 5.390074078246997e-06,
      "loss": 0.7872,
      "step": 3746
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3765966892242432,
      "learning_rate": 5.387958386796096e-06,
      "loss": 0.7923,
      "step": 3747
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3508528470993042,
      "learning_rate": 5.385842625459752e-06,
      "loss": 0.7758,
      "step": 3748
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2574652433395386,
      "learning_rate": 5.383726794619091e-06,
      "loss": 0.8405,
      "step": 3749
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2393215894699097,
      "learning_rate": 5.381610894655251e-06,
      "loss": 0.7894,
      "step": 3750
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2500108480453491,
      "learning_rate": 5.379494925949383e-06,
      "loss": 0.8334,
      "step": 3751
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2452589273452759,
      "learning_rate": 5.37737888888265e-06,
      "loss": 0.7772,
      "step": 3752
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8519881963729858,
      "learning_rate": 5.375262783836226e-06,
      "loss": 1.0414,
      "step": 3753
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.288212776184082,
      "learning_rate": 5.3731466111913e-06,
      "loss": 0.8149,
      "step": 3754
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3221668004989624,
      "learning_rate": 5.371030371329073e-06,
      "loss": 0.7915,
      "step": 3755
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2689995765686035,
      "learning_rate": 5.368914064630755e-06,
      "loss": 0.8284,
      "step": 3756
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3469239473342896,
      "learning_rate": 5.366797691477568e-06,
      "loss": 0.7983,
      "step": 3757
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3214308023452759,
      "learning_rate": 5.364681252250754e-06,
      "loss": 0.7754,
      "step": 3758
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2638441324234009,
      "learning_rate": 5.362564747331556e-06,
      "loss": 0.8015,
      "step": 3759
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3327970504760742,
      "learning_rate": 5.360448177101234e-06,
      "loss": 0.8304,
      "step": 3760
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7457767128944397,
      "learning_rate": 5.358331541941062e-06,
      "loss": 1.0284,
      "step": 3761
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.215772032737732,
      "learning_rate": 5.356214842232319e-06,
      "loss": 0.7802,
      "step": 3762
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.285571813583374,
      "learning_rate": 5.354098078356304e-06,
      "loss": 0.7769,
      "step": 3763
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2749862670898438,
      "learning_rate": 5.3519812506943215e-06,
      "loss": 0.8615,
      "step": 3764
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6390320062637329,
      "learning_rate": 5.349864359627689e-06,
      "loss": 1.0587,
      "step": 3765
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2836456298828125,
      "learning_rate": 5.347747405537735e-06,
      "loss": 0.7595,
      "step": 3766
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6278644800186157,
      "learning_rate": 5.345630388805802e-06,
      "loss": 1.0268,
      "step": 3767
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2509093284606934,
      "learning_rate": 5.343513309813239e-06,
      "loss": 0.7541,
      "step": 3768
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1973501443862915,
      "learning_rate": 5.341396168941415e-06,
      "loss": 0.7683,
      "step": 3769
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2391408681869507,
      "learning_rate": 5.339278966571697e-06,
      "loss": 0.8281,
      "step": 3770
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3818269968032837,
      "learning_rate": 5.337161703085475e-06,
      "loss": 0.7903,
      "step": 3771
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6690825819969177,
      "learning_rate": 5.3350443788641434e-06,
      "loss": 0.9954,
      "step": 3772
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4549592733383179,
      "learning_rate": 5.332926994289109e-06,
      "loss": 0.8117,
      "step": 3773
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6877874135971069,
      "learning_rate": 5.330809549741793e-06,
      "loss": 1.0244,
      "step": 3774
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3264846801757812,
      "learning_rate": 5.32869204560362e-06,
      "loss": 0.813,
      "step": 3775
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6035851240158081,
      "learning_rate": 5.326574482256034e-06,
      "loss": 1.0282,
      "step": 3776
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.30032479763031,
      "learning_rate": 5.324456860080481e-06,
      "loss": 0.8583,
      "step": 3777
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7180482745170593,
      "learning_rate": 5.322339179458427e-06,
      "loss": 1.0117,
      "step": 3778
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7069170475006104,
      "learning_rate": 5.320221440771337e-06,
      "loss": 1.0332,
      "step": 3779
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2909661531448364,
      "learning_rate": 5.318103644400699e-06,
      "loss": 0.7642,
      "step": 3780
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4534013271331787,
      "learning_rate": 5.315985790728002e-06,
      "loss": 0.7896,
      "step": 3781
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6261906623840332,
      "learning_rate": 5.313867880134749e-06,
      "loss": 1.0159,
      "step": 3782
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.390770435333252,
      "learning_rate": 5.3117499130024534e-06,
      "loss": 0.7748,
      "step": 3783
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6549257636070251,
      "learning_rate": 5.3096318897126375e-06,
      "loss": 1.0282,
      "step": 3784
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4725513458251953,
      "learning_rate": 5.307513810646836e-06,
      "loss": 0.7623,
      "step": 3785
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2709755897521973,
      "learning_rate": 5.3053956761865895e-06,
      "loss": 0.8103,
      "step": 3786
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.20332932472229,
      "learning_rate": 5.303277486713454e-06,
      "loss": 0.7905,
      "step": 3787
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.24878990650177,
      "learning_rate": 5.301159242608992e-06,
      "loss": 0.8031,
      "step": 3788
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.277619481086731,
      "learning_rate": 5.2990409442547755e-06,
      "loss": 0.8153,
      "step": 3789
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7168398499488831,
      "learning_rate": 5.296922592032388e-06,
      "loss": 1.036,
      "step": 3790
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7138016223907471,
      "learning_rate": 5.29480418632342e-06,
      "loss": 1.0204,
      "step": 3791
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2391201257705688,
      "learning_rate": 5.292685727509478e-06,
      "loss": 0.7718,
      "step": 3792
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.606196939945221,
      "learning_rate": 5.2905672159721685e-06,
      "loss": 1.0522,
      "step": 3793
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3352771997451782,
      "learning_rate": 5.288448652093115e-06,
      "loss": 0.7768,
      "step": 3794
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2533845901489258,
      "learning_rate": 5.28633003625395e-06,
      "loss": 0.7235,
      "step": 3795
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2730987071990967,
      "learning_rate": 5.2842113688363086e-06,
      "loss": 0.7663,
      "step": 3796
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4030817747116089,
      "learning_rate": 5.282092650221845e-06,
      "loss": 0.7453,
      "step": 3797
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2871862649917603,
      "learning_rate": 5.279973880792214e-06,
      "loss": 0.8159,
      "step": 3798
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.244300127029419,
      "learning_rate": 5.277855060929086e-06,
      "loss": 0.7541,
      "step": 3799
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2733389139175415,
      "learning_rate": 5.275736191014136e-06,
      "loss": 0.8162,
      "step": 3800
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4777616262435913,
      "learning_rate": 5.273617271429049e-06,
      "loss": 0.7363,
      "step": 3801
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3025038242340088,
      "learning_rate": 5.2714983025555226e-06,
      "loss": 0.7989,
      "step": 3802
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.823633074760437,
      "learning_rate": 5.269379284775256e-06,
      "loss": 1.0394,
      "step": 3803
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.28353750705719,
      "learning_rate": 5.267260218469964e-06,
      "loss": 0.8267,
      "step": 3804
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6902365684509277,
      "learning_rate": 5.265141104021369e-06,
      "loss": 1.0371,
      "step": 3805
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6770176887512207,
      "learning_rate": 5.263021941811197e-06,
      "loss": 1.0019,
      "step": 3806
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3091459274291992,
      "learning_rate": 5.260902732221192e-06,
      "loss": 0.8121,
      "step": 3807
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3088184595108032,
      "learning_rate": 5.258783475633094e-06,
      "loss": 0.82,
      "step": 3808
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.223556637763977,
      "learning_rate": 5.256664172428664e-06,
      "loss": 0.8231,
      "step": 3809
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2759417295455933,
      "learning_rate": 5.254544822989662e-06,
      "loss": 0.7931,
      "step": 3810
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2673028707504272,
      "learning_rate": 5.252425427697863e-06,
      "loss": 0.7332,
      "step": 3811
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8796207308769226,
      "learning_rate": 5.250305986935045e-06,
      "loss": 1.0285,
      "step": 3812
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7864622473716736,
      "learning_rate": 5.2481865010829966e-06,
      "loss": 1.0191,
      "step": 3813
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7156237959861755,
      "learning_rate": 5.246066970523517e-06,
      "loss": 1.0292,
      "step": 3814
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3337963819503784,
      "learning_rate": 5.243947395638408e-06,
      "loss": 0.7783,
      "step": 3815
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2779159545898438,
      "learning_rate": 5.241827776809485e-06,
      "loss": 0.772,
      "step": 3816
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8650292158126831,
      "learning_rate": 5.239708114418566e-06,
      "loss": 1.0391,
      "step": 3817
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.370123267173767,
      "learning_rate": 5.237588408847479e-06,
      "loss": 0.8589,
      "step": 3818
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7679318189620972,
      "learning_rate": 5.235468660478065e-06,
      "loss": 1.0285,
      "step": 3819
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4464635848999023,
      "learning_rate": 5.233348869692162e-06,
      "loss": 0.7481,
      "step": 3820
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2924500703811646,
      "learning_rate": 5.2312290368716256e-06,
      "loss": 0.7624,
      "step": 3821
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.767148494720459,
      "learning_rate": 5.229109162398313e-06,
      "loss": 1.0233,
      "step": 3822
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2448554039001465,
      "learning_rate": 5.226989246654091e-06,
      "loss": 0.8188,
      "step": 3823
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2679349184036255,
      "learning_rate": 5.224869290020833e-06,
      "loss": 0.8081,
      "step": 3824
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2584842443466187,
      "learning_rate": 5.2227492928804235e-06,
      "loss": 0.7633,
      "step": 3825
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2847068309783936,
      "learning_rate": 5.220629255614747e-06,
      "loss": 0.7849,
      "step": 3826
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2690999507904053,
      "learning_rate": 5.218509178605704e-06,
      "loss": 0.744,
      "step": 3827
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2438721656799316,
      "learning_rate": 5.216389062235193e-06,
      "loss": 0.7579,
      "step": 3828
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.29109525680542,
      "learning_rate": 5.214268906885128e-06,
      "loss": 0.7879,
      "step": 3829
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2546179294586182,
      "learning_rate": 5.212148712937425e-06,
      "loss": 0.7999,
      "step": 3830
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8683621883392334,
      "learning_rate": 5.210028480774008e-06,
      "loss": 1.0337,
      "step": 3831
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2588444948196411,
      "learning_rate": 5.207908210776807e-06,
      "loss": 0.7992,
      "step": 3832
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1852067708969116,
      "learning_rate": 5.205787903327761e-06,
      "loss": 0.7764,
      "step": 3833
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2479326725006104,
      "learning_rate": 5.203667558808815e-06,
      "loss": 0.7543,
      "step": 3834
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3682698011398315,
      "learning_rate": 5.20154717760192e-06,
      "loss": 0.8088,
      "step": 3835
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.651578426361084,
      "learning_rate": 5.1994267600890336e-06,
      "loss": 1.0251,
      "step": 3836
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3375755548477173,
      "learning_rate": 5.1973063066521205e-06,
      "loss": 0.7947,
      "step": 3837
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2690116167068481,
      "learning_rate": 5.195185817673151e-06,
      "loss": 0.7987,
      "step": 3838
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3173507452011108,
      "learning_rate": 5.193065293534104e-06,
      "loss": 0.7878,
      "step": 3839
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6800269484519958,
      "learning_rate": 5.1909447346169614e-06,
      "loss": 1.0416,
      "step": 3840
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.63582444190979,
      "learning_rate": 5.188824141303715e-06,
      "loss": 1.0123,
      "step": 3841
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2549448013305664,
      "learning_rate": 5.18670351397636e-06,
      "loss": 0.7949,
      "step": 3842
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5923354625701904,
      "learning_rate": 5.184582853016897e-06,
      "loss": 1.0204,
      "step": 3843
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6227216720581055,
      "learning_rate": 5.182462158807338e-06,
      "loss": 1.0244,
      "step": 3844
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2985929250717163,
      "learning_rate": 5.180341431729693e-06,
      "loss": 0.7949,
      "step": 3845
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1932021379470825,
      "learning_rate": 5.178220672165985e-06,
      "loss": 0.745,
      "step": 3846
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2509691715240479,
      "learning_rate": 5.176099880498239e-06,
      "loss": 0.7967,
      "step": 3847
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2513494491577148,
      "learning_rate": 5.173979057108487e-06,
      "loss": 0.7792,
      "step": 3848
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2997262477874756,
      "learning_rate": 5.171858202378769e-06,
      "loss": 0.8282,
      "step": 3849
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7805100083351135,
      "learning_rate": 5.169737316691125e-06,
      "loss": 1.0213,
      "step": 3850
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3112146854400635,
      "learning_rate": 5.167616400427604e-06,
      "loss": 0.8765,
      "step": 3851
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2723106145858765,
      "learning_rate": 5.165495453970262e-06,
      "loss": 0.8364,
      "step": 3852
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2708370685577393,
      "learning_rate": 5.163374477701159e-06,
      "loss": 0.8078,
      "step": 3853
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.384055256843567,
      "learning_rate": 5.161253472002358e-06,
      "loss": 0.856,
      "step": 3854
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2697721719741821,
      "learning_rate": 5.1591324372559294e-06,
      "loss": 0.7791,
      "step": 3855
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.248819351196289,
      "learning_rate": 5.157011373843951e-06,
      "loss": 0.7754,
      "step": 3856
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2608795166015625,
      "learning_rate": 5.154890282148502e-06,
      "loss": 0.7479,
      "step": 3857
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2264108657836914,
      "learning_rate": 5.152769162551671e-06,
      "loss": 0.7739,
      "step": 3858
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2536057233810425,
      "learning_rate": 5.150648015435544e-06,
      "loss": 0.7879,
      "step": 3859
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2230876684188843,
      "learning_rate": 5.148526841182222e-06,
      "loss": 0.7692,
      "step": 3860
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2206013202667236,
      "learning_rate": 5.146405640173802e-06,
      "loss": 0.715,
      "step": 3861
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.235207438468933,
      "learning_rate": 5.1442844127923915e-06,
      "loss": 0.801,
      "step": 3862
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2166924476623535,
      "learning_rate": 5.142163159420102e-06,
      "loss": 0.802,
      "step": 3863
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2193241119384766,
      "learning_rate": 5.140041880439045e-06,
      "loss": 0.8104,
      "step": 3864
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2870852947235107,
      "learning_rate": 5.137920576231342e-06,
      "loss": 0.8178,
      "step": 3865
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2797154188156128,
      "learning_rate": 5.135799247179118e-06,
      "loss": 0.8186,
      "step": 3866
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2475372552871704,
      "learning_rate": 5.133677893664501e-06,
      "loss": 0.8095,
      "step": 3867
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6931846141815186,
      "learning_rate": 5.131556516069625e-06,
      "loss": 0.9908,
      "step": 3868
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2159349918365479,
      "learning_rate": 5.129435114776624e-06,
      "loss": 0.7678,
      "step": 3869
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2639249563217163,
      "learning_rate": 5.127313690167643e-06,
      "loss": 0.8078,
      "step": 3870
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2763736248016357,
      "learning_rate": 5.125192242624828e-06,
      "loss": 0.8384,
      "step": 3871
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2410800457000732,
      "learning_rate": 5.123070772530328e-06,
      "loss": 0.8045,
      "step": 3872
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2148571014404297,
      "learning_rate": 5.1209492802662966e-06,
      "loss": 0.8055,
      "step": 3873
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2677505016326904,
      "learning_rate": 5.118827766214891e-06,
      "loss": 0.7869,
      "step": 3874
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2669345140457153,
      "learning_rate": 5.116706230758276e-06,
      "loss": 0.7856,
      "step": 3875
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2636077404022217,
      "learning_rate": 5.114584674278616e-06,
      "loss": 0.8509,
      "step": 3876
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2154539823532104,
      "learning_rate": 5.11246309715808e-06,
      "loss": 0.8221,
      "step": 3877
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2525230646133423,
      "learning_rate": 5.1103414997788424e-06,
      "loss": 0.8105,
      "step": 3878
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7429220676422119,
      "learning_rate": 5.10821988252308e-06,
      "loss": 1.0381,
      "step": 3879
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2870912551879883,
      "learning_rate": 5.106098245772974e-06,
      "loss": 0.8244,
      "step": 3880
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2794959545135498,
      "learning_rate": 5.1039765899107055e-06,
      "loss": 0.8036,
      "step": 3881
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6076328158378601,
      "learning_rate": 5.101854915318466e-06,
      "loss": 1.0175,
      "step": 3882
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2378766536712646,
      "learning_rate": 5.099733222378443e-06,
      "loss": 0.7782,
      "step": 3883
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2320528030395508,
      "learning_rate": 5.097611511472833e-06,
      "loss": 0.8335,
      "step": 3884
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6701062917709351,
      "learning_rate": 5.095489782983833e-06,
      "loss": 1.0439,
      "step": 3885
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2395163774490356,
      "learning_rate": 5.0933680372936445e-06,
      "loss": 0.8017,
      "step": 3886
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.26939058303833,
      "learning_rate": 5.091246274784468e-06,
      "loss": 0.8103,
      "step": 3887
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2257840633392334,
      "learning_rate": 5.089124495838513e-06,
      "loss": 0.8149,
      "step": 3888
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2312431335449219,
      "learning_rate": 5.087002700837988e-06,
      "loss": 0.7905,
      "step": 3889
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2030047178268433,
      "learning_rate": 5.084880890165107e-06,
      "loss": 0.8107,
      "step": 3890
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.295675277709961,
      "learning_rate": 5.0827590642020856e-06,
      "loss": 0.7761,
      "step": 3891
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8206429481506348,
      "learning_rate": 5.080637223331138e-06,
      "loss": 1.021,
      "step": 3892
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2990373373031616,
      "learning_rate": 5.07851536793449e-06,
      "loss": 0.7626,
      "step": 3893
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2382783889770508,
      "learning_rate": 5.0763934983943634e-06,
      "loss": 0.8286,
      "step": 3894
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2429345846176147,
      "learning_rate": 5.074271615092984e-06,
      "loss": 0.7821,
      "step": 3895
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2102609872817993,
      "learning_rate": 5.072149718412579e-06,
      "loss": 0.7954,
      "step": 3896
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2342019081115723,
      "learning_rate": 5.07002780873538e-06,
      "loss": 0.7744,
      "step": 3897
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3063139915466309,
      "learning_rate": 5.067905886443621e-06,
      "loss": 0.7807,
      "step": 3898
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3380100727081299,
      "learning_rate": 5.0657839519195396e-06,
      "loss": 0.8257,
      "step": 3899
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3510401248931885,
      "learning_rate": 5.06366200554537e-06,
      "loss": 0.8122,
      "step": 3900
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2809278964996338,
      "learning_rate": 5.061540047703354e-06,
      "loss": 0.777,
      "step": 3901
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1796537637710571,
      "learning_rate": 5.0594180787757305e-06,
      "loss": 0.7814,
      "step": 3902
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2758972644805908,
      "learning_rate": 5.057296099144747e-06,
      "loss": 0.7924,
      "step": 3903
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2622474431991577,
      "learning_rate": 5.055174109192649e-06,
      "loss": 0.7876,
      "step": 3904
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7246098518371582,
      "learning_rate": 5.053052109301682e-06,
      "loss": 1.0162,
      "step": 3905
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6747767329216003,
      "learning_rate": 5.0509300998540974e-06,
      "loss": 1.0114,
      "step": 3906
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2310200929641724,
      "learning_rate": 5.048808081232146e-06,
      "loss": 0.7929,
      "step": 3907
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.268760323524475,
      "learning_rate": 5.046686053818081e-06,
      "loss": 0.8022,
      "step": 3908
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3618673086166382,
      "learning_rate": 5.044564017994157e-06,
      "loss": 0.7838,
      "step": 3909
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.28872811794281,
      "learning_rate": 5.042441974142628e-06,
      "loss": 0.7433,
      "step": 3910
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2616114616394043,
      "learning_rate": 5.040319922645752e-06,
      "loss": 0.7696,
      "step": 3911
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2987226247787476,
      "learning_rate": 5.03819786388579e-06,
      "loss": 0.8473,
      "step": 3912
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7576873302459717,
      "learning_rate": 5.0360757982450005e-06,
      "loss": 1.027,
      "step": 3913
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2610727548599243,
      "learning_rate": 5.033953726105646e-06,
      "loss": 0.7829,
      "step": 3914
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.312772512435913,
      "learning_rate": 5.031831647849987e-06,
      "loss": 0.7853,
      "step": 3915
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6908534169197083,
      "learning_rate": 5.0297095638602885e-06,
      "loss": 1.0177,
      "step": 3916
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6095733046531677,
      "learning_rate": 5.0275874745188145e-06,
      "loss": 1.0301,
      "step": 3917
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.306381106376648,
      "learning_rate": 5.025465380207832e-06,
      "loss": 0.7628,
      "step": 3918
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3117127418518066,
      "learning_rate": 5.023343281309607e-06,
      "loss": 0.8054,
      "step": 3919
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2070597410202026,
      "learning_rate": 5.021221178206405e-06,
      "loss": 0.7842,
      "step": 3920
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1751418113708496,
      "learning_rate": 5.019099071280496e-06,
      "loss": 0.802,
      "step": 3921
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.250832200050354,
      "learning_rate": 5.016976960914149e-06,
      "loss": 0.7666,
      "step": 3922
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.343828558921814,
      "learning_rate": 5.014854847489633e-06,
      "loss": 0.8684,
      "step": 3923
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2921674251556396,
      "learning_rate": 5.0127327313892164e-06,
      "loss": 0.8054,
      "step": 3924
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2371894121170044,
      "learning_rate": 5.0106106129951704e-06,
      "loss": 0.7628,
      "step": 3925
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3464391231536865,
      "learning_rate": 5.008488492689767e-06,
      "loss": 0.7485,
      "step": 3926
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2772446870803833,
      "learning_rate": 5.006366370855277e-06,
      "loss": 0.7731,
      "step": 3927
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3139766454696655,
      "learning_rate": 5.0042442478739715e-06,
      "loss": 0.7642,
      "step": 3928
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3282029628753662,
      "learning_rate": 5.002122124128122e-06,
      "loss": 0.8027,
      "step": 3929
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9284089207649231,
      "learning_rate": 5e-06,
      "loss": 1.0226,
      "step": 3930
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3204134702682495,
      "learning_rate": 4.99787787587188e-06,
      "loss": 0.8184,
      "step": 3931
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3584792613983154,
      "learning_rate": 4.995755752126029e-06,
      "loss": 0.8535,
      "step": 3932
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6683178544044495,
      "learning_rate": 4.993633629144726e-06,
      "loss": 1.014,
      "step": 3933
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2459120750427246,
      "learning_rate": 4.991511507310233e-06,
      "loss": 0.7714,
      "step": 3934
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2507306337356567,
      "learning_rate": 4.989389387004831e-06,
      "loss": 0.7955,
      "step": 3935
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2712805271148682,
      "learning_rate": 4.987267268610786e-06,
      "loss": 0.7695,
      "step": 3936
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7454284429550171,
      "learning_rate": 4.985145152510369e-06,
      "loss": 1.0458,
      "step": 3937
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2514287233352661,
      "learning_rate": 4.9830230390858525e-06,
      "loss": 0.7488,
      "step": 3938
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2854379415512085,
      "learning_rate": 4.9809009287195045e-06,
      "loss": 0.7685,
      "step": 3939
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6933979392051697,
      "learning_rate": 4.978778821793596e-06,
      "loss": 1.019,
      "step": 3940
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2673215866088867,
      "learning_rate": 4.976656718690395e-06,
      "loss": 0.8019,
      "step": 3941
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6669439077377319,
      "learning_rate": 4.974534619792169e-06,
      "loss": 1.0381,
      "step": 3942
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2779960632324219,
      "learning_rate": 4.972412525481186e-06,
      "loss": 0.803,
      "step": 3943
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6244248151779175,
      "learning_rate": 4.9702904361397114e-06,
      "loss": 1.0282,
      "step": 3944
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3700767755508423,
      "learning_rate": 4.968168352150014e-06,
      "loss": 0.836,
      "step": 3945
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2068127393722534,
      "learning_rate": 4.966046273894356e-06,
      "loss": 0.7939,
      "step": 3946
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2575324773788452,
      "learning_rate": 4.9639242017549995e-06,
      "loss": 0.7973,
      "step": 3947
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2804938554763794,
      "learning_rate": 4.961802136114211e-06,
      "loss": 0.7693,
      "step": 3948
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2501754760742188,
      "learning_rate": 4.95968007735425e-06,
      "loss": 0.751,
      "step": 3949
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.215990662574768,
      "learning_rate": 4.957558025857374e-06,
      "loss": 0.7803,
      "step": 3950
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7304030060768127,
      "learning_rate": 4.9554359820058465e-06,
      "loss": 1.0389,
      "step": 3951
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.276413917541504,
      "learning_rate": 4.95331394618192e-06,
      "loss": 0.7448,
      "step": 3952
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2655280828475952,
      "learning_rate": 4.951191918767855e-06,
      "loss": 0.8019,
      "step": 3953
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3927288055419922,
      "learning_rate": 4.949069900145904e-06,
      "loss": 0.8116,
      "step": 3954
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2984503507614136,
      "learning_rate": 4.946947890698318e-06,
      "loss": 0.789,
      "step": 3955
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.249932050704956,
      "learning_rate": 4.944825890807353e-06,
      "loss": 0.7903,
      "step": 3956
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2953088283538818,
      "learning_rate": 4.942703900855254e-06,
      "loss": 0.8404,
      "step": 3957
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2030746936798096,
      "learning_rate": 4.94058192122427e-06,
      "loss": 0.811,
      "step": 3958
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2621675729751587,
      "learning_rate": 4.938459952296649e-06,
      "loss": 0.7682,
      "step": 3959
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2249171733856201,
      "learning_rate": 4.936337994454632e-06,
      "loss": 0.7536,
      "step": 3960
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2909228801727295,
      "learning_rate": 4.934216048080462e-06,
      "loss": 0.7754,
      "step": 3961
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2951738834381104,
      "learning_rate": 4.9320941135563786e-06,
      "loss": 0.7254,
      "step": 3962
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.371076226234436,
      "learning_rate": 4.929972191264621e-06,
      "loss": 0.8239,
      "step": 3963
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2818961143493652,
      "learning_rate": 4.927850281587424e-06,
      "loss": 0.8114,
      "step": 3964
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.490452766418457,
      "learning_rate": 4.925728384907018e-06,
      "loss": 0.7802,
      "step": 3965
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6768254637718201,
      "learning_rate": 4.923606501605639e-06,
      "loss": 1.0275,
      "step": 3966
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3683489561080933,
      "learning_rate": 4.92148463206551e-06,
      "loss": 0.8091,
      "step": 3967
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2356826066970825,
      "learning_rate": 4.9193627766688625e-06,
      "loss": 0.7989,
      "step": 3968
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.270551323890686,
      "learning_rate": 4.917240935797918e-06,
      "loss": 0.7128,
      "step": 3969
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2499521970748901,
      "learning_rate": 4.915119109834894e-06,
      "loss": 0.7406,
      "step": 3970
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2026389837265015,
      "learning_rate": 4.9129972991620135e-06,
      "loss": 0.7191,
      "step": 3971
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6517005562782288,
      "learning_rate": 4.910875504161488e-06,
      "loss": 1.0287,
      "step": 3972
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.315311312675476,
      "learning_rate": 4.9087537252155335e-06,
      "loss": 0.7889,
      "step": 3973
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6142386198043823,
      "learning_rate": 4.906631962706359e-06,
      "loss": 1.0332,
      "step": 3974
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3082785606384277,
      "learning_rate": 4.904510217016168e-06,
      "loss": 0.7802,
      "step": 3975
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6482972502708435,
      "learning_rate": 4.9023884885271675e-06,
      "loss": 1.0402,
      "step": 3976
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2335487604141235,
      "learning_rate": 4.900266777621559e-06,
      "loss": 0.7578,
      "step": 3977
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2698118686676025,
      "learning_rate": 4.898145084681536e-06,
      "loss": 0.7545,
      "step": 3978
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2436349391937256,
      "learning_rate": 4.896023410089295e-06,
      "loss": 0.8513,
      "step": 3979
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6528317928314209,
      "learning_rate": 4.893901754227028e-06,
      "loss": 1.0325,
      "step": 3980
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.636203408241272,
      "learning_rate": 4.891780117476921e-06,
      "loss": 1.0045,
      "step": 3981
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.643060564994812,
      "learning_rate": 4.889658500221159e-06,
      "loss": 1.0254,
      "step": 3982
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6067519783973694,
      "learning_rate": 4.887536902841921e-06,
      "loss": 1.0119,
      "step": 3983
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4938888549804688,
      "learning_rate": 4.885415325721386e-06,
      "loss": 0.7907,
      "step": 3984
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2464377880096436,
      "learning_rate": 4.883293769241725e-06,
      "loss": 0.7715,
      "step": 3985
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.22296142578125,
      "learning_rate": 4.88117223378511e-06,
      "loss": 0.772,
      "step": 3986
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3325313329696655,
      "learning_rate": 4.879050719733707e-06,
      "loss": 0.7753,
      "step": 3987
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2488781213760376,
      "learning_rate": 4.876929227469673e-06,
      "loss": 0.7696,
      "step": 3988
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2844139337539673,
      "learning_rate": 4.874807757375174e-06,
      "loss": 0.7319,
      "step": 3989
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8050049543380737,
      "learning_rate": 4.8726863098323566e-06,
      "loss": 1.0046,
      "step": 3990
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2930492162704468,
      "learning_rate": 4.870564885223377e-06,
      "loss": 0.7741,
      "step": 3991
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6497888565063477,
      "learning_rate": 4.868443483930379e-06,
      "loss": 1.0198,
      "step": 3992
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3244891166687012,
      "learning_rate": 4.8663221063355e-06,
      "loss": 0.8026,
      "step": 3993
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6635735630989075,
      "learning_rate": 4.8642007528208835e-06,
      "loss": 1.043,
      "step": 3994
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3248891830444336,
      "learning_rate": 4.862079423768658e-06,
      "loss": 0.7527,
      "step": 3995
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2319886684417725,
      "learning_rate": 4.859958119560956e-06,
      "loss": 0.7398,
      "step": 3996
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1834832429885864,
      "learning_rate": 4.8578368405799e-06,
      "loss": 0.8311,
      "step": 3997
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6896965503692627,
      "learning_rate": 4.8557155872076085e-06,
      "loss": 1.0344,
      "step": 3998
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6278442740440369,
      "learning_rate": 4.8535943598261984e-06,
      "loss": 1.0391,
      "step": 3999
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2621756792068481,
      "learning_rate": 4.8514731588177784e-06,
      "loss": 0.7433,
      "step": 4000
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3094241619110107,
      "learning_rate": 4.849351984564457e-06,
      "loss": 0.8131,
      "step": 4001
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2300398349761963,
      "learning_rate": 4.847230837448331e-06,
      "loss": 0.7785,
      "step": 4002
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6650834083557129,
      "learning_rate": 4.845109717851498e-06,
      "loss": 1.0375,
      "step": 4003
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3689608573913574,
      "learning_rate": 4.842988626156051e-06,
      "loss": 0.7711,
      "step": 4004
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2208569049835205,
      "learning_rate": 4.840867562744073e-06,
      "loss": 0.78,
      "step": 4005
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2094202041625977,
      "learning_rate": 4.8387465279976446e-06,
      "loss": 0.7973,
      "step": 4006
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6549350023269653,
      "learning_rate": 4.836625522298844e-06,
      "loss": 1.0608,
      "step": 4007
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2864599227905273,
      "learning_rate": 4.834504546029739e-06,
      "loss": 0.7784,
      "step": 4008
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2573796510696411,
      "learning_rate": 4.832383599572397e-06,
      "loss": 0.7818,
      "step": 4009
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2664068937301636,
      "learning_rate": 4.830262683308878e-06,
      "loss": 0.8001,
      "step": 4010
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6192852258682251,
      "learning_rate": 4.828141797621232e-06,
      "loss": 1.0224,
      "step": 4011
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2639046907424927,
      "learning_rate": 4.826020942891514e-06,
      "loss": 0.8313,
      "step": 4012
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2525100708007812,
      "learning_rate": 4.823900119501762e-06,
      "loss": 0.7384,
      "step": 4013
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6115729212760925,
      "learning_rate": 4.821779327834017e-06,
      "loss": 1.0444,
      "step": 4014
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2256218194961548,
      "learning_rate": 4.81965856827031e-06,
      "loss": 0.7983,
      "step": 4015
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2261362075805664,
      "learning_rate": 4.817537841192664e-06,
      "loss": 0.7951,
      "step": 4016
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2404677867889404,
      "learning_rate": 4.815417146983104e-06,
      "loss": 0.848,
      "step": 4017
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6063323616981506,
      "learning_rate": 4.813296486023642e-06,
      "loss": 1.0177,
      "step": 4018
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2027007341384888,
      "learning_rate": 4.811175858696286e-06,
      "loss": 0.7236,
      "step": 4019
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6039212346076965,
      "learning_rate": 4.809055265383039e-06,
      "loss": 1.0581,
      "step": 4020
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1724917888641357,
      "learning_rate": 4.806934706465897e-06,
      "loss": 0.7728,
      "step": 4021
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3771122694015503,
      "learning_rate": 4.80481418232685e-06,
      "loss": 0.8152,
      "step": 4022
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5969367623329163,
      "learning_rate": 4.80269369334788e-06,
      "loss": 1.0158,
      "step": 4023
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2036164999008179,
      "learning_rate": 4.800573239910968e-06,
      "loss": 0.7998,
      "step": 4024
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6326904296875,
      "learning_rate": 4.798452822398083e-06,
      "loss": 1.0399,
      "step": 4025
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2756136655807495,
      "learning_rate": 4.796332441191185e-06,
      "loss": 0.7268,
      "step": 4026
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3309866189956665,
      "learning_rate": 4.7942120966722405e-06,
      "loss": 0.8231,
      "step": 4027
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.212033748626709,
      "learning_rate": 4.792091789223193e-06,
      "loss": 0.8267,
      "step": 4028
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4096217155456543,
      "learning_rate": 4.789971519225994e-06,
      "loss": 0.7254,
      "step": 4029
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.666918158531189,
      "learning_rate": 4.7878512870625775e-06,
      "loss": 1.039,
      "step": 4030
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7347897887229919,
      "learning_rate": 4.785731093114873e-06,
      "loss": 1.0445,
      "step": 4031
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6283490657806396,
      "learning_rate": 4.783610937764808e-06,
      "loss": 1.0358,
      "step": 4032
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.277555227279663,
      "learning_rate": 4.781490821394299e-06,
      "loss": 0.7853,
      "step": 4033
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2572298049926758,
      "learning_rate": 4.7793707443852536e-06,
      "loss": 0.7822,
      "step": 4034
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2326244115829468,
      "learning_rate": 4.77725070711958e-06,
      "loss": 0.7218,
      "step": 4035
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2600493431091309,
      "learning_rate": 4.775130709979167e-06,
      "loss": 0.7893,
      "step": 4036
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.287729024887085,
      "learning_rate": 4.7730107533459115e-06,
      "loss": 0.8011,
      "step": 4037
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3366390466690063,
      "learning_rate": 4.770890837601691e-06,
      "loss": 0.81,
      "step": 4038
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2982943058013916,
      "learning_rate": 4.768770963128376e-06,
      "loss": 0.7903,
      "step": 4039
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2723259925842285,
      "learning_rate": 4.76665113030784e-06,
      "loss": 0.8491,
      "step": 4040
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2729005813598633,
      "learning_rate": 4.764531339521938e-06,
      "loss": 0.7867,
      "step": 4041
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2701400518417358,
      "learning_rate": 4.762411591152522e-06,
      "loss": 0.7958,
      "step": 4042
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.362666368484497,
      "learning_rate": 4.7602918855814365e-06,
      "loss": 0.7769,
      "step": 4043
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9508588314056396,
      "learning_rate": 4.7581722231905165e-06,
      "loss": 1.0275,
      "step": 4044
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2917289733886719,
      "learning_rate": 4.756052604361594e-06,
      "loss": 0.8344,
      "step": 4045
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6726950407028198,
      "learning_rate": 4.753933029476484e-06,
      "loss": 1.0145,
      "step": 4046
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3907588720321655,
      "learning_rate": 4.751813498917004e-06,
      "loss": 0.7875,
      "step": 4047
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2739193439483643,
      "learning_rate": 4.749694013064958e-06,
      "loss": 0.8107,
      "step": 4048
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2195879220962524,
      "learning_rate": 4.747574572302138e-06,
      "loss": 0.8597,
      "step": 4049
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9282963275909424,
      "learning_rate": 4.74545517701034e-06,
      "loss": 0.9975,
      "step": 4050
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8401693105697632,
      "learning_rate": 4.743335827571337e-06,
      "loss": 1.0131,
      "step": 4051
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3111371994018555,
      "learning_rate": 4.7412165243669065e-06,
      "loss": 0.8307,
      "step": 4052
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5911361575126648,
      "learning_rate": 4.739097267778811e-06,
      "loss": 1.04,
      "step": 4053
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7097731828689575,
      "learning_rate": 4.736978058188803e-06,
      "loss": 1.0243,
      "step": 4054
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2270416021347046,
      "learning_rate": 4.734858895978633e-06,
      "loss": 0.7877,
      "step": 4055
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7121866941452026,
      "learning_rate": 4.732739781530036e-06,
      "loss": 1.0475,
      "step": 4056
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2926733493804932,
      "learning_rate": 4.730620715224745e-06,
      "loss": 0.8099,
      "step": 4057
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2070709466934204,
      "learning_rate": 4.72850169744448e-06,
      "loss": 0.7791,
      "step": 4058
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3281587362289429,
      "learning_rate": 4.726382728570952e-06,
      "loss": 0.8157,
      "step": 4059
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2836217880249023,
      "learning_rate": 4.724263808985866e-06,
      "loss": 0.813,
      "step": 4060
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2224299907684326,
      "learning_rate": 4.722144939070915e-06,
      "loss": 0.7673,
      "step": 4061
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.274816632270813,
      "learning_rate": 4.720026119207787e-06,
      "loss": 0.7653,
      "step": 4062
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2183983325958252,
      "learning_rate": 4.717907349778156e-06,
      "loss": 0.7519,
      "step": 4063
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.389725685119629,
      "learning_rate": 4.715788631163692e-06,
      "loss": 0.7455,
      "step": 4064
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2622848749160767,
      "learning_rate": 4.713669963746053e-06,
      "loss": 0.7759,
      "step": 4065
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7858260869979858,
      "learning_rate": 4.7115513479068866e-06,
      "loss": 1.0522,
      "step": 4066
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6957390904426575,
      "learning_rate": 4.709432784027832e-06,
      "loss": 1.03,
      "step": 4067
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6230655908584595,
      "learning_rate": 4.707314272490524e-06,
      "loss": 1.0214,
      "step": 4068
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2554478645324707,
      "learning_rate": 4.70519581367658e-06,
      "loss": 0.8351,
      "step": 4069
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2832467555999756,
      "learning_rate": 4.703077407967614e-06,
      "loss": 0.7708,
      "step": 4070
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4429701566696167,
      "learning_rate": 4.700959055745227e-06,
      "loss": 0.7816,
      "step": 4071
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3253861665725708,
      "learning_rate": 4.698840757391009e-06,
      "loss": 0.7621,
      "step": 4072
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3272863626480103,
      "learning_rate": 4.696722513286547e-06,
      "loss": 0.8405,
      "step": 4073
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7994633913040161,
      "learning_rate": 4.6946043238134104e-06,
      "loss": 1.017,
      "step": 4074
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3100004196166992,
      "learning_rate": 4.692486189353166e-06,
      "loss": 0.8066,
      "step": 4075
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2865012884140015,
      "learning_rate": 4.690368110287365e-06,
      "loss": 0.7944,
      "step": 4076
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6387254595756531,
      "learning_rate": 4.688250086997547e-06,
      "loss": 1.0061,
      "step": 4077
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6253063678741455,
      "learning_rate": 4.6861321198652535e-06,
      "loss": 0.9892,
      "step": 4078
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6107132434844971,
      "learning_rate": 4.684014209272e-06,
      "loss": 1.0213,
      "step": 4079
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.253270149230957,
      "learning_rate": 4.681896355599302e-06,
      "loss": 0.7818,
      "step": 4080
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2713943719863892,
      "learning_rate": 4.679778559228664e-06,
      "loss": 0.759,
      "step": 4081
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.308262586593628,
      "learning_rate": 4.677660820541575e-06,
      "loss": 0.8012,
      "step": 4082
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3930857181549072,
      "learning_rate": 4.67554313991952e-06,
      "loss": 0.8356,
      "step": 4083
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2833880186080933,
      "learning_rate": 4.673425517743967e-06,
      "loss": 0.8179,
      "step": 4084
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2065513134002686,
      "learning_rate": 4.67130795439638e-06,
      "loss": 0.7598,
      "step": 4085
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3115438222885132,
      "learning_rate": 4.6691904502582095e-06,
      "loss": 0.7523,
      "step": 4086
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3250644207000732,
      "learning_rate": 4.66707300571089e-06,
      "loss": 0.8324,
      "step": 4087
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8937824368476868,
      "learning_rate": 4.664955621135858e-06,
      "loss": 1.0477,
      "step": 4088
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7836781144142151,
      "learning_rate": 4.662838296914525e-06,
      "loss": 1.0199,
      "step": 4089
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6474027037620544,
      "learning_rate": 4.6607210334283036e-06,
      "loss": 1.0392,
      "step": 4090
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3560330867767334,
      "learning_rate": 4.658603831058588e-06,
      "loss": 0.8011,
      "step": 4091
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.370115041732788,
      "learning_rate": 4.65648669018676e-06,
      "loss": 0.8138,
      "step": 4092
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.328545093536377,
      "learning_rate": 4.6543696111942e-06,
      "loss": 0.7773,
      "step": 4093
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4086238145828247,
      "learning_rate": 4.6522525944622675e-06,
      "loss": 0.8263,
      "step": 4094
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4427883625030518,
      "learning_rate": 4.650135640372313e-06,
      "loss": 0.813,
      "step": 4095
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2603063583374023,
      "learning_rate": 4.648018749305681e-06,
      "loss": 0.8375,
      "step": 4096
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2987245321273804,
      "learning_rate": 4.645901921643697e-06,
      "loss": 0.7522,
      "step": 4097
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.297736644744873,
      "learning_rate": 4.643785157767682e-06,
      "loss": 0.7908,
      "step": 4098
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2150075435638428,
      "learning_rate": 4.641668458058941e-06,
      "loss": 0.7536,
      "step": 4099
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2810428142547607,
      "learning_rate": 4.6395518228987676e-06,
      "loss": 0.7841,
      "step": 4100
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3642916679382324,
      "learning_rate": 4.637435252668446e-06,
      "loss": 0.8019,
      "step": 4101
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1765249967575073,
      "learning_rate": 4.635318747749247e-06,
      "loss": 1.0391,
      "step": 4102
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2607725858688354,
      "learning_rate": 4.633202308522432e-06,
      "loss": 0.7484,
      "step": 4103
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2896921634674072,
      "learning_rate": 4.631085935369247e-06,
      "loss": 0.7668,
      "step": 4104
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.249216079711914,
      "learning_rate": 4.6289696286709284e-06,
      "loss": 0.8141,
      "step": 4105
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7257793545722961,
      "learning_rate": 4.626853388808702e-06,
      "loss": 1.0013,
      "step": 4106
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.359410285949707,
      "learning_rate": 4.624737216163774e-06,
      "loss": 0.8148,
      "step": 4107
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2378969192504883,
      "learning_rate": 4.622621111117352e-06,
      "loss": 0.7613,
      "step": 4108
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6549316644668579,
      "learning_rate": 4.620505074050619e-06,
      "loss": 1.0082,
      "step": 4109
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6378807425498962,
      "learning_rate": 4.6183891053447495e-06,
      "loss": 1.0147,
      "step": 4110
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2986263036727905,
      "learning_rate": 4.616273205380911e-06,
      "loss": 0.7808,
      "step": 4111
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6317775249481201,
      "learning_rate": 4.614157374540248e-06,
      "loss": 1.0289,
      "step": 4112
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2555311918258667,
      "learning_rate": 4.6120416132039055e-06,
      "loss": 0.7864,
      "step": 4113
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2339396476745605,
      "learning_rate": 4.6099259217530055e-06,
      "loss": 0.7081,
      "step": 4114
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3578510284423828,
      "learning_rate": 4.607810300568659e-06,
      "loss": 0.8251,
      "step": 4115
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2237305641174316,
      "learning_rate": 4.605694750031972e-06,
      "loss": 0.7819,
      "step": 4116
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.345240831375122,
      "learning_rate": 4.603579270524028e-06,
      "loss": 0.7879,
      "step": 4117
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.237710952758789,
      "learning_rate": 4.601463862425903e-06,
      "loss": 0.7472,
      "step": 4118
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2585880756378174,
      "learning_rate": 4.59934852611866e-06,
      "loss": 0.8267,
      "step": 4119
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7814757823944092,
      "learning_rate": 4.597233261983347e-06,
      "loss": 1.0083,
      "step": 4120
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2514654397964478,
      "learning_rate": 4.595118070401001e-06,
      "loss": 0.7973,
      "step": 4121
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2721188068389893,
      "learning_rate": 4.593002951752645e-06,
      "loss": 0.7849,
      "step": 4122
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.224090337753296,
      "learning_rate": 4.590887906419288e-06,
      "loss": 0.816,
      "step": 4123
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.252910852432251,
      "learning_rate": 4.588772934781929e-06,
      "loss": 0.8258,
      "step": 4124
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3153252601623535,
      "learning_rate": 4.586658037221546e-06,
      "loss": 0.8305,
      "step": 4125
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2814370393753052,
      "learning_rate": 4.584543214119117e-06,
      "loss": 0.7721,
      "step": 4126
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7055228352546692,
      "learning_rate": 4.582428465855594e-06,
      "loss": 1.0626,
      "step": 4127
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2509406805038452,
      "learning_rate": 4.5803137928119176e-06,
      "loss": 0.7856,
      "step": 4128
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.299137830734253,
      "learning_rate": 4.5781991953690245e-06,
      "loss": 0.8234,
      "step": 4129
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2559025287628174,
      "learning_rate": 4.576084673907824e-06,
      "loss": 0.836,
      "step": 4130
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.278655767440796,
      "learning_rate": 4.573970228809224e-06,
      "loss": 0.7982,
      "step": 4131
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6305714249610901,
      "learning_rate": 4.571855860454111e-06,
      "loss": 1.0439,
      "step": 4132
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2505115270614624,
      "learning_rate": 4.569741569223356e-06,
      "loss": 0.7942,
      "step": 4133
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3627535104751587,
      "learning_rate": 4.567627355497827e-06,
      "loss": 0.8383,
      "step": 4134
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3928757905960083,
      "learning_rate": 4.565513219658364e-06,
      "loss": 0.8157,
      "step": 4135
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6781843900680542,
      "learning_rate": 4.563399162085806e-06,
      "loss": 1.039,
      "step": 4136
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6236845254898071,
      "learning_rate": 4.5612851831609684e-06,
      "loss": 1.0298,
      "step": 4137
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2491496801376343,
      "learning_rate": 4.559171283264654e-06,
      "loss": 0.8288,
      "step": 4138
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.244586706161499,
      "learning_rate": 4.5570574627776584e-06,
      "loss": 0.7359,
      "step": 4139
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.222052812576294,
      "learning_rate": 4.554943722080754e-06,
      "loss": 0.7784,
      "step": 4140
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2392420768737793,
      "learning_rate": 4.5528300615547045e-06,
      "loss": 0.7971,
      "step": 4141
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8012294173240662,
      "learning_rate": 4.550716481580254e-06,
      "loss": 1.02,
      "step": 4142
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.195563554763794,
      "learning_rate": 4.548602982538139e-06,
      "loss": 0.6779,
      "step": 4143
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.409788727760315,
      "learning_rate": 4.546489564809076e-06,
      "loss": 0.8022,
      "step": 4144
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6427168250083923,
      "learning_rate": 4.544376228773766e-06,
      "loss": 1.032,
      "step": 4145
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6070695519447327,
      "learning_rate": 4.542262974812904e-06,
      "loss": 1.0091,
      "step": 4146
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.603972852230072,
      "learning_rate": 4.54014980330716e-06,
      "loss": 1.0122,
      "step": 4147
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6262269616127014,
      "learning_rate": 4.538036714637191e-06,
      "loss": 1.0275,
      "step": 4148
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.243781328201294,
      "learning_rate": 4.5359237091836465e-06,
      "loss": 0.7491,
      "step": 4149
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2791811227798462,
      "learning_rate": 4.533810787327153e-06,
      "loss": 0.8224,
      "step": 4150
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2404377460479736,
      "learning_rate": 4.531697949448323e-06,
      "loss": 0.7719,
      "step": 4151
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.646388053894043,
      "learning_rate": 4.529585195927761e-06,
      "loss": 1.0022,
      "step": 4152
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2629684209823608,
      "learning_rate": 4.527472527146043e-06,
      "loss": 0.8041,
      "step": 4153
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6153344511985779,
      "learning_rate": 4.5253599434837456e-06,
      "loss": 1.0108,
      "step": 4154
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2509803771972656,
      "learning_rate": 4.523247445321418e-06,
      "loss": 0.7447,
      "step": 4155
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.29847252368927,
      "learning_rate": 4.521135033039597e-06,
      "loss": 0.7832,
      "step": 4156
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3260077238082886,
      "learning_rate": 4.519022707018809e-06,
      "loss": 0.7627,
      "step": 4157
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2609169483184814,
      "learning_rate": 4.516910467639556e-06,
      "loss": 0.761,
      "step": 4158
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2306023836135864,
      "learning_rate": 4.514798315282335e-06,
      "loss": 0.7488,
      "step": 4159
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2797033786773682,
      "learning_rate": 4.512686250327618e-06,
      "loss": 0.7668,
      "step": 4160
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2381185293197632,
      "learning_rate": 4.510574273155864e-06,
      "loss": 0.7491,
      "step": 4161
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2362914085388184,
      "learning_rate": 4.5084623841475184e-06,
      "loss": 0.7669,
      "step": 4162
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2196922302246094,
      "learning_rate": 4.50635058368301e-06,
      "loss": 0.7973,
      "step": 4163
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7224158644676208,
      "learning_rate": 4.504238872142751e-06,
      "loss": 1.0278,
      "step": 4164
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.294203519821167,
      "learning_rate": 4.502127249907136e-06,
      "loss": 0.7639,
      "step": 4165
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6473211050033569,
      "learning_rate": 4.500015717356545e-06,
      "loss": 1.0395,
      "step": 4166
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.632973849773407,
      "learning_rate": 4.497904274871346e-06,
      "loss": 1.0477,
      "step": 4167
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2692924737930298,
      "learning_rate": 4.495792922831881e-06,
      "loss": 0.7961,
      "step": 4168
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2909386157989502,
      "learning_rate": 4.493681661618487e-06,
      "loss": 0.8554,
      "step": 4169
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2726658582687378,
      "learning_rate": 4.491570491611475e-06,
      "loss": 0.803,
      "step": 4170
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6605413556098938,
      "learning_rate": 4.489459413191143e-06,
      "loss": 1.0161,
      "step": 4171
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2710844278335571,
      "learning_rate": 4.487348426737779e-06,
      "loss": 0.8191,
      "step": 4172
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2144883871078491,
      "learning_rate": 4.485237532631641e-06,
      "loss": 0.7961,
      "step": 4173
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2851765155792236,
      "learning_rate": 4.483126731252984e-06,
      "loss": 0.8347,
      "step": 4174
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2580515146255493,
      "learning_rate": 4.481016022982039e-06,
      "loss": 0.7644,
      "step": 4175
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2099026441574097,
      "learning_rate": 4.478905408199017e-06,
      "loss": 0.7464,
      "step": 4176
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7326822280883789,
      "learning_rate": 4.476794887284123e-06,
      "loss": 1.0528,
      "step": 4177
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6591681838035583,
      "learning_rate": 4.474684460617536e-06,
      "loss": 1.0115,
      "step": 4178
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.273362398147583,
      "learning_rate": 4.4725741285794175e-06,
      "loss": 0.7595,
      "step": 4179
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2795153856277466,
      "learning_rate": 4.47046389154992e-06,
      "loss": 0.7337,
      "step": 4180
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2909553050994873,
      "learning_rate": 4.468353749909172e-06,
      "loss": 0.7755,
      "step": 4181
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.31020188331604,
      "learning_rate": 4.466243704037287e-06,
      "loss": 0.7422,
      "step": 4182
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.244174838066101,
      "learning_rate": 4.464133754314363e-06,
      "loss": 0.7759,
      "step": 4183
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2500667572021484,
      "learning_rate": 4.462023901120476e-06,
      "loss": 0.7302,
      "step": 4184
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2884786128997803,
      "learning_rate": 4.459914144835689e-06,
      "loss": 0.765,
      "step": 4185
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8284539580345154,
      "learning_rate": 4.457804485840044e-06,
      "loss": 1.0297,
      "step": 4186
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2261161804199219,
      "learning_rate": 4.4556949245135715e-06,
      "loss": 0.8026,
      "step": 4187
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6938318610191345,
      "learning_rate": 4.453585461236277e-06,
      "loss": 1.0021,
      "step": 4188
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2233713865280151,
      "learning_rate": 4.451476096388151e-06,
      "loss": 0.7535,
      "step": 4189
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2445745468139648,
      "learning_rate": 4.449366830349172e-06,
      "loss": 0.7524,
      "step": 4190
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.252293586730957,
      "learning_rate": 4.447257663499289e-06,
      "loss": 0.8233,
      "step": 4191
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2816818952560425,
      "learning_rate": 4.445148596218445e-06,
      "loss": 0.8084,
      "step": 4192
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3243415355682373,
      "learning_rate": 4.443039628886559e-06,
      "loss": 0.7756,
      "step": 4193
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3266507387161255,
      "learning_rate": 4.4409307618835284e-06,
      "loss": 0.8137,
      "step": 4194
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9688791632652283,
      "learning_rate": 4.4388219955892436e-06,
      "loss": 1.0265,
      "step": 4195
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.283086895942688,
      "learning_rate": 4.4367133303835646e-06,
      "loss": 0.7525,
      "step": 4196
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3196889162063599,
      "learning_rate": 4.434604766646344e-06,
      "loss": 0.7764,
      "step": 4197
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2802205085754395,
      "learning_rate": 4.432496304757408e-06,
      "loss": 0.8097,
      "step": 4198
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.335472822189331,
      "learning_rate": 4.430387945096566e-06,
      "loss": 0.8114,
      "step": 4199
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6970697045326233,
      "learning_rate": 4.428279688043613e-06,
      "loss": 1.0643,
      "step": 4200
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2449613809585571,
      "learning_rate": 4.426171533978321e-06,
      "loss": 0.8262,
      "step": 4201
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1862455606460571,
      "learning_rate": 4.4240634832804496e-06,
      "loss": 0.8259,
      "step": 4202
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6285374760627747,
      "learning_rate": 4.42195553632973e-06,
      "loss": 1.0386,
      "step": 4203
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3216147422790527,
      "learning_rate": 4.419847693505883e-06,
      "loss": 0.7394,
      "step": 4204
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.189029574394226,
      "learning_rate": 4.417739955188608e-06,
      "loss": 0.7549,
      "step": 4205
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2878655195236206,
      "learning_rate": 4.415632321757583e-06,
      "loss": 0.8053,
      "step": 4206
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6315605044364929,
      "learning_rate": 4.413524793592475e-06,
      "loss": 1.0243,
      "step": 4207
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2487289905548096,
      "learning_rate": 4.411417371072922e-06,
      "loss": 0.795,
      "step": 4208
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6031628251075745,
      "learning_rate": 4.409310054578547e-06,
      "loss": 1.0295,
      "step": 4209
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.457602620124817,
      "learning_rate": 4.407202844488959e-06,
      "loss": 0.7168,
      "step": 4210
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2218235731124878,
      "learning_rate": 4.40509574118374e-06,
      "loss": 0.7419,
      "step": 4211
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3066248893737793,
      "learning_rate": 4.402988745042454e-06,
      "loss": 0.7805,
      "step": 4212
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3145941495895386,
      "learning_rate": 4.400881856444654e-06,
      "loss": 0.7928,
      "step": 4213
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4073307514190674,
      "learning_rate": 4.398775075769862e-06,
      "loss": 0.7894,
      "step": 4214
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2419986724853516,
      "learning_rate": 4.396668403397589e-06,
      "loss": 0.8261,
      "step": 4215
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6292831301689148,
      "learning_rate": 4.394561839707323e-06,
      "loss": 1.0272,
      "step": 4216
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.350701928138733,
      "learning_rate": 4.392455385078529e-06,
      "loss": 0.7897,
      "step": 4217
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2860591411590576,
      "learning_rate": 4.3903490398906634e-06,
      "loss": 0.7577,
      "step": 4218
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2801066637039185,
      "learning_rate": 4.388242804523149e-06,
      "loss": 0.7498,
      "step": 4219
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6193436980247498,
      "learning_rate": 4.3861366793554e-06,
      "loss": 1.018,
      "step": 4220
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2607603073120117,
      "learning_rate": 4.3840306647668055e-06,
      "loss": 0.7994,
      "step": 4221
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.266189694404602,
      "learning_rate": 4.381924761136733e-06,
      "loss": 0.767,
      "step": 4222
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.621813952922821,
      "learning_rate": 4.379818968844534e-06,
      "loss": 1.0204,
      "step": 4223
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3289934396743774,
      "learning_rate": 4.377713288269539e-06,
      "loss": 0.8438,
      "step": 4224
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2706291675567627,
      "learning_rate": 4.375607719791058e-06,
      "loss": 0.8133,
      "step": 4225
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3821229934692383,
      "learning_rate": 4.37350226378838e-06,
      "loss": 0.8223,
      "step": 4226
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2258172035217285,
      "learning_rate": 4.371396920640773e-06,
      "loss": 0.7951,
      "step": 4227
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.186193585395813,
      "learning_rate": 4.369291690727489e-06,
      "loss": 0.7475,
      "step": 4228
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.292246699333191,
      "learning_rate": 4.367186574427752e-06,
      "loss": 0.8191,
      "step": 4229
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3680673837661743,
      "learning_rate": 4.365081572120776e-06,
      "loss": 0.7878,
      "step": 4230
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2332849502563477,
      "learning_rate": 4.362976684185746e-06,
      "loss": 0.7294,
      "step": 4231
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2469801902770996,
      "learning_rate": 4.360871911001826e-06,
      "loss": 0.7859,
      "step": 4232
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2717230319976807,
      "learning_rate": 4.358767252948167e-06,
      "loss": 0.7554,
      "step": 4233
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6631348729133606,
      "learning_rate": 4.356662710403891e-06,
      "loss": 1.0053,
      "step": 4234
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3697930574417114,
      "learning_rate": 4.354558283748107e-06,
      "loss": 0.7921,
      "step": 4235
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3935019969940186,
      "learning_rate": 4.352453973359897e-06,
      "loss": 0.8174,
      "step": 4236
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2946563959121704,
      "learning_rate": 4.3503497796183204e-06,
      "loss": 0.8087,
      "step": 4237
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2360330820083618,
      "learning_rate": 4.348245702902425e-06,
      "loss": 0.7722,
      "step": 4238
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3040987253189087,
      "learning_rate": 4.346141743591228e-06,
      "loss": 0.8055,
      "step": 4239
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6372134685516357,
      "learning_rate": 4.344037902063728e-06,
      "loss": 1.0487,
      "step": 4240
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2340837717056274,
      "learning_rate": 4.341934178698908e-06,
      "loss": 0.7722,
      "step": 4241
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3963199853897095,
      "learning_rate": 4.339830573875722e-06,
      "loss": 0.789,
      "step": 4242
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3212380409240723,
      "learning_rate": 4.337727087973107e-06,
      "loss": 0.8183,
      "step": 4243
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6378198266029358,
      "learning_rate": 4.335623721369977e-06,
      "loss": 1.0383,
      "step": 4244
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2392820119857788,
      "learning_rate": 4.333520474445225e-06,
      "loss": 0.8388,
      "step": 4245
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5881497859954834,
      "learning_rate": 4.331417347577723e-06,
      "loss": 1.0291,
      "step": 4246
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3711802959442139,
      "learning_rate": 4.329314341146319e-06,
      "loss": 0.7561,
      "step": 4247
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6103939414024353,
      "learning_rate": 4.327211455529844e-06,
      "loss": 1.0117,
      "step": 4248
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5756264925003052,
      "learning_rate": 4.325108691107103e-06,
      "loss": 1.0078,
      "step": 4249
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2859807014465332,
      "learning_rate": 4.323006048256878e-06,
      "loss": 0.8423,
      "step": 4250
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6034691333770752,
      "learning_rate": 4.320903527357936e-06,
      "loss": 1.0516,
      "step": 4251
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3051722049713135,
      "learning_rate": 4.318801128789013e-06,
      "loss": 0.7974,
      "step": 4252
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5885128974914551,
      "learning_rate": 4.316698852928832e-06,
      "loss": 1.0332,
      "step": 4253
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2862894535064697,
      "learning_rate": 4.314596700156088e-06,
      "loss": 0.7498,
      "step": 4254
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2876185178756714,
      "learning_rate": 4.312494670849453e-06,
      "loss": 0.7764,
      "step": 4255
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2751331329345703,
      "learning_rate": 4.310392765387583e-06,
      "loss": 0.7427,
      "step": 4256
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3291751146316528,
      "learning_rate": 4.3082909841491034e-06,
      "loss": 0.7837,
      "step": 4257
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3227781057357788,
      "learning_rate": 4.306189327512626e-06,
      "loss": 0.8143,
      "step": 4258
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2213795185089111,
      "learning_rate": 4.304087795856735e-06,
      "loss": 0.7979,
      "step": 4259
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6659430861473083,
      "learning_rate": 4.301986389559989e-06,
      "loss": 1.0143,
      "step": 4260
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6431261897087097,
      "learning_rate": 4.299885109000931e-06,
      "loss": 1.0383,
      "step": 4261
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6472826600074768,
      "learning_rate": 4.297783954558077e-06,
      "loss": 0.9848,
      "step": 4262
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2396005392074585,
      "learning_rate": 4.295682926609923e-06,
      "loss": 0.8099,
      "step": 4263
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.242674469947815,
      "learning_rate": 4.29358202553494e-06,
      "loss": 0.7654,
      "step": 4264
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6351960897445679,
      "learning_rate": 4.291481251711575e-06,
      "loss": 1.0052,
      "step": 4265
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.245095133781433,
      "learning_rate": 4.289380605518258e-06,
      "loss": 0.7802,
      "step": 4266
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6481583118438721,
      "learning_rate": 4.287280087333388e-06,
      "loss": 1.0138,
      "step": 4267
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6390751600265503,
      "learning_rate": 4.285179697535345e-06,
      "loss": 1.02,
      "step": 4268
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.268486738204956,
      "learning_rate": 4.283079436502489e-06,
      "loss": 0.8094,
      "step": 4269
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3419594764709473,
      "learning_rate": 4.280979304613149e-06,
      "loss": 0.8154,
      "step": 4270
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.29581880569458,
      "learning_rate": 4.278879302245641e-06,
      "loss": 0.8053,
      "step": 4271
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2679646015167236,
      "learning_rate": 4.276779429778248e-06,
      "loss": 0.7615,
      "step": 4272
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.437036395072937,
      "learning_rate": 4.274679687589232e-06,
      "loss": 0.812,
      "step": 4273
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6533953547477722,
      "learning_rate": 4.272580076056838e-06,
      "loss": 1.0165,
      "step": 4274
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2775079011917114,
      "learning_rate": 4.270480595559278e-06,
      "loss": 0.8198,
      "step": 4275
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2480379343032837,
      "learning_rate": 4.268381246474749e-06,
      "loss": 0.8421,
      "step": 4276
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6287219524383545,
      "learning_rate": 4.266282029181419e-06,
      "loss": 1.0119,
      "step": 4277
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2763545513153076,
      "learning_rate": 4.264182944057429e-06,
      "loss": 0.7824,
      "step": 4278
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.197497844696045,
      "learning_rate": 4.262083991480908e-06,
      "loss": 0.7308,
      "step": 4279
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.611443042755127,
      "learning_rate": 4.259985171829949e-06,
      "loss": 1.0088,
      "step": 4280
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5937495231628418,
      "learning_rate": 4.257886485482627e-06,
      "loss": 0.9966,
      "step": 4281
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3207554817199707,
      "learning_rate": 4.255787932816993e-06,
      "loss": 0.8096,
      "step": 4282
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6140526533126831,
      "learning_rate": 4.253689514211071e-06,
      "loss": 1.0102,
      "step": 4283
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3200395107269287,
      "learning_rate": 4.251591230042865e-06,
      "loss": 0.7953,
      "step": 4284
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.28351628780365,
      "learning_rate": 4.249493080690348e-06,
      "loss": 0.7866,
      "step": 4285
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2832263708114624,
      "learning_rate": 4.247395066531479e-06,
      "loss": 0.7932,
      "step": 4286
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2154150009155273,
      "learning_rate": 4.245297187944184e-06,
      "loss": 0.7418,
      "step": 4287
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2557363510131836,
      "learning_rate": 4.243199445306365e-06,
      "loss": 0.7721,
      "step": 4288
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6864755153656006,
      "learning_rate": 4.241101838995907e-06,
      "loss": 1.0129,
      "step": 4289
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.254181981086731,
      "learning_rate": 4.239004369390659e-06,
      "loss": 0.8198,
      "step": 4290
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.403812289237976,
      "learning_rate": 4.236907036868458e-06,
      "loss": 0.7946,
      "step": 4291
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1657756567001343,
      "learning_rate": 4.234809841807108e-06,
      "loss": 0.7667,
      "step": 4292
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6180461645126343,
      "learning_rate": 4.2327127845843874e-06,
      "loss": 1.0313,
      "step": 4293
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2860076427459717,
      "learning_rate": 4.230615865578057e-06,
      "loss": 0.8121,
      "step": 4294
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2455722093582153,
      "learning_rate": 4.228519085165846e-06,
      "loss": 0.8283,
      "step": 4295
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2371667623519897,
      "learning_rate": 4.226422443725458e-06,
      "loss": 0.7932,
      "step": 4296
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3238469362258911,
      "learning_rate": 4.2243259416345805e-06,
      "loss": 0.7944,
      "step": 4297
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7147113084793091,
      "learning_rate": 4.222229579270865e-06,
      "loss": 1.0046,
      "step": 4298
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3682714700698853,
      "learning_rate": 4.220133357011946e-06,
      "loss": 0.7856,
      "step": 4299
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6284421682357788,
      "learning_rate": 4.218037275235429e-06,
      "loss": 1.0086,
      "step": 4300
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2954473495483398,
      "learning_rate": 4.2159413343188935e-06,
      "loss": 0.7817,
      "step": 4301
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2409940958023071,
      "learning_rate": 4.213845534639895e-06,
      "loss": 0.7716,
      "step": 4302
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7037816047668457,
      "learning_rate": 4.211749876575964e-06,
      "loss": 1.0274,
      "step": 4303
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7305372953414917,
      "learning_rate": 4.209654360504605e-06,
      "loss": 1.0371,
      "step": 4304
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6446923017501831,
      "learning_rate": 4.207558986803295e-06,
      "loss": 1.0405,
      "step": 4305
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2555609941482544,
      "learning_rate": 4.20546375584949e-06,
      "loss": 0.774,
      "step": 4306
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6081143021583557,
      "learning_rate": 4.203368668020616e-06,
      "loss": 1.0097,
      "step": 4307
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2853420972824097,
      "learning_rate": 4.201273723694073e-06,
      "loss": 0.8551,
      "step": 4308
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7274729609489441,
      "learning_rate": 4.199178923247241e-06,
      "loss": 1.0222,
      "step": 4309
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7039005160331726,
      "learning_rate": 4.197084267057466e-06,
      "loss": 1.0275,
      "step": 4310
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6237592101097107,
      "learning_rate": 4.194989755502073e-06,
      "loss": 1.0111,
      "step": 4311
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5900949835777283,
      "learning_rate": 4.192895388958361e-06,
      "loss": 1.0222,
      "step": 4312
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2880626916885376,
      "learning_rate": 4.1908011678036e-06,
      "loss": 0.8219,
      "step": 4313
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.307192325592041,
      "learning_rate": 4.188707092415038e-06,
      "loss": 0.8277,
      "step": 4314
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1829150915145874,
      "learning_rate": 4.186613163169892e-06,
      "loss": 0.7554,
      "step": 4315
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4170002937316895,
      "learning_rate": 4.184519380445355e-06,
      "loss": 0.7901,
      "step": 4316
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7254272699356079,
      "learning_rate": 4.182425744618595e-06,
      "loss": 1.0146,
      "step": 4317
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7020663022994995,
      "learning_rate": 4.18033225606675e-06,
      "loss": 1.0245,
      "step": 4318
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2582036256790161,
      "learning_rate": 4.178238915166937e-06,
      "loss": 0.7783,
      "step": 4319
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2693666219711304,
      "learning_rate": 4.176145722296241e-06,
      "loss": 0.8042,
      "step": 4320
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6010758876800537,
      "learning_rate": 4.174052677831722e-06,
      "loss": 1.0312,
      "step": 4321
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3024439811706543,
      "learning_rate": 4.171959782150412e-06,
      "loss": 0.8189,
      "step": 4322
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2107608318328857,
      "learning_rate": 4.1698670356293205e-06,
      "loss": 0.7773,
      "step": 4323
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3903452157974243,
      "learning_rate": 4.167774438645428e-06,
      "loss": 0.7411,
      "step": 4324
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2439777851104736,
      "learning_rate": 4.165681991575686e-06,
      "loss": 0.7879,
      "step": 4325
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2364400625228882,
      "learning_rate": 4.163589694797018e-06,
      "loss": 0.7958,
      "step": 4326
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2568769454956055,
      "learning_rate": 4.161497548686327e-06,
      "loss": 0.7999,
      "step": 4327
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2761938571929932,
      "learning_rate": 4.1594055536204835e-06,
      "loss": 0.8088,
      "step": 4328
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7296198010444641,
      "learning_rate": 4.15731370997633e-06,
      "loss": 1.0338,
      "step": 4329
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2310423851013184,
      "learning_rate": 4.1552220181306875e-06,
      "loss": 0.7482,
      "step": 4330
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2706761360168457,
      "learning_rate": 4.153130478460341e-06,
      "loss": 0.7584,
      "step": 4331
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.4686872959136963,
      "learning_rate": 4.151039091342058e-06,
      "loss": 0.8131,
      "step": 4332
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2298067808151245,
      "learning_rate": 4.148947857152572e-06,
      "loss": 0.7826,
      "step": 4333
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2583550214767456,
      "learning_rate": 4.146856776268588e-06,
      "loss": 0.8183,
      "step": 4334
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2186930179595947,
      "learning_rate": 4.14476584906679e-06,
      "loss": 0.7693,
      "step": 4335
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.270293116569519,
      "learning_rate": 4.142675075923825e-06,
      "loss": 0.7803,
      "step": 4336
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6643315553665161,
      "learning_rate": 4.1405844572163225e-06,
      "loss": 1.0388,
      "step": 4337
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2780301570892334,
      "learning_rate": 4.138493993320877e-06,
      "loss": 0.7644,
      "step": 4338
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1886507272720337,
      "learning_rate": 4.136403684614055e-06,
      "loss": 0.772,
      "step": 4339
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.242806315422058,
      "learning_rate": 4.1343135314724026e-06,
      "loss": 0.7669,
      "step": 4340
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2784168720245361,
      "learning_rate": 4.132223534272427e-06,
      "loss": 0.8106,
      "step": 4341
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5955015420913696,
      "learning_rate": 4.130133693390616e-06,
      "loss": 1.0089,
      "step": 4342
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.313904881477356,
      "learning_rate": 4.128044009203427e-06,
      "loss": 0.8565,
      "step": 4343
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2702360153198242,
      "learning_rate": 4.125954482087284e-06,
      "loss": 0.7906,
      "step": 4344
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3994314670562744,
      "learning_rate": 4.123865112418593e-06,
      "loss": 0.8546,
      "step": 4345
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2211463451385498,
      "learning_rate": 4.121775900573718e-06,
      "loss": 0.7577,
      "step": 4346
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2567033767700195,
      "learning_rate": 4.119686846929009e-06,
      "loss": 0.7852,
      "step": 4347
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6111962795257568,
      "learning_rate": 4.117597951860779e-06,
      "loss": 1.0354,
      "step": 4348
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.276066541671753,
      "learning_rate": 4.11550921574531e-06,
      "loss": 0.895,
      "step": 4349
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6086326241493225,
      "learning_rate": 4.113420638958866e-06,
      "loss": 1.0164,
      "step": 4350
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3251878023147583,
      "learning_rate": 4.111332221877669e-06,
      "loss": 0.8282,
      "step": 4351
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4622042179107666,
      "learning_rate": 4.109243964877926e-06,
      "loss": 0.841,
      "step": 4352
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2746691703796387,
      "learning_rate": 4.1071558683358035e-06,
      "loss": 0.838,
      "step": 4353
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6072760820388794,
      "learning_rate": 4.105067932627443e-06,
      "loss": 1.0359,
      "step": 4354
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2157618999481201,
      "learning_rate": 4.102980158128963e-06,
      "loss": 0.7666,
      "step": 4355
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2379730939865112,
      "learning_rate": 4.1008925452164435e-06,
      "loss": 0.7353,
      "step": 4356
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2480623722076416,
      "learning_rate": 4.098805094265939e-06,
      "loss": 0.8007,
      "step": 4357
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3205004930496216,
      "learning_rate": 4.09671780565348e-06,
      "loss": 0.8362,
      "step": 4358
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.252297043800354,
      "learning_rate": 4.094630679755058e-06,
      "loss": 0.7086,
      "step": 4359
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2777929306030273,
      "learning_rate": 4.092543716946644e-06,
      "loss": 0.7815,
      "step": 4360
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7485051155090332,
      "learning_rate": 4.090456917604176e-06,
      "loss": 0.7841,
      "step": 4361
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.308250904083252,
      "learning_rate": 4.08837028210356e-06,
      "loss": 0.799,
      "step": 4362
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.172869086265564,
      "learning_rate": 4.086283810820677e-06,
      "loss": 0.7925,
      "step": 4363
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6468296051025391,
      "learning_rate": 4.0841975041313766e-06,
      "loss": 1.0201,
      "step": 4364
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6282566785812378,
      "learning_rate": 4.082111362411478e-06,
      "loss": 1.0198,
      "step": 4365
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2512083053588867,
      "learning_rate": 4.0800253860367725e-06,
      "loss": 0.785,
      "step": 4366
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3081696033477783,
      "learning_rate": 4.077939575383017e-06,
      "loss": 0.8468,
      "step": 4367
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2412406206130981,
      "learning_rate": 4.075853930825947e-06,
      "loss": 0.7814,
      "step": 4368
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2312597036361694,
      "learning_rate": 4.073768452741258e-06,
      "loss": 0.7889,
      "step": 4369
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2028567790985107,
      "learning_rate": 4.071683141504625e-06,
      "loss": 0.7825,
      "step": 4370
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.253848910331726,
      "learning_rate": 4.069597997491688e-06,
      "loss": 0.7864,
      "step": 4371
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.238576054573059,
      "learning_rate": 4.0675130210780525e-06,
      "loss": 0.8115,
      "step": 4372
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.273917317390442,
      "learning_rate": 4.065428212639306e-06,
      "loss": 0.8046,
      "step": 4373
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2694628238677979,
      "learning_rate": 4.063343572550992e-06,
      "loss": 0.7697,
      "step": 4374
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2285513877868652,
      "learning_rate": 4.061259101188635e-06,
      "loss": 0.8343,
      "step": 4375
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3278595209121704,
      "learning_rate": 4.0591747989277225e-06,
      "loss": 0.7616,
      "step": 4376
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2690893411636353,
      "learning_rate": 4.057090666143711e-06,
      "loss": 0.7681,
      "step": 4377
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6423009634017944,
      "learning_rate": 4.055006703212033e-06,
      "loss": 1.0123,
      "step": 4378
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4801613092422485,
      "learning_rate": 4.052922910508083e-06,
      "loss": 0.7247,
      "step": 4379
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.32061767578125,
      "learning_rate": 4.0508392884072285e-06,
      "loss": 0.8371,
      "step": 4380
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2452068328857422,
      "learning_rate": 4.048755837284807e-06,
      "loss": 0.7729,
      "step": 4381
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2734899520874023,
      "learning_rate": 4.0466725575161235e-06,
      "loss": 0.8149,
      "step": 4382
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6112028360366821,
      "learning_rate": 4.044589449476451e-06,
      "loss": 1.0283,
      "step": 4383
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3347671031951904,
      "learning_rate": 4.0425065135410364e-06,
      "loss": 0.8242,
      "step": 4384
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2984930276870728,
      "learning_rate": 4.040423750085089e-06,
      "loss": 0.8064,
      "step": 4385
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6201193928718567,
      "learning_rate": 4.038341159483793e-06,
      "loss": 1.027,
      "step": 4386
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.22946035861969,
      "learning_rate": 4.036258742112296e-06,
      "loss": 0.7574,
      "step": 4387
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6021016240119934,
      "learning_rate": 4.03417649834572e-06,
      "loss": 1.0355,
      "step": 4388
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5953752994537354,
      "learning_rate": 4.032094428559153e-06,
      "loss": 1.0185,
      "step": 4389
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4224821329116821,
      "learning_rate": 4.030012533127649e-06,
      "loss": 0.7636,
      "step": 4390
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.348263144493103,
      "learning_rate": 4.027930812426236e-06,
      "loss": 0.8286,
      "step": 4391
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3070666790008545,
      "learning_rate": 4.0258492668299045e-06,
      "loss": 0.8292,
      "step": 4392
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5893195867538452,
      "learning_rate": 4.023767896713621e-06,
      "loss": 1.0154,
      "step": 4393
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2381751537322998,
      "learning_rate": 4.021686702452313e-06,
      "loss": 0.8243,
      "step": 4394
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2512816190719604,
      "learning_rate": 4.01960568442088e-06,
      "loss": 0.8313,
      "step": 4395
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2521708011627197,
      "learning_rate": 4.017524842994191e-06,
      "loss": 0.7867,
      "step": 4396
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6227417588233948,
      "learning_rate": 4.015444178547076e-06,
      "loss": 1.0102,
      "step": 4397
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2813420295715332,
      "learning_rate": 4.013363691454347e-06,
      "loss": 0.7729,
      "step": 4398
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5630403757095337,
      "learning_rate": 4.01128338209077e-06,
      "loss": 0.7375,
      "step": 4399
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2636289596557617,
      "learning_rate": 4.009203250831083e-06,
      "loss": 0.801,
      "step": 4400
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.240052342414856,
      "learning_rate": 4.007123298049998e-06,
      "loss": 0.7363,
      "step": 4401
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2105722427368164,
      "learning_rate": 4.0050435241221876e-06,
      "loss": 0.8125,
      "step": 4402
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6097171902656555,
      "learning_rate": 4.002963929422296e-06,
      "loss": 0.9965,
      "step": 4403
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2586207389831543,
      "learning_rate": 4.0008845143249325e-06,
      "loss": 0.8166,
      "step": 4404
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3099395036697388,
      "learning_rate": 3.998805279204676e-06,
      "loss": 0.7922,
      "step": 4405
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5879127383232117,
      "learning_rate": 3.996726224436075e-06,
      "loss": 1.016,
      "step": 4406
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.274271845817566,
      "learning_rate": 3.994647350393638e-06,
      "loss": 0.7765,
      "step": 4407
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2814013957977295,
      "learning_rate": 3.992568657451852e-06,
      "loss": 0.7932,
      "step": 4408
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2161606550216675,
      "learning_rate": 3.990490145985162e-06,
      "loss": 0.735,
      "step": 4409
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1827926635742188,
      "learning_rate": 3.988411816367981e-06,
      "loss": 0.7807,
      "step": 4410
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6046264171600342,
      "learning_rate": 3.986333668974698e-06,
      "loss": 1.0385,
      "step": 4411
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3398343324661255,
      "learning_rate": 3.9842557041796595e-06,
      "loss": 0.7865,
      "step": 4412
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.167141079902649,
      "learning_rate": 3.982177922357181e-06,
      "loss": 0.7657,
      "step": 4413
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6127106547355652,
      "learning_rate": 3.980100323881551e-06,
      "loss": 1.0257,
      "step": 4414
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2849113941192627,
      "learning_rate": 3.978022909127017e-06,
      "loss": 0.7753,
      "step": 4415
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.222862720489502,
      "learning_rate": 3.9759456784678e-06,
      "loss": 0.8021,
      "step": 4416
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2933906316757202,
      "learning_rate": 3.973868632278084e-06,
      "loss": 0.7892,
      "step": 4417
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5826097726821899,
      "learning_rate": 3.971791770932018e-06,
      "loss": 1.0293,
      "step": 4418
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2797940969467163,
      "learning_rate": 3.969715094803724e-06,
      "loss": 0.7848,
      "step": 4419
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.58852618932724,
      "learning_rate": 3.967638604267286e-06,
      "loss": 1.0113,
      "step": 4420
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5988748669624329,
      "learning_rate": 3.965562299696754e-06,
      "loss": 1.0257,
      "step": 4421
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.269289255142212,
      "learning_rate": 3.963486181466151e-06,
      "loss": 0.7534,
      "step": 4422
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2694138288497925,
      "learning_rate": 3.961410249949455e-06,
      "loss": 0.7769,
      "step": 4423
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.249536156654358,
      "learning_rate": 3.95933450552062e-06,
      "loss": 0.7315,
      "step": 4424
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2681150436401367,
      "learning_rate": 3.957258948553563e-06,
      "loss": 0.8362,
      "step": 4425
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5831018090248108,
      "learning_rate": 3.95518357942217e-06,
      "loss": 1.0362,
      "step": 4426
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2317718267440796,
      "learning_rate": 3.953108398500287e-06,
      "loss": 0.7395,
      "step": 4427
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2195088863372803,
      "learning_rate": 3.951033406161729e-06,
      "loss": 0.8193,
      "step": 4428
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2399790287017822,
      "learning_rate": 3.948958602780283e-06,
      "loss": 0.7842,
      "step": 4429
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.233426809310913,
      "learning_rate": 3.9468839887296895e-06,
      "loss": 0.7493,
      "step": 4430
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6286717653274536,
      "learning_rate": 3.944809564383669e-06,
      "loss": 1.0308,
      "step": 4431
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2379251718521118,
      "learning_rate": 3.942735330115897e-06,
      "loss": 0.87,
      "step": 4432
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2910794019699097,
      "learning_rate": 3.940661286300018e-06,
      "loss": 0.7608,
      "step": 4433
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2999165058135986,
      "learning_rate": 3.938587433309646e-06,
      "loss": 0.7953,
      "step": 4434
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.259970784187317,
      "learning_rate": 3.936513771518354e-06,
      "loss": 0.7616,
      "step": 4435
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3006136417388916,
      "learning_rate": 3.934440301299689e-06,
      "loss": 0.8018,
      "step": 4436
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2971786260604858,
      "learning_rate": 3.932367023027155e-06,
      "loss": 0.7859,
      "step": 4437
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2644495964050293,
      "learning_rate": 3.930293937074224e-06,
      "loss": 0.8006,
      "step": 4438
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.281880497932434,
      "learning_rate": 3.928221043814336e-06,
      "loss": 0.7699,
      "step": 4439
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6401193737983704,
      "learning_rate": 3.926148343620896e-06,
      "loss": 1.026,
      "step": 4440
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.305582046508789,
      "learning_rate": 3.924075836867268e-06,
      "loss": 0.778,
      "step": 4441
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6149711012840271,
      "learning_rate": 3.922003523926791e-06,
      "loss": 1.0107,
      "step": 4442
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2303779125213623,
      "learning_rate": 3.919931405172763e-06,
      "loss": 0.7766,
      "step": 4443
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2160003185272217,
      "learning_rate": 3.917859480978447e-06,
      "loss": 0.7562,
      "step": 4444
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2736496925354004,
      "learning_rate": 3.915787751717073e-06,
      "loss": 0.7573,
      "step": 4445
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.297048807144165,
      "learning_rate": 3.9137162177618335e-06,
      "loss": 0.8182,
      "step": 4446
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.225154161453247,
      "learning_rate": 3.911644879485889e-06,
      "loss": 0.813,
      "step": 4447
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.295322299003601,
      "learning_rate": 3.90957373726236e-06,
      "loss": 0.74,
      "step": 4448
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2419679164886475,
      "learning_rate": 3.907502791464339e-06,
      "loss": 0.7937,
      "step": 4449
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2758582830429077,
      "learning_rate": 3.905432042464877e-06,
      "loss": 0.8036,
      "step": 4450
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3392248153686523,
      "learning_rate": 3.903361490636989e-06,
      "loss": 0.7398,
      "step": 4451
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6997053623199463,
      "learning_rate": 3.90129113635366e-06,
      "loss": 1.0145,
      "step": 4452
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2509957551956177,
      "learning_rate": 3.899220979987834e-06,
      "loss": 0.7654,
      "step": 4453
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2660009860992432,
      "learning_rate": 3.8971510219124246e-06,
      "loss": 0.8087,
      "step": 4454
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6267756223678589,
      "learning_rate": 3.8950812625003044e-06,
      "loss": 1.0177,
      "step": 4455
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2022126913070679,
      "learning_rate": 3.893011702124311e-06,
      "loss": 0.707,
      "step": 4456
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2849295139312744,
      "learning_rate": 3.890942341157251e-06,
      "loss": 0.8111,
      "step": 4457
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6081826090812683,
      "learning_rate": 3.8888731799718885e-06,
      "loss": 1.0128,
      "step": 4458
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6313122510910034,
      "learning_rate": 3.886804218940959e-06,
      "loss": 1.0331,
      "step": 4459
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6220429539680481,
      "learning_rate": 3.884735458437155e-06,
      "loss": 1.0044,
      "step": 4460
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6111922264099121,
      "learning_rate": 3.882666898833135e-06,
      "loss": 1.0331,
      "step": 4461
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.349050760269165,
      "learning_rate": 3.880598540501523e-06,
      "loss": 0.8278,
      "step": 4462
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.599331259727478,
      "learning_rate": 3.878530383814905e-06,
      "loss": 1.0318,
      "step": 4463
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2010552883148193,
      "learning_rate": 3.8764624291458346e-06,
      "loss": 0.7562,
      "step": 4464
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6061270833015442,
      "learning_rate": 3.874394676866821e-06,
      "loss": 1.0306,
      "step": 4465
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2379710674285889,
      "learning_rate": 3.872327127350344e-06,
      "loss": 0.7525,
      "step": 4466
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2527014017105103,
      "learning_rate": 3.870259780968846e-06,
      "loss": 0.7719,
      "step": 4467
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6116620898246765,
      "learning_rate": 3.868192638094727e-06,
      "loss": 1.0377,
      "step": 4468
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2507030963897705,
      "learning_rate": 3.866125699100361e-06,
      "loss": 0.8464,
      "step": 4469
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6570430994033813,
      "learning_rate": 3.864058964358075e-06,
      "loss": 1.0344,
      "step": 4470
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2521042823791504,
      "learning_rate": 3.8619924342401625e-06,
      "loss": 0.7791,
      "step": 4471
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2065141201019287,
      "learning_rate": 3.859926109118884e-06,
      "loss": 0.7852,
      "step": 4472
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.632037341594696,
      "learning_rate": 3.857859989366459e-06,
      "loss": 1.0106,
      "step": 4473
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2408753633499146,
      "learning_rate": 3.855794075355068e-06,
      "loss": 0.8322,
      "step": 4474
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2520227432250977,
      "learning_rate": 3.853728367456862e-06,
      "loss": 0.7384,
      "step": 4475
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2396011352539062,
      "learning_rate": 3.851662866043945e-06,
      "loss": 0.8101,
      "step": 4476
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2729649543762207,
      "learning_rate": 3.849597571488395e-06,
      "loss": 0.8167,
      "step": 4477
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3051421642303467,
      "learning_rate": 3.847532484162244e-06,
      "loss": 0.7616,
      "step": 4478
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2393805980682373,
      "learning_rate": 3.845467604437486e-06,
      "loss": 0.7616,
      "step": 4479
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6773916482925415,
      "learning_rate": 3.843402932686088e-06,
      "loss": 1.0155,
      "step": 4480
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6646068096160889,
      "learning_rate": 3.841338469279967e-06,
      "loss": 1.0175,
      "step": 4481
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2216367721557617,
      "learning_rate": 3.839274214591011e-06,
      "loss": 0.8096,
      "step": 4482
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5955845713615417,
      "learning_rate": 3.8372101689910666e-06,
      "loss": 1.0289,
      "step": 4483
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.187840223312378,
      "learning_rate": 3.835146332851943e-06,
      "loss": 0.7452,
      "step": 4484
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2777193784713745,
      "learning_rate": 3.833082706545414e-06,
      "loss": 0.796,
      "step": 4485
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.240709662437439,
      "learning_rate": 3.831019290443211e-06,
      "loss": 0.8257,
      "step": 4486
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6508936285972595,
      "learning_rate": 3.828956084917035e-06,
      "loss": 1.0088,
      "step": 4487
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3048266172409058,
      "learning_rate": 3.826893090338541e-06,
      "loss": 0.8356,
      "step": 4488
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2714296579360962,
      "learning_rate": 3.824830307079348e-06,
      "loss": 0.7337,
      "step": 4489
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.252669334411621,
      "learning_rate": 3.822767735511043e-06,
      "loss": 0.7856,
      "step": 4490
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2582534551620483,
      "learning_rate": 3.820705376005166e-06,
      "loss": 0.7122,
      "step": 4491
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6588323712348938,
      "learning_rate": 3.818643228933228e-06,
      "loss": 1.0396,
      "step": 4492
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3436838388442993,
      "learning_rate": 3.816581294666693e-06,
      "loss": 0.7582,
      "step": 4493
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6061258912086487,
      "learning_rate": 3.814519573576989e-06,
      "loss": 1.0331,
      "step": 4494
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5961589813232422,
      "learning_rate": 3.8124580660355128e-06,
      "loss": 1.042,
      "step": 4495
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3067779541015625,
      "learning_rate": 3.810396772413611e-06,
      "loss": 0.7969,
      "step": 4496
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.29735267162323,
      "learning_rate": 3.8083356930826025e-06,
      "loss": 0.7204,
      "step": 4497
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2902222871780396,
      "learning_rate": 3.8062748284137612e-06,
      "loss": 0.7968,
      "step": 4498
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2926658391952515,
      "learning_rate": 3.8042141787783215e-06,
      "loss": 0.8681,
      "step": 4499
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2605434656143188,
      "learning_rate": 3.8021537445474854e-06,
      "loss": 0.7988,
      "step": 4500
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6597552299499512,
      "learning_rate": 3.80009352609241e-06,
      "loss": 1.0281,
      "step": 4501
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.274012804031372,
      "learning_rate": 3.798033523784214e-06,
      "loss": 0.8105,
      "step": 4502
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2663239240646362,
      "learning_rate": 3.795973737993983e-06,
      "loss": 0.7737,
      "step": 4503
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2617316246032715,
      "learning_rate": 3.7939141690927567e-06,
      "loss": 0.7105,
      "step": 4504
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6220371127128601,
      "learning_rate": 3.79185481745154e-06,
      "loss": 1.033,
      "step": 4505
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6264622807502747,
      "learning_rate": 3.789795683441295e-06,
      "loss": 1.0254,
      "step": 4506
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1938307285308838,
      "learning_rate": 3.7877367674329484e-06,
      "loss": 0.8271,
      "step": 4507
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2782654762268066,
      "learning_rate": 3.785678069797387e-06,
      "loss": 0.7552,
      "step": 4508
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.599459707736969,
      "learning_rate": 3.7836195909054534e-06,
      "loss": 1.0154,
      "step": 4509
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3331077098846436,
      "learning_rate": 3.78156133112796e-06,
      "loss": 0.7987,
      "step": 4510
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.223341464996338,
      "learning_rate": 3.779503290835671e-06,
      "loss": 0.7309,
      "step": 4511
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6344196200370789,
      "learning_rate": 3.777445470399313e-06,
      "loss": 1.0207,
      "step": 4512
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2901864051818848,
      "learning_rate": 3.7753878701895792e-06,
      "loss": 0.8575,
      "step": 4513
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3037917613983154,
      "learning_rate": 3.773330490577113e-06,
      "loss": 0.8472,
      "step": 4514
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2636650800704956,
      "learning_rate": 3.7712733319325288e-06,
      "loss": 0.7785,
      "step": 4515
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.611797034740448,
      "learning_rate": 3.7692163946263925e-06,
      "loss": 1.013,
      "step": 4516
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6187926530838013,
      "learning_rate": 3.767159679029233e-06,
      "loss": 1.0292,
      "step": 4517
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3056365251541138,
      "learning_rate": 3.765103185511543e-06,
      "loss": 0.7633,
      "step": 4518
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5985475778579712,
      "learning_rate": 3.7630469144437664e-06,
      "loss": 0.9947,
      "step": 4519
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.25906503200531,
      "learning_rate": 3.760990866196319e-06,
      "loss": 0.8092,
      "step": 4520
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6166753172874451,
      "learning_rate": 3.7589350411395674e-06,
      "loss": 1.0188,
      "step": 4521
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1981526613235474,
      "learning_rate": 3.7568794396438387e-06,
      "loss": 0.7552,
      "step": 4522
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3195284605026245,
      "learning_rate": 3.7548240620794228e-06,
      "loss": 0.7757,
      "step": 4523
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2475892305374146,
      "learning_rate": 3.7527689088165687e-06,
      "loss": 0.7796,
      "step": 4524
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.291555404663086,
      "learning_rate": 3.7507139802254855e-06,
      "loss": 0.751,
      "step": 4525
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2761331796646118,
      "learning_rate": 3.7486592766763375e-06,
      "loss": 0.7394,
      "step": 4526
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2527498006820679,
      "learning_rate": 3.746604798539254e-06,
      "loss": 0.7978,
      "step": 4527
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2414051294326782,
      "learning_rate": 3.7445505461843225e-06,
      "loss": 0.8022,
      "step": 4528
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1973457336425781,
      "learning_rate": 3.7424965199815866e-06,
      "loss": 0.7333,
      "step": 4529
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2562918663024902,
      "learning_rate": 3.7404427203010497e-06,
      "loss": 0.7307,
      "step": 4530
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.264275074005127,
      "learning_rate": 3.73838914751268e-06,
      "loss": 0.8102,
      "step": 4531
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2517778873443604,
      "learning_rate": 3.736335801986396e-06,
      "loss": 0.7513,
      "step": 4532
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2538191080093384,
      "learning_rate": 3.7342826840920863e-06,
      "loss": 0.7595,
      "step": 4533
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7641403675079346,
      "learning_rate": 3.7322297941995884e-06,
      "loss": 1.0254,
      "step": 4534
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6924797296524048,
      "learning_rate": 3.7301771326787008e-06,
      "loss": 1.0198,
      "step": 4535
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2992504835128784,
      "learning_rate": 3.728124699899187e-06,
      "loss": 0.8321,
      "step": 4536
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.331688642501831,
      "learning_rate": 3.7260724962307605e-06,
      "loss": 0.7726,
      "step": 4537
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3530185222625732,
      "learning_rate": 3.7240205220431025e-06,
      "loss": 0.7254,
      "step": 4538
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2770986557006836,
      "learning_rate": 3.7219687777058468e-06,
      "loss": 0.7489,
      "step": 4539
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6630486249923706,
      "learning_rate": 3.719917263588584e-06,
      "loss": 1.0321,
      "step": 4540
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2491017580032349,
      "learning_rate": 3.7178659800608706e-06,
      "loss": 0.8029,
      "step": 4541
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2596102952957153,
      "learning_rate": 3.715814927492216e-06,
      "loss": 0.7449,
      "step": 4542
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2395164966583252,
      "learning_rate": 3.7137641062520892e-06,
      "loss": 0.8633,
      "step": 4543
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6351819038391113,
      "learning_rate": 3.7117135167099194e-06,
      "loss": 1.0232,
      "step": 4544
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.663481593132019,
      "learning_rate": 3.7096631592350895e-06,
      "loss": 1.0067,
      "step": 4545
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2156132459640503,
      "learning_rate": 3.7076130341969474e-06,
      "loss": 0.7654,
      "step": 4546
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.296493649482727,
      "learning_rate": 3.705563141964791e-06,
      "loss": 0.7404,
      "step": 4547
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.291279911994934,
      "learning_rate": 3.7035134829078844e-06,
      "loss": 0.8211,
      "step": 4548
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3180776834487915,
      "learning_rate": 3.7014640573954454e-06,
      "loss": 0.8471,
      "step": 4549
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.256178379058838,
      "learning_rate": 3.6994148657966465e-06,
      "loss": 0.7802,
      "step": 4550
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2515007257461548,
      "learning_rate": 3.697365908480627e-06,
      "loss": 0.7974,
      "step": 4551
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2787327766418457,
      "learning_rate": 3.6953171858164745e-06,
      "loss": 0.7183,
      "step": 4552
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.823066234588623,
      "learning_rate": 3.693268698173242e-06,
      "loss": 1.0061,
      "step": 4553
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3125951290130615,
      "learning_rate": 3.6912204459199352e-06,
      "loss": 0.8134,
      "step": 4554
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.254546046257019,
      "learning_rate": 3.689172429425517e-06,
      "loss": 0.7678,
      "step": 4555
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2148382663726807,
      "learning_rate": 3.6871246490589143e-06,
      "loss": 0.7394,
      "step": 4556
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1904499530792236,
      "learning_rate": 3.6850771051890035e-06,
      "loss": 0.7667,
      "step": 4557
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2835021018981934,
      "learning_rate": 3.6830297981846206e-06,
      "loss": 0.7819,
      "step": 4558
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.228991985321045,
      "learning_rate": 3.6809827284145655e-06,
      "loss": 0.7708,
      "step": 4559
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6404527425765991,
      "learning_rate": 3.678935896247584e-06,
      "loss": 1.015,
      "step": 4560
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.253074288368225,
      "learning_rate": 3.676889302052391e-06,
      "loss": 0.8016,
      "step": 4561
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2717260122299194,
      "learning_rate": 3.6748429461976493e-06,
      "loss": 0.8166,
      "step": 4562
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2904090881347656,
      "learning_rate": 3.6727968290519816e-06,
      "loss": 0.8077,
      "step": 4563
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3441901206970215,
      "learning_rate": 3.6707509509839696e-06,
      "loss": 0.7965,
      "step": 4564
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2308604717254639,
      "learning_rate": 3.6687053123621497e-06,
      "loss": 0.7089,
      "step": 4565
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.310370922088623,
      "learning_rate": 3.6666599135550172e-06,
      "loss": 0.8171,
      "step": 4566
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3300609588623047,
      "learning_rate": 3.6646147549310217e-06,
      "loss": 0.804,
      "step": 4567
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2890568971633911,
      "learning_rate": 3.662569836858571e-06,
      "loss": 0.8085,
      "step": 4568
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2030315399169922,
      "learning_rate": 3.660525159706031e-06,
      "loss": 0.7008,
      "step": 4569
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6521837115287781,
      "learning_rate": 3.6584807238417185e-06,
      "loss": 1.0159,
      "step": 4570
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3059253692626953,
      "learning_rate": 3.6564365296339153e-06,
      "loss": 0.8161,
      "step": 4571
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1789906024932861,
      "learning_rate": 3.6543925774508546e-06,
      "loss": 0.7163,
      "step": 4572
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6455694437026978,
      "learning_rate": 3.6523488676607235e-06,
      "loss": 1.0096,
      "step": 4573
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2402852773666382,
      "learning_rate": 3.650305400631672e-06,
      "loss": 0.7821,
      "step": 4574
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1914706230163574,
      "learning_rate": 3.6482621767317995e-06,
      "loss": 0.7551,
      "step": 4575
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6063261032104492,
      "learning_rate": 3.64621919632917e-06,
      "loss": 1.0303,
      "step": 4576
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.601006269454956,
      "learning_rate": 3.6441764597917954e-06,
      "loss": 1.0046,
      "step": 4577
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6227284669876099,
      "learning_rate": 3.642133967487646e-06,
      "loss": 1.0235,
      "step": 4578
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2721959352493286,
      "learning_rate": 3.640091719784653e-06,
      "loss": 0.7806,
      "step": 4579
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2220481634140015,
      "learning_rate": 3.6380497170506957e-06,
      "loss": 0.7523,
      "step": 4580
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2163447141647339,
      "learning_rate": 3.636007959653615e-06,
      "loss": 0.7695,
      "step": 4581
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2578115463256836,
      "learning_rate": 3.6339664479612064e-06,
      "loss": 0.7686,
      "step": 4582
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2818886041641235,
      "learning_rate": 3.6319251823412187e-06,
      "loss": 0.7791,
      "step": 4583
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.211789608001709,
      "learning_rate": 3.6298841631613598e-06,
      "loss": 0.7063,
      "step": 4584
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1886688470840454,
      "learning_rate": 3.6278433907892906e-06,
      "loss": 0.7642,
      "step": 4585
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2879656553268433,
      "learning_rate": 3.62580286559263e-06,
      "loss": 0.8144,
      "step": 4586
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3535242080688477,
      "learning_rate": 3.623762587938951e-06,
      "loss": 0.8839,
      "step": 4587
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2372759580612183,
      "learning_rate": 3.6217225581957784e-06,
      "loss": 0.7967,
      "step": 4588
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4225130081176758,
      "learning_rate": 3.6196827767306003e-06,
      "loss": 0.8163,
      "step": 4589
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2368499040603638,
      "learning_rate": 3.6176432439108545e-06,
      "loss": 0.7686,
      "step": 4590
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.237102746963501,
      "learning_rate": 3.6156039601039315e-06,
      "loss": 0.7732,
      "step": 4591
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3424174785614014,
      "learning_rate": 3.6135649256771865e-06,
      "loss": 0.7654,
      "step": 4592
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.276193380355835,
      "learning_rate": 3.6115261409979183e-06,
      "loss": 0.8,
      "step": 4593
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6699851751327515,
      "learning_rate": 3.6094876064333917e-06,
      "loss": 1.0115,
      "step": 4594
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2706784009933472,
      "learning_rate": 3.607449322350819e-06,
      "loss": 0.8288,
      "step": 4595
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2605488300323486,
      "learning_rate": 3.6054112891173657e-06,
      "loss": 0.8098,
      "step": 4596
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2614697217941284,
      "learning_rate": 3.603373507100162e-06,
      "loss": 0.8015,
      "step": 4597
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2371646165847778,
      "learning_rate": 3.6013359766662814e-06,
      "loss": 0.8316,
      "step": 4598
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2500970363616943,
      "learning_rate": 3.5992986981827615e-06,
      "loss": 0.774,
      "step": 4599
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3903988599777222,
      "learning_rate": 3.597261672016589e-06,
      "loss": 0.7956,
      "step": 4600
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.224275827407837,
      "learning_rate": 3.5952248985347037e-06,
      "loss": 0.775,
      "step": 4601
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2500958442687988,
      "learning_rate": 3.5931883781040078e-06,
      "loss": 0.7247,
      "step": 4602
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2627283334732056,
      "learning_rate": 3.5911521110913483e-06,
      "loss": 0.812,
      "step": 4603
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2083637714385986,
      "learning_rate": 3.5891160978635348e-06,
      "loss": 0.7575,
      "step": 4604
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2446256875991821,
      "learning_rate": 3.5870803387873243e-06,
      "loss": 0.7796,
      "step": 4605
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6250948905944824,
      "learning_rate": 3.585044834229433e-06,
      "loss": 0.989,
      "step": 4606
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6096885800361633,
      "learning_rate": 3.5830095845565297e-06,
      "loss": 1.0329,
      "step": 4607
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2107268571853638,
      "learning_rate": 3.5809745901352344e-06,
      "loss": 0.7658,
      "step": 4608
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2623263597488403,
      "learning_rate": 3.5789398513321284e-06,
      "loss": 0.7657,
      "step": 4609
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2335059642791748,
      "learning_rate": 3.576905368513739e-06,
      "loss": 0.7719,
      "step": 4610
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4278714656829834,
      "learning_rate": 3.5748711420465487e-06,
      "loss": 0.8012,
      "step": 4611
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6258023381233215,
      "learning_rate": 3.572837172297001e-06,
      "loss": 1.0198,
      "step": 4612
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6258543133735657,
      "learning_rate": 3.570803459631484e-06,
      "loss": 1.0098,
      "step": 4613
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2701970338821411,
      "learning_rate": 3.5687700044163464e-06,
      "loss": 0.8451,
      "step": 4614
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.25296151638031,
      "learning_rate": 3.566736807017886e-06,
      "loss": 0.7454,
      "step": 4615
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.255069375038147,
      "learning_rate": 3.5647038678023537e-06,
      "loss": 0.8454,
      "step": 4616
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.286526083946228,
      "learning_rate": 3.5626711871359598e-06,
      "loss": 0.8232,
      "step": 4617
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2411069869995117,
      "learning_rate": 3.5606387653848627e-06,
      "loss": 0.7651,
      "step": 4618
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2501871585845947,
      "learning_rate": 3.5586066029151735e-06,
      "loss": 0.7844,
      "step": 4619
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.243504524230957,
      "learning_rate": 3.556574700092963e-06,
      "loss": 0.7975,
      "step": 4620
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1981695890426636,
      "learning_rate": 3.5545430572842467e-06,
      "loss": 0.7804,
      "step": 4621
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6618080735206604,
      "learning_rate": 3.5525116748549994e-06,
      "loss": 1.0204,
      "step": 4622
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.275306224822998,
      "learning_rate": 3.550480553171148e-06,
      "loss": 0.7967,
      "step": 4623
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1922274827957153,
      "learning_rate": 3.548449692598569e-06,
      "loss": 0.7675,
      "step": 4624
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2576755285263062,
      "learning_rate": 3.5464190935030967e-06,
      "loss": 0.7866,
      "step": 4625
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.213433861732483,
      "learning_rate": 3.5443887562505153e-06,
      "loss": 0.7714,
      "step": 4626
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.302939534187317,
      "learning_rate": 3.5423586812065626e-06,
      "loss": 0.7907,
      "step": 4627
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3064688444137573,
      "learning_rate": 3.5403288687369296e-06,
      "loss": 0.8191,
      "step": 4628
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2422370910644531,
      "learning_rate": 3.538299319207256e-06,
      "loss": 0.7642,
      "step": 4629
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2795745134353638,
      "learning_rate": 3.5362700329831436e-06,
      "loss": 0.7577,
      "step": 4630
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2632269859313965,
      "learning_rate": 3.5342410104301346e-06,
      "loss": 0.7969,
      "step": 4631
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1871572732925415,
      "learning_rate": 3.5322122519137354e-06,
      "loss": 0.7846,
      "step": 4632
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2868850231170654,
      "learning_rate": 3.530183757799397e-06,
      "loss": 0.817,
      "step": 4633
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2487765550613403,
      "learning_rate": 3.5281555284525227e-06,
      "loss": 0.8163,
      "step": 4634
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3424500226974487,
      "learning_rate": 3.5261275642384745e-06,
      "loss": 0.7845,
      "step": 4635
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2427719831466675,
      "learning_rate": 3.5240998655225593e-06,
      "loss": 0.7442,
      "step": 4636
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6532948017120361,
      "learning_rate": 3.5220724326700434e-06,
      "loss": 1.0326,
      "step": 4637
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.33663809299469,
      "learning_rate": 3.5200452660461394e-06,
      "loss": 0.8122,
      "step": 4638
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6176798343658447,
      "learning_rate": 3.5180183660160106e-06,
      "loss": 1.0373,
      "step": 4639
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3475275039672852,
      "learning_rate": 3.5159917329447814e-06,
      "loss": 0.8123,
      "step": 4640
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1944999694824219,
      "learning_rate": 3.5139653671975175e-06,
      "loss": 0.7197,
      "step": 4641
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6137688755989075,
      "learning_rate": 3.511939269139244e-06,
      "loss": 1.0547,
      "step": 4642
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2706997394561768,
      "learning_rate": 3.5099134391349347e-06,
      "loss": 0.8025,
      "step": 4643
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.245895266532898,
      "learning_rate": 3.507887877549514e-06,
      "loss": 0.7646,
      "step": 4644
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2834141254425049,
      "learning_rate": 3.505862584747861e-06,
      "loss": 0.8055,
      "step": 4645
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2491858005523682,
      "learning_rate": 3.5038375610948037e-06,
      "loss": 0.8333,
      "step": 4646
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3552519083023071,
      "learning_rate": 3.501812806955122e-06,
      "loss": 0.8083,
      "step": 4647
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6433742642402649,
      "learning_rate": 3.4997883226935503e-06,
      "loss": 1.0276,
      "step": 4648
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6427686214447021,
      "learning_rate": 3.497764108674768e-06,
      "loss": 1.0033,
      "step": 4649
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2479288578033447,
      "learning_rate": 3.4957401652634147e-06,
      "loss": 0.7706,
      "step": 4650
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3882038593292236,
      "learning_rate": 3.493716492824074e-06,
      "loss": 0.7845,
      "step": 4651
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6059147119522095,
      "learning_rate": 3.4916930917212803e-06,
      "loss": 1.0111,
      "step": 4652
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2102593183517456,
      "learning_rate": 3.4896699623195274e-06,
      "loss": 0.7206,
      "step": 4653
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6094962358474731,
      "learning_rate": 3.4876471049832495e-06,
      "loss": 0.9952,
      "step": 4654
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2160820960998535,
      "learning_rate": 3.4856245200768413e-06,
      "loss": 0.7475,
      "step": 4655
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3245878219604492,
      "learning_rate": 3.4836022079646414e-06,
      "loss": 0.8707,
      "step": 4656
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2243963479995728,
      "learning_rate": 3.481580169010941e-06,
      "loss": 0.7347,
      "step": 4657
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2627086639404297,
      "learning_rate": 3.479558403579987e-06,
      "loss": 0.7965,
      "step": 4658
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.234456181526184,
      "learning_rate": 3.477536912035967e-06,
      "loss": 0.8004,
      "step": 4659
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.181107759475708,
      "learning_rate": 3.4755156947430325e-06,
      "loss": 0.7618,
      "step": 4660
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3265681266784668,
      "learning_rate": 3.4734947520652735e-06,
      "loss": 0.7796,
      "step": 4661
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6271665096282959,
      "learning_rate": 3.471474084366735e-06,
      "loss": 1.0052,
      "step": 4662
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6069362163543701,
      "learning_rate": 3.4694536920114164e-06,
      "loss": 1.0108,
      "step": 4663
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2209573984146118,
      "learning_rate": 3.467433575363261e-06,
      "loss": 0.7334,
      "step": 4664
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2263624668121338,
      "learning_rate": 3.4654137347861676e-06,
      "loss": 0.7811,
      "step": 4665
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.263585090637207,
      "learning_rate": 3.463394170643981e-06,
      "loss": 0.7883,
      "step": 4666
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2822239398956299,
      "learning_rate": 3.461374883300499e-06,
      "loss": 0.7822,
      "step": 4667
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.298230528831482,
      "learning_rate": 3.459355873119471e-06,
      "loss": 0.7928,
      "step": 4668
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2318753004074097,
      "learning_rate": 3.4573371404645895e-06,
      "loss": 0.7441,
      "step": 4669
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.293459415435791,
      "learning_rate": 3.455318685699507e-06,
      "loss": 0.8564,
      "step": 4670
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3221005201339722,
      "learning_rate": 3.453300509187819e-06,
      "loss": 0.7645,
      "step": 4671
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2864007949829102,
      "learning_rate": 3.4512826112930696e-06,
      "loss": 0.796,
      "step": 4672
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6631089448928833,
      "learning_rate": 3.4492649923787603e-06,
      "loss": 1.0086,
      "step": 4673
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3492834568023682,
      "learning_rate": 3.447247652808336e-06,
      "loss": 0.8511,
      "step": 4674
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6142905354499817,
      "learning_rate": 3.44523059294519e-06,
      "loss": 1.0292,
      "step": 4675
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2185897827148438,
      "learning_rate": 3.4432138131526742e-06,
      "loss": 0.7419,
      "step": 4676
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3041914701461792,
      "learning_rate": 3.441197313794079e-06,
      "loss": 0.7755,
      "step": 4677
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.32646906375885,
      "learning_rate": 3.439181095232653e-06,
      "loss": 0.7386,
      "step": 4678
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2636703252792358,
      "learning_rate": 3.437165157831589e-06,
      "loss": 0.7834,
      "step": 4679
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2904990911483765,
      "learning_rate": 3.435149501954029e-06,
      "loss": 0.8018,
      "step": 4680
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2806706428527832,
      "learning_rate": 3.43313412796307e-06,
      "loss": 0.7935,
      "step": 4681
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2443174123764038,
      "learning_rate": 3.431119036221752e-06,
      "loss": 0.7918,
      "step": 4682
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.251995325088501,
      "learning_rate": 3.429104227093066e-06,
      "loss": 0.8016,
      "step": 4683
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7312560677528381,
      "learning_rate": 3.427089700939955e-06,
      "loss": 1.023,
      "step": 4684
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2608952522277832,
      "learning_rate": 3.4250754581253056e-06,
      "loss": 0.731,
      "step": 4685
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6510688662528992,
      "learning_rate": 3.4230614990119564e-06,
      "loss": 1.0393,
      "step": 4686
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2869815826416016,
      "learning_rate": 3.421047823962697e-06,
      "loss": 0.7472,
      "step": 4687
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2522881031036377,
      "learning_rate": 3.419034433340264e-06,
      "loss": 0.7618,
      "step": 4688
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.243262767791748,
      "learning_rate": 3.4170213275073406e-06,
      "loss": 0.7833,
      "step": 4689
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2710721492767334,
      "learning_rate": 3.415008506826558e-06,
      "loss": 0.8032,
      "step": 4690
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2624340057373047,
      "learning_rate": 3.4129959716605054e-06,
      "loss": 0.7539,
      "step": 4691
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2290140390396118,
      "learning_rate": 3.410983722371707e-06,
      "loss": 0.8255,
      "step": 4692
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3578119277954102,
      "learning_rate": 3.4089717593226467e-06,
      "loss": 0.7707,
      "step": 4693
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2722731828689575,
      "learning_rate": 3.406960082875751e-06,
      "loss": 0.761,
      "step": 4694
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2355607748031616,
      "learning_rate": 3.404948693393394e-06,
      "loss": 0.828,
      "step": 4695
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2949128150939941,
      "learning_rate": 3.4029375912379046e-06,
      "loss": 0.7772,
      "step": 4696
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.268575668334961,
      "learning_rate": 3.4009267767715505e-06,
      "loss": 0.7861,
      "step": 4697
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.327236533164978,
      "learning_rate": 3.3989162503565577e-06,
      "loss": 0.8662,
      "step": 4698
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7550416588783264,
      "learning_rate": 3.396906012355092e-06,
      "loss": 1.0422,
      "step": 4699
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2379175424575806,
      "learning_rate": 3.3948960631292705e-06,
      "loss": 0.7687,
      "step": 4700
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.249923586845398,
      "learning_rate": 3.392886403041161e-06,
      "loss": 0.7425,
      "step": 4701
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6712521910667419,
      "learning_rate": 3.3908770324527727e-06,
      "loss": 1.022,
      "step": 4702
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.317999005317688,
      "learning_rate": 3.388867951726069e-06,
      "loss": 0.8313,
      "step": 4703
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2391302585601807,
      "learning_rate": 3.3868591612229595e-06,
      "loss": 0.8229,
      "step": 4704
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2148592472076416,
      "learning_rate": 3.384850661305298e-06,
      "loss": 0.7938,
      "step": 4705
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2300347089767456,
      "learning_rate": 3.38284245233489e-06,
      "loss": 0.7417,
      "step": 4706
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.354499101638794,
      "learning_rate": 3.380834534673486e-06,
      "loss": 0.7726,
      "step": 4707
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.274299144744873,
      "learning_rate": 3.3788269086827856e-06,
      "loss": 0.8021,
      "step": 4708
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.27681565284729,
      "learning_rate": 3.376819574724437e-06,
      "loss": 0.7876,
      "step": 4709
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5012524127960205,
      "learning_rate": 3.374812533160031e-06,
      "loss": 0.7866,
      "step": 4710
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2362210750579834,
      "learning_rate": 3.3728057843511125e-06,
      "loss": 0.7527,
      "step": 4711
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.313598394393921,
      "learning_rate": 3.3707993286591683e-06,
      "loss": 0.8235,
      "step": 4712
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3341675996780396,
      "learning_rate": 3.3687931664456323e-06,
      "loss": 0.7568,
      "step": 4713
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3699601888656616,
      "learning_rate": 3.366787298071892e-06,
      "loss": 0.7757,
      "step": 4714
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8406403064727783,
      "learning_rate": 3.364781723899272e-06,
      "loss": 1.0285,
      "step": 4715
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2475576400756836,
      "learning_rate": 3.362776444289054e-06,
      "loss": 0.7655,
      "step": 4716
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3505759239196777,
      "learning_rate": 3.3607714596024602e-06,
      "loss": 0.7903,
      "step": 4717
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3574188947677612,
      "learning_rate": 3.3587667702006594e-06,
      "loss": 0.7409,
      "step": 4718
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3143432140350342,
      "learning_rate": 3.356762376444773e-06,
      "loss": 0.7284,
      "step": 4719
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6971341967582703,
      "learning_rate": 3.354758278695861e-06,
      "loss": 1.0217,
      "step": 4720
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3648649454116821,
      "learning_rate": 3.352754477314939e-06,
      "loss": 0.7661,
      "step": 4721
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6576515436172485,
      "learning_rate": 3.3507509726629627e-06,
      "loss": 1.0134,
      "step": 4722
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2983229160308838,
      "learning_rate": 3.3487477651008348e-06,
      "loss": 0.868,
      "step": 4723
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6501097679138184,
      "learning_rate": 3.3467448549894064e-06,
      "loss": 1.0244,
      "step": 4724
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.214971661567688,
      "learning_rate": 3.3447422426894773e-06,
      "loss": 0.7873,
      "step": 4725
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3333330154418945,
      "learning_rate": 3.3427399285617884e-06,
      "loss": 0.8304,
      "step": 4726
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7613240480422974,
      "learning_rate": 3.34073791296703e-06,
      "loss": 1.0392,
      "step": 4727
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6854422688484192,
      "learning_rate": 3.338736196265838e-06,
      "loss": 0.9928,
      "step": 4728
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2906620502471924,
      "learning_rate": 3.3367347788187953e-06,
      "loss": 0.8324,
      "step": 4729
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2731208801269531,
      "learning_rate": 3.3347336609864276e-06,
      "loss": 0.7564,
      "step": 4730
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3402388095855713,
      "learning_rate": 3.3327328431292133e-06,
      "loss": 0.7492,
      "step": 4731
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2056578397750854,
      "learning_rate": 3.3307323256075708e-06,
      "loss": 0.8008,
      "step": 4732
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2714917659759521,
      "learning_rate": 3.328732108781863e-06,
      "loss": 0.8075,
      "step": 4733
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.298366665840149,
      "learning_rate": 3.326732193012407e-06,
      "loss": 0.7672,
      "step": 4734
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7189306616783142,
      "learning_rate": 3.3247325786594583e-06,
      "loss": 1.037,
      "step": 4735
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3129512071609497,
      "learning_rate": 3.3227332660832173e-06,
      "loss": 0.8459,
      "step": 4736
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.27616548538208,
      "learning_rate": 3.3207342556438387e-06,
      "loss": 0.8318,
      "step": 4737
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6458540558815002,
      "learning_rate": 3.3187355477014114e-06,
      "loss": 1.0162,
      "step": 4738
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2476736307144165,
      "learning_rate": 3.3167371426159816e-06,
      "loss": 0.754,
      "step": 4739
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7275505065917969,
      "learning_rate": 3.314739040747531e-06,
      "loss": 0.7187,
      "step": 4740
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6331143975257874,
      "learning_rate": 3.31274124245599e-06,
      "loss": 1.0011,
      "step": 4741
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3372784852981567,
      "learning_rate": 3.310743748101237e-06,
      "loss": 0.7678,
      "step": 4742
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2201147079467773,
      "learning_rate": 3.3087465580430927e-06,
      "loss": 0.7467,
      "step": 4743
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6551862359046936,
      "learning_rate": 3.3067496726413236e-06,
      "loss": 1.0076,
      "step": 4744
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1948319673538208,
      "learning_rate": 3.3047530922556424e-06,
      "loss": 0.7947,
      "step": 4745
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2577118873596191,
      "learning_rate": 3.3027568172457047e-06,
      "loss": 0.7818,
      "step": 4746
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6149317622184753,
      "learning_rate": 3.300760847971114e-06,
      "loss": 1.019,
      "step": 4747
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2815366983413696,
      "learning_rate": 3.298765184791414e-06,
      "loss": 0.7853,
      "step": 4748
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6066629886627197,
      "learning_rate": 3.2967698280660997e-06,
      "loss": 1.0106,
      "step": 4749
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1692613363265991,
      "learning_rate": 3.2947747781546068e-06,
      "loss": 0.7513,
      "step": 4750
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2520930767059326,
      "learning_rate": 3.2927800354163143e-06,
      "loss": 0.7206,
      "step": 4751
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.740889072418213,
      "learning_rate": 3.290785600210551e-06,
      "loss": 0.7605,
      "step": 4752
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2375308275222778,
      "learning_rate": 3.2887914728965845e-06,
      "loss": 0.747,
      "step": 4753
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6244728565216064,
      "learning_rate": 3.286797653833633e-06,
      "loss": 1.0146,
      "step": 4754
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.249491810798645,
      "learning_rate": 3.2848041433808554e-06,
      "loss": 0.7645,
      "step": 4755
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2578468322753906,
      "learning_rate": 3.282810941897352e-06,
      "loss": 0.7676,
      "step": 4756
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3174774646759033,
      "learning_rate": 3.2808180497421755e-06,
      "loss": 0.7365,
      "step": 4757
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6431956887245178,
      "learning_rate": 3.2788254672743146e-06,
      "loss": 1.0144,
      "step": 4758
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2439258098602295,
      "learning_rate": 3.27683319485271e-06,
      "loss": 0.7667,
      "step": 4759
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.293735384941101,
      "learning_rate": 3.27484123283624e-06,
      "loss": 0.7865,
      "step": 4760
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2657724618911743,
      "learning_rate": 3.272849581583728e-06,
      "loss": 0.7426,
      "step": 4761
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6030277609825134,
      "learning_rate": 3.2708582414539473e-06,
      "loss": 0.9812,
      "step": 4762
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.265473484992981,
      "learning_rate": 3.2688672128056086e-06,
      "loss": 0.8278,
      "step": 4763
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.223652958869934,
      "learning_rate": 3.2668764959973677e-06,
      "loss": 0.7916,
      "step": 4764
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6037055850028992,
      "learning_rate": 3.2648860913878254e-06,
      "loss": 1.0232,
      "step": 4765
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6307728886604309,
      "learning_rate": 3.262895999335527e-06,
      "loss": 1.023,
      "step": 4766
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5950429439544678,
      "learning_rate": 3.2609062201989618e-06,
      "loss": 1.0341,
      "step": 4767
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2391053438186646,
      "learning_rate": 3.258916754336559e-06,
      "loss": 0.7419,
      "step": 4768
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5945692658424377,
      "learning_rate": 3.2569276021066953e-06,
      "loss": 1.0243,
      "step": 4769
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3206223249435425,
      "learning_rate": 3.25493876386769e-06,
      "loss": 0.7607,
      "step": 4770
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2426681518554688,
      "learning_rate": 3.252950239977802e-06,
      "loss": 0.7664,
      "step": 4771
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2959562540054321,
      "learning_rate": 3.250962030795243e-06,
      "loss": 0.8234,
      "step": 4772
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.295675277709961,
      "learning_rate": 3.248974136678158e-06,
      "loss": 0.7582,
      "step": 4773
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6030348539352417,
      "learning_rate": 3.246986557984637e-06,
      "loss": 0.9949,
      "step": 4774
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3661880493164062,
      "learning_rate": 3.2449992950727207e-06,
      "loss": 0.7522,
      "step": 4775
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.290562391281128,
      "learning_rate": 3.2430123483003816e-06,
      "loss": 0.7871,
      "step": 4776
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2656428813934326,
      "learning_rate": 3.2410257180255477e-06,
      "loss": 0.7521,
      "step": 4777
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2318320274353027,
      "learning_rate": 3.2390394046060796e-06,
      "loss": 0.8044,
      "step": 4778
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2348668575286865,
      "learning_rate": 3.237053408399784e-06,
      "loss": 0.8354,
      "step": 4779
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2417163848876953,
      "learning_rate": 3.235067729764414e-06,
      "loss": 0.781,
      "step": 4780
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2296810150146484,
      "learning_rate": 3.2330823690576596e-06,
      "loss": 0.7821,
      "step": 4781
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2596209049224854,
      "learning_rate": 3.2310973266371603e-06,
      "loss": 0.7729,
      "step": 4782
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2335846424102783,
      "learning_rate": 3.2291126028604926e-06,
      "loss": 0.7582,
      "step": 4783
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6232265830039978,
      "learning_rate": 3.227128198085176e-06,
      "loss": 1.0051,
      "step": 4784
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6311035752296448,
      "learning_rate": 3.225144112668676e-06,
      "loss": 0.9937,
      "step": 4785
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2627136707305908,
      "learning_rate": 3.2231603469683982e-06,
      "loss": 0.7628,
      "step": 4786
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8226425647735596,
      "learning_rate": 3.221176901341693e-06,
      "loss": 0.7386,
      "step": 4787
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2586475610733032,
      "learning_rate": 3.219193776145849e-06,
      "loss": 0.7485,
      "step": 4788
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2202270030975342,
      "learning_rate": 3.2172109717380973e-06,
      "loss": 0.8085,
      "step": 4789
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2063807249069214,
      "learning_rate": 3.215228488475618e-06,
      "loss": 0.7297,
      "step": 4790
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3174899816513062,
      "learning_rate": 3.213246326715526e-06,
      "loss": 0.7671,
      "step": 4791
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2254598140716553,
      "learning_rate": 3.2112644868148794e-06,
      "loss": 0.8027,
      "step": 4792
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3102920055389404,
      "learning_rate": 3.209282969130684e-06,
      "loss": 0.7871,
      "step": 4793
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.269850254058838,
      "learning_rate": 3.2073017740198787e-06,
      "loss": 0.7733,
      "step": 4794
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.272783875465393,
      "learning_rate": 3.205320901839353e-06,
      "loss": 0.7885,
      "step": 4795
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2396421432495117,
      "learning_rate": 3.203340352945932e-06,
      "loss": 0.779,
      "step": 4796
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2233235836029053,
      "learning_rate": 3.2013601276963825e-06,
      "loss": 0.7376,
      "step": 4797
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2815090417861938,
      "learning_rate": 3.1993802264474205e-06,
      "loss": 0.7895,
      "step": 4798
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2681245803833008,
      "learning_rate": 3.1974006495556936e-06,
      "loss": 0.8124,
      "step": 4799
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.172242522239685,
      "learning_rate": 3.195421397377799e-06,
      "loss": 0.745,
      "step": 4800
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.280547857284546,
      "learning_rate": 3.193442470270272e-06,
      "loss": 0.8051,
      "step": 4801
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6764117479324341,
      "learning_rate": 3.191463868589586e-06,
      "loss": 1.01,
      "step": 4802
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6495941281318665,
      "learning_rate": 3.189485592692163e-06,
      "loss": 1.0337,
      "step": 4803
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5928541421890259,
      "learning_rate": 3.187507642934361e-06,
      "loss": 1.0183,
      "step": 4804
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.612328290939331,
      "learning_rate": 3.1855300196724815e-06,
      "loss": 1.0294,
      "step": 4805
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.316052794456482,
      "learning_rate": 3.1835527232627657e-06,
      "loss": 0.7765,
      "step": 4806
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6313083171844482,
      "learning_rate": 3.181575754061397e-06,
      "loss": 1.0057,
      "step": 4807
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6213502883911133,
      "learning_rate": 3.1795991124244996e-06,
      "loss": 1.0196,
      "step": 4808
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3302228450775146,
      "learning_rate": 3.1776227987081375e-06,
      "loss": 0.7686,
      "step": 4809
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.281894326210022,
      "learning_rate": 3.1756468132683206e-06,
      "loss": 0.7494,
      "step": 4810
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3071699142456055,
      "learning_rate": 3.1736711564609924e-06,
      "loss": 0.8063,
      "step": 4811
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6335930228233337,
      "learning_rate": 3.1716958286420397e-06,
      "loss": 1.0241,
      "step": 4812
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2551566362380981,
      "learning_rate": 3.1697208301672957e-06,
      "loss": 0.6921,
      "step": 4813
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2531424760818481,
      "learning_rate": 3.167746161392524e-06,
      "loss": 0.7591,
      "step": 4814
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2898657321929932,
      "learning_rate": 3.16577182267344e-06,
      "loss": 0.7586,
      "step": 4815
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1987560987472534,
      "learning_rate": 3.1637978143656912e-06,
      "loss": 0.7359,
      "step": 4816
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.26264488697052,
      "learning_rate": 3.161824136824867e-06,
      "loss": 0.7741,
      "step": 4817
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2091089487075806,
      "learning_rate": 3.1598507904065013e-06,
      "loss": 0.8009,
      "step": 4818
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.417041540145874,
      "learning_rate": 3.157877775466065e-06,
      "loss": 0.8239,
      "step": 4819
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3592108488082886,
      "learning_rate": 3.155905092358968e-06,
      "loss": 0.8392,
      "step": 4820
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.64969402551651,
      "learning_rate": 3.153932741440565e-06,
      "loss": 1.0091,
      "step": 4821
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6447173357009888,
      "learning_rate": 3.1519607230661465e-06,
      "loss": 1.024,
      "step": 4822
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2374169826507568,
      "learning_rate": 3.1499890375909468e-06,
      "loss": 0.7675,
      "step": 4823
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2331821918487549,
      "learning_rate": 3.148017685370137e-06,
      "loss": 0.7832,
      "step": 4824
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2545559406280518,
      "learning_rate": 3.14604666675883e-06,
      "loss": 0.7209,
      "step": 4825
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2599517107009888,
      "learning_rate": 3.1440759821120764e-06,
      "loss": 0.7915,
      "step": 4826
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5936965942382812,
      "learning_rate": 3.1421056317848698e-06,
      "loss": 1.0456,
      "step": 4827
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6024497747421265,
      "learning_rate": 3.1401356161321433e-06,
      "loss": 1.0112,
      "step": 4828
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6060802936553955,
      "learning_rate": 3.1381659355087666e-06,
      "loss": 1.0219,
      "step": 4829
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6280229091644287,
      "learning_rate": 3.1361965902695503e-06,
      "loss": 1.0246,
      "step": 4830
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2597205638885498,
      "learning_rate": 3.134227580769248e-06,
      "loss": 0.7406,
      "step": 4831
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2247556447982788,
      "learning_rate": 3.1322589073625464e-06,
      "loss": 0.7363,
      "step": 4832
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1695470809936523,
      "learning_rate": 3.1302905704040794e-06,
      "loss": 0.7526,
      "step": 4833
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.275038719177246,
      "learning_rate": 3.1283225702484145e-06,
      "loss": 0.7902,
      "step": 4834
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4335317611694336,
      "learning_rate": 3.1263549072500575e-06,
      "loss": 0.7171,
      "step": 4835
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2713801860809326,
      "learning_rate": 3.1243875817634607e-06,
      "loss": 0.7369,
      "step": 4836
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2902376651763916,
      "learning_rate": 3.1224205941430074e-06,
      "loss": 0.7338,
      "step": 4837
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2578723430633545,
      "learning_rate": 3.1204539447430273e-06,
      "loss": 0.7998,
      "step": 4838
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.218944787979126,
      "learning_rate": 3.1184876339177837e-06,
      "loss": 0.7389,
      "step": 4839
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3035860061645508,
      "learning_rate": 3.116521662021479e-06,
      "loss": 0.7528,
      "step": 4840
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6847445964813232,
      "learning_rate": 3.1145560294082604e-06,
      "loss": 1.0234,
      "step": 4841
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1652636528015137,
      "learning_rate": 3.1125907364322074e-06,
      "loss": 0.7268,
      "step": 4842
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6900144219398499,
      "learning_rate": 3.1106257834473397e-06,
      "loss": 0.9998,
      "step": 4843
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2241472005844116,
      "learning_rate": 3.108661170807621e-06,
      "loss": 0.7242,
      "step": 4844
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2729310989379883,
      "learning_rate": 3.1066968988669456e-06,
      "loss": 0.8036,
      "step": 4845
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3156278133392334,
      "learning_rate": 3.1047329679791522e-06,
      "loss": 0.73,
      "step": 4846
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1574382781982422,
      "learning_rate": 3.1027693784980163e-06,
      "loss": 0.8002,
      "step": 4847
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3125892877578735,
      "learning_rate": 3.100806130777252e-06,
      "loss": 0.7921,
      "step": 4848
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2234405279159546,
      "learning_rate": 3.0988432251705127e-06,
      "loss": 0.7454,
      "step": 4849
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.259756326675415,
      "learning_rate": 3.096880662031385e-06,
      "loss": 0.7356,
      "step": 4850
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.196451187133789,
      "learning_rate": 3.094918441713404e-06,
      "loss": 0.7299,
      "step": 4851
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6715918779373169,
      "learning_rate": 3.092956564570033e-06,
      "loss": 1.0302,
      "step": 4852
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.272396206855774,
      "learning_rate": 3.090995030954678e-06,
      "loss": 0.7904,
      "step": 4853
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.238929271697998,
      "learning_rate": 3.0890338412206845e-06,
      "loss": 0.781,
      "step": 4854
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.277589201927185,
      "learning_rate": 3.087072995721332e-06,
      "loss": 0.7573,
      "step": 4855
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2592519521713257,
      "learning_rate": 3.0851124948098432e-06,
      "loss": 0.7541,
      "step": 4856
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2968788146972656,
      "learning_rate": 3.0831523388393737e-06,
      "loss": 0.7795,
      "step": 4857
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6234224438667297,
      "learning_rate": 3.0811925281630172e-06,
      "loss": 1.0161,
      "step": 4858
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6129544973373413,
      "learning_rate": 3.0792330631338107e-06,
      "loss": 1.0059,
      "step": 4859
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2649368047714233,
      "learning_rate": 3.077273944104723e-06,
      "loss": 0.8511,
      "step": 4860
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2408955097198486,
      "learning_rate": 3.0753151714286644e-06,
      "loss": 0.7657,
      "step": 4861
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2875149250030518,
      "learning_rate": 3.0733567454584805e-06,
      "loss": 0.8083,
      "step": 4862
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4346778392791748,
      "learning_rate": 3.071398666546953e-06,
      "loss": 0.7624,
      "step": 4863
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2335758209228516,
      "learning_rate": 3.069440935046808e-06,
      "loss": 0.8243,
      "step": 4864
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2847357988357544,
      "learning_rate": 3.0674835513106993e-06,
      "loss": 0.6894,
      "step": 4865
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.627963662147522,
      "learning_rate": 3.0655265156912274e-06,
      "loss": 1.0084,
      "step": 4866
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5933164954185486,
      "learning_rate": 3.063569828540922e-06,
      "loss": 1.0101,
      "step": 4867
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5908470749855042,
      "learning_rate": 3.061613490212256e-06,
      "loss": 0.998,
      "step": 4868
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2855405807495117,
      "learning_rate": 3.0596575010576368e-06,
      "loss": 0.8506,
      "step": 4869
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2148734331130981,
      "learning_rate": 3.0577018614294074e-06,
      "loss": 0.7484,
      "step": 4870
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2503341436386108,
      "learning_rate": 3.055746571679854e-06,
      "loss": 0.8166,
      "step": 4871
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6032597422599792,
      "learning_rate": 3.0537916321611925e-06,
      "loss": 1.0273,
      "step": 4872
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1805400848388672,
      "learning_rate": 3.051837043225577e-06,
      "loss": 0.7318,
      "step": 4873
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2482097148895264,
      "learning_rate": 3.049882805225105e-06,
      "loss": 0.8155,
      "step": 4874
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2209006547927856,
      "learning_rate": 3.0479289185118016e-06,
      "loss": 0.763,
      "step": 4875
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.273591160774231,
      "learning_rate": 3.0459753834376364e-06,
      "loss": 0.7465,
      "step": 4876
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2709732055664062,
      "learning_rate": 3.0440222003545113e-06,
      "loss": 0.7657,
      "step": 4877
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.253729224205017,
      "learning_rate": 3.042069369614262e-06,
      "loss": 0.7546,
      "step": 4878
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2644115686416626,
      "learning_rate": 3.0401168915686707e-06,
      "loss": 0.783,
      "step": 4879
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3170890808105469,
      "learning_rate": 3.038164766569447e-06,
      "loss": 0.7797,
      "step": 4880
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6452072262763977,
      "learning_rate": 3.0362129949682368e-06,
      "loss": 1.0551,
      "step": 4881
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.37154221534729,
      "learning_rate": 3.0342615771166303e-06,
      "loss": 0.7799,
      "step": 4882
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2872710227966309,
      "learning_rate": 3.0323105133661457e-06,
      "loss": 0.7958,
      "step": 4883
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.636767566204071,
      "learning_rate": 3.0303598040682413e-06,
      "loss": 0.9976,
      "step": 4884
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6229473352432251,
      "learning_rate": 3.028409449574312e-06,
      "loss": 1.0326,
      "step": 4885
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6247671842575073,
      "learning_rate": 3.026459450235687e-06,
      "loss": 1.0333,
      "step": 4886
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2114536762237549,
      "learning_rate": 3.0245098064036317e-06,
      "loss": 0.7275,
      "step": 4887
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2649861574172974,
      "learning_rate": 3.022560518429348e-06,
      "loss": 0.8372,
      "step": 4888
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2195725440979004,
      "learning_rate": 3.0206115866639762e-06,
      "loss": 0.7533,
      "step": 4889
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.622886061668396,
      "learning_rate": 3.018663011458588e-06,
      "loss": 1.0083,
      "step": 4890
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3596872091293335,
      "learning_rate": 3.0167147931641906e-06,
      "loss": 0.7662,
      "step": 4891
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2425287961959839,
      "learning_rate": 3.014766932131733e-06,
      "loss": 0.7557,
      "step": 4892
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6227298378944397,
      "learning_rate": 3.0128194287120925e-06,
      "loss": 1.0124,
      "step": 4893
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.605133056640625,
      "learning_rate": 3.0108722832560887e-06,
      "loss": 1.007,
      "step": 4894
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2534074783325195,
      "learning_rate": 3.0089254961144725e-06,
      "loss": 0.7525,
      "step": 4895
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5984174013137817,
      "learning_rate": 3.0069790676379286e-06,
      "loss": 1.0028,
      "step": 4896
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2962878942489624,
      "learning_rate": 3.0050329981770833e-06,
      "loss": 0.7941,
      "step": 4897
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2811802625656128,
      "learning_rate": 3.0030872880824914e-06,
      "loss": 0.7935,
      "step": 4898
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6155096888542175,
      "learning_rate": 3.0011419377046496e-06,
      "loss": 1.0044,
      "step": 4899
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3345510959625244,
      "learning_rate": 2.999196947393985e-06,
      "loss": 0.7489,
      "step": 4900
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2521531581878662,
      "learning_rate": 2.9972523175008584e-06,
      "loss": 0.7832,
      "step": 4901
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6172167658805847,
      "learning_rate": 2.9953080483755735e-06,
      "loss": 0.9842,
      "step": 4902
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.235878825187683,
      "learning_rate": 2.99336414036836e-06,
      "loss": 0.7834,
      "step": 4903
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6316982507705688,
      "learning_rate": 2.9914205938293883e-06,
      "loss": 1.0106,
      "step": 4904
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2938560247421265,
      "learning_rate": 2.989477409108763e-06,
      "loss": 0.7303,
      "step": 4905
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2423771619796753,
      "learning_rate": 2.9875345865565198e-06,
      "loss": 0.8462,
      "step": 4906
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2339057922363281,
      "learning_rate": 2.9855921265226353e-06,
      "loss": 0.8102,
      "step": 4907
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5877212882041931,
      "learning_rate": 2.9836500293570123e-06,
      "loss": 0.9819,
      "step": 4908
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6117975115776062,
      "learning_rate": 2.981708295409498e-06,
      "loss": 1.0008,
      "step": 4909
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6132887601852417,
      "learning_rate": 2.9797669250298677e-06,
      "loss": 0.9983,
      "step": 4910
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2381751537322998,
      "learning_rate": 2.9778259185678317e-06,
      "loss": 0.7461,
      "step": 4911
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6027167439460754,
      "learning_rate": 2.9758852763730385e-06,
      "loss": 0.9986,
      "step": 4912
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.349303960800171,
      "learning_rate": 2.9739449987950675e-06,
      "loss": 0.7626,
      "step": 4913
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.268805980682373,
      "learning_rate": 2.9720050861834315e-06,
      "loss": 0.814,
      "step": 4914
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6307265758514404,
      "learning_rate": 2.970065538887583e-06,
      "loss": 1.007,
      "step": 4915
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3440736532211304,
      "learning_rate": 2.9681263572569007e-06,
      "loss": 0.8386,
      "step": 4916
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.638056755065918,
      "learning_rate": 2.9661875416407064e-06,
      "loss": 1.0066,
      "step": 4917
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6421040892601013,
      "learning_rate": 2.9642490923882495e-06,
      "loss": 1.0391,
      "step": 4918
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.256954312324524,
      "learning_rate": 2.962311009848714e-06,
      "loss": 0.7686,
      "step": 4919
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5821592211723328,
      "learning_rate": 2.9603732943712215e-06,
      "loss": 1.0084,
      "step": 4920
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2585747241973877,
      "learning_rate": 2.958435946304823e-06,
      "loss": 0.8069,
      "step": 4921
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1702865362167358,
      "learning_rate": 2.9564989659985078e-06,
      "loss": 0.7875,
      "step": 4922
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.243828535079956,
      "learning_rate": 2.954562353801196e-06,
      "loss": 0.8021,
      "step": 4923
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.639656662940979,
      "learning_rate": 2.95262611006174e-06,
      "loss": 1.0087,
      "step": 4924
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2434536218643188,
      "learning_rate": 2.95069023512893e-06,
      "loss": 0.763,
      "step": 4925
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.413783073425293,
      "learning_rate": 2.9487547293514874e-06,
      "loss": 0.794,
      "step": 4926
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5118893384933472,
      "learning_rate": 2.946819593078067e-06,
      "loss": 0.7917,
      "step": 4927
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.359877347946167,
      "learning_rate": 2.9448848266572562e-06,
      "loss": 0.7478,
      "step": 4928
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6037558913230896,
      "learning_rate": 2.942950430437579e-06,
      "loss": 1.0085,
      "step": 4929
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3157676458358765,
      "learning_rate": 2.9410164047674894e-06,
      "loss": 0.7847,
      "step": 4930
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3028948307037354,
      "learning_rate": 2.9390827499953746e-06,
      "loss": 0.8198,
      "step": 4931
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2958580255508423,
      "learning_rate": 2.93714946646956e-06,
      "loss": 0.7582,
      "step": 4932
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2444956302642822,
      "learning_rate": 2.9352165545382983e-06,
      "loss": 0.723,
      "step": 4933
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4484045505523682,
      "learning_rate": 2.933284014549775e-06,
      "loss": 0.7549,
      "step": 4934
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.272909164428711,
      "learning_rate": 2.931351846852115e-06,
      "loss": 0.6473,
      "step": 4935
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2641793489456177,
      "learning_rate": 2.929420051793371e-06,
      "loss": 0.7917,
      "step": 4936
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2786200046539307,
      "learning_rate": 2.9274886297215273e-06,
      "loss": 0.7668,
      "step": 4937
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3164527416229248,
      "learning_rate": 2.9255575809845067e-06,
      "loss": 0.814,
      "step": 4938
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3485321998596191,
      "learning_rate": 2.923626905930158e-06,
      "loss": 0.7814,
      "step": 4939
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8858088254928589,
      "learning_rate": 2.92169660490627e-06,
      "loss": 0.7741,
      "step": 4940
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.257136344909668,
      "learning_rate": 2.9197666782605588e-06,
      "loss": 0.8277,
      "step": 4941
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6280193328857422,
      "learning_rate": 2.917837126340671e-06,
      "loss": 1.0502,
      "step": 4942
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6199179291725159,
      "learning_rate": 2.9159079494941957e-06,
      "loss": 1.0318,
      "step": 4943
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6328436732292175,
      "learning_rate": 2.913979148068642e-06,
      "loss": 1.0101,
      "step": 4944
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5872544050216675,
      "learning_rate": 2.9120507224114613e-06,
      "loss": 1.0359,
      "step": 4945
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3724433183670044,
      "learning_rate": 2.910122672870032e-06,
      "loss": 0.7756,
      "step": 4946
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.216399908065796,
      "learning_rate": 2.9081949997916642e-06,
      "loss": 0.8075,
      "step": 4947
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6153057217597961,
      "learning_rate": 2.906267703523606e-06,
      "loss": 1.0022,
      "step": 4948
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2518163919448853,
      "learning_rate": 2.904340784413029e-06,
      "loss": 0.7054,
      "step": 4949
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.229977011680603,
      "learning_rate": 2.9024142428070468e-06,
      "loss": 0.7902,
      "step": 4950
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3066202402114868,
      "learning_rate": 2.9004880790526957e-06,
      "loss": 0.7919,
      "step": 4951
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5301135778427124,
      "learning_rate": 2.8985622934969483e-06,
      "loss": 0.7585,
      "step": 4952
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2722079753875732,
      "learning_rate": 2.8966368864867116e-06,
      "loss": 0.8501,
      "step": 4953
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6334863901138306,
      "learning_rate": 2.8947118583688193e-06,
      "loss": 1.0289,
      "step": 4954
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2280725240707397,
      "learning_rate": 2.8927872094900384e-06,
      "loss": 0.727,
      "step": 4955
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6530073285102844,
      "learning_rate": 2.8908629401970706e-06,
      "loss": 1.0089,
      "step": 4956
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.28821861743927,
      "learning_rate": 2.8889390508365462e-06,
      "loss": 0.7638,
      "step": 4957
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2250868082046509,
      "learning_rate": 2.8870155417550274e-06,
      "loss": 0.7973,
      "step": 4958
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3435672521591187,
      "learning_rate": 2.8850924132990056e-06,
      "loss": 0.8508,
      "step": 4959
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.602587878704071,
      "learning_rate": 2.883169665814911e-06,
      "loss": 0.9924,
      "step": 4960
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2109886407852173,
      "learning_rate": 2.8812472996490993e-06,
      "loss": 0.7757,
      "step": 4961
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.322722315788269,
      "learning_rate": 2.8793253151478553e-06,
      "loss": 0.7692,
      "step": 4962
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2047282457351685,
      "learning_rate": 2.8774037126574023e-06,
      "loss": 0.7504,
      "step": 4963
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6342872381210327,
      "learning_rate": 2.8754824925238876e-06,
      "loss": 1.0271,
      "step": 4964
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6023178696632385,
      "learning_rate": 2.8735616550933975e-06,
      "loss": 1.0284,
      "step": 4965
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2320353984832764,
      "learning_rate": 2.8716412007119427e-06,
      "loss": 0.7759,
      "step": 4966
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2357698678970337,
      "learning_rate": 2.869721129725464e-06,
      "loss": 0.8148,
      "step": 4967
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6126973032951355,
      "learning_rate": 2.8678014424798406e-06,
      "loss": 1.0233,
      "step": 4968
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2585217952728271,
      "learning_rate": 2.865882139320877e-06,
      "loss": 0.7824,
      "step": 4969
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2727885246276855,
      "learning_rate": 2.863963220594306e-06,
      "loss": 0.7833,
      "step": 4970
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2874904870986938,
      "learning_rate": 2.8620446866457995e-06,
      "loss": 0.7643,
      "step": 4971
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3317424058914185,
      "learning_rate": 2.8601265378209523e-06,
      "loss": 0.7906,
      "step": 4972
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3123236894607544,
      "learning_rate": 2.8582087744652962e-06,
      "loss": 0.784,
      "step": 4973
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2636865377426147,
      "learning_rate": 2.8562913969242876e-06,
      "loss": 0.8107,
      "step": 4974
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3026585578918457,
      "learning_rate": 2.8543744055433153e-06,
      "loss": 0.8057,
      "step": 4975
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6487959623336792,
      "learning_rate": 2.852457800667703e-06,
      "loss": 1.0233,
      "step": 4976
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6532598733901978,
      "learning_rate": 2.8505415826426963e-06,
      "loss": 1.0146,
      "step": 4977
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5958110094070435,
      "learning_rate": 2.8486257518134795e-06,
      "loss": 1.0163,
      "step": 4978
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2252730131149292,
      "learning_rate": 2.8467103085251633e-06,
      "loss": 0.7532,
      "step": 4979
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.312447190284729,
      "learning_rate": 2.844795253122786e-06,
      "loss": 0.7509,
      "step": 4980
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.237518072128296,
      "learning_rate": 2.842880585951323e-06,
      "loss": 0.7356,
      "step": 4981
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2762742042541504,
      "learning_rate": 2.8409663073556715e-06,
      "loss": 0.788,
      "step": 4982
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2763372659683228,
      "learning_rate": 2.8390524176806665e-06,
      "loss": 0.7747,
      "step": 4983
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2956126928329468,
      "learning_rate": 2.8371389172710674e-06,
      "loss": 0.7557,
      "step": 4984
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.322924256324768,
      "learning_rate": 2.8352258064715638e-06,
      "loss": 0.7105,
      "step": 4985
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2826449871063232,
      "learning_rate": 2.83331308562678e-06,
      "loss": 0.7985,
      "step": 4986
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7246423363685608,
      "learning_rate": 2.8314007550812635e-06,
      "loss": 1.0087,
      "step": 4987
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2426234483718872,
      "learning_rate": 2.829488815179498e-06,
      "loss": 0.738,
      "step": 4988
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2103586196899414,
      "learning_rate": 2.827577266265892e-06,
      "loss": 0.7719,
      "step": 4989
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3002780675888062,
      "learning_rate": 2.825666108684783e-06,
      "loss": 0.8044,
      "step": 4990
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2997679710388184,
      "learning_rate": 2.8237553427804443e-06,
      "loss": 0.7759,
      "step": 4991
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2843266725540161,
      "learning_rate": 2.821844968897072e-06,
      "loss": 0.783,
      "step": 4992
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3407162427902222,
      "learning_rate": 2.819934987378793e-06,
      "loss": 0.8801,
      "step": 4993
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.24105703830719,
      "learning_rate": 2.818025398569667e-06,
      "loss": 0.8032,
      "step": 4994
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6307279467582703,
      "learning_rate": 2.8161162028136803e-06,
      "loss": 1.0279,
      "step": 4995
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2878358364105225,
      "learning_rate": 2.8142074004547447e-06,
      "loss": 0.7978,
      "step": 4996
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5962765216827393,
      "learning_rate": 2.8122989918367106e-06,
      "loss": 1.0237,
      "step": 4997
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2547324895858765,
      "learning_rate": 2.81039097730335e-06,
      "loss": 0.8023,
      "step": 4998
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3246890306472778,
      "learning_rate": 2.808483357198365e-06,
      "loss": 0.7796,
      "step": 4999
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1913502216339111,
      "learning_rate": 2.806576131865385e-06,
      "loss": 0.7187,
      "step": 5000
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3725844621658325,
      "learning_rate": 2.8046693016479763e-06,
      "loss": 0.7593,
      "step": 5001
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.30801260471344,
      "learning_rate": 2.8027628668896263e-06,
      "loss": 0.8384,
      "step": 5002
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3326430320739746,
      "learning_rate": 2.80085682793375e-06,
      "loss": 0.794,
      "step": 5003
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3006480932235718,
      "learning_rate": 2.7989511851237e-06,
      "loss": 0.7881,
      "step": 5004
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6253156661987305,
      "learning_rate": 2.797045938802747e-06,
      "loss": 1.0346,
      "step": 5005
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6609758734703064,
      "learning_rate": 2.7951410893140997e-06,
      "loss": 1.0203,
      "step": 5006
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.21225905418396,
      "learning_rate": 2.793236637000889e-06,
      "loss": 0.7837,
      "step": 5007
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.221370816230774,
      "learning_rate": 2.791332582206174e-06,
      "loss": 0.7517,
      "step": 5008
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3572369813919067,
      "learning_rate": 2.789428925272948e-06,
      "loss": 0.8298,
      "step": 5009
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2587190866470337,
      "learning_rate": 2.7875256665441256e-06,
      "loss": 0.7959,
      "step": 5010
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6104533076286316,
      "learning_rate": 2.7856228063625568e-06,
      "loss": 1.0147,
      "step": 5011
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.233249306678772,
      "learning_rate": 2.783720345071014e-06,
      "loss": 0.7717,
      "step": 5012
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1801992654800415,
      "learning_rate": 2.7818182830121983e-06,
      "loss": 0.764,
      "step": 5013
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6511766314506531,
      "learning_rate": 2.779916620528744e-06,
      "loss": 1.0281,
      "step": 5014
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6090524792671204,
      "learning_rate": 2.7780153579632052e-06,
      "loss": 1.0302,
      "step": 5015
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2393206357955933,
      "learning_rate": 2.776114495658073e-06,
      "loss": 0.7587,
      "step": 5016
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2718124389648438,
      "learning_rate": 2.7742140339557607e-06,
      "loss": 0.7483,
      "step": 5017
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2098186016082764,
      "learning_rate": 2.7723139731986084e-06,
      "loss": 0.7561,
      "step": 5018
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5847419500350952,
      "learning_rate": 2.7704143137288887e-06,
      "loss": 1.0318,
      "step": 5019
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5926461219787598,
      "learning_rate": 2.768515055888798e-06,
      "loss": 1.0067,
      "step": 5020
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3063507080078125,
      "learning_rate": 2.7666162000204645e-06,
      "loss": 0.7606,
      "step": 5021
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2483673095703125,
      "learning_rate": 2.764717746465939e-06,
      "loss": 0.7639,
      "step": 5022
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2769063711166382,
      "learning_rate": 2.7628196955672004e-06,
      "loss": 0.8193,
      "step": 5023
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5884352326393127,
      "learning_rate": 2.760922047666162e-06,
      "loss": 1.0071,
      "step": 5024
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2563320398330688,
      "learning_rate": 2.759024803104656e-06,
      "loss": 0.7575,
      "step": 5025
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2708206176757812,
      "learning_rate": 2.757127962224444e-06,
      "loss": 0.7918,
      "step": 5026
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2277626991271973,
      "learning_rate": 2.755231525367219e-06,
      "loss": 0.8109,
      "step": 5027
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2341209650039673,
      "learning_rate": 2.7533354928745958e-06,
      "loss": 0.7141,
      "step": 5028
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.207831621170044,
      "learning_rate": 2.751439865088123e-06,
      "loss": 0.7667,
      "step": 5029
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3204107284545898,
      "learning_rate": 2.749544642349269e-06,
      "loss": 0.7975,
      "step": 5030
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.359687328338623,
      "learning_rate": 2.747649824999431e-06,
      "loss": 0.7814,
      "step": 5031
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6205031871795654,
      "learning_rate": 2.745755413379939e-06,
      "loss": 1.0069,
      "step": 5032
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2528393268585205,
      "learning_rate": 2.7438614078320413e-06,
      "loss": 0.7836,
      "step": 5033
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2499803304672241,
      "learning_rate": 2.7419678086969216e-06,
      "loss": 0.754,
      "step": 5034
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3222472667694092,
      "learning_rate": 2.7400746163156837e-06,
      "loss": 0.7734,
      "step": 5035
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3024039268493652,
      "learning_rate": 2.7381818310293604e-06,
      "loss": 0.7529,
      "step": 5036
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.619551956653595,
      "learning_rate": 2.7362894531789104e-06,
      "loss": 1.0003,
      "step": 5037
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6086859703063965,
      "learning_rate": 2.734397483105222e-06,
      "loss": 1.0242,
      "step": 5038
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6077716946601868,
      "learning_rate": 2.7325059211491077e-06,
      "loss": 1.0142,
      "step": 5039
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2889853715896606,
      "learning_rate": 2.730614767651306e-06,
      "loss": 0.7712,
      "step": 5040
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.489991307258606,
      "learning_rate": 2.7287240229524813e-06,
      "loss": 0.7869,
      "step": 5041
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6328867673873901,
      "learning_rate": 2.7268336873932285e-06,
      "loss": 1.0093,
      "step": 5042
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2326183319091797,
      "learning_rate": 2.7249437613140622e-06,
      "loss": 0.784,
      "step": 5043
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2679400444030762,
      "learning_rate": 2.7230542450554313e-06,
      "loss": 0.8054,
      "step": 5044
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6083127856254578,
      "learning_rate": 2.7211651389577044e-06,
      "loss": 1.0236,
      "step": 5045
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2662655115127563,
      "learning_rate": 2.7192764433611763e-06,
      "loss": 0.8273,
      "step": 5046
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6105378270149231,
      "learning_rate": 2.7173881586060746e-06,
      "loss": 1.0153,
      "step": 5047
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3610631227493286,
      "learning_rate": 2.7155002850325428e-06,
      "loss": 0.8003,
      "step": 5048
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2091249227523804,
      "learning_rate": 2.7136128229806607e-06,
      "loss": 0.7834,
      "step": 5049
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2643520832061768,
      "learning_rate": 2.7117257727904267e-06,
      "loss": 0.7838,
      "step": 5050
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.604484498500824,
      "learning_rate": 2.7098391348017656e-06,
      "loss": 1.0163,
      "step": 5051
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.157875895500183,
      "learning_rate": 2.707952909354533e-06,
      "loss": 0.8274,
      "step": 5052
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2457526922225952,
      "learning_rate": 2.7060670967885062e-06,
      "loss": 0.7514,
      "step": 5053
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2992885112762451,
      "learning_rate": 2.7041816974433843e-06,
      "loss": 0.8098,
      "step": 5054
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3175652027130127,
      "learning_rate": 2.7022967116588036e-06,
      "loss": 0.814,
      "step": 5055
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3185100555419922,
      "learning_rate": 2.7004121397743123e-06,
      "loss": 0.7904,
      "step": 5056
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2861900329589844,
      "learning_rate": 2.6985279821293954e-06,
      "loss": 0.7892,
      "step": 5057
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2914535999298096,
      "learning_rate": 2.6966442390634557e-06,
      "loss": 0.7618,
      "step": 5058
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2943707704544067,
      "learning_rate": 2.694760910915823e-06,
      "loss": 0.8164,
      "step": 5059
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2443841695785522,
      "learning_rate": 2.6928779980257567e-06,
      "loss": 0.7661,
      "step": 5060
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2648952007293701,
      "learning_rate": 2.690995500732434e-06,
      "loss": 0.7509,
      "step": 5061
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6373929381370544,
      "learning_rate": 2.6891134193749657e-06,
      "loss": 0.9981,
      "step": 5062
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4095370769500732,
      "learning_rate": 2.687231754292381e-06,
      "loss": 0.7661,
      "step": 5063
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.304659128189087,
      "learning_rate": 2.685350505823634e-06,
      "loss": 0.7924,
      "step": 5064
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2451835870742798,
      "learning_rate": 2.68346967430761e-06,
      "loss": 0.765,
      "step": 5065
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2675873041152954,
      "learning_rate": 2.6815892600831124e-06,
      "loss": 0.7368,
      "step": 5066
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2307132482528687,
      "learning_rate": 2.6797092634888754e-06,
      "loss": 0.8138,
      "step": 5067
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2582569122314453,
      "learning_rate": 2.677829684863553e-06,
      "loss": 0.7807,
      "step": 5068
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2868443727493286,
      "learning_rate": 2.675950524545724e-06,
      "loss": 0.8043,
      "step": 5069
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3141999244689941,
      "learning_rate": 2.6740717828738973e-06,
      "loss": 0.7563,
      "step": 5070
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6155031323432922,
      "learning_rate": 2.6721934601864997e-06,
      "loss": 1.0192,
      "step": 5071
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2306923866271973,
      "learning_rate": 2.670315556821888e-06,
      "loss": 0.8063,
      "step": 5072
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2857311964035034,
      "learning_rate": 2.6684380731183403e-06,
      "loss": 0.7626,
      "step": 5073
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6119915246963501,
      "learning_rate": 2.666561009414057e-06,
      "loss": 1.014,
      "step": 5074
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6010538935661316,
      "learning_rate": 2.664684366047171e-06,
      "loss": 1.0022,
      "step": 5075
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2581491470336914,
      "learning_rate": 2.6628081433557307e-06,
      "loss": 0.7278,
      "step": 5076
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.35690176486969,
      "learning_rate": 2.660932341677713e-06,
      "loss": 0.7593,
      "step": 5077
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3000494241714478,
      "learning_rate": 2.6590569613510152e-06,
      "loss": 0.7646,
      "step": 5078
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6439671516418457,
      "learning_rate": 2.6571820027134663e-06,
      "loss": 0.9904,
      "step": 5079
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2890676259994507,
      "learning_rate": 2.6553074661028133e-06,
      "loss": 0.8187,
      "step": 5080
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2874596118927002,
      "learning_rate": 2.6534333518567262e-06,
      "loss": 0.718,
      "step": 5081
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2489808797836304,
      "learning_rate": 2.6515596603128045e-06,
      "loss": 0.8352,
      "step": 5082
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2919707298278809,
      "learning_rate": 2.6496863918085667e-06,
      "loss": 0.8496,
      "step": 5083
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.31493079662323,
      "learning_rate": 2.6478135466814555e-06,
      "loss": 0.7841,
      "step": 5084
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.605559229850769,
      "learning_rate": 2.6459411252688414e-06,
      "loss": 1.0228,
      "step": 5085
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2210056781768799,
      "learning_rate": 2.6440691279080144e-06,
      "loss": 0.7788,
      "step": 5086
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2116611003875732,
      "learning_rate": 2.6421975549361867e-06,
      "loss": 0.7876,
      "step": 5087
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2145729064941406,
      "learning_rate": 2.640326406690502e-06,
      "loss": 0.7392,
      "step": 5088
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2048060894012451,
      "learning_rate": 2.638455683508018e-06,
      "loss": 0.7546,
      "step": 5089
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2171467542648315,
      "learning_rate": 2.6365853857257227e-06,
      "loss": 0.7832,
      "step": 5090
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8574159145355225,
      "learning_rate": 2.6347155136805247e-06,
      "loss": 0.7969,
      "step": 5091
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2745096683502197,
      "learning_rate": 2.6328460677092537e-06,
      "loss": 0.8194,
      "step": 5092
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.195809006690979,
      "learning_rate": 2.630977048148668e-06,
      "loss": 0.7772,
      "step": 5093
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2443238496780396,
      "learning_rate": 2.629108455335443e-06,
      "loss": 0.7354,
      "step": 5094
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2630324363708496,
      "learning_rate": 2.6272402896061846e-06,
      "loss": 0.7732,
      "step": 5095
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6504112482070923,
      "learning_rate": 2.6253725512974144e-06,
      "loss": 0.9923,
      "step": 5096
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3003252744674683,
      "learning_rate": 2.623505240745579e-06,
      "loss": 0.765,
      "step": 5097
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6251606345176697,
      "learning_rate": 2.621638358287053e-06,
      "loss": 1.0132,
      "step": 5098
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6199154257774353,
      "learning_rate": 2.619771904258126e-06,
      "loss": 1.0443,
      "step": 5099
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6001665592193604,
      "learning_rate": 2.6179058789950184e-06,
      "loss": 1.0265,
      "step": 5100
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6026979088783264,
      "learning_rate": 2.6160402828338673e-06,
      "loss": 1.0137,
      "step": 5101
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2737623453140259,
      "learning_rate": 2.614175116110732e-06,
      "loss": 0.7869,
      "step": 5102
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2358123064041138,
      "learning_rate": 2.612310379161601e-06,
      "loss": 0.7871,
      "step": 5103
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2346324920654297,
      "learning_rate": 2.610446072322379e-06,
      "loss": 0.7471,
      "step": 5104
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6432560682296753,
      "learning_rate": 2.608582195928898e-06,
      "loss": 1.0084,
      "step": 5105
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3095595836639404,
      "learning_rate": 2.6067187503169085e-06,
      "loss": 0.8167,
      "step": 5106
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2450006008148193,
      "learning_rate": 2.6048557358220826e-06,
      "loss": 0.7963,
      "step": 5107
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1617594957351685,
      "learning_rate": 2.6029931527800217e-06,
      "loss": 0.7097,
      "step": 5108
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2436615228652954,
      "learning_rate": 2.601131001526241e-06,
      "loss": 0.7893,
      "step": 5109
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2280622720718384,
      "learning_rate": 2.5992692823961853e-06,
      "loss": 0.7939,
      "step": 5110
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.308534026145935,
      "learning_rate": 2.597407995725216e-06,
      "loss": 0.7766,
      "step": 5111
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2762627601623535,
      "learning_rate": 2.5955471418486167e-06,
      "loss": 0.8352,
      "step": 5112
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.625436544418335,
      "learning_rate": 2.5936867211015993e-06,
      "loss": 0.9989,
      "step": 5113
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2190107107162476,
      "learning_rate": 2.591826733819291e-06,
      "loss": 0.779,
      "step": 5114
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6254538893699646,
      "learning_rate": 2.589967180336742e-06,
      "loss": 1.0042,
      "step": 5115
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.604223906993866,
      "learning_rate": 2.5881080609889286e-06,
      "loss": 0.9996,
      "step": 5116
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2446013689041138,
      "learning_rate": 2.5862493761107448e-06,
      "loss": 0.7101,
      "step": 5117
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5791917443275452,
      "learning_rate": 2.584391126037007e-06,
      "loss": 1.0061,
      "step": 5118
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.295188307762146,
      "learning_rate": 2.582533311102452e-06,
      "loss": 0.7398,
      "step": 5119
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.251255989074707,
      "learning_rate": 2.580675931641744e-06,
      "loss": 0.8151,
      "step": 5120
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3605035543441772,
      "learning_rate": 2.5788189879894625e-06,
      "loss": 0.7788,
      "step": 5121
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2867852449417114,
      "learning_rate": 2.576962480480109e-06,
      "loss": 0.7381,
      "step": 5122
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3044195175170898,
      "learning_rate": 2.575106409448112e-06,
      "loss": 0.8158,
      "step": 5123
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.220288634300232,
      "learning_rate": 2.5732507752278157e-06,
      "loss": 0.75,
      "step": 5124
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6728734374046326,
      "learning_rate": 2.571395578153485e-06,
      "loss": 1.0196,
      "step": 5125
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2856239080429077,
      "learning_rate": 2.5695408185593126e-06,
      "loss": 0.7575,
      "step": 5126
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3043938875198364,
      "learning_rate": 2.5676864967794047e-06,
      "loss": 0.7801,
      "step": 5127
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3989336490631104,
      "learning_rate": 2.5658326131477955e-06,
      "loss": 0.7539,
      "step": 5128
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2685327529907227,
      "learning_rate": 2.563979167998436e-06,
      "loss": 0.8046,
      "step": 5129
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5992125272750854,
      "learning_rate": 2.5621261616651965e-06,
      "loss": 0.9934,
      "step": 5130
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2140151262283325,
      "learning_rate": 2.5602735944818754e-06,
      "loss": 0.7418,
      "step": 5131
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2289789915084839,
      "learning_rate": 2.558421466782183e-06,
      "loss": 0.7417,
      "step": 5132
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5849961638450623,
      "learning_rate": 2.5565697788997595e-06,
      "loss": 1.0223,
      "step": 5133
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3271580934524536,
      "learning_rate": 2.5547185311681588e-06,
      "loss": 0.7627,
      "step": 5134
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2921016216278076,
      "learning_rate": 2.552867723920857e-06,
      "loss": 0.7925,
      "step": 5135
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.275116205215454,
      "learning_rate": 2.5510173574912554e-06,
      "loss": 0.8572,
      "step": 5136
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1924703121185303,
      "learning_rate": 2.549167432212668e-06,
      "loss": 0.7511,
      "step": 5137
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6111853122711182,
      "learning_rate": 2.5473179484183386e-06,
      "loss": 1.0157,
      "step": 5138
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3770304918289185,
      "learning_rate": 2.5454689064414233e-06,
      "loss": 0.7652,
      "step": 5139
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.31877863407135,
      "learning_rate": 2.5436203066150017e-06,
      "loss": 0.8411,
      "step": 5140
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2376312017440796,
      "learning_rate": 2.541772149272077e-06,
      "loss": 0.7768,
      "step": 5141
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3149731159210205,
      "learning_rate": 2.5399244347455677e-06,
      "loss": 0.78,
      "step": 5142
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6264121532440186,
      "learning_rate": 2.538077163368312e-06,
      "loss": 1.0175,
      "step": 5143
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6118509769439697,
      "learning_rate": 2.5362303354730768e-06,
      "loss": 1.0218,
      "step": 5144
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.215552568435669,
      "learning_rate": 2.534383951392536e-06,
      "loss": 0.7293,
      "step": 5145
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2354772090911865,
      "learning_rate": 2.532538011459298e-06,
      "loss": 0.7933,
      "step": 5146
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.332767367362976,
      "learning_rate": 2.5306925160058794e-06,
      "loss": 0.7794,
      "step": 5147
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5935001373291016,
      "learning_rate": 2.52884746536472e-06,
      "loss": 1.0213,
      "step": 5148
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2777093648910522,
      "learning_rate": 2.5270028598681846e-06,
      "loss": 0.7839,
      "step": 5149
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.286042332649231,
      "learning_rate": 2.52515869984855e-06,
      "loss": 0.814,
      "step": 5150
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3403371572494507,
      "learning_rate": 2.523314985638021e-06,
      "loss": 0.6997,
      "step": 5151
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3295950889587402,
      "learning_rate": 2.521471717568715e-06,
      "loss": 0.8591,
      "step": 5152
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.274156093597412,
      "learning_rate": 2.51962889597267e-06,
      "loss": 0.7055,
      "step": 5153
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6513662934303284,
      "learning_rate": 2.5177865211818486e-06,
      "loss": 1.0035,
      "step": 5154
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.210255742073059,
      "learning_rate": 2.515944593528129e-06,
      "loss": 0.7522,
      "step": 5155
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2418100833892822,
      "learning_rate": 2.514103113343307e-06,
      "loss": 0.7117,
      "step": 5156
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3242814540863037,
      "learning_rate": 2.512262080959105e-06,
      "loss": 0.7747,
      "step": 5157
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6031501293182373,
      "learning_rate": 2.5104214967071554e-06,
      "loss": 1.0128,
      "step": 5158
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.251117467880249,
      "learning_rate": 2.5085813609190173e-06,
      "loss": 0.751,
      "step": 5159
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2355458736419678,
      "learning_rate": 2.506741673926163e-06,
      "loss": 0.7755,
      "step": 5160
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2082983255386353,
      "learning_rate": 2.5049024360599914e-06,
      "loss": 0.7686,
      "step": 5161
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2413945198059082,
      "learning_rate": 2.5030636476518145e-06,
      "loss": 0.7815,
      "step": 5162
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2238038778305054,
      "learning_rate": 2.501225309032862e-06,
      "loss": 0.7637,
      "step": 5163
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2485148906707764,
      "learning_rate": 2.499387420534291e-06,
      "loss": 0.8029,
      "step": 5164
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2351861000061035,
      "learning_rate": 2.497549982487168e-06,
      "loss": 0.7381,
      "step": 5165
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3400518894195557,
      "learning_rate": 2.495712995222486e-06,
      "loss": 0.7787,
      "step": 5166
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6250477433204651,
      "learning_rate": 2.493876459071151e-06,
      "loss": 0.996,
      "step": 5167
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6371719241142273,
      "learning_rate": 2.4920403743639893e-06,
      "loss": 1.0224,
      "step": 5168
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.185378909111023,
      "learning_rate": 2.4902047414317497e-06,
      "loss": 0.7734,
      "step": 5169
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.590591311454773,
      "learning_rate": 2.4883695606050952e-06,
      "loss": 1.0242,
      "step": 5170
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2603707313537598,
      "learning_rate": 2.4865348322146064e-06,
      "loss": 0.7391,
      "step": 5171
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3887660503387451,
      "learning_rate": 2.4847005565907886e-06,
      "loss": 0.7745,
      "step": 5172
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2776691913604736,
      "learning_rate": 2.482866734064058e-06,
      "loss": 0.7365,
      "step": 5173
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6120586395263672,
      "learning_rate": 2.4810333649647565e-06,
      "loss": 0.9978,
      "step": 5174
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2820292711257935,
      "learning_rate": 2.479200449623139e-06,
      "loss": 0.8064,
      "step": 5175
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3241784572601318,
      "learning_rate": 2.4773679883693785e-06,
      "loss": 0.8394,
      "step": 5176
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2656718492507935,
      "learning_rate": 2.475535981533572e-06,
      "loss": 0.8246,
      "step": 5177
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2504407167434692,
      "learning_rate": 2.473704429445726e-06,
      "loss": 0.6875,
      "step": 5178
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6186150312423706,
      "learning_rate": 2.4718733324357746e-06,
      "loss": 0.9908,
      "step": 5179
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3069876432418823,
      "learning_rate": 2.4700426908335627e-06,
      "loss": 0.7662,
      "step": 5180
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6401224732398987,
      "learning_rate": 2.4682125049688536e-06,
      "loss": 1.0237,
      "step": 5181
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2574293613433838,
      "learning_rate": 2.466382775171335e-06,
      "loss": 0.7934,
      "step": 5182
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3139100074768066,
      "learning_rate": 2.464553501770603e-06,
      "loss": 0.7322,
      "step": 5183
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2715566158294678,
      "learning_rate": 2.4627246850961804e-06,
      "loss": 0.7976,
      "step": 5184
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2543237209320068,
      "learning_rate": 2.4608963254775027e-06,
      "loss": 0.7795,
      "step": 5185
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3125097751617432,
      "learning_rate": 2.4590684232439216e-06,
      "loss": 0.7689,
      "step": 5186
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2557841539382935,
      "learning_rate": 2.4572409787247127e-06,
      "loss": 0.7929,
      "step": 5187
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.689008355140686,
      "learning_rate": 2.4554139922490617e-06,
      "loss": 1.0454,
      "step": 5188
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2314845323562622,
      "learning_rate": 2.453587464146079e-06,
      "loss": 0.7565,
      "step": 5189
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6191551089286804,
      "learning_rate": 2.4517613947447867e-06,
      "loss": 1.0159,
      "step": 5190
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.282225251197815,
      "learning_rate": 2.449935784374125e-06,
      "loss": 0.8047,
      "step": 5191
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2590107917785645,
      "learning_rate": 2.4481106333629573e-06,
      "loss": 0.7333,
      "step": 5192
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5954828262329102,
      "learning_rate": 2.4462859420400547e-06,
      "loss": 1.0135,
      "step": 5193
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2943716049194336,
      "learning_rate": 2.4444617107341157e-06,
      "loss": 0.7606,
      "step": 5194
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2703241109848022,
      "learning_rate": 2.4426379397737474e-06,
      "loss": 0.7513,
      "step": 5195
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6007758975028992,
      "learning_rate": 2.4408146294874785e-06,
      "loss": 1.0104,
      "step": 5196
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.369024395942688,
      "learning_rate": 2.4389917802037518e-06,
      "loss": 0.7608,
      "step": 5197
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5913619995117188,
      "learning_rate": 2.437169392250932e-06,
      "loss": 1.0253,
      "step": 5198
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3042306900024414,
      "learning_rate": 2.4353474659572962e-06,
      "loss": 0.7517,
      "step": 5199
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5958995223045349,
      "learning_rate": 2.43352600165104e-06,
      "loss": 1.0199,
      "step": 5200
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2878892421722412,
      "learning_rate": 2.4317049996602727e-06,
      "loss": 0.824,
      "step": 5201
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2954381704330444,
      "learning_rate": 2.429884460313027e-06,
      "loss": 0.8183,
      "step": 5202
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2788056135177612,
      "learning_rate": 2.4280643839372476e-06,
      "loss": 0.7722,
      "step": 5203
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6153697967529297,
      "learning_rate": 2.426244770860794e-06,
      "loss": 0.9928,
      "step": 5204
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6286501288414001,
      "learning_rate": 2.424425621411448e-06,
      "loss": 1.021,
      "step": 5205
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2247422933578491,
      "learning_rate": 2.4226069359169015e-06,
      "loss": 0.7864,
      "step": 5206
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2087454795837402,
      "learning_rate": 2.4207887147047696e-06,
      "loss": 0.8332,
      "step": 5207
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2851771116256714,
      "learning_rate": 2.4189709581025794e-06,
      "loss": 0.7783,
      "step": 5208
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2380365133285522,
      "learning_rate": 2.4171536664377714e-06,
      "loss": 0.7269,
      "step": 5209
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3075902462005615,
      "learning_rate": 2.4153368400377114e-06,
      "loss": 0.7848,
      "step": 5210
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2590104341506958,
      "learning_rate": 2.4135204792296714e-06,
      "loss": 0.7731,
      "step": 5211
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2832525968551636,
      "learning_rate": 2.4117045843408476e-06,
      "loss": 0.7769,
      "step": 5212
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2206757068634033,
      "learning_rate": 2.4098891556983483e-06,
      "loss": 0.7315,
      "step": 5213
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6169692277908325,
      "learning_rate": 2.4080741936291953e-06,
      "loss": 1.0041,
      "step": 5214
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3273130655288696,
      "learning_rate": 2.406259698460333e-06,
      "loss": 0.7636,
      "step": 5215
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.198750615119934,
      "learning_rate": 2.404445670518616e-06,
      "loss": 0.7976,
      "step": 5216
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2575448751449585,
      "learning_rate": 2.4026321101308186e-06,
      "loss": 0.7504,
      "step": 5217
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2114417552947998,
      "learning_rate": 2.4008190176236287e-06,
      "loss": 0.7771,
      "step": 5218
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2759703397750854,
      "learning_rate": 2.3990063933236475e-06,
      "loss": 0.7445,
      "step": 5219
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3529016971588135,
      "learning_rate": 2.397194237557399e-06,
      "loss": 0.8119,
      "step": 5220
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.250257134437561,
      "learning_rate": 2.3953825506513145e-06,
      "loss": 0.7721,
      "step": 5221
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.321168303489685,
      "learning_rate": 2.3935713329317486e-06,
      "loss": 0.8134,
      "step": 5222
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3414198160171509,
      "learning_rate": 2.391760584724966e-06,
      "loss": 0.7969,
      "step": 5223
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1924318075180054,
      "learning_rate": 2.3899503063571463e-06,
      "loss": 0.7616,
      "step": 5224
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6108638048171997,
      "learning_rate": 2.3881404981543898e-06,
      "loss": 1.0081,
      "step": 5225
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1743873357772827,
      "learning_rate": 2.3863311604427063e-06,
      "loss": 0.731,
      "step": 5226
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1832444667816162,
      "learning_rate": 2.3845222935480266e-06,
      "loss": 0.8012,
      "step": 5227
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2532862424850464,
      "learning_rate": 2.3827138977961915e-06,
      "loss": 0.7466,
      "step": 5228
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.231605887413025,
      "learning_rate": 2.3809059735129575e-06,
      "loss": 0.7995,
      "step": 5229
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3876796960830688,
      "learning_rate": 2.379098521024001e-06,
      "loss": 0.7908,
      "step": 5230
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2436553239822388,
      "learning_rate": 2.3772915406549084e-06,
      "loss": 0.7887,
      "step": 5231
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2286953926086426,
      "learning_rate": 2.375485032731181e-06,
      "loss": 0.7505,
      "step": 5232
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2307490110397339,
      "learning_rate": 2.3736789975782404e-06,
      "loss": 0.7829,
      "step": 5233
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2602202892303467,
      "learning_rate": 2.3718734355214157e-06,
      "loss": 0.7907,
      "step": 5234
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6229403614997864,
      "learning_rate": 2.3700683468859576e-06,
      "loss": 1.0189,
      "step": 5235
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2296342849731445,
      "learning_rate": 2.3682637319970265e-06,
      "loss": 0.8109,
      "step": 5236
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2403603792190552,
      "learning_rate": 2.3664595911797e-06,
      "loss": 0.7831,
      "step": 5237
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2986030578613281,
      "learning_rate": 2.364655924758967e-06,
      "loss": 0.8076,
      "step": 5238
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2766821384429932,
      "learning_rate": 2.3628527330597385e-06,
      "loss": 0.8298,
      "step": 5239
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.264847755432129,
      "learning_rate": 2.3610500164068317e-06,
      "loss": 0.8263,
      "step": 5240
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6334629058837891,
      "learning_rate": 2.3592477751249827e-06,
      "loss": 1.0254,
      "step": 5241
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.262593388557434,
      "learning_rate": 2.3574460095388375e-06,
      "loss": 0.7531,
      "step": 5242
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2405871152877808,
      "learning_rate": 2.3556447199729648e-06,
      "loss": 0.7843,
      "step": 5243
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3035106658935547,
      "learning_rate": 2.3538439067518388e-06,
      "loss": 0.695,
      "step": 5244
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2739816904067993,
      "learning_rate": 2.3520435701998544e-06,
      "loss": 0.7332,
      "step": 5245
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3286389112472534,
      "learning_rate": 2.350243710641317e-06,
      "loss": 0.809,
      "step": 5246
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2270303964614868,
      "learning_rate": 2.348444328400444e-06,
      "loss": 0.7944,
      "step": 5247
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2666205167770386,
      "learning_rate": 2.3466454238013743e-06,
      "loss": 0.7909,
      "step": 5248
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3780910968780518,
      "learning_rate": 2.3448469971681514e-06,
      "loss": 0.7689,
      "step": 5249
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6055684089660645,
      "learning_rate": 2.3430490488247427e-06,
      "loss": 1.0091,
      "step": 5250
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2784286737442017,
      "learning_rate": 2.3412515790950207e-06,
      "loss": 0.7404,
      "step": 5251
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3054139614105225,
      "learning_rate": 2.3394545883027747e-06,
      "loss": 0.8598,
      "step": 5252
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2939015626907349,
      "learning_rate": 2.337658076771711e-06,
      "loss": 0.7242,
      "step": 5253
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2505793571472168,
      "learning_rate": 2.3358620448254437e-06,
      "loss": 0.7778,
      "step": 5254
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5907544493675232,
      "learning_rate": 2.3340664927875067e-06,
      "loss": 1.0194,
      "step": 5255
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2781684398651123,
      "learning_rate": 2.3322714209813433e-06,
      "loss": 0.8012,
      "step": 5256
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2018053531646729,
      "learning_rate": 2.3304768297303094e-06,
      "loss": 0.7598,
      "step": 5257
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2259536981582642,
      "learning_rate": 2.328682719357679e-06,
      "loss": 0.7256,
      "step": 5258
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2142311334609985,
      "learning_rate": 2.326889090186636e-06,
      "loss": 0.8019,
      "step": 5259
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.233819842338562,
      "learning_rate": 2.325095942540276e-06,
      "loss": 0.7735,
      "step": 5260
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.248152732849121,
      "learning_rate": 2.3233032767416136e-06,
      "loss": 0.8029,
      "step": 5261
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2710072994232178,
      "learning_rate": 2.3215110931135702e-06,
      "loss": 0.7583,
      "step": 5262
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2572985887527466,
      "learning_rate": 2.3197193919789873e-06,
      "loss": 0.7536,
      "step": 5263
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2037001848220825,
      "learning_rate": 2.317928173660613e-06,
      "loss": 0.7424,
      "step": 5264
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6621717810630798,
      "learning_rate": 2.316137438481109e-06,
      "loss": 1.0032,
      "step": 5265
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6212566494941711,
      "learning_rate": 2.3143471867630563e-06,
      "loss": 0.9949,
      "step": 5266
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.281865119934082,
      "learning_rate": 2.3125574188289408e-06,
      "loss": 0.7747,
      "step": 5267
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2964951992034912,
      "learning_rate": 2.3107681350011677e-06,
      "loss": 0.8359,
      "step": 5268
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2247596979141235,
      "learning_rate": 2.308979335602052e-06,
      "loss": 0.7533,
      "step": 5269
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5851088762283325,
      "learning_rate": 2.3071910209538177e-06,
      "loss": 0.9965,
      "step": 5270
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2172107696533203,
      "learning_rate": 2.3054031913786113e-06,
      "loss": 0.7538,
      "step": 5271
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6153619885444641,
      "learning_rate": 2.303615847198481e-06,
      "loss": 1.0165,
      "step": 5272
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2547911405563354,
      "learning_rate": 2.3018289887353964e-06,
      "loss": 0.7262,
      "step": 5273
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2816203832626343,
      "learning_rate": 2.3000426163112342e-06,
      "loss": 0.7659,
      "step": 5274
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4008393287658691,
      "learning_rate": 2.298256730247784e-06,
      "loss": 0.8552,
      "step": 5275
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.270085334777832,
      "learning_rate": 2.2964713308667514e-06,
      "loss": 0.7311,
      "step": 5276
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3590394258499146,
      "learning_rate": 2.2946864184897505e-06,
      "loss": 0.7972,
      "step": 5277
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.260683536529541,
      "learning_rate": 2.2929019934383095e-06,
      "loss": 0.7557,
      "step": 5278
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2290949821472168,
      "learning_rate": 2.291118056033866e-06,
      "loss": 0.7198,
      "step": 5279
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3399142026901245,
      "learning_rate": 2.2893346065977764e-06,
      "loss": 0.7984,
      "step": 5280
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3127957582473755,
      "learning_rate": 2.2875516454513023e-06,
      "loss": 0.831,
      "step": 5281
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6141186356544495,
      "learning_rate": 2.285769172915619e-06,
      "loss": 1.014,
      "step": 5282
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.278220534324646,
      "learning_rate": 2.2839871893118175e-06,
      "loss": 0.7846,
      "step": 5283
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5934454798698425,
      "learning_rate": 2.282205694960897e-06,
      "loss": 0.9965,
      "step": 5284
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2484790086746216,
      "learning_rate": 2.280424690183768e-06,
      "loss": 0.7708,
      "step": 5285
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.270189642906189,
      "learning_rate": 2.2786441753012565e-06,
      "loss": 0.7616,
      "step": 5286
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.272720217704773,
      "learning_rate": 2.2768641506340977e-06,
      "loss": 0.8145,
      "step": 5287
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1931935548782349,
      "learning_rate": 2.275084616502936e-06,
      "loss": 0.7335,
      "step": 5288
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.318842887878418,
      "learning_rate": 2.2733055732283356e-06,
      "loss": 0.77,
      "step": 5289
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5999537706375122,
      "learning_rate": 2.271527021130762e-06,
      "loss": 1.0047,
      "step": 5290
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.270455241203308,
      "learning_rate": 2.2697489605306007e-06,
      "loss": 0.8425,
      "step": 5291
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2778741121292114,
      "learning_rate": 2.2679713917481438e-06,
      "loss": 0.7481,
      "step": 5292
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.257005214691162,
      "learning_rate": 2.266194315103595e-06,
      "loss": 0.7604,
      "step": 5293
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1782395839691162,
      "learning_rate": 2.2644177309170733e-06,
      "loss": 0.7244,
      "step": 5294
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2332631349563599,
      "learning_rate": 2.262641639508603e-06,
      "loss": 0.7114,
      "step": 5295
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5995087623596191,
      "learning_rate": 2.260866041198127e-06,
      "loss": 1.0202,
      "step": 5296
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2764114141464233,
      "learning_rate": 2.259090936305493e-06,
      "loss": 0.8252,
      "step": 5297
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.271798849105835,
      "learning_rate": 2.25731632515046e-06,
      "loss": 0.8432,
      "step": 5298
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.606604278087616,
      "learning_rate": 2.255542208052704e-06,
      "loss": 1.0143,
      "step": 5299
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2890875339508057,
      "learning_rate": 2.2537685853318046e-06,
      "loss": 0.7966,
      "step": 5300
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2796354293823242,
      "learning_rate": 2.2519954573072596e-06,
      "loss": 0.7789,
      "step": 5301
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5763800144195557,
      "learning_rate": 2.250222824298472e-06,
      "loss": 0.7963,
      "step": 5302
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2477045059204102,
      "learning_rate": 2.2484506866247556e-06,
      "loss": 0.7798,
      "step": 5303
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2760920524597168,
      "learning_rate": 2.2466790446053412e-06,
      "loss": 0.7849,
      "step": 5304
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.25416898727417,
      "learning_rate": 2.2449078985593614e-06,
      "loss": 0.7558,
      "step": 5305
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3065638542175293,
      "learning_rate": 2.243137248805868e-06,
      "loss": 0.8307,
      "step": 5306
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2780187129974365,
      "learning_rate": 2.2413670956638193e-06,
      "loss": 0.8442,
      "step": 5307
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6139873266220093,
      "learning_rate": 2.239597439452081e-06,
      "loss": 1.0184,
      "step": 5308
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2667962312698364,
      "learning_rate": 2.237828280489437e-06,
      "loss": 0.793,
      "step": 5309
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3504562377929688,
      "learning_rate": 2.2360596190945727e-06,
      "loss": 0.7832,
      "step": 5310
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2225260734558105,
      "learning_rate": 2.2342914555860934e-06,
      "loss": 0.7734,
      "step": 5311
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6087744235992432,
      "learning_rate": 2.232523790282507e-06,
      "loss": 1.0108,
      "step": 5312
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5935739874839783,
      "learning_rate": 2.230756623502233e-06,
      "loss": 1.0126,
      "step": 5313
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3246102333068848,
      "learning_rate": 2.2289899555636062e-06,
      "loss": 0.7777,
      "step": 5314
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2985965013504028,
      "learning_rate": 2.2272237867848664e-06,
      "loss": 0.785,
      "step": 5315
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3135305643081665,
      "learning_rate": 2.225458117484162e-06,
      "loss": 0.7578,
      "step": 5316
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.248063087463379,
      "learning_rate": 2.2236929479795592e-06,
      "loss": 0.7418,
      "step": 5317
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5826264023780823,
      "learning_rate": 2.2219282785890273e-06,
      "loss": 0.9968,
      "step": 5318
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1871614456176758,
      "learning_rate": 2.220164109630447e-06,
      "loss": 0.7434,
      "step": 5319
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2904231548309326,
      "learning_rate": 2.218400441421608e-06,
      "loss": 0.7779,
      "step": 5320
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2578577995300293,
      "learning_rate": 2.216637274280215e-06,
      "loss": 0.7644,
      "step": 5321
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5875464081764221,
      "learning_rate": 2.2148746085238763e-06,
      "loss": 1.0273,
      "step": 5322
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2467536926269531,
      "learning_rate": 2.213112444470111e-06,
      "loss": 0.6935,
      "step": 5323
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2463585138320923,
      "learning_rate": 2.211350782436353e-06,
      "loss": 0.7972,
      "step": 5324
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2720166444778442,
      "learning_rate": 2.209589622739938e-06,
      "loss": 0.7734,
      "step": 5325
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2898868322372437,
      "learning_rate": 2.2078289656981163e-06,
      "loss": 0.7959,
      "step": 5326
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.311921238899231,
      "learning_rate": 2.206068811628047e-06,
      "loss": 0.7927,
      "step": 5327
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3132591247558594,
      "learning_rate": 2.204309160846797e-06,
      "loss": 0.7751,
      "step": 5328
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2443312406539917,
      "learning_rate": 2.2025500136713453e-06,
      "loss": 0.7225,
      "step": 5329
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5853533744812012,
      "learning_rate": 2.2007913704185774e-06,
      "loss": 0.9812,
      "step": 5330
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2791571617126465,
      "learning_rate": 2.1990332314052866e-06,
      "loss": 0.7093,
      "step": 5331
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2932723760604858,
      "learning_rate": 2.1972755969481823e-06,
      "loss": 0.762,
      "step": 5332
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.280156135559082,
      "learning_rate": 2.1955184673638745e-06,
      "loss": 0.7994,
      "step": 5333
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2376627922058105,
      "learning_rate": 2.1937618429688896e-06,
      "loss": 0.7987,
      "step": 5334
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2239711284637451,
      "learning_rate": 2.1920057240796583e-06,
      "loss": 0.7014,
      "step": 5335
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2874077558517456,
      "learning_rate": 2.1902501110125196e-06,
      "loss": 0.6948,
      "step": 5336
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2795815467834473,
      "learning_rate": 2.188495004083727e-06,
      "loss": 0.769,
      "step": 5337
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.265648365020752,
      "learning_rate": 2.186740403609436e-06,
      "loss": 0.7553,
      "step": 5338
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2855440378189087,
      "learning_rate": 2.1849863099057167e-06,
      "loss": 0.7906,
      "step": 5339
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2422605752944946,
      "learning_rate": 2.1832327232885442e-06,
      "loss": 0.8403,
      "step": 5340
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1795392036437988,
      "learning_rate": 2.1814796440738016e-06,
      "loss": 0.7661,
      "step": 5341
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3105933666229248,
      "learning_rate": 2.1797270725772856e-06,
      "loss": 0.7581,
      "step": 5342
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2773793935775757,
      "learning_rate": 2.177975009114695e-06,
      "loss": 0.8181,
      "step": 5343
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1975582838058472,
      "learning_rate": 2.1762234540016434e-06,
      "loss": 0.8079,
      "step": 5344
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1841917037963867,
      "learning_rate": 2.1744724075536488e-06,
      "loss": 0.7997,
      "step": 5345
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6031394004821777,
      "learning_rate": 2.172721870086136e-06,
      "loss": 1.0074,
      "step": 5346
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.343579888343811,
      "learning_rate": 2.170971841914444e-06,
      "loss": 0.8229,
      "step": 5347
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2315493822097778,
      "learning_rate": 2.169222323353815e-06,
      "loss": 0.7887,
      "step": 5348
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4810912609100342,
      "learning_rate": 2.1674733147193995e-06,
      "loss": 0.7121,
      "step": 5349
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5809697508811951,
      "learning_rate": 2.1657248163262607e-06,
      "loss": 1.0317,
      "step": 5350
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3286535739898682,
      "learning_rate": 2.163976828489364e-06,
      "loss": 0.8059,
      "step": 5351
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2310247421264648,
      "learning_rate": 2.1622293515235893e-06,
      "loss": 0.786,
      "step": 5352
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2098881006240845,
      "learning_rate": 2.1604823857437185e-06,
      "loss": 0.7414,
      "step": 5353
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5778074860572815,
      "learning_rate": 2.1587359314644423e-06,
      "loss": 1.0137,
      "step": 5354
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2885030508041382,
      "learning_rate": 2.156989989000364e-06,
      "loss": 0.7984,
      "step": 5355
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2377132177352905,
      "learning_rate": 2.1552445586659904e-06,
      "loss": 0.7542,
      "step": 5356
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.218826413154602,
      "learning_rate": 2.1534996407757354e-06,
      "loss": 0.7565,
      "step": 5357
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2295186519622803,
      "learning_rate": 2.151755235643925e-06,
      "loss": 0.7907,
      "step": 5358
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.288374900817871,
      "learning_rate": 2.150011343584788e-06,
      "loss": 0.7784,
      "step": 5359
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.360758662223816,
      "learning_rate": 2.148267964912464e-06,
      "loss": 0.7864,
      "step": 5360
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2045947313308716,
      "learning_rate": 2.1465250999409975e-06,
      "loss": 0.7441,
      "step": 5361
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2064958810806274,
      "learning_rate": 2.144782748984345e-06,
      "loss": 0.7475,
      "step": 5362
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2748483419418335,
      "learning_rate": 2.1430409123563652e-06,
      "loss": 0.7346,
      "step": 5363
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2890205383300781,
      "learning_rate": 2.141299590370825e-06,
      "loss": 0.7437,
      "step": 5364
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3640092611312866,
      "learning_rate": 2.1395587833414033e-06,
      "loss": 0.7639,
      "step": 5365
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2147560119628906,
      "learning_rate": 2.1378184915816795e-06,
      "loss": 0.7277,
      "step": 5366
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3455567359924316,
      "learning_rate": 2.136078715405147e-06,
      "loss": 0.8344,
      "step": 5367
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6189049482345581,
      "learning_rate": 2.1343394551252018e-06,
      "loss": 1.0323,
      "step": 5368
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.318284511566162,
      "learning_rate": 2.132600711055145e-06,
      "loss": 0.7991,
      "step": 5369
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.284911036491394,
      "learning_rate": 2.1308624835081924e-06,
      "loss": 0.7308,
      "step": 5370
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6105337142944336,
      "learning_rate": 2.129124772797458e-06,
      "loss": 1.0163,
      "step": 5371
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4094289541244507,
      "learning_rate": 2.1273875792359704e-06,
      "loss": 0.8039,
      "step": 5372
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2978050708770752,
      "learning_rate": 2.12565090313666e-06,
      "loss": 0.7235,
      "step": 5373
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2730940580368042,
      "learning_rate": 2.1239147448123637e-06,
      "loss": 0.7497,
      "step": 5374
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2267868518829346,
      "learning_rate": 2.1221791045758306e-06,
      "loss": 0.7872,
      "step": 5375
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.252976655960083,
      "learning_rate": 2.12044398273971e-06,
      "loss": 0.8008,
      "step": 5376
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2576665878295898,
      "learning_rate": 2.118709379616559e-06,
      "loss": 0.7213,
      "step": 5377
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4186972379684448,
      "learning_rate": 2.1169752955188476e-06,
      "loss": 0.757,
      "step": 5378
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2437533140182495,
      "learning_rate": 2.115241730758943e-06,
      "loss": 0.7936,
      "step": 5379
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2305667400360107,
      "learning_rate": 2.1135086856491273e-06,
      "loss": 0.8363,
      "step": 5380
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2427443265914917,
      "learning_rate": 2.111776160501583e-06,
      "loss": 0.7769,
      "step": 5381
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.661024570465088,
      "learning_rate": 2.1100441556283992e-06,
      "loss": 0.7791,
      "step": 5382
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2475954294204712,
      "learning_rate": 2.1083126713415767e-06,
      "loss": 0.7943,
      "step": 5383
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2397780418395996,
      "learning_rate": 2.106581707953016e-06,
      "loss": 0.7394,
      "step": 5384
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1995660066604614,
      "learning_rate": 2.104851265774529e-06,
      "loss": 0.7587,
      "step": 5385
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.220292568206787,
      "learning_rate": 2.1031213451178312e-06,
      "loss": 0.7292,
      "step": 5386
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.60846346616745,
      "learning_rate": 2.101391946294541e-06,
      "loss": 1.0012,
      "step": 5387
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6099743843078613,
      "learning_rate": 2.099663069616191e-06,
      "loss": 1.0185,
      "step": 5388
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2675905227661133,
      "learning_rate": 2.0979347153942105e-06,
      "loss": 0.828,
      "step": 5389
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.226714849472046,
      "learning_rate": 2.0962068839399423e-06,
      "loss": 0.7879,
      "step": 5390
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5946820378303528,
      "learning_rate": 2.0944795755646315e-06,
      "loss": 1.0234,
      "step": 5391
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2703710794448853,
      "learning_rate": 2.0927527905794258e-06,
      "loss": 0.7522,
      "step": 5392
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5808026790618896,
      "learning_rate": 2.091026529295387e-06,
      "loss": 0.9959,
      "step": 5393
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2855578660964966,
      "learning_rate": 2.089300792023473e-06,
      "loss": 0.757,
      "step": 5394
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.208638072013855,
      "learning_rate": 2.0875755790745573e-06,
      "loss": 0.7551,
      "step": 5395
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5713147521018982,
      "learning_rate": 2.0858508907594092e-06,
      "loss": 1.021,
      "step": 5396
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3025709390640259,
      "learning_rate": 2.0841267273887108e-06,
      "loss": 0.7417,
      "step": 5397
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2783756256103516,
      "learning_rate": 2.082403089273042e-06,
      "loss": 0.7468,
      "step": 5398
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5793789625167847,
      "learning_rate": 2.080679976722899e-06,
      "loss": 1.0107,
      "step": 5399
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.604304850101471,
      "learning_rate": 2.078957390048674e-06,
      "loss": 1.0027,
      "step": 5400
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5924301743507385,
      "learning_rate": 2.0772353295606684e-06,
      "loss": 1.0141,
      "step": 5401
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.199090838432312,
      "learning_rate": 2.075513795569085e-06,
      "loss": 0.7942,
      "step": 5402
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2199188470840454,
      "learning_rate": 2.0737927883840397e-06,
      "loss": 0.7611,
      "step": 5403
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.228361964225769,
      "learning_rate": 2.0720723083155464e-06,
      "loss": 0.782,
      "step": 5404
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2276890277862549,
      "learning_rate": 2.070352355673524e-06,
      "loss": 0.7777,
      "step": 5405
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3119890689849854,
      "learning_rate": 2.0686329307678025e-06,
      "loss": 0.7815,
      "step": 5406
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6146221160888672,
      "learning_rate": 2.06691403390811e-06,
      "loss": 1.0014,
      "step": 5407
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2221046686172485,
      "learning_rate": 2.0651956654040855e-06,
      "loss": 0.8196,
      "step": 5408
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2640891075134277,
      "learning_rate": 2.0634778255652676e-06,
      "loss": 0.7492,
      "step": 5409
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2761809825897217,
      "learning_rate": 2.061760514701101e-06,
      "loss": 0.7411,
      "step": 5410
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2723116874694824,
      "learning_rate": 2.060043733120939e-06,
      "loss": 0.7977,
      "step": 5411
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6017179489135742,
      "learning_rate": 2.0583274811340327e-06,
      "loss": 1.0107,
      "step": 5412
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5866907238960266,
      "learning_rate": 2.0566117590495454e-06,
      "loss": 0.9933,
      "step": 5413
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2802077531814575,
      "learning_rate": 2.05489656717654e-06,
      "loss": 0.7653,
      "step": 5414
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.224570870399475,
      "learning_rate": 2.053181905823982e-06,
      "loss": 0.7487,
      "step": 5415
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2716566324234009,
      "learning_rate": 2.051467775300749e-06,
      "loss": 0.8027,
      "step": 5416
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2314726114273071,
      "learning_rate": 2.049754175915615e-06,
      "loss": 0.7566,
      "step": 5417
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.245422601699829,
      "learning_rate": 2.0480411079772645e-06,
      "loss": 0.7825,
      "step": 5418
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.252686619758606,
      "learning_rate": 2.0463285717942816e-06,
      "loss": 0.7947,
      "step": 5419
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2354434728622437,
      "learning_rate": 2.044616567675155e-06,
      "loss": 0.7892,
      "step": 5420
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6257315874099731,
      "learning_rate": 2.0429050959282832e-06,
      "loss": 1.0031,
      "step": 5421
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.298337697982788,
      "learning_rate": 2.0411941568619602e-06,
      "loss": 0.7366,
      "step": 5422
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2629770040512085,
      "learning_rate": 2.039483750784392e-06,
      "loss": 0.8121,
      "step": 5423
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2176088094711304,
      "learning_rate": 2.0377738780036843e-06,
      "loss": 0.7516,
      "step": 5424
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2327625751495361,
      "learning_rate": 2.036064538827845e-06,
      "loss": 0.6882,
      "step": 5425
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3248683214187622,
      "learning_rate": 2.0343557335647916e-06,
      "loss": 0.7709,
      "step": 5426
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2983531951904297,
      "learning_rate": 2.032647462522339e-06,
      "loss": 0.8067,
      "step": 5427
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.256270170211792,
      "learning_rate": 2.030939726008213e-06,
      "loss": 0.8266,
      "step": 5428
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.317942500114441,
      "learning_rate": 2.029232524330037e-06,
      "loss": 0.8017,
      "step": 5429
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3060598373413086,
      "learning_rate": 2.0275258577953383e-06,
      "loss": 0.764,
      "step": 5430
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2730660438537598,
      "learning_rate": 2.0258197267115538e-06,
      "loss": 0.7863,
      "step": 5431
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2671676874160767,
      "learning_rate": 2.024114131386017e-06,
      "loss": 0.7929,
      "step": 5432
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2353335618972778,
      "learning_rate": 2.0224090721259675e-06,
      "loss": 0.7615,
      "step": 5433
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2929469347000122,
      "learning_rate": 2.0207045492385514e-06,
      "loss": 0.7823,
      "step": 5434
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.624534010887146,
      "learning_rate": 2.0190005630308113e-06,
      "loss": 1.0074,
      "step": 5435
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.212786316871643,
      "learning_rate": 2.017297113809702e-06,
      "loss": 0.7711,
      "step": 5436
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2484580278396606,
      "learning_rate": 2.015594201882074e-06,
      "loss": 0.7614,
      "step": 5437
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2441428899765015,
      "learning_rate": 2.013891827554684e-06,
      "loss": 0.8163,
      "step": 5438
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3221452236175537,
      "learning_rate": 2.01218999113419e-06,
      "loss": 0.8258,
      "step": 5439
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2753206491470337,
      "learning_rate": 2.010488692927159e-06,
      "loss": 0.7794,
      "step": 5440
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.587138831615448,
      "learning_rate": 2.0087879332400533e-06,
      "loss": 1.0061,
      "step": 5441
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5973618626594543,
      "learning_rate": 2.007087712379242e-06,
      "loss": 1.0339,
      "step": 5442
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.610686719417572,
      "learning_rate": 2.005388030650999e-06,
      "loss": 1.021,
      "step": 5443
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2742785215377808,
      "learning_rate": 2.003688888361497e-06,
      "loss": 0.7589,
      "step": 5444
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.599727988243103,
      "learning_rate": 2.001990285816813e-06,
      "loss": 1.0411,
      "step": 5445
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3790132999420166,
      "learning_rate": 2.0002922233229288e-06,
      "loss": 0.8279,
      "step": 5446
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3310842514038086,
      "learning_rate": 1.9985947011857276e-06,
      "loss": 0.7809,
      "step": 5447
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3128223419189453,
      "learning_rate": 1.9968977197109916e-06,
      "loss": 0.7528,
      "step": 5448
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6131588816642761,
      "learning_rate": 1.995201279204414e-06,
      "loss": 0.9782,
      "step": 5449
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6013650298118591,
      "learning_rate": 1.993505379971581e-06,
      "loss": 1.0113,
      "step": 5450
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4085191488265991,
      "learning_rate": 1.99181002231799e-06,
      "loss": 0.8091,
      "step": 5451
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2093983888626099,
      "learning_rate": 1.9901152065490337e-06,
      "loss": 0.6934,
      "step": 5452
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.258054494857788,
      "learning_rate": 1.98842093297001e-06,
      "loss": 0.75,
      "step": 5453
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.215470790863037,
      "learning_rate": 1.986727201886122e-06,
      "loss": 0.7866,
      "step": 5454
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5921594500541687,
      "learning_rate": 1.9850340136024688e-06,
      "loss": 1.0198,
      "step": 5455
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5871026515960693,
      "learning_rate": 1.9833413684240593e-06,
      "loss": 1.0092,
      "step": 5456
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5901731252670288,
      "learning_rate": 1.9816492666557983e-06,
      "loss": 1.0075,
      "step": 5457
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5969985723495483,
      "learning_rate": 1.979957708602494e-06,
      "loss": 1.0206,
      "step": 5458
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.230157494544983,
      "learning_rate": 1.97826669456886e-06,
      "loss": 0.778,
      "step": 5459
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2991132736206055,
      "learning_rate": 1.976576224859509e-06,
      "loss": 0.7796,
      "step": 5460
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.566985011100769,
      "learning_rate": 1.974886299778954e-06,
      "loss": 1.0008,
      "step": 5461
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3094046115875244,
      "learning_rate": 1.9731969196316146e-06,
      "loss": 0.804,
      "step": 5462
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.196552038192749,
      "learning_rate": 1.971508084721808e-06,
      "loss": 0.7204,
      "step": 5463
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2497011423110962,
      "learning_rate": 1.9698197953537566e-06,
      "loss": 0.8304,
      "step": 5464
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3012455701828003,
      "learning_rate": 1.968132051831583e-06,
      "loss": 0.7572,
      "step": 5465
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4913074970245361,
      "learning_rate": 1.9664448544593072e-06,
      "loss": 0.768,
      "step": 5466
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3220536708831787,
      "learning_rate": 1.964758203540861e-06,
      "loss": 0.7454,
      "step": 5467
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3011573553085327,
      "learning_rate": 1.9630720993800667e-06,
      "loss": 0.7999,
      "step": 5468
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2998398542404175,
      "learning_rate": 1.961386542280657e-06,
      "loss": 0.7157,
      "step": 5469
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6041449904441833,
      "learning_rate": 1.9597015325462597e-06,
      "loss": 1.0159,
      "step": 5470
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6417827010154724,
      "learning_rate": 1.9580170704804058e-06,
      "loss": 1.034,
      "step": 5471
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2560802698135376,
      "learning_rate": 1.956333156386531e-06,
      "loss": 0.795,
      "step": 5472
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.246209740638733,
      "learning_rate": 1.954649790567967e-06,
      "loss": 0.7189,
      "step": 5473
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6053828001022339,
      "learning_rate": 1.952966973327952e-06,
      "loss": 1.0189,
      "step": 5474
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2197949886322021,
      "learning_rate": 1.951284704969621e-06,
      "loss": 0.7771,
      "step": 5475
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5868723392486572,
      "learning_rate": 1.9496029857960103e-06,
      "loss": 1.0007,
      "step": 5476
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2617356777191162,
      "learning_rate": 1.947921816110062e-06,
      "loss": 0.7555,
      "step": 5477
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2315185070037842,
      "learning_rate": 1.946241196214615e-06,
      "loss": 0.7716,
      "step": 5478
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6074092984199524,
      "learning_rate": 1.944561126412409e-06,
      "loss": 1.0158,
      "step": 5479
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5857641100883484,
      "learning_rate": 1.942881607006086e-06,
      "loss": 0.9862,
      "step": 5480
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5808943510055542,
      "learning_rate": 1.9412026382981897e-06,
      "loss": 1.0086,
      "step": 5481
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.581732451915741,
      "learning_rate": 1.9395242205911634e-06,
      "loss": 1.0086,
      "step": 5482
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3749700784683228,
      "learning_rate": 1.9378463541873494e-06,
      "loss": 0.7784,
      "step": 5483
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2492907047271729,
      "learning_rate": 1.936169039388996e-06,
      "loss": 0.7652,
      "step": 5484
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5974879264831543,
      "learning_rate": 1.934492276498246e-06,
      "loss": 1.0166,
      "step": 5485
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.260878562927246,
      "learning_rate": 1.932816065817145e-06,
      "loss": 0.7528,
      "step": 5486
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4152414798736572,
      "learning_rate": 1.9311404076476433e-06,
      "loss": 0.7789,
      "step": 5487
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6306127309799194,
      "learning_rate": 1.929465302291583e-06,
      "loss": 1.0316,
      "step": 5488
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2694306373596191,
      "learning_rate": 1.9277907500507166e-06,
      "loss": 0.7567,
      "step": 5489
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3274922370910645,
      "learning_rate": 1.9261167512266893e-06,
      "loss": 0.747,
      "step": 5490
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2009193897247314,
      "learning_rate": 1.9244433061210477e-06,
      "loss": 0.7343,
      "step": 5491
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2653199434280396,
      "learning_rate": 1.9227704150352437e-06,
      "loss": 0.7774,
      "step": 5492
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5777532458305359,
      "learning_rate": 1.921098078270625e-06,
      "loss": 1.0271,
      "step": 5493
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2397375106811523,
      "learning_rate": 1.919426296128438e-06,
      "loss": 0.7909,
      "step": 5494
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2865478992462158,
      "learning_rate": 1.917755068909835e-06,
      "loss": 0.813,
      "step": 5495
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5882442593574524,
      "learning_rate": 1.9160843969158615e-06,
      "loss": 1.0359,
      "step": 5496
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3193516731262207,
      "learning_rate": 1.91441428044747e-06,
      "loss": 0.8025,
      "step": 5497
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2250264883041382,
      "learning_rate": 1.9127447198055076e-06,
      "loss": 0.7427,
      "step": 5498
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5942315459251404,
      "learning_rate": 1.9110757152907215e-06,
      "loss": 1.0102,
      "step": 5499
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2797253131866455,
      "learning_rate": 1.909407267203763e-06,
      "loss": 0.7868,
      "step": 5500
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2275002002716064,
      "learning_rate": 1.9077393758451777e-06,
      "loss": 0.8096,
      "step": 5501
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1693791151046753,
      "learning_rate": 1.9060720415154165e-06,
      "loss": 0.8054,
      "step": 5502
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.397975206375122,
      "learning_rate": 1.904405264514826e-06,
      "loss": 0.8034,
      "step": 5503
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2666314840316772,
      "learning_rate": 1.9027390451436505e-06,
      "loss": 0.7605,
      "step": 5504
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2342267036437988,
      "learning_rate": 1.9010733837020417e-06,
      "loss": 0.7713,
      "step": 5505
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1561325788497925,
      "learning_rate": 1.8994082804900422e-06,
      "loss": 0.7547,
      "step": 5506
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5889041423797607,
      "learning_rate": 1.8977437358076002e-06,
      "loss": 1.0365,
      "step": 5507
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.238444209098816,
      "learning_rate": 1.8960797499545602e-06,
      "loss": 0.7651,
      "step": 5508
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2260109186172485,
      "learning_rate": 1.8944163232306645e-06,
      "loss": 0.7769,
      "step": 5509
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2094635963439941,
      "learning_rate": 1.8927534559355603e-06,
      "loss": 0.8084,
      "step": 5510
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.265152931213379,
      "learning_rate": 1.8910911483687872e-06,
      "loss": 0.7763,
      "step": 5511
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6073710918426514,
      "learning_rate": 1.889429400829791e-06,
      "loss": 1.0162,
      "step": 5512
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5997208952903748,
      "learning_rate": 1.8877682136179115e-06,
      "loss": 1.0128,
      "step": 5513
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3320162296295166,
      "learning_rate": 1.8861075870323864e-06,
      "loss": 0.7368,
      "step": 5514
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3476722240447998,
      "learning_rate": 1.8844475213723595e-06,
      "loss": 0.7938,
      "step": 5515
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2478259801864624,
      "learning_rate": 1.8827880169368667e-06,
      "loss": 0.7862,
      "step": 5516
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2152057886123657,
      "learning_rate": 1.881129074024844e-06,
      "loss": 0.7325,
      "step": 5517
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2711238861083984,
      "learning_rate": 1.8794706929351303e-06,
      "loss": 0.7676,
      "step": 5518
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.420460820198059,
      "learning_rate": 1.8778128739664591e-06,
      "loss": 0.7568,
      "step": 5519
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2108947038650513,
      "learning_rate": 1.8761556174174644e-06,
      "loss": 0.7625,
      "step": 5520
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2470985651016235,
      "learning_rate": 1.8744989235866766e-06,
      "loss": 0.7802,
      "step": 5521
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2365151643753052,
      "learning_rate": 1.8728427927725295e-06,
      "loss": 0.8312,
      "step": 5522
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2384852170944214,
      "learning_rate": 1.8711872252733514e-06,
      "loss": 0.7645,
      "step": 5523
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.252018690109253,
      "learning_rate": 1.8695322213873684e-06,
      "loss": 0.7923,
      "step": 5524
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6122907400131226,
      "learning_rate": 1.8678777814127108e-06,
      "loss": 1.0151,
      "step": 5525
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2316710948944092,
      "learning_rate": 1.8662239056474013e-06,
      "loss": 0.7593,
      "step": 5526
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6054991483688354,
      "learning_rate": 1.8645705943893615e-06,
      "loss": 1.0036,
      "step": 5527
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.594476580619812,
      "learning_rate": 1.8629178479364174e-06,
      "loss": 0.9913,
      "step": 5528
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2395732402801514,
      "learning_rate": 1.8612656665862843e-06,
      "loss": 0.7934,
      "step": 5529
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2737114429473877,
      "learning_rate": 1.8596140506365839e-06,
      "loss": 0.784,
      "step": 5530
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5711726546287537,
      "learning_rate": 1.8579630003848309e-06,
      "loss": 0.9945,
      "step": 5531
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5956981778144836,
      "learning_rate": 1.8563125161284384e-06,
      "loss": 1.0054,
      "step": 5532
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.273410677909851,
      "learning_rate": 1.8546625981647215e-06,
      "loss": 0.7333,
      "step": 5533
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2920230627059937,
      "learning_rate": 1.8530132467908868e-06,
      "loss": 0.7942,
      "step": 5534
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5989683866500854,
      "learning_rate": 1.8513644623040472e-06,
      "loss": 1.0157,
      "step": 5535
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2772318124771118,
      "learning_rate": 1.8497162450012062e-06,
      "loss": 0.8,
      "step": 5536
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5884794592857361,
      "learning_rate": 1.8480685951792659e-06,
      "loss": 0.9991,
      "step": 5537
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2502107620239258,
      "learning_rate": 1.8464215131350327e-06,
      "loss": 0.8156,
      "step": 5538
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2578885555267334,
      "learning_rate": 1.8447749991652014e-06,
      "loss": 0.7136,
      "step": 5539
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2926766872406006,
      "learning_rate": 1.8431290535663737e-06,
      "loss": 0.7919,
      "step": 5540
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3718088865280151,
      "learning_rate": 1.841483676635042e-06,
      "loss": 0.7765,
      "step": 5541
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3137909173965454,
      "learning_rate": 1.8398388686675965e-06,
      "loss": 0.8246,
      "step": 5542
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2827142477035522,
      "learning_rate": 1.8381946299603316e-06,
      "loss": 0.7489,
      "step": 5543
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2044105529785156,
      "learning_rate": 1.83655096080943e-06,
      "loss": 0.7328,
      "step": 5544
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6098160743713379,
      "learning_rate": 1.8349078615109805e-06,
      "loss": 1.0066,
      "step": 5545
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2364282608032227,
      "learning_rate": 1.8332653323609638e-06,
      "loss": 0.7923,
      "step": 5546
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2129778861999512,
      "learning_rate": 1.8316233736552568e-06,
      "loss": 0.7489,
      "step": 5547
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.281783103942871,
      "learning_rate": 1.82998198568964e-06,
      "loss": 0.8053,
      "step": 5548
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.240446925163269,
      "learning_rate": 1.8283411687597851e-06,
      "loss": 0.7693,
      "step": 5549
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2336539030075073,
      "learning_rate": 1.826700923161262e-06,
      "loss": 0.8271,
      "step": 5550
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2960071563720703,
      "learning_rate": 1.8250612491895415e-06,
      "loss": 0.8139,
      "step": 5551
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2088488340377808,
      "learning_rate": 1.8234221471399849e-06,
      "loss": 0.7365,
      "step": 5552
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6033432483673096,
      "learning_rate": 1.8217836173078584e-06,
      "loss": 1.0038,
      "step": 5553
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1996077299118042,
      "learning_rate": 1.8201456599883188e-06,
      "loss": 0.7577,
      "step": 5554
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3019474744796753,
      "learning_rate": 1.8185082754764205e-06,
      "loss": 0.7448,
      "step": 5555
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4028193950653076,
      "learning_rate": 1.8168714640671193e-06,
      "loss": 0.8237,
      "step": 5556
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.270087718963623,
      "learning_rate": 1.815235226055263e-06,
      "loss": 0.7826,
      "step": 5557
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2124840021133423,
      "learning_rate": 1.8135995617355957e-06,
      "loss": 0.8011,
      "step": 5558
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2844257354736328,
      "learning_rate": 1.811964471402764e-06,
      "loss": 0.8141,
      "step": 5559
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2674741744995117,
      "learning_rate": 1.8103299553513048e-06,
      "loss": 0.7995,
      "step": 5560
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2542394399642944,
      "learning_rate": 1.8086960138756531e-06,
      "loss": 0.7446,
      "step": 5561
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6074081659317017,
      "learning_rate": 1.8070626472701446e-06,
      "loss": 1.0405,
      "step": 5562
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6199937462806702,
      "learning_rate": 1.8054298558290063e-06,
      "loss": 0.997,
      "step": 5563
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2236264944076538,
      "learning_rate": 1.803797639846363e-06,
      "loss": 0.76,
      "step": 5564
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1667534112930298,
      "learning_rate": 1.802165999616235e-06,
      "loss": 0.7477,
      "step": 5565
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6000190377235413,
      "learning_rate": 1.8005349354325435e-06,
      "loss": 0.9913,
      "step": 5566
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3061120510101318,
      "learning_rate": 1.7989044475891e-06,
      "loss": 0.698,
      "step": 5567
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2291243076324463,
      "learning_rate": 1.7972745363796168e-06,
      "loss": 0.7569,
      "step": 5568
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3201086521148682,
      "learning_rate": 1.795645202097699e-06,
      "loss": 0.7893,
      "step": 5569
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2314988374710083,
      "learning_rate": 1.794016445036848e-06,
      "loss": 0.76,
      "step": 5570
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2941545248031616,
      "learning_rate": 1.792388265490465e-06,
      "loss": 0.7699,
      "step": 5571
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3168450593948364,
      "learning_rate": 1.7907606637518415e-06,
      "loss": 0.8206,
      "step": 5572
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3442631959915161,
      "learning_rate": 1.7891336401141712e-06,
      "loss": 0.7249,
      "step": 5573
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2761623859405518,
      "learning_rate": 1.7875071948705386e-06,
      "loss": 0.7841,
      "step": 5574
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2100216150283813,
      "learning_rate": 1.7858813283139237e-06,
      "loss": 0.769,
      "step": 5575
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2781983613967896,
      "learning_rate": 1.7842560407372083e-06,
      "loss": 0.7479,
      "step": 5576
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3111598491668701,
      "learning_rate": 1.7826313324331646e-06,
      "loss": 0.8155,
      "step": 5577
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6008740067481995,
      "learning_rate": 1.7810072036944588e-06,
      "loss": 1.0109,
      "step": 5578
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2721902132034302,
      "learning_rate": 1.7793836548136595e-06,
      "loss": 0.7532,
      "step": 5579
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2981436252593994,
      "learning_rate": 1.7777606860832242e-06,
      "loss": 0.7941,
      "step": 5580
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5895186066627502,
      "learning_rate": 1.7761382977955117e-06,
      "loss": 1.0081,
      "step": 5581
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2921713590621948,
      "learning_rate": 1.774516490242772e-06,
      "loss": 0.7426,
      "step": 5582
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2856087684631348,
      "learning_rate": 1.7728952637171488e-06,
      "loss": 0.8184,
      "step": 5583
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6028676629066467,
      "learning_rate": 1.7712746185106888e-06,
      "loss": 0.9988,
      "step": 5584
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4075050354003906,
      "learning_rate": 1.7696545549153255e-06,
      "loss": 0.7645,
      "step": 5585
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2716208696365356,
      "learning_rate": 1.7680350732228945e-06,
      "loss": 0.779,
      "step": 5586
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.300072431564331,
      "learning_rate": 1.7664161737251217e-06,
      "loss": 0.7693,
      "step": 5587
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2777032852172852,
      "learning_rate": 1.7647978567136292e-06,
      "loss": 0.7422,
      "step": 5588
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.232588291168213,
      "learning_rate": 1.7631801224799367e-06,
      "loss": 0.7836,
      "step": 5589
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.298367977142334,
      "learning_rate": 1.7615629713154548e-06,
      "loss": 0.7843,
      "step": 5590
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5885098576545715,
      "learning_rate": 1.7599464035114944e-06,
      "loss": 1.0092,
      "step": 5591
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2702140808105469,
      "learning_rate": 1.7583304193592566e-06,
      "loss": 0.7495,
      "step": 5592
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1761629581451416,
      "learning_rate": 1.7567150191498372e-06,
      "loss": 0.7248,
      "step": 5593
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5818374752998352,
      "learning_rate": 1.7551002031742325e-06,
      "loss": 1.0154,
      "step": 5594
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2308485507965088,
      "learning_rate": 1.753485971723326e-06,
      "loss": 0.7463,
      "step": 5595
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2158557176589966,
      "learning_rate": 1.7518723250879027e-06,
      "loss": 0.7511,
      "step": 5596
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4545587301254272,
      "learning_rate": 1.750259263558638e-06,
      "loss": 0.7943,
      "step": 5597
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5836336016654968,
      "learning_rate": 1.7486467874261025e-06,
      "loss": 1.0053,
      "step": 5598
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3435367345809937,
      "learning_rate": 1.747034896980761e-06,
      "loss": 0.7282,
      "step": 5599
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.311076045036316,
      "learning_rate": 1.7454235925129764e-06,
      "loss": 0.7684,
      "step": 5600
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1952574253082275,
      "learning_rate": 1.7438128743130018e-06,
      "loss": 0.8092,
      "step": 5601
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2611124515533447,
      "learning_rate": 1.7422027426709848e-06,
      "loss": 0.7309,
      "step": 5602
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2694995403289795,
      "learning_rate": 1.7405931978769719e-06,
      "loss": 0.7726,
      "step": 5603
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6150548458099365,
      "learning_rate": 1.738984240220899e-06,
      "loss": 1.0324,
      "step": 5604
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.286224603652954,
      "learning_rate": 1.737375869992597e-06,
      "loss": 0.7566,
      "step": 5605
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1955751180648804,
      "learning_rate": 1.7357680874817944e-06,
      "loss": 0.7583,
      "step": 5606
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2574462890625,
      "learning_rate": 1.7341608929781107e-06,
      "loss": 0.8149,
      "step": 5607
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1730197668075562,
      "learning_rate": 1.7325542867710576e-06,
      "loss": 0.7562,
      "step": 5608
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2636357545852661,
      "learning_rate": 1.7309482691500473e-06,
      "loss": 0.7494,
      "step": 5609
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2212390899658203,
      "learning_rate": 1.72934284040438e-06,
      "loss": 0.7433,
      "step": 5610
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2411298751831055,
      "learning_rate": 1.7277380008232508e-06,
      "loss": 0.7532,
      "step": 5611
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2138620615005493,
      "learning_rate": 1.7261337506957532e-06,
      "loss": 0.7386,
      "step": 5612
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6112498641014099,
      "learning_rate": 1.7245300903108669e-06,
      "loss": 1.0017,
      "step": 5613
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2249940633773804,
      "learning_rate": 1.7229270199574743e-06,
      "loss": 0.7847,
      "step": 5614
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5953035950660706,
      "learning_rate": 1.7213245399243439e-06,
      "loss": 1.0107,
      "step": 5615
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.238617181777954,
      "learning_rate": 1.719722650500139e-06,
      "loss": 0.7935,
      "step": 5616
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.230312466621399,
      "learning_rate": 1.7181213519734225e-06,
      "loss": 0.7495,
      "step": 5617
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5721988081932068,
      "learning_rate": 1.7165206446326433e-06,
      "loss": 1.0192,
      "step": 5618
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1694501638412476,
      "learning_rate": 1.7149205287661492e-06,
      "loss": 0.75,
      "step": 5619
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.23813796043396,
      "learning_rate": 1.713321004662179e-06,
      "loss": 0.8236,
      "step": 5620
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5856726765632629,
      "learning_rate": 1.7117220726088628e-06,
      "loss": 0.9838,
      "step": 5621
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5868394374847412,
      "learning_rate": 1.7101237328942305e-06,
      "loss": 0.9837,
      "step": 5622
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3215734958648682,
      "learning_rate": 1.7085259858061977e-06,
      "loss": 0.8014,
      "step": 5623
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.824876308441162,
      "learning_rate": 1.7069288316325799e-06,
      "loss": 0.8283,
      "step": 5624
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3115689754486084,
      "learning_rate": 1.705332270661082e-06,
      "loss": 0.7816,
      "step": 5625
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.220325231552124,
      "learning_rate": 1.7037363031793003e-06,
      "loss": 0.7702,
      "step": 5626
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5904744267463684,
      "learning_rate": 1.7021409294747299e-06,
      "loss": 1.0318,
      "step": 5627
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3039708137512207,
      "learning_rate": 1.7005461498347532e-06,
      "loss": 0.8155,
      "step": 5628
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5772900581359863,
      "learning_rate": 1.6989519645466507e-06,
      "loss": 0.9961,
      "step": 5629
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3248592615127563,
      "learning_rate": 1.6973583738975925e-06,
      "loss": 0.8234,
      "step": 5630
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2737103700637817,
      "learning_rate": 1.6957653781746397e-06,
      "loss": 0.7734,
      "step": 5631
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2584316730499268,
      "learning_rate": 1.6941729776647524e-06,
      "loss": 0.8447,
      "step": 5632
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2319622039794922,
      "learning_rate": 1.6925811726547774e-06,
      "loss": 0.8175,
      "step": 5633
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2396454811096191,
      "learning_rate": 1.6909899634314592e-06,
      "loss": 0.8458,
      "step": 5634
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2845096588134766,
      "learning_rate": 1.6893993502814316e-06,
      "loss": 0.7476,
      "step": 5635
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3153326511383057,
      "learning_rate": 1.6878093334912199e-06,
      "loss": 0.753,
      "step": 5636
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2615342140197754,
      "learning_rate": 1.6862199133472473e-06,
      "loss": 0.7938,
      "step": 5637
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3169479370117188,
      "learning_rate": 1.6846310901358248e-06,
      "loss": 0.7662,
      "step": 5638
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2809809446334839,
      "learning_rate": 1.683042864143158e-06,
      "loss": 0.7288,
      "step": 5639
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5881851315498352,
      "learning_rate": 1.6814552356553416e-06,
      "loss": 1.0321,
      "step": 5640
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1984119415283203,
      "learning_rate": 1.6798682049583687e-06,
      "loss": 0.7908,
      "step": 5641
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.227935552597046,
      "learning_rate": 1.6782817723381206e-06,
      "loss": 0.81,
      "step": 5642
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2502161264419556,
      "learning_rate": 1.6766959380803693e-06,
      "loss": 0.7512,
      "step": 5643
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6011190414428711,
      "learning_rate": 1.6751107024707846e-06,
      "loss": 1.0104,
      "step": 5644
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2321573495864868,
      "learning_rate": 1.6735260657949243e-06,
      "loss": 0.7814,
      "step": 5645
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5758342146873474,
      "learning_rate": 1.6719420283382366e-06,
      "loss": 1.0326,
      "step": 5646
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2678906917572021,
      "learning_rate": 1.6703585903860681e-06,
      "loss": 0.7904,
      "step": 5647
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2642533779144287,
      "learning_rate": 1.668775752223652e-06,
      "loss": 0.7801,
      "step": 5648
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.227104902267456,
      "learning_rate": 1.6671935141361134e-06,
      "loss": 0.7644,
      "step": 5649
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2188750505447388,
      "learning_rate": 1.6656118764084745e-06,
      "loss": 0.8066,
      "step": 5650
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2233806848526,
      "learning_rate": 1.6640308393256427e-06,
      "loss": 0.7492,
      "step": 5651
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2449893951416016,
      "learning_rate": 1.662450403172423e-06,
      "loss": 0.8103,
      "step": 5652
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.22795569896698,
      "learning_rate": 1.6608705682335092e-06,
      "loss": 0.7564,
      "step": 5653
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2453103065490723,
      "learning_rate": 1.6592913347934836e-06,
      "loss": 0.7852,
      "step": 5654
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3759288787841797,
      "learning_rate": 1.657712703136829e-06,
      "loss": 0.8402,
      "step": 5655
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2220240831375122,
      "learning_rate": 1.65613467354791e-06,
      "loss": 0.7518,
      "step": 5656
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2573468685150146,
      "learning_rate": 1.6545572463109904e-06,
      "loss": 0.7497,
      "step": 5657
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.192051649093628,
      "learning_rate": 1.6529804217102214e-06,
      "loss": 0.7696,
      "step": 5658
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2422268390655518,
      "learning_rate": 1.6514042000296448e-06,
      "loss": 0.7152,
      "step": 5659
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.239172101020813,
      "learning_rate": 1.6498285815531984e-06,
      "loss": 0.7199,
      "step": 5660
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2343714237213135,
      "learning_rate": 1.6482535665647053e-06,
      "loss": 0.7382,
      "step": 5661
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2913858890533447,
      "learning_rate": 1.6466791553478872e-06,
      "loss": 0.7782,
      "step": 5662
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2246607542037964,
      "learning_rate": 1.6451053481863498e-06,
      "loss": 0.748,
      "step": 5663
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3216115236282349,
      "learning_rate": 1.6435321453635934e-06,
      "loss": 0.6981,
      "step": 5664
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2693971395492554,
      "learning_rate": 1.6419595471630107e-06,
      "loss": 0.8361,
      "step": 5665
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2110072374343872,
      "learning_rate": 1.6403875538678827e-06,
      "loss": 0.8142,
      "step": 5666
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3023045063018799,
      "learning_rate": 1.6388161657613816e-06,
      "loss": 0.7789,
      "step": 5667
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3216955661773682,
      "learning_rate": 1.6372453831265744e-06,
      "loss": 0.8844,
      "step": 5668
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2493466138839722,
      "learning_rate": 1.6356752062464137e-06,
      "loss": 0.7651,
      "step": 5669
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5995171666145325,
      "learning_rate": 1.6341056354037482e-06,
      "loss": 1.0218,
      "step": 5670
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3596453666687012,
      "learning_rate": 1.6325366708813135e-06,
      "loss": 0.7776,
      "step": 5671
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.59131920337677,
      "learning_rate": 1.630968312961735e-06,
      "loss": 1.0463,
      "step": 5672
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2221970558166504,
      "learning_rate": 1.6294005619275354e-06,
      "loss": 0.7843,
      "step": 5673
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1987863779067993,
      "learning_rate": 1.62783341806112e-06,
      "loss": 0.7831,
      "step": 5674
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.218514323234558,
      "learning_rate": 1.6262668816447918e-06,
      "loss": 0.7984,
      "step": 5675
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3699824810028076,
      "learning_rate": 1.6247009529607394e-06,
      "loss": 0.8321,
      "step": 5676
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5979382395744324,
      "learning_rate": 1.6231356322910418e-06,
      "loss": 0.9878,
      "step": 5677
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.19979727268219,
      "learning_rate": 1.6215709199176744e-06,
      "loss": 0.7371,
      "step": 5678
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3292899131774902,
      "learning_rate": 1.6200068161224963e-06,
      "loss": 0.783,
      "step": 5679
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5911884903907776,
      "learning_rate": 1.6184433211872597e-06,
      "loss": 1.0055,
      "step": 5680
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3061025142669678,
      "learning_rate": 1.6168804353936063e-06,
      "loss": 0.8606,
      "step": 5681
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2698955535888672,
      "learning_rate": 1.615318159023071e-06,
      "loss": 0.8239,
      "step": 5682
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2559200525283813,
      "learning_rate": 1.613756492357076e-06,
      "loss": 0.739,
      "step": 5683
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2471566200256348,
      "learning_rate": 1.6121954356769314e-06,
      "loss": 0.7391,
      "step": 5684
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5882711410522461,
      "learning_rate": 1.6106349892638451e-06,
      "loss": 1.0157,
      "step": 5685
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3027070760726929,
      "learning_rate": 1.6090751533989084e-06,
      "loss": 0.7862,
      "step": 5686
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3319787979125977,
      "learning_rate": 1.607515928363102e-06,
      "loss": 0.8075,
      "step": 5687
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5751706957817078,
      "learning_rate": 1.6059573144373035e-06,
      "loss": 1.005,
      "step": 5688
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.222191333770752,
      "learning_rate": 1.6043993119022722e-06,
      "loss": 0.7386,
      "step": 5689
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5899686813354492,
      "learning_rate": 1.6028419210386648e-06,
      "loss": 0.9968,
      "step": 5690
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2880207300186157,
      "learning_rate": 1.601285142127022e-06,
      "loss": 0.8474,
      "step": 5691
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.581082284450531,
      "learning_rate": 1.5997289754477758e-06,
      "loss": 0.9949,
      "step": 5692
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3034355640411377,
      "learning_rate": 1.5981734212812506e-06,
      "loss": 0.7532,
      "step": 5693
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.576894223690033,
      "learning_rate": 1.5966184799076567e-06,
      "loss": 1.0077,
      "step": 5694
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5636464953422546,
      "learning_rate": 1.5950641516070953e-06,
      "loss": 1.031,
      "step": 5695
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3118470907211304,
      "learning_rate": 1.59351043665956e-06,
      "loss": 0.8217,
      "step": 5696
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2961382865905762,
      "learning_rate": 1.591957335344928e-06,
      "loss": 0.7879,
      "step": 5697
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2926583290100098,
      "learning_rate": 1.5904048479429734e-06,
      "loss": 0.751,
      "step": 5698
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2390692234039307,
      "learning_rate": 1.5888529747333537e-06,
      "loss": 0.7933,
      "step": 5699
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.228803038597107,
      "learning_rate": 1.5873017159956167e-06,
      "loss": 0.6786,
      "step": 5700
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2766815423965454,
      "learning_rate": 1.585751072009203e-06,
      "loss": 0.7367,
      "step": 5701
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1955140829086304,
      "learning_rate": 1.584201043053437e-06,
      "loss": 0.7113,
      "step": 5702
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3527686595916748,
      "learning_rate": 1.5826516294075394e-06,
      "loss": 0.7432,
      "step": 5703
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.286595106124878,
      "learning_rate": 1.5811028313506138e-06,
      "loss": 0.7755,
      "step": 5704
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5995773077011108,
      "learning_rate": 1.5795546491616537e-06,
      "loss": 1.0027,
      "step": 5705
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2418183088302612,
      "learning_rate": 1.5780070831195466e-06,
      "loss": 0.7957,
      "step": 5706
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2357510328292847,
      "learning_rate": 1.5764601335030615e-06,
      "loss": 0.8256,
      "step": 5707
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2113958597183228,
      "learning_rate": 1.5749138005908648e-06,
      "loss": 0.7544,
      "step": 5708
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5949121713638306,
      "learning_rate": 1.573368084661505e-06,
      "loss": 1.0037,
      "step": 5709
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.30686616897583,
      "learning_rate": 1.571822985993421e-06,
      "loss": 0.7995,
      "step": 5710
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.304824948310852,
      "learning_rate": 1.5702785048649432e-06,
      "loss": 0.7874,
      "step": 5711
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5785019397735596,
      "learning_rate": 1.568734641554287e-06,
      "loss": 1.0124,
      "step": 5712
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5896837115287781,
      "learning_rate": 1.5671913963395612e-06,
      "loss": 1.0003,
      "step": 5713
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2540864944458008,
      "learning_rate": 1.5656487694987593e-06,
      "loss": 0.8581,
      "step": 5714
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5759269595146179,
      "learning_rate": 1.5641067613097621e-06,
      "loss": 1.029,
      "step": 5715
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2473039627075195,
      "learning_rate": 1.5625653720503458e-06,
      "loss": 0.7461,
      "step": 5716
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2423595190048218,
      "learning_rate": 1.561024601998168e-06,
      "loss": 0.8179,
      "step": 5717
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3430575132369995,
      "learning_rate": 1.5594844514307767e-06,
      "loss": 0.8106,
      "step": 5718
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2766727209091187,
      "learning_rate": 1.557944920625612e-06,
      "loss": 0.7766,
      "step": 5719
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2600923776626587,
      "learning_rate": 1.5564060098599986e-06,
      "loss": 0.7597,
      "step": 5720
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2788870334625244,
      "learning_rate": 1.5548677194111477e-06,
      "loss": 0.8458,
      "step": 5721
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2887474298477173,
      "learning_rate": 1.5533300495561653e-06,
      "loss": 0.765,
      "step": 5722
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2365366220474243,
      "learning_rate": 1.5517930005720404e-06,
      "loss": 0.7734,
      "step": 5723
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4395517110824585,
      "learning_rate": 1.550256572735651e-06,
      "loss": 0.8303,
      "step": 5724
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6421467661857605,
      "learning_rate": 1.5487207663237625e-06,
      "loss": 1.0015,
      "step": 5725
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3956772089004517,
      "learning_rate": 1.5471855816130328e-06,
      "loss": 0.7893,
      "step": 5726
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6220532059669495,
      "learning_rate": 1.5456510188800023e-06,
      "loss": 1.0,
      "step": 5727
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3270045518875122,
      "learning_rate": 1.544117078401101e-06,
      "loss": 0.7407,
      "step": 5728
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.237687587738037,
      "learning_rate": 1.5425837604526506e-06,
      "loss": 0.764,
      "step": 5729
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2582720518112183,
      "learning_rate": 1.541051065310854e-06,
      "loss": 0.744,
      "step": 5730
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6090282797813416,
      "learning_rate": 1.5395189932518085e-06,
      "loss": 0.9972,
      "step": 5731
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2418161630630493,
      "learning_rate": 1.5379875445514947e-06,
      "loss": 0.759,
      "step": 5732
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3309707641601562,
      "learning_rate": 1.536456719485781e-06,
      "loss": 0.7778,
      "step": 5733
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5751979351043701,
      "learning_rate": 1.5349265183304273e-06,
      "loss": 1.0096,
      "step": 5734
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.23201584815979,
      "learning_rate": 1.5333969413610766e-06,
      "loss": 0.7651,
      "step": 5735
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2576080560684204,
      "learning_rate": 1.5318679888532633e-06,
      "loss": 0.7253,
      "step": 5736
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3521788120269775,
      "learning_rate": 1.5303396610824061e-06,
      "loss": 0.8079,
      "step": 5737
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.230859637260437,
      "learning_rate": 1.528811958323812e-06,
      "loss": 0.7775,
      "step": 5738
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2766460180282593,
      "learning_rate": 1.527284880852678e-06,
      "loss": 0.7779,
      "step": 5739
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4151685237884521,
      "learning_rate": 1.5257584289440835e-06,
      "loss": 0.7976,
      "step": 5740
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6244418621063232,
      "learning_rate": 1.5242326028730015e-06,
      "loss": 0.9993,
      "step": 5741
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6098215579986572,
      "learning_rate": 1.5227074029142874e-06,
      "loss": 1.0142,
      "step": 5742
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6008574366569519,
      "learning_rate": 1.5211828293426833e-06,
      "loss": 1.0122,
      "step": 5743
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2957478761672974,
      "learning_rate": 1.519658882432824e-06,
      "loss": 0.7942,
      "step": 5744
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.265913486480713,
      "learning_rate": 1.5181355624592242e-06,
      "loss": 0.7879,
      "step": 5745
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5777853727340698,
      "learning_rate": 1.5166128696962929e-06,
      "loss": 1.0188,
      "step": 5746
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3165504932403564,
      "learning_rate": 1.515090804418321e-06,
      "loss": 0.7193,
      "step": 5747
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2810860872268677,
      "learning_rate": 1.5135693668994866e-06,
      "loss": 0.8109,
      "step": 5748
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6012042164802551,
      "learning_rate": 1.5120485574138583e-06,
      "loss": 1.0084,
      "step": 5749
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3056166172027588,
      "learning_rate": 1.5105283762353866e-06,
      "loss": 0.7772,
      "step": 5750
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.247025966644287,
      "learning_rate": 1.5090088236379151e-06,
      "loss": 0.736,
      "step": 5751
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2173174619674683,
      "learning_rate": 1.5074898998951686e-06,
      "loss": 0.7899,
      "step": 5752
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3046835660934448,
      "learning_rate": 1.5059716052807593e-06,
      "loss": 0.7855,
      "step": 5753
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2549844980239868,
      "learning_rate": 1.5044539400681901e-06,
      "loss": 0.7871,
      "step": 5754
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2692615985870361,
      "learning_rate": 1.5029369045308466e-06,
      "loss": 0.7707,
      "step": 5755
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2191135883331299,
      "learning_rate": 1.5014204989420005e-06,
      "loss": 0.7702,
      "step": 5756
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6046841144561768,
      "learning_rate": 1.4999047235748149e-06,
      "loss": 0.9869,
      "step": 5757
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2292041778564453,
      "learning_rate": 1.4983895787023345e-06,
      "loss": 0.7111,
      "step": 5758
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2120620012283325,
      "learning_rate": 1.496875064597491e-06,
      "loss": 0.7519,
      "step": 5759
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5875847935676575,
      "learning_rate": 1.495361181533106e-06,
      "loss": 1.0099,
      "step": 5760
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5861993432044983,
      "learning_rate": 1.4938479297818837e-06,
      "loss": 1.0221,
      "step": 5761
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2416399717330933,
      "learning_rate": 1.4923353096164144e-06,
      "loss": 0.8054,
      "step": 5762
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2427459955215454,
      "learning_rate": 1.4908233213091793e-06,
      "loss": 0.7684,
      "step": 5763
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.24257230758667,
      "learning_rate": 1.4893119651325405e-06,
      "loss": 0.7861,
      "step": 5764
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6026312708854675,
      "learning_rate": 1.4878012413587495e-06,
      "loss": 1.001,
      "step": 5765
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.294411063194275,
      "learning_rate": 1.4862911502599404e-06,
      "loss": 0.7745,
      "step": 5766
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3093774318695068,
      "learning_rate": 1.4847816921081382e-06,
      "loss": 0.7966,
      "step": 5767
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2726181745529175,
      "learning_rate": 1.4832728671752488e-06,
      "loss": 0.7893,
      "step": 5768
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5953940749168396,
      "learning_rate": 1.48176467573307e-06,
      "loss": 0.9994,
      "step": 5769
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5910455584526062,
      "learning_rate": 1.4802571180532798e-06,
      "loss": 1.0175,
      "step": 5770
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2365061044692993,
      "learning_rate": 1.4787501944074435e-06,
      "loss": 0.7408,
      "step": 5771
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.235034704208374,
      "learning_rate": 1.4772439050670157e-06,
      "loss": 0.7694,
      "step": 5772
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.582173228263855,
      "learning_rate": 1.47573825030333e-06,
      "loss": 1.0295,
      "step": 5773
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5678975582122803,
      "learning_rate": 1.4742332303876144e-06,
      "loss": 1.0336,
      "step": 5774
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2626159191131592,
      "learning_rate": 1.472728845590975e-06,
      "loss": 0.7206,
      "step": 5775
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.592038631439209,
      "learning_rate": 1.4712250961844054e-06,
      "loss": 0.9965,
      "step": 5776
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5956904888153076,
      "learning_rate": 1.4697219824387887e-06,
      "loss": 1.0148,
      "step": 5777
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.294520616531372,
      "learning_rate": 1.4682195046248871e-06,
      "loss": 0.8228,
      "step": 5778
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2559081315994263,
      "learning_rate": 1.466717663013355e-06,
      "loss": 0.8655,
      "step": 5779
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3453701734542847,
      "learning_rate": 1.4652164578747274e-06,
      "loss": 0.7539,
      "step": 5780
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3222846984863281,
      "learning_rate": 1.463715889479424e-06,
      "loss": 0.7238,
      "step": 5781
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2784478664398193,
      "learning_rate": 1.4622159580977546e-06,
      "loss": 0.761,
      "step": 5782
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.255242943763733,
      "learning_rate": 1.4607166639999114e-06,
      "loss": 0.7929,
      "step": 5783
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2399097681045532,
      "learning_rate": 1.4592180074559692e-06,
      "loss": 0.7102,
      "step": 5784
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2073203325271606,
      "learning_rate": 1.4577199887358939e-06,
      "loss": 0.7295,
      "step": 5785
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5825672745704651,
      "learning_rate": 1.4562226081095304e-06,
      "loss": 1.0224,
      "step": 5786
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.193888545036316,
      "learning_rate": 1.4547258658466147e-06,
      "loss": 0.7332,
      "step": 5787
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.301507592201233,
      "learning_rate": 1.453229762216763e-06,
      "loss": 0.8025,
      "step": 5788
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2446705102920532,
      "learning_rate": 1.451734297489476e-06,
      "loss": 0.7447,
      "step": 5789
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2537908554077148,
      "learning_rate": 1.450239471934145e-06,
      "loss": 0.8123,
      "step": 5790
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2238765954971313,
      "learning_rate": 1.4487452858200395e-06,
      "loss": 0.8177,
      "step": 5791
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2474431991577148,
      "learning_rate": 1.44725173941632e-06,
      "loss": 0.8223,
      "step": 5792
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.21596360206604,
      "learning_rate": 1.445758832992027e-06,
      "loss": 0.7453,
      "step": 5793
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5917344689369202,
      "learning_rate": 1.4442665668160855e-06,
      "loss": 1.0046,
      "step": 5794
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.276322364807129,
      "learning_rate": 1.44277494115731e-06,
      "loss": 0.8309,
      "step": 5795
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.576990008354187,
      "learning_rate": 1.4412839562843944e-06,
      "loss": 1.016,
      "step": 5796
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6073499917984009,
      "learning_rate": 1.4397936124659218e-06,
      "loss": 0.9962,
      "step": 5797
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2411810159683228,
      "learning_rate": 1.438303909970356e-06,
      "loss": 0.778,
      "step": 5798
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.584256649017334,
      "learning_rate": 1.436814849066047e-06,
      "loss": 1.0177,
      "step": 5799
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2144520282745361,
      "learning_rate": 1.4353264300212273e-06,
      "loss": 0.8238,
      "step": 5800
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2514103651046753,
      "learning_rate": 1.4338386531040176e-06,
      "loss": 0.755,
      "step": 5801
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2897248268127441,
      "learning_rate": 1.4323515185824204e-06,
      "loss": 0.6772,
      "step": 5802
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2240180969238281,
      "learning_rate": 1.4308650267243213e-06,
      "loss": 0.7609,
      "step": 5803
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5125356912612915,
      "learning_rate": 1.4293791777974936e-06,
      "loss": 0.8281,
      "step": 5804
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2238847017288208,
      "learning_rate": 1.4278939720695923e-06,
      "loss": 0.7798,
      "step": 5805
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2995526790618896,
      "learning_rate": 1.4264094098081554e-06,
      "loss": 0.7479,
      "step": 5806
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.184018850326538,
      "learning_rate": 1.4249254912806092e-06,
      "loss": 0.7445,
      "step": 5807
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2325108051300049,
      "learning_rate": 1.4234422167542604e-06,
      "loss": 0.7722,
      "step": 5808
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3555915355682373,
      "learning_rate": 1.4219595864962993e-06,
      "loss": 0.7311,
      "step": 5809
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2570210695266724,
      "learning_rate": 1.420477600773804e-06,
      "loss": 0.8331,
      "step": 5810
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.307297945022583,
      "learning_rate": 1.4189962598537338e-06,
      "loss": 0.7863,
      "step": 5811
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.178979754447937,
      "learning_rate": 1.4175155640029298e-06,
      "loss": 0.7504,
      "step": 5812
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2898280620574951,
      "learning_rate": 1.4160355134881227e-06,
      "loss": 0.7753,
      "step": 5813
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2134121656417847,
      "learning_rate": 1.4145561085759201e-06,
      "loss": 0.7544,
      "step": 5814
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.273449182510376,
      "learning_rate": 1.41307734953282e-06,
      "loss": 0.7605,
      "step": 5815
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1992524862289429,
      "learning_rate": 1.4115992366251996e-06,
      "loss": 0.7445,
      "step": 5816
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2527893781661987,
      "learning_rate": 1.4101217701193188e-06,
      "loss": 0.8214,
      "step": 5817
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2429112195968628,
      "learning_rate": 1.4086449502813265e-06,
      "loss": 0.7688,
      "step": 5818
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2563687562942505,
      "learning_rate": 1.4071687773772485e-06,
      "loss": 0.726,
      "step": 5819
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6316378116607666,
      "learning_rate": 1.405693251673001e-06,
      "loss": 1.016,
      "step": 5820
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6017209887504578,
      "learning_rate": 1.4042183734343779e-06,
      "loss": 0.9996,
      "step": 5821
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3186335563659668,
      "learning_rate": 1.4027441429270566e-06,
      "loss": 0.813,
      "step": 5822
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.290588140487671,
      "learning_rate": 1.4012705604166038e-06,
      "loss": 0.7792,
      "step": 5823
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2421995401382446,
      "learning_rate": 1.3997976261684615e-06,
      "loss": 0.7829,
      "step": 5824
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2470922470092773,
      "learning_rate": 1.3983253404479625e-06,
      "loss": 0.7433,
      "step": 5825
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3801188468933105,
      "learning_rate": 1.3968537035203173e-06,
      "loss": 0.8029,
      "step": 5826
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3616156578063965,
      "learning_rate": 1.3953827156506194e-06,
      "loss": 0.7799,
      "step": 5827
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5854344367980957,
      "learning_rate": 1.3939123771038515e-06,
      "loss": 1.0316,
      "step": 5828
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1764551401138306,
      "learning_rate": 1.392442688144871e-06,
      "loss": 0.7602,
      "step": 5829
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1766631603240967,
      "learning_rate": 1.3909736490384258e-06,
      "loss": 0.7558,
      "step": 5830
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.21517014503479,
      "learning_rate": 1.3895052600491421e-06,
      "loss": 0.8261,
      "step": 5831
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.368865966796875,
      "learning_rate": 1.3880375214415281e-06,
      "loss": 0.7956,
      "step": 5832
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2667652368545532,
      "learning_rate": 1.3865704334799807e-06,
      "loss": 0.7584,
      "step": 5833
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.587931215763092,
      "learning_rate": 1.3851039964287733e-06,
      "loss": 0.994,
      "step": 5834
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3582805395126343,
      "learning_rate": 1.3836382105520663e-06,
      "loss": 0.7868,
      "step": 5835
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2393195629119873,
      "learning_rate": 1.3821730761139008e-06,
      "loss": 0.7119,
      "step": 5836
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6027059555053711,
      "learning_rate": 1.3807085933781988e-06,
      "loss": 1.0068,
      "step": 5837
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2463430166244507,
      "learning_rate": 1.3792447626087702e-06,
      "loss": 0.8247,
      "step": 5838
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5985650420188904,
      "learning_rate": 1.3777815840693026e-06,
      "loss": 1.0088,
      "step": 5839
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.242266058921814,
      "learning_rate": 1.3763190580233687e-06,
      "loss": 0.7329,
      "step": 5840
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.252046823501587,
      "learning_rate": 1.37485718473442e-06,
      "loss": 0.834,
      "step": 5841
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1983906030654907,
      "learning_rate": 1.3733959644657964e-06,
      "loss": 0.7197,
      "step": 5842
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2620935440063477,
      "learning_rate": 1.3719353974807159e-06,
      "loss": 0.7626,
      "step": 5843
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.314968466758728,
      "learning_rate": 1.370475484042278e-06,
      "loss": 0.7867,
      "step": 5844
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5865930914878845,
      "learning_rate": 1.3690162244134697e-06,
      "loss": 1.0274,
      "step": 5845
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2549858093261719,
      "learning_rate": 1.3675576188571548e-06,
      "loss": 0.771,
      "step": 5846
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2406854629516602,
      "learning_rate": 1.36609966763608e-06,
      "loss": 0.7572,
      "step": 5847
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1990805864334106,
      "learning_rate": 1.3646423710128786e-06,
      "loss": 0.7423,
      "step": 5848
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5769811868667603,
      "learning_rate": 1.363185729250061e-06,
      "loss": 1.0213,
      "step": 5849
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2317051887512207,
      "learning_rate": 1.3617297426100208e-06,
      "loss": 0.7804,
      "step": 5850
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5957320332527161,
      "learning_rate": 1.360274411355036e-06,
      "loss": 1.0077,
      "step": 5851
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1919655799865723,
      "learning_rate": 1.3588197357472628e-06,
      "loss": 0.756,
      "step": 5852
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2748295068740845,
      "learning_rate": 1.3573657160487436e-06,
      "loss": 0.7295,
      "step": 5853
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3254317045211792,
      "learning_rate": 1.3559123525213997e-06,
      "loss": 0.7819,
      "step": 5854
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2075042724609375,
      "learning_rate": 1.354459645427032e-06,
      "loss": 0.7551,
      "step": 5855
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2652699947357178,
      "learning_rate": 1.3530075950273303e-06,
      "loss": 0.8259,
      "step": 5856
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5929760336875916,
      "learning_rate": 1.3515562015838585e-06,
      "loss": 1.0263,
      "step": 5857
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2449631690979004,
      "learning_rate": 1.3501054653580675e-06,
      "loss": 0.7822,
      "step": 5858
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3283485174179077,
      "learning_rate": 1.3486553866112872e-06,
      "loss": 0.8001,
      "step": 5859
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2467658519744873,
      "learning_rate": 1.347205965604728e-06,
      "loss": 0.7477,
      "step": 5860
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.282764196395874,
      "learning_rate": 1.3457572025994864e-06,
      "loss": 0.8309,
      "step": 5861
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.274813175201416,
      "learning_rate": 1.3443090978565344e-06,
      "loss": 0.7877,
      "step": 5862
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2562625408172607,
      "learning_rate": 1.3428616516367316e-06,
      "loss": 0.7718,
      "step": 5863
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2939400672912598,
      "learning_rate": 1.3414148642008134e-06,
      "loss": 0.7504,
      "step": 5864
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2387818098068237,
      "learning_rate": 1.3399687358093987e-06,
      "loss": 0.7486,
      "step": 5865
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1988240480422974,
      "learning_rate": 1.3385232667229904e-06,
      "loss": 0.7262,
      "step": 5866
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2074470520019531,
      "learning_rate": 1.3370784572019663e-06,
      "loss": 0.7887,
      "step": 5867
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.601064920425415,
      "learning_rate": 1.3356343075065937e-06,
      "loss": 1.0247,
      "step": 5868
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2632828950881958,
      "learning_rate": 1.3341908178970141e-06,
      "loss": 0.7827,
      "step": 5869
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6107639670372009,
      "learning_rate": 1.3327479886332522e-06,
      "loss": 1.0198,
      "step": 5870
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2783292531967163,
      "learning_rate": 1.3313058199752154e-06,
      "loss": 0.7297,
      "step": 5871
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3226466178894043,
      "learning_rate": 1.329864312182691e-06,
      "loss": 0.7967,
      "step": 5872
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.196691870689392,
      "learning_rate": 1.3284234655153443e-06,
      "loss": 0.7405,
      "step": 5873
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2469594478607178,
      "learning_rate": 1.326983280232728e-06,
      "loss": 0.7529,
      "step": 5874
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.863938808441162,
      "learning_rate": 1.3255437565942692e-06,
      "loss": 0.7685,
      "step": 5875
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.26416015625,
      "learning_rate": 1.3241048948592805e-06,
      "loss": 0.7288,
      "step": 5876
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2154086828231812,
      "learning_rate": 1.3226666952869532e-06,
      "loss": 0.735,
      "step": 5877
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5877107381820679,
      "learning_rate": 1.3212291581363567e-06,
      "loss": 1.0249,
      "step": 5878
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2102124691009521,
      "learning_rate": 1.3197922836664473e-06,
      "loss": 0.7958,
      "step": 5879
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2675890922546387,
      "learning_rate": 1.3183560721360577e-06,
      "loss": 0.8142,
      "step": 5880
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2890366315841675,
      "learning_rate": 1.316920523803899e-06,
      "loss": 0.8181,
      "step": 5881
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2915743589401245,
      "learning_rate": 1.3154856389285702e-06,
      "loss": 0.8271,
      "step": 5882
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2676018476486206,
      "learning_rate": 1.3140514177685438e-06,
      "loss": 0.7407,
      "step": 5883
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1840311288833618,
      "learning_rate": 1.3126178605821754e-06,
      "loss": 0.7622,
      "step": 5884
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3284869194030762,
      "learning_rate": 1.3111849676277e-06,
      "loss": 0.7307,
      "step": 5885
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2621445655822754,
      "learning_rate": 1.3097527391632365e-06,
      "loss": 0.787,
      "step": 5886
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2648125886917114,
      "learning_rate": 1.3083211754467801e-06,
      "loss": 0.7139,
      "step": 5887
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2770894765853882,
      "learning_rate": 1.3068902767362057e-06,
      "loss": 0.7496,
      "step": 5888
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5796307325363159,
      "learning_rate": 1.3054600432892734e-06,
      "loss": 1.0021,
      "step": 5889
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2431846857070923,
      "learning_rate": 1.304030475363618e-06,
      "loss": 0.7519,
      "step": 5890
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.26494300365448,
      "learning_rate": 1.3026015732167586e-06,
      "loss": 0.7489,
      "step": 5891
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1818901300430298,
      "learning_rate": 1.301173337106092e-06,
      "loss": 0.7243,
      "step": 5892
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2385988235473633,
      "learning_rate": 1.299745767288894e-06,
      "loss": 0.7959,
      "step": 5893
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2297186851501465,
      "learning_rate": 1.2983188640223248e-06,
      "loss": 0.7357,
      "step": 5894
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1882661581039429,
      "learning_rate": 1.296892627563418e-06,
      "loss": 0.7614,
      "step": 5895
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.244681715965271,
      "learning_rate": 1.2954670581690947e-06,
      "loss": 0.7629,
      "step": 5896
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8240935802459717,
      "learning_rate": 1.2940421560961501e-06,
      "loss": 0.7299,
      "step": 5897
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2562545537948608,
      "learning_rate": 1.2926179216012596e-06,
      "loss": 0.7988,
      "step": 5898
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2515828609466553,
      "learning_rate": 1.2911943549409827e-06,
      "loss": 0.7854,
      "step": 5899
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2500617504119873,
      "learning_rate": 1.289771456371754e-06,
      "loss": 0.7877,
      "step": 5900
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2534674406051636,
      "learning_rate": 1.2883492261498876e-06,
      "loss": 0.7352,
      "step": 5901
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2515137195587158,
      "learning_rate": 1.286927664531582e-06,
      "loss": 0.7426,
      "step": 5902
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2510874271392822,
      "learning_rate": 1.2855067717729103e-06,
      "loss": 0.7805,
      "step": 5903
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6139270663261414,
      "learning_rate": 1.2840865481298288e-06,
      "loss": 1.0237,
      "step": 5904
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2682602405548096,
      "learning_rate": 1.2826669938581704e-06,
      "loss": 0.8395,
      "step": 5905
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2423068284988403,
      "learning_rate": 1.2812481092136476e-06,
      "loss": 0.773,
      "step": 5906
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5982129573822021,
      "learning_rate": 1.2798298944518555e-06,
      "loss": 0.9942,
      "step": 5907
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.276837706565857,
      "learning_rate": 1.278412349828264e-06,
      "loss": 0.8049,
      "step": 5908
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2231166362762451,
      "learning_rate": 1.2769954755982268e-06,
      "loss": 0.7619,
      "step": 5909
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.267663598060608,
      "learning_rate": 1.2755792720169736e-06,
      "loss": 0.7731,
      "step": 5910
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2553287744522095,
      "learning_rate": 1.2741637393396122e-06,
      "loss": 0.7934,
      "step": 5911
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2945387363433838,
      "learning_rate": 1.2727488778211356e-06,
      "loss": 0.756,
      "step": 5912
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2628566026687622,
      "learning_rate": 1.271334687716408e-06,
      "loss": 0.8009,
      "step": 5913
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.331648588180542,
      "learning_rate": 1.2699211692801794e-06,
      "loss": 0.7668,
      "step": 5914
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5812333822250366,
      "learning_rate": 1.2685083227670757e-06,
      "loss": 1.001,
      "step": 5915
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2800406217575073,
      "learning_rate": 1.2670961484315996e-06,
      "loss": 0.7376,
      "step": 5916
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2063730955123901,
      "learning_rate": 1.265684646528138e-06,
      "loss": 0.8166,
      "step": 5917
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3056235313415527,
      "learning_rate": 1.2642738173109531e-06,
      "loss": 0.7848,
      "step": 5918
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5690938234329224,
      "learning_rate": 1.262863661034185e-06,
      "loss": 1.019,
      "step": 5919
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.564055860042572,
      "learning_rate": 1.2614541779518563e-06,
      "loss": 0.9963,
      "step": 5920
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2386395931243896,
      "learning_rate": 1.2600453683178659e-06,
      "loss": 0.7903,
      "step": 5921
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2802183628082275,
      "learning_rate": 1.2586372323859896e-06,
      "loss": 0.7488,
      "step": 5922
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.321629524230957,
      "learning_rate": 1.2572297704098873e-06,
      "loss": 0.7859,
      "step": 5923
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2765295505523682,
      "learning_rate": 1.2558229826430922e-06,
      "loss": 0.6984,
      "step": 5924
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3328198194503784,
      "learning_rate": 1.2544168693390191e-06,
      "loss": 0.7522,
      "step": 5925
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2922062873840332,
      "learning_rate": 1.2530114307509577e-06,
      "loss": 0.728,
      "step": 5926
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2637810707092285,
      "learning_rate": 1.251606667132082e-06,
      "loss": 0.825,
      "step": 5927
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2938339710235596,
      "learning_rate": 1.25020257873544e-06,
      "loss": 0.7762,
      "step": 5928
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5926676392555237,
      "learning_rate": 1.248799165813957e-06,
      "loss": 1.0198,
      "step": 5929
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2714976072311401,
      "learning_rate": 1.2473964286204426e-06,
      "loss": 0.7738,
      "step": 5930
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6079380512237549,
      "learning_rate": 1.245994367407577e-06,
      "loss": 1.0182,
      "step": 5931
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2417138814926147,
      "learning_rate": 1.2445929824279262e-06,
      "loss": 0.7561,
      "step": 5932
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2213342189788818,
      "learning_rate": 1.2431922739339293e-06,
      "loss": 0.7307,
      "step": 5933
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5945495367050171,
      "learning_rate": 1.2417922421779027e-06,
      "loss": 1.0199,
      "step": 5934
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2574489116668701,
      "learning_rate": 1.240392887412047e-06,
      "loss": 0.7636,
      "step": 5935
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.206933856010437,
      "learning_rate": 1.2389942098884334e-06,
      "loss": 0.7757,
      "step": 5936
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5940548181533813,
      "learning_rate": 1.2375962098590178e-06,
      "loss": 1.0626,
      "step": 5937
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.208079218864441,
      "learning_rate": 1.2361988875756292e-06,
      "loss": 0.7386,
      "step": 5938
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2291145324707031,
      "learning_rate": 1.2348022432899754e-06,
      "loss": 0.7335,
      "step": 5939
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3481673002243042,
      "learning_rate": 1.2334062772536454e-06,
      "loss": 0.8173,
      "step": 5940
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3167362213134766,
      "learning_rate": 1.232010989718101e-06,
      "loss": 0.8118,
      "step": 5941
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3003424406051636,
      "learning_rate": 1.2306163809346867e-06,
      "loss": 0.8001,
      "step": 5942
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5917959213256836,
      "learning_rate": 1.2292224511546219e-06,
      "loss": 1.0253,
      "step": 5943
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5196220874786377,
      "learning_rate": 1.2278292006290015e-06,
      "loss": 0.7887,
      "step": 5944
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2534762620925903,
      "learning_rate": 1.2264366296088043e-06,
      "loss": 0.7303,
      "step": 5945
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2910912036895752,
      "learning_rate": 1.2250447383448798e-06,
      "loss": 0.8486,
      "step": 5946
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2634797096252441,
      "learning_rate": 1.2236535270879617e-06,
      "loss": 0.8151,
      "step": 5947
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5818764567375183,
      "learning_rate": 1.2222629960886561e-06,
      "loss": 1.0153,
      "step": 5948
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1708300113677979,
      "learning_rate": 1.2208731455974466e-06,
      "loss": 0.741,
      "step": 5949
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2576998472213745,
      "learning_rate": 1.2194839758646999e-06,
      "loss": 0.7403,
      "step": 5950
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.237318515777588,
      "learning_rate": 1.2180954871406514e-06,
      "loss": 0.7623,
      "step": 5951
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2569022178649902,
      "learning_rate": 1.216707679675423e-06,
      "loss": 0.718,
      "step": 5952
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2304062843322754,
      "learning_rate": 1.2153205537190077e-06,
      "loss": 0.7952,
      "step": 5953
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2082984447479248,
      "learning_rate": 1.2139341095212754e-06,
      "loss": 0.7356,
      "step": 5954
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3688241243362427,
      "learning_rate": 1.2125483473319782e-06,
      "loss": 0.7861,
      "step": 5955
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2327347993850708,
      "learning_rate": 1.2111632674007412e-06,
      "loss": 0.7145,
      "step": 5956
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2428884506225586,
      "learning_rate": 1.2097788699770658e-06,
      "loss": 0.7797,
      "step": 5957
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.188982605934143,
      "learning_rate": 1.2083951553103362e-06,
      "loss": 0.7292,
      "step": 5958
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3425737619400024,
      "learning_rate": 1.2070121236498078e-06,
      "loss": 0.8273,
      "step": 5959
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2856277227401733,
      "learning_rate": 1.2056297752446139e-06,
      "loss": 0.787,
      "step": 5960
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1931555271148682,
      "learning_rate": 1.2042481103437686e-06,
      "loss": 0.7525,
      "step": 5961
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2530128955841064,
      "learning_rate": 1.202867129196158e-06,
      "loss": 0.8068,
      "step": 5962
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2384523153305054,
      "learning_rate": 1.2014868320505468e-06,
      "loss": 0.7738,
      "step": 5963
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2948594093322754,
      "learning_rate": 1.200107219155579e-06,
      "loss": 0.7669,
      "step": 5964
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3117775917053223,
      "learning_rate": 1.198728290759772e-06,
      "loss": 0.7789,
      "step": 5965
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5888672471046448,
      "learning_rate": 1.19735004711152e-06,
      "loss": 1.0215,
      "step": 5966
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2808135747909546,
      "learning_rate": 1.1959724884590957e-06,
      "loss": 0.7221,
      "step": 5967
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1784956455230713,
      "learning_rate": 1.1945956150506482e-06,
      "loss": 0.7355,
      "step": 5968
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.340888500213623,
      "learning_rate": 1.1932194271342006e-06,
      "loss": 0.7791,
      "step": 5969
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2602298259735107,
      "learning_rate": 1.191843924957658e-06,
      "loss": 0.8406,
      "step": 5970
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3050870895385742,
      "learning_rate": 1.1904691087687964e-06,
      "loss": 0.7907,
      "step": 5971
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.189098834991455,
      "learning_rate": 1.189094978815269e-06,
      "loss": 0.765,
      "step": 5972
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2369831800460815,
      "learning_rate": 1.1877215353446092e-06,
      "loss": 0.7788,
      "step": 5973
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2167439460754395,
      "learning_rate": 1.1863487786042215e-06,
      "loss": 0.7413,
      "step": 5974
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6104679107666016,
      "learning_rate": 1.1849767088413926e-06,
      "loss": 0.9784,
      "step": 5975
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2462605237960815,
      "learning_rate": 1.1836053263032804e-06,
      "loss": 0.7874,
      "step": 5976
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1694631576538086,
      "learning_rate": 1.1822346312369198e-06,
      "loss": 0.7516,
      "step": 5977
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5917034149169922,
      "learning_rate": 1.1808646238892257e-06,
      "loss": 1.0256,
      "step": 5978
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.234242558479309,
      "learning_rate": 1.1794953045069835e-06,
      "loss": 0.7353,
      "step": 5979
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5745362639427185,
      "learning_rate": 1.1781266733368602e-06,
      "loss": 1.0239,
      "step": 5980
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.182173728942871,
      "learning_rate": 1.1767587306253952e-06,
      "loss": 0.7423,
      "step": 5981
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2613998651504517,
      "learning_rate": 1.1753914766190028e-06,
      "loss": 0.7801,
      "step": 5982
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.589870274066925,
      "learning_rate": 1.174024911563978e-06,
      "loss": 1.0237,
      "step": 5983
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2510019540786743,
      "learning_rate": 1.172659035706487e-06,
      "loss": 0.7824,
      "step": 5984
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5904861092567444,
      "learning_rate": 1.1712938492925757e-06,
      "loss": 0.9887,
      "step": 5985
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.24056875705719,
      "learning_rate": 1.1699293525681632e-06,
      "loss": 0.7533,
      "step": 5986
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.585634171962738,
      "learning_rate": 1.1685655457790435e-06,
      "loss": 1.0143,
      "step": 5987
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2293832302093506,
      "learning_rate": 1.1672024291708906e-06,
      "loss": 0.7477,
      "step": 5988
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.233406901359558,
      "learning_rate": 1.1658400029892502e-06,
      "loss": 0.7718,
      "step": 5989
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1795316934585571,
      "learning_rate": 1.1644782674795435e-06,
      "loss": 0.7303,
      "step": 5990
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.245106816291809,
      "learning_rate": 1.1631172228870706e-06,
      "loss": 0.8114,
      "step": 5991
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5716187953948975,
      "learning_rate": 1.161756869457004e-06,
      "loss": 1.0048,
      "step": 5992
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5710926651954651,
      "learning_rate": 1.160397207434395e-06,
      "loss": 1.0277,
      "step": 5993
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.27336585521698,
      "learning_rate": 1.1590382370641663e-06,
      "loss": 0.7429,
      "step": 5994
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3584057092666626,
      "learning_rate": 1.1576799585911175e-06,
      "loss": 0.8424,
      "step": 5995
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2037463188171387,
      "learning_rate": 1.1563223722599266e-06,
      "loss": 0.7548,
      "step": 5996
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.306707739830017,
      "learning_rate": 1.154965478315141e-06,
      "loss": 0.7639,
      "step": 5997
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2317514419555664,
      "learning_rate": 1.1536092770011898e-06,
      "loss": 0.7914,
      "step": 5998
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2464995384216309,
      "learning_rate": 1.1522537685623735e-06,
      "loss": 0.7505,
      "step": 5999
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2895756959915161,
      "learning_rate": 1.1508989532428683e-06,
      "loss": 0.7814,
      "step": 6000
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2346208095550537,
      "learning_rate": 1.1495448312867235e-06,
      "loss": 0.7397,
      "step": 6001
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2839123010635376,
      "learning_rate": 1.1481914029378698e-06,
      "loss": 0.7948,
      "step": 6002
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.482804298400879,
      "learning_rate": 1.1468386684401074e-06,
      "loss": 0.7702,
      "step": 6003
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.286879062652588,
      "learning_rate": 1.1454866280371107e-06,
      "loss": 0.7636,
      "step": 6004
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2205536365509033,
      "learning_rate": 1.1441352819724355e-06,
      "loss": 0.7487,
      "step": 6005
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.213417887687683,
      "learning_rate": 1.1427846304895063e-06,
      "loss": 0.78,
      "step": 6006
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2470735311508179,
      "learning_rate": 1.1414346738316234e-06,
      "loss": 0.7865,
      "step": 6007
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1860934495925903,
      "learning_rate": 1.1400854122419659e-06,
      "loss": 0.7269,
      "step": 6008
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2501194477081299,
      "learning_rate": 1.138736845963584e-06,
      "loss": 0.79,
      "step": 6009
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1580541133880615,
      "learning_rate": 1.1373889752394018e-06,
      "loss": 0.756,
      "step": 6010
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6048233509063721,
      "learning_rate": 1.1360418003122231e-06,
      "loss": 1.0311,
      "step": 6011
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.24027681350708,
      "learning_rate": 1.1346953214247197e-06,
      "loss": 0.796,
      "step": 6012
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2151273488998413,
      "learning_rate": 1.1333495388194454e-06,
      "loss": 0.7273,
      "step": 6013
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5810208916664124,
      "learning_rate": 1.1320044527388223e-06,
      "loss": 1.0036,
      "step": 6014
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2928338050842285,
      "learning_rate": 1.1306600634251486e-06,
      "loss": 0.7944,
      "step": 6015
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5654194951057434,
      "learning_rate": 1.1293163711206e-06,
      "loss": 0.9768,
      "step": 6016
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5829293131828308,
      "learning_rate": 1.1279733760672235e-06,
      "loss": 1.0134,
      "step": 6017
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2259575128555298,
      "learning_rate": 1.1266310785069406e-06,
      "loss": 0.7208,
      "step": 6018
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4677814245224,
      "learning_rate": 1.1252894786815494e-06,
      "loss": 0.7668,
      "step": 6019
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2141531705856323,
      "learning_rate": 1.1239485768327195e-06,
      "loss": 0.7327,
      "step": 6020
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5738003253936768,
      "learning_rate": 1.1226083732019981e-06,
      "loss": 1.03,
      "step": 6021
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2837172746658325,
      "learning_rate": 1.1212688680308042e-06,
      "loss": 0.7939,
      "step": 6022
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5754179358482361,
      "learning_rate": 1.1199300615604291e-06,
      "loss": 1.0053,
      "step": 6023
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.313095211982727,
      "learning_rate": 1.118591954032044e-06,
      "loss": 0.8062,
      "step": 6024
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2474380731582642,
      "learning_rate": 1.1172545456866873e-06,
      "loss": 0.8211,
      "step": 6025
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2616050243377686,
      "learning_rate": 1.115917836765279e-06,
      "loss": 0.7807,
      "step": 6026
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2574464082717896,
      "learning_rate": 1.1145818275086068e-06,
      "loss": 0.7689,
      "step": 6027
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5774536728858948,
      "learning_rate": 1.1132465181573332e-06,
      "loss": 1.0314,
      "step": 6028
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.280050277709961,
      "learning_rate": 1.1119119089519997e-06,
      "loss": 0.8135,
      "step": 6029
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2760396003723145,
      "learning_rate": 1.1105780001330147e-06,
      "loss": 0.7506,
      "step": 6030
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4828758239746094,
      "learning_rate": 1.1092447919406662e-06,
      "loss": 0.7943,
      "step": 6031
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.272443413734436,
      "learning_rate": 1.107912284615113e-06,
      "loss": 0.791,
      "step": 6032
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5626997351646423,
      "learning_rate": 1.1065804783963856e-06,
      "loss": 0.9875,
      "step": 6033
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2639647722244263,
      "learning_rate": 1.1052493735243952e-06,
      "loss": 0.7639,
      "step": 6034
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2864513397216797,
      "learning_rate": 1.1039189702389187e-06,
      "loss": 0.7477,
      "step": 6035
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5696431398391724,
      "learning_rate": 1.102589268779613e-06,
      "loss": 1.008,
      "step": 6036
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.235195279121399,
      "learning_rate": 1.1012602693860047e-06,
      "loss": 0.7699,
      "step": 6037
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2430665493011475,
      "learning_rate": 1.0999319722974928e-06,
      "loss": 0.7871,
      "step": 6038
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1733163595199585,
      "learning_rate": 1.0986043777533556e-06,
      "loss": 0.747,
      "step": 6039
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.225063443183899,
      "learning_rate": 1.0972774859927398e-06,
      "loss": 0.837,
      "step": 6040
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5943523645401001,
      "learning_rate": 1.095951297254665e-06,
      "loss": 0.9943,
      "step": 6041
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5878843665122986,
      "learning_rate": 1.0946258117780294e-06,
      "loss": 1.0381,
      "step": 6042
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2432698011398315,
      "learning_rate": 1.0933010298016e-06,
      "loss": 0.7706,
      "step": 6043
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2977514266967773,
      "learning_rate": 1.0919769515640183e-06,
      "loss": 0.777,
      "step": 6044
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.580843448638916,
      "learning_rate": 1.0906535773037969e-06,
      "loss": 1.0152,
      "step": 6045
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5844226479530334,
      "learning_rate": 1.0893309072593273e-06,
      "loss": 1.0145,
      "step": 6046
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5817234516143799,
      "learning_rate": 1.0880089416688682e-06,
      "loss": 1.023,
      "step": 6047
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2609734535217285,
      "learning_rate": 1.086687680770554e-06,
      "loss": 0.7559,
      "step": 6048
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1998498439788818,
      "learning_rate": 1.0853671248023934e-06,
      "loss": 0.7697,
      "step": 6049
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.262387752532959,
      "learning_rate": 1.0840472740022657e-06,
      "loss": 0.8188,
      "step": 6050
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3059308528900146,
      "learning_rate": 1.082728128607922e-06,
      "loss": 0.7005,
      "step": 6051
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5741438269615173,
      "learning_rate": 1.0814096888569931e-06,
      "loss": 0.9923,
      "step": 6052
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5602342486381531,
      "learning_rate": 1.0800919549869737e-06,
      "loss": 0.9971,
      "step": 6053
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3986696004867554,
      "learning_rate": 1.0787749272352381e-06,
      "loss": 0.7403,
      "step": 6054
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5738655924797058,
      "learning_rate": 1.0774586058390314e-06,
      "loss": 1.0113,
      "step": 6055
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.570705771446228,
      "learning_rate": 1.0761429910354682e-06,
      "loss": 0.9991,
      "step": 6056
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.370617151260376,
      "learning_rate": 1.074828083061542e-06,
      "loss": 0.7611,
      "step": 6057
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5880171656608582,
      "learning_rate": 1.0735138821541125e-06,
      "loss": 1.0101,
      "step": 6058
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1805975437164307,
      "learning_rate": 1.0722003885499184e-06,
      "loss": 0.7275,
      "step": 6059
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.222243070602417,
      "learning_rate": 1.0708876024855664e-06,
      "loss": 0.754,
      "step": 6060
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.592744767665863,
      "learning_rate": 1.0695755241975347e-06,
      "loss": 1.0413,
      "step": 6061
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.254606008529663,
      "learning_rate": 1.0682641539221805e-06,
      "loss": 0.7604,
      "step": 6062
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5922152996063232,
      "learning_rate": 1.0669534918957258e-06,
      "loss": 0.9851,
      "step": 6063
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.221655011177063,
      "learning_rate": 1.0656435383542724e-06,
      "loss": 0.753,
      "step": 6064
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5840696692466736,
      "learning_rate": 1.0643342935337876e-06,
      "loss": 1.0103,
      "step": 6065
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2113956212997437,
      "learning_rate": 1.0630257576701142e-06,
      "loss": 0.7897,
      "step": 6066
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2421727180480957,
      "learning_rate": 1.0617179309989688e-06,
      "loss": 0.8448,
      "step": 6067
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4497134685516357,
      "learning_rate": 1.060410813755937e-06,
      "loss": 0.8,
      "step": 6068
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.334606409072876,
      "learning_rate": 1.0591044061764804e-06,
      "loss": 0.7523,
      "step": 6069
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5721689462661743,
      "learning_rate": 1.0577987084959296e-06,
      "loss": 1.0412,
      "step": 6070
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5671486258506775,
      "learning_rate": 1.0564937209494863e-06,
      "loss": 0.9783,
      "step": 6071
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1810177564620972,
      "learning_rate": 1.0551894437722294e-06,
      "loss": 0.7733,
      "step": 6072
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5842350721359253,
      "learning_rate": 1.0538858771991056e-06,
      "loss": 0.986,
      "step": 6073
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2172130346298218,
      "learning_rate": 1.052583021464933e-06,
      "loss": 0.7318,
      "step": 6074
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2040005922317505,
      "learning_rate": 1.0512808768044063e-06,
      "loss": 0.7717,
      "step": 6075
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5764269232749939,
      "learning_rate": 1.0499794434520867e-06,
      "loss": 1.0214,
      "step": 6076
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2042211294174194,
      "learning_rate": 1.048678721642412e-06,
      "loss": 0.7467,
      "step": 6077
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2645466327667236,
      "learning_rate": 1.0473787116096885e-06,
      "loss": 0.748,
      "step": 6078
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5730329751968384,
      "learning_rate": 1.0460794135880941e-06,
      "loss": 0.9878,
      "step": 6079
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.285593867301941,
      "learning_rate": 1.0447808278116828e-06,
      "loss": 0.7632,
      "step": 6080
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5773106217384338,
      "learning_rate": 1.0434829545143748e-06,
      "loss": 1.0005,
      "step": 6081
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2318154573440552,
      "learning_rate": 1.042185793929964e-06,
      "loss": 0.7477,
      "step": 6082
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5933464169502258,
      "learning_rate": 1.0408893462921187e-06,
      "loss": 1.0256,
      "step": 6083
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3249648809432983,
      "learning_rate": 1.0395936118343747e-06,
      "loss": 0.8015,
      "step": 6084
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2546439170837402,
      "learning_rate": 1.0382985907901416e-06,
      "loss": 0.8155,
      "step": 6085
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2757360935211182,
      "learning_rate": 1.037004283392698e-06,
      "loss": 0.7946,
      "step": 6086
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2356650829315186,
      "learning_rate": 1.035710689875199e-06,
      "loss": 0.7811,
      "step": 6087
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2544080018997192,
      "learning_rate": 1.034417810470666e-06,
      "loss": 0.7704,
      "step": 6088
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.253559947013855,
      "learning_rate": 1.0331256454119927e-06,
      "loss": 0.7425,
      "step": 6089
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3421374559402466,
      "learning_rate": 1.0318341949319478e-06,
      "loss": 0.8256,
      "step": 6090
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.241698980331421,
      "learning_rate": 1.0305434592631658e-06,
      "loss": 0.7615,
      "step": 6091
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2771583795547485,
      "learning_rate": 1.0292534386381581e-06,
      "loss": 0.7793,
      "step": 6092
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1744177341461182,
      "learning_rate": 1.0279641332893032e-06,
      "loss": 0.7743,
      "step": 6093
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5825190544128418,
      "learning_rate": 1.0266755434488501e-06,
      "loss": 1.0022,
      "step": 6094
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5612584352493286,
      "learning_rate": 1.0253876693489234e-06,
      "loss": 1.0364,
      "step": 6095
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2098802328109741,
      "learning_rate": 1.024100511221514e-06,
      "loss": 0.7543,
      "step": 6096
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.580446720123291,
      "learning_rate": 1.022814069298489e-06,
      "loss": 1.0083,
      "step": 6097
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.757831335067749,
      "learning_rate": 1.0215283438115813e-06,
      "loss": 0.7627,
      "step": 6098
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5713570713996887,
      "learning_rate": 1.0202433349923957e-06,
      "loss": 1.0209,
      "step": 6099
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2689359188079834,
      "learning_rate": 1.0189590430724123e-06,
      "loss": 0.7887,
      "step": 6100
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2643638849258423,
      "learning_rate": 1.0176754682829759e-06,
      "loss": 0.8033,
      "step": 6101
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2798880338668823,
      "learning_rate": 1.016392610855308e-06,
      "loss": 0.8079,
      "step": 6102
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2640728950500488,
      "learning_rate": 1.0151104710204957e-06,
      "loss": 0.8139,
      "step": 6103
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2229266166687012,
      "learning_rate": 1.0138290490094992e-06,
      "loss": 0.7815,
      "step": 6104
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2438170909881592,
      "learning_rate": 1.0125483450531509e-06,
      "loss": 0.7519,
      "step": 6105
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2313755750656128,
      "learning_rate": 1.011268359382151e-06,
      "loss": 0.8097,
      "step": 6106
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2148154973983765,
      "learning_rate": 1.009989092227071e-06,
      "loss": 0.7486,
      "step": 6107
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2358647584915161,
      "learning_rate": 1.0087105438183554e-06,
      "loss": 0.7738,
      "step": 6108
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2185307741165161,
      "learning_rate": 1.007432714386315e-06,
      "loss": 0.7744,
      "step": 6109
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.177597165107727,
      "learning_rate": 1.0061556041611358e-06,
      "loss": 0.7032,
      "step": 6110
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2749850749969482,
      "learning_rate": 1.0048792133728713e-06,
      "loss": 0.6883,
      "step": 6111
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2622840404510498,
      "learning_rate": 1.003603542251444e-06,
      "loss": 0.7723,
      "step": 6112
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2614030838012695,
      "learning_rate": 1.0023285910266517e-06,
      "loss": 0.7712,
      "step": 6113
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2781574726104736,
      "learning_rate": 1.0010543599281569e-06,
      "loss": 0.7707,
      "step": 6114
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2703438997268677,
      "learning_rate": 9.997808491854977e-07,
      "loss": 0.724,
      "step": 6115
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2877440452575684,
      "learning_rate": 9.985080590280783e-07,
      "loss": 0.7439,
      "step": 6116
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6243894696235657,
      "learning_rate": 9.972359896851736e-07,
      "loss": 1.0199,
      "step": 6117
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.232998013496399,
      "learning_rate": 9.95964641385932e-07,
      "loss": 0.7582,
      "step": 6118
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2437834739685059,
      "learning_rate": 9.94694014359368e-07,
      "loss": 0.8292,
      "step": 6119
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.274243712425232,
      "learning_rate": 9.934241088343671e-07,
      "loss": 0.7839,
      "step": 6120
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2085055112838745,
      "learning_rate": 9.921549250396884e-07,
      "loss": 0.759,
      "step": 6121
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6040817499160767,
      "learning_rate": 9.90886463203956e-07,
      "loss": 0.9963,
      "step": 6122
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5983880162239075,
      "learning_rate": 9.896187235556653e-07,
      "loss": 1.0211,
      "step": 6123
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2290199995040894,
      "learning_rate": 9.88351706323185e-07,
      "loss": 0.7271,
      "step": 6124
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5816089510917664,
      "learning_rate": 9.8708541173475e-07,
      "loss": 1.0129,
      "step": 6125
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5792880058288574,
      "learning_rate": 9.85819840018466e-07,
      "loss": 0.9985,
      "step": 6126
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5658020973205566,
      "learning_rate": 9.845549914023067e-07,
      "loss": 1.0052,
      "step": 6127
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2409294843673706,
      "learning_rate": 9.832908661141204e-07,
      "loss": 0.7892,
      "step": 6128
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.594115138053894,
      "learning_rate": 9.820274643816197e-07,
      "loss": 1.0167,
      "step": 6129
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.237195611000061,
      "learning_rate": 9.80764786432392e-07,
      "loss": 0.744,
      "step": 6130
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2588070631027222,
      "learning_rate": 9.795028324938899e-07,
      "loss": 0.6981,
      "step": 6131
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2793253660202026,
      "learning_rate": 9.78241602793436e-07,
      "loss": 0.7986,
      "step": 6132
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.312507152557373,
      "learning_rate": 9.769810975582268e-07,
      "loss": 0.8294,
      "step": 6133
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1960704326629639,
      "learning_rate": 9.757213170153234e-07,
      "loss": 0.7097,
      "step": 6134
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2488648891448975,
      "learning_rate": 9.744622613916572e-07,
      "loss": 0.7711,
      "step": 6135
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3049644231796265,
      "learning_rate": 9.73203930914033e-07,
      "loss": 0.7731,
      "step": 6136
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2831465005874634,
      "learning_rate": 9.719463258091182e-07,
      "loss": 0.7311,
      "step": 6137
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5755881667137146,
      "learning_rate": 9.706894463034567e-07,
      "loss": 1.0077,
      "step": 6138
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.240189552307129,
      "learning_rate": 9.694332926234567e-07,
      "loss": 0.8354,
      "step": 6139
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3179783821105957,
      "learning_rate": 9.681778649953966e-07,
      "loss": 0.7874,
      "step": 6140
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3246604204177856,
      "learning_rate": 9.66923163645427e-07,
      "loss": 0.7423,
      "step": 6141
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.252906322479248,
      "learning_rate": 9.656691887995618e-07,
      "loss": 0.7907,
      "step": 6142
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2641472816467285,
      "learning_rate": 9.644159406836912e-07,
      "loss": 0.7789,
      "step": 6143
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5777742266654968,
      "learning_rate": 9.631634195235696e-07,
      "loss": 1.0121,
      "step": 6144
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.297264575958252,
      "learning_rate": 9.619116255448196e-07,
      "loss": 0.8155,
      "step": 6145
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2150232791900635,
      "learning_rate": 9.606605589729385e-07,
      "loss": 0.7531,
      "step": 6146
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3643358945846558,
      "learning_rate": 9.594102200332855e-07,
      "loss": 0.7999,
      "step": 6147
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2535529136657715,
      "learning_rate": 9.581606089510952e-07,
      "loss": 0.7337,
      "step": 6148
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.238136887550354,
      "learning_rate": 9.569117259514666e-07,
      "loss": 0.7561,
      "step": 6149
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5733403563499451,
      "learning_rate": 9.556635712593682e-07,
      "loss": 1.0087,
      "step": 6150
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2714749574661255,
      "learning_rate": 9.544161450996402e-07,
      "loss": 0.747,
      "step": 6151
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2754385471343994,
      "learning_rate": 9.531694476969872e-07,
      "loss": 0.7545,
      "step": 6152
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2750009298324585,
      "learning_rate": 9.51923479275988e-07,
      "loss": 0.8122,
      "step": 6153
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2323163747787476,
      "learning_rate": 9.506782400610842e-07,
      "loss": 0.8237,
      "step": 6154
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2567646503448486,
      "learning_rate": 9.494337302765883e-07,
      "loss": 0.769,
      "step": 6155
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5766451954841614,
      "learning_rate": 9.481899501466846e-07,
      "loss": 1.0091,
      "step": 6156
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3527729511260986,
      "learning_rate": 9.469468998954206e-07,
      "loss": 0.781,
      "step": 6157
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5690637826919556,
      "learning_rate": 9.457045797467173e-07,
      "loss": 1.0056,
      "step": 6158
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2058274745941162,
      "learning_rate": 9.444629899243607e-07,
      "loss": 0.7219,
      "step": 6159
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.216794490814209,
      "learning_rate": 9.43222130652005e-07,
      "loss": 0.7519,
      "step": 6160
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2174261808395386,
      "learning_rate": 9.419820021531767e-07,
      "loss": 0.7772,
      "step": 6161
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2254085540771484,
      "learning_rate": 9.407426046512669e-07,
      "loss": 0.764,
      "step": 6162
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2334624528884888,
      "learning_rate": 9.395039383695359e-07,
      "loss": 0.7532,
      "step": 6163
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1784956455230713,
      "learning_rate": 9.382660035311125e-07,
      "loss": 0.7695,
      "step": 6164
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2340115308761597,
      "learning_rate": 9.37028800358995e-07,
      "loss": 0.7015,
      "step": 6165
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5848467946052551,
      "learning_rate": 9.357923290760484e-07,
      "loss": 1.0089,
      "step": 6166
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6041578650474548,
      "learning_rate": 9.345565899050058e-07,
      "loss": 1.0205,
      "step": 6167
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3501543998718262,
      "learning_rate": 9.333215830684678e-07,
      "loss": 0.7869,
      "step": 6168
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2389349937438965,
      "learning_rate": 9.320873087889071e-07,
      "loss": 0.7874,
      "step": 6169
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2402023077011108,
      "learning_rate": 9.308537672886581e-07,
      "loss": 0.8004,
      "step": 6170
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.262833833694458,
      "learning_rate": 9.296209587899291e-07,
      "loss": 0.7727,
      "step": 6171
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2643550634384155,
      "learning_rate": 9.283888835147936e-07,
      "loss": 0.7122,
      "step": 6172
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2529100179672241,
      "learning_rate": 9.271575416851908e-07,
      "loss": 0.7427,
      "step": 6173
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5603122115135193,
      "learning_rate": 9.259269335229338e-07,
      "loss": 0.991,
      "step": 6174
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3624049425125122,
      "learning_rate": 9.246970592496968e-07,
      "loss": 0.7969,
      "step": 6175
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2480878829956055,
      "learning_rate": 9.234679190870277e-07,
      "loss": 0.8141,
      "step": 6176
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5923499464988708,
      "learning_rate": 9.222395132563378e-07,
      "loss": 1.018,
      "step": 6177
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.229502558708191,
      "learning_rate": 9.210118419789071e-07,
      "loss": 0.7871,
      "step": 6178
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2735346555709839,
      "learning_rate": 9.197849054758861e-07,
      "loss": 0.8191,
      "step": 6179
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2685277462005615,
      "learning_rate": 9.185587039682886e-07,
      "loss": 0.7647,
      "step": 6180
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2183332443237305,
      "learning_rate": 9.173332376770006e-07,
      "loss": 0.7891,
      "step": 6181
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2863651514053345,
      "learning_rate": 9.16108506822771e-07,
      "loss": 0.8124,
      "step": 6182
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2178469896316528,
      "learning_rate": 9.148845116262184e-07,
      "loss": 0.7718,
      "step": 6183
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5856855511665344,
      "learning_rate": 9.136612523078314e-07,
      "loss": 1.0009,
      "step": 6184
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2546467781066895,
      "learning_rate": 9.124387290879605e-07,
      "loss": 0.8081,
      "step": 6185
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1928123235702515,
      "learning_rate": 9.112169421868289e-07,
      "loss": 0.762,
      "step": 6186
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4265172481536865,
      "learning_rate": 9.099958918245249e-07,
      "loss": 0.8117,
      "step": 6187
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2607570886611938,
      "learning_rate": 9.087755782210017e-07,
      "loss": 0.8366,
      "step": 6188
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.258047103881836,
      "learning_rate": 9.075560015960855e-07,
      "loss": 0.7521,
      "step": 6189
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2882225513458252,
      "learning_rate": 9.06337162169465e-07,
      "loss": 0.7215,
      "step": 6190
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2458018064498901,
      "learning_rate": 9.051190601606968e-07,
      "loss": 0.7075,
      "step": 6191
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.344533085823059,
      "learning_rate": 9.039016957892071e-07,
      "loss": 0.7905,
      "step": 6192
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2268275022506714,
      "learning_rate": 9.026850692742856e-07,
      "loss": 0.7476,
      "step": 6193
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2766854763031006,
      "learning_rate": 9.014691808350934e-07,
      "loss": 0.7672,
      "step": 6194
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2496181726455688,
      "learning_rate": 9.002540306906549e-07,
      "loss": 0.8051,
      "step": 6195
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.262463092803955,
      "learning_rate": 8.990396190598622e-07,
      "loss": 0.7086,
      "step": 6196
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2038543224334717,
      "learning_rate": 8.978259461614769e-07,
      "loss": 0.7933,
      "step": 6197
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5975993275642395,
      "learning_rate": 8.966130122141242e-07,
      "loss": 1.0112,
      "step": 6198
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2532886266708374,
      "learning_rate": 8.954008174362994e-07,
      "loss": 0.7992,
      "step": 6199
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2052299976348877,
      "learning_rate": 8.94189362046362e-07,
      "loss": 0.8246,
      "step": 6200
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5866808891296387,
      "learning_rate": 8.929786462625379e-07,
      "loss": 1.0215,
      "step": 6201
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2437057495117188,
      "learning_rate": 8.917686703029233e-07,
      "loss": 0.7528,
      "step": 6202
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2275382280349731,
      "learning_rate": 8.905594343854785e-07,
      "loss": 0.7608,
      "step": 6203
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2433485984802246,
      "learning_rate": 8.893509387280302e-07,
      "loss": 0.8286,
      "step": 6204
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1834847927093506,
      "learning_rate": 8.881431835482718e-07,
      "loss": 0.6836,
      "step": 6205
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3725941181182861,
      "learning_rate": 8.869361690637668e-07,
      "loss": 0.8152,
      "step": 6206
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2647581100463867,
      "learning_rate": 8.857298954919402e-07,
      "loss": 0.8019,
      "step": 6207
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2404444217681885,
      "learning_rate": 8.84524363050085e-07,
      "loss": 0.7229,
      "step": 6208
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2473735809326172,
      "learning_rate": 8.833195719553645e-07,
      "loss": 0.7599,
      "step": 6209
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2365760803222656,
      "learning_rate": 8.821155224248035e-07,
      "loss": 0.7574,
      "step": 6210
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.292245626449585,
      "learning_rate": 8.809122146752946e-07,
      "loss": 0.7696,
      "step": 6211
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2835503816604614,
      "learning_rate": 8.797096489235995e-07,
      "loss": 0.7616,
      "step": 6212
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2565333843231201,
      "learning_rate": 8.785078253863422e-07,
      "loss": 0.8224,
      "step": 6213
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5707911849021912,
      "learning_rate": 8.773067442800164e-07,
      "loss": 1.019,
      "step": 6214
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2732210159301758,
      "learning_rate": 8.7610640582098e-07,
      "loss": 0.7868,
      "step": 6215
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5971196889877319,
      "learning_rate": 8.749068102254559e-07,
      "loss": 0.9763,
      "step": 6216
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5660398006439209,
      "learning_rate": 8.737079577095386e-07,
      "loss": 1.0231,
      "step": 6217
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6059620976448059,
      "learning_rate": 8.725098484891825e-07,
      "loss": 0.9899,
      "step": 6218
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5769706964492798,
      "learning_rate": 8.713124827802105e-07,
      "loss": 1.0058,
      "step": 6219
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2364919185638428,
      "learning_rate": 8.70115860798314e-07,
      "loss": 0.8086,
      "step": 6220
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2719018459320068,
      "learning_rate": 8.689199827590455e-07,
      "loss": 0.7912,
      "step": 6221
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5839936137199402,
      "learning_rate": 8.67724848877829e-07,
      "loss": 0.9864,
      "step": 6222
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2662992477416992,
      "learning_rate": 8.665304593699503e-07,
      "loss": 0.7781,
      "step": 6223
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2184665203094482,
      "learning_rate": 8.653368144505614e-07,
      "loss": 0.7744,
      "step": 6224
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.289175033569336,
      "learning_rate": 8.64143914334683e-07,
      "loss": 0.869,
      "step": 6225
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3183491230010986,
      "learning_rate": 8.629517592371983e-07,
      "loss": 0.8252,
      "step": 6226
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5813025236129761,
      "learning_rate": 8.617603493728604e-07,
      "loss": 1.022,
      "step": 6227
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2373607158660889,
      "learning_rate": 8.605696849562834e-07,
      "loss": 0.7101,
      "step": 6228
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1389763355255127,
      "learning_rate": 8.593797662019493e-07,
      "loss": 0.7149,
      "step": 6229
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2681350708007812,
      "learning_rate": 8.581905933242068e-07,
      "loss": 0.7731,
      "step": 6230
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5894423127174377,
      "learning_rate": 8.570021665372686e-07,
      "loss": 1.0113,
      "step": 6231
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3108205795288086,
      "learning_rate": 8.558144860552147e-07,
      "loss": 0.7739,
      "step": 6232
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.271172285079956,
      "learning_rate": 8.546275520919894e-07,
      "loss": 0.7364,
      "step": 6233
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2232272624969482,
      "learning_rate": 8.534413648614009e-07,
      "loss": 0.7707,
      "step": 6234
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2538989782333374,
      "learning_rate": 8.522559245771272e-07,
      "loss": 0.7934,
      "step": 6235
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2162898778915405,
      "learning_rate": 8.51071231452707e-07,
      "loss": 0.8521,
      "step": 6236
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.281671166419983,
      "learning_rate": 8.498872857015495e-07,
      "loss": 0.7528,
      "step": 6237
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2719167470932007,
      "learning_rate": 8.487040875369251e-07,
      "loss": 0.7876,
      "step": 6238
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5858556032180786,
      "learning_rate": 8.475216371719691e-07,
      "loss": 1.0237,
      "step": 6239
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5776805281639099,
      "learning_rate": 8.463399348196871e-07,
      "loss": 0.9796,
      "step": 6240
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1938292980194092,
      "learning_rate": 8.45158980692945e-07,
      "loss": 0.7245,
      "step": 6241
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2441564798355103,
      "learning_rate": 8.439787750044742e-07,
      "loss": 0.7959,
      "step": 6242
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2130519151687622,
      "learning_rate": 8.427993179668764e-07,
      "loss": 0.7307,
      "step": 6243
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2391656637191772,
      "learning_rate": 8.416206097926122e-07,
      "loss": 0.7651,
      "step": 6244
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1812318563461304,
      "learning_rate": 8.404426506940111e-07,
      "loss": 0.7448,
      "step": 6245
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2042617797851562,
      "learning_rate": 8.39265440883264e-07,
      "loss": 0.7691,
      "step": 6246
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2520818710327148,
      "learning_rate": 8.380889805724329e-07,
      "loss": 0.7814,
      "step": 6247
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2604591846466064,
      "learning_rate": 8.369132699734394e-07,
      "loss": 0.7558,
      "step": 6248
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2618006467819214,
      "learning_rate": 8.357383092980703e-07,
      "loss": 0.752,
      "step": 6249
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2305225133895874,
      "learning_rate": 8.345640987579817e-07,
      "loss": 0.7634,
      "step": 6250
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2327359914779663,
      "learning_rate": 8.333906385646906e-07,
      "loss": 0.6788,
      "step": 6251
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4132529497146606,
      "learning_rate": 8.322179289295784e-07,
      "loss": 0.7716,
      "step": 6252
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.213515043258667,
      "learning_rate": 8.310459700638957e-07,
      "loss": 0.7901,
      "step": 6253
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3056501150131226,
      "learning_rate": 8.298747621787517e-07,
      "loss": 0.8028,
      "step": 6254
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2133870124816895,
      "learning_rate": 8.287043054851268e-07,
      "loss": 0.7721,
      "step": 6255
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2318037748336792,
      "learning_rate": 8.275346001938617e-07,
      "loss": 0.7986,
      "step": 6256
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5747051239013672,
      "learning_rate": 8.263656465156611e-07,
      "loss": 0.9966,
      "step": 6257
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2313789129257202,
      "learning_rate": 8.251974446610994e-07,
      "loss": 0.7505,
      "step": 6258
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3119581937789917,
      "learning_rate": 8.240299948406094e-07,
      "loss": 0.7894,
      "step": 6259
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2435972690582275,
      "learning_rate": 8.22863297264494e-07,
      "loss": 0.7341,
      "step": 6260
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2844724655151367,
      "learning_rate": 8.216973521429161e-07,
      "loss": 0.7851,
      "step": 6261
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2854686975479126,
      "learning_rate": 8.205321596859045e-07,
      "loss": 0.7648,
      "step": 6262
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.295350193977356,
      "learning_rate": 8.193677201033551e-07,
      "loss": 0.8181,
      "step": 6263
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5790804624557495,
      "learning_rate": 8.182040336050234e-07,
      "loss": 0.9944,
      "step": 6264
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2229784727096558,
      "learning_rate": 8.170411004005341e-07,
      "loss": 0.8006,
      "step": 6265
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.196417212486267,
      "learning_rate": 8.158789206993728e-07,
      "loss": 0.7454,
      "step": 6266
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6036532521247864,
      "learning_rate": 8.147174947108888e-07,
      "loss": 0.9889,
      "step": 6267
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5755158066749573,
      "learning_rate": 8.135568226442997e-07,
      "loss": 0.9869,
      "step": 6268
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5787367820739746,
      "learning_rate": 8.123969047086828e-07,
      "loss": 1.0213,
      "step": 6269
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2243324518203735,
      "learning_rate": 8.11237741112984e-07,
      "loss": 0.7909,
      "step": 6270
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5812839865684509,
      "learning_rate": 8.100793320660095e-07,
      "loss": 1.0297,
      "step": 6271
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2713812589645386,
      "learning_rate": 8.089216777764297e-07,
      "loss": 0.7862,
      "step": 6272
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2652333974838257,
      "learning_rate": 8.077647784527831e-07,
      "loss": 0.7509,
      "step": 6273
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2113920450210571,
      "learning_rate": 8.066086343034663e-07,
      "loss": 0.7963,
      "step": 6274
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5713061690330505,
      "learning_rate": 8.054532455367459e-07,
      "loss": 1.0036,
      "step": 6275
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.260642409324646,
      "learning_rate": 8.042986123607488e-07,
      "loss": 0.7777,
      "step": 6276
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2486991882324219,
      "learning_rate": 8.031447349834643e-07,
      "loss": 0.7656,
      "step": 6277
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.212592601776123,
      "learning_rate": 8.019916136127509e-07,
      "loss": 0.7514,
      "step": 6278
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3620580434799194,
      "learning_rate": 8.00839248456326e-07,
      "loss": 0.7832,
      "step": 6279
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5657126307487488,
      "learning_rate": 7.996876397217723e-07,
      "loss": 0.9997,
      "step": 6280
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2765262126922607,
      "learning_rate": 7.985367876165378e-07,
      "loss": 0.8028,
      "step": 6281
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5812993049621582,
      "learning_rate": 7.973866923479323e-07,
      "loss": 1.0013,
      "step": 6282
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5782318711280823,
      "learning_rate": 7.962373541231283e-07,
      "loss": 1.0156,
      "step": 6283
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2103127241134644,
      "learning_rate": 7.950887731491669e-07,
      "loss": 0.809,
      "step": 6284
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5249820947647095,
      "learning_rate": 7.939409496329465e-07,
      "loss": 0.7437,
      "step": 6285
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.257939338684082,
      "learning_rate": 7.927938837812338e-07,
      "loss": 0.8223,
      "step": 6286
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.579791784286499,
      "learning_rate": 7.916475758006543e-07,
      "loss": 1.0003,
      "step": 6287
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5821329951286316,
      "learning_rate": 7.905020258977031e-07,
      "loss": 1.0123,
      "step": 6288
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5104490518569946,
      "learning_rate": 7.893572342787342e-07,
      "loss": 0.7454,
      "step": 6289
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2576497793197632,
      "learning_rate": 7.882132011499654e-07,
      "loss": 0.7442,
      "step": 6290
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5857726335525513,
      "learning_rate": 7.870699267174809e-07,
      "loss": 1.0168,
      "step": 6291
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5778794288635254,
      "learning_rate": 7.859274111872228e-07,
      "loss": 1.0195,
      "step": 6292
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5638222694396973,
      "learning_rate": 7.84785654765004e-07,
      "loss": 0.9848,
      "step": 6293
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2372852563858032,
      "learning_rate": 7.836446576564932e-07,
      "loss": 0.8171,
      "step": 6294
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.581695020198822,
      "learning_rate": 7.82504420067226e-07,
      "loss": 1.0105,
      "step": 6295
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1641072034835815,
      "learning_rate": 7.813649422026026e-07,
      "loss": 0.7211,
      "step": 6296
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2172366380691528,
      "learning_rate": 7.802262242678815e-07,
      "loss": 0.7319,
      "step": 6297
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5757519006729126,
      "learning_rate": 7.790882664681909e-07,
      "loss": 1.0062,
      "step": 6298
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2248553037643433,
      "learning_rate": 7.779510690085157e-07,
      "loss": 0.779,
      "step": 6299
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1979612112045288,
      "learning_rate": 7.76814632093707e-07,
      "loss": 0.7446,
      "step": 6300
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5613813400268555,
      "learning_rate": 7.7567895592848e-07,
      "loss": 1.0133,
      "step": 6301
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2535418272018433,
      "learning_rate": 7.745440407174088e-07,
      "loss": 0.7017,
      "step": 6302
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5696107149124146,
      "learning_rate": 7.734098866649358e-07,
      "loss": 1.0039,
      "step": 6303
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5621254444122314,
      "learning_rate": 7.722764939753624e-07,
      "loss": 1.0071,
      "step": 6304
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6085169911384583,
      "learning_rate": 7.711438628528523e-07,
      "loss": 1.0198,
      "step": 6305
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2582478523254395,
      "learning_rate": 7.700119935014367e-07,
      "loss": 0.8128,
      "step": 6306
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2843117713928223,
      "learning_rate": 7.688808861250047e-07,
      "loss": 0.7411,
      "step": 6307
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2363402843475342,
      "learning_rate": 7.677505409273084e-07,
      "loss": 0.7407,
      "step": 6308
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.565606415271759,
      "learning_rate": 7.666209581119677e-07,
      "loss": 0.9915,
      "step": 6309
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2962517738342285,
      "learning_rate": 7.654921378824582e-07,
      "loss": 0.7994,
      "step": 6310
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5681315064430237,
      "learning_rate": 7.643640804421248e-07,
      "loss": 0.9959,
      "step": 6311
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5760475993156433,
      "learning_rate": 7.632367859941698e-07,
      "loss": 0.9968,
      "step": 6312
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2928664684295654,
      "learning_rate": 7.621102547416592e-07,
      "loss": 0.7456,
      "step": 6313
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3532929420471191,
      "learning_rate": 7.609844868875249e-07,
      "loss": 0.8201,
      "step": 6314
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5610832571983337,
      "learning_rate": 7.598594826345557e-07,
      "loss": 1.0194,
      "step": 6315
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1946433782577515,
      "learning_rate": 7.587352421854088e-07,
      "loss": 0.79,
      "step": 6316
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2436795234680176,
      "learning_rate": 7.576117657425997e-07,
      "loss": 0.7387,
      "step": 6317
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3125436305999756,
      "learning_rate": 7.564890535085056e-07,
      "loss": 0.7743,
      "step": 6318
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2459627389907837,
      "learning_rate": 7.553671056853712e-07,
      "loss": 0.7834,
      "step": 6319
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2510021924972534,
      "learning_rate": 7.542459224752973e-07,
      "loss": 0.7506,
      "step": 6320
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5633784532546997,
      "learning_rate": 7.53125504080252e-07,
      "loss": 0.993,
      "step": 6321
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5818407535552979,
      "learning_rate": 7.520058507020627e-07,
      "loss": 1.0111,
      "step": 6322
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.257135272026062,
      "learning_rate": 7.508869625424198e-07,
      "loss": 0.7916,
      "step": 6323
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.271890640258789,
      "learning_rate": 7.497688398028746e-07,
      "loss": 0.764,
      "step": 6324
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.237687110900879,
      "learning_rate": 7.48651482684844e-07,
      "loss": 0.8018,
      "step": 6325
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5719300508499146,
      "learning_rate": 7.47534891389603e-07,
      "loss": 0.9776,
      "step": 6326
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2597534656524658,
      "learning_rate": 7.464190661182918e-07,
      "loss": 0.7732,
      "step": 6327
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.203161597251892,
      "learning_rate": 7.453040070719087e-07,
      "loss": 0.7445,
      "step": 6328
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2744677066802979,
      "learning_rate": 7.441897144513188e-07,
      "loss": 0.7301,
      "step": 6329
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.305294394493103,
      "learning_rate": 7.430761884572457e-07,
      "loss": 0.7856,
      "step": 6330
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2986432313919067,
      "learning_rate": 7.419634292902772e-07,
      "loss": 0.7937,
      "step": 6331
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2767337560653687,
      "learning_rate": 7.408514371508613e-07,
      "loss": 0.7751,
      "step": 6332
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.214685082435608,
      "learning_rate": 7.397402122393066e-07,
      "loss": 0.7541,
      "step": 6333
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2247445583343506,
      "learning_rate": 7.386297547557874e-07,
      "loss": 0.7241,
      "step": 6334
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5924798250198364,
      "learning_rate": 7.375200649003372e-07,
      "loss": 1.0131,
      "step": 6335
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5728161931037903,
      "learning_rate": 7.364111428728504e-07,
      "loss": 1.0128,
      "step": 6336
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.229867696762085,
      "learning_rate": 7.353029888730856e-07,
      "loss": 0.7584,
      "step": 6337
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5899827480316162,
      "learning_rate": 7.341956031006597e-07,
      "loss": 1.0132,
      "step": 6338
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.227046251296997,
      "learning_rate": 7.330889857550566e-07,
      "loss": 0.7376,
      "step": 6339
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3237175941467285,
      "learning_rate": 7.319831370356162e-07,
      "loss": 0.7883,
      "step": 6340
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.218469500541687,
      "learning_rate": 7.308780571415408e-07,
      "loss": 0.7614,
      "step": 6341
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2100774049758911,
      "learning_rate": 7.297737462718985e-07,
      "loss": 0.7169,
      "step": 6342
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2045890092849731,
      "learning_rate": 7.286702046256139e-07,
      "loss": 0.8055,
      "step": 6343
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2601563930511475,
      "learning_rate": 7.275674324014764e-07,
      "loss": 0.7785,
      "step": 6344
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3154020309448242,
      "learning_rate": 7.264654297981355e-07,
      "loss": 0.7097,
      "step": 6345
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2853033542633057,
      "learning_rate": 7.253641970140995e-07,
      "loss": 0.7448,
      "step": 6346
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2182868719100952,
      "learning_rate": 7.242637342477437e-07,
      "loss": 0.7585,
      "step": 6347
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3342427015304565,
      "learning_rate": 7.231640416972996e-07,
      "loss": 0.7352,
      "step": 6348
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2163881063461304,
      "learning_rate": 7.220651195608635e-07,
      "loss": 0.717,
      "step": 6349
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2093534469604492,
      "learning_rate": 7.209669680363901e-07,
      "loss": 0.7354,
      "step": 6350
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2632659673690796,
      "learning_rate": 7.19869587321696e-07,
      "loss": 0.7918,
      "step": 6351
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2730766534805298,
      "learning_rate": 7.187729776144608e-07,
      "loss": 0.7324,
      "step": 6352
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2929044961929321,
      "learning_rate": 7.176771391122228e-07,
      "loss": 0.7934,
      "step": 6353
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5660268068313599,
      "learning_rate": 7.16582072012384e-07,
      "loss": 1.0069,
      "step": 6354
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.336453914642334,
      "learning_rate": 7.154877765122048e-07,
      "loss": 0.822,
      "step": 6355
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1900736093521118,
      "learning_rate": 7.143942528088066e-07,
      "loss": 0.7195,
      "step": 6356
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3364237546920776,
      "learning_rate": 7.13301501099175e-07,
      "loss": 0.7559,
      "step": 6357
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3165518045425415,
      "learning_rate": 7.122095215801522e-07,
      "loss": 0.7729,
      "step": 6358
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.601372241973877,
      "learning_rate": 7.111183144484462e-07,
      "loss": 1.0118,
      "step": 6359
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5745439529418945,
      "learning_rate": 7.100278799006216e-07,
      "loss": 1.0043,
      "step": 6360
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.284464716911316,
      "learning_rate": 7.089382181331045e-07,
      "loss": 0.734,
      "step": 6361
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2354402542114258,
      "learning_rate": 7.078493293421845e-07,
      "loss": 0.794,
      "step": 6362
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2653223276138306,
      "learning_rate": 7.067612137240098e-07,
      "loss": 0.8015,
      "step": 6363
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2410345077514648,
      "learning_rate": 7.056738714745887e-07,
      "loss": 0.7105,
      "step": 6364
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2137384414672852,
      "learning_rate": 7.045873027897909e-07,
      "loss": 0.8055,
      "step": 6365
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2030478715896606,
      "learning_rate": 7.03501507865349e-07,
      "loss": 0.7789,
      "step": 6366
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2701289653778076,
      "learning_rate": 7.024164868968525e-07,
      "loss": 0.789,
      "step": 6367
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.574211061000824,
      "learning_rate": 7.01332240079754e-07,
      "loss": 1.0072,
      "step": 6368
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3424556255340576,
      "learning_rate": 7.002487676093645e-07,
      "loss": 0.8081,
      "step": 6369
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2393511533737183,
      "learning_rate": 6.991660696808594e-07,
      "loss": 0.8009,
      "step": 6370
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5867096781730652,
      "learning_rate": 6.98084146489269e-07,
      "loss": 0.9949,
      "step": 6371
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.243097186088562,
      "learning_rate": 6.970029982294902e-07,
      "loss": 0.8087,
      "step": 6372
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5732131004333496,
      "learning_rate": 6.959226250962759e-07,
      "loss": 1.0066,
      "step": 6373
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5742697715759277,
      "learning_rate": 6.948430272842388e-07,
      "loss": 1.0291,
      "step": 6374
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2657780647277832,
      "learning_rate": 6.937642049878568e-07,
      "loss": 0.8176,
      "step": 6375
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.183072566986084,
      "learning_rate": 6.926861584014627e-07,
      "loss": 0.7464,
      "step": 6376
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5833141207695007,
      "learning_rate": 6.916088877192545e-07,
      "loss": 1.0278,
      "step": 6377
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3138309717178345,
      "learning_rate": 6.905323931352864e-07,
      "loss": 0.8095,
      "step": 6378
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2728105783462524,
      "learning_rate": 6.894566748434734e-07,
      "loss": 0.8092,
      "step": 6379
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2398250102996826,
      "learning_rate": 6.883817330375931e-07,
      "loss": 0.7811,
      "step": 6380
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.214350700378418,
      "learning_rate": 6.873075679112806e-07,
      "loss": 0.7254,
      "step": 6381
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2333194017410278,
      "learning_rate": 6.862341796580335e-07,
      "loss": 0.7725,
      "step": 6382
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2580220699310303,
      "learning_rate": 6.851615684712077e-07,
      "loss": 0.7628,
      "step": 6383
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5666014552116394,
      "learning_rate": 6.840897345440173e-07,
      "loss": 0.9951,
      "step": 6384
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.223960041999817,
      "learning_rate": 6.830186780695425e-07,
      "loss": 0.7893,
      "step": 6385
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1950008869171143,
      "learning_rate": 6.819483992407166e-07,
      "loss": 0.7728,
      "step": 6386
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2357147932052612,
      "learning_rate": 6.808788982503379e-07,
      "loss": 0.7646,
      "step": 6387
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.206006407737732,
      "learning_rate": 6.798101752910613e-07,
      "loss": 0.6984,
      "step": 6388
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2314093112945557,
      "learning_rate": 6.787422305554015e-07,
      "loss": 0.7513,
      "step": 6389
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2608187198638916,
      "learning_rate": 6.776750642357372e-07,
      "loss": 0.7867,
      "step": 6390
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2236700057983398,
      "learning_rate": 6.766086765243013e-07,
      "loss": 0.7877,
      "step": 6391
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.288336992263794,
      "learning_rate": 6.755430676131919e-07,
      "loss": 0.7583,
      "step": 6392
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2583695650100708,
      "learning_rate": 6.744782376943615e-07,
      "loss": 0.7455,
      "step": 6393
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2994428873062134,
      "learning_rate": 6.73414186959625e-07,
      "loss": 0.7581,
      "step": 6394
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2940154075622559,
      "learning_rate": 6.723509156006585e-07,
      "loss": 0.7477,
      "step": 6395
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2793439626693726,
      "learning_rate": 6.712884238089945e-07,
      "loss": 0.746,
      "step": 6396
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5675020217895508,
      "learning_rate": 6.702267117760253e-07,
      "loss": 1.004,
      "step": 6397
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5754805207252502,
      "learning_rate": 6.691657796930074e-07,
      "loss": 0.9929,
      "step": 6398
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.420832633972168,
      "learning_rate": 6.681056277510495e-07,
      "loss": 0.7348,
      "step": 6399
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2523247003555298,
      "learning_rate": 6.670462561411262e-07,
      "loss": 0.7608,
      "step": 6400
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5921109914779663,
      "learning_rate": 6.659876650540687e-07,
      "loss": 0.7839,
      "step": 6401
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5820190906524658,
      "learning_rate": 6.64929854680566e-07,
      "loss": 1.02,
      "step": 6402
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.233698844909668,
      "learning_rate": 6.638728252111704e-07,
      "loss": 0.7682,
      "step": 6403
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.262176513671875,
      "learning_rate": 6.62816576836291e-07,
      "loss": 0.7472,
      "step": 6404
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5748112797737122,
      "learning_rate": 6.617611097461962e-07,
      "loss": 1.0109,
      "step": 6405
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.195081353187561,
      "learning_rate": 6.607064241310124e-07,
      "loss": 0.7379,
      "step": 6406
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5749443173408508,
      "learning_rate": 6.596525201807302e-07,
      "loss": 1.0093,
      "step": 6407
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3223142623901367,
      "learning_rate": 6.585993980851945e-07,
      "loss": 0.8717,
      "step": 6408
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3567867279052734,
      "learning_rate": 6.575470580341098e-07,
      "loss": 0.7275,
      "step": 6409
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2237929105758667,
      "learning_rate": 6.564955002170425e-07,
      "loss": 0.7482,
      "step": 6410
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2378875017166138,
      "learning_rate": 6.55444724823417e-07,
      "loss": 0.8409,
      "step": 6411
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2051478624343872,
      "learning_rate": 6.543947320425137e-07,
      "loss": 0.7825,
      "step": 6412
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5824782252311707,
      "learning_rate": 6.533455220634771e-07,
      "loss": 1.0092,
      "step": 6413
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3520275354385376,
      "learning_rate": 6.522970950753066e-07,
      "loss": 0.8586,
      "step": 6414
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2600843906402588,
      "learning_rate": 6.512494512668633e-07,
      "loss": 0.7471,
      "step": 6415
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5750939846038818,
      "learning_rate": 6.502025908268655e-07,
      "loss": 0.996,
      "step": 6416
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3270355463027954,
      "learning_rate": 6.491565139438899e-07,
      "loss": 0.7611,
      "step": 6417
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2415300607681274,
      "learning_rate": 6.48111220806375e-07,
      "loss": 0.8019,
      "step": 6418
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5811151266098022,
      "learning_rate": 6.470667116026141e-07,
      "loss": 1.0311,
      "step": 6419
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3026386499404907,
      "learning_rate": 6.460229865207635e-07,
      "loss": 0.7988,
      "step": 6420
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.26333749294281,
      "learning_rate": 6.449800457488354e-07,
      "loss": 0.7614,
      "step": 6421
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5912108421325684,
      "learning_rate": 6.439378894746995e-07,
      "loss": 1.0242,
      "step": 6422
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.237837791442871,
      "learning_rate": 6.428965178860891e-07,
      "loss": 0.7252,
      "step": 6423
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5531255006790161,
      "learning_rate": 6.418559311705919e-07,
      "loss": 1.013,
      "step": 6424
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.265287160873413,
      "learning_rate": 6.408161295156545e-07,
      "loss": 0.733,
      "step": 6425
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2158478498458862,
      "learning_rate": 6.397771131085845e-07,
      "loss": 0.7839,
      "step": 6426
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1802716255187988,
      "learning_rate": 6.387388821365453e-07,
      "loss": 0.7369,
      "step": 6427
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.297518253326416,
      "learning_rate": 6.37701436786562e-07,
      "loss": 0.7584,
      "step": 6428
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3160804510116577,
      "learning_rate": 6.366647772455154e-07,
      "loss": 0.8183,
      "step": 6429
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.565650224685669,
      "learning_rate": 6.35628903700144e-07,
      "loss": 0.9992,
      "step": 6430
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2441540956497192,
      "learning_rate": 6.345938163370491e-07,
      "loss": 0.7383,
      "step": 6431
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3189200162887573,
      "learning_rate": 6.335595153426854e-07,
      "loss": 0.7695,
      "step": 6432
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6493254899978638,
      "learning_rate": 6.325260009033707e-07,
      "loss": 0.7755,
      "step": 6433
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.315501093864441,
      "learning_rate": 6.314932732052764e-07,
      "loss": 0.7474,
      "step": 6434
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2538013458251953,
      "learning_rate": 6.304613324344344e-07,
      "loss": 0.758,
      "step": 6435
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.355974555015564,
      "learning_rate": 6.294301787767365e-07,
      "loss": 0.7428,
      "step": 6436
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.233171820640564,
      "learning_rate": 6.283998124179292e-07,
      "loss": 0.8158,
      "step": 6437
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2705127000808716,
      "learning_rate": 6.27370233543621e-07,
      "loss": 0.7311,
      "step": 6438
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2106101512908936,
      "learning_rate": 6.263414423392755e-07,
      "loss": 0.8198,
      "step": 6439
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5768383741378784,
      "learning_rate": 6.253134389902143e-07,
      "loss": 1.0134,
      "step": 6440
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2451270818710327,
      "learning_rate": 6.242862236816205e-07,
      "loss": 0.7812,
      "step": 6441
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2299306392669678,
      "learning_rate": 6.232597965985321e-07,
      "loss": 0.7646,
      "step": 6442
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5834564566612244,
      "learning_rate": 6.222341579258451e-07,
      "loss": 1.0424,
      "step": 6443
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5711174011230469,
      "learning_rate": 6.212093078483155e-07,
      "loss": 1.023,
      "step": 6444
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5794382095336914,
      "learning_rate": 6.201852465505564e-07,
      "loss": 1.0055,
      "step": 6445
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.280770182609558,
      "learning_rate": 6.19161974217038e-07,
      "loss": 0.7842,
      "step": 6446
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5738340616226196,
      "learning_rate": 6.181394910320876e-07,
      "loss": 0.9977,
      "step": 6447
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5770959854125977,
      "learning_rate": 6.171177971798942e-07,
      "loss": 1.0189,
      "step": 6448
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.274145483970642,
      "learning_rate": 6.16096892844501e-07,
      "loss": 0.8055,
      "step": 6449
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2647395133972168,
      "learning_rate": 6.150767782098088e-07,
      "loss": 0.7837,
      "step": 6450
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2350717782974243,
      "learning_rate": 6.1405745345958e-07,
      "loss": 0.7072,
      "step": 6451
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.258805513381958,
      "learning_rate": 6.130389187774305e-07,
      "loss": 0.7625,
      "step": 6452
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2154771089553833,
      "learning_rate": 6.12021174346834e-07,
      "loss": 0.7171,
      "step": 6453
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2360156774520874,
      "learning_rate": 6.110042203511268e-07,
      "loss": 0.8102,
      "step": 6454
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.35529363155365,
      "learning_rate": 6.099880569734967e-07,
      "loss": 0.7661,
      "step": 6455
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2275495529174805,
      "learning_rate": 6.08972684396994e-07,
      "loss": 0.7228,
      "step": 6456
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5823635458946228,
      "learning_rate": 6.079581028045223e-07,
      "loss": 0.9848,
      "step": 6457
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3409698009490967,
      "learning_rate": 6.069443123788449e-07,
      "loss": 0.8088,
      "step": 6458
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2248932123184204,
      "learning_rate": 6.059313133025841e-07,
      "loss": 0.7623,
      "step": 6459
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.586155354976654,
      "learning_rate": 6.049191057582155e-07,
      "loss": 1.0182,
      "step": 6460
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3263267278671265,
      "learning_rate": 6.039076899280766e-07,
      "loss": 0.7697,
      "step": 6461
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3014456033706665,
      "learning_rate": 6.028970659943601e-07,
      "loss": 0.7943,
      "step": 6462
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2467703819274902,
      "learning_rate": 6.018872341391141e-07,
      "loss": 0.735,
      "step": 6463
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.233271837234497,
      "learning_rate": 6.008781945442488e-07,
      "loss": 0.7349,
      "step": 6464
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2540370225906372,
      "learning_rate": 5.99869947391527e-07,
      "loss": 0.7994,
      "step": 6465
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2237956523895264,
      "learning_rate": 5.988624928625725e-07,
      "loss": 0.7429,
      "step": 6466
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2696397304534912,
      "learning_rate": 5.978558311388639e-07,
      "loss": 0.8009,
      "step": 6467
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.564990222454071,
      "learning_rate": 5.96849962401736e-07,
      "loss": 1.0045,
      "step": 6468
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5635812878608704,
      "learning_rate": 5.958448868323846e-07,
      "loss": 1.0127,
      "step": 6469
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5861135721206665,
      "learning_rate": 5.948406046118594e-07,
      "loss": 1.0157,
      "step": 6470
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3560283184051514,
      "learning_rate": 5.938371159210687e-07,
      "loss": 0.7879,
      "step": 6471
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.27101469039917,
      "learning_rate": 5.928344209407777e-07,
      "loss": 0.7394,
      "step": 6472
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1964601278305054,
      "learning_rate": 5.918325198516067e-07,
      "loss": 0.8019,
      "step": 6473
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2405264377593994,
      "learning_rate": 5.908314128340364e-07,
      "loss": 0.7833,
      "step": 6474
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2958906888961792,
      "learning_rate": 5.89831100068401e-07,
      "loss": 0.7371,
      "step": 6475
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.220310091972351,
      "learning_rate": 5.888315817348955e-07,
      "loss": 0.7955,
      "step": 6476
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2954535484313965,
      "learning_rate": 5.878328580135679e-07,
      "loss": 0.8101,
      "step": 6477
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.29776132106781,
      "learning_rate": 5.868349290843239e-07,
      "loss": 0.7622,
      "step": 6478
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.581624448299408,
      "learning_rate": 5.858377951269295e-07,
      "loss": 1.0165,
      "step": 6479
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2460458278656006,
      "learning_rate": 5.848414563210036e-07,
      "loss": 0.8204,
      "step": 6480
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2007702589035034,
      "learning_rate": 5.838459128460216e-07,
      "loss": 0.7976,
      "step": 6481
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2502366304397583,
      "learning_rate": 5.828511648813201e-07,
      "loss": 0.7743,
      "step": 6482
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5768781304359436,
      "learning_rate": 5.818572126060873e-07,
      "loss": 1.0446,
      "step": 6483
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.317731261253357,
      "learning_rate": 5.8086405619937e-07,
      "loss": 0.7742,
      "step": 6484
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.280988097190857,
      "learning_rate": 5.798716958400735e-07,
      "loss": 0.7577,
      "step": 6485
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2875726222991943,
      "learning_rate": 5.788801317069575e-07,
      "loss": 0.7831,
      "step": 6486
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.227494478225708,
      "learning_rate": 5.778893639786381e-07,
      "loss": 0.7876,
      "step": 6487
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2347631454467773,
      "learning_rate": 5.768993928335887e-07,
      "loss": 0.7324,
      "step": 6488
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2057714462280273,
      "learning_rate": 5.759102184501403e-07,
      "loss": 0.7061,
      "step": 6489
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.284531593322754,
      "learning_rate": 5.749218410064789e-07,
      "loss": 0.7785,
      "step": 6490
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5718621015548706,
      "learning_rate": 5.739342606806453e-07,
      "loss": 1.0114,
      "step": 6491
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2527151107788086,
      "learning_rate": 5.729474776505423e-07,
      "loss": 0.8094,
      "step": 6492
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.205153226852417,
      "learning_rate": 5.719614920939215e-07,
      "loss": 0.7939,
      "step": 6493
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2720767259597778,
      "learning_rate": 5.709763041883987e-07,
      "loss": 0.7011,
      "step": 6494
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.264554500579834,
      "learning_rate": 5.699919141114401e-07,
      "loss": 0.8088,
      "step": 6495
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5627804398536682,
      "learning_rate": 5.690083220403692e-07,
      "loss": 1.0018,
      "step": 6496
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.276000738143921,
      "learning_rate": 5.680255281523694e-07,
      "loss": 0.7426,
      "step": 6497
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2804818153381348,
      "learning_rate": 5.67043532624475e-07,
      "loss": 0.7797,
      "step": 6498
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3486486673355103,
      "learning_rate": 5.660623356335815e-07,
      "loss": 0.8022,
      "step": 6499
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2792710065841675,
      "learning_rate": 5.650819373564376e-07,
      "loss": 0.7869,
      "step": 6500
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.275658369064331,
      "learning_rate": 5.641023379696469e-07,
      "loss": 0.7655,
      "step": 6501
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5697952508926392,
      "learning_rate": 5.631235376496735e-07,
      "loss": 0.9971,
      "step": 6502
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2631275653839111,
      "learning_rate": 5.621455365728329e-07,
      "loss": 0.7704,
      "step": 6503
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2409659624099731,
      "learning_rate": 5.611683349153013e-07,
      "loss": 0.7606,
      "step": 6504
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.214661955833435,
      "learning_rate": 5.601919328531064e-07,
      "loss": 0.7804,
      "step": 6505
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5734946727752686,
      "learning_rate": 5.592163305621329e-07,
      "loss": 0.996,
      "step": 6506
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2328684329986572,
      "learning_rate": 5.582415282181253e-07,
      "loss": 0.7196,
      "step": 6507
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2688486576080322,
      "learning_rate": 5.572675259966775e-07,
      "loss": 0.7576,
      "step": 6508
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5797882080078125,
      "learning_rate": 5.562943240732466e-07,
      "loss": 1.015,
      "step": 6509
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1902248859405518,
      "learning_rate": 5.553219226231393e-07,
      "loss": 0.7306,
      "step": 6510
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1811941862106323,
      "learning_rate": 5.543503218215196e-07,
      "loss": 0.7522,
      "step": 6511
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5676934719085693,
      "learning_rate": 5.533795218434113e-07,
      "loss": 1.0209,
      "step": 6512
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2832444906234741,
      "learning_rate": 5.524095228636889e-07,
      "loss": 0.8404,
      "step": 6513
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2808082103729248,
      "learning_rate": 5.514403250570832e-07,
      "loss": 0.7995,
      "step": 6514
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5675670504570007,
      "learning_rate": 5.504719285981858e-07,
      "loss": 1.0104,
      "step": 6515
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2426213026046753,
      "learning_rate": 5.495043336614364e-07,
      "loss": 0.8185,
      "step": 6516
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2047683000564575,
      "learning_rate": 5.485375404211369e-07,
      "loss": 0.7394,
      "step": 6517
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.228071928024292,
      "learning_rate": 5.475715490514405e-07,
      "loss": 0.7885,
      "step": 6518
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5591835975646973,
      "learning_rate": 5.466063597263571e-07,
      "loss": 1.0128,
      "step": 6519
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2084403038024902,
      "learning_rate": 5.456419726197542e-07,
      "loss": 0.7216,
      "step": 6520
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2145061492919922,
      "learning_rate": 5.446783879053513e-07,
      "loss": 0.778,
      "step": 6521
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3752555847167969,
      "learning_rate": 5.437156057567267e-07,
      "loss": 0.7968,
      "step": 6522
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5654351711273193,
      "learning_rate": 5.427536263473115e-07,
      "loss": 0.9813,
      "step": 6523
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.166237711906433,
      "learning_rate": 5.417924498503935e-07,
      "loss": 0.8149,
      "step": 6524
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2519972324371338,
      "learning_rate": 5.408320764391146e-07,
      "loss": 0.7814,
      "step": 6525
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5709391832351685,
      "learning_rate": 5.398725062864751e-07,
      "loss": 1.0192,
      "step": 6526
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5560889840126038,
      "learning_rate": 5.389137395653271e-07,
      "loss": 1.0141,
      "step": 6527
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.261387825012207,
      "learning_rate": 5.3795577644838e-07,
      "loss": 0.813,
      "step": 6528
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.581031084060669,
      "learning_rate": 5.36998617108197e-07,
      "loss": 1.0079,
      "step": 6529
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5637762546539307,
      "learning_rate": 5.36042261717199e-07,
      "loss": 1.0012,
      "step": 6530
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.260305404663086,
      "learning_rate": 5.350867104476581e-07,
      "loss": 0.7827,
      "step": 6531
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2548034191131592,
      "learning_rate": 5.341319634717063e-07,
      "loss": 0.7826,
      "step": 6532
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2262802124023438,
      "learning_rate": 5.331780209613275e-07,
      "loss": 0.8019,
      "step": 6533
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3520357608795166,
      "learning_rate": 5.322248830883603e-07,
      "loss": 0.8238,
      "step": 6534
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2953659296035767,
      "learning_rate": 5.31272550024502e-07,
      "loss": 0.7732,
      "step": 6535
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2703219652175903,
      "learning_rate": 5.303210219412996e-07,
      "loss": 0.7126,
      "step": 6536
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2151844501495361,
      "learning_rate": 5.293702990101607e-07,
      "loss": 0.7827,
      "step": 6537
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.302179217338562,
      "learning_rate": 5.284203814023436e-07,
      "loss": 0.7956,
      "step": 6538
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2512210607528687,
      "learning_rate": 5.274712692889627e-07,
      "loss": 0.7614,
      "step": 6539
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5679236650466919,
      "learning_rate": 5.265229628409891e-07,
      "loss": 0.99,
      "step": 6540
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2869406938552856,
      "learning_rate": 5.255754622292464e-07,
      "loss": 0.8031,
      "step": 6541
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2145280838012695,
      "learning_rate": 5.246287676244138e-07,
      "loss": 0.7074,
      "step": 6542
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.233572244644165,
      "learning_rate": 5.236828791970266e-07,
      "loss": 0.7736,
      "step": 6543
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2170562744140625,
      "learning_rate": 5.227377971174718e-07,
      "loss": 0.7709,
      "step": 6544
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5755438804626465,
      "learning_rate": 5.217935215559961e-07,
      "loss": 0.9952,
      "step": 6545
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1776360273361206,
      "learning_rate": 5.208500526826954e-07,
      "loss": 0.697,
      "step": 6546
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2015173435211182,
      "learning_rate": 5.199073906675234e-07,
      "loss": 0.756,
      "step": 6547
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2407379150390625,
      "learning_rate": 5.189655356802886e-07,
      "loss": 0.8178,
      "step": 6548
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2821261882781982,
      "learning_rate": 5.18024487890652e-07,
      "loss": 0.8032,
      "step": 6549
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2858558893203735,
      "learning_rate": 5.170842474681326e-07,
      "loss": 0.7682,
      "step": 6550
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.214975118637085,
      "learning_rate": 5.161448145821013e-07,
      "loss": 0.7973,
      "step": 6551
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.57706618309021,
      "learning_rate": 5.152061894017823e-07,
      "loss": 1.0011,
      "step": 6552
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5708666443824768,
      "learning_rate": 5.142683720962588e-07,
      "loss": 1.0284,
      "step": 6553
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5904618501663208,
      "learning_rate": 5.133313628344638e-07,
      "loss": 1.0037,
      "step": 6554
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2195758819580078,
      "learning_rate": 5.123951617851891e-07,
      "loss": 0.7737,
      "step": 6555
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1492726802825928,
      "learning_rate": 5.11459769117077e-07,
      "loss": 0.7,
      "step": 6556
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5779885649681091,
      "learning_rate": 5.105251849986249e-07,
      "loss": 1.0129,
      "step": 6557
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.298802375793457,
      "learning_rate": 5.095914095981879e-07,
      "loss": 0.8451,
      "step": 6558
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3691399097442627,
      "learning_rate": 5.086584430839708e-07,
      "loss": 0.8228,
      "step": 6559
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2268351316452026,
      "learning_rate": 5.077262856240367e-07,
      "loss": 0.7107,
      "step": 6560
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.275433897972107,
      "learning_rate": 5.067949373863007e-07,
      "loss": 0.8083,
      "step": 6561
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.560110867023468,
      "learning_rate": 5.058643985385303e-07,
      "loss": 1.0091,
      "step": 6562
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2122267484664917,
      "learning_rate": 5.049346692483525e-07,
      "loss": 0.7913,
      "step": 6563
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5784462094306946,
      "learning_rate": 5.040057496832446e-07,
      "loss": 1.018,
      "step": 6564
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5795724391937256,
      "learning_rate": 5.03077640010538e-07,
      "loss": 1.043,
      "step": 6565
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5710545182228088,
      "learning_rate": 5.021503403974187e-07,
      "loss": 1.0014,
      "step": 6566
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2173755168914795,
      "learning_rate": 5.01223851010929e-07,
      "loss": 0.7837,
      "step": 6567
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5572419166564941,
      "learning_rate": 5.002981720179622e-07,
      "loss": 0.9993,
      "step": 6568
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.578493058681488,
      "learning_rate": 4.993733035852666e-07,
      "loss": 1.0111,
      "step": 6569
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2358276844024658,
      "learning_rate": 4.984492458794454e-07,
      "loss": 0.7824,
      "step": 6570
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.29184091091156,
      "learning_rate": 4.975259990669551e-07,
      "loss": 0.7678,
      "step": 6571
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2487049102783203,
      "learning_rate": 4.966035633141047e-07,
      "loss": 0.7734,
      "step": 6572
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2505913972854614,
      "learning_rate": 4.956819387870604e-07,
      "loss": 0.7565,
      "step": 6573
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.285399317741394,
      "learning_rate": 4.947611256518392e-07,
      "loss": 0.7573,
      "step": 6574
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2704254388809204,
      "learning_rate": 4.93841124074313e-07,
      "loss": 0.8159,
      "step": 6575
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2991161346435547,
      "learning_rate": 4.929219342202091e-07,
      "loss": 0.7646,
      "step": 6576
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3942068815231323,
      "learning_rate": 4.920035562551045e-07,
      "loss": 0.8154,
      "step": 6577
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2983121871948242,
      "learning_rate": 4.910859903444354e-07,
      "loss": 0.7933,
      "step": 6578
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5949007868766785,
      "learning_rate": 4.901692366534872e-07,
      "loss": 1.0102,
      "step": 6579
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2714898586273193,
      "learning_rate": 4.892532953474e-07,
      "loss": 0.773,
      "step": 6580
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.274544358253479,
      "learning_rate": 4.883381665911708e-07,
      "loss": 0.7936,
      "step": 6581
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2333978414535522,
      "learning_rate": 4.874238505496442e-07,
      "loss": 0.7922,
      "step": 6582
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5897946953773499,
      "learning_rate": 4.865103473875248e-07,
      "loss": 1.0156,
      "step": 6583
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5659948587417603,
      "learning_rate": 4.855976572693666e-07,
      "loss": 1.0126,
      "step": 6584
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5855599641799927,
      "learning_rate": 4.846857803595778e-07,
      "loss": 1.0082,
      "step": 6585
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2468281984329224,
      "learning_rate": 4.837747168224222e-07,
      "loss": 0.8025,
      "step": 6586
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.285059928894043,
      "learning_rate": 4.828644668220134e-07,
      "loss": 0.7305,
      "step": 6587
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2308632135391235,
      "learning_rate": 4.819550305223225e-07,
      "loss": 0.7295,
      "step": 6588
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5900975465774536,
      "learning_rate": 4.810464080871719e-07,
      "loss": 1.0208,
      "step": 6589
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5931254625320435,
      "learning_rate": 4.80138599680236e-07,
      "loss": 1.0112,
      "step": 6590
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5736281871795654,
      "learning_rate": 4.792316054650458e-07,
      "loss": 1.0046,
      "step": 6591
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.339024543762207,
      "learning_rate": 4.78325425604983e-07,
      "loss": 0.7669,
      "step": 6592
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2361751794815063,
      "learning_rate": 4.774200602632851e-07,
      "loss": 0.7854,
      "step": 6593
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4353351593017578,
      "learning_rate": 4.765155096030405e-07,
      "loss": 0.6613,
      "step": 6594
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.205812692642212,
      "learning_rate": 4.7561177378719016e-07,
      "loss": 0.7638,
      "step": 6595
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2323895692825317,
      "learning_rate": 4.7470885297853265e-07,
      "loss": 0.7563,
      "step": 6596
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5708876252174377,
      "learning_rate": 4.738067473397157e-07,
      "loss": 1.0203,
      "step": 6597
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.299358606338501,
      "learning_rate": 4.7290545703324033e-07,
      "loss": 0.7897,
      "step": 6598
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5700481534004211,
      "learning_rate": 4.720049822214634e-07,
      "loss": 1.0056,
      "step": 6599
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2347513437271118,
      "learning_rate": 4.7110532306659184e-07,
      "loss": 0.7615,
      "step": 6600
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5799998044967651,
      "learning_rate": 4.702064797306888e-07,
      "loss": 1.015,
      "step": 6601
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2982556819915771,
      "learning_rate": 4.693084523756675e-07,
      "loss": 0.7911,
      "step": 6602
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1859965324401855,
      "learning_rate": 4.684112411632946e-07,
      "loss": 0.7184,
      "step": 6603
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2616331577301025,
      "learning_rate": 4.675148462551926e-07,
      "loss": 0.7623,
      "step": 6604
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2417110204696655,
      "learning_rate": 4.666192678128334e-07,
      "loss": 0.7842,
      "step": 6605
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.585343062877655,
      "learning_rate": 4.6572450599754403e-07,
      "loss": 0.9805,
      "step": 6606
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.15427827835083,
      "learning_rate": 4.6483056097050286e-07,
      "loss": 0.6981,
      "step": 6607
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.273724913597107,
      "learning_rate": 4.639374328927426e-07,
      "loss": 0.7657,
      "step": 6608
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.204558253288269,
      "learning_rate": 4.630451219251486e-07,
      "loss": 0.7552,
      "step": 6609
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2218949794769287,
      "learning_rate": 4.621536282284572e-07,
      "loss": 0.7128,
      "step": 6610
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.267396330833435,
      "learning_rate": 4.6126295196326e-07,
      "loss": 0.8531,
      "step": 6611
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2619339227676392,
      "learning_rate": 4.6037309329000077e-07,
      "loss": 0.7349,
      "step": 6612
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2980663776397705,
      "learning_rate": 4.5948405236897366e-07,
      "loss": 0.7809,
      "step": 6613
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5678025484085083,
      "learning_rate": 4.5859582936032877e-07,
      "loss": 0.9956,
      "step": 6614
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2187678813934326,
      "learning_rate": 4.577084244240665e-07,
      "loss": 0.7535,
      "step": 6615
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.578671395778656,
      "learning_rate": 4.568218377200423e-07,
      "loss": 1.0101,
      "step": 6616
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5653660893440247,
      "learning_rate": 4.5593606940796175e-07,
      "loss": 0.991,
      "step": 6617
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3388733863830566,
      "learning_rate": 4.550511196473828e-07,
      "loss": 0.7656,
      "step": 6618
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.225724697113037,
      "learning_rate": 4.541669885977196e-07,
      "loss": 0.756,
      "step": 6619
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2566709518432617,
      "learning_rate": 4.532836764182347e-07,
      "loss": 0.808,
      "step": 6620
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2045938968658447,
      "learning_rate": 4.524011832680458e-07,
      "loss": 0.8284,
      "step": 6621
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2482060194015503,
      "learning_rate": 4.5151950930612144e-07,
      "loss": 0.7529,
      "step": 6622
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2301157712936401,
      "learning_rate": 4.5063865469128276e-07,
      "loss": 0.8118,
      "step": 6623
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3081018924713135,
      "learning_rate": 4.497586195822046e-07,
      "loss": 0.7195,
      "step": 6624
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3766489028930664,
      "learning_rate": 4.488794041374128e-07,
      "loss": 0.758,
      "step": 6625
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2517253160476685,
      "learning_rate": 4.480010085152875e-07,
      "loss": 0.7927,
      "step": 6626
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2352551221847534,
      "learning_rate": 4.471234328740581e-07,
      "loss": 0.7431,
      "step": 6627
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3611223697662354,
      "learning_rate": 4.4624667737180827e-07,
      "loss": 0.8093,
      "step": 6628
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2440592050552368,
      "learning_rate": 4.4537074216647434e-07,
      "loss": 0.7615,
      "step": 6629
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5668907165527344,
      "learning_rate": 4.4449562741584353e-07,
      "loss": 1.0149,
      "step": 6630
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5687164068222046,
      "learning_rate": 4.436213332775557e-07,
      "loss": 1.013,
      "step": 6631
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2176631689071655,
      "learning_rate": 4.42747859909104e-07,
      "loss": 0.7035,
      "step": 6632
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.292076826095581,
      "learning_rate": 4.4187520746783177e-07,
      "loss": 0.779,
      "step": 6633
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5619764924049377,
      "learning_rate": 4.410033761109361e-07,
      "loss": 1.0024,
      "step": 6634
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5784081816673279,
      "learning_rate": 4.401323659954665e-07,
      "loss": 1.0296,
      "step": 6635
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.235222339630127,
      "learning_rate": 4.392621772783212e-07,
      "loss": 0.8332,
      "step": 6636
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3787764310836792,
      "learning_rate": 4.383928101162549e-07,
      "loss": 0.756,
      "step": 6637
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2993556261062622,
      "learning_rate": 4.3752426466587174e-07,
      "loss": 0.7364,
      "step": 6638
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.242666244506836,
      "learning_rate": 4.3665654108362877e-07,
      "loss": 0.7381,
      "step": 6639
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2616852521896362,
      "learning_rate": 4.357896395258343e-07,
      "loss": 0.7586,
      "step": 6640
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5576772093772888,
      "learning_rate": 4.349235601486479e-07,
      "loss": 1.0015,
      "step": 6641
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2194777727127075,
      "learning_rate": 4.340583031080836e-07,
      "loss": 0.7701,
      "step": 6642
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3528231382369995,
      "learning_rate": 4.3319386856000565e-07,
      "loss": 0.7867,
      "step": 6643
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.238764762878418,
      "learning_rate": 4.323302566601284e-07,
      "loss": 0.7504,
      "step": 6644
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.249118447303772,
      "learning_rate": 4.3146746756402236e-07,
      "loss": 0.7234,
      "step": 6645
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2874908447265625,
      "learning_rate": 4.30605501427106e-07,
      "loss": 0.7041,
      "step": 6646
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.337647557258606,
      "learning_rate": 4.297443584046507e-07,
      "loss": 0.7647,
      "step": 6647
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2293111085891724,
      "learning_rate": 4.28884038651779e-07,
      "loss": 0.8177,
      "step": 6648
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2161970138549805,
      "learning_rate": 4.280245423234675e-07,
      "loss": 0.7627,
      "step": 6649
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.204355001449585,
      "learning_rate": 4.2716586957454174e-07,
      "loss": 0.6958,
      "step": 6650
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2572802305221558,
      "learning_rate": 4.263080205596798e-07,
      "loss": 0.7669,
      "step": 6651
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.306933879852295,
      "learning_rate": 4.254509954334124e-07,
      "loss": 0.7998,
      "step": 6652
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5692976713180542,
      "learning_rate": 4.2459479435011995e-07,
      "loss": 0.9984,
      "step": 6653
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5765652656555176,
      "learning_rate": 4.237394174640369e-07,
      "loss": 0.9819,
      "step": 6654
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.248899221420288,
      "learning_rate": 4.228848649292472e-07,
      "loss": 0.7982,
      "step": 6655
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2393594980239868,
      "learning_rate": 4.22031136899686e-07,
      "loss": 0.7384,
      "step": 6656
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5699577927589417,
      "learning_rate": 4.21178233529142e-07,
      "loss": 0.9948,
      "step": 6657
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.210028886795044,
      "learning_rate": 4.2032615497125396e-07,
      "loss": 0.7717,
      "step": 6658
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5942990183830261,
      "learning_rate": 4.1947490137951076e-07,
      "loss": 1.0308,
      "step": 6659
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2718348503112793,
      "learning_rate": 4.186244729072564e-07,
      "loss": 0.8009,
      "step": 6660
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5846972465515137,
      "learning_rate": 4.177748697076822e-07,
      "loss": 1.0006,
      "step": 6661
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5618342161178589,
      "learning_rate": 4.169260919338347e-07,
      "loss": 1.0271,
      "step": 6662
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1949639320373535,
      "learning_rate": 4.160781397386082e-07,
      "loss": 0.7738,
      "step": 6663
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5639432668685913,
      "learning_rate": 4.1523101327474945e-07,
      "loss": 1.0101,
      "step": 6664
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2461563348770142,
      "learning_rate": 4.143847126948586e-07,
      "loss": 0.7525,
      "step": 6665
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.219457983970642,
      "learning_rate": 4.135392381513831e-07,
      "loss": 0.7189,
      "step": 6666
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2686213254928589,
      "learning_rate": 4.126945897966256e-07,
      "loss": 0.7696,
      "step": 6667
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2212337255477905,
      "learning_rate": 4.118507677827371e-07,
      "loss": 0.7247,
      "step": 6668
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5804188847541809,
      "learning_rate": 4.110077722617195e-07,
      "loss": 0.996,
      "step": 6669
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2012314796447754,
      "learning_rate": 4.1016560338542954e-07,
      "loss": 0.762,
      "step": 6670
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5772923827171326,
      "learning_rate": 4.093242613055709e-07,
      "loss": 0.9949,
      "step": 6671
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2897937297821045,
      "learning_rate": 4.084837461737007e-07,
      "loss": 0.8252,
      "step": 6672
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2708014249801636,
      "learning_rate": 4.0764405814122556e-07,
      "loss": 0.7973,
      "step": 6673
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.230891466140747,
      "learning_rate": 4.0680519735940394e-07,
      "loss": 0.7643,
      "step": 6674
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5638816356658936,
      "learning_rate": 4.0596716397934665e-07,
      "loss": 0.9804,
      "step": 6675
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2143296003341675,
      "learning_rate": 4.0512995815201185e-07,
      "loss": 0.7207,
      "step": 6676
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5865045785903931,
      "learning_rate": 4.0429358002821275e-07,
      "loss": 1.0389,
      "step": 6677
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1685093641281128,
      "learning_rate": 4.034580297586105e-07,
      "loss": 0.7777,
      "step": 6678
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.220349669456482,
      "learning_rate": 4.0262330749371805e-07,
      "loss": 0.7417,
      "step": 6679
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5574803948402405,
      "learning_rate": 4.017894133839001e-07,
      "loss": 1.0131,
      "step": 6680
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2943605184555054,
      "learning_rate": 4.009563475793704e-07,
      "loss": 0.8133,
      "step": 6681
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5829062461853027,
      "learning_rate": 4.001241102301956e-07,
      "loss": 0.9887,
      "step": 6682
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.258069634437561,
      "learning_rate": 3.992927014862913e-07,
      "loss": 0.8382,
      "step": 6683
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2369861602783203,
      "learning_rate": 3.9846212149742433e-07,
      "loss": 0.774,
      "step": 6684
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5536277294158936,
      "learning_rate": 3.9763237041321233e-07,
      "loss": 0.9859,
      "step": 6685
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2478246688842773,
      "learning_rate": 3.9680344838312456e-07,
      "loss": 0.733,
      "step": 6686
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2391077280044556,
      "learning_rate": 3.959753555564794e-07,
      "loss": 0.8201,
      "step": 6687
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2115610837936401,
      "learning_rate": 3.9514809208244685e-07,
      "loss": 0.7284,
      "step": 6688
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3042914867401123,
      "learning_rate": 3.943216581100462e-07,
      "loss": 0.7836,
      "step": 6689
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5699750185012817,
      "learning_rate": 3.934960537881499e-07,
      "loss": 0.9953,
      "step": 6690
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2743703126907349,
      "learning_rate": 3.9267127926547856e-07,
      "loss": 0.8048,
      "step": 6691
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2330232858657837,
      "learning_rate": 3.918473346906032e-07,
      "loss": 0.738,
      "step": 6692
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5727026462554932,
      "learning_rate": 3.9102422021194853e-07,
      "loss": 1.0177,
      "step": 6693
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2804151773452759,
      "learning_rate": 3.902019359777848e-07,
      "loss": 0.7946,
      "step": 6694
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2521865367889404,
      "learning_rate": 3.893804821362379e-07,
      "loss": 0.7526,
      "step": 6695
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2518911361694336,
      "learning_rate": 3.885598588352807e-07,
      "loss": 0.841,
      "step": 6696
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.214052677154541,
      "learning_rate": 3.8774006622273597e-07,
      "loss": 0.7998,
      "step": 6697
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2113348245620728,
      "learning_rate": 3.869211044462806e-07,
      "loss": 0.759,
      "step": 6698
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3098355531692505,
      "learning_rate": 3.861029736534366e-07,
      "loss": 0.7635,
      "step": 6699
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4996604919433594,
      "learning_rate": 3.8528567399158225e-07,
      "loss": 0.7737,
      "step": 6700
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2557615041732788,
      "learning_rate": 3.844692056079413e-07,
      "loss": 0.7531,
      "step": 6701
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2717037200927734,
      "learning_rate": 3.8365356864958903e-07,
      "loss": 0.8023,
      "step": 6702
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2796545028686523,
      "learning_rate": 3.8283876326345283e-07,
      "loss": 0.7629,
      "step": 6703
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2756237983703613,
      "learning_rate": 3.8202478959630694e-07,
      "loss": 0.8604,
      "step": 6704
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2171376943588257,
      "learning_rate": 3.812116477947803e-07,
      "loss": 0.7499,
      "step": 6705
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.23923659324646,
      "learning_rate": 3.8039933800534724e-07,
      "loss": 0.7753,
      "step": 6706
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.232079029083252,
      "learning_rate": 3.7958786037433413e-07,
      "loss": 0.7702,
      "step": 6707
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5892322659492493,
      "learning_rate": 3.7877721504791963e-07,
      "loss": 1.0277,
      "step": 6708
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2205450534820557,
      "learning_rate": 3.7796740217212804e-07,
      "loss": 0.7406,
      "step": 6709
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.581059992313385,
      "learning_rate": 3.771584218928387e-07,
      "loss": 0.9952,
      "step": 6710
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.292738914489746,
      "learning_rate": 3.7635027435577734e-07,
      "loss": 0.7372,
      "step": 6711
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4716858863830566,
      "learning_rate": 3.755429597065197e-07,
      "loss": 0.7705,
      "step": 6712
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2997649908065796,
      "learning_rate": 3.747364780904944e-07,
      "loss": 0.8035,
      "step": 6713
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2536205053329468,
      "learning_rate": 3.7393082965297754e-07,
      "loss": 0.7593,
      "step": 6714
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.292948842048645,
      "learning_rate": 3.731260145390941e-07,
      "loss": 0.7479,
      "step": 6715
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.296531319618225,
      "learning_rate": 3.7232203289382374e-07,
      "loss": 0.7813,
      "step": 6716
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5692836046218872,
      "learning_rate": 3.715188848619894e-07,
      "loss": 1.0027,
      "step": 6717
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2319753170013428,
      "learning_rate": 3.707165705882704e-07,
      "loss": 0.7287,
      "step": 6718
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.232141137123108,
      "learning_rate": 3.6991509021719163e-07,
      "loss": 0.737,
      "step": 6719
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.253011703491211,
      "learning_rate": 3.6911444389312766e-07,
      "loss": 0.7828,
      "step": 6720
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.180871605873108,
      "learning_rate": 3.683146317603059e-07,
      "loss": 0.7561,
      "step": 6721
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3433966636657715,
      "learning_rate": 3.6751565396280055e-07,
      "loss": 0.7511,
      "step": 6722
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2599942684173584,
      "learning_rate": 3.6671751064453756e-07,
      "loss": 0.8026,
      "step": 6723
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5847632884979248,
      "learning_rate": 3.659202019492908e-07,
      "loss": 1.0224,
      "step": 6724
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3713915348052979,
      "learning_rate": 3.6512372802068486e-07,
      "loss": 0.858,
      "step": 6725
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2011635303497314,
      "learning_rate": 3.643280890021933e-07,
      "loss": 0.728,
      "step": 6726
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5593814849853516,
      "learning_rate": 3.6353328503714036e-07,
      "loss": 1.0248,
      "step": 6727
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2331876754760742,
      "learning_rate": 3.6273931626869927e-07,
      "loss": 0.7876,
      "step": 6728
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.585059642791748,
      "learning_rate": 3.6194618283989183e-07,
      "loss": 0.9727,
      "step": 6729
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2473398447036743,
      "learning_rate": 3.6115388489359137e-07,
      "loss": 0.7706,
      "step": 6730
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.228637933731079,
      "learning_rate": 3.603624225725194e-07,
      "loss": 0.7486,
      "step": 6731
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.574200451374054,
      "learning_rate": 3.5957179601924574e-07,
      "loss": 1.0269,
      "step": 6732
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.354030728340149,
      "learning_rate": 3.587820053761931e-07,
      "loss": 0.8,
      "step": 6733
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2419865131378174,
      "learning_rate": 3.5799305078563104e-07,
      "loss": 0.7672,
      "step": 6734
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3159197568893433,
      "learning_rate": 3.57204932389677e-07,
      "loss": 0.738,
      "step": 6735
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2834558486938477,
      "learning_rate": 3.5641765033030294e-07,
      "loss": 0.7454,
      "step": 6736
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2086180448532104,
      "learning_rate": 3.5563120474932435e-07,
      "loss": 0.7756,
      "step": 6737
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5713205933570862,
      "learning_rate": 3.548455957884106e-07,
      "loss": 1.0128,
      "step": 6738
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2852287292480469,
      "learning_rate": 3.54060823589078e-07,
      "loss": 0.7627,
      "step": 6739
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2425315380096436,
      "learning_rate": 3.532768882926918e-07,
      "loss": 0.6823,
      "step": 6740
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5778658986091614,
      "learning_rate": 3.5249379004046847e-07,
      "loss": 1.037,
      "step": 6741
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.231349229812622,
      "learning_rate": 3.5171152897347125e-07,
      "loss": 0.7343,
      "step": 6742
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.565989077091217,
      "learning_rate": 3.5093010523261583e-07,
      "loss": 1.0056,
      "step": 6743
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5655950307846069,
      "learning_rate": 3.5014951895866347e-07,
      "loss": 1.0136,
      "step": 6744
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2623529434204102,
      "learning_rate": 3.4936977029222607e-07,
      "loss": 0.7516,
      "step": 6745
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5639002919197083,
      "learning_rate": 3.485908593737658e-07,
      "loss": 1.0002,
      "step": 6746
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5782889723777771,
      "learning_rate": 3.478127863435926e-07,
      "loss": 1.0129,
      "step": 6747
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3790699243545532,
      "learning_rate": 3.4703555134186496e-07,
      "loss": 0.7225,
      "step": 6748
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5675606727600098,
      "learning_rate": 3.462591545085925e-07,
      "loss": 0.9843,
      "step": 6749
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2624812126159668,
      "learning_rate": 3.454835959836317e-07,
      "loss": 0.8131,
      "step": 6750
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2095379829406738,
      "learning_rate": 3.447088759066891e-07,
      "loss": 0.7084,
      "step": 6751
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.264265775680542,
      "learning_rate": 3.439349944173209e-07,
      "loss": 0.7693,
      "step": 6752
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2913591861724854,
      "learning_rate": 3.4316195165492947e-07,
      "loss": 0.8197,
      "step": 6753
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3146791458129883,
      "learning_rate": 3.4238974775876954e-07,
      "loss": 0.7938,
      "step": 6754
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5633770823478699,
      "learning_rate": 3.4161838286794256e-07,
      "loss": 0.9797,
      "step": 6755
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2373526096343994,
      "learning_rate": 3.4084785712139967e-07,
      "loss": 0.765,
      "step": 6756
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.266555666923523,
      "learning_rate": 3.4007817065794146e-07,
      "loss": 0.8065,
      "step": 6757
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2456018924713135,
      "learning_rate": 3.3930932361621426e-07,
      "loss": 0.7623,
      "step": 6758
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5773100256919861,
      "learning_rate": 3.385413161347173e-07,
      "loss": 1.0082,
      "step": 6759
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5777111053466797,
      "learning_rate": 3.37774148351796e-07,
      "loss": 1.0091,
      "step": 6760
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2582662105560303,
      "learning_rate": 3.3700782040564594e-07,
      "loss": 0.7899,
      "step": 6761
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5622374415397644,
      "learning_rate": 3.3624233243431004e-07,
      "loss": 1.0009,
      "step": 6762
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2936961650848389,
      "learning_rate": 3.3547768457568017e-07,
      "loss": 0.7323,
      "step": 6763
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1933408975601196,
      "learning_rate": 3.3471387696749844e-07,
      "loss": 0.718,
      "step": 6764
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.180311679840088,
      "learning_rate": 3.3395090974735423e-07,
      "loss": 0.7374,
      "step": 6765
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5754979848861694,
      "learning_rate": 3.331887830526853e-07,
      "loss": 0.9917,
      "step": 6766
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3695228099822998,
      "learning_rate": 3.324274970207775e-07,
      "loss": 0.7245,
      "step": 6767
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2561180591583252,
      "learning_rate": 3.3166705178876825e-07,
      "loss": 0.7717,
      "step": 6768
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2297881841659546,
      "learning_rate": 3.309074474936402e-07,
      "loss": 0.7508,
      "step": 6769
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5823480486869812,
      "learning_rate": 3.3014868427222513e-07,
      "loss": 0.9866,
      "step": 6770
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5659955739974976,
      "learning_rate": 3.2939076226120524e-07,
      "loss": 1.0064,
      "step": 6771
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3084532022476196,
      "learning_rate": 3.2863368159710975e-07,
      "loss": 0.7788,
      "step": 6772
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2568517923355103,
      "learning_rate": 3.2787744241631556e-07,
      "loss": 0.781,
      "step": 6773
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.217248558998108,
      "learning_rate": 3.2712204485504994e-07,
      "loss": 0.7926,
      "step": 6774
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.574140727519989,
      "learning_rate": 3.263674890493873e-07,
      "loss": 0.9914,
      "step": 6775
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2815370559692383,
      "learning_rate": 3.256137751352495e-07,
      "loss": 0.7793,
      "step": 6776
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5616647005081177,
      "learning_rate": 3.2486090324840957e-07,
      "loss": 1.0248,
      "step": 6777
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.248412847518921,
      "learning_rate": 3.241088735244857e-07,
      "loss": 0.7337,
      "step": 6778
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2689138650894165,
      "learning_rate": 3.2335768609894735e-07,
      "loss": 0.7202,
      "step": 6779
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1723521947860718,
      "learning_rate": 3.226073411071096e-07,
      "loss": 0.796,
      "step": 6780
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3194137811660767,
      "learning_rate": 3.21857838684137e-07,
      "loss": 0.8218,
      "step": 6781
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.266641616821289,
      "learning_rate": 3.2110917896504277e-07,
      "loss": 0.7722,
      "step": 6782
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2645349502563477,
      "learning_rate": 3.2036136208468684e-07,
      "loss": 0.7333,
      "step": 6783
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.228433609008789,
      "learning_rate": 3.196143881777802e-07,
      "loss": 0.8039,
      "step": 6784
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3127747774124146,
      "learning_rate": 3.188682573788787e-07,
      "loss": 0.7794,
      "step": 6785
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.249667763710022,
      "learning_rate": 3.18122969822387e-07,
      "loss": 0.7062,
      "step": 6786
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2881689071655273,
      "learning_rate": 3.173785256425599e-07,
      "loss": 0.7003,
      "step": 6787
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3167767524719238,
      "learning_rate": 3.166349249734979e-07,
      "loss": 0.7456,
      "step": 6788
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5750754475593567,
      "learning_rate": 3.158921679491522e-07,
      "loss": 1.008,
      "step": 6789
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3267277479171753,
      "learning_rate": 3.151502547033186e-07,
      "loss": 0.8172,
      "step": 6790
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.280099630355835,
      "learning_rate": 3.144091853696435e-07,
      "loss": 0.7514,
      "step": 6791
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2184265851974487,
      "learning_rate": 3.1366896008162064e-07,
      "loss": 0.7671,
      "step": 6792
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1905155181884766,
      "learning_rate": 3.129295789725906e-07,
      "loss": 0.749,
      "step": 6793
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5576725602149963,
      "learning_rate": 3.121910421757446e-07,
      "loss": 1.0019,
      "step": 6794
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5642749071121216,
      "learning_rate": 3.1145334982411845e-07,
      "loss": 0.9872,
      "step": 6795
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5882588028907776,
      "learning_rate": 3.1071650205059755e-07,
      "loss": 0.9809,
      "step": 6796
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5466628670692444,
      "learning_rate": 3.0998049898791685e-07,
      "loss": 0.9902,
      "step": 6797
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3032149076461792,
      "learning_rate": 3.0924534076865464e-07,
      "loss": 0.7843,
      "step": 6798
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2863703966140747,
      "learning_rate": 3.0851102752524176e-07,
      "loss": 0.7629,
      "step": 6799
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2187360525131226,
      "learning_rate": 3.07777559389954e-07,
      "loss": 0.7446,
      "step": 6800
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5760177373886108,
      "learning_rate": 3.070449364949152e-07,
      "loss": 1.0056,
      "step": 6801
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.186623454093933,
      "learning_rate": 3.0631315897209854e-07,
      "loss": 0.8116,
      "step": 6802
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3210936784744263,
      "learning_rate": 3.055822269533237e-07,
      "loss": 0.7879,
      "step": 6803
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.206268310546875,
      "learning_rate": 3.048521405702565e-07,
      "loss": 0.7482,
      "step": 6804
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.551679790019989,
      "learning_rate": 3.0412289995441437e-07,
      "loss": 1.0143,
      "step": 6805
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2412502765655518,
      "learning_rate": 3.0339450523715896e-07,
      "loss": 0.746,
      "step": 6806
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5746703743934631,
      "learning_rate": 3.026669565497009e-07,
      "loss": 1.0033,
      "step": 6807
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2491235733032227,
      "learning_rate": 3.0194025402309757e-07,
      "loss": 0.7898,
      "step": 6808
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.578374981880188,
      "learning_rate": 3.012143977882553e-07,
      "loss": 1.0339,
      "step": 6809
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2607454061508179,
      "learning_rate": 3.00489387975928e-07,
      "loss": 0.7794,
      "step": 6810
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.256473422050476,
      "learning_rate": 2.997652247167138e-07,
      "loss": 0.77,
      "step": 6811
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2935141324996948,
      "learning_rate": 2.990419081410639e-07,
      "loss": 0.8032,
      "step": 6812
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3388158082962036,
      "learning_rate": 2.9831943837927257e-07,
      "loss": 0.7796,
      "step": 6813
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2942581176757812,
      "learning_rate": 2.9759781556148224e-07,
      "loss": 0.751,
      "step": 6814
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3011196851730347,
      "learning_rate": 2.968770398176851e-07,
      "loss": 0.794,
      "step": 6815
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2484960556030273,
      "learning_rate": 2.961571112777173e-07,
      "loss": 0.7387,
      "step": 6816
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.567099928855896,
      "learning_rate": 2.9543803007126613e-07,
      "loss": 1.009,
      "step": 6817
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3148030042648315,
      "learning_rate": 2.947197963278631e-07,
      "loss": 0.6972,
      "step": 6818
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2622193098068237,
      "learning_rate": 2.940024101768879e-07,
      "loss": 0.7683,
      "step": 6819
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2379364967346191,
      "learning_rate": 2.932858717475695e-07,
      "loss": 0.8407,
      "step": 6820
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5689172148704529,
      "learning_rate": 2.9257018116898074e-07,
      "loss": 1.013,
      "step": 6821
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.568686306476593,
      "learning_rate": 2.918553385700451e-07,
      "loss": 1.0278,
      "step": 6822
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1881120204925537,
      "learning_rate": 2.911413440795313e-07,
      "loss": 0.7557,
      "step": 6823
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.277739405632019,
      "learning_rate": 2.904281978260548e-07,
      "loss": 0.7773,
      "step": 6824
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3256138563156128,
      "learning_rate": 2.897158999380806e-07,
      "loss": 0.7727,
      "step": 6825
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5600088238716125,
      "learning_rate": 2.8900445054391877e-07,
      "loss": 1.0239,
      "step": 6826
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2402626276016235,
      "learning_rate": 2.8829384977172747e-07,
      "loss": 0.7564,
      "step": 6827
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.250677466392517,
      "learning_rate": 2.875840977495126e-07,
      "loss": 0.7702,
      "step": 6828
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.566483199596405,
      "learning_rate": 2.86875194605124e-07,
      "loss": 0.9753,
      "step": 6829
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2067053318023682,
      "learning_rate": 2.861671404662636e-07,
      "loss": 0.7743,
      "step": 6830
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3336296081542969,
      "learning_rate": 2.85459935460477e-07,
      "loss": 0.8051,
      "step": 6831
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.258565902709961,
      "learning_rate": 2.847535797151563e-07,
      "loss": 0.7867,
      "step": 6832
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.247872233390808,
      "learning_rate": 2.8404807335754346e-07,
      "loss": 0.7382,
      "step": 6833
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2460753917694092,
      "learning_rate": 2.8334341651472477e-07,
      "loss": 0.7709,
      "step": 6834
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2986698150634766,
      "learning_rate": 2.8263960931363524e-07,
      "loss": 0.8555,
      "step": 6835
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.249463677406311,
      "learning_rate": 2.819366518810568e-07,
      "loss": 0.6974,
      "step": 6836
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.564952552318573,
      "learning_rate": 2.8123454434361597e-07,
      "loss": 1.0038,
      "step": 6837
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2763733863830566,
      "learning_rate": 2.8053328682778936e-07,
      "loss": 0.7628,
      "step": 6838
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2177176475524902,
      "learning_rate": 2.7983287945989813e-07,
      "loss": 0.7905,
      "step": 6839
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.319801926612854,
      "learning_rate": 2.7913332236611243e-07,
      "loss": 0.7931,
      "step": 6840
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5648277401924133,
      "learning_rate": 2.784346156724471e-07,
      "loss": 1.0038,
      "step": 6841
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.257209062576294,
      "learning_rate": 2.7773675950476366e-07,
      "loss": 0.7933,
      "step": 6842
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2376515865325928,
      "learning_rate": 2.770397539887737e-07,
      "loss": 0.7165,
      "step": 6843
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2240307331085205,
      "learning_rate": 2.763435992500318e-07,
      "loss": 0.7526,
      "step": 6844
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.194719910621643,
      "learning_rate": 2.7564829541394057e-07,
      "loss": 0.7057,
      "step": 6845
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.576127827167511,
      "learning_rate": 2.7495384260575077e-07,
      "loss": 1.0038,
      "step": 6846
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.237623691558838,
      "learning_rate": 2.7426024095055794e-07,
      "loss": 0.8029,
      "step": 6847
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5776860117912292,
      "learning_rate": 2.7356749057330433e-07,
      "loss": 1.0057,
      "step": 6848
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2263394594192505,
      "learning_rate": 2.7287559159878076e-07,
      "loss": 0.7662,
      "step": 6849
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2015115022659302,
      "learning_rate": 2.72184544151623e-07,
      "loss": 0.7626,
      "step": 6850
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2768445014953613,
      "learning_rate": 2.7149434835631426e-07,
      "loss": 0.8048,
      "step": 6851
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2882134914398193,
      "learning_rate": 2.708050043371824e-07,
      "loss": 0.7876,
      "step": 6852
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.321570873260498,
      "learning_rate": 2.7011651221840574e-07,
      "loss": 0.7212,
      "step": 6853
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3432049751281738,
      "learning_rate": 2.694288721240046e-07,
      "loss": 0.7589,
      "step": 6854
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3570663928985596,
      "learning_rate": 2.687420841778493e-07,
      "loss": 0.7975,
      "step": 6855
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2623823881149292,
      "learning_rate": 2.6805614850365536e-07,
      "loss": 0.7612,
      "step": 6856
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2057629823684692,
      "learning_rate": 2.673710652249839e-07,
      "loss": 0.7925,
      "step": 6857
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2369623184204102,
      "learning_rate": 2.6668683446524456e-07,
      "loss": 0.7763,
      "step": 6858
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2451075315475464,
      "learning_rate": 2.660034563476921e-07,
      "loss": 0.8009,
      "step": 6859
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2834852933883667,
      "learning_rate": 2.653209309954263e-07,
      "loss": 0.7678,
      "step": 6860
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1795929670333862,
      "learning_rate": 2.6463925853139673e-07,
      "loss": 0.7173,
      "step": 6861
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5803419947624207,
      "learning_rate": 2.639584390783956e-07,
      "loss": 1.0305,
      "step": 6862
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3120834827423096,
      "learning_rate": 2.632784727590654e-07,
      "loss": 0.7724,
      "step": 6863
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.191565990447998,
      "learning_rate": 2.6259935969589203e-07,
      "loss": 0.7237,
      "step": 6864
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3418090343475342,
      "learning_rate": 2.6192110001120707e-07,
      "loss": 0.7517,
      "step": 6865
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3266451358795166,
      "learning_rate": 2.612436938271917e-07,
      "loss": 0.7833,
      "step": 6866
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3101129531860352,
      "learning_rate": 2.605671412658706e-07,
      "loss": 0.8076,
      "step": 6867
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2846250534057617,
      "learning_rate": 2.598914424491161e-07,
      "loss": 0.7899,
      "step": 6868
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2302207946777344,
      "learning_rate": 2.592165974986455e-07,
      "loss": 0.7671,
      "step": 6869
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2162821292877197,
      "learning_rate": 2.5854260653602306e-07,
      "loss": 0.7563,
      "step": 6870
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5675669312477112,
      "learning_rate": 2.578694696826595e-07,
      "loss": 1.0127,
      "step": 6871
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2704130411148071,
      "learning_rate": 2.5719718705981055e-07,
      "loss": 0.745,
      "step": 6872
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2027393579483032,
      "learning_rate": 2.5652575878858e-07,
      "loss": 0.788,
      "step": 6873
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2683405876159668,
      "learning_rate": 2.5585518498991604e-07,
      "loss": 0.8058,
      "step": 6874
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2423839569091797,
      "learning_rate": 2.5518546578461256e-07,
      "loss": 0.6913,
      "step": 6875
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3060508966445923,
      "learning_rate": 2.5451660129331204e-07,
      "loss": 0.7439,
      "step": 6876
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2674893140792847,
      "learning_rate": 2.5384859163649966e-07,
      "loss": 0.7757,
      "step": 6877
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2401233911514282,
      "learning_rate": 2.531814369345098e-07,
      "loss": 0.7433,
      "step": 6878
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2932156324386597,
      "learning_rate": 2.525151373075202e-07,
      "loss": 0.7871,
      "step": 6879
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2668280601501465,
      "learning_rate": 2.5184969287555595e-07,
      "loss": 0.7794,
      "step": 6880
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2132108211517334,
      "learning_rate": 2.5118510375848834e-07,
      "loss": 0.7455,
      "step": 6881
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5694855451583862,
      "learning_rate": 2.5052137007603337e-07,
      "loss": 1.0112,
      "step": 6882
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2953110933303833,
      "learning_rate": 2.498584919477548e-07,
      "loss": 0.832,
      "step": 6883
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3116189241409302,
      "learning_rate": 2.491964694930604e-07,
      "loss": 0.7991,
      "step": 6884
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2215734720230103,
      "learning_rate": 2.485353028312043e-07,
      "loss": 0.7117,
      "step": 6885
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5642130374908447,
      "learning_rate": 2.4787499208128616e-07,
      "loss": 0.9925,
      "step": 6886
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.184691071510315,
      "learning_rate": 2.472155373622537e-07,
      "loss": 0.7489,
      "step": 6887
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2232792377471924,
      "learning_rate": 2.4655693879289787e-07,
      "loss": 0.7543,
      "step": 6888
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5902747511863708,
      "learning_rate": 2.4589919649185556e-07,
      "loss": 1.0295,
      "step": 6889
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5620668530464172,
      "learning_rate": 2.4524231057761136e-07,
      "loss": 0.9601,
      "step": 6890
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3214738368988037,
      "learning_rate": 2.44586281168494e-07,
      "loss": 0.8329,
      "step": 6891
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2859951257705688,
      "learning_rate": 2.439311083826779e-07,
      "loss": 0.79,
      "step": 6892
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5751363039016724,
      "learning_rate": 2.4327679233818294e-07,
      "loss": 1.0153,
      "step": 6893
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.294214129447937,
      "learning_rate": 2.426233331528771e-07,
      "loss": 0.7622,
      "step": 6894
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.243955135345459,
      "learning_rate": 2.419707309444702e-07,
      "loss": 0.7885,
      "step": 6895
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.253617286682129,
      "learning_rate": 2.413189858305215e-07,
      "loss": 0.7369,
      "step": 6896
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2623567581176758,
      "learning_rate": 2.406680979284337e-07,
      "loss": 0.7583,
      "step": 6897
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5633556842803955,
      "learning_rate": 2.4001806735545365e-07,
      "loss": 1.0037,
      "step": 6898
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5988162159919739,
      "learning_rate": 2.393688942286776e-07,
      "loss": 1.0005,
      "step": 6899
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1738522052764893,
      "learning_rate": 2.387205786650437e-07,
      "loss": 0.7601,
      "step": 6900
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2345346212387085,
      "learning_rate": 2.3807312078133914e-07,
      "loss": 0.8364,
      "step": 6901
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5684312582015991,
      "learning_rate": 2.3742652069419337e-07,
      "loss": 1.0107,
      "step": 6902
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2635847330093384,
      "learning_rate": 2.3678077852008206e-07,
      "loss": 0.7492,
      "step": 6903
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5836601257324219,
      "learning_rate": 2.3613589437532835e-07,
      "loss": 1.0231,
      "step": 6904
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2192109823226929,
      "learning_rate": 2.3549186837609817e-07,
      "loss": 0.7816,
      "step": 6905
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3173807859420776,
      "learning_rate": 2.3484870063840483e-07,
      "loss": 0.7682,
      "step": 6906
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5628971457481384,
      "learning_rate": 2.3420639127810618e-07,
      "loss": 1.0317,
      "step": 6907
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2358897924423218,
      "learning_rate": 2.3356494041090527e-07,
      "loss": 0.7824,
      "step": 6908
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2567378282546997,
      "learning_rate": 2.329243481523513e-07,
      "loss": 0.7788,
      "step": 6909
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.568396270275116,
      "learning_rate": 2.3228461461783702e-07,
      "loss": 1.0123,
      "step": 6910
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5682054758071899,
      "learning_rate": 2.3164573992260352e-07,
      "loss": 1.0132,
      "step": 6911
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2693440914154053,
      "learning_rate": 2.3100772418173434e-07,
      "loss": 0.7831,
      "step": 6912
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5799645781517029,
      "learning_rate": 2.3037056751015864e-07,
      "loss": 1.0076,
      "step": 6913
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2265769243240356,
      "learning_rate": 2.2973427002265348e-07,
      "loss": 0.7379,
      "step": 6914
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2516292333602905,
      "learning_rate": 2.2909883183383774e-07,
      "loss": 0.7836,
      "step": 6915
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.252631664276123,
      "learning_rate": 2.2846425305817766e-07,
      "loss": 0.7554,
      "step": 6916
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2690248489379883,
      "learning_rate": 2.27830533809984e-07,
      "loss": 0.7708,
      "step": 6917
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.563795268535614,
      "learning_rate": 2.2719767420341155e-07,
      "loss": 1.0122,
      "step": 6918
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2405246496200562,
      "learning_rate": 2.2656567435246357e-07,
      "loss": 0.7335,
      "step": 6919
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3547673225402832,
      "learning_rate": 2.2593453437098457e-07,
      "loss": 0.7762,
      "step": 6920
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3219497203826904,
      "learning_rate": 2.253042543726658e-07,
      "loss": 0.802,
      "step": 6921
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2609754800796509,
      "learning_rate": 2.2467483447104476e-07,
      "loss": 0.7563,
      "step": 6922
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2231664657592773,
      "learning_rate": 2.2404627477950135e-07,
      "loss": 0.8287,
      "step": 6923
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2958598136901855,
      "learning_rate": 2.2341857541126445e-07,
      "loss": 0.7004,
      "step": 6924
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5802488327026367,
      "learning_rate": 2.2279173647940354e-07,
      "loss": 1.0183,
      "step": 6925
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3096650838851929,
      "learning_rate": 2.2216575809683617e-07,
      "loss": 0.7794,
      "step": 6926
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.569980263710022,
      "learning_rate": 2.215406403763226e-07,
      "loss": 1.0103,
      "step": 6927
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.260528802871704,
      "learning_rate": 2.2091638343047062e-07,
      "loss": 0.7422,
      "step": 6928
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2135826349258423,
      "learning_rate": 2.2029298737173188e-07,
      "loss": 0.7782,
      "step": 6929
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5624490976333618,
      "learning_rate": 2.1967045231240103e-07,
      "loss": 1.0092,
      "step": 6930
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2364617586135864,
      "learning_rate": 2.1904877836462059e-07,
      "loss": 0.7885,
      "step": 6931
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.167583703994751,
      "learning_rate": 2.1842796564037706e-07,
      "loss": 0.7499,
      "step": 6932
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5721542835235596,
      "learning_rate": 2.178080142514999e-07,
      "loss": 1.0157,
      "step": 6933
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2992308139801025,
      "learning_rate": 2.1718892430966698e-07,
      "loss": 0.7382,
      "step": 6934
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5767437219619751,
      "learning_rate": 2.1657069592639747e-07,
      "loss": 1.0026,
      "step": 6935
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2597885131835938,
      "learning_rate": 2.1595332921305666e-07,
      "loss": 0.747,
      "step": 6936
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5649868845939636,
      "learning_rate": 2.1533682428085568e-07,
      "loss": 1.0203,
      "step": 6937
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5829665064811707,
      "learning_rate": 2.1472118124084895e-07,
      "loss": 0.9997,
      "step": 6938
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3751294612884521,
      "learning_rate": 2.1410640020393726e-07,
      "loss": 0.7892,
      "step": 6939
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3317140340805054,
      "learning_rate": 2.134924812808642e-07,
      "loss": 0.7639,
      "step": 6940
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4212541580200195,
      "learning_rate": 2.12879424582218e-07,
      "loss": 0.7314,
      "step": 6941
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3707940578460693,
      "learning_rate": 2.1226723021843475e-07,
      "loss": 0.754,
      "step": 6942
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2835241556167603,
      "learning_rate": 2.1165589829979072e-07,
      "loss": 0.8014,
      "step": 6943
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5759059190750122,
      "learning_rate": 2.1104542893641055e-07,
      "loss": 0.9939,
      "step": 6944
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2734674215316772,
      "learning_rate": 2.1043582223826185e-07,
      "loss": 0.7128,
      "step": 6945
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5777814984321594,
      "learning_rate": 2.0982707831515625e-07,
      "loss": 1.0017,
      "step": 6946
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5687837600708008,
      "learning_rate": 2.0921919727675154e-07,
      "loss": 0.9997,
      "step": 6947
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.258292317390442,
      "learning_rate": 2.0861217923254906e-07,
      "loss": 0.7762,
      "step": 6948
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3004114627838135,
      "learning_rate": 2.080060242918941e-07,
      "loss": 0.7538,
      "step": 6949
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2650374174118042,
      "learning_rate": 2.0740073256397876e-07,
      "loss": 0.7446,
      "step": 6950
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2404569387435913,
      "learning_rate": 2.0679630415783637e-07,
      "loss": 0.8144,
      "step": 6951
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.293296456336975,
      "learning_rate": 2.0619273918234762e-07,
      "loss": 0.8011,
      "step": 6952
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2361748218536377,
      "learning_rate": 2.0559003774623664e-07,
      "loss": 0.7247,
      "step": 6953
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2366244792938232,
      "learning_rate": 2.0498819995807107e-07,
      "loss": 0.7802,
      "step": 6954
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3370544910430908,
      "learning_rate": 2.0438722592626524e-07,
      "loss": 0.748,
      "step": 6955
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2342112064361572,
      "learning_rate": 2.0378711575907428e-07,
      "loss": 0.8023,
      "step": 6956
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.355546236038208,
      "learning_rate": 2.0318786956460223e-07,
      "loss": 0.7896,
      "step": 6957
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5543166995048523,
      "learning_rate": 2.0258948745079444e-07,
      "loss": 1.0179,
      "step": 6958
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2395302057266235,
      "learning_rate": 2.019919695254402e-07,
      "loss": 0.8073,
      "step": 6959
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5802670121192932,
      "learning_rate": 2.0139531589617567e-07,
      "loss": 1.0076,
      "step": 6960
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.25019371509552,
      "learning_rate": 2.0079952667047875e-07,
      "loss": 0.8084,
      "step": 6961
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2753424644470215,
      "learning_rate": 2.002046019556736e-07,
      "loss": 0.837,
      "step": 6962
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2156568765640259,
      "learning_rate": 1.9961054185892836e-07,
      "loss": 0.7794,
      "step": 6963
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5621374845504761,
      "learning_rate": 1.9901734648725303e-07,
      "loss": 0.9583,
      "step": 6964
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3225938081741333,
      "learning_rate": 1.984250159475054e-07,
      "loss": 0.7934,
      "step": 6965
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2028117179870605,
      "learning_rate": 1.9783355034638574e-07,
      "loss": 0.798,
      "step": 6966
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1627802848815918,
      "learning_rate": 1.972429497904377e-07,
      "loss": 0.7352,
      "step": 6967
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2187951803207397,
      "learning_rate": 1.9665321438605e-07,
      "loss": 0.7433,
      "step": 6968
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5675551891326904,
      "learning_rate": 1.960643442394561e-07,
      "loss": 0.9844,
      "step": 6969
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1983072757720947,
      "learning_rate": 1.9547633945673272e-07,
      "loss": 0.7198,
      "step": 6970
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.207226037979126,
      "learning_rate": 1.948892001438002e-07,
      "loss": 0.754,
      "step": 6971
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2388226985931396,
      "learning_rate": 1.94302926406425e-07,
      "loss": 0.7333,
      "step": 6972
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.34602689743042,
      "learning_rate": 1.9371751835021603e-07,
      "loss": 0.7797,
      "step": 6973
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5599948763847351,
      "learning_rate": 1.9313297608062608e-07,
      "loss": 1.0059,
      "step": 6974
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.356958031654358,
      "learning_rate": 1.925492997029532e-07,
      "loss": 0.7698,
      "step": 6975
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2530081272125244,
      "learning_rate": 1.919664893223383e-07,
      "loss": 0.8141,
      "step": 6976
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.354762315750122,
      "learning_rate": 1.9138454504376625e-07,
      "loss": 0.6948,
      "step": 6977
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2316080331802368,
      "learning_rate": 1.9080346697206765e-07,
      "loss": 0.7924,
      "step": 6978
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5791056156158447,
      "learning_rate": 1.9022325521191432e-07,
      "loss": 0.9887,
      "step": 6979
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5732427835464478,
      "learning_rate": 1.8964390986782545e-07,
      "loss": 0.9921,
      "step": 6980
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1930307149887085,
      "learning_rate": 1.8906543104416086e-07,
      "loss": 0.7825,
      "step": 6981
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5679134726524353,
      "learning_rate": 1.8848781884512613e-07,
      "loss": 1.0185,
      "step": 6982
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5728141069412231,
      "learning_rate": 1.8791107337477022e-07,
      "loss": 1.0034,
      "step": 6983
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2694242000579834,
      "learning_rate": 1.8733519473698557e-07,
      "loss": 0.8046,
      "step": 6984
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2298860549926758,
      "learning_rate": 1.867601830355098e-07,
      "loss": 0.7825,
      "step": 6985
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5706016421318054,
      "learning_rate": 1.8618603837392336e-07,
      "loss": 0.994,
      "step": 6986
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2241120338439941,
      "learning_rate": 1.8561276085564906e-07,
      "loss": 0.7751,
      "step": 6987
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.228574275970459,
      "learning_rate": 1.8504035058395708e-07,
      "loss": 0.7845,
      "step": 6988
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2220734357833862,
      "learning_rate": 1.8446880766195828e-07,
      "loss": 0.8334,
      "step": 6989
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5784609317779541,
      "learning_rate": 1.8389813219260865e-07,
      "loss": 1.0042,
      "step": 6990
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1910916566848755,
      "learning_rate": 1.8332832427870817e-07,
      "loss": 0.8091,
      "step": 6991
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5750253796577454,
      "learning_rate": 1.827593840228986e-07,
      "loss": 1.031,
      "step": 6992
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3148775100708008,
      "learning_rate": 1.8219131152766855e-07,
      "loss": 0.7939,
      "step": 6993
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5663288235664368,
      "learning_rate": 1.816241068953467e-07,
      "loss": 1.0018,
      "step": 6994
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.459954857826233,
      "learning_rate": 1.8105777022810911e-07,
      "loss": 0.8252,
      "step": 6995
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5743618011474609,
      "learning_rate": 1.8049230162797305e-07,
      "loss": 1.0022,
      "step": 6996
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3274370431900024,
      "learning_rate": 1.799277011967987e-07,
      "loss": 0.8445,
      "step": 6997
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2214587926864624,
      "learning_rate": 1.7936396903629304e-07,
      "loss": 0.7116,
      "step": 6998
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2930915355682373,
      "learning_rate": 1.788011052480032e-07,
      "loss": 0.8086,
      "step": 6999
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2270666360855103,
      "learning_rate": 1.7823910993332306e-07,
      "loss": 0.7657,
      "step": 7000
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5732661485671997,
      "learning_rate": 1.776779831934877e-07,
      "loss": 1.0161,
      "step": 7001
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5762140154838562,
      "learning_rate": 1.771177251295758e-07,
      "loss": 1.0129,
      "step": 7002
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.588917076587677,
      "learning_rate": 1.7655833584251157e-07,
      "loss": 1.0104,
      "step": 7003
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2556419372558594,
      "learning_rate": 1.759998154330611e-07,
      "loss": 0.78,
      "step": 7004
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3553391695022583,
      "learning_rate": 1.7544216400183277e-07,
      "loss": 0.7303,
      "step": 7005
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.271083950996399,
      "learning_rate": 1.748853816492818e-07,
      "loss": 0.7979,
      "step": 7006
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3056391477584839,
      "learning_rate": 1.7432946847570408e-07,
      "loss": 0.7809,
      "step": 7007
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2208571434020996,
      "learning_rate": 1.737744245812395e-07,
      "loss": 0.8411,
      "step": 7008
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2530319690704346,
      "learning_rate": 1.7322025006587306e-07,
      "loss": 0.7449,
      "step": 7009
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3041489124298096,
      "learning_rate": 1.7266694502943048e-07,
      "loss": 0.7751,
      "step": 7010
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5701492428779602,
      "learning_rate": 1.721145095715826e-07,
      "loss": 1.0014,
      "step": 7011
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2845494747161865,
      "learning_rate": 1.7156294379184257e-07,
      "loss": 0.7773,
      "step": 7012
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5681664943695068,
      "learning_rate": 1.7101224778956927e-07,
      "loss": 1.0014,
      "step": 7013
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.580683708190918,
      "learning_rate": 1.704624216639611e-07,
      "loss": 1.0194,
      "step": 7014
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.312604546546936,
      "learning_rate": 1.6991346551406275e-07,
      "loss": 0.8213,
      "step": 7015
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2584666013717651,
      "learning_rate": 1.693653794387612e-07,
      "loss": 0.7473,
      "step": 7016
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2517735958099365,
      "learning_rate": 1.6881816353678583e-07,
      "loss": 0.7287,
      "step": 7017
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5810706615447998,
      "learning_rate": 1.682718179067122e-07,
      "loss": 1.0209,
      "step": 7018
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2040737867355347,
      "learning_rate": 1.6772634264695553e-07,
      "loss": 0.7566,
      "step": 7019
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5839853882789612,
      "learning_rate": 1.6718173785577496e-07,
      "loss": 1.0056,
      "step": 7020
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5672755837440491,
      "learning_rate": 1.6663800363127592e-07,
      "loss": 1.0222,
      "step": 7021
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2901657819747925,
      "learning_rate": 1.6609514007140282e-07,
      "loss": 0.7492,
      "step": 7022
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.314499855041504,
      "learning_rate": 1.655531472739469e-07,
      "loss": 0.7849,
      "step": 7023
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2942478656768799,
      "learning_rate": 1.6501202533653948e-07,
      "loss": 0.7839,
      "step": 7024
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4342368841171265,
      "learning_rate": 1.644717743566565e-07,
      "loss": 0.7308,
      "step": 7025
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.23953115940094,
      "learning_rate": 1.6393239443161735e-07,
      "loss": 0.729,
      "step": 7026
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5807210206985474,
      "learning_rate": 1.6339388565858317e-07,
      "loss": 0.9764,
      "step": 7027
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5737016201019287,
      "learning_rate": 1.6285624813456025e-07,
      "loss": 1.007,
      "step": 7028
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5674349069595337,
      "learning_rate": 1.6231948195639613e-07,
      "loss": 0.9868,
      "step": 7029
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2541131973266602,
      "learning_rate": 1.6178358722078125e-07,
      "loss": 0.767,
      "step": 7030
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2775415182113647,
      "learning_rate": 1.612485640242506e-07,
      "loss": 0.7548,
      "step": 7031
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.317542314529419,
      "learning_rate": 1.60714412463181e-07,
      "loss": 0.7499,
      "step": 7032
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5722711086273193,
      "learning_rate": 1.601811326337932e-07,
      "loss": 0.9905,
      "step": 7033
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5715471506118774,
      "learning_rate": 1.5964872463214985e-07,
      "loss": 1.0044,
      "step": 7034
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.561942458152771,
      "learning_rate": 1.5911718855415648e-07,
      "loss": 0.9981,
      "step": 7035
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5739449262619019,
      "learning_rate": 1.5858652449556368e-07,
      "loss": 1.0006,
      "step": 7036
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1866217851638794,
      "learning_rate": 1.5805673255196164e-07,
      "loss": 0.7292,
      "step": 7037
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.261564016342163,
      "learning_rate": 1.5752781281878626e-07,
      "loss": 0.7397,
      "step": 7038
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2468185424804688,
      "learning_rate": 1.5699976539131467e-07,
      "loss": 0.7629,
      "step": 7039
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.274024486541748,
      "learning_rate": 1.56472590364668e-07,
      "loss": 0.7962,
      "step": 7040
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.269607663154602,
      "learning_rate": 1.5594628783380917e-07,
      "loss": 0.7689,
      "step": 7041
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5474952459335327,
      "learning_rate": 1.554208578935451e-07,
      "loss": 0.7307,
      "step": 7042
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2558015584945679,
      "learning_rate": 1.5489630063852402e-07,
      "loss": 0.7578,
      "step": 7043
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.203432559967041,
      "learning_rate": 1.5437261616323861e-07,
      "loss": 0.7411,
      "step": 7044
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5831084847450256,
      "learning_rate": 1.5384980456202348e-07,
      "loss": 0.9987,
      "step": 7045
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.23021399974823,
      "learning_rate": 1.533278659290549e-07,
      "loss": 0.8168,
      "step": 7046
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2430833578109741,
      "learning_rate": 1.5280680035835438e-07,
      "loss": 0.7106,
      "step": 7047
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2916924953460693,
      "learning_rate": 1.522866079437846e-07,
      "loss": 0.8011,
      "step": 7048
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3003385066986084,
      "learning_rate": 1.5176728877905e-07,
      "loss": 0.8305,
      "step": 7049
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.244003176689148,
      "learning_rate": 1.512488429577008e-07,
      "loss": 0.7517,
      "step": 7050
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7333214282989502,
      "learning_rate": 1.507312705731262e-07,
      "loss": 0.7818,
      "step": 7051
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.232099175453186,
      "learning_rate": 1.5021457171856103e-07,
      "loss": 0.8781,
      "step": 7052
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2674227952957153,
      "learning_rate": 1.4969874648708082e-07,
      "loss": 0.7418,
      "step": 7053
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2697936296463013,
      "learning_rate": 1.4918379497160517e-07,
      "loss": 0.8042,
      "step": 7054
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2112449407577515,
      "learning_rate": 1.4866971726489486e-07,
      "loss": 0.7521,
      "step": 7055
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2264424562454224,
      "learning_rate": 1.4815651345955474e-07,
      "loss": 0.7571,
      "step": 7056
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.252151608467102,
      "learning_rate": 1.4764418364803134e-07,
      "loss": 0.7935,
      "step": 7057
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5754520893096924,
      "learning_rate": 1.4713272792261368e-07,
      "loss": 0.998,
      "step": 7058
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3270137310028076,
      "learning_rate": 1.466221463754336e-07,
      "loss": 0.7991,
      "step": 7059
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3478715419769287,
      "learning_rate": 1.4611243909846529e-07,
      "loss": 0.7516,
      "step": 7060
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1910123825073242,
      "learning_rate": 1.4560360618352643e-07,
      "loss": 0.7343,
      "step": 7061
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5682963728904724,
      "learning_rate": 1.4509564772227535e-07,
      "loss": 0.9885,
      "step": 7062
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2862368822097778,
      "learning_rate": 1.445885638062139e-07,
      "loss": 0.7883,
      "step": 7063
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2309409379959106,
      "learning_rate": 1.4408235452668783e-07,
      "loss": 0.7549,
      "step": 7064
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2083839178085327,
      "learning_rate": 1.435770199748826e-07,
      "loss": 0.7921,
      "step": 7065
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5735665559768677,
      "learning_rate": 1.4307256024182703e-07,
      "loss": 1.0027,
      "step": 7066
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5754279494285583,
      "learning_rate": 1.4256897541839343e-07,
      "loss": 0.9873,
      "step": 7067
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2788057327270508,
      "learning_rate": 1.420662655952959e-07,
      "loss": 0.7643,
      "step": 7068
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2466046810150146,
      "learning_rate": 1.4156443086309034e-07,
      "loss": 0.7823,
      "step": 7069
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.500586748123169,
      "learning_rate": 1.4106347131217612e-07,
      "loss": 0.7778,
      "step": 7070
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2359322309494019,
      "learning_rate": 1.4056338703279326e-07,
      "loss": 0.7537,
      "step": 7071
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5605948567390442,
      "learning_rate": 1.4006417811502582e-07,
      "loss": 1.0224,
      "step": 7072
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2038037776947021,
      "learning_rate": 1.3956584464879964e-07,
      "loss": 0.7864,
      "step": 7073
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3246732950210571,
      "learning_rate": 1.3906838672388233e-07,
      "loss": 0.7897,
      "step": 7074
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2526482343673706,
      "learning_rate": 1.3857180442988506e-07,
      "loss": 0.7991,
      "step": 7075
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5715946555137634,
      "learning_rate": 1.3807609785625897e-07,
      "loss": 0.9986,
      "step": 7076
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2875423431396484,
      "learning_rate": 1.3758126709229992e-07,
      "loss": 0.7668,
      "step": 7077
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5544647574424744,
      "learning_rate": 1.370873122271449e-07,
      "loss": 1.009,
      "step": 7078
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5561962127685547,
      "learning_rate": 1.3659423334977274e-07,
      "loss": 0.9985,
      "step": 7079
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.275428056716919,
      "learning_rate": 1.3610203054900571e-07,
      "loss": 0.7586,
      "step": 7080
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3175348043441772,
      "learning_rate": 1.3561070391350573e-07,
      "loss": 0.7656,
      "step": 7081
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2516887187957764,
      "learning_rate": 1.351202535317808e-07,
      "loss": 0.7538,
      "step": 7082
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2900279760360718,
      "learning_rate": 1.3463067949217756e-07,
      "loss": 0.8192,
      "step": 7083
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2846121788024902,
      "learning_rate": 1.3414198188288708e-07,
      "loss": 0.8288,
      "step": 7084
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2860984802246094,
      "learning_rate": 1.336541607919406e-07,
      "loss": 0.7516,
      "step": 7085
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2572460174560547,
      "learning_rate": 1.3316721630721285e-07,
      "loss": 0.85,
      "step": 7086
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2508713006973267,
      "learning_rate": 1.3268114851642032e-07,
      "loss": 0.7997,
      "step": 7087
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.396107792854309,
      "learning_rate": 1.3219595750712133e-07,
      "loss": 0.732,
      "step": 7088
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.278878927230835,
      "learning_rate": 1.3171164336671705e-07,
      "loss": 0.8109,
      "step": 7089
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5720974206924438,
      "learning_rate": 1.3122820618244936e-07,
      "loss": 0.9926,
      "step": 7090
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2171379327774048,
      "learning_rate": 1.3074564604140362e-07,
      "loss": 0.744,
      "step": 7091
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.210352897644043,
      "learning_rate": 1.3026396303050637e-07,
      "loss": 0.8076,
      "step": 7092
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.195624589920044,
      "learning_rate": 1.2978315723652601e-07,
      "loss": 0.7832,
      "step": 7093
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2993748188018799,
      "learning_rate": 1.2930322874607272e-07,
      "loss": 0.7712,
      "step": 7094
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3378344774246216,
      "learning_rate": 1.288241776455995e-07,
      "loss": 0.7561,
      "step": 7095
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1965129375457764,
      "learning_rate": 1.2834600402140128e-07,
      "loss": 0.7389,
      "step": 7096
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2444480657577515,
      "learning_rate": 1.278687079596147e-07,
      "loss": 0.7767,
      "step": 7097
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3309383392333984,
      "learning_rate": 1.2739228954621764e-07,
      "loss": 0.7894,
      "step": 7098
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5628092885017395,
      "learning_rate": 1.2691674886702976e-07,
      "loss": 0.9921,
      "step": 7099
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1849347352981567,
      "learning_rate": 1.2644208600771534e-07,
      "loss": 0.7492,
      "step": 7100
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2432551383972168,
      "learning_rate": 1.25968301053776e-07,
      "loss": 0.8492,
      "step": 7101
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7567859888076782,
      "learning_rate": 1.2549539409055955e-07,
      "loss": 0.7952,
      "step": 7102
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3274284601211548,
      "learning_rate": 1.250233652032534e-07,
      "loss": 0.7561,
      "step": 7103
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2822901010513306,
      "learning_rate": 1.2455221447688571e-07,
      "loss": 0.75,
      "step": 7104
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5505884885787964,
      "learning_rate": 1.2408194199633016e-07,
      "loss": 1.0172,
      "step": 7105
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5761300325393677,
      "learning_rate": 1.2361254784629794e-07,
      "loss": 0.9842,
      "step": 7106
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2652887105941772,
      "learning_rate": 1.2314403211134528e-07,
      "loss": 0.7169,
      "step": 7107
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5794368982315063,
      "learning_rate": 1.2267639487586858e-07,
      "loss": 1.0082,
      "step": 7108
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2556899785995483,
      "learning_rate": 1.2220963622410598e-07,
      "loss": 0.8164,
      "step": 7109
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.248097538948059,
      "learning_rate": 1.2174375624013858e-07,
      "loss": 0.7579,
      "step": 7110
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1637647151947021,
      "learning_rate": 1.2127875500788754e-07,
      "loss": 0.7124,
      "step": 7111
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2230685949325562,
      "learning_rate": 1.2081463261111704e-07,
      "loss": 0.8179,
      "step": 7112
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2609987258911133,
      "learning_rate": 1.2035138913343236e-07,
      "loss": 0.8032,
      "step": 7113
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1769816875457764,
      "learning_rate": 1.1988902465828013e-07,
      "loss": 0.729,
      "step": 7114
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2806968688964844,
      "learning_rate": 1.194275392689498e-07,
      "loss": 0.7753,
      "step": 7115
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4314970970153809,
      "learning_rate": 1.1896693304857044e-07,
      "loss": 0.8078,
      "step": 7116
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5549329519271851,
      "learning_rate": 1.1850720608011513e-07,
      "loss": 0.9989,
      "step": 7117
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2993996143341064,
      "learning_rate": 1.1804835844639761e-07,
      "loss": 0.7791,
      "step": 7118
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.256688117980957,
      "learning_rate": 1.175903902300718e-07,
      "loss": 0.7631,
      "step": 7119
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2359033823013306,
      "learning_rate": 1.1713330151363555e-07,
      "loss": 0.6983,
      "step": 7120
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3053051233291626,
      "learning_rate": 1.1667709237942637e-07,
      "loss": 0.766,
      "step": 7121
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2871901988983154,
      "learning_rate": 1.1622176290962517e-07,
      "loss": 0.8044,
      "step": 7122
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2189171314239502,
      "learning_rate": 1.1576731318625301e-07,
      "loss": 0.7953,
      "step": 7123
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2637745141983032,
      "learning_rate": 1.1531374329117217e-07,
      "loss": 0.7724,
      "step": 7124
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2414933443069458,
      "learning_rate": 1.1486105330608844e-07,
      "loss": 0.7678,
      "step": 7125
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2197929620742798,
      "learning_rate": 1.1440924331254655e-07,
      "loss": 0.786,
      "step": 7126
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3142056465148926,
      "learning_rate": 1.1395831339193419e-07,
      "loss": 0.8434,
      "step": 7127
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5608037710189819,
      "learning_rate": 1.1350826362548028e-07,
      "loss": 1.0236,
      "step": 7128
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3057233095169067,
      "learning_rate": 1.1305909409425608e-07,
      "loss": 0.8309,
      "step": 7129
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.291688323020935,
      "learning_rate": 1.1261080487917241e-07,
      "loss": 0.778,
      "step": 7130
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2778558731079102,
      "learning_rate": 1.1216339606098247e-07,
      "loss": 0.8282,
      "step": 7131
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2612894773483276,
      "learning_rate": 1.1171686772028123e-07,
      "loss": 0.7804,
      "step": 7132
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5706038475036621,
      "learning_rate": 1.112712199375049e-07,
      "loss": 1.0058,
      "step": 7133
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3005239963531494,
      "learning_rate": 1.1082645279293036e-07,
      "loss": 0.8312,
      "step": 7134
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2146899700164795,
      "learning_rate": 1.1038256636667687e-07,
      "loss": 0.774,
      "step": 7135
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2178218364715576,
      "learning_rate": 1.0993956073870438e-07,
      "loss": 0.8015,
      "step": 7136
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5613469481468201,
      "learning_rate": 1.0949743598881401e-07,
      "loss": 0.9991,
      "step": 7137
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2772176265716553,
      "learning_rate": 1.0905619219664932e-07,
      "loss": 0.7623,
      "step": 7138
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5728772878646851,
      "learning_rate": 1.0861582944169335e-07,
      "loss": 1.01,
      "step": 7139
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2996430397033691,
      "learning_rate": 1.0817634780327213e-07,
      "loss": 0.7242,
      "step": 7140
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3239085674285889,
      "learning_rate": 1.0773774736055287e-07,
      "loss": 0.7765,
      "step": 7141
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2305928468704224,
      "learning_rate": 1.0730002819254236e-07,
      "loss": 0.7969,
      "step": 7142
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2764052152633667,
      "learning_rate": 1.0686319037809034e-07,
      "loss": 0.7578,
      "step": 7143
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2568796873092651,
      "learning_rate": 1.0642723399588662e-07,
      "loss": 0.7396,
      "step": 7144
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3146469593048096,
      "learning_rate": 1.0599215912446447e-07,
      "loss": 0.7696,
      "step": 7145
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.288205862045288,
      "learning_rate": 1.0555796584219514e-07,
      "loss": 0.7441,
      "step": 7146
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5707981586456299,
      "learning_rate": 1.0512465422729267e-07,
      "loss": 1.0133,
      "step": 7147
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2686504125595093,
      "learning_rate": 1.0469222435781357e-07,
      "loss": 0.7586,
      "step": 7148
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5547546148300171,
      "learning_rate": 1.0426067631165271e-07,
      "loss": 1.0002,
      "step": 7149
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2974121570587158,
      "learning_rate": 1.0383001016654848e-07,
      "loss": 0.7645,
      "step": 7150
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2760910987854004,
      "learning_rate": 1.0340022600007992e-07,
      "loss": 0.7998,
      "step": 7151
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5811741948127747,
      "learning_rate": 1.0297132388966513e-07,
      "loss": 0.992,
      "step": 7152
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2780338525772095,
      "learning_rate": 1.0254330391256729e-07,
      "loss": 0.8142,
      "step": 7153
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.245665192604065,
      "learning_rate": 1.0211616614588693e-07,
      "loss": 0.7719,
      "step": 7154
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4243202209472656,
      "learning_rate": 1.0168991066656698e-07,
      "loss": 0.7778,
      "step": 7155
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2606933116912842,
      "learning_rate": 1.0126453755139266e-07,
      "loss": 0.7688,
      "step": 7156
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3628782033920288,
      "learning_rate": 1.0084004687698823e-07,
      "loss": 0.8157,
      "step": 7157
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3358063697814941,
      "learning_rate": 1.0041643871982032e-07,
      "loss": 0.8297,
      "step": 7158
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.197135090827942,
      "learning_rate": 9.999371315619622e-08,
      "loss": 0.7004,
      "step": 7159
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2454767227172852,
      "learning_rate": 9.957187026226389e-08,
      "loss": 0.774,
      "step": 7160
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2214688062667847,
      "learning_rate": 9.915091011401313e-08,
      "loss": 0.8127,
      "step": 7161
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5709989666938782,
      "learning_rate": 9.87308327872738e-08,
      "loss": 1.0035,
      "step": 7162
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5598495602607727,
      "learning_rate": 9.831163835771706e-08,
      "loss": 1.0188,
      "step": 7163
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5594305396080017,
      "learning_rate": 9.789332690085585e-08,
      "loss": 1.018,
      "step": 7164
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5527370572090149,
      "learning_rate": 9.74758984920421e-08,
      "loss": 1.0221,
      "step": 7165
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2508347034454346,
      "learning_rate": 9.70593532064712e-08,
      "loss": 0.7844,
      "step": 7166
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.229332447052002,
      "learning_rate": 9.664369111917704e-08,
      "loss": 0.7359,
      "step": 7167
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2316197156906128,
      "learning_rate": 9.622891230503584e-08,
      "loss": 0.7658,
      "step": 7168
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2668105363845825,
      "learning_rate": 9.581501683876504e-08,
      "loss": 0.756,
      "step": 7169
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2360731363296509,
      "learning_rate": 9.540200479492112e-08,
      "loss": 0.8001,
      "step": 7170
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.20785391330719,
      "learning_rate": 9.498987624790401e-08,
      "loss": 0.7623,
      "step": 7171
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2390047311782837,
      "learning_rate": 9.457863127195099e-08,
      "loss": 0.7904,
      "step": 7172
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.239530086517334,
      "learning_rate": 9.416826994114447e-08,
      "loss": 0.7545,
      "step": 7173
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.555097758769989,
      "learning_rate": 9.375879232940477e-08,
      "loss": 0.9911,
      "step": 7174
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2265547513961792,
      "learning_rate": 9.335019851049288e-08,
      "loss": 0.7717,
      "step": 7175
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3205231428146362,
      "learning_rate": 9.294248855801269e-08,
      "loss": 0.7519,
      "step": 7176
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2117925882339478,
      "learning_rate": 9.253566254540603e-08,
      "loss": 0.7237,
      "step": 7177
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.287742018699646,
      "learning_rate": 9.21297205459587e-08,
      "loss": 0.7819,
      "step": 7178
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5691709518432617,
      "learning_rate": 9.172466263279555e-08,
      "loss": 1.0152,
      "step": 7179
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5758416652679443,
      "learning_rate": 9.132048887888101e-08,
      "loss": 1.011,
      "step": 7180
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2813704013824463,
      "learning_rate": 9.091719935702236e-08,
      "loss": 0.7626,
      "step": 7181
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2640337944030762,
      "learning_rate": 9.051479413986708e-08,
      "loss": 0.7501,
      "step": 7182
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.34196138381958,
      "learning_rate": 9.01132732999016e-08,
      "loss": 0.7517,
      "step": 7183
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5693212151527405,
      "learning_rate": 8.971263690945587e-08,
      "loss": 1.0291,
      "step": 7184
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2196894884109497,
      "learning_rate": 8.931288504069879e-08,
      "loss": 0.7646,
      "step": 7185
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2376229763031006,
      "learning_rate": 8.891401776564001e-08,
      "loss": 0.7657,
      "step": 7186
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3058072328567505,
      "learning_rate": 8.851603515613038e-08,
      "loss": 0.7516,
      "step": 7187
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5712016820907593,
      "learning_rate": 8.811893728386033e-08,
      "loss": 0.9799,
      "step": 7188
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2618329524993896,
      "learning_rate": 8.772272422036265e-08,
      "loss": 0.7673,
      "step": 7189
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.331047534942627,
      "learning_rate": 8.732739603700857e-08,
      "loss": 0.7784,
      "step": 7190
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2726683616638184,
      "learning_rate": 8.693295280501224e-08,
      "loss": 0.801,
      "step": 7191
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.261330485343933,
      "learning_rate": 8.653939459542682e-08,
      "loss": 0.7506,
      "step": 7192
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5751417875289917,
      "learning_rate": 8.614672147914616e-08,
      "loss": 0.9798,
      "step": 7193
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2836430072784424,
      "learning_rate": 8.57549335269059e-08,
      "loss": 0.7127,
      "step": 7194
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2843540906906128,
      "learning_rate": 8.536403080928013e-08,
      "loss": 0.7963,
      "step": 7195
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2353520393371582,
      "learning_rate": 8.497401339668531e-08,
      "loss": 0.7836,
      "step": 7196
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2865768671035767,
      "learning_rate": 8.458488135937859e-08,
      "loss": 0.7565,
      "step": 7197
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2588807344436646,
      "learning_rate": 8.419663476745499e-08,
      "loss": 0.7658,
      "step": 7198
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.575009822845459,
      "learning_rate": 8.3809273690853e-08,
      "loss": 1.0096,
      "step": 7199
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2193058729171753,
      "learning_rate": 8.342279819935073e-08,
      "loss": 0.7677,
      "step": 7200
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5749539732933044,
      "learning_rate": 8.303720836256579e-08,
      "loss": 1.0006,
      "step": 7201
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2836368083953857,
      "learning_rate": 8.265250424995763e-08,
      "loss": 0.8293,
      "step": 7202
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3510665893554688,
      "learning_rate": 8.22686859308247e-08,
      "loss": 0.8017,
      "step": 7203
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2589210271835327,
      "learning_rate": 8.188575347430783e-08,
      "loss": 0.7597,
      "step": 7204
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2441874742507935,
      "learning_rate": 8.150370694938626e-08,
      "loss": 0.8103,
      "step": 7205
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5637435913085938,
      "learning_rate": 8.112254642487993e-08,
      "loss": 1.0026,
      "step": 7206
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2448853254318237,
      "learning_rate": 8.074227196945061e-08,
      "loss": 0.758,
      "step": 7207
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2975795269012451,
      "learning_rate": 8.03628836515996e-08,
      "loss": 0.766,
      "step": 7208
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5661070346832275,
      "learning_rate": 7.998438153966837e-08,
      "loss": 1.0123,
      "step": 7209
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5683702230453491,
      "learning_rate": 7.960676570183901e-08,
      "loss": 1.0066,
      "step": 7210
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5604120492935181,
      "learning_rate": 7.923003620613379e-08,
      "loss": 1.0053,
      "step": 7211
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1495403051376343,
      "learning_rate": 7.885419312041509e-08,
      "loss": 0.7653,
      "step": 7212
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2515181303024292,
      "learning_rate": 7.847923651238654e-08,
      "loss": 0.7474,
      "step": 7213
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1812450885772705,
      "learning_rate": 7.810516644959131e-08,
      "loss": 0.7523,
      "step": 7214
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5530091524124146,
      "learning_rate": 7.773198299941276e-08,
      "loss": 1.0329,
      "step": 7215
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5665895342826843,
      "learning_rate": 7.735968622907486e-08,
      "loss": 0.9959,
      "step": 7216
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2190781831741333,
      "learning_rate": 7.698827620564287e-08,
      "loss": 0.7944,
      "step": 7217
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5521810054779053,
      "learning_rate": 7.661775299601936e-08,
      "loss": 1.0149,
      "step": 7218
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.222106695175171,
      "learning_rate": 7.624811666695098e-08,
      "loss": 0.7297,
      "step": 7219
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2498116493225098,
      "learning_rate": 7.587936728502166e-08,
      "loss": 0.734,
      "step": 7220
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.567383348941803,
      "learning_rate": 7.55115049166566e-08,
      "loss": 0.9997,
      "step": 7221
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5614198446273804,
      "learning_rate": 7.514452962812113e-08,
      "loss": 1.0178,
      "step": 7222
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1826279163360596,
      "learning_rate": 7.477844148552127e-08,
      "loss": 0.7541,
      "step": 7223
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5614771842956543,
      "learning_rate": 7.441324055480314e-08,
      "loss": 1.0322,
      "step": 7224
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1776273250579834,
      "learning_rate": 7.404892690175247e-08,
      "loss": 0.7868,
      "step": 7225
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1970936059951782,
      "learning_rate": 7.368550059199453e-08,
      "loss": 0.7719,
      "step": 7226
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2676191329956055,
      "learning_rate": 7.332296169099695e-08,
      "loss": 0.7728,
      "step": 7227
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2053229808807373,
      "learning_rate": 7.296131026406472e-08,
      "loss": 0.7523,
      "step": 7228
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2316893339157104,
      "learning_rate": 7.260054637634628e-08,
      "loss": 0.7752,
      "step": 7229
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5641615986824036,
      "learning_rate": 7.224067009282742e-08,
      "loss": 1.0151,
      "step": 7230
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.258497714996338,
      "learning_rate": 7.188168147833464e-08,
      "loss": 0.7168,
      "step": 7231
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1886992454528809,
      "learning_rate": 7.152358059753506e-08,
      "loss": 0.7497,
      "step": 7232
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1994702816009521,
      "learning_rate": 7.1166367514936e-08,
      "loss": 0.7418,
      "step": 7233
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5702007412910461,
      "learning_rate": 7.081004229488431e-08,
      "loss": 1.0064,
      "step": 7234
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2678492069244385,
      "learning_rate": 7.045460500156753e-08,
      "loss": 0.7972,
      "step": 7235
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2311395406723022,
      "learning_rate": 7.010005569901223e-08,
      "loss": 0.7634,
      "step": 7236
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2753602266311646,
      "learning_rate": 6.974639445108621e-08,
      "loss": 0.7756,
      "step": 7237
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2633306980133057,
      "learning_rate": 6.93936213214963e-08,
      "loss": 0.758,
      "step": 7238
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5363807678222656,
      "learning_rate": 6.904173637378997e-08,
      "loss": 0.7636,
      "step": 7239
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2838612794876099,
      "learning_rate": 6.869073967135487e-08,
      "loss": 0.7667,
      "step": 7240
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5769206285476685,
      "learning_rate": 6.834063127741764e-08,
      "loss": 1.0181,
      "step": 7241
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1771519184112549,
      "learning_rate": 6.799141125504671e-08,
      "loss": 0.7905,
      "step": 7242
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2482746839523315,
      "learning_rate": 6.76430796671479e-08,
      "loss": 0.7682,
      "step": 7243
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.286785364151001,
      "learning_rate": 6.729563657646931e-08,
      "loss": 0.8035,
      "step": 7244
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.206430435180664,
      "learning_rate": 6.694908204559758e-08,
      "loss": 0.7413,
      "step": 7245
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2527387142181396,
      "learning_rate": 6.660341613696053e-08,
      "loss": 0.7633,
      "step": 7246
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1854509115219116,
      "learning_rate": 6.625863891282447e-08,
      "loss": 0.7626,
      "step": 7247
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2937471866607666,
      "learning_rate": 6.591475043529638e-08,
      "loss": 0.8047,
      "step": 7248
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5771689414978027,
      "learning_rate": 6.557175076632394e-08,
      "loss": 1.005,
      "step": 7249
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5641588568687439,
      "learning_rate": 6.522963996769272e-08,
      "loss": 1.0154,
      "step": 7250
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.18429434299469,
      "learning_rate": 6.488841810103008e-08,
      "loss": 0.7765,
      "step": 7251
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2729463577270508,
      "learning_rate": 6.454808522780298e-08,
      "loss": 0.7606,
      "step": 7252
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2029266357421875,
      "learning_rate": 6.420864140931682e-08,
      "loss": 0.7465,
      "step": 7253
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.231174111366272,
      "learning_rate": 6.387008670671769e-08,
      "loss": 0.783,
      "step": 7254
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2343902587890625,
      "learning_rate": 6.353242118099234e-08,
      "loss": 0.7584,
      "step": 7255
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3370429277420044,
      "learning_rate": 6.319564489296603e-08,
      "loss": 0.7848,
      "step": 7256
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2463997602462769,
      "learning_rate": 6.285975790330522e-08,
      "loss": 0.7957,
      "step": 7257
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2370854616165161,
      "learning_rate": 6.252476027251487e-08,
      "loss": 0.7218,
      "step": 7258
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.258797526359558,
      "learning_rate": 6.219065206094e-08,
      "loss": 0.7709,
      "step": 7259
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3256345987319946,
      "learning_rate": 6.185743332876637e-08,
      "loss": 0.8194,
      "step": 7260
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1916167736053467,
      "learning_rate": 6.15251041360182e-08,
      "loss": 0.762,
      "step": 7261
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.268054485321045,
      "learning_rate": 6.119366454256037e-08,
      "loss": 0.7666,
      "step": 7262
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2051132917404175,
      "learning_rate": 6.086311460809791e-08,
      "loss": 0.7519,
      "step": 7263
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2753477096557617,
      "learning_rate": 6.053345439217318e-08,
      "loss": 0.8251,
      "step": 7264
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2559223175048828,
      "learning_rate": 6.020468395417145e-08,
      "loss": 0.7966,
      "step": 7265
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2587777376174927,
      "learning_rate": 5.987680335331592e-08,
      "loss": 0.7677,
      "step": 7266
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4071860313415527,
      "learning_rate": 5.954981264866988e-08,
      "loss": 0.8017,
      "step": 7267
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2064377069473267,
      "learning_rate": 5.922371189913567e-08,
      "loss": 0.7805,
      "step": 7268
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2758687734603882,
      "learning_rate": 5.889850116345686e-08,
      "loss": 0.7578,
      "step": 7269
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2665337324142456,
      "learning_rate": 5.857418050021491e-08,
      "loss": 0.6943,
      "step": 7270
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2251302003860474,
      "learning_rate": 5.825074996783253e-08,
      "loss": 0.8031,
      "step": 7271
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5735700130462646,
      "learning_rate": 5.7928209624570356e-08,
      "loss": 0.9834,
      "step": 7272
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2161065340042114,
      "learning_rate": 5.760655952853078e-08,
      "loss": 0.7643,
      "step": 7273
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2212857007980347,
      "learning_rate": 5.728579973765413e-08,
      "loss": 0.7719,
      "step": 7274
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5628806948661804,
      "learning_rate": 5.6965930309720866e-08,
      "loss": 1.0195,
      "step": 7275
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2751410007476807,
      "learning_rate": 5.664695130235154e-08,
      "loss": 0.7907,
      "step": 7276
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5656262636184692,
      "learning_rate": 5.63288627730052e-08,
      "loss": 1.0275,
      "step": 7277
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.334579348564148,
      "learning_rate": 5.601166477898212e-08,
      "loss": 0.8256,
      "step": 7278
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2237099409103394,
      "learning_rate": 5.569535737742049e-08,
      "loss": 0.7144,
      "step": 7279
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2355451583862305,
      "learning_rate": 5.537994062529917e-08,
      "loss": 0.84,
      "step": 7280
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2942644357681274,
      "learning_rate": 5.506541457943548e-08,
      "loss": 0.7979,
      "step": 7281
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5644991397857666,
      "learning_rate": 5.4751779296488006e-08,
      "loss": 0.9965,
      "step": 7282
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.273149847984314,
      "learning_rate": 5.4439034832953764e-08,
      "loss": 0.7224,
      "step": 7283
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5631281733512878,
      "learning_rate": 5.412718124516825e-08,
      "loss": 0.9936,
      "step": 7284
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3270840644836426,
      "learning_rate": 5.3816218589309297e-08,
      "loss": 0.7359,
      "step": 7285
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5710548758506775,
      "learning_rate": 5.350614692139211e-08,
      "loss": 0.9837,
      "step": 7286
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2440639734268188,
      "learning_rate": 5.319696629727089e-08,
      "loss": 0.7567,
      "step": 7287
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5633500218391418,
      "learning_rate": 5.2888676772641644e-08,
      "loss": 1.0167,
      "step": 7288
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3262183666229248,
      "learning_rate": 5.258127840303773e-08,
      "loss": 0.7798,
      "step": 7289
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2727769613265991,
      "learning_rate": 5.227477124383318e-08,
      "loss": 0.7515,
      "step": 7290
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2452826499938965,
      "learning_rate": 5.196915535024105e-08,
      "loss": 0.8114,
      "step": 7291
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2786604166030884,
      "learning_rate": 5.1664430777313424e-08,
      "loss": 0.8002,
      "step": 7292
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2661612033843994,
      "learning_rate": 5.1360597579943606e-08,
      "loss": 0.8143,
      "step": 7293
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5762155055999756,
      "learning_rate": 5.105765581286115e-08,
      "loss": 1.0116,
      "step": 7294
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5748745799064636,
      "learning_rate": 5.075560553063852e-08,
      "loss": 1.0126,
      "step": 7295
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3924992084503174,
      "learning_rate": 5.045444678768496e-08,
      "loss": 0.7552,
      "step": 7296
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5741152167320251,
      "learning_rate": 5.0154179638250424e-08,
      "loss": 1.001,
      "step": 7297
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2326644659042358,
      "learning_rate": 4.985480413642496e-08,
      "loss": 0.7344,
      "step": 7298
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2988470792770386,
      "learning_rate": 4.955632033613544e-08,
      "loss": 0.724,
      "step": 7299
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5590829253196716,
      "learning_rate": 4.925872829114997e-08,
      "loss": 1.0172,
      "step": 7300
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.249543309211731,
      "learning_rate": 4.896202805507677e-08,
      "loss": 0.7615,
      "step": 7301
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1700316667556763,
      "learning_rate": 4.866621968136142e-08,
      "loss": 0.7288,
      "step": 7302
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2949941158294678,
      "learning_rate": 4.83713032232902e-08,
      "loss": 0.7932,
      "step": 7303
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2542035579681396,
      "learning_rate": 4.807727873398838e-08,
      "loss": 0.6698,
      "step": 7304
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5787044167518616,
      "learning_rate": 4.7784146266419695e-08,
      "loss": 1.0045,
      "step": 7305
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5701503753662109,
      "learning_rate": 4.7491905873389144e-08,
      "loss": 0.9838,
      "step": 7306
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5760389566421509,
      "learning_rate": 4.72005576075385e-08,
      "loss": 0.9988,
      "step": 7307
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2856543064117432,
      "learning_rate": 4.691010152135189e-08,
      "loss": 0.7813,
      "step": 7308
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2278836965560913,
      "learning_rate": 4.662053766714969e-08,
      "loss": 0.7303,
      "step": 7309
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.172978401184082,
      "learning_rate": 4.6331866097093505e-08,
      "loss": 0.7519,
      "step": 7310
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5689111948013306,
      "learning_rate": 4.6044086863183403e-08,
      "loss": 1.0051,
      "step": 7311
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2541755437850952,
      "learning_rate": 4.575720001725903e-08,
      "loss": 0.8036,
      "step": 7312
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.238824486732483,
      "learning_rate": 4.547120561099905e-08,
      "loss": 0.7431,
      "step": 7313
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5573675632476807,
      "learning_rate": 4.518610369592169e-08,
      "loss": 1.0305,
      "step": 7314
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5560218691825867,
      "learning_rate": 4.490189432338365e-08,
      "loss": 0.986,
      "step": 7315
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4816715717315674,
      "learning_rate": 4.461857754458232e-08,
      "loss": 0.8067,
      "step": 7316
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.579484224319458,
      "learning_rate": 4.4336153410552416e-08,
      "loss": 1.009,
      "step": 7317
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1839683055877686,
      "learning_rate": 4.405462197216937e-08,
      "loss": 0.7237,
      "step": 7318
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2338820695877075,
      "learning_rate": 4.377398328014704e-08,
      "loss": 0.7982,
      "step": 7319
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2967664003372192,
      "learning_rate": 4.349423738503888e-08,
      "loss": 0.7663,
      "step": 7320
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5733811259269714,
      "learning_rate": 4.321538433723682e-08,
      "loss": 1.0052,
      "step": 7321
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5632733106613159,
      "learning_rate": 4.293742418697344e-08,
      "loss": 1.0086,
      "step": 7322
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2319848537445068,
      "learning_rate": 4.26603569843187e-08,
      "loss": 0.8155,
      "step": 7323
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2341625690460205,
      "learning_rate": 4.238418277918266e-08,
      "loss": 0.7348,
      "step": 7324
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3274893760681152,
      "learning_rate": 4.210890162131387e-08,
      "loss": 0.7768,
      "step": 7325
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2140252590179443,
      "learning_rate": 4.1834513560301545e-08,
      "loss": 0.7291,
      "step": 7326
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3071541786193848,
      "learning_rate": 4.156101864557227e-08,
      "loss": 0.8123,
      "step": 7327
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5717107057571411,
      "learning_rate": 4.128841692639274e-08,
      "loss": 1.0131,
      "step": 7328
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2613635063171387,
      "learning_rate": 4.101670845186812e-08,
      "loss": 0.8137,
      "step": 7329
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.267999529838562,
      "learning_rate": 4.074589327094369e-08,
      "loss": 0.807,
      "step": 7330
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2434148788452148,
      "learning_rate": 4.0475971432402675e-08,
      "loss": 0.8258,
      "step": 7331
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4703402519226074,
      "learning_rate": 4.020694298486727e-08,
      "loss": 0.733,
      "step": 7332
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.320422649383545,
      "learning_rate": 3.993880797679983e-08,
      "loss": 0.7899,
      "step": 7333
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2368584871292114,
      "learning_rate": 3.967156645650172e-08,
      "loss": 0.7719,
      "step": 7334
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5574426651000977,
      "learning_rate": 3.940521847211221e-08,
      "loss": 1.002,
      "step": 7335
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2510218620300293,
      "learning_rate": 3.913976407161069e-08,
      "loss": 0.7804,
      "step": 7336
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2320683002471924,
      "learning_rate": 3.8875203302815024e-08,
      "loss": 0.7797,
      "step": 7337
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2606600522994995,
      "learning_rate": 3.861153621338154e-08,
      "loss": 0.7687,
      "step": 7338
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.216097116470337,
      "learning_rate": 3.8348762850807794e-08,
      "loss": 0.8045,
      "step": 7339
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5694848895072937,
      "learning_rate": 3.808688326242704e-08,
      "loss": 1.0159,
      "step": 7340
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1513087749481201,
      "learning_rate": 3.7825897495414874e-08,
      "loss": 0.6889,
      "step": 7341
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2743085622787476,
      "learning_rate": 3.756580559678369e-08,
      "loss": 0.7795,
      "step": 7342
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2165368795394897,
      "learning_rate": 3.730660761338545e-08,
      "loss": 0.7695,
      "step": 7343
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2090388536453247,
      "learning_rate": 3.7048303591911695e-08,
      "loss": 0.7695,
      "step": 7344
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5537940859794617,
      "learning_rate": 3.6790893578891874e-08,
      "loss": 1.0022,
      "step": 7345
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5618104934692383,
      "learning_rate": 3.6534377620695e-08,
      "loss": 1.0387,
      "step": 7346
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2479208707809448,
      "learning_rate": 3.627875576352857e-08,
      "loss": 0.8105,
      "step": 7347
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.25295889377594,
      "learning_rate": 3.602402805344018e-08,
      "loss": 0.7378,
      "step": 7348
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.238407015800476,
      "learning_rate": 3.5770194536314803e-08,
      "loss": 0.7476,
      "step": 7349
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2647438049316406,
      "learning_rate": 3.551725525787808e-08,
      "loss": 0.7628,
      "step": 7350
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2500615119934082,
      "learning_rate": 3.526521026369245e-08,
      "loss": 0.7683,
      "step": 7351
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2863566875457764,
      "learning_rate": 3.50140595991616e-08,
      "loss": 0.7634,
      "step": 7352
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1983613967895508,
      "learning_rate": 3.4763803309525446e-08,
      "loss": 0.7138,
      "step": 7353
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.183271050453186,
      "learning_rate": 3.451444143986571e-08,
      "loss": 0.7587,
      "step": 7354
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2248144149780273,
      "learning_rate": 3.42659740351009e-08,
      "loss": 0.7233,
      "step": 7355
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1938763856887817,
      "learning_rate": 3.401840113998855e-08,
      "loss": 0.7259,
      "step": 7356
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3767777681350708,
      "learning_rate": 3.3771722799126885e-08,
      "loss": 0.7096,
      "step": 7357
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.283310890197754,
      "learning_rate": 3.3525939056949786e-08,
      "loss": 0.735,
      "step": 7358
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2531009912490845,
      "learning_rate": 3.328104995773407e-08,
      "loss": 0.7479,
      "step": 7359
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5720188021659851,
      "learning_rate": 3.303705554559167e-08,
      "loss": 1.0013,
      "step": 7360
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5657689571380615,
      "learning_rate": 3.27939558644752e-08,
      "loss": 1.0306,
      "step": 7361
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.180722713470459,
      "learning_rate": 3.255175095817575e-08,
      "loss": 0.7988,
      "step": 7362
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5631593465805054,
      "learning_rate": 3.2310440870323426e-08,
      "loss": 1.0057,
      "step": 7363
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5665018558502197,
      "learning_rate": 3.207002564438677e-08,
      "loss": 1.0286,
      "step": 7364
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.279119849205017,
      "learning_rate": 3.183050532367338e-08,
      "loss": 0.7345,
      "step": 7365
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2534196376800537,
      "learning_rate": 3.159187995132984e-08,
      "loss": 0.7068,
      "step": 7366
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3180525302886963,
      "learning_rate": 3.135414957034122e-08,
      "loss": 0.8176,
      "step": 7367
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.181904911994934,
      "learning_rate": 3.1117314223531036e-08,
      "loss": 0.6996,
      "step": 7368
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3831583261489868,
      "learning_rate": 3.088137395356239e-08,
      "loss": 0.7917,
      "step": 7369
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3037112951278687,
      "learning_rate": 3.0646328802936274e-08,
      "loss": 0.7681,
      "step": 7370
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2341855764389038,
      "learning_rate": 3.0412178813993274e-08,
      "loss": 0.7967,
      "step": 7371
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3372960090637207,
      "learning_rate": 3.017892402891243e-08,
      "loss": 0.7673,
      "step": 7372
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2562566995620728,
      "learning_rate": 2.994656448971123e-08,
      "loss": 0.7608,
      "step": 7373
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1718307733535767,
      "learning_rate": 2.9715100238245642e-08,
      "loss": 0.7401,
      "step": 7374
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2044824361801147,
      "learning_rate": 2.948453131621176e-08,
      "loss": 0.7548,
      "step": 7375
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2297581434249878,
      "learning_rate": 2.9254857765142474e-08,
      "loss": 0.7941,
      "step": 7376
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.257210612297058,
      "learning_rate": 2.902607962641135e-08,
      "loss": 0.78,
      "step": 7377
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.319414734840393,
      "learning_rate": 2.8798196941228762e-08,
      "loss": 0.8257,
      "step": 7378
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2586110830307007,
      "learning_rate": 2.8571209750645202e-08,
      "loss": 0.8113,
      "step": 7379
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2030692100524902,
      "learning_rate": 2.8345118095549073e-08,
      "loss": 0.7869,
      "step": 7380
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2715171575546265,
      "learning_rate": 2.81199220166678e-08,
      "loss": 0.8195,
      "step": 7381
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5576772093772888,
      "learning_rate": 2.789562155456782e-08,
      "loss": 0.984,
      "step": 7382
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3403459787368774,
      "learning_rate": 2.7672216749652925e-08,
      "loss": 0.7366,
      "step": 7383
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3304492235183716,
      "learning_rate": 2.744970764216759e-08,
      "loss": 0.7393,
      "step": 7384
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3126716613769531,
      "learning_rate": 2.7228094272193086e-08,
      "loss": 0.7868,
      "step": 7385
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5615561604499817,
      "learning_rate": 2.70073766796497e-08,
      "loss": 1.0177,
      "step": 7386
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2444697618484497,
      "learning_rate": 2.6787554904297852e-08,
      "loss": 0.8082,
      "step": 7387
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5652669072151184,
      "learning_rate": 2.65686289857342e-08,
      "loss": 1.0009,
      "step": 7388
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5771145820617676,
      "learning_rate": 2.635059896339609e-08,
      "loss": 1.0217,
      "step": 7389
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5840969681739807,
      "learning_rate": 2.6133464876558768e-08,
      "loss": 1.0056,
      "step": 7390
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3764984607696533,
      "learning_rate": 2.591722676433539e-08,
      "loss": 0.7707,
      "step": 7391
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2576385736465454,
      "learning_rate": 2.570188466567869e-08,
      "loss": 0.7616,
      "step": 7392
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5597712993621826,
      "learning_rate": 2.5487438619379303e-08,
      "loss": 0.9818,
      "step": 7393
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3207708597183228,
      "learning_rate": 2.5273888664067437e-08,
      "loss": 0.7931,
      "step": 7394
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2220736742019653,
      "learning_rate": 2.5061234838210656e-08,
      "loss": 0.7824,
      "step": 7395
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2627556324005127,
      "learning_rate": 2.4849477180115544e-08,
      "loss": 0.7534,
      "step": 7396
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5753888487815857,
      "learning_rate": 2.463861572792825e-08,
      "loss": 1.0297,
      "step": 7397
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2733098268508911,
      "learning_rate": 2.4428650519631725e-08,
      "loss": 0.8062,
      "step": 7398
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2561826705932617,
      "learning_rate": 2.421958159304849e-08,
      "loss": 0.7952,
      "step": 7399
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2680838108062744,
      "learning_rate": 2.4011408985839535e-08,
      "loss": 0.7226,
      "step": 7400
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3324546813964844,
      "learning_rate": 2.380413273550486e-08,
      "loss": 0.7572,
      "step": 7401
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2905718088150024,
      "learning_rate": 2.3597752879381818e-08,
      "loss": 0.7416,
      "step": 7402
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.225698471069336,
      "learning_rate": 2.339226945464734e-08,
      "loss": 0.7774,
      "step": 7403
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2624272108078003,
      "learning_rate": 2.318768249831571e-08,
      "loss": 0.7668,
      "step": 7404
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5791789293289185,
      "learning_rate": 2.2983992047241334e-08,
      "loss": 0.9839,
      "step": 7405
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5753722190856934,
      "learning_rate": 2.278119813811652e-08,
      "loss": 1.0354,
      "step": 7406
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3244823217391968,
      "learning_rate": 2.257930080747095e-08,
      "loss": 0.777,
      "step": 7407
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2671468257904053,
      "learning_rate": 2.2378300091673853e-08,
      "loss": 0.7316,
      "step": 7408
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.568333625793457,
      "learning_rate": 2.2178196026933495e-08,
      "loss": 0.9974,
      "step": 7409
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2392429113388062,
      "learning_rate": 2.197898864929493e-08,
      "loss": 0.7645,
      "step": 7410
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2764290571212769,
      "learning_rate": 2.1780677994643895e-08,
      "loss": 0.7727,
      "step": 7411
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2156274318695068,
      "learning_rate": 2.1583264098702374e-08,
      "loss": 0.7768,
      "step": 7412
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2103575468063354,
      "learning_rate": 2.1386746997031914e-08,
      "loss": 0.7623,
      "step": 7413
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5768426656723022,
      "learning_rate": 2.1191126725031984e-08,
      "loss": 1.0154,
      "step": 7414
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1979036331176758,
      "learning_rate": 2.0996403317942172e-08,
      "loss": 0.745,
      "step": 7415
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2146711349487305,
      "learning_rate": 2.080257681083886e-08,
      "loss": 0.785,
      "step": 7416
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1999447345733643,
      "learning_rate": 2.0609647238636343e-08,
      "loss": 0.8027,
      "step": 7417
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.243836522102356,
      "learning_rate": 2.04176146360896e-08,
      "loss": 0.7541,
      "step": 7418
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1938802003860474,
      "learning_rate": 2.0226479037789292e-08,
      "loss": 0.7115,
      "step": 7419
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2516834735870361,
      "learning_rate": 2.0036240478167323e-08,
      "loss": 0.8083,
      "step": 7420
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2855100631713867,
      "learning_rate": 1.9846898991491283e-08,
      "loss": 0.7837,
      "step": 7421
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2351640462875366,
      "learning_rate": 1.965845461186944e-08,
      "loss": 0.7352,
      "step": 7422
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2459781169891357,
      "learning_rate": 1.947090737324686e-08,
      "loss": 0.7566,
      "step": 7423
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5660906434059143,
      "learning_rate": 1.9284257309408194e-08,
      "loss": 1.0198,
      "step": 7424
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5679795145988464,
      "learning_rate": 1.909850445397543e-08,
      "loss": 1.0071,
      "step": 7425
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5587965846061707,
      "learning_rate": 1.8913648840410136e-08,
      "loss": 1.013,
      "step": 7426
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2683254480361938,
      "learning_rate": 1.872969050201068e-08,
      "loss": 0.7877,
      "step": 7427
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3237617015838623,
      "learning_rate": 1.854662947191499e-08,
      "loss": 0.8158,
      "step": 7428
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2965842485427856,
      "learning_rate": 1.8364465783098918e-08,
      "loss": 0.7581,
      "step": 7429
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2880173921585083,
      "learning_rate": 1.8183199468377322e-08,
      "loss": 0.7685,
      "step": 7430
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.315850853919983,
      "learning_rate": 1.800283056040242e-08,
      "loss": 0.7554,
      "step": 7431
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2129253149032593,
      "learning_rate": 1.7823359091664882e-08,
      "loss": 0.7183,
      "step": 7432
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.227662444114685,
      "learning_rate": 1.7644785094494964e-08,
      "loss": 0.781,
      "step": 7433
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2318832874298096,
      "learning_rate": 1.7467108601059713e-08,
      "loss": 0.733,
      "step": 7434
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2542024850845337,
      "learning_rate": 1.729032964336519e-08,
      "loss": 0.6917,
      "step": 7435
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3112897872924805,
      "learning_rate": 1.7114448253255924e-08,
      "loss": 0.7629,
      "step": 7436
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5620604157447815,
      "learning_rate": 1.6939464462414347e-08,
      "loss": 0.988,
      "step": 7437
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2424790859222412,
      "learning_rate": 1.6765378302361357e-08,
      "loss": 0.7715,
      "step": 7438
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2650384902954102,
      "learning_rate": 1.659218980445687e-08,
      "loss": 0.7844,
      "step": 7439
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2376717329025269,
      "learning_rate": 1.64198989998976e-08,
      "loss": 0.7472,
      "step": 7440
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2699979543685913,
      "learning_rate": 1.6248505919719825e-08,
      "loss": 0.7668,
      "step": 7441
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2314971685409546,
      "learning_rate": 1.6078010594797742e-08,
      "loss": 0.7051,
      "step": 7442
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2825673818588257,
      "learning_rate": 1.590841305584345e-08,
      "loss": 0.8287,
      "step": 7443
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.568397581577301,
      "learning_rate": 1.5739713333408068e-08,
      "loss": 1.0059,
      "step": 7444
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3104456663131714,
      "learning_rate": 1.557191145788006e-08,
      "loss": 0.7939,
      "step": 7445
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3835557699203491,
      "learning_rate": 1.5405007459487475e-08,
      "loss": 0.738,
      "step": 7446
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.240615725517273,
      "learning_rate": 1.5239001368294593e-08,
      "loss": 0.7009,
      "step": 7447
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2781656980514526,
      "learning_rate": 1.5073893214206383e-08,
      "loss": 0.7884,
      "step": 7448
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2943209409713745,
      "learning_rate": 1.490968302696405e-08,
      "loss": 0.7928,
      "step": 7449
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5839306116104126,
      "learning_rate": 1.4746370836147827e-08,
      "loss": 0.9922,
      "step": 7450
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.293074131011963,
      "learning_rate": 1.4583956671176957e-08,
      "loss": 0.7436,
      "step": 7451
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2129868268966675,
      "learning_rate": 1.4422440561307482e-08,
      "loss": 0.7263,
      "step": 7452
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2409818172454834,
      "learning_rate": 1.4261822535633907e-08,
      "loss": 0.7671,
      "step": 7453
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5707188844680786,
      "learning_rate": 1.4102102623090863e-08,
      "loss": 1.0144,
      "step": 7454
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2085742950439453,
      "learning_rate": 1.3943280852448116e-08,
      "loss": 0.8085,
      "step": 7455
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2581233978271484,
      "learning_rate": 1.3785357252316666e-08,
      "loss": 0.8067,
      "step": 7456
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.202066421508789,
      "learning_rate": 1.3628331851142651e-08,
      "loss": 0.7724,
      "step": 7457
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2602910995483398,
      "learning_rate": 1.3472204677213995e-08,
      "loss": 0.8179,
      "step": 7458
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3133583068847656,
      "learning_rate": 1.3316975758653761e-08,
      "loss": 0.7854,
      "step": 7459
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2661954164505005,
      "learning_rate": 1.3162645123424022e-08,
      "loss": 0.7278,
      "step": 7460
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4150642156600952,
      "learning_rate": 1.3009212799325877e-08,
      "loss": 0.7776,
      "step": 7461
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3689614534378052,
      "learning_rate": 1.2856678813998325e-08,
      "loss": 0.8217,
      "step": 7462
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2102023363113403,
      "learning_rate": 1.2705043194918276e-08,
      "loss": 0.8297,
      "step": 7463
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2442960739135742,
      "learning_rate": 1.2554305969399993e-08,
      "loss": 0.8003,
      "step": 7464
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5742360949516296,
      "learning_rate": 1.2404467164597311e-08,
      "loss": 1.0398,
      "step": 7465
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1918227672576904,
      "learning_rate": 1.2255526807501971e-08,
      "loss": 0.7142,
      "step": 7466
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.472584843635559,
      "learning_rate": 1.210748492494307e-08,
      "loss": 0.7766,
      "step": 7467
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1959319114685059,
      "learning_rate": 1.1960341543588716e-08,
      "loss": 0.7928,
      "step": 7468
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.345343828201294,
      "learning_rate": 1.1814096689944932e-08,
      "loss": 0.7566,
      "step": 7469
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2659082412719727,
      "learning_rate": 1.1668750390354533e-08,
      "loss": 0.7332,
      "step": 7470
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2265554666519165,
      "learning_rate": 1.152430267100102e-08,
      "loss": 0.7391,
      "step": 7471
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2915449142456055,
      "learning_rate": 1.138075355790469e-08,
      "loss": 0.7703,
      "step": 7472
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.234918236732483,
      "learning_rate": 1.1238103076922635e-08,
      "loss": 0.8341,
      "step": 7473
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2832226753234863,
      "learning_rate": 1.1096351253752635e-08,
      "loss": 0.7442,
      "step": 7474
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2222830057144165,
      "learning_rate": 1.0955498113929264e-08,
      "loss": 0.7673,
      "step": 7475
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2435142993927002,
      "learning_rate": 1.0815543682825003e-08,
      "loss": 0.786,
      "step": 7476
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5813614130020142,
      "learning_rate": 1.0676487985650242e-08,
      "loss": 0.9957,
      "step": 7477
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2562472820281982,
      "learning_rate": 1.0538331047454941e-08,
      "loss": 0.7574,
      "step": 7478
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2786985635757446,
      "learning_rate": 1.0401072893125863e-08,
      "loss": 0.7475,
      "step": 7479
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3053114414215088,
      "learning_rate": 1.0264713547388227e-08,
      "loss": 0.7777,
      "step": 7480
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2193562984466553,
      "learning_rate": 1.0129253034805165e-08,
      "loss": 0.7936,
      "step": 7481
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2928563356399536,
      "learning_rate": 9.99469137977771e-09,
      "loss": 0.8285,
      "step": 7482
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1994177103042603,
      "learning_rate": 9.86102860654592e-09,
      "loss": 0.8042,
      "step": 7483
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3306337594985962,
      "learning_rate": 9.728264739187198e-09,
      "loss": 0.7845,
      "step": 7484
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2698959112167358,
      "learning_rate": 9.596399801616862e-09,
      "loss": 0.8003,
      "step": 7485
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3499712944030762,
      "learning_rate": 9.465433817589243e-09,
      "loss": 0.8091,
      "step": 7486
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5576027035713196,
      "learning_rate": 9.335366810695467e-09,
      "loss": 1.0137,
      "step": 7487
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2379485368728638,
      "learning_rate": 9.206198804365684e-09,
      "loss": 0.7459,
      "step": 7488
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.242552638053894,
      "learning_rate": 9.077929821867392e-09,
      "loss": 0.8125,
      "step": 7489
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2122585773468018,
      "learning_rate": 8.950559886306553e-09,
      "loss": 0.7724,
      "step": 7490
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3392245769500732,
      "learning_rate": 8.82408902062759e-09,
      "loss": 0.8084,
      "step": 7491
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.228517770767212,
      "learning_rate": 8.698517247612282e-09,
      "loss": 0.6941,
      "step": 7492
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3005326986312866,
      "learning_rate": 8.573844589880308e-09,
      "loss": 0.7631,
      "step": 7493
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1978235244750977,
      "learning_rate": 8.450071069890375e-09,
      "loss": 0.7517,
      "step": 7494
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2184295654296875,
      "learning_rate": 8.327196709937978e-09,
      "loss": 0.7998,
      "step": 7495
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2291874885559082,
      "learning_rate": 8.205221532158192e-09,
      "loss": 0.7544,
      "step": 7496
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5721341967582703,
      "learning_rate": 8.084145558522327e-09,
      "loss": 1.0021,
      "step": 7497
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2529269456863403,
      "learning_rate": 7.963968810840162e-09,
      "loss": 0.7456,
      "step": 7498
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.208557367324829,
      "learning_rate": 7.8446913107616e-09,
      "loss": 0.7386,
      "step": 7499
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.269126057624817,
      "learning_rate": 7.726313079771119e-09,
      "loss": 0.8088,
      "step": 7500
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5644829273223877,
      "learning_rate": 7.608834139194443e-09,
      "loss": 1.0064,
      "step": 7501
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3067302703857422,
      "learning_rate": 7.49225451019242e-09,
      "loss": 0.7755,
      "step": 7502
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2348525524139404,
      "learning_rate": 7.37657421376603e-09,
      "loss": 0.7831,
      "step": 7503
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2160097360610962,
      "learning_rate": 7.261793270753603e-09,
      "loss": 0.7628,
      "step": 7504
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.25216543674469,
      "learning_rate": 7.147911701830823e-09,
      "loss": 0.7699,
      "step": 7505
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.247066617012024,
      "learning_rate": 7.034929527512946e-09,
      "loss": 0.7997,
      "step": 7506
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2190814018249512,
      "learning_rate": 6.9228467681509145e-09,
      "loss": 0.7606,
      "step": 7507
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5763190984725952,
      "learning_rate": 6.811663443935801e-09,
      "loss": 1.0049,
      "step": 7508
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1997668743133545,
      "learning_rate": 6.701379574894917e-09,
      "loss": 0.7991,
      "step": 7509
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5735092163085938,
      "learning_rate": 6.591995180895705e-09,
      "loss": 1.0274,
      "step": 7510
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2346644401550293,
      "learning_rate": 6.483510281641292e-09,
      "loss": 0.8045,
      "step": 7511
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5612821578979492,
      "learning_rate": 6.375924896673824e-09,
      "loss": 1.0017,
      "step": 7512
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2646558284759521,
      "learning_rate": 6.2692390453733545e-09,
      "loss": 0.7937,
      "step": 7513
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2001278400421143,
      "learning_rate": 6.163452746958398e-09,
      "loss": 0.7105,
      "step": 7514
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5723273158073425,
      "learning_rate": 6.058566020484824e-09,
      "loss": 1.0264,
      "step": 7515
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5691769123077393,
      "learning_rate": 5.954578884845852e-09,
      "loss": 1.0072,
      "step": 7516
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2392823696136475,
      "learning_rate": 5.8514913587737195e-09,
      "loss": 0.7649,
      "step": 7517
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5638213157653809,
      "learning_rate": 5.749303460838573e-09,
      "loss": 1.0134,
      "step": 7518
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3421586751937866,
      "learning_rate": 5.648015209448465e-09,
      "loss": 0.8061,
      "step": 7519
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.557209849357605,
      "learning_rate": 5.547626622848246e-09,
      "loss": 1.0103,
      "step": 7520
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5575226545333862,
      "learning_rate": 5.448137719121782e-09,
      "loss": 1.0146,
      "step": 7521
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5683830380439758,
      "learning_rate": 5.34954851619085e-09,
      "loss": 1.0141,
      "step": 7522
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.26517915725708,
      "learning_rate": 5.251859031814577e-09,
      "loss": 0.7009,
      "step": 7523
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5704751014709473,
      "learning_rate": 5.155069283591108e-09,
      "loss": 1.0078,
      "step": 7524
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5742433071136475,
      "learning_rate": 5.0591792889553845e-09,
      "loss": 1.0165,
      "step": 7525
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5840357542037964,
      "learning_rate": 4.964189065180258e-09,
      "loss": 1.0232,
      "step": 7526
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5844436883926392,
      "learning_rate": 4.870098629377596e-09,
      "loss": 1.0226,
      "step": 7527
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3384798765182495,
      "learning_rate": 4.776907998496616e-09,
      "loss": 0.7866,
      "step": 7528
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5783049464225769,
      "learning_rate": 4.684617189323892e-09,
      "loss": 1.0207,
      "step": 7529
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2171913385391235,
      "learning_rate": 4.593226218483904e-09,
      "loss": 0.7736,
      "step": 7530
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1595131158828735,
      "learning_rate": 4.502735102440703e-09,
      "loss": 0.7529,
      "step": 7531
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5558248162269592,
      "learning_rate": 4.413143857494584e-09,
      "loss": 1.0201,
      "step": 7532
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1700738668441772,
      "learning_rate": 4.324452499783749e-09,
      "loss": 0.7636,
      "step": 7533
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5903946161270142,
      "learning_rate": 4.2366610452848625e-09,
      "loss": 1.0169,
      "step": 7534
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.311608910560608,
      "learning_rate": 4.1497695098124954e-09,
      "loss": 0.7751,
      "step": 7535
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5771519541740417,
      "learning_rate": 4.063777909019129e-09,
      "loss": 0.9852,
      "step": 7536
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.308107852935791,
      "learning_rate": 3.978686258395148e-09,
      "loss": 0.7828,
      "step": 7537
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2475478649139404,
      "learning_rate": 3.8944945732682925e-09,
      "loss": 0.7417,
      "step": 7538
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2194442749023438,
      "learning_rate": 3.811202868804209e-09,
      "loss": 0.7504,
      "step": 7539
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2654216289520264,
      "learning_rate": 3.728811160007562e-09,
      "loss": 0.7786,
      "step": 7540
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3333873748779297,
      "learning_rate": 3.6473194617198117e-09,
      "loss": 0.7825,
      "step": 7541
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5768585801124573,
      "learning_rate": 3.5667277886203277e-09,
      "loss": 1.0146,
      "step": 7542
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2214949131011963,
      "learning_rate": 3.4870361552269415e-09,
      "loss": 0.7284,
      "step": 7543
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2336219549179077,
      "learning_rate": 3.408244575894837e-09,
      "loss": 0.7309,
      "step": 7544
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1985067129135132,
      "learning_rate": 3.3303530648171047e-09,
      "loss": 0.7691,
      "step": 7545
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1956251859664917,
      "learning_rate": 3.253361636024743e-09,
      "loss": 0.7699,
      "step": 7546
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2278048992156982,
      "learning_rate": 3.1772703033877693e-09,
      "loss": 0.8084,
      "step": 7547
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5766283869743347,
      "learning_rate": 3.1020790806118863e-09,
      "loss": 0.9904,
      "step": 7548
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5628784894943237,
      "learning_rate": 3.0277879812418145e-09,
      "loss": 1.0044,
      "step": 7549
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2175804376602173,
      "learning_rate": 2.9543970186607375e-09,
      "loss": 0.8346,
      "step": 7550
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5648717880249023,
      "learning_rate": 2.881906206088081e-09,
      "loss": 0.725,
      "step": 7551
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2983065843582153,
      "learning_rate": 2.8103155565833982e-09,
      "loss": 0.7895,
      "step": 7552
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2678651809692383,
      "learning_rate": 2.7396250830419302e-09,
      "loss": 0.7228,
      "step": 7553
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2427194118499756,
      "learning_rate": 2.6698347981973794e-09,
      "loss": 0.7714,
      "step": 7554
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2144362926483154,
      "learning_rate": 2.6009447146224664e-09,
      "loss": 0.77,
      "step": 7555
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.223490595817566,
      "learning_rate": 2.532954844725599e-09,
      "loss": 0.7638,
      "step": 7556
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1634142398834229,
      "learning_rate": 2.465865200755313e-09,
      "loss": 0.6845,
      "step": 7557
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5786891579627991,
      "learning_rate": 2.3996757947963855e-09,
      "loss": 1.0118,
      "step": 7558
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5772786736488342,
      "learning_rate": 2.334386638771502e-09,
      "loss": 1.0063,
      "step": 7559
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.184624195098877,
      "learning_rate": 2.26999774444292e-09,
      "loss": 0.7378,
      "step": 7560
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5685511827468872,
      "learning_rate": 2.2065091234080295e-09,
      "loss": 0.9966,
      "step": 7561
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5776368379592896,
      "learning_rate": 2.1439207871043477e-09,
      "loss": 1.0269,
      "step": 7562
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.195753574371338,
      "learning_rate": 2.0822327468056347e-09,
      "loss": 0.7509,
      "step": 7563
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.173051118850708,
      "learning_rate": 2.0214450136246676e-09,
      "loss": 0.733,
      "step": 7564
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.240083932876587,
      "learning_rate": 1.9615575985115766e-09,
      "loss": 0.7972,
      "step": 7565
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3166013956069946,
      "learning_rate": 1.902570512253843e-09,
      "loss": 0.8524,
      "step": 7566
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1947652101516724,
      "learning_rate": 1.8444837654774118e-09,
      "loss": 0.7697,
      "step": 7567
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5810166001319885,
      "learning_rate": 1.7872973686461348e-09,
      "loss": 1.0081,
      "step": 7568
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5524392127990723,
      "learning_rate": 1.7310113320606613e-09,
      "loss": 1.0065,
      "step": 7569
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5665906667709351,
      "learning_rate": 1.6756256658612135e-09,
      "loss": 1.0193,
      "step": 7570
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.58530592918396,
      "learning_rate": 1.6211403800237003e-09,
      "loss": 1.024,
      "step": 7571
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.221932053565979,
      "learning_rate": 1.5675554843630481e-09,
      "loss": 0.7174,
      "step": 7572
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.176681399345398,
      "learning_rate": 1.5148709885326463e-09,
      "loss": 0.695,
      "step": 7573
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2821367979049683,
      "learning_rate": 1.463086902021571e-09,
      "loss": 0.8179,
      "step": 7574
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.251056432723999,
      "learning_rate": 1.4122032341595816e-09,
      "loss": 0.7988,
      "step": 7575
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5550667643547058,
      "learning_rate": 1.3622199941115689e-09,
      "loss": 1.0029,
      "step": 7576
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5645877718925476,
      "learning_rate": 1.3131371908819968e-09,
      "loss": 1.0101,
      "step": 7577
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5547069907188416,
      "learning_rate": 1.2649548333115713e-09,
      "loss": 1.0205,
      "step": 7578
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2634998559951782,
      "learning_rate": 1.2176729300811264e-09,
      "loss": 0.7428,
      "step": 7579
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.219232201576233,
      "learning_rate": 1.1712914897060723e-09,
      "loss": 0.781,
      "step": 7580
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5745155215263367,
      "learning_rate": 1.125810520543058e-09,
      "loss": 1.0135,
      "step": 7581
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2316869497299194,
      "learning_rate": 1.0812300307838642e-09,
      "loss": 0.7309,
      "step": 7582
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2912050485610962,
      "learning_rate": 1.0375500284592888e-09,
      "loss": 0.7926,
      "step": 7583
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2868348360061646,
      "learning_rate": 9.947705214380377e-10,
      "loss": 0.7943,
      "step": 7584
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2166104316711426,
      "learning_rate": 9.528915174256137e-10,
      "loss": 0.7691,
      "step": 7585
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2117267847061157,
      "learning_rate": 9.119130239665374e-10,
      "loss": 0.7729,
      "step": 7586
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2707136869430542,
      "learning_rate": 8.718350484421267e-10,
      "loss": 0.7842,
      "step": 7587
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2731891870498657,
      "learning_rate": 8.326575980716067e-10,
      "loss": 0.7789,
      "step": 7588
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5814482569694519,
      "learning_rate": 7.943806799132203e-10,
      "loss": 1.0197,
      "step": 7589
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2790875434875488,
      "learning_rate": 7.570043008614525e-10,
      "loss": 0.7527,
      "step": 7590
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2553828954696655,
      "learning_rate": 7.205284676492508e-10,
      "loss": 0.7851,
      "step": 7591
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3162018060684204,
      "learning_rate": 6.849531868474702e-10,
      "loss": 0.8083,
      "step": 7592
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1943509578704834,
      "learning_rate": 6.502784648637628e-10,
      "loss": 0.7162,
      "step": 7593
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3235958814620972,
      "learning_rate": 6.165043079447986e-10,
      "loss": 0.7542,
      "step": 7594
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.277804970741272,
      "learning_rate": 5.836307221751547e-10,
      "loss": 0.8187,
      "step": 7595
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5715505480766296,
      "learning_rate": 5.516577134750955e-10,
      "loss": 0.9947,
      "step": 7596
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5541532039642334,
      "learning_rate": 5.205852876055684e-10,
      "loss": 0.9922,
      "step": 7597
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2688450813293457,
      "learning_rate": 4.904134501632074e-10,
      "loss": 0.7997,
      "step": 7598
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2815550565719604,
      "learning_rate": 4.6114220658310946e-10,
      "loss": 0.7066,
      "step": 7599
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1898329257965088,
      "learning_rate": 4.3277156213772377e-10,
      "loss": 0.6804,
      "step": 7600
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.255784511566162,
      "learning_rate": 4.053015219379619e-10,
      "loss": 0.7553,
      "step": 7601
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2314550876617432,
      "learning_rate": 3.787320909326431e-10,
      "loss": 0.737,
      "step": 7602
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2022907733917236,
      "learning_rate": 3.530632739073836e-10,
      "loss": 0.7763,
      "step": 7603
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5552709698677063,
      "learning_rate": 3.282950754862624e-10,
      "loss": 1.005,
      "step": 7604
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5703383088111877,
      "learning_rate": 3.0442750013071065e-10,
      "loss": 0.9982,
      "step": 7605
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2731218338012695,
      "learning_rate": 2.8146055214006705e-10,
      "loss": 0.8196,
      "step": 7606
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.257799506187439,
      "learning_rate": 2.5939423565157773e-10,
      "loss": 0.7965,
      "step": 7607
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2903183698654175,
      "learning_rate": 2.382285546409513e-10,
      "loss": 0.7226,
      "step": 7608
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2824689149856567,
      "learning_rate": 2.179635129195834e-10,
      "loss": 0.7562,
      "step": 7609
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.260946273803711,
      "learning_rate": 1.9859911413899758e-10,
      "loss": 0.7869,
      "step": 7610
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.268721342086792,
      "learning_rate": 1.8013536178695946e-10,
      "loss": 0.7971,
      "step": 7611
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3054540157318115,
      "learning_rate": 1.625722591896972e-10,
      "loss": 0.7639,
      "step": 7612
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2163220643997192,
      "learning_rate": 1.4590980951079137e-10,
      "loss": 0.7154,
      "step": 7613
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2461177110671997,
      "learning_rate": 1.3014801575228497e-10,
      "loss": 0.7786,
      "step": 7614
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.238411545753479,
      "learning_rate": 1.1528688075246319e-10,
      "loss": 0.7549,
      "step": 7615
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2469273805618286,
      "learning_rate": 1.0132640718862885e-10,
      "loss": 0.7562,
      "step": 7616
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3064460754394531,
      "learning_rate": 8.82665975759922e-11,
      "loss": 0.7705,
      "step": 7617
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5634201169013977,
      "learning_rate": 7.610745426711585e-11,
      "loss": 1.0057,
      "step": 7618
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5683512687683105,
      "learning_rate": 6.484897945191471e-11,
      "loss": 0.9807,
      "step": 7619
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.239810585975647,
      "learning_rate": 5.449117515876623e-11,
      "loss": 0.7696,
      "step": 7620
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3013932704925537,
      "learning_rate": 4.5034043253400214e-11,
      "loss": 0.7837,
      "step": 7621
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.333762288093567,
      "learning_rate": 3.64775854394539e-11,
      "loss": 0.7884,
      "step": 7622
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5617364645004272,
      "learning_rate": 2.8821803257916835e-11,
      "loss": 0.9991,
      "step": 7623
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.17570960521698,
      "learning_rate": 2.2066698087686022e-11,
      "loss": 0.7831,
      "step": 7624
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.227004885673523,
      "learning_rate": 1.6212271146676116e-11,
      "loss": 0.761,
      "step": 7625
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.578382670879364,
      "learning_rate": 1.1258523488488771e-11,
      "loss": 1.0091,
      "step": 7626
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.227470874786377,
      "learning_rate": 7.205456006298406e-12,
      "loss": 0.767,
      "step": 7627
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5830634236335754,
      "learning_rate": 4.05306942952155e-12,
      "loss": 0.9831,
      "step": 7628
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2981268167495728,
      "learning_rate": 1.8013643260372804e-12,
      "loss": 0.8166,
      "step": 7629
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5550422072410583,
      "learning_rate": 4.5034110163211243e-13,
      "loss": 0.9921,
      "step": 7630
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3351916074752808,
      "learning_rate": 0.0,
      "loss": 0.6742,
      "step": 7631
    },
    {
      "epoch": 1.0,
      "step": 7631,
      "total_flos": 8.50898129102517e+18,
      "train_loss": 0.8704985497804226,
      "train_runtime": 132327.0695,
      "train_samples_per_second": 29.529,
      "train_steps_per_second": 0.058
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 7631,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 8.50898129102517e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}