{
  "best_global_step": 21000,
  "best_metric": 0.876753368482751,
  "best_model_checkpoint": "./saved_best/sst2_XLnet-20000/full_text/checkpoint-21000",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 21000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0014285714285714286,
      "grad_norm": 2353427.5,
      "learning_rate": 1.9991428571428572e-05,
      "loss": 0.734,
      "step": 10
    },
    {
      "epoch": 0.002857142857142857,
      "grad_norm": 3877230.25,
      "learning_rate": 1.9981904761904764e-05,
      "loss": 0.8142,
      "step": 20
    },
    {
      "epoch": 0.004285714285714286,
      "grad_norm": 1722789.625,
      "learning_rate": 1.9972380952380953e-05,
      "loss": 0.691,
      "step": 30
    },
    {
      "epoch": 0.005714285714285714,
      "grad_norm": 491525.375,
      "learning_rate": 1.9962857142857145e-05,
      "loss": 0.9021,
      "step": 40
    },
    {
      "epoch": 0.007142857142857143,
      "grad_norm": 653827.6875,
      "learning_rate": 1.9953333333333334e-05,
      "loss": 1.3132,
      "step": 50
    },
    {
      "epoch": 0.008571428571428572,
      "grad_norm": 7243415.5,
      "learning_rate": 1.9943809523809527e-05,
      "loss": 1.6558,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 56343.32421875,
      "learning_rate": 1.9934285714285716e-05,
      "loss": 1.5856,
      "step": 70
    },
    {
      "epoch": 0.011428571428571429,
      "grad_norm": 7940638.0,
      "learning_rate": 1.9924761904761908e-05,
      "loss": 2.3503,
      "step": 80
    },
    {
      "epoch": 0.012857142857142857,
      "grad_norm": 5312922.0,
      "learning_rate": 1.9915238095238097e-05,
      "loss": 2.7958,
      "step": 90
    },
    {
      "epoch": 0.014285714285714285,
      "grad_norm": 3296663.75,
      "learning_rate": 1.990571428571429e-05,
      "loss": 0.7183,
      "step": 100
    },
    {
      "epoch": 0.015714285714285715,
      "grad_norm": 487267.90625,
      "learning_rate": 1.9896190476190478e-05,
      "loss": 0.7908,
      "step": 110
    },
    {
      "epoch": 0.017142857142857144,
      "grad_norm": 166188.921875,
      "learning_rate": 1.9886666666666667e-05,
      "loss": 1.946,
      "step": 120
    },
    {
      "epoch": 0.018571428571428572,
      "grad_norm": 47725.7890625,
      "learning_rate": 1.987714285714286e-05,
      "loss": 0.8051,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 10994127.0,
      "learning_rate": 1.986761904761905e-05,
      "loss": 3.054,
      "step": 140
    },
    {
      "epoch": 0.02142857142857143,
      "grad_norm": 116536.4921875,
      "learning_rate": 1.985809523809524e-05,
      "loss": 3.7001,
      "step": 150
    },
    {
      "epoch": 0.022857142857142857,
      "grad_norm": 137400.578125,
      "learning_rate": 1.984857142857143e-05,
      "loss": 1.8729,
      "step": 160
    },
    {
      "epoch": 0.024285714285714285,
      "grad_norm": 5900089.5,
      "learning_rate": 1.983904761904762e-05,
      "loss": 2.1636,
      "step": 170
    },
    {
      "epoch": 0.025714285714285714,
      "grad_norm": 5998738.5,
      "learning_rate": 1.982952380952381e-05,
      "loss": 3.1394,
      "step": 180
    },
    {
      "epoch": 0.027142857142857142,
      "grad_norm": 22422.69140625,
      "learning_rate": 1.982e-05,
      "loss": 3.1351,
      "step": 190
    },
    {
      "epoch": 0.02857142857142857,
      "grad_norm": 5135892.0,
      "learning_rate": 1.9810476190476192e-05,
      "loss": 2.0144,
      "step": 200
    },
    {
      "epoch": 0.03,
      "grad_norm": 5801.9951171875,
      "learning_rate": 1.980095238095238e-05,
      "loss": 2.1827,
      "step": 210
    },
    {
      "epoch": 0.03142857142857143,
      "grad_norm": 2392.868408203125,
      "learning_rate": 1.9791428571428574e-05,
      "loss": 3.2605,
      "step": 220
    },
    {
      "epoch": 0.032857142857142856,
      "grad_norm": 1334.1170654296875,
      "learning_rate": 1.9781904761904766e-05,
      "loss": 4.5156,
      "step": 230
    },
    {
      "epoch": 0.03428571428571429,
      "grad_norm": 36640.8359375,
      "learning_rate": 1.9772380952380955e-05,
      "loss": 6.1001,
      "step": 240
    },
    {
      "epoch": 0.03571428571428571,
      "grad_norm": 457097.28125,
      "learning_rate": 1.9762857142857144e-05,
      "loss": 1.6054,
      "step": 250
    },
    {
      "epoch": 0.037142857142857144,
      "grad_norm": 34546.21875,
      "learning_rate": 1.9753333333333336e-05,
      "loss": 2.4868,
      "step": 260
    },
    {
      "epoch": 0.03857142857142857,
      "grad_norm": 5782654.0,
      "learning_rate": 1.9743809523809525e-05,
      "loss": 3.3626,
      "step": 270
    },
    {
      "epoch": 0.04,
      "grad_norm": 6972.009765625,
      "learning_rate": 1.9734285714285717e-05,
      "loss": 3.3807,
      "step": 280
    },
    {
      "epoch": 0.041428571428571426,
      "grad_norm": 12274402.0,
      "learning_rate": 1.9724761904761906e-05,
      "loss": 2.5003,
      "step": 290
    },
    {
      "epoch": 0.04285714285714286,
      "grad_norm": 37185.46484375,
      "learning_rate": 1.9715238095238095e-05,
      "loss": 2.6575,
      "step": 300
    },
    {
      "epoch": 0.04428571428571428,
      "grad_norm": 5052950.0,
      "learning_rate": 1.9705714285714288e-05,
      "loss": 1.6708,
      "step": 310
    },
    {
      "epoch": 0.045714285714285714,
      "grad_norm": 41157.984375,
      "learning_rate": 1.9696190476190477e-05,
      "loss": 3.516,
      "step": 320
    },
    {
      "epoch": 0.047142857142857146,
      "grad_norm": 276855.625,
      "learning_rate": 1.9686666666666666e-05,
      "loss": 2.7894,
      "step": 330
    },
    {
      "epoch": 0.04857142857142857,
      "grad_norm": 163346.875,
      "learning_rate": 1.9677142857142858e-05,
      "loss": 2.8045,
      "step": 340
    },
    {
      "epoch": 0.05,
      "grad_norm": 181578.46875,
      "learning_rate": 1.966761904761905e-05,
      "loss": 1.143,
      "step": 350
    },
    {
      "epoch": 0.05142857142857143,
      "grad_norm": 6600286.0,
      "learning_rate": 1.965809523809524e-05,
      "loss": 1.3318,
      "step": 360
    },
    {
      "epoch": 0.05285714285714286,
      "grad_norm": 99727.0625,
      "learning_rate": 1.964857142857143e-05,
      "loss": 1.5101,
      "step": 370
    },
    {
      "epoch": 0.054285714285714284,
      "grad_norm": 15291124.0,
      "learning_rate": 1.963904761904762e-05,
      "loss": 3.3314,
      "step": 380
    },
    {
      "epoch": 0.055714285714285716,
      "grad_norm": 4876.53271484375,
      "learning_rate": 1.9629523809523813e-05,
      "loss": 5.328,
      "step": 390
    },
    {
      "epoch": 0.05714285714285714,
      "grad_norm": 556938.25,
      "learning_rate": 1.9620000000000002e-05,
      "loss": 1.6572,
      "step": 400
    },
    {
      "epoch": 0.05857142857142857,
      "grad_norm": 5173910.0,
      "learning_rate": 1.9610476190476194e-05,
      "loss": 1.9942,
      "step": 410
    },
    {
      "epoch": 0.06,
      "grad_norm": 271407.6875,
      "learning_rate": 1.9600952380952383e-05,
      "loss": 0.7475,
      "step": 420
    },
    {
      "epoch": 0.06142857142857143,
      "grad_norm": 405646.65625,
      "learning_rate": 1.9591428571428572e-05,
      "loss": 1.1938,
      "step": 430
    },
    {
      "epoch": 0.06285714285714286,
      "grad_norm": 241159.953125,
      "learning_rate": 1.9581904761904764e-05,
      "loss": 2.0702,
      "step": 440
    },
    {
      "epoch": 0.06428571428571428,
      "grad_norm": 4451660.0,
      "learning_rate": 1.9572380952380953e-05,
      "loss": 1.4764,
      "step": 450
    },
    {
      "epoch": 0.06571428571428571,
      "grad_norm": 121948.9296875,
      "learning_rate": 1.9562857142857142e-05,
      "loss": 1.9151,
      "step": 460
    },
    {
      "epoch": 0.06714285714285714,
      "grad_norm": 4201188.5,
      "learning_rate": 1.9553333333333335e-05,
      "loss": 1.1726,
      "step": 470
    },
    {
      "epoch": 0.06857142857142857,
      "grad_norm": 72867.046875,
      "learning_rate": 1.9543809523809524e-05,
      "loss": 2.6849,
      "step": 480
    },
    {
      "epoch": 0.07,
      "grad_norm": 89064.578125,
      "learning_rate": 1.9534285714285716e-05,
      "loss": 2.1194,
      "step": 490
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 96635.90625,
      "learning_rate": 1.9524761904761905e-05,
      "loss": 2.524,
      "step": 500
    },
    {
      "epoch": 0.07285714285714286,
      "grad_norm": 4096215.0,
      "learning_rate": 1.9515238095238097e-05,
      "loss": 1.687,
      "step": 510
    },
    {
      "epoch": 0.07428571428571429,
      "grad_norm": 4344023.5,
      "learning_rate": 1.9505714285714286e-05,
      "loss": 1.7832,
      "step": 520
    },
    {
      "epoch": 0.07571428571428572,
      "grad_norm": 43894.7734375,
      "learning_rate": 1.949619047619048e-05,
      "loss": 2.3537,
      "step": 530
    },
    {
      "epoch": 0.07714285714285714,
      "grad_norm": 66251.3125,
      "learning_rate": 1.948666666666667e-05,
      "loss": 1.5205,
      "step": 540
    },
    {
      "epoch": 0.07857142857142857,
      "grad_norm": 48226.3359375,
      "learning_rate": 1.947714285714286e-05,
      "loss": 1.6408,
      "step": 550
    },
    {
      "epoch": 0.08,
      "grad_norm": 63618.09375,
      "learning_rate": 1.946761904761905e-05,
      "loss": 2.0727,
      "step": 560
    },
    {
      "epoch": 0.08142857142857143,
      "grad_norm": 4043522.75,
      "learning_rate": 1.945809523809524e-05,
      "loss": 2.0688,
      "step": 570
    },
    {
      "epoch": 0.08285714285714285,
      "grad_norm": 3876487.5,
      "learning_rate": 1.944857142857143e-05,
      "loss": 2.743,
      "step": 580
    },
    {
      "epoch": 0.08428571428571428,
      "grad_norm": 3808414.5,
      "learning_rate": 1.943904761904762e-05,
      "loss": 1.3212,
      "step": 590
    },
    {
      "epoch": 0.08571428571428572,
      "grad_norm": 3669863.5,
      "learning_rate": 1.942952380952381e-05,
      "loss": 1.5305,
      "step": 600
    },
    {
      "epoch": 0.08714285714285715,
      "grad_norm": 2005332.25,
      "learning_rate": 1.942e-05,
      "loss": 1.1047,
      "step": 610
    },
    {
      "epoch": 0.08857142857142856,
      "grad_norm": 20659.7734375,
      "learning_rate": 1.9410476190476193e-05,
      "loss": 1.1888,
      "step": 620
    },
    {
      "epoch": 0.09,
      "grad_norm": 1288264.125,
      "learning_rate": 1.9400952380952382e-05,
      "loss": 1.2519,
      "step": 630
    },
    {
      "epoch": 0.09142857142857143,
      "grad_norm": 8332.625,
      "learning_rate": 1.939142857142857e-05,
      "loss": 2.5215,
      "step": 640
    },
    {
      "epoch": 0.09285714285714286,
      "grad_norm": 4794.73583984375,
      "learning_rate": 1.9381904761904763e-05,
      "loss": 2.4632,
      "step": 650
    },
    {
      "epoch": 0.09428571428571429,
      "grad_norm": 5967.048828125,
      "learning_rate": 1.9372380952380952e-05,
      "loss": 1.5709,
      "step": 660
    },
    {
      "epoch": 0.09571428571428571,
      "grad_norm": 1716.23486328125,
      "learning_rate": 1.9362857142857144e-05,
      "loss": 2.464,
      "step": 670
    },
    {
      "epoch": 0.09714285714285714,
      "grad_norm": 2786.732421875,
      "learning_rate": 1.9353333333333337e-05,
      "loss": 2.1597,
      "step": 680
    },
    {
      "epoch": 0.09857142857142857,
      "grad_norm": 1405.267822265625,
      "learning_rate": 1.9343809523809526e-05,
      "loss": 2.3262,
      "step": 690
    },
    {
      "epoch": 0.1,
      "grad_norm": 1107.4581298828125,
      "learning_rate": 1.9334285714285718e-05,
      "loss": 1.6042,
      "step": 700
    },
    {
      "epoch": 0.10142857142857142,
      "grad_norm": 363.9830017089844,
      "learning_rate": 1.9324761904761907e-05,
      "loss": 5.3205,
      "step": 710
    },
    {
      "epoch": 0.10285714285714286,
      "grad_norm": 3714561.75,
      "learning_rate": 1.9315238095238096e-05,
      "loss": 5.2108,
      "step": 720
    },
    {
      "epoch": 0.10428571428571429,
      "grad_norm": 3698512.5,
      "learning_rate": 1.9305714285714288e-05,
      "loss": 4.0755,
      "step": 730
    },
    {
      "epoch": 0.10571428571428572,
      "grad_norm": 1785.8067626953125,
      "learning_rate": 1.9296190476190477e-05,
      "loss": 2.3997,
      "step": 740
    },
    {
      "epoch": 0.10714285714285714,
      "grad_norm": 623.1176147460938,
      "learning_rate": 1.928666666666667e-05,
      "loss": 3.542,
      "step": 750
    },
    {
      "epoch": 0.10857142857142857,
      "grad_norm": 3403546.25,
      "learning_rate": 1.927714285714286e-05,
      "loss": 6.3852,
      "step": 760
    },
    {
      "epoch": 0.11,
      "grad_norm": 3194865.25,
      "learning_rate": 1.9267619047619047e-05,
      "loss": 4.657,
      "step": 770
    },
    {
      "epoch": 0.11142857142857143,
      "grad_norm": 3499253.25,
      "learning_rate": 1.925809523809524e-05,
      "loss": 6.5549,
      "step": 780
    },
    {
      "epoch": 0.11285714285714285,
      "grad_norm": 3554522.25,
      "learning_rate": 1.924857142857143e-05,
      "loss": 2.3156,
      "step": 790
    },
    {
      "epoch": 0.11428571428571428,
      "grad_norm": 3435890.5,
      "learning_rate": 1.923904761904762e-05,
      "loss": 3.9629,
      "step": 800
    },
    {
      "epoch": 0.11571428571428571,
      "grad_norm": 1248.7685546875,
      "learning_rate": 1.922952380952381e-05,
      "loss": 2.4083,
      "step": 810
    },
    {
      "epoch": 0.11714285714285715,
      "grad_norm": 979.3628540039062,
      "learning_rate": 1.9220000000000002e-05,
      "loss": 3.2426,
      "step": 820
    },
    {
      "epoch": 0.11857142857142858,
      "grad_norm": 947.1708984375,
      "learning_rate": 1.921047619047619e-05,
      "loss": 4.2141,
      "step": 830
    },
    {
      "epoch": 0.12,
      "grad_norm": 497.06683349609375,
      "learning_rate": 1.9200952380952384e-05,
      "loss": 2.5158,
      "step": 840
    },
    {
      "epoch": 0.12142857142857143,
      "grad_norm": 627.3834838867188,
      "learning_rate": 1.9191428571428573e-05,
      "loss": 4.3296,
      "step": 850
    },
    {
      "epoch": 0.12285714285714286,
      "grad_norm": 272.04144287109375,
      "learning_rate": 1.9181904761904765e-05,
      "loss": 1.7343,
      "step": 860
    },
    {
      "epoch": 0.12428571428571429,
      "grad_norm": 3250077.0,
      "learning_rate": 1.9172380952380954e-05,
      "loss": 7.2134,
      "step": 870
    },
    {
      "epoch": 0.12571428571428572,
      "grad_norm": 327.62847900390625,
      "learning_rate": 1.9162857142857146e-05,
      "loss": 4.1679,
      "step": 880
    },
    {
      "epoch": 0.12714285714285714,
      "grad_norm": 531.8444213867188,
      "learning_rate": 1.9153333333333335e-05,
      "loss": 3.5677,
      "step": 890
    },
    {
      "epoch": 0.12857142857142856,
      "grad_norm": 2975953.25,
      "learning_rate": 1.9143809523809524e-05,
      "loss": 5.2461,
      "step": 900
    },
    {
      "epoch": 0.13,
      "grad_norm": 3058841.75,
      "learning_rate": 1.9134285714285716e-05,
      "loss": 1.7504,
      "step": 910
    },
    {
      "epoch": 0.13142857142857142,
      "grad_norm": 3079184.25,
      "learning_rate": 1.9124761904761905e-05,
      "loss": 3.659,
      "step": 920
    },
    {
      "epoch": 0.13285714285714287,
      "grad_norm": 3073773.75,
      "learning_rate": 1.9115238095238098e-05,
      "loss": 2.8085,
      "step": 930
    },
    {
      "epoch": 0.13428571428571429,
      "grad_norm": 213.323974609375,
      "learning_rate": 1.9105714285714287e-05,
      "loss": 6.6365,
      "step": 940
    },
    {
      "epoch": 0.1357142857142857,
      "grad_norm": 3187626.0,
      "learning_rate": 1.9096190476190476e-05,
      "loss": 5.5579,
      "step": 950
    },
    {
      "epoch": 0.13714285714285715,
      "grad_norm": 2987384.75,
      "learning_rate": 1.9086666666666668e-05,
      "loss": 5.4104,
      "step": 960
    },
    {
      "epoch": 0.13857142857142857,
      "grad_norm": 444.3898010253906,
      "learning_rate": 1.9077142857142857e-05,
      "loss": 4.3498,
      "step": 970
    },
    {
      "epoch": 0.14,
      "grad_norm": 3211776.5,
      "learning_rate": 1.906761904761905e-05,
      "loss": 5.2822,
      "step": 980
    },
    {
      "epoch": 0.14142857142857143,
      "grad_norm": 472.4419250488281,
      "learning_rate": 1.905809523809524e-05,
      "loss": 0.8663,
      "step": 990
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 197.6053924560547,
      "learning_rate": 1.904857142857143e-05,
      "loss": 4.6225,
      "step": 1000
    },
    {
      "epoch": 0.1442857142857143,
      "grad_norm": 3039000.75,
      "learning_rate": 1.9039047619047623e-05,
      "loss": 6.4161,
      "step": 1010
    },
    {
      "epoch": 0.1457142857142857,
      "grad_norm": 258.9091796875,
      "learning_rate": 1.9029523809523812e-05,
      "loss": 4.6638,
      "step": 1020
    },
    {
      "epoch": 0.14714285714285713,
      "grad_norm": 454.4802551269531,
      "learning_rate": 1.902e-05,
      "loss": 3.6012,
      "step": 1030
    },
    {
      "epoch": 0.14857142857142858,
      "grad_norm": 2940198.0,
      "learning_rate": 1.9010476190476193e-05,
      "loss": 3.7449,
      "step": 1040
    },
    {
      "epoch": 0.15,
      "grad_norm": 3040586.0,
      "learning_rate": 1.9000952380952382e-05,
      "loss": 5.4813,
      "step": 1050
    },
    {
      "epoch": 0.15142857142857144,
      "grad_norm": 344.3110046386719,
      "learning_rate": 1.8991428571428574e-05,
      "loss": 3.6115,
      "step": 1060
    },
    {
      "epoch": 0.15285714285714286,
      "grad_norm": 410.5352783203125,
      "learning_rate": 1.8981904761904763e-05,
      "loss": 3.6529,
      "step": 1070
    },
    {
      "epoch": 0.15428571428571428,
      "grad_norm": 220.04129028320312,
      "learning_rate": 1.8972380952380952e-05,
      "loss": 2.7209,
      "step": 1080
    },
    {
      "epoch": 0.15571428571428572,
      "grad_norm": 166.97799682617188,
      "learning_rate": 1.8962857142857145e-05,
      "loss": 2.7928,
      "step": 1090
    },
    {
      "epoch": 0.15714285714285714,
      "grad_norm": 3046999.0,
      "learning_rate": 1.8953333333333334e-05,
      "loss": 6.9501,
      "step": 1100
    },
    {
      "epoch": 0.15857142857142856,
      "grad_norm": 3067341.25,
      "learning_rate": 1.8943809523809523e-05,
      "loss": 2.0356,
      "step": 1110
    },
    {
      "epoch": 0.16,
      "grad_norm": 138.7490997314453,
      "learning_rate": 1.8934285714285715e-05,
      "loss": 4.0999,
      "step": 1120
    },
    {
      "epoch": 0.16142857142857142,
      "grad_norm": 3417015.0,
      "learning_rate": 1.8924761904761907e-05,
      "loss": 6.1333,
      "step": 1130
    },
    {
      "epoch": 0.16285714285714287,
      "grad_norm": 3648903.5,
      "learning_rate": 1.8915238095238096e-05,
      "loss": 6.9597,
      "step": 1140
    },
    {
      "epoch": 0.16428571428571428,
      "grad_norm": 3591961.0,
      "learning_rate": 1.890571428571429e-05,
      "loss": 5.7271,
      "step": 1150
    },
    {
      "epoch": 0.1657142857142857,
      "grad_norm": 4246559.5,
      "learning_rate": 1.8896190476190478e-05,
      "loss": 4.4086,
      "step": 1160
    },
    {
      "epoch": 0.16714285714285715,
      "grad_norm": 4116118.75,
      "learning_rate": 1.888666666666667e-05,
      "loss": 6.195,
      "step": 1170
    },
    {
      "epoch": 0.16857142857142857,
      "grad_norm": 3530923.0,
      "learning_rate": 1.887714285714286e-05,
      "loss": 4.3632,
      "step": 1180
    },
    {
      "epoch": 0.17,
      "grad_norm": 339.5002746582031,
      "learning_rate": 1.886761904761905e-05,
      "loss": 1.704,
      "step": 1190
    },
    {
      "epoch": 0.17142857142857143,
      "grad_norm": 179.62376403808594,
      "learning_rate": 1.885809523809524e-05,
      "loss": 4.6432,
      "step": 1200
    },
    {
      "epoch": 0.17285714285714285,
      "grad_norm": 412.82916259765625,
      "learning_rate": 1.884857142857143e-05,
      "loss": 5.6369,
      "step": 1210
    },
    {
      "epoch": 0.1742857142857143,
      "grad_norm": 307.7279357910156,
      "learning_rate": 1.883904761904762e-05,
      "loss": 5.6237,
      "step": 1220
    },
    {
      "epoch": 0.1757142857142857,
      "grad_norm": 3377791.25,
      "learning_rate": 1.882952380952381e-05,
      "loss": 4.6557,
      "step": 1230
    },
    {
      "epoch": 0.17714285714285713,
      "grad_norm": 3291173.5,
      "learning_rate": 1.882e-05,
      "loss": 6.2411,
      "step": 1240
    },
    {
      "epoch": 0.17857142857142858,
      "grad_norm": 1721.5784912109375,
      "learning_rate": 1.8810476190476192e-05,
      "loss": 4.0696,
      "step": 1250
    },
    {
      "epoch": 0.18,
      "grad_norm": 4803452.0,
      "learning_rate": 1.880095238095238e-05,
      "loss": 3.7285,
      "step": 1260
    },
    {
      "epoch": 0.18142857142857144,
      "grad_norm": 4309866.0,
      "learning_rate": 1.8791428571428573e-05,
      "loss": 3.4967,
      "step": 1270
    },
    {
      "epoch": 0.18285714285714286,
      "grad_norm": 3802055.0,
      "learning_rate": 1.8781904761904762e-05,
      "loss": 4.7738,
      "step": 1280
    },
    {
      "epoch": 0.18428571428571427,
      "grad_norm": 317.4769287109375,
      "learning_rate": 1.8772380952380954e-05,
      "loss": 1.9971,
      "step": 1290
    },
    {
      "epoch": 0.18571428571428572,
      "grad_norm": 31.29709815979004,
      "learning_rate": 1.8762857142857143e-05,
      "loss": 2.0222,
      "step": 1300
    },
    {
      "epoch": 0.18714285714285714,
      "grad_norm": 15021.96484375,
      "learning_rate": 1.8753333333333336e-05,
      "loss": 3.7272,
      "step": 1310
    },
    {
      "epoch": 0.18857142857142858,
      "grad_norm": 7305956.0,
      "learning_rate": 1.8743809523809528e-05,
      "loss": 6.2566,
      "step": 1320
    },
    {
      "epoch": 0.19,
      "grad_norm": 117.37650299072266,
      "learning_rate": 1.8734285714285717e-05,
      "loss": 3.499,
      "step": 1330
    },
    {
      "epoch": 0.19142857142857142,
      "grad_norm": 46.29367446899414,
      "learning_rate": 1.8724761904761906e-05,
      "loss": 6.7519,
      "step": 1340
    },
    {
      "epoch": 0.19285714285714287,
      "grad_norm": 3842110.75,
      "learning_rate": 1.8715238095238098e-05,
      "loss": 8.3382,
      "step": 1350
    },
    {
      "epoch": 0.19428571428571428,
      "grad_norm": 3527373.0,
      "learning_rate": 1.8705714285714287e-05,
      "loss": 2.9076,
      "step": 1360
    },
    {
      "epoch": 0.1957142857142857,
      "grad_norm": 3416911.75,
      "learning_rate": 1.8696190476190476e-05,
      "loss": 5.5127,
      "step": 1370
    },
    {
      "epoch": 0.19714285714285715,
      "grad_norm": 186.9600830078125,
      "learning_rate": 1.868666666666667e-05,
      "loss": 5.1474,
      "step": 1380
    },
    {
      "epoch": 0.19857142857142857,
      "grad_norm": 4047857.5,
      "learning_rate": 1.8677142857142857e-05,
      "loss": 5.701,
      "step": 1390
    },
    {
      "epoch": 0.2,
      "grad_norm": 163.80221557617188,
      "learning_rate": 1.866761904761905e-05,
      "loss": 4.7802,
      "step": 1400
    },
    {
      "epoch": 0.20142857142857143,
      "grad_norm": 3628152.25,
      "learning_rate": 1.865809523809524e-05,
      "loss": 3.274,
      "step": 1410
    },
    {
      "epoch": 0.20285714285714285,
      "grad_norm": 4659968.5,
      "learning_rate": 1.8648571428571428e-05,
      "loss": 5.3987,
      "step": 1420
    },
    {
      "epoch": 0.2042857142857143,
      "grad_norm": 364.80841064453125,
      "learning_rate": 1.863904761904762e-05,
      "loss": 7.5868,
      "step": 1430
    },
    {
      "epoch": 0.2057142857142857,
      "grad_norm": 699.7898559570312,
      "learning_rate": 1.8629523809523812e-05,
      "loss": 3.6566,
      "step": 1440
    },
    {
      "epoch": 0.20714285714285716,
      "grad_norm": 416.3044128417969,
      "learning_rate": 1.862e-05,
      "loss": 4.6997,
      "step": 1450
    },
    {
      "epoch": 0.20857142857142857,
      "grad_norm": 124.31083679199219,
      "learning_rate": 1.8610476190476194e-05,
      "loss": 3.0336,
      "step": 1460
    },
    {
      "epoch": 0.21,
      "grad_norm": 86.59341430664062,
      "learning_rate": 1.8600952380952383e-05,
      "loss": 4.6071,
      "step": 1470
    },
    {
      "epoch": 0.21142857142857144,
      "grad_norm": 35.801513671875,
      "learning_rate": 1.8591428571428575e-05,
      "loss": 8.1925,
      "step": 1480
    },
    {
      "epoch": 0.21285714285714286,
      "grad_norm": 151.98065185546875,
      "learning_rate": 1.8581904761904764e-05,
      "loss": 5.0046,
      "step": 1490
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 1860.28076171875,
      "learning_rate": 1.8572380952380953e-05,
      "loss": 3.9033,
      "step": 1500
    },
    {
      "epoch": 0.21571428571428572,
      "grad_norm": 3926798.0,
      "learning_rate": 1.8562857142857145e-05,
      "loss": 2.7764,
      "step": 1510
    },
    {
      "epoch": 0.21714285714285714,
      "grad_norm": 3614959.25,
      "learning_rate": 1.8553333333333334e-05,
      "loss": 4.0617,
      "step": 1520
    },
    {
      "epoch": 0.21857142857142858,
      "grad_norm": 332.648193359375,
      "learning_rate": 1.8543809523809526e-05,
      "loss": 5.3247,
      "step": 1530
    },
    {
      "epoch": 0.22,
      "grad_norm": 9262.7578125,
      "learning_rate": 1.8534285714285715e-05,
      "loss": 4.7147,
      "step": 1540
    },
    {
      "epoch": 0.22142857142857142,
      "grad_norm": 559.4579467773438,
      "learning_rate": 1.8524761904761904e-05,
      "loss": 4.683,
      "step": 1550
    },
    {
      "epoch": 0.22285714285714286,
      "grad_norm": 376.1788635253906,
      "learning_rate": 1.8515238095238097e-05,
      "loss": 5.6564,
      "step": 1560
    },
    {
      "epoch": 0.22428571428571428,
      "grad_norm": 16.28071403503418,
      "learning_rate": 1.8505714285714286e-05,
      "loss": 5.7237,
      "step": 1570
    },
    {
      "epoch": 0.2257142857142857,
      "grad_norm": 3411980.0,
      "learning_rate": 1.8496190476190478e-05,
      "loss": 6.0559,
      "step": 1580
    },
    {
      "epoch": 0.22714285714285715,
      "grad_norm": 15.78839111328125,
      "learning_rate": 1.8486666666666667e-05,
      "loss": 4.2503,
      "step": 1590
    },
    {
      "epoch": 0.22857142857142856,
      "grad_norm": 4429952.5,
      "learning_rate": 1.847714285714286e-05,
      "loss": 4.4403,
      "step": 1600
    },
    {
      "epoch": 0.23,
      "grad_norm": 5530591.5,
      "learning_rate": 1.8467619047619048e-05,
      "loss": 6.3761,
      "step": 1610
    },
    {
      "epoch": 0.23142857142857143,
      "grad_norm": 7.323903560638428,
      "learning_rate": 1.845809523809524e-05,
      "loss": 0.0,
      "step": 1620
    },
    {
      "epoch": 0.23285714285714285,
      "grad_norm": 3800930.5,
      "learning_rate": 1.844857142857143e-05,
      "loss": 6.431,
      "step": 1630
    },
    {
      "epoch": 0.2342857142857143,
      "grad_norm": 10.597685813903809,
      "learning_rate": 1.8439047619047622e-05,
      "loss": 4.9521,
      "step": 1640
    },
    {
      "epoch": 0.2357142857142857,
      "grad_norm": 5045.07470703125,
      "learning_rate": 1.842952380952381e-05,
      "loss": 1.7803,
      "step": 1650
    },
    {
      "epoch": 0.23714285714285716,
      "grad_norm": 49637592.0,
      "learning_rate": 1.8420000000000003e-05,
      "loss": 4.0582,
      "step": 1660
    },
    {
      "epoch": 0.23857142857142857,
      "grad_norm": 0.6540566086769104,
      "learning_rate": 1.8410476190476192e-05,
      "loss": 1.518,
      "step": 1670
    },
    {
      "epoch": 0.24,
      "grad_norm": 3117930.25,
      "learning_rate": 1.840095238095238e-05,
      "loss": 3.0968,
      "step": 1680
    },
    {
      "epoch": 0.24142857142857144,
      "grad_norm": 0.39475953578948975,
      "learning_rate": 1.8391428571428573e-05,
      "loss": 3.336,
      "step": 1690
    },
    {
      "epoch": 0.24285714285714285,
      "grad_norm": 202.1228790283203,
      "learning_rate": 1.8381904761904762e-05,
      "loss": 1.6399,
      "step": 1700
    },
    {
      "epoch": 0.24428571428571427,
      "grad_norm": 26007860.0,
      "learning_rate": 1.8372380952380955e-05,
      "loss": 3.5599,
      "step": 1710
    },
    {
      "epoch": 0.24571428571428572,
      "grad_norm": 122523.5625,
      "learning_rate": 1.8362857142857144e-05,
      "loss": 1.0542,
      "step": 1720
    },
    {
      "epoch": 0.24714285714285714,
      "grad_norm": 18631.544921875,
      "learning_rate": 1.8353333333333333e-05,
      "loss": 1.2353,
      "step": 1730
    },
    {
      "epoch": 0.24857142857142858,
      "grad_norm": 5729.6953125,
      "learning_rate": 1.8343809523809525e-05,
      "loss": 1.2454,
      "step": 1740
    },
    {
      "epoch": 0.25,
      "grad_norm": 6366465.0,
      "learning_rate": 1.8334285714285714e-05,
      "loss": 3.3708,
      "step": 1750
    },
    {
      "epoch": 0.25142857142857145,
      "grad_norm": 5909998.0,
      "learning_rate": 1.8324761904761906e-05,
      "loss": 4.298,
      "step": 1760
    },
    {
      "epoch": 0.25285714285714284,
      "grad_norm": 4360192.5,
      "learning_rate": 1.83152380952381e-05,
      "loss": 3.3291,
      "step": 1770
    },
    {
      "epoch": 0.2542857142857143,
      "grad_norm": 2.2309699058532715,
      "learning_rate": 1.8305714285714288e-05,
      "loss": 1.5079,
      "step": 1780
    },
    {
      "epoch": 0.2557142857142857,
      "grad_norm": 15763.150390625,
      "learning_rate": 1.829619047619048e-05,
      "loss": 4.4763,
      "step": 1790
    },
    {
      "epoch": 0.2571428571428571,
      "grad_norm": 4511289.0,
      "learning_rate": 1.828666666666667e-05,
      "loss": 3.9169,
      "step": 1800
    },
    {
      "epoch": 0.25857142857142856,
      "grad_norm": 988.843994140625,
      "learning_rate": 1.8277142857142858e-05,
      "loss": 4.0668,
      "step": 1810
    },
    {
      "epoch": 0.26,
      "grad_norm": 5752921.0,
      "learning_rate": 1.826761904761905e-05,
      "loss": 4.3689,
      "step": 1820
    },
    {
      "epoch": 0.26142857142857145,
      "grad_norm": 1034.303466796875,
      "learning_rate": 1.825809523809524e-05,
      "loss": 3.6404,
      "step": 1830
    },
    {
      "epoch": 0.26285714285714284,
      "grad_norm": 4683658.0,
      "learning_rate": 1.824857142857143e-05,
      "loss": 5.1516,
      "step": 1840
    },
    {
      "epoch": 0.2642857142857143,
      "grad_norm": 4343246.0,
      "learning_rate": 1.823904761904762e-05,
      "loss": 4.8976,
      "step": 1850
    },
    {
      "epoch": 0.26571428571428574,
      "grad_norm": 4432690.5,
      "learning_rate": 1.822952380952381e-05,
      "loss": 4.7781,
      "step": 1860
    },
    {
      "epoch": 0.2671428571428571,
      "grad_norm": 2846.268310546875,
      "learning_rate": 1.8220000000000002e-05,
      "loss": 2.8631,
      "step": 1870
    },
    {
      "epoch": 0.26857142857142857,
      "grad_norm": 5005105.0,
      "learning_rate": 1.821047619047619e-05,
      "loss": 2.9285,
      "step": 1880
    },
    {
      "epoch": 0.27,
      "grad_norm": 1939443.125,
      "learning_rate": 1.8200952380952383e-05,
      "loss": 3.7188,
      "step": 1890
    },
    {
      "epoch": 0.2714285714285714,
      "grad_norm": 7336.20849609375,
      "learning_rate": 1.8191428571428572e-05,
      "loss": 2.8425,
      "step": 1900
    },
    {
      "epoch": 0.27285714285714285,
      "grad_norm": 1052.8658447265625,
      "learning_rate": 1.8181904761904764e-05,
      "loss": 2.4512,
      "step": 1910
    },
    {
      "epoch": 0.2742857142857143,
      "grad_norm": 709.6981201171875,
      "learning_rate": 1.8172380952380953e-05,
      "loss": 3.1909,
      "step": 1920
    },
    {
      "epoch": 0.2757142857142857,
      "grad_norm": 1475.0738525390625,
      "learning_rate": 1.8162857142857146e-05,
      "loss": 4.1909,
      "step": 1930
    },
    {
      "epoch": 0.27714285714285714,
      "grad_norm": 824.7996215820312,
      "learning_rate": 1.8153333333333335e-05,
      "loss": 4.6443,
      "step": 1940
    },
    {
      "epoch": 0.2785714285714286,
      "grad_norm": 758.540283203125,
      "learning_rate": 1.8143809523809527e-05,
      "loss": 1.6537,
      "step": 1950
    },
    {
      "epoch": 0.28,
      "grad_norm": 4110761.5,
      "learning_rate": 1.8134285714285716e-05,
      "loss": 2.6131,
      "step": 1960
    },
    {
      "epoch": 0.2814285714285714,
      "grad_norm": 283.4450988769531,
      "learning_rate": 1.8124761904761908e-05,
      "loss": 4.4712,
      "step": 1970
    },
    {
      "epoch": 0.28285714285714286,
      "grad_norm": 3958263.25,
      "learning_rate": 1.8115238095238097e-05,
      "loss": 6.2166,
      "step": 1980
    },
    {
      "epoch": 0.2842857142857143,
      "grad_norm": 4028225.0,
      "learning_rate": 1.8105714285714286e-05,
      "loss": 2.705,
      "step": 1990
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 319.6245422363281,
      "learning_rate": 1.809619047619048e-05,
      "loss": 4.2966,
      "step": 2000
    },
    {
      "epoch": 0.28714285714285714,
      "grad_norm": 510.7989196777344,
      "learning_rate": 1.8086666666666667e-05,
      "loss": 4.8344,
      "step": 2010
    },
    {
      "epoch": 0.2885714285714286,
      "grad_norm": 4121417.0,
      "learning_rate": 1.8077142857142856e-05,
      "loss": 2.8616,
      "step": 2020
    },
    {
      "epoch": 0.29,
      "grad_norm": 954.7911987304688,
      "learning_rate": 1.806761904761905e-05,
      "loss": 2.7943,
      "step": 2030
    },
    {
      "epoch": 0.2914285714285714,
      "grad_norm": 14565.4482421875,
      "learning_rate": 1.8058095238095238e-05,
      "loss": 5.1063,
      "step": 2040
    },
    {
      "epoch": 0.29285714285714287,
      "grad_norm": 3.5497238636016846,
      "learning_rate": 1.804857142857143e-05,
      "loss": 3.664,
      "step": 2050
    },
    {
      "epoch": 0.29428571428571426,
      "grad_norm": 4154798.5,
      "learning_rate": 1.803904761904762e-05,
      "loss": 3.6396,
      "step": 2060
    },
    {
      "epoch": 0.2957142857142857,
      "grad_norm": 9222.0986328125,
      "learning_rate": 1.802952380952381e-05,
      "loss": 4.8363,
      "step": 2070
    },
    {
      "epoch": 0.29714285714285715,
      "grad_norm": 0.054987553507089615,
      "learning_rate": 1.802e-05,
      "loss": 2.6011,
      "step": 2080
    },
    {
      "epoch": 0.2985714285714286,
      "grad_norm": 4229905.0,
      "learning_rate": 1.8010476190476193e-05,
      "loss": 5.4407,
      "step": 2090
    },
    {
      "epoch": 0.3,
      "grad_norm": 4184132.0,
      "learning_rate": 1.8000952380952385e-05,
      "loss": 2.62,
      "step": 2100
    },
    {
      "epoch": 0.30142857142857143,
      "grad_norm": 0.1274380385875702,
      "learning_rate": 1.7991428571428574e-05,
      "loss": 5.3366,
      "step": 2110
    },
    {
      "epoch": 0.3028571428571429,
      "grad_norm": 1710.6993408203125,
      "learning_rate": 1.7981904761904763e-05,
      "loss": 2.4021,
      "step": 2120
    },
    {
      "epoch": 0.30428571428571427,
      "grad_norm": 0.0698726698756218,
      "learning_rate": 1.7972380952380955e-05,
      "loss": 2.9571,
      "step": 2130
    },
    {
      "epoch": 0.3057142857142857,
      "grad_norm": 7477561.5,
      "learning_rate": 1.7962857142857144e-05,
      "loss": 4.1564,
      "step": 2140
    },
    {
      "epoch": 0.30714285714285716,
      "grad_norm": 0.07299598306417465,
      "learning_rate": 1.7953333333333333e-05,
      "loss": 1.7192,
      "step": 2150
    },
    {
      "epoch": 0.30857142857142855,
      "grad_norm": 0.0664527490735054,
      "learning_rate": 1.7943809523809525e-05,
      "loss": 5.4473,
      "step": 2160
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.053739454597234726,
      "learning_rate": 1.7934285714285714e-05,
      "loss": 3.1096,
      "step": 2170
    },
    {
      "epoch": 0.31142857142857144,
      "grad_norm": 3163786.75,
      "learning_rate": 1.7924761904761907e-05,
      "loss": 2.5493,
      "step": 2180
    },
    {
      "epoch": 0.31285714285714283,
      "grad_norm": 7008768.5,
      "learning_rate": 1.7915238095238096e-05,
      "loss": 3.9422,
      "step": 2190
    },
    {
      "epoch": 0.3142857142857143,
      "grad_norm": 0.46840348839759827,
      "learning_rate": 1.7905714285714285e-05,
      "loss": 3.2266,
      "step": 2200
    },
    {
      "epoch": 0.3157142857142857,
      "grad_norm": 5359808.0,
      "learning_rate": 1.7896190476190477e-05,
      "loss": 4.2985,
      "step": 2210
    },
    {
      "epoch": 0.3171428571428571,
      "grad_norm": 429208.75,
      "learning_rate": 1.788666666666667e-05,
      "loss": 1.5939,
      "step": 2220
    },
    {
      "epoch": 0.31857142857142856,
      "grad_norm": 4877.22216796875,
      "learning_rate": 1.7877142857142858e-05,
      "loss": 2.91,
      "step": 2230
    },
    {
      "epoch": 0.32,
      "grad_norm": 12945198.0,
      "learning_rate": 1.786761904761905e-05,
      "loss": 3.1451,
      "step": 2240
    },
    {
      "epoch": 0.32142857142857145,
      "grad_norm": 13.191624641418457,
      "learning_rate": 1.785809523809524e-05,
      "loss": 3.1815,
      "step": 2250
    },
    {
      "epoch": 0.32285714285714284,
      "grad_norm": 0.13054807484149933,
      "learning_rate": 1.7848571428571432e-05,
      "loss": 2.9765,
      "step": 2260
    },
    {
      "epoch": 0.3242857142857143,
      "grad_norm": 0.08028245717287064,
      "learning_rate": 1.783904761904762e-05,
      "loss": 5.012,
      "step": 2270
    },
    {
      "epoch": 0.32571428571428573,
      "grad_norm": 0.7741024494171143,
      "learning_rate": 1.782952380952381e-05,
      "loss": 1.6718,
      "step": 2280
    },
    {
      "epoch": 0.3271428571428571,
      "grad_norm": 0.3882508873939514,
      "learning_rate": 1.7820000000000002e-05,
      "loss": 3.201,
      "step": 2290
    },
    {
      "epoch": 0.32857142857142857,
      "grad_norm": 0.11589991301298141,
      "learning_rate": 1.781047619047619e-05,
      "loss": 4.7258,
      "step": 2300
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6033210158348083,
      "learning_rate": 1.7800952380952383e-05,
      "loss": 4.9354,
      "step": 2310
    },
    {
      "epoch": 0.3314285714285714,
      "grad_norm": 0.11468209326267242,
      "learning_rate": 1.7791428571428572e-05,
      "loss": 0.5545,
      "step": 2320
    },
    {
      "epoch": 0.33285714285714285,
      "grad_norm": 0.1605638712644577,
      "learning_rate": 1.778190476190476e-05,
      "loss": 3.1421,
      "step": 2330
    },
    {
      "epoch": 0.3342857142857143,
      "grad_norm": 359176160.0,
      "learning_rate": 1.7772380952380954e-05,
      "loss": 10.0451,
      "step": 2340
    },
    {
      "epoch": 0.3357142857142857,
      "grad_norm": 7101215.0,
      "learning_rate": 1.7762857142857143e-05,
      "loss": 8.9785,
      "step": 2350
    },
    {
      "epoch": 0.33714285714285713,
      "grad_norm": 435591424.0,
      "learning_rate": 1.7753333333333335e-05,
      "loss": 3.7182,
      "step": 2360
    },
    {
      "epoch": 0.3385714285714286,
      "grad_norm": 2.6491806507110596,
      "learning_rate": 1.7743809523809524e-05,
      "loss": 4.2943,
      "step": 2370
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4358234703540802,
      "learning_rate": 1.7734285714285716e-05,
      "loss": 4.5147,
      "step": 2380
    },
    {
      "epoch": 0.3414285714285714,
      "grad_norm": 2922213.0,
      "learning_rate": 1.7724761904761905e-05,
      "loss": 9.4796,
      "step": 2390
    },
    {
      "epoch": 0.34285714285714286,
      "grad_norm": 4287912.5,
      "learning_rate": 1.7715238095238098e-05,
      "loss": 10.3668,
      "step": 2400
    },
    {
      "epoch": 0.3442857142857143,
      "grad_norm": 24.69457244873047,
      "learning_rate": 1.7705714285714287e-05,
      "loss": 6.6808,
      "step": 2410
    },
    {
      "epoch": 0.3457142857142857,
      "grad_norm": 525.4225463867188,
      "learning_rate": 1.769619047619048e-05,
      "loss": 4.8519,
      "step": 2420
    },
    {
      "epoch": 0.34714285714285714,
      "grad_norm": 8149897.0,
      "learning_rate": 1.7686666666666668e-05,
      "loss": 4.4684,
      "step": 2430
    },
    {
      "epoch": 0.3485714285714286,
      "grad_norm": 4240109.5,
      "learning_rate": 1.767714285714286e-05,
      "loss": 7.0125,
      "step": 2440
    },
    {
      "epoch": 0.35,
      "grad_norm": 6134.9345703125,
      "learning_rate": 1.766761904761905e-05,
      "loss": 6.3019,
      "step": 2450
    },
    {
      "epoch": 0.3514285714285714,
      "grad_norm": 24740158.0,
      "learning_rate": 1.7658095238095238e-05,
      "loss": 3.9963,
      "step": 2460
    },
    {
      "epoch": 0.35285714285714287,
      "grad_norm": 8690219.0,
      "learning_rate": 1.764857142857143e-05,
      "loss": 3.4266,
      "step": 2470
    },
    {
      "epoch": 0.35428571428571426,
      "grad_norm": 184204.015625,
      "learning_rate": 1.763904761904762e-05,
      "loss": 3.3633,
      "step": 2480
    },
    {
      "epoch": 0.3557142857142857,
      "grad_norm": 0.29240667819976807,
      "learning_rate": 1.7629523809523812e-05,
      "loss": 0.5219,
      "step": 2490
    },
    {
      "epoch": 0.35714285714285715,
      "grad_norm": 4005113.5,
      "learning_rate": 1.762e-05,
      "loss": 3.7876,
      "step": 2500
    },
    {
      "epoch": 0.3585714285714286,
      "grad_norm": 0.5350474119186401,
      "learning_rate": 1.761047619047619e-05,
      "loss": 3.9895,
      "step": 2510
    },
    {
      "epoch": 0.36,
      "grad_norm": 32.12314224243164,
      "learning_rate": 1.7600952380952382e-05,
      "loss": 4.8755,
      "step": 2520
    },
    {
      "epoch": 0.36142857142857143,
      "grad_norm": 0.0640583336353302,
      "learning_rate": 1.759142857142857e-05,
      "loss": 3.6898,
      "step": 2530
    },
    {
      "epoch": 0.3628571428571429,
      "grad_norm": 0.6821596622467041,
      "learning_rate": 1.7581904761904763e-05,
      "loss": 4.3172,
      "step": 2540
    },
    {
      "epoch": 0.36428571428571427,
      "grad_norm": 3126204.0,
      "learning_rate": 1.7572380952380956e-05,
      "loss": 6.7888,
      "step": 2550
    },
    {
      "epoch": 0.3657142857142857,
      "grad_norm": 0.13851745426654816,
      "learning_rate": 1.7562857142857145e-05,
      "loss": 6.0354,
      "step": 2560
    },
    {
      "epoch": 0.36714285714285716,
      "grad_norm": 61.239349365234375,
      "learning_rate": 1.7553333333333337e-05,
      "loss": 4.8022,
      "step": 2570
    },
    {
      "epoch": 0.36857142857142855,
      "grad_norm": 85.11653900146484,
      "learning_rate": 1.7543809523809526e-05,
      "loss": 7.9517,
      "step": 2580
    },
    {
      "epoch": 0.37,
      "grad_norm": 21.397722244262695,
      "learning_rate": 1.7534285714285715e-05,
      "loss": 2.9253,
      "step": 2590
    },
    {
      "epoch": 0.37142857142857144,
      "grad_norm": 2.1357691287994385,
      "learning_rate": 1.7524761904761907e-05,
      "loss": 4.4413,
      "step": 2600
    },
    {
      "epoch": 0.37285714285714283,
      "grad_norm": 2482.100830078125,
      "learning_rate": 1.7515238095238096e-05,
      "loss": 3.2157,
      "step": 2610
    },
    {
      "epoch": 0.3742857142857143,
      "grad_norm": 0.126044362783432,
      "learning_rate": 1.750571428571429e-05,
      "loss": 3.3892,
      "step": 2620
    },
    {
      "epoch": 0.3757142857142857,
      "grad_norm": 201.6053466796875,
      "learning_rate": 1.7496190476190477e-05,
      "loss": 3.3951,
      "step": 2630
    },
    {
      "epoch": 0.37714285714285717,
      "grad_norm": 3601545.25,
      "learning_rate": 1.7486666666666666e-05,
      "loss": 4.0292,
      "step": 2640
    },
    {
      "epoch": 0.37857142857142856,
      "grad_norm": 2746637.75,
      "learning_rate": 1.747714285714286e-05,
      "loss": 3.2878,
      "step": 2650
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2718538045883179,
      "learning_rate": 1.7467619047619048e-05,
      "loss": 6.2071,
      "step": 2660
    },
    {
      "epoch": 0.38142857142857145,
      "grad_norm": 0.38920724391937256,
      "learning_rate": 1.745809523809524e-05,
      "loss": 4.8498,
      "step": 2670
    },
    {
      "epoch": 0.38285714285714284,
      "grad_norm": 0.2845199704170227,
      "learning_rate": 1.744857142857143e-05,
      "loss": 4.9194,
      "step": 2680
    },
    {
      "epoch": 0.3842857142857143,
      "grad_norm": 0.30018651485443115,
      "learning_rate": 1.743904761904762e-05,
      "loss": 3.3057,
      "step": 2690
    },
    {
      "epoch": 0.38571428571428573,
      "grad_norm": 2774187.25,
      "learning_rate": 1.742952380952381e-05,
      "loss": 3.2186,
      "step": 2700
    },
    {
      "epoch": 0.3871428571428571,
      "grad_norm": 2917556.75,
      "learning_rate": 1.7420000000000003e-05,
      "loss": 3.1508,
      "step": 2710
    },
    {
      "epoch": 0.38857142857142857,
      "grad_norm": 2966650.5,
      "learning_rate": 1.741047619047619e-05,
      "loss": 8.0832,
      "step": 2720
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.35328835248947144,
      "learning_rate": 1.7400952380952384e-05,
      "loss": 3.039,
      "step": 2730
    },
    {
      "epoch": 0.3914285714285714,
      "grad_norm": 0.6767052412033081,
      "learning_rate": 1.7391428571428573e-05,
      "loss": 6.309,
      "step": 2740
    },
    {
      "epoch": 0.39285714285714285,
      "grad_norm": 0.42332029342651367,
      "learning_rate": 1.7381904761904765e-05,
      "loss": 3.051,
      "step": 2750
    },
    {
      "epoch": 0.3942857142857143,
      "grad_norm": 0.35865992307662964,
      "learning_rate": 1.7372380952380954e-05,
      "loss": 3.1789,
      "step": 2760
    },
    {
      "epoch": 0.39571428571428574,
      "grad_norm": 2720728.0,
      "learning_rate": 1.7362857142857143e-05,
      "loss": 4.8344,
      "step": 2770
    },
    {
      "epoch": 0.39714285714285713,
      "grad_norm": 0.1094881147146225,
      "learning_rate": 1.7353333333333335e-05,
      "loss": 0.0,
      "step": 2780
    },
    {
      "epoch": 0.3985714285714286,
      "grad_norm": 0.10334648936986923,
      "learning_rate": 1.7343809523809524e-05,
      "loss": 6.4609,
      "step": 2790
    },
    {
      "epoch": 0.4,
      "grad_norm": 2929750.0,
      "learning_rate": 1.7334285714285713e-05,
      "loss": 6.3594,
      "step": 2800
    },
    {
      "epoch": 0.4014285714285714,
      "grad_norm": 3.4951541423797607,
      "learning_rate": 1.7324761904761906e-05,
      "loss": 6.3399,
      "step": 2810
    },
    {
      "epoch": 0.40285714285714286,
      "grad_norm": 81.97663879394531,
      "learning_rate": 1.7315238095238095e-05,
      "loss": 8.841,
      "step": 2820
    },
    {
      "epoch": 0.4042857142857143,
      "grad_norm": 0.9684914946556091,
      "learning_rate": 1.7305714285714287e-05,
      "loss": 1.4786,
      "step": 2830
    },
    {
      "epoch": 0.4057142857142857,
      "grad_norm": 2980061.75,
      "learning_rate": 1.7296190476190476e-05,
      "loss": 7.6426,
      "step": 2840
    },
    {
      "epoch": 0.40714285714285714,
      "grad_norm": 3.7037432193756104,
      "learning_rate": 1.7286666666666668e-05,
      "loss": 4.3771,
      "step": 2850
    },
    {
      "epoch": 0.4085714285714286,
      "grad_norm": 0.4184705317020416,
      "learning_rate": 1.727714285714286e-05,
      "loss": 4.6235,
      "step": 2860
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.30692318081855774,
      "learning_rate": 1.726761904761905e-05,
      "loss": 4.5594,
      "step": 2870
    },
    {
      "epoch": 0.4114285714285714,
      "grad_norm": 6.405407428741455,
      "learning_rate": 1.7258095238095242e-05,
      "loss": 4.7619,
      "step": 2880
    },
    {
      "epoch": 0.41285714285714287,
      "grad_norm": 0.329047292470932,
      "learning_rate": 1.724857142857143e-05,
      "loss": 3.1981,
      "step": 2890
    },
    {
      "epoch": 0.4142857142857143,
      "grad_norm": 0.3733239769935608,
      "learning_rate": 1.723904761904762e-05,
      "loss": 1.6408,
      "step": 2900
    },
    {
      "epoch": 0.4157142857142857,
      "grad_norm": 0.33396831154823303,
      "learning_rate": 1.7229523809523812e-05,
      "loss": 7.9085,
      "step": 2910
    },
    {
      "epoch": 0.41714285714285715,
      "grad_norm": 0.6095501780509949,
      "learning_rate": 1.722e-05,
      "loss": 4.8165,
      "step": 2920
    },
    {
      "epoch": 0.4185714285714286,
      "grad_norm": 0.27922093868255615,
      "learning_rate": 1.721047619047619e-05,
      "loss": 3.302,
      "step": 2930
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3227151930332184,
      "learning_rate": 1.7200952380952382e-05,
      "loss": 8.1122,
      "step": 2940
    },
    {
      "epoch": 0.42142857142857143,
      "grad_norm": 0.32717186212539673,
      "learning_rate": 1.719142857142857e-05,
      "loss": 0.0,
      "step": 2950
    },
    {
      "epoch": 0.4228571428571429,
      "grad_norm": 0.32608669996261597,
      "learning_rate": 1.7181904761904764e-05,
      "loss": 6.544,
      "step": 2960
    },
    {
      "epoch": 0.42428571428571427,
      "grad_norm": 0.3141319155693054,
      "learning_rate": 1.7172380952380953e-05,
      "loss": 4.793,
      "step": 2970
    },
    {
      "epoch": 0.4257142857142857,
      "grad_norm": 0.30614757537841797,
      "learning_rate": 1.716285714285714e-05,
      "loss": 3.2015,
      "step": 2980
    },
    {
      "epoch": 0.42714285714285716,
      "grad_norm": 0.2856157124042511,
      "learning_rate": 1.7153333333333334e-05,
      "loss": 4.8088,
      "step": 2990
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.29165908694267273,
      "learning_rate": 1.7143809523809526e-05,
      "loss": 3.2519,
      "step": 3000
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3221181035041809,
      "learning_rate": 1.7134285714285715e-05,
      "loss": 3.209,
      "step": 3010
    },
    {
      "epoch": 0.43142857142857144,
      "grad_norm": 0.2804566025733948,
      "learning_rate": 1.7124761904761908e-05,
      "loss": 1.5999,
      "step": 3020
    },
    {
      "epoch": 0.4328571428571429,
      "grad_norm": 2879736.5,
      "learning_rate": 1.7115238095238097e-05,
      "loss": 4.9121,
      "step": 3030
    },
    {
      "epoch": 0.4342857142857143,
      "grad_norm": 0.3359666168689728,
      "learning_rate": 1.710571428571429e-05,
      "loss": 3.2839,
      "step": 3040
    },
    {
      "epoch": 0.4357142857142857,
      "grad_norm": 0.3057880401611328,
      "learning_rate": 1.7096190476190478e-05,
      "loss": 1.5798,
      "step": 3050
    },
    {
      "epoch": 0.43714285714285717,
      "grad_norm": 2721627.0,
      "learning_rate": 1.7086666666666667e-05,
      "loss": 4.8672,
      "step": 3060
    },
    {
      "epoch": 0.43857142857142856,
      "grad_norm": 2738713.25,
      "learning_rate": 1.707714285714286e-05,
      "loss": 3.1883,
      "step": 3070
    },
    {
      "epoch": 0.44,
      "grad_norm": 2733585.75,
      "learning_rate": 1.7067619047619048e-05,
      "loss": 7.6691,
      "step": 3080
    },
    {
      "epoch": 0.44142857142857145,
      "grad_norm": 0.5917357206344604,
      "learning_rate": 1.705809523809524e-05,
      "loss": 4.7835,
      "step": 3090
    },
    {
      "epoch": 0.44285714285714284,
      "grad_norm": 0.7100284099578857,
      "learning_rate": 1.704857142857143e-05,
      "loss": 4.7959,
      "step": 3100
    },
    {
      "epoch": 0.4442857142857143,
      "grad_norm": 0.30806609988212585,
      "learning_rate": 1.703904761904762e-05,
      "loss": 4.7337,
      "step": 3110
    },
    {
      "epoch": 0.44571428571428573,
      "grad_norm": 2760101.25,
      "learning_rate": 1.702952380952381e-05,
      "loss": 7.8048,
      "step": 3120
    },
    {
      "epoch": 0.4471428571428571,
      "grad_norm": 3180642.5,
      "learning_rate": 1.702e-05,
      "loss": 7.8035,
      "step": 3130
    },
    {
      "epoch": 0.44857142857142857,
      "grad_norm": 12.569872856140137,
      "learning_rate": 1.7010476190476192e-05,
      "loss": 0.0,
      "step": 3140
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.28412607312202454,
      "learning_rate": 1.700095238095238e-05,
      "loss": 4.6876,
      "step": 3150
    },
    {
      "epoch": 0.4514285714285714,
      "grad_norm": 0.26430854201316833,
      "learning_rate": 1.6991428571428573e-05,
      "loss": 3.1382,
      "step": 3160
    },
    {
      "epoch": 0.45285714285714285,
      "grad_norm": 0.6566365957260132,
      "learning_rate": 1.6981904761904762e-05,
      "loss": 1.4797,
      "step": 3170
    },
    {
      "epoch": 0.4542857142857143,
      "grad_norm": 0.09471367299556732,
      "learning_rate": 1.6972380952380955e-05,
      "loss": 1.5373,
      "step": 3180
    },
    {
      "epoch": 0.45571428571428574,
      "grad_norm": 2678757.0,
      "learning_rate": 1.6962857142857147e-05,
      "loss": 3.219,
      "step": 3190
    },
    {
      "epoch": 0.45714285714285713,
      "grad_norm": 0.5344274640083313,
      "learning_rate": 1.6953333333333336e-05,
      "loss": 3.2999,
      "step": 3200
    },
    {
      "epoch": 0.4585714285714286,
      "grad_norm": 0.6432442665100098,
      "learning_rate": 1.6943809523809525e-05,
      "loss": 6.5689,
      "step": 3210
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.27346470952033997,
      "learning_rate": 1.6934285714285717e-05,
      "loss": 3.2394,
      "step": 3220
    },
    {
      "epoch": 0.4614285714285714,
      "grad_norm": 0.27335426211357117,
      "learning_rate": 1.6924761904761906e-05,
      "loss": 7.9203,
      "step": 3230
    },
    {
      "epoch": 0.46285714285714286,
      "grad_norm": 986.945556640625,
      "learning_rate": 1.6915238095238095e-05,
      "loss": 0.0,
      "step": 3240
    },
    {
      "epoch": 0.4642857142857143,
      "grad_norm": 0.11446543037891388,
      "learning_rate": 1.6905714285714287e-05,
      "loss": 1.4975,
      "step": 3250
    },
    {
      "epoch": 0.4657142857142857,
      "grad_norm": 0.19220978021621704,
      "learning_rate": 1.6896190476190476e-05,
      "loss": 0.0,
      "step": 3260
    },
    {
      "epoch": 0.46714285714285714,
      "grad_norm": 0.3593844771385193,
      "learning_rate": 1.688666666666667e-05,
      "loss": 4.9805,
      "step": 3270
    },
    {
      "epoch": 0.4685714285714286,
      "grad_norm": 0.11434998363256454,
      "learning_rate": 1.6877142857142858e-05,
      "loss": 1.727,
      "step": 3280
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.13721834123134613,
      "learning_rate": 1.6867619047619047e-05,
      "loss": 4.9358,
      "step": 3290
    },
    {
      "epoch": 0.4714285714285714,
      "grad_norm": 0.5732265114784241,
      "learning_rate": 1.685809523809524e-05,
      "loss": 0.0,
      "step": 3300
    },
    {
      "epoch": 0.47285714285714286,
      "grad_norm": 0.0743197649717331,
      "learning_rate": 1.684857142857143e-05,
      "loss": 8.125,
      "step": 3310
    },
    {
      "epoch": 0.4742857142857143,
      "grad_norm": 0.08995280414819717,
      "learning_rate": 1.683904761904762e-05,
      "loss": 3.259,
      "step": 3320
    },
    {
      "epoch": 0.4757142857142857,
      "grad_norm": 0.31050562858581543,
      "learning_rate": 1.6829523809523813e-05,
      "loss": 3.117,
      "step": 3330
    },
    {
      "epoch": 0.47714285714285715,
      "grad_norm": 0.6722134351730347,
      "learning_rate": 1.682e-05,
      "loss": 4.7582,
      "step": 3340
    },
    {
      "epoch": 0.4785714285714286,
      "grad_norm": 2842511.75,
      "learning_rate": 1.6810476190476194e-05,
      "loss": 1.7488,
      "step": 3350
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.11184928566217422,
      "learning_rate": 1.6800952380952383e-05,
      "loss": 0.0,
      "step": 3360
    },
    {
      "epoch": 0.48142857142857143,
      "grad_norm": 0.08614572137594223,
      "learning_rate": 1.6791428571428572e-05,
      "loss": 1.7377,
      "step": 3370
    },
    {
      "epoch": 0.4828571428571429,
      "grad_norm": 0.11438575387001038,
      "learning_rate": 1.6781904761904764e-05,
      "loss": 3.4229,
      "step": 3380
    },
    {
      "epoch": 0.48428571428571426,
      "grad_norm": 2896058.5,
      "learning_rate": 1.6772380952380953e-05,
      "loss": 3.4226,
      "step": 3390
    },
    {
      "epoch": 0.4857142857142857,
      "grad_norm": 0.28166908025741577,
      "learning_rate": 1.6762857142857145e-05,
      "loss": 1.6505,
      "step": 3400
    },
    {
      "epoch": 0.48714285714285716,
      "grad_norm": 0.08594900369644165,
      "learning_rate": 1.6753333333333334e-05,
      "loss": 1.687,
      "step": 3410
    },
    {
      "epoch": 0.48857142857142855,
      "grad_norm": 0.2854008972644806,
      "learning_rate": 1.6743809523809523e-05,
      "loss": 1.6237,
      "step": 3420
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.10124766081571579,
      "learning_rate": 1.6734285714285716e-05,
      "loss": 3.3169,
      "step": 3430
    },
    {
      "epoch": 0.49142857142857144,
      "grad_norm": 0.25782522559165955,
      "learning_rate": 1.6724761904761905e-05,
      "loss": 3.2946,
      "step": 3440
    },
    {
      "epoch": 0.4928571428571429,
      "grad_norm": 2700739.5,
      "learning_rate": 1.6715238095238097e-05,
      "loss": 1.6761,
      "step": 3450
    },
    {
      "epoch": 0.4942857142857143,
      "grad_norm": 0.3033803701400757,
      "learning_rate": 1.6705714285714286e-05,
      "loss": 4.9379,
      "step": 3460
    },
    {
      "epoch": 0.4957142857142857,
      "grad_norm": 0.28263384103775024,
      "learning_rate": 1.669619047619048e-05,
      "loss": 1.6481,
      "step": 3470
    },
    {
      "epoch": 0.49714285714285716,
      "grad_norm": 0.10296333581209183,
      "learning_rate": 1.6686666666666667e-05,
      "loss": 3.2453,
      "step": 3480
    },
    {
      "epoch": 0.49857142857142855,
      "grad_norm": 2657473.75,
      "learning_rate": 1.667714285714286e-05,
      "loss": 4.9642,
      "step": 3490
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.11391142755746841,
      "learning_rate": 1.666761904761905e-05,
      "loss": 0.0,
      "step": 3500
    },
    {
      "epoch": 0.5014285714285714,
      "grad_norm": 0.25250235199928284,
      "learning_rate": 1.665809523809524e-05,
      "loss": 3.2724,
      "step": 3510
    },
    {
      "epoch": 0.5028571428571429,
      "grad_norm": 0.12918342649936676,
      "learning_rate": 1.664857142857143e-05,
      "loss": 1.654,
      "step": 3520
    },
    {
      "epoch": 0.5042857142857143,
      "grad_norm": 0.3773730397224426,
      "learning_rate": 1.6639047619047622e-05,
      "loss": 4.9128,
      "step": 3530
    },
    {
      "epoch": 0.5057142857142857,
      "grad_norm": 0.08973002433776855,
      "learning_rate": 1.662952380952381e-05,
      "loss": 4.8723,
      "step": 3540
    },
    {
      "epoch": 0.5071428571428571,
      "grad_norm": 2782727.5,
      "learning_rate": 1.662e-05,
      "loss": 3.1804,
      "step": 3550
    },
    {
      "epoch": 0.5085714285714286,
      "grad_norm": 0.0862429067492485,
      "learning_rate": 1.6610476190476192e-05,
      "loss": 3.2415,
      "step": 3560
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.08203268051147461,
      "learning_rate": 1.660095238095238e-05,
      "loss": 4.8707,
      "step": 3570
    },
    {
      "epoch": 0.5114285714285715,
      "grad_norm": 0.424404501914978,
      "learning_rate": 1.6591428571428574e-05,
      "loss": 3.2748,
      "step": 3580
    },
    {
      "epoch": 0.5128571428571429,
      "grad_norm": 0.10740084946155548,
      "learning_rate": 1.6581904761904763e-05,
      "loss": 4.7792,
      "step": 3590
    },
    {
      "epoch": 0.5142857142857142,
      "grad_norm": 0.32315415143966675,
      "learning_rate": 1.657238095238095e-05,
      "loss": 7.869,
      "step": 3600
    },
    {
      "epoch": 0.5157142857142857,
      "grad_norm": 0.3686506748199463,
      "learning_rate": 1.6562857142857144e-05,
      "loss": 0.0,
      "step": 3610
    },
    {
      "epoch": 0.5171428571428571,
      "grad_norm": 0.07097970694303513,
      "learning_rate": 1.6553333333333333e-05,
      "loss": 1.4624,
      "step": 3620
    },
    {
      "epoch": 0.5185714285714286,
      "grad_norm": 0.437063992023468,
      "learning_rate": 1.6543809523809525e-05,
      "loss": 4.9358,
      "step": 3630
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.27448156476020813,
      "learning_rate": 1.6534285714285718e-05,
      "loss": 6.6685,
      "step": 3640
    },
    {
      "epoch": 0.5214285714285715,
      "grad_norm": 0.36948761343955994,
      "learning_rate": 1.6524761904761907e-05,
      "loss": 3.1539,
      "step": 3650
    },
    {
      "epoch": 0.5228571428571429,
      "grad_norm": 0.703045666217804,
      "learning_rate": 1.65152380952381e-05,
      "loss": 3.3015,
      "step": 3660
    },
    {
      "epoch": 0.5242857142857142,
      "grad_norm": 0.27631494402885437,
      "learning_rate": 1.6505714285714288e-05,
      "loss": 4.6988,
      "step": 3670
    },
    {
      "epoch": 0.5257142857142857,
      "grad_norm": 0.39693719148635864,
      "learning_rate": 1.6496190476190477e-05,
      "loss": 1.6412,
      "step": 3680
    },
    {
      "epoch": 0.5271428571428571,
      "grad_norm": 0.26690158247947693,
      "learning_rate": 1.648666666666667e-05,
      "loss": 0.0,
      "step": 3690
    },
    {
      "epoch": 0.5285714285714286,
      "grad_norm": 0.0701303705573082,
      "learning_rate": 1.6477142857142858e-05,
      "loss": 1.6379,
      "step": 3700
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.07016340643167496,
      "learning_rate": 1.646761904761905e-05,
      "loss": 0.0,
      "step": 3710
    },
    {
      "epoch": 0.5314285714285715,
      "grad_norm": 2810315.75,
      "learning_rate": 1.645809523809524e-05,
      "loss": 9.9234,
      "step": 3720
    },
    {
      "epoch": 0.5328571428571428,
      "grad_norm": 0.06698928773403168,
      "learning_rate": 1.644857142857143e-05,
      "loss": 0.0,
      "step": 3730
    },
    {
      "epoch": 0.5342857142857143,
      "grad_norm": 2845855.0,
      "learning_rate": 1.643904761904762e-05,
      "loss": 3.284,
      "step": 3740
    },
    {
      "epoch": 0.5357142857142857,
      "grad_norm": 0.3084893822669983,
      "learning_rate": 1.642952380952381e-05,
      "loss": 3.3501,
      "step": 3750
    },
    {
      "epoch": 0.5371428571428571,
      "grad_norm": 0.28991755843162537,
      "learning_rate": 1.6420000000000002e-05,
      "loss": 4.9608,
      "step": 3760
    },
    {
      "epoch": 0.5385714285714286,
      "grad_norm": 0.3004452586174011,
      "learning_rate": 1.641047619047619e-05,
      "loss": 1.6338,
      "step": 3770
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.325141578912735,
      "learning_rate": 1.6400952380952383e-05,
      "loss": 6.5229,
      "step": 3780
    },
    {
      "epoch": 0.5414285714285715,
      "grad_norm": 0.36467835307121277,
      "learning_rate": 1.6391428571428572e-05,
      "loss": 1.5888,
      "step": 3790
    },
    {
      "epoch": 0.5428571428571428,
      "grad_norm": 0.30043649673461914,
      "learning_rate": 1.6381904761904765e-05,
      "loss": 8.1121,
      "step": 3800
    },
    {
      "epoch": 0.5442857142857143,
      "grad_norm": 0.3594944179058075,
      "learning_rate": 1.6372380952380954e-05,
      "loss": 4.7767,
      "step": 3810
    },
    {
      "epoch": 0.5457142857142857,
      "grad_norm": 0.32140281796455383,
      "learning_rate": 1.6362857142857146e-05,
      "loss": 3.1709,
      "step": 3820
    },
    {
      "epoch": 0.5471428571428572,
      "grad_norm": 0.28547146916389465,
      "learning_rate": 1.6353333333333335e-05,
      "loss": 4.7352,
      "step": 3830
    },
    {
      "epoch": 0.5485714285714286,
      "grad_norm": 0.302616149187088,
      "learning_rate": 1.6343809523809527e-05,
      "loss": 3.0553,
      "step": 3840
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.10340223461389542,
      "learning_rate": 1.6334285714285716e-05,
      "loss": 1.5902,
      "step": 3850
    },
    {
      "epoch": 0.5514285714285714,
      "grad_norm": 0.2624635398387909,
      "learning_rate": 1.6324761904761905e-05,
      "loss": 9.5575,
      "step": 3860
    },
    {
      "epoch": 0.5528571428571428,
      "grad_norm": 0.34584909677505493,
      "learning_rate": 1.6315238095238097e-05,
      "loss": 4.5919,
      "step": 3870
    },
    {
      "epoch": 0.5542857142857143,
      "grad_norm": 2975829.0,
      "learning_rate": 1.6305714285714286e-05,
      "loss": 3.1257,
      "step": 3880
    },
    {
      "epoch": 0.5557142857142857,
      "grad_norm": 3027449.5,
      "learning_rate": 1.6296190476190475e-05,
      "loss": 4.6669,
      "step": 3890
    },
    {
      "epoch": 0.5571428571428572,
      "grad_norm": 0.3074139356613159,
      "learning_rate": 1.6286666666666668e-05,
      "loss": 3.0835,
      "step": 3900
    },
    {
      "epoch": 0.5585714285714286,
      "grad_norm": 0.6759763360023499,
      "learning_rate": 1.6277142857142857e-05,
      "loss": 4.6413,
      "step": 3910
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.13771891593933105,
      "learning_rate": 1.626761904761905e-05,
      "loss": 3.1148,
      "step": 3920
    },
    {
      "epoch": 0.5614285714285714,
      "grad_norm": 2683646.75,
      "learning_rate": 1.6258095238095238e-05,
      "loss": 3.1832,
      "step": 3930
    },
    {
      "epoch": 0.5628571428571428,
      "grad_norm": 0.27801549434661865,
      "learning_rate": 1.624857142857143e-05,
      "loss": 4.7509,
      "step": 3940
    },
    {
      "epoch": 0.5642857142857143,
      "grad_norm": 83098.1796875,
      "learning_rate": 1.623904761904762e-05,
      "loss": 4.7437,
      "step": 3950
    },
    {
      "epoch": 0.5657142857142857,
      "grad_norm": 2785081.0,
      "learning_rate": 1.622952380952381e-05,
      "loss": 3.1561,
      "step": 3960
    },
    {
      "epoch": 0.5671428571428572,
      "grad_norm": 0.07642267644405365,
      "learning_rate": 1.6220000000000004e-05,
      "loss": 3.224,
      "step": 3970
    },
    {
      "epoch": 0.5685714285714286,
      "grad_norm": 2736332.25,
      "learning_rate": 1.6210476190476193e-05,
      "loss": 3.3139,
      "step": 3980
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.06574222445487976,
      "learning_rate": 1.6200952380952382e-05,
      "loss": 4.7511,
      "step": 3990
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.06725277006626129,
      "learning_rate": 1.6191428571428574e-05,
      "loss": 3.0476,
      "step": 4000
    },
    {
      "epoch": 0.5728571428571428,
      "grad_norm": 2739118.0,
      "learning_rate": 1.6181904761904763e-05,
      "loss": 3.1532,
      "step": 4010
    },
    {
      "epoch": 0.5742857142857143,
      "grad_norm": 0.37368476390838623,
      "learning_rate": 1.6172380952380952e-05,
      "loss": 3.094,
      "step": 4020
    },
    {
      "epoch": 0.5757142857142857,
      "grad_norm": 0.5981259942054749,
      "learning_rate": 1.6162857142857144e-05,
      "loss": 1.5825,
      "step": 4030
    },
    {
      "epoch": 0.5771428571428572,
      "grad_norm": 0.038421131670475006,
      "learning_rate": 1.6153333333333333e-05,
      "loss": 1.5637,
      "step": 4040
    },
    {
      "epoch": 0.5785714285714286,
      "grad_norm": 0.5585459470748901,
      "learning_rate": 1.6143809523809526e-05,
      "loss": 3.2513,
      "step": 4050
    },
    {
      "epoch": 0.58,
      "grad_norm": 2897875.25,
      "learning_rate": 1.6134285714285715e-05,
      "loss": 6.5978,
      "step": 4060
    },
    {
      "epoch": 0.5814285714285714,
      "grad_norm": 0.06783616542816162,
      "learning_rate": 1.6124761904761904e-05,
      "loss": 1.6161,
      "step": 4070
    },
    {
      "epoch": 0.5828571428571429,
      "grad_norm": 2668624.25,
      "learning_rate": 1.6115238095238096e-05,
      "loss": 4.817,
      "step": 4080
    },
    {
      "epoch": 0.5842857142857143,
      "grad_norm": 0.37836769223213196,
      "learning_rate": 1.610571428571429e-05,
      "loss": 4.8278,
      "step": 4090
    },
    {
      "epoch": 0.5857142857142857,
      "grad_norm": 0.052158188074827194,
      "learning_rate": 1.6096190476190477e-05,
      "loss": 1.5347,
      "step": 4100
    },
    {
      "epoch": 0.5871428571428572,
      "grad_norm": 3082697.25,
      "learning_rate": 1.608666666666667e-05,
      "loss": 3.0838,
      "step": 4110
    },
    {
      "epoch": 0.5885714285714285,
      "grad_norm": 0.42292892932891846,
      "learning_rate": 1.607714285714286e-05,
      "loss": 3.1075,
      "step": 4120
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.09839190542697906,
      "learning_rate": 1.606761904761905e-05,
      "loss": 1.9352,
      "step": 4130
    },
    {
      "epoch": 0.5914285714285714,
      "grad_norm": 2718823.25,
      "learning_rate": 1.605809523809524e-05,
      "loss": 3.345,
      "step": 4140
    },
    {
      "epoch": 0.5928571428571429,
      "grad_norm": 0.27444520592689514,
      "learning_rate": 1.604857142857143e-05,
      "loss": 4.9641,
      "step": 4150
    },
    {
      "epoch": 0.5942857142857143,
      "grad_norm": 0.3214452564716339,
      "learning_rate": 1.603904761904762e-05,
      "loss": 1.6393,
      "step": 4160
    },
    {
      "epoch": 0.5957142857142858,
      "grad_norm": 0.25701478123664856,
      "learning_rate": 1.602952380952381e-05,
      "loss": 1.6509,
      "step": 4170
    },
    {
      "epoch": 0.5971428571428572,
      "grad_norm": 0.07796379923820496,
      "learning_rate": 1.6020000000000002e-05,
      "loss": 3.3024,
      "step": 4180
    },
    {
      "epoch": 0.5985714285714285,
      "grad_norm": 0.10487744212150574,
      "learning_rate": 1.601047619047619e-05,
      "loss": 4.8168,
      "step": 4190
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.06666464358568192,
      "learning_rate": 1.600095238095238e-05,
      "loss": 1.6254,
      "step": 4200
    },
    {
      "epoch": 0.6014285714285714,
      "grad_norm": 0.35322555899620056,
      "learning_rate": 1.5991428571428573e-05,
      "loss": 6.3597,
      "step": 4210
    },
    {
      "epoch": 0.6028571428571429,
      "grad_norm": 0.3549623489379883,
      "learning_rate": 1.598190476190476e-05,
      "loss": 4.7849,
      "step": 4220
    },
    {
      "epoch": 0.6042857142857143,
      "grad_norm": 0.37740358710289,
      "learning_rate": 1.5972380952380954e-05,
      "loss": 6.3263,
      "step": 4230
    },
    {
      "epoch": 0.6057142857142858,
      "grad_norm": 2733393.5,
      "learning_rate": 1.5962857142857143e-05,
      "loss": 1.5586,
      "step": 4240
    },
    {
      "epoch": 0.6071428571428571,
      "grad_norm": 2747028.0,
      "learning_rate": 1.5953333333333335e-05,
      "loss": 4.7417,
      "step": 4250
    },
    {
      "epoch": 0.6085714285714285,
      "grad_norm": 0.07712685316801071,
      "learning_rate": 1.5943809523809524e-05,
      "loss": 3.1359,
      "step": 4260
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4122738838195801,
      "learning_rate": 1.5934285714285717e-05,
      "loss": 1.55,
      "step": 4270
    },
    {
      "epoch": 0.6114285714285714,
      "grad_norm": 0.2621636986732483,
      "learning_rate": 1.5924761904761906e-05,
      "loss": 1.5178,
      "step": 4280
    },
    {
      "epoch": 0.6128571428571429,
      "grad_norm": 2746450.25,
      "learning_rate": 1.5915238095238098e-05,
      "loss": 4.9315,
      "step": 4290
    },
    {
      "epoch": 0.6142857142857143,
      "grad_norm": 0.08654768019914627,
      "learning_rate": 1.5905714285714287e-05,
      "loss": 0.0,
      "step": 4300
    },
    {
      "epoch": 0.6157142857142858,
      "grad_norm": 0.30355894565582275,
      "learning_rate": 1.589619047619048e-05,
      "loss": 0.0,
      "step": 4310
    },
    {
      "epoch": 0.6171428571428571,
      "grad_norm": 2722167.5,
      "learning_rate": 1.5886666666666668e-05,
      "loss": 4.7661,
      "step": 4320
    },
    {
      "epoch": 0.6185714285714285,
      "grad_norm": 2732705.25,
      "learning_rate": 1.5877142857142857e-05,
      "loss": 4.7851,
      "step": 4330
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.08668097108602524,
      "learning_rate": 1.586761904761905e-05,
      "loss": 3.3159,
      "step": 4340
    },
    {
      "epoch": 0.6214285714285714,
      "grad_norm": 0.3422408699989319,
      "learning_rate": 1.585809523809524e-05,
      "loss": 0.0,
      "step": 4350
    },
    {
      "epoch": 0.6228571428571429,
      "grad_norm": 0.9455875754356384,
      "learning_rate": 1.584857142857143e-05,
      "loss": 1.5694,
      "step": 4360
    },
    {
      "epoch": 0.6242857142857143,
      "grad_norm": 0.044478509575128555,
      "learning_rate": 1.583904761904762e-05,
      "loss": 1.6057,
      "step": 4370
    },
    {
      "epoch": 0.6257142857142857,
      "grad_norm": 0.08381345123052597,
      "learning_rate": 1.582952380952381e-05,
      "loss": 1.5658,
      "step": 4380
    },
    {
      "epoch": 0.6271428571428571,
      "grad_norm": 0.32129600644111633,
      "learning_rate": 1.582e-05,
      "loss": 3.2293,
      "step": 4390
    },
    {
      "epoch": 0.6285714285714286,
      "grad_norm": 0.5672987103462219,
      "learning_rate": 1.581047619047619e-05,
      "loss": 1.5346,
      "step": 4400
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3494936227798462,
      "learning_rate": 1.5800952380952382e-05,
      "loss": 1.5564,
      "step": 4410
    },
    {
      "epoch": 0.6314285714285715,
      "grad_norm": 0.10821928828954697,
      "learning_rate": 1.5791428571428575e-05,
      "loss": 1.5642,
      "step": 4420
    },
    {
      "epoch": 0.6328571428571429,
      "grad_norm": 2737073.5,
      "learning_rate": 1.5781904761904764e-05,
      "loss": 3.1981,
      "step": 4430
    },
    {
      "epoch": 0.6342857142857142,
      "grad_norm": 0.09739159047603607,
      "learning_rate": 1.5772380952380956e-05,
      "loss": 4.7397,
      "step": 4440
    },
    {
      "epoch": 0.6357142857142857,
      "grad_norm": 0.34721335768699646,
      "learning_rate": 1.5762857142857145e-05,
      "loss": 1.6853,
      "step": 4450
    },
    {
      "epoch": 0.6371428571428571,
      "grad_norm": 0.06305290013551712,
      "learning_rate": 1.5753333333333334e-05,
      "loss": 1.6395,
      "step": 4460
    },
    {
      "epoch": 0.6385714285714286,
      "grad_norm": 0.06435007601976395,
      "learning_rate": 1.5743809523809526e-05,
      "loss": 3.3868,
      "step": 4470
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3210868537425995,
      "learning_rate": 1.5734285714285715e-05,
      "loss": 1.7447,
      "step": 4480
    },
    {
      "epoch": 0.6414285714285715,
      "grad_norm": 0.10086046904325485,
      "learning_rate": 1.5724761904761907e-05,
      "loss": 3.2896,
      "step": 4490
    },
    {
      "epoch": 0.6428571428571429,
      "grad_norm": 0.2700510025024414,
      "learning_rate": 1.5715238095238096e-05,
      "loss": 1.6653,
      "step": 4500
    },
    {
      "epoch": 0.6442857142857142,
      "grad_norm": 0.26571059226989746,
      "learning_rate": 1.5705714285714285e-05,
      "loss": 3.3421,
      "step": 4510
    },
    {
      "epoch": 0.6457142857142857,
      "grad_norm": 0.2759547531604767,
      "learning_rate": 1.5696190476190478e-05,
      "loss": 8.101,
      "step": 4520
    },
    {
      "epoch": 0.6471428571428571,
      "grad_norm": 3767085.75,
      "learning_rate": 1.5686666666666667e-05,
      "loss": 10.9414,
      "step": 4530
    },
    {
      "epoch": 0.6485714285714286,
      "grad_norm": 26.828210830688477,
      "learning_rate": 1.567714285714286e-05,
      "loss": 3.6594,
      "step": 4540
    },
    {
      "epoch": 0.65,
      "grad_norm": 2940143.0,
      "learning_rate": 1.5667619047619048e-05,
      "loss": 6.0135,
      "step": 4550
    },
    {
      "epoch": 0.6514285714285715,
      "grad_norm": 2778777.0,
      "learning_rate": 1.565809523809524e-05,
      "loss": 6.0806,
      "step": 4560
    },
    {
      "epoch": 0.6528571428571428,
      "grad_norm": 2719365.25,
      "learning_rate": 1.564857142857143e-05,
      "loss": 4.1958,
      "step": 4570
    },
    {
      "epoch": 0.6542857142857142,
      "grad_norm": 2667017.0,
      "learning_rate": 1.563904761904762e-05,
      "loss": 5.7083,
      "step": 4580
    },
    {
      "epoch": 0.6557142857142857,
      "grad_norm": 0.29587244987487793,
      "learning_rate": 1.562952380952381e-05,
      "loss": 4.7635,
      "step": 4590
    },
    {
      "epoch": 0.6571428571428571,
      "grad_norm": 0.31774067878723145,
      "learning_rate": 1.5620000000000003e-05,
      "loss": 1.6037,
      "step": 4600
    },
    {
      "epoch": 0.6585714285714286,
      "grad_norm": 0.2913525104522705,
      "learning_rate": 1.5610476190476192e-05,
      "loss": 0.0,
      "step": 4610
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35932010412216187,
      "learning_rate": 1.5600952380952384e-05,
      "loss": 6.4895,
      "step": 4620
    },
    {
      "epoch": 0.6614285714285715,
      "grad_norm": 0.37067627906799316,
      "learning_rate": 1.5591428571428573e-05,
      "loss": 6.4868,
      "step": 4630
    },
    {
      "epoch": 0.6628571428571428,
      "grad_norm": 0.26529204845428467,
      "learning_rate": 1.5581904761904762e-05,
      "loss": 4.7807,
      "step": 4640
    },
    {
      "epoch": 0.6642857142857143,
      "grad_norm": 0.3446073532104492,
      "learning_rate": 1.5572380952380954e-05,
      "loss": 4.7454,
      "step": 4650
    },
    {
      "epoch": 0.6657142857142857,
      "grad_norm": 2982641.5,
      "learning_rate": 1.5562857142857143e-05,
      "loss": 4.5512,
      "step": 4660
    },
    {
      "epoch": 0.6671428571428571,
      "grad_norm": 3166771.5,
      "learning_rate": 1.5553333333333332e-05,
      "loss": 4.56,
      "step": 4670
    },
    {
      "epoch": 0.6685714285714286,
      "grad_norm": 2850972.75,
      "learning_rate": 1.5543809523809525e-05,
      "loss": 1.4864,
      "step": 4680
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7112411856651306,
      "learning_rate": 1.5534285714285714e-05,
      "loss": 7.5089,
      "step": 4690
    },
    {
      "epoch": 0.6714285714285714,
      "grad_norm": 0.1557735651731491,
      "learning_rate": 1.5524761904761906e-05,
      "loss": 4.608,
      "step": 4700
    },
    {
      "epoch": 0.6728571428571428,
      "grad_norm": 0.28362518548965454,
      "learning_rate": 1.5515238095238095e-05,
      "loss": 0.0,
      "step": 4710
    },
    {
      "epoch": 0.6742857142857143,
      "grad_norm": 0.07087372988462448,
      "learning_rate": 1.5505714285714287e-05,
      "loss": 1.5179,
      "step": 4720
    },
    {
      "epoch": 0.6757142857142857,
      "grad_norm": 0.47066980600357056,
      "learning_rate": 1.549619047619048e-05,
      "loss": 0.0,
      "step": 4730
    },
    {
      "epoch": 0.6771428571428572,
      "grad_norm": 0.28070592880249023,
      "learning_rate": 1.548666666666667e-05,
      "loss": 1.6121,
      "step": 4740
    },
    {
      "epoch": 0.6785714285714286,
      "grad_norm": 0.07762420177459717,
      "learning_rate": 1.547714285714286e-05,
      "loss": 3.2431,
      "step": 4750
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.06042889878153801,
      "learning_rate": 1.546761904761905e-05,
      "loss": 1.7179,
      "step": 4760
    },
    {
      "epoch": 0.6814285714285714,
      "grad_norm": 0.05443624407052994,
      "learning_rate": 1.545809523809524e-05,
      "loss": 1.7934,
      "step": 4770
    },
    {
      "epoch": 0.6828571428571428,
      "grad_norm": 0.7387009263038635,
      "learning_rate": 1.544857142857143e-05,
      "loss": 1.6111,
      "step": 4780
    },
    {
      "epoch": 0.6842857142857143,
      "grad_norm": 0.3175390660762787,
      "learning_rate": 1.543904761904762e-05,
      "loss": 6.5942,
      "step": 4790
    },
    {
      "epoch": 0.6857142857142857,
      "grad_norm": 3083573.75,
      "learning_rate": 1.542952380952381e-05,
      "loss": 3.3563,
      "step": 4800
    },
    {
      "epoch": 0.6871428571428572,
      "grad_norm": 0.10209000110626221,
      "learning_rate": 1.542e-05,
      "loss": 1.6737,
      "step": 4810
    },
    {
      "epoch": 0.6885714285714286,
      "grad_norm": 0.2936196029186249,
      "learning_rate": 1.541047619047619e-05,
      "loss": 1.6377,
      "step": 4820
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.31702372431755066,
      "learning_rate": 1.5400952380952383e-05,
      "loss": 1.6359,
      "step": 4830
    },
    {
      "epoch": 0.6914285714285714,
      "grad_norm": 0.3880065977573395,
      "learning_rate": 1.539142857142857e-05,
      "loss": 6.5765,
      "step": 4840
    },
    {
      "epoch": 0.6928571428571428,
      "grad_norm": 0.3540211319923401,
      "learning_rate": 1.5381904761904764e-05,
      "loss": 6.3685,
      "step": 4850
    },
    {
      "epoch": 0.6942857142857143,
      "grad_norm": 0.2726327180862427,
      "learning_rate": 1.5372380952380953e-05,
      "loss": 3.1723,
      "step": 4860
    },
    {
      "epoch": 0.6957142857142857,
      "grad_norm": 0.30269956588745117,
      "learning_rate": 1.5362857142857145e-05,
      "loss": 4.5404,
      "step": 4870
    },
    {
      "epoch": 0.6971428571428572,
      "grad_norm": 0.40749019384384155,
      "learning_rate": 1.5353333333333334e-05,
      "loss": 2.7564,
      "step": 4880
    },
    {
      "epoch": 0.6985714285714286,
      "grad_norm": 0.3330423831939697,
      "learning_rate": 1.5343809523809527e-05,
      "loss": 4.8656,
      "step": 4890
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.04657755792140961,
      "learning_rate": 1.5334285714285716e-05,
      "loss": 0.0,
      "step": 4900
    },
    {
      "epoch": 0.7014285714285714,
      "grad_norm": 2813467.75,
      "learning_rate": 1.5324761904761908e-05,
      "loss": 1.7619,
      "step": 4910
    },
    {
      "epoch": 0.7028571428571428,
      "grad_norm": 0.31519854068756104,
      "learning_rate": 1.5315238095238097e-05,
      "loss": 6.525,
      "step": 4920
    },
    {
      "epoch": 0.7042857142857143,
      "grad_norm": 0.3445568382740021,
      "learning_rate": 1.5305714285714286e-05,
      "loss": 4.6138,
      "step": 4930
    },
    {
      "epoch": 0.7057142857142857,
      "grad_norm": 2657612.25,
      "learning_rate": 1.5296190476190478e-05,
      "loss": 7.9655,
      "step": 4940
    },
    {
      "epoch": 0.7071428571428572,
      "grad_norm": 0.3534267544746399,
      "learning_rate": 1.5286666666666667e-05,
      "loss": 3.1716,
      "step": 4950
    },
    {
      "epoch": 0.7085714285714285,
      "grad_norm": 0.40478354692459106,
      "learning_rate": 1.527714285714286e-05,
      "loss": 3.2771,
      "step": 4960
    },
    {
      "epoch": 0.71,
      "grad_norm": 2983096.25,
      "learning_rate": 1.526761904761905e-05,
      "loss": 8.0337,
      "step": 4970
    },
    {
      "epoch": 0.7114285714285714,
      "grad_norm": 3016300.0,
      "learning_rate": 1.5258095238095239e-05,
      "loss": 4.9061,
      "step": 4980
    },
    {
      "epoch": 0.7128571428571429,
      "grad_norm": 0.30139225721359253,
      "learning_rate": 1.524857142857143e-05,
      "loss": 3.1314,
      "step": 4990
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 252.80076599121094,
      "learning_rate": 1.523904761904762e-05,
      "loss": 9.5534,
      "step": 5000
    },
    {
      "epoch": 0.7157142857142857,
      "grad_norm": 2851133.5,
      "learning_rate": 1.5229523809523811e-05,
      "loss": 2.7331,
      "step": 5010
    },
    {
      "epoch": 0.7171428571428572,
      "grad_norm": 2792782.5,
      "learning_rate": 1.5220000000000002e-05,
      "loss": 6.0593,
      "step": 5020
    },
    {
      "epoch": 0.7185714285714285,
      "grad_norm": 2816984.75,
      "learning_rate": 1.521047619047619e-05,
      "loss": 4.7333,
      "step": 5030
    },
    {
      "epoch": 0.72,
      "grad_norm": 5705384.0,
      "learning_rate": 1.5200952380952383e-05,
      "loss": 4.6904,
      "step": 5040
    },
    {
      "epoch": 0.7214285714285714,
      "grad_norm": 253380976.0,
      "learning_rate": 1.5191428571428572e-05,
      "loss": 2.4527,
      "step": 5050
    },
    {
      "epoch": 0.7228571428571429,
      "grad_norm": 0.34645339846611023,
      "learning_rate": 1.5181904761904763e-05,
      "loss": 1.5568,
      "step": 5060
    },
    {
      "epoch": 0.7242857142857143,
      "grad_norm": 6945812.5,
      "learning_rate": 1.5172380952380953e-05,
      "loss": 3.2194,
      "step": 5070
    },
    {
      "epoch": 0.7257142857142858,
      "grad_norm": 1.0328357219696045,
      "learning_rate": 1.5162857142857144e-05,
      "loss": 1.5854,
      "step": 5080
    },
    {
      "epoch": 0.7271428571428571,
      "grad_norm": 0.11636336892843246,
      "learning_rate": 1.5153333333333336e-05,
      "loss": 2.5473,
      "step": 5090
    },
    {
      "epoch": 0.7285714285714285,
      "grad_norm": 0.26487699151039124,
      "learning_rate": 1.5143809523809525e-05,
      "loss": 1.5927,
      "step": 5100
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.0850910171866417,
      "learning_rate": 1.5134285714285714e-05,
      "loss": 1.6618,
      "step": 5110
    },
    {
      "epoch": 0.7314285714285714,
      "grad_norm": 2701409.75,
      "learning_rate": 1.5124761904761906e-05,
      "loss": 3.3317,
      "step": 5120
    },
    {
      "epoch": 0.7328571428571429,
      "grad_norm": 0.09510695189237595,
      "learning_rate": 1.5115238095238095e-05,
      "loss": 3.3969,
      "step": 5130
    },
    {
      "epoch": 0.7342857142857143,
      "grad_norm": 0.28927356004714966,
      "learning_rate": 1.5105714285714288e-05,
      "loss": 1.673,
      "step": 5140
    },
    {
      "epoch": 0.7357142857142858,
      "grad_norm": 4078158.25,
      "learning_rate": 1.5096190476190478e-05,
      "loss": 8.1363,
      "step": 5150
    },
    {
      "epoch": 0.7371428571428571,
      "grad_norm": 1.030771255493164,
      "learning_rate": 1.5086666666666667e-05,
      "loss": 3.2727,
      "step": 5160
    },
    {
      "epoch": 0.7385714285714285,
      "grad_norm": 2792047.0,
      "learning_rate": 1.507714285714286e-05,
      "loss": 1.5996,
      "step": 5170
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2912902235984802,
      "learning_rate": 1.5067619047619049e-05,
      "loss": 4.9679,
      "step": 5180
    },
    {
      "epoch": 0.7414285714285714,
      "grad_norm": 0.0894903838634491,
      "learning_rate": 1.5058095238095238e-05,
      "loss": 4.4445,
      "step": 5190
    },
    {
      "epoch": 0.7428571428571429,
      "grad_norm": 0.26367828249931335,
      "learning_rate": 1.504857142857143e-05,
      "loss": 6.2724,
      "step": 5200
    },
    {
      "epoch": 0.7442857142857143,
      "grad_norm": 0.2959534227848053,
      "learning_rate": 1.503904761904762e-05,
      "loss": 4.777,
      "step": 5210
    },
    {
      "epoch": 0.7457142857142857,
      "grad_norm": 0.08696381747722626,
      "learning_rate": 1.5029523809523811e-05,
      "loss": 3.1746,
      "step": 5220
    },
    {
      "epoch": 0.7471428571428571,
      "grad_norm": 0.06944651901721954,
      "learning_rate": 1.5020000000000002e-05,
      "loss": 1.5825,
      "step": 5230
    },
    {
      "epoch": 0.7485714285714286,
      "grad_norm": 0.375199556350708,
      "learning_rate": 1.501047619047619e-05,
      "loss": 1.6651,
      "step": 5240
    },
    {
      "epoch": 0.75,
      "grad_norm": 2686806.75,
      "learning_rate": 1.5000952380952383e-05,
      "loss": 4.9131,
      "step": 5250
    },
    {
      "epoch": 0.7514285714285714,
      "grad_norm": 0.11285000294446945,
      "learning_rate": 1.4991428571428572e-05,
      "loss": 1.5947,
      "step": 5260
    },
    {
      "epoch": 0.7528571428571429,
      "grad_norm": 0.27778133749961853,
      "learning_rate": 1.4981904761904764e-05,
      "loss": 2.2366,
      "step": 5270
    },
    {
      "epoch": 0.7542857142857143,
      "grad_norm": 4737705.5,
      "learning_rate": 1.4972380952380953e-05,
      "loss": 2.8466,
      "step": 5280
    },
    {
      "epoch": 0.7557142857142857,
      "grad_norm": 39.194828033447266,
      "learning_rate": 1.4962857142857144e-05,
      "loss": 2.2701,
      "step": 5290
    },
    {
      "epoch": 0.7571428571428571,
      "grad_norm": 0.10251382738351822,
      "learning_rate": 1.4953333333333335e-05,
      "loss": 0.0,
      "step": 5300
    },
    {
      "epoch": 0.7585714285714286,
      "grad_norm": 0.30918964743614197,
      "learning_rate": 1.4943809523809525e-05,
      "loss": 3.2972,
      "step": 5310
    },
    {
      "epoch": 0.76,
      "grad_norm": 4088495.75,
      "learning_rate": 1.4934285714285714e-05,
      "loss": 4.8362,
      "step": 5320
    },
    {
      "epoch": 0.7614285714285715,
      "grad_norm": 0.29832854866981506,
      "learning_rate": 1.4924761904761907e-05,
      "loss": 4.8289,
      "step": 5330
    },
    {
      "epoch": 0.7628571428571429,
      "grad_norm": 2729317.5,
      "learning_rate": 1.4915238095238096e-05,
      "loss": 2.4568,
      "step": 5340
    },
    {
      "epoch": 0.7642857142857142,
      "grad_norm": 0.39662009477615356,
      "learning_rate": 1.4905714285714288e-05,
      "loss": 4.8555,
      "step": 5350
    },
    {
      "epoch": 0.7657142857142857,
      "grad_norm": 2635886.75,
      "learning_rate": 1.4896190476190477e-05,
      "loss": 3.1437,
      "step": 5360
    },
    {
      "epoch": 0.7671428571428571,
      "grad_norm": 2707647.5,
      "learning_rate": 1.4886666666666668e-05,
      "loss": 6.4845,
      "step": 5370
    },
    {
      "epoch": 0.7685714285714286,
      "grad_norm": 0.3359176516532898,
      "learning_rate": 1.4877142857142858e-05,
      "loss": 1.5753,
      "step": 5380
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.29091107845306396,
      "learning_rate": 1.4867619047619049e-05,
      "loss": 3.1229,
      "step": 5390
    },
    {
      "epoch": 0.7714285714285715,
      "grad_norm": 0.32749322056770325,
      "learning_rate": 1.485809523809524e-05,
      "loss": 3.0864,
      "step": 5400
    },
    {
      "epoch": 0.7728571428571429,
      "grad_norm": 2699685.75,
      "learning_rate": 1.484857142857143e-05,
      "loss": 4.7574,
      "step": 5410
    },
    {
      "epoch": 0.7742857142857142,
      "grad_norm": 0.07796455174684525,
      "learning_rate": 1.4839047619047619e-05,
      "loss": 1.6946,
      "step": 5420
    },
    {
      "epoch": 0.7757142857142857,
      "grad_norm": 2724882.0,
      "learning_rate": 1.4829523809523811e-05,
      "loss": 3.2277,
      "step": 5430
    },
    {
      "epoch": 0.7771428571428571,
      "grad_norm": 0.33651071786880493,
      "learning_rate": 1.482e-05,
      "loss": 1.5901,
      "step": 5440
    },
    {
      "epoch": 0.7785714285714286,
      "grad_norm": 0.0654456615447998,
      "learning_rate": 1.4810476190476191e-05,
      "loss": 1.5858,
      "step": 5450
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3859398365020752,
      "learning_rate": 1.4800952380952382e-05,
      "loss": 1.5631,
      "step": 5460
    },
    {
      "epoch": 0.7814285714285715,
      "grad_norm": 0.06587772816419601,
      "learning_rate": 1.4791428571428572e-05,
      "loss": 1.5905,
      "step": 5470
    },
    {
      "epoch": 0.7828571428571428,
      "grad_norm": 0.3137292265892029,
      "learning_rate": 1.4781904761904765e-05,
      "loss": 4.8618,
      "step": 5480
    },
    {
      "epoch": 0.7842857142857143,
      "grad_norm": 0.06515590101480484,
      "learning_rate": 1.4772380952380954e-05,
      "loss": 0.0,
      "step": 5490
    },
    {
      "epoch": 0.7857142857142857,
      "grad_norm": 0.05615342780947685,
      "learning_rate": 1.4762857142857143e-05,
      "loss": 3.1921,
      "step": 5500
    },
    {
      "epoch": 0.7871428571428571,
      "grad_norm": 0.15226885676383972,
      "learning_rate": 1.4753333333333335e-05,
      "loss": 1.7092,
      "step": 5510
    },
    {
      "epoch": 0.7885714285714286,
      "grad_norm": 0.06395590305328369,
      "learning_rate": 1.4743809523809524e-05,
      "loss": 0.0,
      "step": 5520
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.27748334407806396,
      "learning_rate": 1.4734285714285716e-05,
      "loss": 3.4013,
      "step": 5530
    },
    {
      "epoch": 0.7914285714285715,
      "grad_norm": 10113510.0,
      "learning_rate": 1.4724761904761907e-05,
      "loss": 3.3054,
      "step": 5540
    },
    {
      "epoch": 0.7928571428571428,
      "grad_norm": 0.2876303791999817,
      "learning_rate": 1.4715238095238096e-05,
      "loss": 3.2482,
      "step": 5550
    },
    {
      "epoch": 0.7942857142857143,
      "grad_norm": 0.31878435611724854,
      "learning_rate": 1.4705714285714288e-05,
      "loss": 4.819,
      "step": 5560
    },
    {
      "epoch": 0.7957142857142857,
      "grad_norm": 2.456526279449463,
      "learning_rate": 1.4696190476190477e-05,
      "loss": 4.7089,
      "step": 5570
    },
    {
      "epoch": 0.7971428571428572,
      "grad_norm": 0.09629660844802856,
      "learning_rate": 1.4686666666666666e-05,
      "loss": 3.2786,
      "step": 5580
    },
    {
      "epoch": 0.7985714285714286,
      "grad_norm": 2734056.5,
      "learning_rate": 1.4677142857142858e-05,
      "loss": 1.552,
      "step": 5590
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6233723163604736,
      "learning_rate": 1.4667619047619049e-05,
      "loss": 3.257,
      "step": 5600
    },
    {
      "epoch": 0.8014285714285714,
      "grad_norm": 0.6156706213951111,
      "learning_rate": 1.465809523809524e-05,
      "loss": 1.5507,
      "step": 5610
    },
    {
      "epoch": 0.8028571428571428,
      "grad_norm": 0.5814512968063354,
      "learning_rate": 1.464857142857143e-05,
      "loss": 1.7091,
      "step": 5620
    },
    {
      "epoch": 0.8042857142857143,
      "grad_norm": 0.28188368678092957,
      "learning_rate": 1.463904761904762e-05,
      "loss": 3.2031,
      "step": 5630
    },
    {
      "epoch": 0.8057142857142857,
      "grad_norm": 0.2728634774684906,
      "learning_rate": 1.4629523809523812e-05,
      "loss": 3.3682,
      "step": 5640
    },
    {
      "epoch": 0.8071428571428572,
      "grad_norm": 2754549.25,
      "learning_rate": 1.462e-05,
      "loss": 3.1798,
      "step": 5650
    },
    {
      "epoch": 0.8085714285714286,
      "grad_norm": 0.31991660594940186,
      "learning_rate": 1.4610476190476193e-05,
      "loss": 3.2029,
      "step": 5660
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.0930977538228035,
      "learning_rate": 1.4600952380952382e-05,
      "loss": 0.0,
      "step": 5670
    },
    {
      "epoch": 0.8114285714285714,
      "grad_norm": 6.9387898445129395,
      "learning_rate": 1.4591428571428573e-05,
      "loss": 0.0,
      "step": 5680
    },
    {
      "epoch": 0.8128571428571428,
      "grad_norm": 0.6195544004440308,
      "learning_rate": 1.4581904761904763e-05,
      "loss": 5.0807,
      "step": 5690
    },
    {
      "epoch": 0.8142857142857143,
      "grad_norm": 0.3995417654514313,
      "learning_rate": 1.4572380952380954e-05,
      "loss": 3.26,
      "step": 5700
    },
    {
      "epoch": 0.8157142857142857,
      "grad_norm": 0.3031126856803894,
      "learning_rate": 1.4562857142857143e-05,
      "loss": 1.63,
      "step": 5710
    },
    {
      "epoch": 0.8171428571428572,
      "grad_norm": 2756448.25,
      "learning_rate": 1.4553333333333335e-05,
      "loss": 3.3938,
      "step": 5720
    },
    {
      "epoch": 0.8185714285714286,
      "grad_norm": 0.14320529997348785,
      "learning_rate": 1.4543809523809524e-05,
      "loss": 3.2711,
      "step": 5730
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2604595422744751,
      "learning_rate": 1.4534285714285716e-05,
      "loss": 6.4388,
      "step": 5740
    },
    {
      "epoch": 0.8214285714285714,
      "grad_norm": 3532353.75,
      "learning_rate": 1.4524761904761905e-05,
      "loss": 3.1896,
      "step": 5750
    },
    {
      "epoch": 0.8228571428571428,
      "grad_norm": 2707096.5,
      "learning_rate": 1.4515238095238096e-05,
      "loss": 7.9528,
      "step": 5760
    },
    {
      "epoch": 0.8242857142857143,
      "grad_norm": 0.30123114585876465,
      "learning_rate": 1.4505714285714287e-05,
      "loss": 1.6161,
      "step": 5770
    },
    {
      "epoch": 0.8257142857142857,
      "grad_norm": 2720902.75,
      "learning_rate": 1.4496190476190477e-05,
      "loss": 4.7777,
      "step": 5780
    },
    {
      "epoch": 0.8271428571428572,
      "grad_norm": 1.1321107149124146,
      "learning_rate": 1.448666666666667e-05,
      "loss": 0.0,
      "step": 5790
    },
    {
      "epoch": 0.8285714285714286,
      "grad_norm": 0.27708855271339417,
      "learning_rate": 1.4477142857142859e-05,
      "loss": 1.6006,
      "step": 5800
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5602757930755615,
      "learning_rate": 1.4467619047619048e-05,
      "loss": 1.6543,
      "step": 5810
    },
    {
      "epoch": 0.8314285714285714,
      "grad_norm": 0.11507312208414078,
      "learning_rate": 1.445809523809524e-05,
      "loss": 1.3813,
      "step": 5820
    },
    {
      "epoch": 0.8328571428571429,
      "grad_norm": 0.269243985414505,
      "learning_rate": 1.4448571428571429e-05,
      "loss": 0.0,
      "step": 5830
    },
    {
      "epoch": 0.8342857142857143,
      "grad_norm": 0.29347893595695496,
      "learning_rate": 1.443904761904762e-05,
      "loss": 1.748,
      "step": 5840
    },
    {
      "epoch": 0.8357142857142857,
      "grad_norm": 0.1597544401884079,
      "learning_rate": 1.442952380952381e-05,
      "loss": 0.0,
      "step": 5850
    },
    {
      "epoch": 0.8371428571428572,
      "grad_norm": 3236185.25,
      "learning_rate": 1.4420000000000001e-05,
      "loss": 3.3906,
      "step": 5860
    },
    {
      "epoch": 0.8385714285714285,
      "grad_norm": 0.09639938175678253,
      "learning_rate": 1.4410476190476193e-05,
      "loss": 3.4192,
      "step": 5870
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.0977713018655777,
      "learning_rate": 1.4400952380952382e-05,
      "loss": 3.3191,
      "step": 5880
    },
    {
      "epoch": 0.8414285714285714,
      "grad_norm": 0.2681606709957123,
      "learning_rate": 1.4391428571428571e-05,
      "loss": 1.6438,
      "step": 5890
    },
    {
      "epoch": 0.8428571428571429,
      "grad_norm": 0.26125088334083557,
      "learning_rate": 1.4381904761904763e-05,
      "loss": 1.6295,
      "step": 5900
    },
    {
      "epoch": 0.8442857142857143,
      "grad_norm": 0.32890695333480835,
      "learning_rate": 1.4372380952380952e-05,
      "loss": 6.4595,
      "step": 5910
    },
    {
      "epoch": 0.8457142857142858,
      "grad_norm": 0.4383538067340851,
      "learning_rate": 1.4362857142857145e-05,
      "loss": 1.6019,
      "step": 5920
    },
    {
      "epoch": 0.8471428571428572,
      "grad_norm": 0.2845783829689026,
      "learning_rate": 1.4353333333333335e-05,
      "loss": 3.3259,
      "step": 5930
    },
    {
      "epoch": 0.8485714285714285,
      "grad_norm": 0.26797017455101013,
      "learning_rate": 1.4343809523809524e-05,
      "loss": 3.0579,
      "step": 5940
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.28775840997695923,
      "learning_rate": 1.4334285714285717e-05,
      "loss": 3.0357,
      "step": 5950
    },
    {
      "epoch": 0.8514285714285714,
      "grad_norm": 0.08319329470396042,
      "learning_rate": 1.4324761904761906e-05,
      "loss": 6.4019,
      "step": 5960
    },
    {
      "epoch": 0.8528571428571429,
      "grad_norm": 0.25417858362197876,
      "learning_rate": 1.4315238095238095e-05,
      "loss": 2.8936,
      "step": 5970
    },
    {
      "epoch": 0.8542857142857143,
      "grad_norm": 0.11786254495382309,
      "learning_rate": 1.4305714285714287e-05,
      "loss": 3.4023,
      "step": 5980
    },
    {
      "epoch": 0.8557142857142858,
      "grad_norm": 0.07068348675966263,
      "learning_rate": 1.4296190476190478e-05,
      "loss": 2.9997,
      "step": 5990
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.08457542955875397,
      "learning_rate": 1.4286666666666668e-05,
      "loss": 3.379,
      "step": 6000
    },
    {
      "epoch": 0.8585714285714285,
      "grad_norm": 0.088286854326725,
      "learning_rate": 1.4277142857142859e-05,
      "loss": 1.5187,
      "step": 6010
    },
    {
      "epoch": 0.86,
      "grad_norm": 2672551.25,
      "learning_rate": 1.4267619047619048e-05,
      "loss": 1.6514,
      "step": 6020
    },
    {
      "epoch": 0.8614285714285714,
      "grad_norm": 0.27820804715156555,
      "learning_rate": 1.425809523809524e-05,
      "loss": 1.6875,
      "step": 6030
    },
    {
      "epoch": 0.8628571428571429,
      "grad_norm": 551591.6875,
      "learning_rate": 1.4248571428571429e-05,
      "loss": 4.8571,
      "step": 6040
    },
    {
      "epoch": 0.8642857142857143,
      "grad_norm": 2726342.0,
      "learning_rate": 1.4239047619047621e-05,
      "loss": 4.8465,
      "step": 6050
    },
    {
      "epoch": 0.8657142857142858,
      "grad_norm": 0.09061988443136215,
      "learning_rate": 1.422952380952381e-05,
      "loss": 2.8706,
      "step": 6060
    },
    {
      "epoch": 0.8671428571428571,
      "grad_norm": 9246068.0,
      "learning_rate": 1.4220000000000001e-05,
      "loss": 6.1455,
      "step": 6070
    },
    {
      "epoch": 0.8685714285714285,
      "grad_norm": 0.10027223825454712,
      "learning_rate": 1.4210476190476192e-05,
      "loss": 6.2273,
      "step": 6080
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.33077114820480347,
      "learning_rate": 1.4200952380952382e-05,
      "loss": 3.0691,
      "step": 6090
    },
    {
      "epoch": 0.8714285714285714,
      "grad_norm": 2868339.0,
      "learning_rate": 1.4191428571428571e-05,
      "loss": 4.8066,
      "step": 6100
    },
    {
      "epoch": 0.8728571428571429,
      "grad_norm": 0.09321148693561554,
      "learning_rate": 1.4181904761904764e-05,
      "loss": 0.0,
      "step": 6110
    },
    {
      "epoch": 0.8742857142857143,
      "grad_norm": 0.11052189767360687,
      "learning_rate": 1.4172380952380953e-05,
      "loss": 1.5083,
      "step": 6120
    },
    {
      "epoch": 0.8757142857142857,
      "grad_norm": 0.08915243297815323,
      "learning_rate": 1.4162857142857145e-05,
      "loss": 1.6506,
      "step": 6130
    },
    {
      "epoch": 0.8771428571428571,
      "grad_norm": 0.3687570095062256,
      "learning_rate": 1.4153333333333334e-05,
      "loss": 2.8012,
      "step": 6140
    },
    {
      "epoch": 0.8785714285714286,
      "grad_norm": 0.0709046721458435,
      "learning_rate": 1.4143809523809525e-05,
      "loss": 1.3274,
      "step": 6150
    },
    {
      "epoch": 0.88,
      "grad_norm": 3269600.75,
      "learning_rate": 1.4134285714285715e-05,
      "loss": 1.7143,
      "step": 6160
    },
    {
      "epoch": 0.8814285714285715,
      "grad_norm": 374.8963317871094,
      "learning_rate": 1.4124761904761906e-05,
      "loss": 1.6709,
      "step": 6170
    },
    {
      "epoch": 0.8828571428571429,
      "grad_norm": 0.09369340538978577,
      "learning_rate": 1.4115238095238098e-05,
      "loss": 0.0,
      "step": 6180
    },
    {
      "epoch": 0.8842857142857142,
      "grad_norm": 0.10248035192489624,
      "learning_rate": 1.4105714285714287e-05,
      "loss": 3.9159,
      "step": 6190
    },
    {
      "epoch": 0.8857142857142857,
      "grad_norm": 0.279605895280838,
      "learning_rate": 1.4096190476190476e-05,
      "loss": 6.7725,
      "step": 6200
    },
    {
      "epoch": 0.8871428571428571,
      "grad_norm": 0.26080915331840515,
      "learning_rate": 1.4086666666666668e-05,
      "loss": 2.7467,
      "step": 6210
    },
    {
      "epoch": 0.8885714285714286,
      "grad_norm": 2703822.75,
      "learning_rate": 1.4077142857142857e-05,
      "loss": 5.0613,
      "step": 6220
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1007886677980423,
      "learning_rate": 1.4067619047619048e-05,
      "loss": 4.7497,
      "step": 6230
    },
    {
      "epoch": 0.8914285714285715,
      "grad_norm": 0.26448556780815125,
      "learning_rate": 1.405809523809524e-05,
      "loss": 4.9351,
      "step": 6240
    },
    {
      "epoch": 0.8928571428571429,
      "grad_norm": 0.06660294532775879,
      "learning_rate": 1.404857142857143e-05,
      "loss": 1.5638,
      "step": 6250
    },
    {
      "epoch": 0.8942857142857142,
      "grad_norm": 3115689.5,
      "learning_rate": 1.4039047619047622e-05,
      "loss": 4.4773,
      "step": 6260
    },
    {
      "epoch": 0.8957142857142857,
      "grad_norm": 0.28107452392578125,
      "learning_rate": 1.402952380952381e-05,
      "loss": 4.9004,
      "step": 6270
    },
    {
      "epoch": 0.8971428571428571,
      "grad_norm": 0.2956908047199249,
      "learning_rate": 1.402e-05,
      "loss": 1.6683,
      "step": 6280
    },
    {
      "epoch": 0.8985714285714286,
      "grad_norm": 0.2785201966762543,
      "learning_rate": 1.4010476190476192e-05,
      "loss": 1.6772,
      "step": 6290
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.07782118022441864,
      "learning_rate": 1.4000952380952381e-05,
      "loss": 3.2304,
      "step": 6300
    },
    {
      "epoch": 0.9014285714285715,
      "grad_norm": 0.10485462099313736,
      "learning_rate": 1.3991428571428573e-05,
      "loss": 1.2258,
      "step": 6310
    },
    {
      "epoch": 0.9028571428571428,
      "grad_norm": 0.6598716974258423,
      "learning_rate": 1.3981904761904764e-05,
      "loss": 4.8716,
      "step": 6320
    },
    {
      "epoch": 0.9042857142857142,
      "grad_norm": 0.08638031035661697,
      "learning_rate": 1.3972380952380953e-05,
      "loss": 0.0,
      "step": 6330
    },
    {
      "epoch": 0.9057142857142857,
      "grad_norm": 0.2575370967388153,
      "learning_rate": 1.3962857142857145e-05,
      "loss": 1.6317,
      "step": 6340
    },
    {
      "epoch": 0.9071428571428571,
      "grad_norm": 0.06752851605415344,
      "learning_rate": 1.3953333333333334e-05,
      "loss": 5.0037,
      "step": 6350
    },
    {
      "epoch": 0.9085714285714286,
      "grad_norm": 2999873.0,
      "learning_rate": 1.3943809523809523e-05,
      "loss": 6.6131,
      "step": 6360
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2685863673686981,
      "learning_rate": 1.3934285714285715e-05,
      "loss": 1.3691,
      "step": 6370
    },
    {
      "epoch": 0.9114285714285715,
      "grad_norm": 3009460.0,
      "learning_rate": 1.3924761904761906e-05,
      "loss": 3.2383,
      "step": 6380
    },
    {
      "epoch": 0.9128571428571428,
      "grad_norm": 0.2823067903518677,
      "learning_rate": 1.3915238095238097e-05,
      "loss": 0.0,
      "step": 6390
    },
    {
      "epoch": 0.9142857142857143,
      "grad_norm": 2987691.0,
      "learning_rate": 1.3905714285714287e-05,
      "loss": 3.286,
      "step": 6400
    },
    {
      "epoch": 0.9157142857142857,
      "grad_norm": 0.1049148440361023,
      "learning_rate": 1.3896190476190476e-05,
      "loss": 1.6467,
      "step": 6410
    },
    {
      "epoch": 0.9171428571428571,
      "grad_norm": 0.26472726464271545,
      "learning_rate": 1.3886666666666669e-05,
      "loss": 3.1949,
      "step": 6420
    },
    {
      "epoch": 0.9185714285714286,
      "grad_norm": 0.09823822230100632,
      "learning_rate": 1.3877142857142858e-05,
      "loss": 0.0,
      "step": 6430
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2547095715999603,
      "learning_rate": 1.386761904761905e-05,
      "loss": 4.3471,
      "step": 6440
    },
    {
      "epoch": 0.9214285714285714,
      "grad_norm": 0.08986924588680267,
      "learning_rate": 1.3858095238095239e-05,
      "loss": 2.5032,
      "step": 6450
    },
    {
      "epoch": 0.9228571428571428,
      "grad_norm": 0.09011940658092499,
      "learning_rate": 1.384857142857143e-05,
      "loss": 1.3814,
      "step": 6460
    },
    {
      "epoch": 0.9242857142857143,
      "grad_norm": 0.05266342684626579,
      "learning_rate": 1.383904761904762e-05,
      "loss": 1.7143,
      "step": 6470
    },
    {
      "epoch": 0.9257142857142857,
      "grad_norm": 0.06448287516832352,
      "learning_rate": 1.3829523809523811e-05,
      "loss": 0.5855,
      "step": 6480
    },
    {
      "epoch": 0.9271428571428572,
      "grad_norm": 0.06454791873693466,
      "learning_rate": 1.382e-05,
      "loss": 1.6781,
      "step": 6490
    },
    {
      "epoch": 0.9285714285714286,
      "grad_norm": 0.08691821247339249,
      "learning_rate": 1.3810476190476192e-05,
      "loss": 1.3742,
      "step": 6500
    },
    {
      "epoch": 0.93,
      "grad_norm": 652300.4375,
      "learning_rate": 1.3800952380952381e-05,
      "loss": 3.4434,
      "step": 6510
    },
    {
      "epoch": 0.9314285714285714,
      "grad_norm": 2695130.75,
      "learning_rate": 1.3791428571428573e-05,
      "loss": 3.415,
      "step": 6520
    },
    {
      "epoch": 0.9328571428571428,
      "grad_norm": 0.0784001275897026,
      "learning_rate": 1.3781904761904762e-05,
      "loss": 3.3946,
      "step": 6530
    },
    {
      "epoch": 0.9342857142857143,
      "grad_norm": 0.09227895736694336,
      "learning_rate": 1.3772380952380953e-05,
      "loss": 0.0,
      "step": 6540
    },
    {
      "epoch": 0.9357142857142857,
      "grad_norm": 107089576.0,
      "learning_rate": 1.3762857142857144e-05,
      "loss": 2.849,
      "step": 6550
    },
    {
      "epoch": 0.9371428571428572,
      "grad_norm": 0.28994736075401306,
      "learning_rate": 1.3753333333333334e-05,
      "loss": 1.659,
      "step": 6560
    },
    {
      "epoch": 0.9385714285714286,
      "grad_norm": 0.09471923857927322,
      "learning_rate": 1.3743809523809527e-05,
      "loss": 0.0,
      "step": 6570
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.07917433232069016,
      "learning_rate": 1.3734285714285716e-05,
      "loss": 5.0736,
      "step": 6580
    },
    {
      "epoch": 0.9414285714285714,
      "grad_norm": 0.26266103982925415,
      "learning_rate": 1.3724761904761905e-05,
      "loss": 0.879,
      "step": 6590
    },
    {
      "epoch": 0.9428571428571428,
      "grad_norm": 0.09929802268743515,
      "learning_rate": 1.3715238095238097e-05,
      "loss": 3.3946,
      "step": 6600
    },
    {
      "epoch": 0.9442857142857143,
      "grad_norm": 0.06964003294706345,
      "learning_rate": 1.3705714285714286e-05,
      "loss": 3.3028,
      "step": 6610
    },
    {
      "epoch": 0.9457142857142857,
      "grad_norm": 0.07238195091485977,
      "learning_rate": 1.3696190476190477e-05,
      "loss": 1.7332,
      "step": 6620
    },
    {
      "epoch": 0.9471428571428572,
      "grad_norm": 0.054564666002988815,
      "learning_rate": 1.3686666666666669e-05,
      "loss": 5.0476,
      "step": 6630
    },
    {
      "epoch": 0.9485714285714286,
      "grad_norm": 0.08676331490278244,
      "learning_rate": 1.3677142857142858e-05,
      "loss": 1.6078,
      "step": 6640
    },
    {
      "epoch": 0.95,
      "grad_norm": 2706806.25,
      "learning_rate": 1.366761904761905e-05,
      "loss": 3.3588,
      "step": 6650
    },
    {
      "epoch": 0.9514285714285714,
      "grad_norm": 2787061.75,
      "learning_rate": 1.3658095238095239e-05,
      "loss": 4.7973,
      "step": 6660
    },
    {
      "epoch": 0.9528571428571428,
      "grad_norm": 0.11435998231172562,
      "learning_rate": 1.3648571428571428e-05,
      "loss": 1.9771,
      "step": 6670
    },
    {
      "epoch": 0.9542857142857143,
      "grad_norm": 0.29006391763687134,
      "learning_rate": 1.363904761904762e-05,
      "loss": 0.0,
      "step": 6680
    },
    {
      "epoch": 0.9557142857142857,
      "grad_norm": 0.25920921564102173,
      "learning_rate": 1.3629523809523811e-05,
      "loss": 1.7126,
      "step": 6690
    },
    {
      "epoch": 0.9571428571428572,
      "grad_norm": 0.14795933663845062,
      "learning_rate": 1.3620000000000002e-05,
      "loss": 1.6559,
      "step": 6700
    },
    {
      "epoch": 0.9585714285714285,
      "grad_norm": 0.2678607404232025,
      "learning_rate": 1.3610476190476192e-05,
      "loss": 1.6604,
      "step": 6710
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.09658209979534149,
      "learning_rate": 1.3600952380952381e-05,
      "loss": 1.6584,
      "step": 6720
    },
    {
      "epoch": 0.9614285714285714,
      "grad_norm": 0.09817372262477875,
      "learning_rate": 1.3591428571428574e-05,
      "loss": 1.7098,
      "step": 6730
    },
    {
      "epoch": 0.9628571428571429,
      "grad_norm": 0.11151159554719925,
      "learning_rate": 1.3581904761904763e-05,
      "loss": 3.3146,
      "step": 6740
    },
    {
      "epoch": 0.9642857142857143,
      "grad_norm": 0.1236787810921669,
      "learning_rate": 1.3572380952380953e-05,
      "loss": 5.0312,
      "step": 6750
    },
    {
      "epoch": 0.9657142857142857,
      "grad_norm": 0.2585096061229706,
      "learning_rate": 1.3562857142857144e-05,
      "loss": 1.6187,
      "step": 6760
    },
    {
      "epoch": 0.9671428571428572,
      "grad_norm": 0.07166355103254318,
      "learning_rate": 1.3553333333333335e-05,
      "loss": 3.3816,
      "step": 6770
    },
    {
      "epoch": 0.9685714285714285,
      "grad_norm": 0.2716030776500702,
      "learning_rate": 1.3543809523809525e-05,
      "loss": 0.0,
      "step": 6780
    },
    {
      "epoch": 0.97,
      "grad_norm": 2931336.5,
      "learning_rate": 1.3534285714285716e-05,
      "loss": 2.7326,
      "step": 6790
    },
    {
      "epoch": 0.9714285714285714,
      "grad_norm": 0.08315255492925644,
      "learning_rate": 1.3524761904761905e-05,
      "loss": 4.9144,
      "step": 6800
    },
    {
      "epoch": 0.9728571428571429,
      "grad_norm": 0.33772969245910645,
      "learning_rate": 1.3515238095238097e-05,
      "loss": 1.6123,
      "step": 6810
    },
    {
      "epoch": 0.9742857142857143,
      "grad_norm": 0.10215848684310913,
      "learning_rate": 1.3505714285714286e-05,
      "loss": 0.4177,
      "step": 6820
    },
    {
      "epoch": 0.9757142857142858,
      "grad_norm": 0.3583911955356598,
      "learning_rate": 1.3496190476190478e-05,
      "loss": 3.3612,
      "step": 6830
    },
    {
      "epoch": 0.9771428571428571,
      "grad_norm": 2759828.5,
      "learning_rate": 1.3486666666666667e-05,
      "loss": 8.3191,
      "step": 6840
    },
    {
      "epoch": 0.9785714285714285,
      "grad_norm": 0.307038277387619,
      "learning_rate": 1.3477142857142858e-05,
      "loss": 4.7668,
      "step": 6850
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.28703778982162476,
      "learning_rate": 1.3467619047619049e-05,
      "loss": 1.6188,
      "step": 6860
    },
    {
      "epoch": 0.9814285714285714,
      "grad_norm": 0.10246041417121887,
      "learning_rate": 1.345809523809524e-05,
      "loss": 0.012,
      "step": 6870
    },
    {
      "epoch": 0.9828571428571429,
      "grad_norm": 0.27164363861083984,
      "learning_rate": 1.3448571428571428e-05,
      "loss": 1.6945,
      "step": 6880
    },
    {
      "epoch": 0.9842857142857143,
      "grad_norm": 0.09343203157186508,
      "learning_rate": 1.343904761904762e-05,
      "loss": 0.0,
      "step": 6890
    },
    {
      "epoch": 0.9857142857142858,
      "grad_norm": 0.0890515074133873,
      "learning_rate": 1.342952380952381e-05,
      "loss": 0.0,
      "step": 6900
    },
    {
      "epoch": 0.9871428571428571,
      "grad_norm": 2649089.25,
      "learning_rate": 1.3420000000000002e-05,
      "loss": 5.0504,
      "step": 6910
    },
    {
      "epoch": 0.9885714285714285,
      "grad_norm": 0.28675147891044617,
      "learning_rate": 1.3410476190476191e-05,
      "loss": 5.0506,
      "step": 6920
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3538757860660553,
      "learning_rate": 1.3400952380952382e-05,
      "loss": 6.5864,
      "step": 6930
    },
    {
      "epoch": 0.9914285714285714,
      "grad_norm": 0.2661507725715637,
      "learning_rate": 1.3391428571428572e-05,
      "loss": 0.0,
      "step": 6940
    },
    {
      "epoch": 0.9928571428571429,
      "grad_norm": 0.31976190209388733,
      "learning_rate": 1.3381904761904763e-05,
      "loss": 1.6721,
      "step": 6950
    },
    {
      "epoch": 0.9942857142857143,
      "grad_norm": 42689024.0,
      "learning_rate": 1.3372380952380955e-05,
      "loss": 1.5449,
      "step": 6960
    },
    {
      "epoch": 0.9957142857142857,
      "grad_norm": 0.28577619791030884,
      "learning_rate": 1.3362857142857144e-05,
      "loss": 1.621,
      "step": 6970
    },
    {
      "epoch": 0.9971428571428571,
      "grad_norm": 0.2734738886356354,
      "learning_rate": 1.3353333333333333e-05,
      "loss": 1.3187,
      "step": 6980
    },
    {
      "epoch": 0.9985714285714286,
      "grad_norm": 0.2536512613296509,
      "learning_rate": 1.3343809523809525e-05,
      "loss": 1.6528,
      "step": 6990
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.08253481984138489,
      "learning_rate": 1.3334285714285714e-05,
      "loss": 0.0,
      "step": 7000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8581666666666666,
      "eval_f1_macro": 0.8568519913791592,
      "eval_f1_weighted": 0.8582786998650802,
      "eval_loss": 2.2743473052978516,
      "eval_runtime": 254.3442,
      "eval_samples_per_second": 23.59,
      "eval_steps_per_second": 11.795,
      "step": 7000
    },
    {
      "epoch": 1.0014285714285713,
      "grad_norm": 0.06524629145860672,
      "learning_rate": 1.3324761904761905e-05,
      "loss": 1.719,
      "step": 7010
    },
    {
      "epoch": 1.002857142857143,
      "grad_norm": 0.07639039307832718,
      "learning_rate": 1.3315238095238097e-05,
      "loss": 1.1487,
      "step": 7020
    },
    {
      "epoch": 1.0042857142857142,
      "grad_norm": 10.756919860839844,
      "learning_rate": 1.3305714285714286e-05,
      "loss": 1.6901,
      "step": 7030
    },
    {
      "epoch": 1.0057142857142858,
      "grad_norm": 0.2843257486820221,
      "learning_rate": 1.3296190476190479e-05,
      "loss": 0.0,
      "step": 7040
    },
    {
      "epoch": 1.0071428571428571,
      "grad_norm": 0.07870824635028839,
      "learning_rate": 1.3286666666666668e-05,
      "loss": 1.6764,
      "step": 7050
    },
    {
      "epoch": 1.0085714285714287,
      "grad_norm": 0.08176097273826599,
      "learning_rate": 1.3277142857142857e-05,
      "loss": 1.6902,
      "step": 7060
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.05623985081911087,
      "learning_rate": 1.3267619047619049e-05,
      "loss": 0.0,
      "step": 7070
    },
    {
      "epoch": 1.0114285714285713,
      "grad_norm": 0.10545434802770615,
      "learning_rate": 1.325809523809524e-05,
      "loss": 1.7213,
      "step": 7080
    },
    {
      "epoch": 1.012857142857143,
      "grad_norm": 0.2660452425479889,
      "learning_rate": 1.324857142857143e-05,
      "loss": 0.205,
      "step": 7090
    },
    {
      "epoch": 1.0142857142857142,
      "grad_norm": 3539808.75,
      "learning_rate": 1.3239047619047621e-05,
      "loss": 4.9952,
      "step": 7100
    },
    {
      "epoch": 1.0157142857142858,
      "grad_norm": 0.1172022894024849,
      "learning_rate": 1.322952380952381e-05,
      "loss": 1.7411,
      "step": 7110
    },
    {
      "epoch": 1.0171428571428571,
      "grad_norm": 0.280821293592453,
      "learning_rate": 1.3220000000000002e-05,
      "loss": 3.371,
      "step": 7120
    },
    {
      "epoch": 1.0185714285714287,
      "grad_norm": 0.36806708574295044,
      "learning_rate": 1.3210476190476191e-05,
      "loss": 1.6237,
      "step": 7130
    },
    {
      "epoch": 1.02,
      "grad_norm": 2906476.25,
      "learning_rate": 1.3200952380952382e-05,
      "loss": 4.9218,
      "step": 7140
    },
    {
      "epoch": 1.0214285714285714,
      "grad_norm": 0.06572098284959793,
      "learning_rate": 1.3191428571428572e-05,
      "loss": 0.0,
      "step": 7150
    },
    {
      "epoch": 1.022857142857143,
      "grad_norm": 0.08447570353746414,
      "learning_rate": 1.3181904761904763e-05,
      "loss": 1.2031,
      "step": 7160
    },
    {
      "epoch": 1.0242857142857142,
      "grad_norm": 0.08211159706115723,
      "learning_rate": 1.3172380952380954e-05,
      "loss": 1.702,
      "step": 7170
    },
    {
      "epoch": 1.0257142857142858,
      "grad_norm": 0.078719861805439,
      "learning_rate": 1.3162857142857144e-05,
      "loss": 1.6854,
      "step": 7180
    },
    {
      "epoch": 1.0271428571428571,
      "grad_norm": 0.07860417664051056,
      "learning_rate": 1.3153333333333333e-05,
      "loss": 6.3192,
      "step": 7190
    },
    {
      "epoch": 1.0285714285714285,
      "grad_norm": 0.3389703035354614,
      "learning_rate": 1.3143809523809526e-05,
      "loss": 1.6538,
      "step": 7200
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.0831182524561882,
      "learning_rate": 1.3134285714285715e-05,
      "loss": 3.5917,
      "step": 7210
    },
    {
      "epoch": 1.0314285714285714,
      "grad_norm": 1.7051129341125488,
      "learning_rate": 1.3124761904761907e-05,
      "loss": 1.029,
      "step": 7220
    },
    {
      "epoch": 1.032857142857143,
      "grad_norm": 0.07892736792564392,
      "learning_rate": 1.3115238095238096e-05,
      "loss": 1.6251,
      "step": 7230
    },
    {
      "epoch": 1.0342857142857143,
      "grad_norm": 0.07784108817577362,
      "learning_rate": 1.3105714285714287e-05,
      "loss": 1.6807,
      "step": 7240
    },
    {
      "epoch": 1.0357142857142858,
      "grad_norm": 0.0752163752913475,
      "learning_rate": 1.3096190476190477e-05,
      "loss": 0.0,
      "step": 7250
    },
    {
      "epoch": 1.0371428571428571,
      "grad_norm": 0.0473162867128849,
      "learning_rate": 1.3086666666666668e-05,
      "loss": 4.9185,
      "step": 7260
    },
    {
      "epoch": 1.0385714285714285,
      "grad_norm": 3190552.75,
      "learning_rate": 1.3077142857142857e-05,
      "loss": 4.9521,
      "step": 7270
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7256755828857422,
      "learning_rate": 1.306761904761905e-05,
      "loss": 1.6564,
      "step": 7280
    },
    {
      "epoch": 1.0414285714285714,
      "grad_norm": 0.3497380018234253,
      "learning_rate": 1.3058095238095238e-05,
      "loss": 0.0,
      "step": 7290
    },
    {
      "epoch": 1.042857142857143,
      "grad_norm": 0.11230095475912094,
      "learning_rate": 1.304857142857143e-05,
      "loss": 0.0,
      "step": 7300
    },
    {
      "epoch": 1.0442857142857143,
      "grad_norm": 0.09371715039014816,
      "learning_rate": 1.303904761904762e-05,
      "loss": 0.0,
      "step": 7310
    },
    {
      "epoch": 1.0457142857142858,
      "grad_norm": 0.06134510040283203,
      "learning_rate": 1.302952380952381e-05,
      "loss": 1.7166,
      "step": 7320
    },
    {
      "epoch": 1.0471428571428572,
      "grad_norm": 0.0936863049864769,
      "learning_rate": 1.302e-05,
      "loss": 1.7146,
      "step": 7330
    },
    {
      "epoch": 1.0485714285714285,
      "grad_norm": 0.058681171387434006,
      "learning_rate": 1.3010476190476191e-05,
      "loss": 1.6684,
      "step": 7340
    },
    {
      "epoch": 1.05,
      "grad_norm": 2863945.5,
      "learning_rate": 1.3000952380952384e-05,
      "loss": 3.2896,
      "step": 7350
    },
    {
      "epoch": 1.0514285714285714,
      "grad_norm": 0.04654539376497269,
      "learning_rate": 1.2991428571428573e-05,
      "loss": 3.3953,
      "step": 7360
    },
    {
      "epoch": 1.052857142857143,
      "grad_norm": 0.04255078732967377,
      "learning_rate": 1.2981904761904762e-05,
      "loss": 1.6735,
      "step": 7370
    },
    {
      "epoch": 1.0542857142857143,
      "grad_norm": 0.7997064590454102,
      "learning_rate": 1.2972380952380954e-05,
      "loss": 2.934,
      "step": 7380
    },
    {
      "epoch": 1.0557142857142856,
      "grad_norm": 2697320.75,
      "learning_rate": 1.2962857142857143e-05,
      "loss": 5.0927,
      "step": 7390
    },
    {
      "epoch": 1.0571428571428572,
      "grad_norm": 2.4146807193756104,
      "learning_rate": 1.2953333333333334e-05,
      "loss": 0.9925,
      "step": 7400
    },
    {
      "epoch": 1.0585714285714285,
      "grad_norm": 8682957.0,
      "learning_rate": 1.2943809523809526e-05,
      "loss": 4.8148,
      "step": 7410
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.2796831727027893,
      "learning_rate": 1.2934285714285715e-05,
      "loss": 5.1053,
      "step": 7420
    },
    {
      "epoch": 1.0614285714285714,
      "grad_norm": 0.09321077913045883,
      "learning_rate": 1.2924761904761907e-05,
      "loss": 0.0,
      "step": 7430
    },
    {
      "epoch": 1.062857142857143,
      "grad_norm": 0.279420405626297,
      "learning_rate": 1.2915238095238096e-05,
      "loss": 1.7014,
      "step": 7440
    },
    {
      "epoch": 1.0642857142857143,
      "grad_norm": 0.06750282645225525,
      "learning_rate": 1.2905714285714285e-05,
      "loss": 1.6365,
      "step": 7450
    },
    {
      "epoch": 1.0657142857142856,
      "grad_norm": 2700408.5,
      "learning_rate": 1.2896190476190477e-05,
      "loss": 3.3975,
      "step": 7460
    },
    {
      "epoch": 1.0671428571428572,
      "grad_norm": 0.07228533178567886,
      "learning_rate": 1.2886666666666668e-05,
      "loss": 0.0,
      "step": 7470
    },
    {
      "epoch": 1.0685714285714285,
      "grad_norm": 0.08984232693910599,
      "learning_rate": 1.2877142857142859e-05,
      "loss": 0.0,
      "step": 7480
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.06164906173944473,
      "learning_rate": 1.286761904761905e-05,
      "loss": 3.0807,
      "step": 7490
    },
    {
      "epoch": 1.0714285714285714,
      "grad_norm": 0.05915174260735512,
      "learning_rate": 1.2858095238095238e-05,
      "loss": 0.0,
      "step": 7500
    },
    {
      "epoch": 1.072857142857143,
      "grad_norm": 0.05659664794802666,
      "learning_rate": 1.284857142857143e-05,
      "loss": 0.5366,
      "step": 7510
    },
    {
      "epoch": 1.0742857142857143,
      "grad_norm": 0.11282779276371002,
      "learning_rate": 1.283904761904762e-05,
      "loss": 3.0671,
      "step": 7520
    },
    {
      "epoch": 1.0757142857142856,
      "grad_norm": 19.017818450927734,
      "learning_rate": 1.282952380952381e-05,
      "loss": 3.3296,
      "step": 7530
    },
    {
      "epoch": 1.0771428571428572,
      "grad_norm": 0.07697110623121262,
      "learning_rate": 1.2820000000000001e-05,
      "loss": 0.0,
      "step": 7540
    },
    {
      "epoch": 1.0785714285714285,
      "grad_norm": 0.4340910017490387,
      "learning_rate": 1.2810476190476192e-05,
      "loss": 1.653,
      "step": 7550
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.079569011926651,
      "learning_rate": 1.2800952380952382e-05,
      "loss": 1.6302,
      "step": 7560
    },
    {
      "epoch": 1.0814285714285714,
      "grad_norm": 0.26589831709861755,
      "learning_rate": 1.2791428571428573e-05,
      "loss": 3.2626,
      "step": 7570
    },
    {
      "epoch": 1.0828571428571427,
      "grad_norm": 0.32339560985565186,
      "learning_rate": 1.2781904761904762e-05,
      "loss": 0.0,
      "step": 7580
    },
    {
      "epoch": 1.0842857142857143,
      "grad_norm": 0.07803523540496826,
      "learning_rate": 1.2772380952380954e-05,
      "loss": 4.8012,
      "step": 7590
    },
    {
      "epoch": 1.0857142857142856,
      "grad_norm": 2867901.5,
      "learning_rate": 1.2762857142857143e-05,
      "loss": 3.0946,
      "step": 7600
    },
    {
      "epoch": 1.0871428571428572,
      "grad_norm": 0.07056844979524612,
      "learning_rate": 1.2753333333333335e-05,
      "loss": 1.6903,
      "step": 7610
    },
    {
      "epoch": 1.0885714285714285,
      "grad_norm": 0.1086554005742073,
      "learning_rate": 1.2743809523809524e-05,
      "loss": 3.2361,
      "step": 7620
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.0701163038611412,
      "learning_rate": 1.2734285714285715e-05,
      "loss": 4.5889,
      "step": 7630
    },
    {
      "epoch": 1.0914285714285714,
      "grad_norm": 11624050.0,
      "learning_rate": 1.2724761904761906e-05,
      "loss": 3.2511,
      "step": 7640
    },
    {
      "epoch": 1.092857142857143,
      "grad_norm": 0.09211303293704987,
      "learning_rate": 1.2715238095238096e-05,
      "loss": 1.6567,
      "step": 7650
    },
    {
      "epoch": 1.0942857142857143,
      "grad_norm": 0.2591836452484131,
      "learning_rate": 1.2705714285714285e-05,
      "loss": 1.6629,
      "step": 7660
    },
    {
      "epoch": 1.0957142857142856,
      "grad_norm": 0.11361389607191086,
      "learning_rate": 1.2696190476190478e-05,
      "loss": 3.3657,
      "step": 7670
    },
    {
      "epoch": 1.0971428571428572,
      "grad_norm": 0.07622842490673065,
      "learning_rate": 1.2686666666666667e-05,
      "loss": 0.0,
      "step": 7680
    },
    {
      "epoch": 1.0985714285714285,
      "grad_norm": 0.2590929865837097,
      "learning_rate": 1.2677142857142859e-05,
      "loss": 1.7153,
      "step": 7690
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.06755845248699188,
      "learning_rate": 1.2667619047619048e-05,
      "loss": 3.3219,
      "step": 7700
    },
    {
      "epoch": 1.1014285714285714,
      "grad_norm": 0.10156223177909851,
      "learning_rate": 1.2658095238095239e-05,
      "loss": 3.1703,
      "step": 7710
    },
    {
      "epoch": 1.1028571428571428,
      "grad_norm": 0.07093604654073715,
      "learning_rate": 1.2648571428571431e-05,
      "loss": 5.0764,
      "step": 7720
    },
    {
      "epoch": 1.1042857142857143,
      "grad_norm": 0.30190858244895935,
      "learning_rate": 1.263904761904762e-05,
      "loss": 0.0,
      "step": 7730
    },
    {
      "epoch": 1.1057142857142856,
      "grad_norm": 0.09366699308156967,
      "learning_rate": 1.2629523809523812e-05,
      "loss": 1.42,
      "step": 7740
    },
    {
      "epoch": 1.1071428571428572,
      "grad_norm": 0.09598547220230103,
      "learning_rate": 1.2620000000000001e-05,
      "loss": 3.3778,
      "step": 7750
    },
    {
      "epoch": 1.1085714285714285,
      "grad_norm": 2745100.75,
      "learning_rate": 1.261047619047619e-05,
      "loss": 3.158,
      "step": 7760
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.25911951065063477,
      "learning_rate": 1.2600952380952382e-05,
      "loss": 4.9961,
      "step": 7770
    },
    {
      "epoch": 1.1114285714285714,
      "grad_norm": 2696361.75,
      "learning_rate": 1.2591428571428571e-05,
      "loss": 5.8073,
      "step": 7780
    },
    {
      "epoch": 1.1128571428571428,
      "grad_norm": 0.3467887341976166,
      "learning_rate": 1.2581904761904762e-05,
      "loss": 3.2126,
      "step": 7790
    },
    {
      "epoch": 1.1142857142857143,
      "grad_norm": 0.2688603401184082,
      "learning_rate": 1.2572380952380954e-05,
      "loss": 0.0,
      "step": 7800
    },
    {
      "epoch": 1.1157142857142857,
      "grad_norm": 35454340.0,
      "learning_rate": 1.2562857142857143e-05,
      "loss": 1.5833,
      "step": 7810
    },
    {
      "epoch": 1.1171428571428572,
      "grad_norm": 0.2633322477340698,
      "learning_rate": 1.2553333333333336e-05,
      "loss": 3.2492,
      "step": 7820
    },
    {
      "epoch": 1.1185714285714285,
      "grad_norm": 0.09941697120666504,
      "learning_rate": 1.2543809523809525e-05,
      "loss": 1.7284,
      "step": 7830
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.2717256247997284,
      "learning_rate": 1.2534285714285714e-05,
      "loss": 3.2648,
      "step": 7840
    },
    {
      "epoch": 1.1214285714285714,
      "grad_norm": 3656444.0,
      "learning_rate": 1.2524761904761906e-05,
      "loss": 4.3964,
      "step": 7850
    },
    {
      "epoch": 1.1228571428571428,
      "grad_norm": 0.2762461304664612,
      "learning_rate": 1.2515238095238097e-05,
      "loss": 1.029,
      "step": 7860
    },
    {
      "epoch": 1.1242857142857143,
      "grad_norm": 0.08754371851682663,
      "learning_rate": 1.2505714285714287e-05,
      "loss": 0.0248,
      "step": 7870
    },
    {
      "epoch": 1.1257142857142857,
      "grad_norm": 0.08332744985818863,
      "learning_rate": 1.2496190476190478e-05,
      "loss": 3.3295,
      "step": 7880
    },
    {
      "epoch": 1.1271428571428572,
      "grad_norm": 0.07152831554412842,
      "learning_rate": 1.2486666666666667e-05,
      "loss": 3.103,
      "step": 7890
    },
    {
      "epoch": 1.1285714285714286,
      "grad_norm": 0.06777605414390564,
      "learning_rate": 1.247714285714286e-05,
      "loss": 3.3063,
      "step": 7900
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.11003103107213974,
      "learning_rate": 1.2467619047619048e-05,
      "loss": 0.0,
      "step": 7910
    },
    {
      "epoch": 1.1314285714285715,
      "grad_norm": 0.0832381621003151,
      "learning_rate": 1.2458095238095239e-05,
      "loss": 1.5967,
      "step": 7920
    },
    {
      "epoch": 1.1328571428571428,
      "grad_norm": 0.263389527797699,
      "learning_rate": 1.244857142857143e-05,
      "loss": 1.677,
      "step": 7930
    },
    {
      "epoch": 1.1342857142857143,
      "grad_norm": 2752875.25,
      "learning_rate": 1.243904761904762e-05,
      "loss": 3.3576,
      "step": 7940
    },
    {
      "epoch": 1.1357142857142857,
      "grad_norm": 2763614.0,
      "learning_rate": 1.242952380952381e-05,
      "loss": 5.013,
      "step": 7950
    },
    {
      "epoch": 1.1371428571428572,
      "grad_norm": 1515.222900390625,
      "learning_rate": 1.2420000000000001e-05,
      "loss": 3.3172,
      "step": 7960
    },
    {
      "epoch": 1.1385714285714286,
      "grad_norm": 0.08413535356521606,
      "learning_rate": 1.241047619047619e-05,
      "loss": 3.2049,
      "step": 7970
    },
    {
      "epoch": 1.1400000000000001,
      "grad_norm": 0.43008384108543396,
      "learning_rate": 1.2400952380952383e-05,
      "loss": 6.4701,
      "step": 7980
    },
    {
      "epoch": 1.1414285714285715,
      "grad_norm": 0.3334710896015167,
      "learning_rate": 1.2391428571428572e-05,
      "loss": 1.5362,
      "step": 7990
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 0.38969752192497253,
      "learning_rate": 1.2381904761904764e-05,
      "loss": 3.0478,
      "step": 8000
    },
    {
      "epoch": 1.1442857142857144,
      "grad_norm": 0.07484236359596252,
      "learning_rate": 1.2372380952380953e-05,
      "loss": 0.0,
      "step": 8010
    },
    {
      "epoch": 1.1457142857142857,
      "grad_norm": 2762522.75,
      "learning_rate": 1.2362857142857144e-05,
      "loss": 4.7588,
      "step": 8020
    },
    {
      "epoch": 1.1471428571428572,
      "grad_norm": 2789983.25,
      "learning_rate": 1.2353333333333334e-05,
      "loss": 1.5842,
      "step": 8030
    },
    {
      "epoch": 1.1485714285714286,
      "grad_norm": 0.048898447304964066,
      "learning_rate": 1.2343809523809525e-05,
      "loss": 1.5802,
      "step": 8040
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.07350628823041916,
      "learning_rate": 1.2334285714285714e-05,
      "loss": 1.748,
      "step": 8050
    },
    {
      "epoch": 1.1514285714285715,
      "grad_norm": 0.2740125060081482,
      "learning_rate": 1.2324761904761906e-05,
      "loss": 4.8298,
      "step": 8060
    },
    {
      "epoch": 1.1528571428571428,
      "grad_norm": 2741855.5,
      "learning_rate": 1.2315238095238095e-05,
      "loss": 5.0608,
      "step": 8070
    },
    {
      "epoch": 1.1542857142857144,
      "grad_norm": 20284618.0,
      "learning_rate": 1.2305714285714287e-05,
      "loss": 8.305,
      "step": 8080
    },
    {
      "epoch": 1.1557142857142857,
      "grad_norm": 0.31602948904037476,
      "learning_rate": 1.2296190476190476e-05,
      "loss": 2.9488,
      "step": 8090
    },
    {
      "epoch": 1.157142857142857,
      "grad_norm": 0.11067284643650055,
      "learning_rate": 1.2286666666666667e-05,
      "loss": 3.1949,
      "step": 8100
    },
    {
      "epoch": 1.1585714285714286,
      "grad_norm": 112.5748519897461,
      "learning_rate": 1.227714285714286e-05,
      "loss": 1.5847,
      "step": 8110
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.2927233576774597,
      "learning_rate": 1.2267619047619048e-05,
      "loss": 1.6538,
      "step": 8120
    },
    {
      "epoch": 1.1614285714285715,
      "grad_norm": 0.09973874688148499,
      "learning_rate": 1.225809523809524e-05,
      "loss": 3.2714,
      "step": 8130
    },
    {
      "epoch": 1.1628571428571428,
      "grad_norm": 3246513.75,
      "learning_rate": 1.224857142857143e-05,
      "loss": 4.8792,
      "step": 8140
    },
    {
      "epoch": 1.1642857142857144,
      "grad_norm": 0.31462693214416504,
      "learning_rate": 1.2239047619047619e-05,
      "loss": 1.5942,
      "step": 8150
    },
    {
      "epoch": 1.1657142857142857,
      "grad_norm": 0.26772210001945496,
      "learning_rate": 1.2229523809523811e-05,
      "loss": 0.0,
      "step": 8160
    },
    {
      "epoch": 1.1671428571428573,
      "grad_norm": 0.09527409076690674,
      "learning_rate": 1.2220000000000002e-05,
      "loss": 3.2938,
      "step": 8170
    },
    {
      "epoch": 1.1685714285714286,
      "grad_norm": 0.2835180163383484,
      "learning_rate": 1.221047619047619e-05,
      "loss": 1.67,
      "step": 8180
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.103312186896801,
      "learning_rate": 1.2200952380952383e-05,
      "loss": 1.729,
      "step": 8190
    },
    {
      "epoch": 1.1714285714285715,
      "grad_norm": 0.07407815009355545,
      "learning_rate": 1.2191428571428572e-05,
      "loss": 3.3544,
      "step": 8200
    },
    {
      "epoch": 1.1728571428571428,
      "grad_norm": 0.04922327771782875,
      "learning_rate": 1.2181904761904764e-05,
      "loss": 1.7425,
      "step": 8210
    },
    {
      "epoch": 1.1742857142857144,
      "grad_norm": 2665470.0,
      "learning_rate": 1.2172380952380953e-05,
      "loss": 3.3905,
      "step": 8220
    },
    {
      "epoch": 1.1757142857142857,
      "grad_norm": 2681768.0,
      "learning_rate": 1.2162857142857142e-05,
      "loss": 3.3385,
      "step": 8230
    },
    {
      "epoch": 1.177142857142857,
      "grad_norm": 0.053927231580019,
      "learning_rate": 1.2153333333333334e-05,
      "loss": 0.0,
      "step": 8240
    },
    {
      "epoch": 1.1785714285714286,
      "grad_norm": 3329077.5,
      "learning_rate": 1.2143809523809525e-05,
      "loss": 6.6314,
      "step": 8250
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.2745104432106018,
      "learning_rate": 1.2134285714285716e-05,
      "loss": 3.2413,
      "step": 8260
    },
    {
      "epoch": 1.1814285714285715,
      "grad_norm": 2725320.5,
      "learning_rate": 1.2124761904761906e-05,
      "loss": 3.2662,
      "step": 8270
    },
    {
      "epoch": 1.1828571428571428,
      "grad_norm": 0.2871527075767517,
      "learning_rate": 1.2115238095238095e-05,
      "loss": 7.937,
      "step": 8280
    },
    {
      "epoch": 1.1842857142857142,
      "grad_norm": 0.30602556467056274,
      "learning_rate": 1.2105714285714288e-05,
      "loss": 1.5795,
      "step": 8290
    },
    {
      "epoch": 1.1857142857142857,
      "grad_norm": 0.11223508417606354,
      "learning_rate": 1.2096190476190477e-05,
      "loss": 1.6302,
      "step": 8300
    },
    {
      "epoch": 1.187142857142857,
      "grad_norm": 0.08333160728216171,
      "learning_rate": 1.2086666666666667e-05,
      "loss": 3.2025,
      "step": 8310
    },
    {
      "epoch": 1.1885714285714286,
      "grad_norm": 0.06381603330373764,
      "learning_rate": 1.2077142857142858e-05,
      "loss": 0.0,
      "step": 8320
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.08236698061227798,
      "learning_rate": 1.2067619047619049e-05,
      "loss": 4.8874,
      "step": 8330
    },
    {
      "epoch": 1.1914285714285715,
      "grad_norm": 0.08024781942367554,
      "learning_rate": 1.205809523809524e-05,
      "loss": 3.2252,
      "step": 8340
    },
    {
      "epoch": 1.1928571428571428,
      "grad_norm": 0.2607198655605316,
      "learning_rate": 1.204857142857143e-05,
      "loss": 0.0,
      "step": 8350
    },
    {
      "epoch": 1.1942857142857144,
      "grad_norm": 0.35439208149909973,
      "learning_rate": 1.2039047619047619e-05,
      "loss": 3.385,
      "step": 8360
    },
    {
      "epoch": 1.1957142857142857,
      "grad_norm": 7788365.5,
      "learning_rate": 1.2029523809523811e-05,
      "loss": 5.6777,
      "step": 8370
    },
    {
      "epoch": 1.197142857142857,
      "grad_norm": 0.054066404700279236,
      "learning_rate": 1.202e-05,
      "loss": 0.0,
      "step": 8380
    },
    {
      "epoch": 1.1985714285714286,
      "grad_norm": 0.08664076775312424,
      "learning_rate": 1.2010476190476192e-05,
      "loss": 3.2902,
      "step": 8390
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.09972183406352997,
      "learning_rate": 1.2000952380952381e-05,
      "loss": 4.6255,
      "step": 8400
    },
    {
      "epoch": 1.2014285714285715,
      "grad_norm": 0.08062146604061127,
      "learning_rate": 1.1991428571428572e-05,
      "loss": 0.0,
      "step": 8410
    },
    {
      "epoch": 1.2028571428571428,
      "grad_norm": 2714186.5,
      "learning_rate": 1.1981904761904763e-05,
      "loss": 3.3599,
      "step": 8420
    },
    {
      "epoch": 1.2042857142857142,
      "grad_norm": 0.08378219604492188,
      "learning_rate": 1.1972380952380953e-05,
      "loss": 0.0,
      "step": 8430
    },
    {
      "epoch": 1.2057142857142857,
      "grad_norm": 0.10963328182697296,
      "learning_rate": 1.1962857142857146e-05,
      "loss": 0.0,
      "step": 8440
    },
    {
      "epoch": 1.207142857142857,
      "grad_norm": 0.07506141066551208,
      "learning_rate": 1.1953333333333335e-05,
      "loss": 0.0,
      "step": 8450
    },
    {
      "epoch": 1.2085714285714286,
      "grad_norm": 0.11016018688678741,
      "learning_rate": 1.1943809523809524e-05,
      "loss": 0.0,
      "step": 8460
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.08467579632997513,
      "learning_rate": 1.1934285714285716e-05,
      "loss": 3.3044,
      "step": 8470
    },
    {
      "epoch": 1.2114285714285715,
      "grad_norm": 0.06293339282274246,
      "learning_rate": 1.1924761904761905e-05,
      "loss": 0.0,
      "step": 8480
    },
    {
      "epoch": 1.2128571428571429,
      "grad_norm": 0.09842652082443237,
      "learning_rate": 1.1915238095238096e-05,
      "loss": 1.708,
      "step": 8490
    },
    {
      "epoch": 1.2142857142857142,
      "grad_norm": 0.10157168656587601,
      "learning_rate": 1.1905714285714288e-05,
      "loss": 0.0,
      "step": 8500
    },
    {
      "epoch": 1.2157142857142857,
      "grad_norm": 0.07940179109573364,
      "learning_rate": 1.1896190476190477e-05,
      "loss": 1.7338,
      "step": 8510
    },
    {
      "epoch": 1.217142857142857,
      "grad_norm": 0.05702076852321625,
      "learning_rate": 1.188666666666667e-05,
      "loss": 1.7185,
      "step": 8520
    },
    {
      "epoch": 1.2185714285714286,
      "grad_norm": 2789342.25,
      "learning_rate": 1.1877142857142858e-05,
      "loss": 3.4338,
      "step": 8530
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.04681907594203949,
      "learning_rate": 1.1867619047619047e-05,
      "loss": 1.6446,
      "step": 8540
    },
    {
      "epoch": 1.2214285714285715,
      "grad_norm": 0.06916458904743195,
      "learning_rate": 1.185809523809524e-05,
      "loss": 1.6501,
      "step": 8550
    },
    {
      "epoch": 1.2228571428571429,
      "grad_norm": 0.06291616708040237,
      "learning_rate": 1.184857142857143e-05,
      "loss": 1.74,
      "step": 8560
    },
    {
      "epoch": 1.2242857142857142,
      "grad_norm": 0.270444393157959,
      "learning_rate": 1.183904761904762e-05,
      "loss": 1.6978,
      "step": 8570
    },
    {
      "epoch": 1.2257142857142858,
      "grad_norm": 0.10983391106128693,
      "learning_rate": 1.1829523809523811e-05,
      "loss": 3.3699,
      "step": 8580
    },
    {
      "epoch": 1.227142857142857,
      "grad_norm": 0.09734371304512024,
      "learning_rate": 1.182e-05,
      "loss": 3.8659,
      "step": 8590
    },
    {
      "epoch": 1.2285714285714286,
      "grad_norm": 0.06517007201910019,
      "learning_rate": 1.1810476190476193e-05,
      "loss": 1.6832,
      "step": 8600
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.08319639414548874,
      "learning_rate": 1.1800952380952382e-05,
      "loss": 3.3265,
      "step": 8610
    },
    {
      "epoch": 1.2314285714285713,
      "grad_norm": 0.33907169103622437,
      "learning_rate": 1.1791428571428572e-05,
      "loss": 3.3481,
      "step": 8620
    },
    {
      "epoch": 1.2328571428571429,
      "grad_norm": 2698721.0,
      "learning_rate": 1.1781904761904763e-05,
      "loss": 3.3683,
      "step": 8630
    },
    {
      "epoch": 1.2342857142857142,
      "grad_norm": 0.3113778233528137,
      "learning_rate": 1.1772380952380954e-05,
      "loss": 0.0,
      "step": 8640
    },
    {
      "epoch": 1.2357142857142858,
      "grad_norm": 2735741.25,
      "learning_rate": 1.1762857142857144e-05,
      "loss": 3.252,
      "step": 8650
    },
    {
      "epoch": 1.237142857142857,
      "grad_norm": 0.07275033742189407,
      "learning_rate": 1.1753333333333335e-05,
      "loss": 3.3158,
      "step": 8660
    },
    {
      "epoch": 1.2385714285714287,
      "grad_norm": 2724238.0,
      "learning_rate": 1.1743809523809524e-05,
      "loss": 4.9737,
      "step": 8670
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.27408841252326965,
      "learning_rate": 1.1734285714285716e-05,
      "loss": 1.7225,
      "step": 8680
    },
    {
      "epoch": 1.2414285714285715,
      "grad_norm": 3167011.75,
      "learning_rate": 1.1724761904761905e-05,
      "loss": 1.5233,
      "step": 8690
    },
    {
      "epoch": 1.2428571428571429,
      "grad_norm": 0.08655983209609985,
      "learning_rate": 1.1715238095238097e-05,
      "loss": 3.3664,
      "step": 8700
    },
    {
      "epoch": 1.2442857142857142,
      "grad_norm": 0.07498035579919815,
      "learning_rate": 1.1705714285714286e-05,
      "loss": 3.2522,
      "step": 8710
    },
    {
      "epoch": 1.2457142857142858,
      "grad_norm": 0.3165728747844696,
      "learning_rate": 1.1696190476190477e-05,
      "loss": 0.0,
      "step": 8720
    },
    {
      "epoch": 1.247142857142857,
      "grad_norm": 0.2656264007091522,
      "learning_rate": 1.1686666666666668e-05,
      "loss": 0.0,
      "step": 8730
    },
    {
      "epoch": 1.2485714285714287,
      "grad_norm": 0.08162735402584076,
      "learning_rate": 1.1677142857142858e-05,
      "loss": 4.9558,
      "step": 8740
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.10963474959135056,
      "learning_rate": 1.1667619047619047e-05,
      "loss": 1.7131,
      "step": 8750
    },
    {
      "epoch": 1.2514285714285713,
      "grad_norm": 0.07270567864179611,
      "learning_rate": 1.165809523809524e-05,
      "loss": 1.7098,
      "step": 8760
    },
    {
      "epoch": 1.252857142857143,
      "grad_norm": 0.1089026927947998,
      "learning_rate": 1.1648571428571429e-05,
      "loss": 0.0,
      "step": 8770
    },
    {
      "epoch": 1.2542857142857142,
      "grad_norm": 0.08585276454687119,
      "learning_rate": 1.1639047619047621e-05,
      "loss": 1.6807,
      "step": 8780
    },
    {
      "epoch": 1.2557142857142858,
      "grad_norm": 0.2645547389984131,
      "learning_rate": 1.162952380952381e-05,
      "loss": 3.3498,
      "step": 8790
    },
    {
      "epoch": 1.2571428571428571,
      "grad_norm": 0.05401886627078056,
      "learning_rate": 1.162e-05,
      "loss": 1.6255,
      "step": 8800
    },
    {
      "epoch": 1.2585714285714285,
      "grad_norm": 0.11345364898443222,
      "learning_rate": 1.1610476190476191e-05,
      "loss": 1.6805,
      "step": 8810
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.06991084665060043,
      "learning_rate": 1.1600952380952382e-05,
      "loss": 3.3292,
      "step": 8820
    },
    {
      "epoch": 1.2614285714285716,
      "grad_norm": 3570265.0,
      "learning_rate": 1.1591428571428574e-05,
      "loss": 1.6384,
      "step": 8830
    },
    {
      "epoch": 1.262857142857143,
      "grad_norm": 0.06628427654504776,
      "learning_rate": 1.1581904761904763e-05,
      "loss": 0.0,
      "step": 8840
    },
    {
      "epoch": 1.2642857142857142,
      "grad_norm": 0.10371081531047821,
      "learning_rate": 1.1572380952380952e-05,
      "loss": 1.6581,
      "step": 8850
    },
    {
      "epoch": 1.2657142857142858,
      "grad_norm": 0.2678084373474121,
      "learning_rate": 1.1562857142857144e-05,
      "loss": 1.6734,
      "step": 8860
    },
    {
      "epoch": 1.2671428571428571,
      "grad_norm": 0.04184602573513985,
      "learning_rate": 1.1553333333333333e-05,
      "loss": 0.0,
      "step": 8870
    },
    {
      "epoch": 1.2685714285714287,
      "grad_norm": 0.3470386862754822,
      "learning_rate": 1.1543809523809524e-05,
      "loss": 0.0,
      "step": 8880
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.05774245038628578,
      "learning_rate": 1.1534285714285716e-05,
      "loss": 6.3737,
      "step": 8890
    },
    {
      "epoch": 1.2714285714285714,
      "grad_norm": 0.07692985981702805,
      "learning_rate": 1.1524761904761905e-05,
      "loss": 3.3805,
      "step": 8900
    },
    {
      "epoch": 1.272857142857143,
      "grad_norm": 0.07886411994695663,
      "learning_rate": 1.1515238095238098e-05,
      "loss": 0.0,
      "step": 8910
    },
    {
      "epoch": 1.2742857142857142,
      "grad_norm": 0.06742313504219055,
      "learning_rate": 1.1505714285714287e-05,
      "loss": 0.0,
      "step": 8920
    },
    {
      "epoch": 1.2757142857142858,
      "grad_norm": 0.09272091090679169,
      "learning_rate": 1.1496190476190476e-05,
      "loss": 3.318,
      "step": 8930
    },
    {
      "epoch": 1.2771428571428571,
      "grad_norm": 0.27502772212028503,
      "learning_rate": 1.1486666666666668e-05,
      "loss": 1.7077,
      "step": 8940
    },
    {
      "epoch": 1.2785714285714285,
      "grad_norm": 3127166.0,
      "learning_rate": 1.1477142857142859e-05,
      "loss": 3.2867,
      "step": 8950
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.1033775806427002,
      "learning_rate": 1.146761904761905e-05,
      "loss": 0.0,
      "step": 8960
    },
    {
      "epoch": 1.2814285714285714,
      "grad_norm": 13785108480.0,
      "learning_rate": 1.145809523809524e-05,
      "loss": 2.1908,
      "step": 8970
    },
    {
      "epoch": 1.282857142857143,
      "grad_norm": 0.05370601639151573,
      "learning_rate": 1.1448571428571429e-05,
      "loss": 1.6406,
      "step": 8980
    },
    {
      "epoch": 1.2842857142857143,
      "grad_norm": 0.06493110209703445,
      "learning_rate": 1.1439047619047621e-05,
      "loss": 1.6006,
      "step": 8990
    },
    {
      "epoch": 1.2857142857142856,
      "grad_norm": 0.05978746712207794,
      "learning_rate": 1.142952380952381e-05,
      "loss": 0.0,
      "step": 9000
    },
    {
      "epoch": 1.2871428571428571,
      "grad_norm": 0.04053087159991264,
      "learning_rate": 1.142e-05,
      "loss": 3.3989,
      "step": 9010
    },
    {
      "epoch": 1.2885714285714287,
      "grad_norm": 3077025.5,
      "learning_rate": 1.1410476190476191e-05,
      "loss": 1.6508,
      "step": 9020
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.27468204498291016,
      "learning_rate": 1.1400952380952382e-05,
      "loss": 1.6712,
      "step": 9030
    },
    {
      "epoch": 1.2914285714285714,
      "grad_norm": 3892680.0,
      "learning_rate": 1.1391428571428573e-05,
      "loss": 1.6122,
      "step": 9040
    },
    {
      "epoch": 1.292857142857143,
      "grad_norm": 0.4008592367172241,
      "learning_rate": 1.1381904761904763e-05,
      "loss": 0.0,
      "step": 9050
    },
    {
      "epoch": 1.2942857142857143,
      "grad_norm": 0.2640068829059601,
      "learning_rate": 1.1372380952380952e-05,
      "loss": 1.7453,
      "step": 9060
    },
    {
      "epoch": 1.2957142857142858,
      "grad_norm": 2713210.0,
      "learning_rate": 1.1362857142857145e-05,
      "loss": 1.668,
      "step": 9070
    },
    {
      "epoch": 1.2971428571428572,
      "grad_norm": 0.09720293432474136,
      "learning_rate": 1.1353333333333334e-05,
      "loss": 3.2868,
      "step": 9080
    },
    {
      "epoch": 1.2985714285714285,
      "grad_norm": 0.07251011580228806,
      "learning_rate": 1.1343809523809526e-05,
      "loss": 0.0,
      "step": 9090
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.052912719547748566,
      "learning_rate": 1.1334285714285715e-05,
      "loss": 1.6235,
      "step": 9100
    },
    {
      "epoch": 1.3014285714285714,
      "grad_norm": 0.45247340202331543,
      "learning_rate": 1.1324761904761906e-05,
      "loss": 3.3708,
      "step": 9110
    },
    {
      "epoch": 1.302857142857143,
      "grad_norm": 0.10093789547681808,
      "learning_rate": 1.1315238095238096e-05,
      "loss": 3.3363,
      "step": 9120
    },
    {
      "epoch": 1.3042857142857143,
      "grad_norm": 0.26444703340530396,
      "learning_rate": 1.1305714285714287e-05,
      "loss": 1.6147,
      "step": 9130
    },
    {
      "epoch": 1.3057142857142856,
      "grad_norm": 0.11089300364255905,
      "learning_rate": 1.1296190476190476e-05,
      "loss": 1.6522,
      "step": 9140
    },
    {
      "epoch": 1.3071428571428572,
      "grad_norm": 0.044950030744075775,
      "learning_rate": 1.1286666666666668e-05,
      "loss": 1.6975,
      "step": 9150
    },
    {
      "epoch": 1.3085714285714285,
      "grad_norm": 0.08187732845544815,
      "learning_rate": 1.1277142857142857e-05,
      "loss": 3.3664,
      "step": 9160
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.5063349008560181,
      "learning_rate": 1.126761904761905e-05,
      "loss": 3.1996,
      "step": 9170
    },
    {
      "epoch": 1.3114285714285714,
      "grad_norm": 0.07828164845705032,
      "learning_rate": 1.1258095238095238e-05,
      "loss": 3.0276,
      "step": 9180
    },
    {
      "epoch": 1.3128571428571427,
      "grad_norm": 0.06560466438531876,
      "learning_rate": 1.1248571428571429e-05,
      "loss": 0.0,
      "step": 9190
    },
    {
      "epoch": 1.3142857142857143,
      "grad_norm": 0.04751639440655708,
      "learning_rate": 1.123904761904762e-05,
      "loss": 3.1322,
      "step": 9200
    },
    {
      "epoch": 1.3157142857142858,
      "grad_norm": 2715818.75,
      "learning_rate": 1.122952380952381e-05,
      "loss": 2.4901,
      "step": 9210
    },
    {
      "epoch": 1.3171428571428572,
      "grad_norm": 0.11855614930391312,
      "learning_rate": 1.1220000000000003e-05,
      "loss": 1.6606,
      "step": 9220
    },
    {
      "epoch": 1.3185714285714285,
      "grad_norm": 0.059449855238199234,
      "learning_rate": 1.1210476190476192e-05,
      "loss": 1.1706,
      "step": 9230
    },
    {
      "epoch": 1.32,
      "grad_norm": 5901075.0,
      "learning_rate": 1.120095238095238e-05,
      "loss": 2.456,
      "step": 9240
    },
    {
      "epoch": 1.3214285714285714,
      "grad_norm": 2696627.75,
      "learning_rate": 1.1191428571428573e-05,
      "loss": 1.6543,
      "step": 9250
    },
    {
      "epoch": 1.322857142857143,
      "grad_norm": 0.10897507518529892,
      "learning_rate": 1.1181904761904762e-05,
      "loss": 0.0008,
      "step": 9260
    },
    {
      "epoch": 1.3242857142857143,
      "grad_norm": 0.05380300059914589,
      "learning_rate": 1.1172380952380953e-05,
      "loss": 3.3091,
      "step": 9270
    },
    {
      "epoch": 1.3257142857142856,
      "grad_norm": 0.29344090819358826,
      "learning_rate": 1.1162857142857145e-05,
      "loss": 1.661,
      "step": 9280
    },
    {
      "epoch": 1.3271428571428572,
      "grad_norm": 0.2629215121269226,
      "learning_rate": 1.1153333333333334e-05,
      "loss": 0.7251,
      "step": 9290
    },
    {
      "epoch": 1.3285714285714285,
      "grad_norm": 1.1299355030059814,
      "learning_rate": 1.1143809523809526e-05,
      "loss": 3.182,
      "step": 9300
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.11254875361919403,
      "learning_rate": 1.1134285714285715e-05,
      "loss": 3.4176,
      "step": 9310
    },
    {
      "epoch": 1.3314285714285714,
      "grad_norm": 0.05811762064695358,
      "learning_rate": 1.1124761904761904e-05,
      "loss": 3.5219,
      "step": 9320
    },
    {
      "epoch": 1.3328571428571427,
      "grad_norm": 0.2905595302581787,
      "learning_rate": 1.1115238095238096e-05,
      "loss": 0.0,
      "step": 9330
    },
    {
      "epoch": 1.3342857142857143,
      "grad_norm": 0.10448836535215378,
      "learning_rate": 1.1105714285714287e-05,
      "loss": 3.3943,
      "step": 9340
    },
    {
      "epoch": 1.3357142857142856,
      "grad_norm": 0.1012033075094223,
      "learning_rate": 1.1096190476190478e-05,
      "loss": 0.0,
      "step": 9350
    },
    {
      "epoch": 1.3371428571428572,
      "grad_norm": 0.06508680433034897,
      "learning_rate": 1.1086666666666668e-05,
      "loss": 1.6768,
      "step": 9360
    },
    {
      "epoch": 1.3385714285714285,
      "grad_norm": 2774042.75,
      "learning_rate": 1.1077142857142857e-05,
      "loss": 3.4558,
      "step": 9370
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.07781711965799332,
      "learning_rate": 1.106761904761905e-05,
      "loss": 2.5208,
      "step": 9380
    },
    {
      "epoch": 1.3414285714285714,
      "grad_norm": 0.07038260996341705,
      "learning_rate": 1.1058095238095239e-05,
      "loss": 1.6794,
      "step": 9390
    },
    {
      "epoch": 1.342857142857143,
      "grad_norm": 0.10716129094362259,
      "learning_rate": 1.104857142857143e-05,
      "loss": 2.1389,
      "step": 9400
    },
    {
      "epoch": 1.3442857142857143,
      "grad_norm": 2741737.5,
      "learning_rate": 1.103904761904762e-05,
      "loss": 3.3241,
      "step": 9410
    },
    {
      "epoch": 1.3457142857142856,
      "grad_norm": 0.09748080372810364,
      "learning_rate": 1.102952380952381e-05,
      "loss": 3.4857,
      "step": 9420
    },
    {
      "epoch": 1.3471428571428572,
      "grad_norm": 0.29078954458236694,
      "learning_rate": 1.1020000000000001e-05,
      "loss": 3.3007,
      "step": 9430
    },
    {
      "epoch": 1.3485714285714285,
      "grad_norm": 0.2853606939315796,
      "learning_rate": 1.1010476190476192e-05,
      "loss": 1.645,
      "step": 9440
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.2755785286426544,
      "learning_rate": 1.100095238095238e-05,
      "loss": 0.0064,
      "step": 9450
    },
    {
      "epoch": 1.3514285714285714,
      "grad_norm": 7033528.0,
      "learning_rate": 1.0991428571428573e-05,
      "loss": 4.5467,
      "step": 9460
    },
    {
      "epoch": 1.3528571428571428,
      "grad_norm": 0.09846706688404083,
      "learning_rate": 1.0981904761904762e-05,
      "loss": 2.5916,
      "step": 9470
    },
    {
      "epoch": 1.3542857142857143,
      "grad_norm": 0.08780983835458755,
      "learning_rate": 1.0972380952380954e-05,
      "loss": 4.2724,
      "step": 9480
    },
    {
      "epoch": 1.3557142857142856,
      "grad_norm": 0.2654746174812317,
      "learning_rate": 1.0962857142857143e-05,
      "loss": 0.0,
      "step": 9490
    },
    {
      "epoch": 1.3571428571428572,
      "grad_norm": 0.10320671647787094,
      "learning_rate": 1.0953333333333334e-05,
      "loss": 1.4394,
      "step": 9500
    },
    {
      "epoch": 1.3585714285714285,
      "grad_norm": 0.11850814521312714,
      "learning_rate": 1.0943809523809525e-05,
      "loss": 0.0,
      "step": 9510
    },
    {
      "epoch": 1.3599999999999999,
      "grad_norm": 0.05300489813089371,
      "learning_rate": 1.0934285714285715e-05,
      "loss": 0.0,
      "step": 9520
    },
    {
      "epoch": 1.3614285714285714,
      "grad_norm": 0.09572378545999527,
      "learning_rate": 1.0924761904761904e-05,
      "loss": 1.7299,
      "step": 9530
    },
    {
      "epoch": 1.362857142857143,
      "grad_norm": 0.27864083647727966,
      "learning_rate": 1.0915238095238097e-05,
      "loss": 0.7996,
      "step": 9540
    },
    {
      "epoch": 1.3642857142857143,
      "grad_norm": 0.053130242973566055,
      "learning_rate": 1.0905714285714286e-05,
      "loss": 3.3998,
      "step": 9550
    },
    {
      "epoch": 1.3657142857142857,
      "grad_norm": 0.08635593205690384,
      "learning_rate": 1.0896190476190478e-05,
      "loss": 0.0,
      "step": 9560
    },
    {
      "epoch": 1.3671428571428572,
      "grad_norm": 0.0628146082162857,
      "learning_rate": 1.0886666666666667e-05,
      "loss": 1.0156,
      "step": 9570
    },
    {
      "epoch": 1.3685714285714285,
      "grad_norm": 0.05476335808634758,
      "learning_rate": 1.0877142857142858e-05,
      "loss": 0.0,
      "step": 9580
    },
    {
      "epoch": 1.37,
      "grad_norm": 2730955.5,
      "learning_rate": 1.086761904761905e-05,
      "loss": 5.0477,
      "step": 9590
    },
    {
      "epoch": 1.3714285714285714,
      "grad_norm": 0.11138815432786942,
      "learning_rate": 1.0858095238095239e-05,
      "loss": 0.0,
      "step": 9600
    },
    {
      "epoch": 1.3728571428571428,
      "grad_norm": 0.10123737156391144,
      "learning_rate": 1.0848571428571431e-05,
      "loss": 3.3628,
      "step": 9610
    },
    {
      "epoch": 1.3742857142857143,
      "grad_norm": 0.2688646912574768,
      "learning_rate": 1.083904761904762e-05,
      "loss": 0.0,
      "step": 9620
    },
    {
      "epoch": 1.3757142857142857,
      "grad_norm": 2708468.5,
      "learning_rate": 1.0829523809523809e-05,
      "loss": 1.7138,
      "step": 9630
    },
    {
      "epoch": 1.3771428571428572,
      "grad_norm": 0.10289940237998962,
      "learning_rate": 1.0820000000000001e-05,
      "loss": 1.7179,
      "step": 9640
    },
    {
      "epoch": 1.3785714285714286,
      "grad_norm": 0.10597999393939972,
      "learning_rate": 1.081047619047619e-05,
      "loss": 3.3695,
      "step": 9650
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.26770731806755066,
      "learning_rate": 1.0800952380952381e-05,
      "loss": 3.3875,
      "step": 9660
    },
    {
      "epoch": 1.3814285714285715,
      "grad_norm": 0.08173449337482452,
      "learning_rate": 1.0791428571428573e-05,
      "loss": 1.7264,
      "step": 9670
    },
    {
      "epoch": 1.3828571428571428,
      "grad_norm": 0.10454538464546204,
      "learning_rate": 1.0781904761904762e-05,
      "loss": 3.3471,
      "step": 9680
    },
    {
      "epoch": 1.3842857142857143,
      "grad_norm": 0.10582536458969116,
      "learning_rate": 1.0772380952380955e-05,
      "loss": 0.0,
      "step": 9690
    },
    {
      "epoch": 1.3857142857142857,
      "grad_norm": 0.07889352738857269,
      "learning_rate": 1.0762857142857144e-05,
      "loss": 1.6205,
      "step": 9700
    },
    {
      "epoch": 1.387142857142857,
      "grad_norm": 0.2699141204357147,
      "learning_rate": 1.0753333333333333e-05,
      "loss": 0.0,
      "step": 9710
    },
    {
      "epoch": 1.3885714285714286,
      "grad_norm": 0.08278677612543106,
      "learning_rate": 1.0743809523809525e-05,
      "loss": 2.9465,
      "step": 9720
    },
    {
      "epoch": 1.3900000000000001,
      "grad_norm": 0.08412046730518341,
      "learning_rate": 1.0734285714285716e-05,
      "loss": 8.4142,
      "step": 9730
    },
    {
      "epoch": 1.3914285714285715,
      "grad_norm": 0.2662345767021179,
      "learning_rate": 1.0724761904761906e-05,
      "loss": 5.0325,
      "step": 9740
    },
    {
      "epoch": 1.3928571428571428,
      "grad_norm": 0.39172878861427307,
      "learning_rate": 1.0715238095238097e-05,
      "loss": 1.5191,
      "step": 9750
    },
    {
      "epoch": 1.3942857142857144,
      "grad_norm": 2753334.25,
      "learning_rate": 1.0705714285714286e-05,
      "loss": 1.6599,
      "step": 9760
    },
    {
      "epoch": 1.3957142857142857,
      "grad_norm": 0.05460904538631439,
      "learning_rate": 1.0696190476190478e-05,
      "loss": 4.9734,
      "step": 9770
    },
    {
      "epoch": 1.3971428571428572,
      "grad_norm": 0.3562275469303131,
      "learning_rate": 1.0686666666666667e-05,
      "loss": 1.6976,
      "step": 9780
    },
    {
      "epoch": 1.3985714285714286,
      "grad_norm": 0.294485479593277,
      "learning_rate": 1.0677142857142858e-05,
      "loss": 1.7568,
      "step": 9790
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.060677725821733475,
      "learning_rate": 1.0667619047619048e-05,
      "loss": 1.7144,
      "step": 9800
    },
    {
      "epoch": 1.4014285714285715,
      "grad_norm": 0.27143120765686035,
      "learning_rate": 1.0658095238095239e-05,
      "loss": 1.6438,
      "step": 9810
    },
    {
      "epoch": 1.4028571428571428,
      "grad_norm": 0.3038036525249481,
      "learning_rate": 1.064857142857143e-05,
      "loss": 1.6369,
      "step": 9820
    },
    {
      "epoch": 1.4042857142857144,
      "grad_norm": 0.09655313193798065,
      "learning_rate": 1.063904761904762e-05,
      "loss": 3.2169,
      "step": 9830
    },
    {
      "epoch": 1.4057142857142857,
      "grad_norm": 2712677.75,
      "learning_rate": 1.062952380952381e-05,
      "loss": 1.703,
      "step": 9840
    },
    {
      "epoch": 1.407142857142857,
      "grad_norm": 39904204.0,
      "learning_rate": 1.0620000000000002e-05,
      "loss": 3.2809,
      "step": 9850
    },
    {
      "epoch": 1.4085714285714286,
      "grad_norm": 0.2771344780921936,
      "learning_rate": 1.061047619047619e-05,
      "loss": 3.2498,
      "step": 9860
    },
    {
      "epoch": 1.41,
      "grad_norm": 2707100.75,
      "learning_rate": 1.0600952380952383e-05,
      "loss": 3.29,
      "step": 9870
    },
    {
      "epoch": 1.4114285714285715,
      "grad_norm": 2767808.0,
      "learning_rate": 1.0591428571428572e-05,
      "loss": 1.6274,
      "step": 9880
    },
    {
      "epoch": 1.4128571428571428,
      "grad_norm": 0.12089500576257706,
      "learning_rate": 1.0581904761904763e-05,
      "loss": 4.9311,
      "step": 9890
    },
    {
      "epoch": 1.4142857142857144,
      "grad_norm": 0.09278760850429535,
      "learning_rate": 1.0572380952380953e-05,
      "loss": 0.0,
      "step": 9900
    },
    {
      "epoch": 1.4157142857142857,
      "grad_norm": 0.26354560256004333,
      "learning_rate": 1.0562857142857144e-05,
      "loss": 0.0,
      "step": 9910
    },
    {
      "epoch": 1.4171428571428573,
      "grad_norm": 2781253.75,
      "learning_rate": 1.0553333333333333e-05,
      "loss": 1.6931,
      "step": 9920
    },
    {
      "epoch": 1.4185714285714286,
      "grad_norm": 0.29739901423454285,
      "learning_rate": 1.0543809523809525e-05,
      "loss": 3.4094,
      "step": 9930
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.3024948835372925,
      "learning_rate": 1.0534285714285714e-05,
      "loss": 3.2903,
      "step": 9940
    },
    {
      "epoch": 1.4214285714285715,
      "grad_norm": 0.3286880850791931,
      "learning_rate": 1.0524761904761906e-05,
      "loss": 4.9201,
      "step": 9950
    },
    {
      "epoch": 1.4228571428571428,
      "grad_norm": 0.2658393383026123,
      "learning_rate": 1.0515238095238095e-05,
      "loss": 3.2615,
      "step": 9960
    },
    {
      "epoch": 1.4242857142857144,
      "grad_norm": 0.05200805142521858,
      "learning_rate": 1.0505714285714286e-05,
      "loss": 3.226,
      "step": 9970
    },
    {
      "epoch": 1.4257142857142857,
      "grad_norm": 0.27344799041748047,
      "learning_rate": 1.0496190476190478e-05,
      "loss": 1.668,
      "step": 9980
    },
    {
      "epoch": 1.427142857142857,
      "grad_norm": 0.2534730136394501,
      "learning_rate": 1.0486666666666667e-05,
      "loss": 3.2739,
      "step": 9990
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 0.31795477867126465,
      "learning_rate": 1.047714285714286e-05,
      "loss": 0.0,
      "step": 10000
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.26867735385894775,
      "learning_rate": 1.0467619047619049e-05,
      "loss": 1.6451,
      "step": 10010
    },
    {
      "epoch": 1.4314285714285715,
      "grad_norm": 0.08070123195648193,
      "learning_rate": 1.0458095238095238e-05,
      "loss": 3.2891,
      "step": 10020
    },
    {
      "epoch": 1.4328571428571428,
      "grad_norm": 2731371.25,
      "learning_rate": 1.044857142857143e-05,
      "loss": 1.6419,
      "step": 10030
    },
    {
      "epoch": 1.4342857142857142,
      "grad_norm": 0.2641814351081848,
      "learning_rate": 1.043904761904762e-05,
      "loss": 1.6313,
      "step": 10040
    },
    {
      "epoch": 1.4357142857142857,
      "grad_norm": 0.056718435138463974,
      "learning_rate": 1.042952380952381e-05,
      "loss": 1.6379,
      "step": 10050
    },
    {
      "epoch": 1.4371428571428573,
      "grad_norm": 0.2766065299510956,
      "learning_rate": 1.0420000000000002e-05,
      "loss": 3.3978,
      "step": 10060
    },
    {
      "epoch": 1.4385714285714286,
      "grad_norm": 0.08112350106239319,
      "learning_rate": 1.041047619047619e-05,
      "loss": 1.7308,
      "step": 10070
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.2616097033023834,
      "learning_rate": 1.0400952380952383e-05,
      "loss": 3.2874,
      "step": 10080
    },
    {
      "epoch": 1.4414285714285715,
      "grad_norm": 2713811.75,
      "learning_rate": 1.0391428571428572e-05,
      "loss": 3.1964,
      "step": 10090
    },
    {
      "epoch": 1.4428571428571428,
      "grad_norm": 0.0946865901350975,
      "learning_rate": 1.0381904761904763e-05,
      "loss": 3.2559,
      "step": 10100
    },
    {
      "epoch": 1.4442857142857144,
      "grad_norm": 0.09109173715114594,
      "learning_rate": 1.0372380952380953e-05,
      "loss": 1.6641,
      "step": 10110
    },
    {
      "epoch": 1.4457142857142857,
      "grad_norm": 0.27739623188972473,
      "learning_rate": 1.0362857142857144e-05,
      "loss": 3.4267,
      "step": 10120
    },
    {
      "epoch": 1.447142857142857,
      "grad_norm": 2958976.5,
      "learning_rate": 1.0353333333333335e-05,
      "loss": 3.245,
      "step": 10130
    },
    {
      "epoch": 1.4485714285714286,
      "grad_norm": 0.2762155532836914,
      "learning_rate": 1.0343809523809525e-05,
      "loss": 4.9211,
      "step": 10140
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.28874868154525757,
      "learning_rate": 1.0334285714285714e-05,
      "loss": 0.0,
      "step": 10150
    },
    {
      "epoch": 1.4514285714285715,
      "grad_norm": 0.308319628238678,
      "learning_rate": 1.0324761904761907e-05,
      "loss": 1.6631,
      "step": 10160
    },
    {
      "epoch": 1.4528571428571428,
      "grad_norm": 0.317165344953537,
      "learning_rate": 1.0315238095238096e-05,
      "loss": 3.3344,
      "step": 10170
    },
    {
      "epoch": 1.4542857142857142,
      "grad_norm": 0.08171477168798447,
      "learning_rate": 1.0305714285714286e-05,
      "loss": 3.3422,
      "step": 10180
    },
    {
      "epoch": 1.4557142857142857,
      "grad_norm": 0.3036486804485321,
      "learning_rate": 1.0296190476190477e-05,
      "loss": 3.2622,
      "step": 10190
    },
    {
      "epoch": 1.457142857142857,
      "grad_norm": 0.2635881006717682,
      "learning_rate": 1.0286666666666668e-05,
      "loss": 1.6914,
      "step": 10200
    },
    {
      "epoch": 1.4585714285714286,
      "grad_norm": 0.10043521225452423,
      "learning_rate": 1.0277142857142858e-05,
      "loss": 3.3638,
      "step": 10210
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.07477544993162155,
      "learning_rate": 1.0267619047619049e-05,
      "loss": 3.3052,
      "step": 10220
    },
    {
      "epoch": 1.4614285714285713,
      "grad_norm": 2667536.25,
      "learning_rate": 1.0258095238095238e-05,
      "loss": 4.8246,
      "step": 10230
    },
    {
      "epoch": 1.4628571428571429,
      "grad_norm": 2786455.5,
      "learning_rate": 1.024857142857143e-05,
      "loss": 6.7562,
      "step": 10240
    },
    {
      "epoch": 1.4642857142857144,
      "grad_norm": 0.11023322492837906,
      "learning_rate": 1.0239047619047619e-05,
      "loss": 3.1375,
      "step": 10250
    },
    {
      "epoch": 1.4657142857142857,
      "grad_norm": 0.27548524737358093,
      "learning_rate": 1.0229523809523811e-05,
      "loss": 4.7676,
      "step": 10260
    },
    {
      "epoch": 1.467142857142857,
      "grad_norm": 0.3202376663684845,
      "learning_rate": 1.022e-05,
      "loss": 1.635,
      "step": 10270
    },
    {
      "epoch": 1.4685714285714286,
      "grad_norm": 0.2710201144218445,
      "learning_rate": 1.0210476190476191e-05,
      "loss": 3.1309,
      "step": 10280
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.09601007401943207,
      "learning_rate": 1.0200952380952382e-05,
      "loss": 1.5473,
      "step": 10290
    },
    {
      "epoch": 1.4714285714285715,
      "grad_norm": 0.2820023000240326,
      "learning_rate": 1.0191428571428572e-05,
      "loss": 4.5161,
      "step": 10300
    },
    {
      "epoch": 1.4728571428571429,
      "grad_norm": 0.06702522933483124,
      "learning_rate": 1.0181904761904761e-05,
      "loss": 1.3664,
      "step": 10310
    },
    {
      "epoch": 1.4742857142857142,
      "grad_norm": 0.06650152802467346,
      "learning_rate": 1.0172380952380954e-05,
      "loss": 1.6613,
      "step": 10320
    },
    {
      "epoch": 1.4757142857142858,
      "grad_norm": 0.2606026828289032,
      "learning_rate": 1.0162857142857143e-05,
      "loss": 3.1784,
      "step": 10330
    },
    {
      "epoch": 1.477142857142857,
      "grad_norm": 2671037.0,
      "learning_rate": 1.0153333333333335e-05,
      "loss": 4.5592,
      "step": 10340
    },
    {
      "epoch": 1.4785714285714286,
      "grad_norm": 0.6079274415969849,
      "learning_rate": 1.0143809523809524e-05,
      "loss": 3.2082,
      "step": 10350
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.07919321209192276,
      "learning_rate": 1.0134285714285715e-05,
      "loss": 1.6181,
      "step": 10360
    },
    {
      "epoch": 1.4814285714285713,
      "grad_norm": 0.0897681787610054,
      "learning_rate": 1.0124761904761907e-05,
      "loss": 4.8578,
      "step": 10370
    },
    {
      "epoch": 1.4828571428571429,
      "grad_norm": 0.2620407044887543,
      "learning_rate": 1.0115238095238096e-05,
      "loss": 3.0442,
      "step": 10380
    },
    {
      "epoch": 1.4842857142857142,
      "grad_norm": 0.2619856297969818,
      "learning_rate": 1.0105714285714288e-05,
      "loss": 3.1243,
      "step": 10390
    },
    {
      "epoch": 1.4857142857142858,
      "grad_norm": 0.04875782132148743,
      "learning_rate": 1.0096190476190477e-05,
      "loss": 0.0,
      "step": 10400
    },
    {
      "epoch": 1.487142857142857,
      "grad_norm": 3411169.5,
      "learning_rate": 1.0086666666666666e-05,
      "loss": 1.4949,
      "step": 10410
    },
    {
      "epoch": 1.4885714285714284,
      "grad_norm": 0.05112254247069359,
      "learning_rate": 1.0077142857142858e-05,
      "loss": 3.3021,
      "step": 10420
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.3294416069984436,
      "learning_rate": 1.0067619047619049e-05,
      "loss": 0.0,
      "step": 10430
    },
    {
      "epoch": 1.4914285714285715,
      "grad_norm": 0.06153273954987526,
      "learning_rate": 1.0058095238095238e-05,
      "loss": 1.6716,
      "step": 10440
    },
    {
      "epoch": 1.4928571428571429,
      "grad_norm": 0.10072437673807144,
      "learning_rate": 1.004857142857143e-05,
      "loss": 3.1545,
      "step": 10450
    },
    {
      "epoch": 1.4942857142857142,
      "grad_norm": 0.0685887560248375,
      "learning_rate": 1.003904761904762e-05,
      "loss": 3.2866,
      "step": 10460
    },
    {
      "epoch": 1.4957142857142858,
      "grad_norm": 2676873.5,
      "learning_rate": 1.0029523809523812e-05,
      "loss": 1.6319,
      "step": 10470
    },
    {
      "epoch": 1.497142857142857,
      "grad_norm": 0.05984823405742645,
      "learning_rate": 1.002e-05,
      "loss": 1.6442,
      "step": 10480
    },
    {
      "epoch": 1.4985714285714287,
      "grad_norm": 0.058897506445646286,
      "learning_rate": 1.0010476190476191e-05,
      "loss": 6.632,
      "step": 10490
    },
    {
      "epoch": 1.5,
      "grad_norm": 2642462.75,
      "learning_rate": 1.0000952380952382e-05,
      "loss": 1.6322,
      "step": 10500
    },
    {
      "epoch": 1.5014285714285713,
      "grad_norm": 0.30095428228378296,
      "learning_rate": 9.991428571428573e-06,
      "loss": 0.0,
      "step": 10510
    },
    {
      "epoch": 1.502857142857143,
      "grad_norm": 2681716.5,
      "learning_rate": 9.981904761904763e-06,
      "loss": 4.7128,
      "step": 10520
    },
    {
      "epoch": 1.5042857142857144,
      "grad_norm": 0.26892247796058655,
      "learning_rate": 9.972380952380954e-06,
      "loss": 0.0,
      "step": 10530
    },
    {
      "epoch": 1.5057142857142858,
      "grad_norm": 0.03928595781326294,
      "learning_rate": 9.962857142857145e-06,
      "loss": 1.5848,
      "step": 10540
    },
    {
      "epoch": 1.5071428571428571,
      "grad_norm": 0.08657500147819519,
      "learning_rate": 9.953333333333333e-06,
      "loss": 1.6292,
      "step": 10550
    },
    {
      "epoch": 1.5085714285714285,
      "grad_norm": 0.2712164521217346,
      "learning_rate": 9.943809523809524e-06,
      "loss": 1.6508,
      "step": 10560
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.3049302399158478,
      "learning_rate": 9.934285714285715e-06,
      "loss": 0.0,
      "step": 10570
    },
    {
      "epoch": 1.5114285714285716,
      "grad_norm": 2713888.5,
      "learning_rate": 9.924761904761905e-06,
      "loss": 6.5702,
      "step": 10580
    },
    {
      "epoch": 1.512857142857143,
      "grad_norm": 0.2597413659095764,
      "learning_rate": 9.915238095238096e-06,
      "loss": 2.9842,
      "step": 10590
    },
    {
      "epoch": 1.5142857142857142,
      "grad_norm": 0.29552143812179565,
      "learning_rate": 9.905714285714287e-06,
      "loss": 4.7169,
      "step": 10600
    },
    {
      "epoch": 1.5157142857142856,
      "grad_norm": 0.07085546851158142,
      "learning_rate": 9.896190476190477e-06,
      "loss": 1.5911,
      "step": 10610
    },
    {
      "epoch": 1.5171428571428571,
      "grad_norm": 0.2682349681854248,
      "learning_rate": 9.886666666666668e-06,
      "loss": 1.6477,
      "step": 10620
    },
    {
      "epoch": 1.5185714285714287,
      "grad_norm": 0.08038995414972305,
      "learning_rate": 9.877142857142859e-06,
      "loss": 3.3006,
      "step": 10630
    },
    {
      "epoch": 1.52,
      "grad_norm": 3025682.25,
      "learning_rate": 9.867619047619048e-06,
      "loss": 3.232,
      "step": 10640
    },
    {
      "epoch": 1.5214285714285714,
      "grad_norm": 0.27009913325309753,
      "learning_rate": 9.858095238095238e-06,
      "loss": 1.5048,
      "step": 10650
    },
    {
      "epoch": 1.522857142857143,
      "grad_norm": 0.08110564947128296,
      "learning_rate": 9.848571428571429e-06,
      "loss": 1.6755,
      "step": 10660
    },
    {
      "epoch": 1.5242857142857142,
      "grad_norm": 0.2743569016456604,
      "learning_rate": 9.83904761904762e-06,
      "loss": 3.232,
      "step": 10670
    },
    {
      "epoch": 1.5257142857142858,
      "grad_norm": 0.10605532675981522,
      "learning_rate": 9.82952380952381e-06,
      "loss": 0.0,
      "step": 10680
    },
    {
      "epoch": 1.5271428571428571,
      "grad_norm": 0.06257695704698563,
      "learning_rate": 9.820000000000001e-06,
      "loss": 5.0636,
      "step": 10690
    },
    {
      "epoch": 1.5285714285714285,
      "grad_norm": 0.27065590023994446,
      "learning_rate": 9.810476190476191e-06,
      "loss": 0.0,
      "step": 10700
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.06909879297018051,
      "learning_rate": 9.800952380952382e-06,
      "loss": 1.7121,
      "step": 10710
    },
    {
      "epoch": 1.5314285714285716,
      "grad_norm": 0.11823401600122452,
      "learning_rate": 9.791428571428571e-06,
      "loss": 0.0,
      "step": 10720
    },
    {
      "epoch": 1.532857142857143,
      "grad_norm": 0.0677570030093193,
      "learning_rate": 9.781904761904762e-06,
      "loss": 0.0,
      "step": 10730
    },
    {
      "epoch": 1.5342857142857143,
      "grad_norm": 0.0697219967842102,
      "learning_rate": 9.772380952380952e-06,
      "loss": 3.4049,
      "step": 10740
    },
    {
      "epoch": 1.5357142857142856,
      "grad_norm": 0.08251184225082397,
      "learning_rate": 9.762857142857145e-06,
      "loss": 1.7132,
      "step": 10750
    },
    {
      "epoch": 1.5371428571428571,
      "grad_norm": 0.06704151630401611,
      "learning_rate": 9.753333333333335e-06,
      "loss": 4.9368,
      "step": 10760
    },
    {
      "epoch": 1.5385714285714287,
      "grad_norm": 0.2837602198123932,
      "learning_rate": 9.743809523809524e-06,
      "loss": 1.6555,
      "step": 10770
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.10235035419464111,
      "learning_rate": 9.734285714285715e-06,
      "loss": 5.0211,
      "step": 10780
    },
    {
      "epoch": 1.5414285714285714,
      "grad_norm": 2666126.25,
      "learning_rate": 9.724761904761906e-06,
      "loss": 1.7125,
      "step": 10790
    },
    {
      "epoch": 1.5428571428571427,
      "grad_norm": 0.2633870542049408,
      "learning_rate": 9.715238095238096e-06,
      "loss": 3.3516,
      "step": 10800
    },
    {
      "epoch": 1.5442857142857143,
      "grad_norm": 0.25902074575424194,
      "learning_rate": 9.705714285714287e-06,
      "loss": 0.0,
      "step": 10810
    },
    {
      "epoch": 1.5457142857142858,
      "grad_norm": 0.27103033661842346,
      "learning_rate": 9.696190476190478e-06,
      "loss": 1.6591,
      "step": 10820
    },
    {
      "epoch": 1.5471428571428572,
      "grad_norm": 0.09405016154050827,
      "learning_rate": 9.686666666666668e-06,
      "loss": 3.3324,
      "step": 10830
    },
    {
      "epoch": 1.5485714285714285,
      "grad_norm": 0.09749303758144379,
      "learning_rate": 9.677142857142859e-06,
      "loss": 3.2968,
      "step": 10840
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.10208931565284729,
      "learning_rate": 9.667619047619048e-06,
      "loss": 3.2748,
      "step": 10850
    },
    {
      "epoch": 1.5514285714285714,
      "grad_norm": 0.2785692512989044,
      "learning_rate": 9.658095238095238e-06,
      "loss": 1.6322,
      "step": 10860
    },
    {
      "epoch": 1.552857142857143,
      "grad_norm": 0.32484328746795654,
      "learning_rate": 9.648571428571429e-06,
      "loss": 1.6536,
      "step": 10870
    },
    {
      "epoch": 1.5542857142857143,
      "grad_norm": 0.06453529000282288,
      "learning_rate": 9.63904761904762e-06,
      "loss": 4.9877,
      "step": 10880
    },
    {
      "epoch": 1.5557142857142856,
      "grad_norm": 0.2611559331417084,
      "learning_rate": 9.62952380952381e-06,
      "loss": 3.3223,
      "step": 10890
    },
    {
      "epoch": 1.5571428571428572,
      "grad_norm": 0.3171072006225586,
      "learning_rate": 9.620000000000001e-06,
      "loss": 1.5969,
      "step": 10900
    },
    {
      "epoch": 1.5585714285714287,
      "grad_norm": 0.0565175861120224,
      "learning_rate": 9.610476190476192e-06,
      "loss": 1.6092,
      "step": 10910
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.3171681761741638,
      "learning_rate": 9.600952380952382e-06,
      "loss": 8.1145,
      "step": 10920
    },
    {
      "epoch": 1.5614285714285714,
      "grad_norm": 0.07752358913421631,
      "learning_rate": 9.591428571428573e-06,
      "loss": 4.7216,
      "step": 10930
    },
    {
      "epoch": 1.5628571428571427,
      "grad_norm": 0.27256450057029724,
      "learning_rate": 9.581904761904762e-06,
      "loss": 1.5947,
      "step": 10940
    },
    {
      "epoch": 1.5642857142857143,
      "grad_norm": 0.06036011502146721,
      "learning_rate": 9.572380952380953e-06,
      "loss": 4.7386,
      "step": 10950
    },
    {
      "epoch": 1.5657142857142858,
      "grad_norm": 0.09984336048364639,
      "learning_rate": 9.562857142857143e-06,
      "loss": 3.1955,
      "step": 10960
    },
    {
      "epoch": 1.5671428571428572,
      "grad_norm": 0.37131956219673157,
      "learning_rate": 9.553333333333334e-06,
      "loss": 0.0,
      "step": 10970
    },
    {
      "epoch": 1.5685714285714285,
      "grad_norm": 0.40252453088760376,
      "learning_rate": 9.543809523809525e-06,
      "loss": 3.1284,
      "step": 10980
    },
    {
      "epoch": 1.5699999999999998,
      "grad_norm": 0.09780517220497131,
      "learning_rate": 9.534285714285715e-06,
      "loss": 0.0,
      "step": 10990
    },
    {
      "epoch": 1.5714285714285714,
      "grad_norm": 0.3709656894207001,
      "learning_rate": 9.524761904761906e-06,
      "loss": 0.0,
      "step": 11000
    },
    {
      "epoch": 1.572857142857143,
      "grad_norm": 0.03868282958865166,
      "learning_rate": 9.515238095238097e-06,
      "loss": 4.8826,
      "step": 11010
    },
    {
      "epoch": 1.5742857142857143,
      "grad_norm": 0.3405112326145172,
      "learning_rate": 9.505714285714287e-06,
      "loss": 0.0,
      "step": 11020
    },
    {
      "epoch": 1.5757142857142856,
      "grad_norm": 0.2938917279243469,
      "learning_rate": 9.496190476190476e-06,
      "loss": 1.6296,
      "step": 11030
    },
    {
      "epoch": 1.5771428571428572,
      "grad_norm": 0.2774594724178314,
      "learning_rate": 9.486666666666667e-06,
      "loss": 0.0,
      "step": 11040
    },
    {
      "epoch": 1.5785714285714287,
      "grad_norm": 0.13591164350509644,
      "learning_rate": 9.477142857142857e-06,
      "loss": 1.7018,
      "step": 11050
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.15303824841976166,
      "learning_rate": 9.467619047619048e-06,
      "loss": 1.6693,
      "step": 11060
    },
    {
      "epoch": 1.5814285714285714,
      "grad_norm": 0.05637514591217041,
      "learning_rate": 9.458095238095239e-06,
      "loss": 3.2566,
      "step": 11070
    },
    {
      "epoch": 1.5828571428571427,
      "grad_norm": 0.36354464292526245,
      "learning_rate": 9.44857142857143e-06,
      "loss": 3.1413,
      "step": 11080
    },
    {
      "epoch": 1.5842857142857143,
      "grad_norm": 2769474.0,
      "learning_rate": 9.43904761904762e-06,
      "loss": 6.5301,
      "step": 11090
    },
    {
      "epoch": 1.5857142857142859,
      "grad_norm": 0.040934786200523376,
      "learning_rate": 9.42952380952381e-06,
      "loss": 1.527,
      "step": 11100
    },
    {
      "epoch": 1.5871428571428572,
      "grad_norm": 0.07159456610679626,
      "learning_rate": 9.42e-06,
      "loss": 3.2047,
      "step": 11110
    },
    {
      "epoch": 1.5885714285714285,
      "grad_norm": 0.38398194313049316,
      "learning_rate": 9.41047619047619e-06,
      "loss": 3.1988,
      "step": 11120
    },
    {
      "epoch": 1.5899999999999999,
      "grad_norm": 2702241.5,
      "learning_rate": 9.400952380952381e-06,
      "loss": 8.0312,
      "step": 11130
    },
    {
      "epoch": 1.5914285714285714,
      "grad_norm": 2674230.25,
      "learning_rate": 9.391428571428573e-06,
      "loss": 7.9047,
      "step": 11140
    },
    {
      "epoch": 1.592857142857143,
      "grad_norm": 2656620.5,
      "learning_rate": 9.381904761904764e-06,
      "loss": 7.6423,
      "step": 11150
    },
    {
      "epoch": 1.5942857142857143,
      "grad_norm": 2660058.0,
      "learning_rate": 9.372380952380953e-06,
      "loss": 2.9715,
      "step": 11160
    },
    {
      "epoch": 1.5957142857142856,
      "grad_norm": 0.724397599697113,
      "learning_rate": 9.362857142857143e-06,
      "loss": 4.5151,
      "step": 11170
    },
    {
      "epoch": 1.5971428571428572,
      "grad_norm": 0.8835497498512268,
      "learning_rate": 9.353333333333334e-06,
      "loss": 1.5566,
      "step": 11180
    },
    {
      "epoch": 1.5985714285714285,
      "grad_norm": 0.8843017220497131,
      "learning_rate": 9.343809523809525e-06,
      "loss": 3.0691,
      "step": 11190
    },
    {
      "epoch": 1.6,
      "grad_norm": 2867607.25,
      "learning_rate": 9.334285714285715e-06,
      "loss": 3.2367,
      "step": 11200
    },
    {
      "epoch": 1.6014285714285714,
      "grad_norm": 2640246.25,
      "learning_rate": 9.324761904761906e-06,
      "loss": 4.7148,
      "step": 11210
    },
    {
      "epoch": 1.6028571428571428,
      "grad_norm": 0.5753086805343628,
      "learning_rate": 9.315238095238097e-06,
      "loss": 4.6919,
      "step": 11220
    },
    {
      "epoch": 1.6042857142857143,
      "grad_norm": 0.5726458430290222,
      "learning_rate": 9.305714285714287e-06,
      "loss": 4.7807,
      "step": 11230
    },
    {
      "epoch": 1.6057142857142859,
      "grad_norm": 0.5944982767105103,
      "learning_rate": 9.296190476190476e-06,
      "loss": 3.0359,
      "step": 11240
    },
    {
      "epoch": 1.6071428571428572,
      "grad_norm": 0.3677980303764343,
      "learning_rate": 9.286666666666667e-06,
      "loss": 3.0885,
      "step": 11250
    },
    {
      "epoch": 1.6085714285714285,
      "grad_norm": 0.32756364345550537,
      "learning_rate": 9.277142857142858e-06,
      "loss": 1.5781,
      "step": 11260
    },
    {
      "epoch": 1.6099999999999999,
      "grad_norm": 0.3306446671485901,
      "learning_rate": 9.267619047619048e-06,
      "loss": 3.2742,
      "step": 11270
    },
    {
      "epoch": 1.6114285714285714,
      "grad_norm": 0.06524766236543655,
      "learning_rate": 9.258095238095239e-06,
      "loss": 4.7407,
      "step": 11280
    },
    {
      "epoch": 1.612857142857143,
      "grad_norm": 0.3035062551498413,
      "learning_rate": 9.24857142857143e-06,
      "loss": 4.7347,
      "step": 11290
    },
    {
      "epoch": 1.6142857142857143,
      "grad_norm": 0.5922884345054626,
      "learning_rate": 9.23904761904762e-06,
      "loss": 3.1709,
      "step": 11300
    },
    {
      "epoch": 1.6157142857142857,
      "grad_norm": 0.2667209208011627,
      "learning_rate": 9.229523809523811e-06,
      "loss": 4.6432,
      "step": 11310
    },
    {
      "epoch": 1.617142857142857,
      "grad_norm": 0.2659376859664917,
      "learning_rate": 9.220000000000002e-06,
      "loss": 3.303,
      "step": 11320
    },
    {
      "epoch": 1.6185714285714285,
      "grad_norm": 0.37805140018463135,
      "learning_rate": 9.21047619047619e-06,
      "loss": 6.2364,
      "step": 11330
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.072052001953125,
      "learning_rate": 9.200952380952381e-06,
      "loss": 3.1294,
      "step": 11340
    },
    {
      "epoch": 1.6214285714285714,
      "grad_norm": 0.09726820886135101,
      "learning_rate": 9.191428571428572e-06,
      "loss": 0.0,
      "step": 11350
    },
    {
      "epoch": 1.6228571428571428,
      "grad_norm": 0.3344334065914154,
      "learning_rate": 9.181904761904762e-06,
      "loss": 0.0,
      "step": 11360
    },
    {
      "epoch": 1.6242857142857143,
      "grad_norm": 0.04846709594130516,
      "learning_rate": 9.172380952380953e-06,
      "loss": 1.5919,
      "step": 11370
    },
    {
      "epoch": 1.6257142857142857,
      "grad_norm": 0.33496832847595215,
      "learning_rate": 9.162857142857144e-06,
      "loss": 3.2341,
      "step": 11380
    },
    {
      "epoch": 1.6271428571428572,
      "grad_norm": 2682696.0,
      "learning_rate": 9.153333333333334e-06,
      "loss": 3.2106,
      "step": 11390
    },
    {
      "epoch": 1.6285714285714286,
      "grad_norm": 0.04830352962017059,
      "learning_rate": 9.143809523809525e-06,
      "loss": 6.4007,
      "step": 11400
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.05872783809900284,
      "learning_rate": 9.134285714285716e-06,
      "loss": 3.1665,
      "step": 11410
    },
    {
      "epoch": 1.6314285714285715,
      "grad_norm": 0.3770410716533661,
      "learning_rate": 9.124761904761905e-06,
      "loss": 8.0562,
      "step": 11420
    },
    {
      "epoch": 1.632857142857143,
      "grad_norm": 0.37001657485961914,
      "learning_rate": 9.115238095238095e-06,
      "loss": 4.6637,
      "step": 11430
    },
    {
      "epoch": 1.6342857142857143,
      "grad_norm": 0.05787251144647598,
      "learning_rate": 9.105714285714286e-06,
      "loss": 0.0,
      "step": 11440
    },
    {
      "epoch": 1.6357142857142857,
      "grad_norm": 0.312743604183197,
      "learning_rate": 9.096190476190477e-06,
      "loss": 1.5346,
      "step": 11450
    },
    {
      "epoch": 1.637142857142857,
      "grad_norm": 0.3573249280452728,
      "learning_rate": 9.086666666666667e-06,
      "loss": 3.2194,
      "step": 11460
    },
    {
      "epoch": 1.6385714285714286,
      "grad_norm": 0.10668041557073593,
      "learning_rate": 9.077142857142858e-06,
      "loss": 1.5438,
      "step": 11470
    },
    {
      "epoch": 1.6400000000000001,
      "grad_norm": 0.3361459970474243,
      "learning_rate": 9.067619047619048e-06,
      "loss": 3.1392,
      "step": 11480
    },
    {
      "epoch": 1.6414285714285715,
      "grad_norm": 0.07028315961360931,
      "learning_rate": 9.058095238095239e-06,
      "loss": 3.2891,
      "step": 11490
    },
    {
      "epoch": 1.6428571428571428,
      "grad_norm": 0.26114028692245483,
      "learning_rate": 9.048571428571428e-06,
      "loss": 0.0,
      "step": 11500
    },
    {
      "epoch": 1.6442857142857141,
      "grad_norm": 0.563707172870636,
      "learning_rate": 9.039047619047619e-06,
      "loss": 1.5707,
      "step": 11510
    },
    {
      "epoch": 1.6457142857142857,
      "grad_norm": 0.28197112679481506,
      "learning_rate": 9.029523809523811e-06,
      "loss": 4.9381,
      "step": 11520
    },
    {
      "epoch": 1.6471428571428572,
      "grad_norm": 0.051052261143922806,
      "learning_rate": 9.020000000000002e-06,
      "loss": 3.188,
      "step": 11530
    },
    {
      "epoch": 1.6485714285714286,
      "grad_norm": 0.10858014971017838,
      "learning_rate": 9.010476190476192e-06,
      "loss": 4.9463,
      "step": 11540
    },
    {
      "epoch": 1.65,
      "grad_norm": 2683343.0,
      "learning_rate": 9.000952380952381e-06,
      "loss": 3.1929,
      "step": 11550
    },
    {
      "epoch": 1.6514285714285715,
      "grad_norm": 0.29845935106277466,
      "learning_rate": 8.991428571428572e-06,
      "loss": 3.1048,
      "step": 11560
    },
    {
      "epoch": 1.6528571428571428,
      "grad_norm": 0.06873109936714172,
      "learning_rate": 8.981904761904763e-06,
      "loss": 4.8623,
      "step": 11570
    },
    {
      "epoch": 1.6542857142857144,
      "grad_norm": 0.30705177783966064,
      "learning_rate": 8.972380952380953e-06,
      "loss": 1.5586,
      "step": 11580
    },
    {
      "epoch": 1.6557142857142857,
      "grad_norm": 0.30086109042167664,
      "learning_rate": 8.962857142857144e-06,
      "loss": 0.0,
      "step": 11590
    },
    {
      "epoch": 1.657142857142857,
      "grad_norm": 0.2873622179031372,
      "learning_rate": 8.953333333333335e-06,
      "loss": 1.5877,
      "step": 11600
    },
    {
      "epoch": 1.6585714285714286,
      "grad_norm": 0.03723100200295448,
      "learning_rate": 8.943809523809525e-06,
      "loss": 6.5836,
      "step": 11610
    },
    {
      "epoch": 1.6600000000000001,
      "grad_norm": 0.3903443515300751,
      "learning_rate": 8.934285714285716e-06,
      "loss": 1.5757,
      "step": 11620
    },
    {
      "epoch": 1.6614285714285715,
      "grad_norm": 0.5677575469017029,
      "learning_rate": 8.924761904761905e-06,
      "loss": 3.183,
      "step": 11630
    },
    {
      "epoch": 1.6628571428571428,
      "grad_norm": 0.3414389491081238,
      "learning_rate": 8.915238095238095e-06,
      "loss": 3.1337,
      "step": 11640
    },
    {
      "epoch": 1.6642857142857141,
      "grad_norm": 2675426.25,
      "learning_rate": 8.905714285714286e-06,
      "loss": 7.9987,
      "step": 11650
    },
    {
      "epoch": 1.6657142857142857,
      "grad_norm": 0.29813891649246216,
      "learning_rate": 8.896190476190477e-06,
      "loss": 1.6998,
      "step": 11660
    },
    {
      "epoch": 1.6671428571428573,
      "grad_norm": 0.31175410747528076,
      "learning_rate": 8.886666666666667e-06,
      "loss": 3.2234,
      "step": 11670
    },
    {
      "epoch": 1.6685714285714286,
      "grad_norm": 0.07375010848045349,
      "learning_rate": 8.877142857142858e-06,
      "loss": 4.7543,
      "step": 11680
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.3406822681427002,
      "learning_rate": 8.867619047619049e-06,
      "loss": 3.162,
      "step": 11690
    },
    {
      "epoch": 1.6714285714285713,
      "grad_norm": 0.06374247372150421,
      "learning_rate": 8.85809523809524e-06,
      "loss": 0.0,
      "step": 11700
    },
    {
      "epoch": 1.6728571428571428,
      "grad_norm": 0.05138038098812103,
      "learning_rate": 8.84857142857143e-06,
      "loss": 1.5096,
      "step": 11710
    },
    {
      "epoch": 1.6742857142857144,
      "grad_norm": 0.049153972417116165,
      "learning_rate": 8.839047619047619e-06,
      "loss": 3.2054,
      "step": 11720
    },
    {
      "epoch": 1.6757142857142857,
      "grad_norm": 0.42527759075164795,
      "learning_rate": 8.82952380952381e-06,
      "loss": 0.0,
      "step": 11730
    },
    {
      "epoch": 1.677142857142857,
      "grad_norm": 0.3604915738105774,
      "learning_rate": 8.82e-06,
      "loss": 1.7411,
      "step": 11740
    },
    {
      "epoch": 1.6785714285714286,
      "grad_norm": 0.368735134601593,
      "learning_rate": 8.810476190476191e-06,
      "loss": 1.5477,
      "step": 11750
    },
    {
      "epoch": 1.6800000000000002,
      "grad_norm": 0.05079406872391701,
      "learning_rate": 8.800952380952382e-06,
      "loss": 0.0,
      "step": 11760
    },
    {
      "epoch": 1.6814285714285715,
      "grad_norm": 0.09589863568544388,
      "learning_rate": 8.791428571428572e-06,
      "loss": 0.0,
      "step": 11770
    },
    {
      "epoch": 1.6828571428571428,
      "grad_norm": 0.05556400492787361,
      "learning_rate": 8.781904761904763e-06,
      "loss": 1.579,
      "step": 11780
    },
    {
      "epoch": 1.6842857142857142,
      "grad_norm": 0.057199202477931976,
      "learning_rate": 8.772380952380953e-06,
      "loss": 1.7623,
      "step": 11790
    },
    {
      "epoch": 1.6857142857142857,
      "grad_norm": 0.3160739541053772,
      "learning_rate": 8.762857142857144e-06,
      "loss": 0.0,
      "step": 11800
    },
    {
      "epoch": 1.6871428571428573,
      "grad_norm": 2627229.0,
      "learning_rate": 8.753333333333333e-06,
      "loss": 4.7954,
      "step": 11810
    },
    {
      "epoch": 1.6885714285714286,
      "grad_norm": 0.2876618504524231,
      "learning_rate": 8.743809523809524e-06,
      "loss": 1.6298,
      "step": 11820
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.27944937348365784,
      "learning_rate": 8.734285714285714e-06,
      "loss": 1.7804,
      "step": 11830
    },
    {
      "epoch": 1.6914285714285713,
      "grad_norm": 0.045453909784555435,
      "learning_rate": 8.724761904761905e-06,
      "loss": 0.0,
      "step": 11840
    },
    {
      "epoch": 1.6928571428571428,
      "grad_norm": 0.07794566452503204,
      "learning_rate": 8.715238095238096e-06,
      "loss": 1.6377,
      "step": 11850
    },
    {
      "epoch": 1.6942857142857144,
      "grad_norm": 0.2940438985824585,
      "learning_rate": 8.705714285714286e-06,
      "loss": 3.2646,
      "step": 11860
    },
    {
      "epoch": 1.6957142857142857,
      "grad_norm": 2683619.25,
      "learning_rate": 8.696190476190477e-06,
      "loss": 4.7708,
      "step": 11870
    },
    {
      "epoch": 1.697142857142857,
      "grad_norm": 2707529.5,
      "learning_rate": 8.686666666666668e-06,
      "loss": 3.3368,
      "step": 11880
    },
    {
      "epoch": 1.6985714285714286,
      "grad_norm": 0.06268660724163055,
      "learning_rate": 8.677142857142857e-06,
      "loss": 3.2559,
      "step": 11890
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.0952715203166008,
      "learning_rate": 8.667619047619047e-06,
      "loss": 3.189,
      "step": 11900
    },
    {
      "epoch": 1.7014285714285715,
      "grad_norm": 2160970.25,
      "learning_rate": 8.65809523809524e-06,
      "loss": 4.7074,
      "step": 11910
    },
    {
      "epoch": 1.7028571428571428,
      "grad_norm": 0.0613769106566906,
      "learning_rate": 8.64857142857143e-06,
      "loss": 4.8656,
      "step": 11920
    },
    {
      "epoch": 1.7042857142857142,
      "grad_norm": 0.27360329031944275,
      "learning_rate": 8.639047619047621e-06,
      "loss": 3.3129,
      "step": 11930
    },
    {
      "epoch": 1.7057142857142857,
      "grad_norm": 0.27994686365127563,
      "learning_rate": 8.62952380952381e-06,
      "loss": 1.7698,
      "step": 11940
    },
    {
      "epoch": 1.7071428571428573,
      "grad_norm": 0.08520540595054626,
      "learning_rate": 8.62e-06,
      "loss": 1.6039,
      "step": 11950
    },
    {
      "epoch": 1.7085714285714286,
      "grad_norm": 2642050.0,
      "learning_rate": 8.610476190476191e-06,
      "loss": 4.8126,
      "step": 11960
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.30390113592147827,
      "learning_rate": 8.600952380952382e-06,
      "loss": 3.2868,
      "step": 11970
    },
    {
      "epoch": 1.7114285714285713,
      "grad_norm": 0.27206140756607056,
      "learning_rate": 8.591428571428572e-06,
      "loss": 3.1549,
      "step": 11980
    },
    {
      "epoch": 1.7128571428571429,
      "grad_norm": 2682314.75,
      "learning_rate": 8.581904761904763e-06,
      "loss": 1.6383,
      "step": 11990
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 0.27989575266838074,
      "learning_rate": 8.572380952380954e-06,
      "loss": 1.585,
      "step": 12000
    },
    {
      "epoch": 1.7157142857142857,
      "grad_norm": 0.3239697217941284,
      "learning_rate": 8.562857142857144e-06,
      "loss": 6.4754,
      "step": 12010
    },
    {
      "epoch": 1.717142857142857,
      "grad_norm": 0.3855680525302887,
      "learning_rate": 8.553333333333333e-06,
      "loss": 0.0,
      "step": 12020
    },
    {
      "epoch": 1.7185714285714284,
      "grad_norm": 2666456.0,
      "learning_rate": 8.543809523809524e-06,
      "loss": 3.1497,
      "step": 12030
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.28660938143730164,
      "learning_rate": 8.534285714285715e-06,
      "loss": 0.0,
      "step": 12040
    },
    {
      "epoch": 1.7214285714285715,
      "grad_norm": 0.07276833057403564,
      "learning_rate": 8.524761904761905e-06,
      "loss": 0.0,
      "step": 12050
    },
    {
      "epoch": 1.7228571428571429,
      "grad_norm": 0.08704174309968948,
      "learning_rate": 8.515238095238096e-06,
      "loss": 6.5417,
      "step": 12060
    },
    {
      "epoch": 1.7242857142857142,
      "grad_norm": 0.4294500946998596,
      "learning_rate": 8.505714285714287e-06,
      "loss": 0.0,
      "step": 12070
    },
    {
      "epoch": 1.7257142857142858,
      "grad_norm": 2691145.25,
      "learning_rate": 8.496190476190477e-06,
      "loss": 1.7157,
      "step": 12080
    },
    {
      "epoch": 1.727142857142857,
      "grad_norm": 2680514.0,
      "learning_rate": 8.486666666666668e-06,
      "loss": 4.7792,
      "step": 12090
    },
    {
      "epoch": 1.7285714285714286,
      "grad_norm": 0.04227988421916962,
      "learning_rate": 8.477142857142859e-06,
      "loss": 1.5928,
      "step": 12100
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.3111175000667572,
      "learning_rate": 8.467619047619047e-06,
      "loss": 3.2171,
      "step": 12110
    },
    {
      "epoch": 1.7314285714285713,
      "grad_norm": 0.05176758021116257,
      "learning_rate": 8.458095238095238e-06,
      "loss": 1.5541,
      "step": 12120
    },
    {
      "epoch": 1.7328571428571429,
      "grad_norm": 0.3198159635066986,
      "learning_rate": 8.448571428571429e-06,
      "loss": 0.0,
      "step": 12130
    },
    {
      "epoch": 1.7342857142857144,
      "grad_norm": 0.04010104015469551,
      "learning_rate": 8.43904761904762e-06,
      "loss": 1.609,
      "step": 12140
    },
    {
      "epoch": 1.7357142857142858,
      "grad_norm": 0.07411621510982513,
      "learning_rate": 8.42952380952381e-06,
      "loss": 0.0,
      "step": 12150
    },
    {
      "epoch": 1.737142857142857,
      "grad_norm": 0.2752339839935303,
      "learning_rate": 8.42e-06,
      "loss": 1.5226,
      "step": 12160
    },
    {
      "epoch": 1.7385714285714284,
      "grad_norm": 0.3299456238746643,
      "learning_rate": 8.410476190476191e-06,
      "loss": 3.3204,
      "step": 12170
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.29449018836021423,
      "learning_rate": 8.400952380952382e-06,
      "loss": 1.536,
      "step": 12180
    },
    {
      "epoch": 1.7414285714285715,
      "grad_norm": 0.05400171875953674,
      "learning_rate": 8.391428571428573e-06,
      "loss": 3.2784,
      "step": 12190
    },
    {
      "epoch": 1.7428571428571429,
      "grad_norm": 0.033352531492710114,
      "learning_rate": 8.381904761904762e-06,
      "loss": 0.0,
      "step": 12200
    },
    {
      "epoch": 1.7442857142857142,
      "grad_norm": 0.11174055933952332,
      "learning_rate": 8.372380952380952e-06,
      "loss": 3.205,
      "step": 12210
    },
    {
      "epoch": 1.7457142857142856,
      "grad_norm": 0.3957633674144745,
      "learning_rate": 8.362857142857143e-06,
      "loss": 4.796,
      "step": 12220
    },
    {
      "epoch": 1.747142857142857,
      "grad_norm": 0.2935469150543213,
      "learning_rate": 8.353333333333335e-06,
      "loss": 0.0,
      "step": 12230
    },
    {
      "epoch": 1.7485714285714287,
      "grad_norm": 0.06805076450109482,
      "learning_rate": 8.343809523809524e-06,
      "loss": 1.5788,
      "step": 12240
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.08289561420679092,
      "learning_rate": 8.334285714285715e-06,
      "loss": 0.0,
      "step": 12250
    },
    {
      "epoch": 1.7514285714285713,
      "grad_norm": 0.09809799492359161,
      "learning_rate": 8.324761904761905e-06,
      "loss": 1.6706,
      "step": 12260
    },
    {
      "epoch": 1.752857142857143,
      "grad_norm": 0.26419609785079956,
      "learning_rate": 8.315238095238096e-06,
      "loss": 1.6684,
      "step": 12270
    },
    {
      "epoch": 1.7542857142857144,
      "grad_norm": 0.28544360399246216,
      "learning_rate": 8.305714285714285e-06,
      "loss": 1.6442,
      "step": 12280
    },
    {
      "epoch": 1.7557142857142858,
      "grad_norm": 0.07606469094753265,
      "learning_rate": 8.296190476190476e-06,
      "loss": 4.819,
      "step": 12290
    },
    {
      "epoch": 1.7571428571428571,
      "grad_norm": 0.057824235409498215,
      "learning_rate": 8.286666666666668e-06,
      "loss": 1.5591,
      "step": 12300
    },
    {
      "epoch": 1.7585714285714285,
      "grad_norm": 0.25446897745132446,
      "learning_rate": 8.277142857142859e-06,
      "loss": 3.2317,
      "step": 12310
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.04909758269786835,
      "learning_rate": 8.26761904761905e-06,
      "loss": 0.0,
      "step": 12320
    },
    {
      "epoch": 1.7614285714285716,
      "grad_norm": 2703655.5,
      "learning_rate": 8.258095238095238e-06,
      "loss": 3.1315,
      "step": 12330
    },
    {
      "epoch": 1.762857142857143,
      "grad_norm": 0.05647187680006027,
      "learning_rate": 8.248571428571429e-06,
      "loss": 6.4626,
      "step": 12340
    },
    {
      "epoch": 1.7642857142857142,
      "grad_norm": 0.10961688309907913,
      "learning_rate": 8.23904761904762e-06,
      "loss": 1.5746,
      "step": 12350
    },
    {
      "epoch": 1.7657142857142856,
      "grad_norm": 0.28542935848236084,
      "learning_rate": 8.22952380952381e-06,
      "loss": 6.4393,
      "step": 12360
    },
    {
      "epoch": 1.7671428571428571,
      "grad_norm": 0.34944427013397217,
      "learning_rate": 8.220000000000001e-06,
      "loss": 0.0,
      "step": 12370
    },
    {
      "epoch": 1.7685714285714287,
      "grad_norm": 2684455.75,
      "learning_rate": 8.210476190476192e-06,
      "loss": 1.6304,
      "step": 12380
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.27027326822280884,
      "learning_rate": 8.200952380952382e-06,
      "loss": 1.5629,
      "step": 12390
    },
    {
      "epoch": 1.7714285714285714,
      "grad_norm": 0.04507613554596901,
      "learning_rate": 8.191428571428573e-06,
      "loss": 1.5693,
      "step": 12400
    },
    {
      "epoch": 1.772857142857143,
      "grad_norm": 0.3361146152019501,
      "learning_rate": 8.181904761904764e-06,
      "loss": 3.1569,
      "step": 12410
    },
    {
      "epoch": 1.7742857142857142,
      "grad_norm": 0.2798568606376648,
      "learning_rate": 8.172380952380952e-06,
      "loss": 3.1066,
      "step": 12420
    },
    {
      "epoch": 1.7757142857142858,
      "grad_norm": 0.3398832082748413,
      "learning_rate": 8.162857142857143e-06,
      "loss": 3.1583,
      "step": 12430
    },
    {
      "epoch": 1.7771428571428571,
      "grad_norm": 0.3240271806716919,
      "learning_rate": 8.153333333333334e-06,
      "loss": 3.1214,
      "step": 12440
    },
    {
      "epoch": 1.7785714285714285,
      "grad_norm": 0.06398267298936844,
      "learning_rate": 8.143809523809524e-06,
      "loss": 3.2925,
      "step": 12450
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.35680556297302246,
      "learning_rate": 8.134285714285715e-06,
      "loss": 0.0,
      "step": 12460
    },
    {
      "epoch": 1.7814285714285716,
      "grad_norm": 0.08929067105054855,
      "learning_rate": 8.124761904761906e-06,
      "loss": 0.0,
      "step": 12470
    },
    {
      "epoch": 1.782857142857143,
      "grad_norm": 0.3254115581512451,
      "learning_rate": 8.115238095238096e-06,
      "loss": 4.8466,
      "step": 12480
    },
    {
      "epoch": 1.7842857142857143,
      "grad_norm": 0.29027268290519714,
      "learning_rate": 8.105714285714287e-06,
      "loss": 0.0,
      "step": 12490
    },
    {
      "epoch": 1.7857142857142856,
      "grad_norm": 0.2918083369731903,
      "learning_rate": 8.096190476190476e-06,
      "loss": 0.0,
      "step": 12500
    },
    {
      "epoch": 1.7871428571428571,
      "grad_norm": 0.08205100148916245,
      "learning_rate": 8.086666666666667e-06,
      "loss": 4.7859,
      "step": 12510
    },
    {
      "epoch": 1.7885714285714287,
      "grad_norm": 0.31172558665275574,
      "learning_rate": 8.077142857142857e-06,
      "loss": 8.0491,
      "step": 12520
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.2989666163921356,
      "learning_rate": 8.067619047619048e-06,
      "loss": 6.3448,
      "step": 12530
    },
    {
      "epoch": 1.7914285714285714,
      "grad_norm": 2681001.5,
      "learning_rate": 8.058095238095239e-06,
      "loss": 4.75,
      "step": 12540
    },
    {
      "epoch": 1.7928571428571427,
      "grad_norm": 0.32386988401412964,
      "learning_rate": 8.04857142857143e-06,
      "loss": 4.6664,
      "step": 12550
    },
    {
      "epoch": 1.7942857142857143,
      "grad_norm": 2685160.5,
      "learning_rate": 8.03904761904762e-06,
      "loss": 6.1749,
      "step": 12560
    },
    {
      "epoch": 1.7957142857142858,
      "grad_norm": 0.6991725564002991,
      "learning_rate": 8.02952380952381e-06,
      "loss": 3.2701,
      "step": 12570
    },
    {
      "epoch": 1.7971428571428572,
      "grad_norm": 0.9692411422729492,
      "learning_rate": 8.020000000000001e-06,
      "loss": 9.2649,
      "step": 12580
    },
    {
      "epoch": 1.7985714285714285,
      "grad_norm": 0.04359370470046997,
      "learning_rate": 8.01047619047619e-06,
      "loss": 1.5142,
      "step": 12590
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7130703330039978,
      "learning_rate": 8.00095238095238e-06,
      "loss": 4.7197,
      "step": 12600
    },
    {
      "epoch": 1.8014285714285714,
      "grad_norm": 0.06616736203432083,
      "learning_rate": 7.991428571428571e-06,
      "loss": 1.5027,
      "step": 12610
    },
    {
      "epoch": 1.802857142857143,
      "grad_norm": 0.9018630385398865,
      "learning_rate": 7.981904761904764e-06,
      "loss": 1.5424,
      "step": 12620
    },
    {
      "epoch": 1.8042857142857143,
      "grad_norm": 0.08448358625173569,
      "learning_rate": 7.972380952380953e-06,
      "loss": 0.0,
      "step": 12630
    },
    {
      "epoch": 1.8057142857142856,
      "grad_norm": 0.060722704976797104,
      "learning_rate": 7.962857142857143e-06,
      "loss": 0.0,
      "step": 12640
    },
    {
      "epoch": 1.8071428571428572,
      "grad_norm": 0.08254537731409073,
      "learning_rate": 7.953333333333334e-06,
      "loss": 3.4139,
      "step": 12650
    },
    {
      "epoch": 1.8085714285714287,
      "grad_norm": 0.3203651010990143,
      "learning_rate": 7.943809523809525e-06,
      "loss": 4.7859,
      "step": 12660
    },
    {
      "epoch": 1.81,
      "grad_norm": 2660634.75,
      "learning_rate": 7.934285714285715e-06,
      "loss": 3.1989,
      "step": 12670
    },
    {
      "epoch": 1.8114285714285714,
      "grad_norm": 0.08424670249223709,
      "learning_rate": 7.924761904761906e-06,
      "loss": 1.6311,
      "step": 12680
    },
    {
      "epoch": 1.8128571428571427,
      "grad_norm": 0.45503053069114685,
      "learning_rate": 7.915238095238097e-06,
      "loss": 1.5398,
      "step": 12690
    },
    {
      "epoch": 1.8142857142857143,
      "grad_norm": 0.28270775079727173,
      "learning_rate": 7.905714285714287e-06,
      "loss": 3.2011,
      "step": 12700
    },
    {
      "epoch": 1.8157142857142858,
      "grad_norm": 0.32408952713012695,
      "learning_rate": 7.896190476190478e-06,
      "loss": 1.6321,
      "step": 12710
    },
    {
      "epoch": 1.8171428571428572,
      "grad_norm": 2638489.25,
      "learning_rate": 7.886666666666667e-06,
      "loss": 8.0434,
      "step": 12720
    },
    {
      "epoch": 1.8185714285714285,
      "grad_norm": 0.5758330225944519,
      "learning_rate": 7.877142857142857e-06,
      "loss": 3.2946,
      "step": 12730
    },
    {
      "epoch": 1.8199999999999998,
      "grad_norm": 0.6163074374198914,
      "learning_rate": 7.867619047619048e-06,
      "loss": 1.7208,
      "step": 12740
    },
    {
      "epoch": 1.8214285714285714,
      "grad_norm": 0.11204085499048233,
      "learning_rate": 7.858095238095239e-06,
      "loss": 1.548,
      "step": 12750
    },
    {
      "epoch": 1.822857142857143,
      "grad_norm": 0.3880802094936371,
      "learning_rate": 7.84857142857143e-06,
      "loss": 6.2363,
      "step": 12760
    },
    {
      "epoch": 1.8242857142857143,
      "grad_norm": 2649566.5,
      "learning_rate": 7.83904761904762e-06,
      "loss": 1.5715,
      "step": 12770
    },
    {
      "epoch": 1.8257142857142856,
      "grad_norm": 0.31142643094062805,
      "learning_rate": 7.82952380952381e-06,
      "loss": 1.5796,
      "step": 12780
    },
    {
      "epoch": 1.8271428571428572,
      "grad_norm": 0.33521008491516113,
      "learning_rate": 7.820000000000001e-06,
      "loss": 1.6016,
      "step": 12790
    },
    {
      "epoch": 1.8285714285714287,
      "grad_norm": 2917802.25,
      "learning_rate": 7.810476190476192e-06,
      "loss": 3.1619,
      "step": 12800
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.27896031737327576,
      "learning_rate": 7.800952380952381e-06,
      "loss": 0.0,
      "step": 12810
    },
    {
      "epoch": 1.8314285714285714,
      "grad_norm": 2630128.5,
      "learning_rate": 7.791428571428572e-06,
      "loss": 3.1421,
      "step": 12820
    },
    {
      "epoch": 1.8328571428571427,
      "grad_norm": 0.289754182100296,
      "learning_rate": 7.781904761904762e-06,
      "loss": 0.0,
      "step": 12830
    },
    {
      "epoch": 1.8342857142857143,
      "grad_norm": 0.06289057433605194,
      "learning_rate": 7.772380952380953e-06,
      "loss": 0.0,
      "step": 12840
    },
    {
      "epoch": 1.8357142857142859,
      "grad_norm": 0.2782083749771118,
      "learning_rate": 7.762857142857144e-06,
      "loss": 1.5919,
      "step": 12850
    },
    {
      "epoch": 1.8371428571428572,
      "grad_norm": 0.056629862636327744,
      "learning_rate": 7.753333333333334e-06,
      "loss": 1.7148,
      "step": 12860
    },
    {
      "epoch": 1.8385714285714285,
      "grad_norm": 0.09039817005395889,
      "learning_rate": 7.743809523809525e-06,
      "loss": 3.313,
      "step": 12870
    },
    {
      "epoch": 1.8399999999999999,
      "grad_norm": 0.043976668268442154,
      "learning_rate": 7.734285714285715e-06,
      "loss": 0.0,
      "step": 12880
    },
    {
      "epoch": 1.8414285714285714,
      "grad_norm": 0.318976491689682,
      "learning_rate": 7.724761904761904e-06,
      "loss": 1.6282,
      "step": 12890
    },
    {
      "epoch": 1.842857142857143,
      "grad_norm": 0.2572118639945984,
      "learning_rate": 7.715238095238095e-06,
      "loss": 0.0,
      "step": 12900
    },
    {
      "epoch": 1.8442857142857143,
      "grad_norm": 0.26584768295288086,
      "learning_rate": 7.705714285714286e-06,
      "loss": 1.6662,
      "step": 12910
    },
    {
      "epoch": 1.8457142857142856,
      "grad_norm": 2708454.5,
      "learning_rate": 7.696190476190476e-06,
      "loss": 3.3723,
      "step": 12920
    },
    {
      "epoch": 1.8471428571428572,
      "grad_norm": 0.03493857383728027,
      "learning_rate": 7.686666666666667e-06,
      "loss": 1.7183,
      "step": 12930
    },
    {
      "epoch": 1.8485714285714285,
      "grad_norm": 2672767.5,
      "learning_rate": 7.677142857142858e-06,
      "loss": 3.2476,
      "step": 12940
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.32090887427330017,
      "learning_rate": 7.667619047619048e-06,
      "loss": 4.9479,
      "step": 12950
    },
    {
      "epoch": 1.8514285714285714,
      "grad_norm": 0.0972277969121933,
      "learning_rate": 7.658095238095239e-06,
      "loss": 1.6144,
      "step": 12960
    },
    {
      "epoch": 1.8528571428571428,
      "grad_norm": 0.31062665581703186,
      "learning_rate": 7.64857142857143e-06,
      "loss": 1.6499,
      "step": 12970
    },
    {
      "epoch": 1.8542857142857143,
      "grad_norm": 2684994.5,
      "learning_rate": 7.639047619047619e-06,
      "loss": 3.2699,
      "step": 12980
    },
    {
      "epoch": 1.8557142857142859,
      "grad_norm": 0.2671312987804413,
      "learning_rate": 7.62952380952381e-06,
      "loss": 3.2334,
      "step": 12990
    },
    {
      "epoch": 1.8571428571428572,
      "grad_norm": 0.08492816984653473,
      "learning_rate": 7.620000000000001e-06,
      "loss": 3.2258,
      "step": 13000
    },
    {
      "epoch": 1.8585714285714285,
      "grad_norm": 0.08103000372648239,
      "learning_rate": 7.610476190476191e-06,
      "loss": 1.5958,
      "step": 13010
    },
    {
      "epoch": 1.8599999999999999,
      "grad_norm": 0.04117155820131302,
      "learning_rate": 7.600952380952381e-06,
      "loss": 3.2399,
      "step": 13020
    },
    {
      "epoch": 1.8614285714285714,
      "grad_norm": 0.05828747898340225,
      "learning_rate": 7.591428571428572e-06,
      "loss": 1.6104,
      "step": 13030
    },
    {
      "epoch": 1.862857142857143,
      "grad_norm": 0.06786676496267319,
      "learning_rate": 7.5819047619047625e-06,
      "loss": 1.5893,
      "step": 13040
    },
    {
      "epoch": 1.8642857142857143,
      "grad_norm": 0.2744838297367096,
      "learning_rate": 7.572380952380953e-06,
      "loss": 1.5602,
      "step": 13050
    },
    {
      "epoch": 1.8657142857142857,
      "grad_norm": 0.2825525403022766,
      "learning_rate": 7.562857142857144e-06,
      "loss": 3.2027,
      "step": 13060
    },
    {
      "epoch": 1.867142857142857,
      "grad_norm": 0.2589586675167084,
      "learning_rate": 7.553333333333334e-06,
      "loss": 4.8492,
      "step": 13070
    },
    {
      "epoch": 1.8685714285714285,
      "grad_norm": 0.049171384423971176,
      "learning_rate": 7.543809523809524e-06,
      "loss": 3.2046,
      "step": 13080
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.2702985405921936,
      "learning_rate": 7.534285714285715e-06,
      "loss": 0.0,
      "step": 13090
    },
    {
      "epoch": 1.8714285714285714,
      "grad_norm": 0.3705894351005554,
      "learning_rate": 7.5247619047619055e-06,
      "loss": 3.2352,
      "step": 13100
    },
    {
      "epoch": 1.8728571428571428,
      "grad_norm": 0.027351519092917442,
      "learning_rate": 7.515238095238095e-06,
      "loss": 3.234,
      "step": 13110
    },
    {
      "epoch": 1.8742857142857143,
      "grad_norm": 0.05229870229959488,
      "learning_rate": 7.505714285714286e-06,
      "loss": 4.8652,
      "step": 13120
    },
    {
      "epoch": 1.8757142857142857,
      "grad_norm": 0.0326702743768692,
      "learning_rate": 7.496190476190477e-06,
      "loss": 3.2007,
      "step": 13130
    },
    {
      "epoch": 1.8771428571428572,
      "grad_norm": 0.0747067853808403,
      "learning_rate": 7.486666666666667e-06,
      "loss": 4.9172,
      "step": 13140
    },
    {
      "epoch": 1.8785714285714286,
      "grad_norm": 0.5954580307006836,
      "learning_rate": 7.477142857142857e-06,
      "loss": 1.5616,
      "step": 13150
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.29769167304039,
      "learning_rate": 7.467619047619048e-06,
      "loss": 3.3552,
      "step": 13160
    },
    {
      "epoch": 1.8814285714285715,
      "grad_norm": 0.044067785143852234,
      "learning_rate": 7.458095238095238e-06,
      "loss": 6.4895,
      "step": 13170
    },
    {
      "epoch": 1.882857142857143,
      "grad_norm": 0.10143379867076874,
      "learning_rate": 7.44857142857143e-06,
      "loss": 1.5352,
      "step": 13180
    },
    {
      "epoch": 1.8842857142857143,
      "grad_norm": 0.06495194882154465,
      "learning_rate": 7.4390476190476205e-06,
      "loss": 0.0,
      "step": 13190
    },
    {
      "epoch": 1.8857142857142857,
      "grad_norm": 0.3443673849105835,
      "learning_rate": 7.4295238095238095e-06,
      "loss": 3.2785,
      "step": 13200
    },
    {
      "epoch": 1.887142857142857,
      "grad_norm": 0.2742867171764374,
      "learning_rate": 7.420000000000001e-06,
      "loss": 0.0,
      "step": 13210
    },
    {
      "epoch": 1.8885714285714286,
      "grad_norm": 0.08340451121330261,
      "learning_rate": 7.410476190476192e-06,
      "loss": 4.8955,
      "step": 13220
    },
    {
      "epoch": 1.8900000000000001,
      "grad_norm": 0.06812101602554321,
      "learning_rate": 7.400952380952382e-06,
      "loss": 3.2671,
      "step": 13230
    },
    {
      "epoch": 1.8914285714285715,
      "grad_norm": 2805775.25,
      "learning_rate": 7.391428571428571e-06,
      "loss": 3.2055,
      "step": 13240
    },
    {
      "epoch": 1.8928571428571428,
      "grad_norm": 0.27274516224861145,
      "learning_rate": 7.381904761904763e-06,
      "loss": 1.5733,
      "step": 13250
    },
    {
      "epoch": 1.8942857142857141,
      "grad_norm": 0.04588813707232475,
      "learning_rate": 7.372380952380953e-06,
      "loss": 0.0,
      "step": 13260
    },
    {
      "epoch": 1.8957142857142857,
      "grad_norm": 0.26624786853790283,
      "learning_rate": 7.362857142857144e-06,
      "loss": 0.0,
      "step": 13270
    },
    {
      "epoch": 1.8971428571428572,
      "grad_norm": 0.11292257905006409,
      "learning_rate": 7.353333333333334e-06,
      "loss": 1.5655,
      "step": 13280
    },
    {
      "epoch": 1.8985714285714286,
      "grad_norm": 0.06331798434257507,
      "learning_rate": 7.3438095238095244e-06,
      "loss": 4.8867,
      "step": 13290
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.3434509336948395,
      "learning_rate": 7.334285714285715e-06,
      "loss": 0.0,
      "step": 13300
    },
    {
      "epoch": 1.9014285714285715,
      "grad_norm": 0.2621558904647827,
      "learning_rate": 7.324761904761906e-06,
      "loss": 4.7859,
      "step": 13310
    },
    {
      "epoch": 1.9028571428571428,
      "grad_norm": 0.04607006907463074,
      "learning_rate": 7.315238095238096e-06,
      "loss": 3.2777,
      "step": 13320
    },
    {
      "epoch": 1.9042857142857144,
      "grad_norm": 0.03129078820347786,
      "learning_rate": 7.305714285714286e-06,
      "loss": 1.585,
      "step": 13330
    },
    {
      "epoch": 1.9057142857142857,
      "grad_norm": 0.33169806003570557,
      "learning_rate": 7.296190476190477e-06,
      "loss": 4.8275,
      "step": 13340
    },
    {
      "epoch": 1.907142857142857,
      "grad_norm": 0.2668425142765045,
      "learning_rate": 7.2866666666666675e-06,
      "loss": 1.7285,
      "step": 13350
    },
    {
      "epoch": 1.9085714285714286,
      "grad_norm": 0.3559333086013794,
      "learning_rate": 7.277142857142858e-06,
      "loss": 1.5807,
      "step": 13360
    },
    {
      "epoch": 1.9100000000000001,
      "grad_norm": 2693553.25,
      "learning_rate": 7.267619047619048e-06,
      "loss": 1.6263,
      "step": 13370
    },
    {
      "epoch": 1.9114285714285715,
      "grad_norm": 0.033235613256692886,
      "learning_rate": 7.258095238095239e-06,
      "loss": 4.9197,
      "step": 13380
    },
    {
      "epoch": 1.9128571428571428,
      "grad_norm": 2699659.75,
      "learning_rate": 7.248571428571429e-06,
      "loss": 8.1446,
      "step": 13390
    },
    {
      "epoch": 1.9142857142857141,
      "grad_norm": 0.10074140876531601,
      "learning_rate": 7.23904761904762e-06,
      "loss": 1.6298,
      "step": 13400
    },
    {
      "epoch": 1.9157142857142857,
      "grad_norm": 2666896.5,
      "learning_rate": 7.22952380952381e-06,
      "loss": 4.9334,
      "step": 13410
    },
    {
      "epoch": 1.9171428571428573,
      "grad_norm": 0.08268934488296509,
      "learning_rate": 7.22e-06,
      "loss": 1.6271,
      "step": 13420
    },
    {
      "epoch": 1.9185714285714286,
      "grad_norm": 2841545.5,
      "learning_rate": 7.210476190476191e-06,
      "loss": 5.0148,
      "step": 13430
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.1122441217303276,
      "learning_rate": 7.200952380952382e-06,
      "loss": 4.9148,
      "step": 13440
    },
    {
      "epoch": 1.9214285714285713,
      "grad_norm": 0.29528385400772095,
      "learning_rate": 7.191428571428572e-06,
      "loss": 8.0884,
      "step": 13450
    },
    {
      "epoch": 1.9228571428571428,
      "grad_norm": 2920895.5,
      "learning_rate": 7.181904761904762e-06,
      "loss": 9.5128,
      "step": 13460
    },
    {
      "epoch": 1.9242857142857144,
      "grad_norm": 0.27183178067207336,
      "learning_rate": 7.172380952380953e-06,
      "loss": 3.0097,
      "step": 13470
    },
    {
      "epoch": 1.9257142857142857,
      "grad_norm": 6.8426899909973145,
      "learning_rate": 7.162857142857143e-06,
      "loss": 8.7023,
      "step": 13480
    },
    {
      "epoch": 1.927142857142857,
      "grad_norm": 5102857.5,
      "learning_rate": 7.153333333333334e-06,
      "loss": 7.4458,
      "step": 13490
    },
    {
      "epoch": 1.9285714285714286,
      "grad_norm": 7050508.0,
      "learning_rate": 7.143809523809524e-06,
      "loss": 3.4189,
      "step": 13500
    },
    {
      "epoch": 1.9300000000000002,
      "grad_norm": 0.09868761152029037,
      "learning_rate": 7.1342857142857145e-06,
      "loss": 0.6336,
      "step": 13510
    },
    {
      "epoch": 1.9314285714285715,
      "grad_norm": 3388047.25,
      "learning_rate": 7.124761904761905e-06,
      "loss": 2.1834,
      "step": 13520
    },
    {
      "epoch": 1.9328571428571428,
      "grad_norm": 0.048632293939590454,
      "learning_rate": 7.115238095238096e-06,
      "loss": 0.0006,
      "step": 13530
    },
    {
      "epoch": 1.9342857142857142,
      "grad_norm": 0.2727932631969452,
      "learning_rate": 7.1057142857142856e-06,
      "loss": 6.7373,
      "step": 13540
    },
    {
      "epoch": 1.9357142857142857,
      "grad_norm": 0.0744365006685257,
      "learning_rate": 7.096190476190476e-06,
      "loss": 1.5732,
      "step": 13550
    },
    {
      "epoch": 1.9371428571428573,
      "grad_norm": 0.27282628417015076,
      "learning_rate": 7.086666666666667e-06,
      "loss": 0.0,
      "step": 13560
    },
    {
      "epoch": 1.9385714285714286,
      "grad_norm": 0.3557356297969818,
      "learning_rate": 7.077142857142858e-06,
      "loss": 0.0,
      "step": 13570
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.09538581222295761,
      "learning_rate": 7.067619047619049e-06,
      "loss": 1.5536,
      "step": 13580
    },
    {
      "epoch": 1.9414285714285713,
      "grad_norm": 0.27574536204338074,
      "learning_rate": 7.058095238095238e-06,
      "loss": 3.3373,
      "step": 13590
    },
    {
      "epoch": 1.9428571428571428,
      "grad_norm": 0.07006961107254028,
      "learning_rate": 7.0485714285714295e-06,
      "loss": 1.7258,
      "step": 13600
    },
    {
      "epoch": 1.9442857142857144,
      "grad_norm": 0.3057045638561249,
      "learning_rate": 7.03904761904762e-06,
      "loss": 0.0,
      "step": 13610
    },
    {
      "epoch": 1.9457142857142857,
      "grad_norm": 0.2658008337020874,
      "learning_rate": 7.029523809523811e-06,
      "loss": 1.6119,
      "step": 13620
    },
    {
      "epoch": 1.947142857142857,
      "grad_norm": 0.08354052901268005,
      "learning_rate": 7.0200000000000006e-06,
      "loss": 1.6587,
      "step": 13630
    },
    {
      "epoch": 1.9485714285714286,
      "grad_norm": 0.2791806757450104,
      "learning_rate": 7.010476190476191e-06,
      "loss": 3.3207,
      "step": 13640
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.0807982087135315,
      "learning_rate": 7.000952380952382e-06,
      "loss": 1.6239,
      "step": 13650
    },
    {
      "epoch": 1.9514285714285715,
      "grad_norm": 0.29967960715293884,
      "learning_rate": 6.9914285714285725e-06,
      "loss": 1.6197,
      "step": 13660
    },
    {
      "epoch": 1.9528571428571428,
      "grad_norm": 0.060215212404727936,
      "learning_rate": 6.981904761904762e-06,
      "loss": 3.3209,
      "step": 13670
    },
    {
      "epoch": 1.9542857142857142,
      "grad_norm": 0.25531065464019775,
      "learning_rate": 6.972380952380953e-06,
      "loss": 1.5919,
      "step": 13680
    },
    {
      "epoch": 1.9557142857142857,
      "grad_norm": 0.12100515514612198,
      "learning_rate": 6.962857142857144e-06,
      "loss": 1.5945,
      "step": 13690
    },
    {
      "epoch": 1.9571428571428573,
      "grad_norm": 0.332278847694397,
      "learning_rate": 6.953333333333334e-06,
      "loss": 3.263,
      "step": 13700
    },
    {
      "epoch": 1.9585714285714286,
      "grad_norm": 0.08925547450780869,
      "learning_rate": 6.943809523809525e-06,
      "loss": 0.0,
      "step": 13710
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.263852059841156,
      "learning_rate": 6.934285714285715e-06,
      "loss": 1.6165,
      "step": 13720
    },
    {
      "epoch": 1.9614285714285713,
      "grad_norm": 0.12665005028247833,
      "learning_rate": 6.924761904761905e-06,
      "loss": 0.0,
      "step": 13730
    },
    {
      "epoch": 1.9628571428571429,
      "grad_norm": 0.2846013009548187,
      "learning_rate": 6.915238095238096e-06,
      "loss": 0.0,
      "step": 13740
    },
    {
      "epoch": 1.9642857142857144,
      "grad_norm": 2780921.75,
      "learning_rate": 6.905714285714287e-06,
      "loss": 6.7779,
      "step": 13750
    },
    {
      "epoch": 1.9657142857142857,
      "grad_norm": 2723070.75,
      "learning_rate": 6.8961904761904764e-06,
      "loss": 3.2882,
      "step": 13760
    },
    {
      "epoch": 1.967142857142857,
      "grad_norm": 0.07653206586837769,
      "learning_rate": 6.886666666666667e-06,
      "loss": 1.6122,
      "step": 13770
    },
    {
      "epoch": 1.9685714285714284,
      "grad_norm": 2765774.5,
      "learning_rate": 6.877142857142858e-06,
      "loss": 4.8689,
      "step": 13780
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.2873535454273224,
      "learning_rate": 6.867619047619048e-06,
      "loss": 4.7469,
      "step": 13790
    },
    {
      "epoch": 1.9714285714285715,
      "grad_norm": 0.07044811546802521,
      "learning_rate": 6.858095238095238e-06,
      "loss": 4.779,
      "step": 13800
    },
    {
      "epoch": 1.9728571428571429,
      "grad_norm": 0.0695636123418808,
      "learning_rate": 6.848571428571429e-06,
      "loss": 0.0,
      "step": 13810
    },
    {
      "epoch": 1.9742857142857142,
      "grad_norm": 3053003.75,
      "learning_rate": 6.8390476190476195e-06,
      "loss": 4.8695,
      "step": 13820
    },
    {
      "epoch": 1.9757142857142858,
      "grad_norm": 0.06094406917691231,
      "learning_rate": 6.82952380952381e-06,
      "loss": 3.1374,
      "step": 13830
    },
    {
      "epoch": 1.977142857142857,
      "grad_norm": 2737894.5,
      "learning_rate": 6.820000000000001e-06,
      "loss": 6.1736,
      "step": 13840
    },
    {
      "epoch": 1.9785714285714286,
      "grad_norm": 0.08146178722381592,
      "learning_rate": 6.8104761904761906e-06,
      "loss": 0.0,
      "step": 13850
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.07577681541442871,
      "learning_rate": 6.800952380952381e-06,
      "loss": 0.0,
      "step": 13860
    },
    {
      "epoch": 1.9814285714285713,
      "grad_norm": 0.2732942998409271,
      "learning_rate": 6.791428571428572e-06,
      "loss": 0.0,
      "step": 13870
    },
    {
      "epoch": 1.9828571428571429,
      "grad_norm": 0.0666847825050354,
      "learning_rate": 6.7819047619047625e-06,
      "loss": 1.1028,
      "step": 13880
    },
    {
      "epoch": 1.9842857142857144,
      "grad_norm": 0.0670706257224083,
      "learning_rate": 6.772380952380952e-06,
      "loss": 3.2564,
      "step": 13890
    },
    {
      "epoch": 1.9857142857142858,
      "grad_norm": 0.3063766658306122,
      "learning_rate": 6.762857142857143e-06,
      "loss": 1.6813,
      "step": 13900
    },
    {
      "epoch": 1.987142857142857,
      "grad_norm": 0.07315870374441147,
      "learning_rate": 6.753333333333334e-06,
      "loss": 3.1108,
      "step": 13910
    },
    {
      "epoch": 1.9885714285714284,
      "grad_norm": 0.11694533377885818,
      "learning_rate": 6.743809523809525e-06,
      "loss": 0.0,
      "step": 13920
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.07647539675235748,
      "learning_rate": 6.734285714285714e-06,
      "loss": 1.5608,
      "step": 13930
    },
    {
      "epoch": 1.9914285714285715,
      "grad_norm": 0.10481464117765427,
      "learning_rate": 6.724761904761905e-06,
      "loss": 0.0,
      "step": 13940
    },
    {
      "epoch": 1.9928571428571429,
      "grad_norm": 0.0954461544752121,
      "learning_rate": 6.715238095238095e-06,
      "loss": 1.7095,
      "step": 13950
    },
    {
      "epoch": 1.9942857142857142,
      "grad_norm": 0.05502492934465408,
      "learning_rate": 6.705714285714287e-06,
      "loss": 0.0,
      "step": 13960
    },
    {
      "epoch": 1.9957142857142856,
      "grad_norm": 0.11214712262153625,
      "learning_rate": 6.6961904761904775e-06,
      "loss": 3.1152,
      "step": 13970
    },
    {
      "epoch": 1.997142857142857,
      "grad_norm": 0.211678147315979,
      "learning_rate": 6.6866666666666665e-06,
      "loss": 0.0,
      "step": 13980
    },
    {
      "epoch": 1.9985714285714287,
      "grad_norm": 0.3000403940677643,
      "learning_rate": 6.677142857142858e-06,
      "loss": 3.4827,
      "step": 13990
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.06649123877286911,
      "learning_rate": 6.667619047619049e-06,
      "loss": 2.55,
      "step": 14000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.859,
      "eval_f1_macro": 0.8589141571026995,
      "eval_f1_weighted": 0.8592760893183452,
      "eval_loss": 1.7680644989013672,
      "eval_runtime": 255.48,
      "eval_samples_per_second": 23.485,
      "eval_steps_per_second": 11.743,
      "step": 14000
    },
    {
      "epoch": 2.0014285714285713,
      "grad_norm": 0.07783756405115128,
      "learning_rate": 6.658095238095239e-06,
      "loss": 0.073,
      "step": 14010
    },
    {
      "epoch": 2.0028571428571427,
      "grad_norm": 0.2640673816204071,
      "learning_rate": 6.648571428571429e-06,
      "loss": 1.5738,
      "step": 14020
    },
    {
      "epoch": 2.0042857142857144,
      "grad_norm": 0.06619155406951904,
      "learning_rate": 6.63904761904762e-06,
      "loss": 0.0,
      "step": 14030
    },
    {
      "epoch": 2.005714285714286,
      "grad_norm": 0.05236930027604103,
      "learning_rate": 6.62952380952381e-06,
      "loss": 3.2776,
      "step": 14040
    },
    {
      "epoch": 2.007142857142857,
      "grad_norm": 0.14656458795070648,
      "learning_rate": 6.620000000000001e-06,
      "loss": 0.0,
      "step": 14050
    },
    {
      "epoch": 2.0085714285714285,
      "grad_norm": 0.12643378973007202,
      "learning_rate": 6.610476190476191e-06,
      "loss": 0.913,
      "step": 14060
    },
    {
      "epoch": 2.01,
      "grad_norm": 3132056.75,
      "learning_rate": 6.6009523809523814e-06,
      "loss": 1.6993,
      "step": 14070
    },
    {
      "epoch": 2.0114285714285716,
      "grad_norm": 0.10983424633741379,
      "learning_rate": 6.591428571428572e-06,
      "loss": 4.7271,
      "step": 14080
    },
    {
      "epoch": 2.012857142857143,
      "grad_norm": 0.32419082522392273,
      "learning_rate": 6.581904761904763e-06,
      "loss": 1.6729,
      "step": 14090
    },
    {
      "epoch": 2.0142857142857142,
      "grad_norm": 2.1024200916290283,
      "learning_rate": 6.572380952380953e-06,
      "loss": 5.1711,
      "step": 14100
    },
    {
      "epoch": 2.0157142857142856,
      "grad_norm": 0.1469053328037262,
      "learning_rate": 6.562857142857143e-06,
      "loss": 0.4048,
      "step": 14110
    },
    {
      "epoch": 2.0171428571428573,
      "grad_norm": 0.07334276288747787,
      "learning_rate": 6.553333333333334e-06,
      "loss": 1.7111,
      "step": 14120
    },
    {
      "epoch": 2.0185714285714287,
      "grad_norm": 0.29361841082572937,
      "learning_rate": 6.5438095238095245e-06,
      "loss": 1.6859,
      "step": 14130
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.1298183798789978,
      "learning_rate": 6.534285714285715e-06,
      "loss": 3.3206,
      "step": 14140
    },
    {
      "epoch": 2.0214285714285714,
      "grad_norm": 0.0568879060447216,
      "learning_rate": 6.524761904761905e-06,
      "loss": 4.5887,
      "step": 14150
    },
    {
      "epoch": 2.0228571428571427,
      "grad_norm": 3638.93408203125,
      "learning_rate": 6.515238095238096e-06,
      "loss": 1.1823,
      "step": 14160
    },
    {
      "epoch": 2.0242857142857145,
      "grad_norm": 0.10810936987400055,
      "learning_rate": 6.505714285714286e-06,
      "loss": 1.7179,
      "step": 14170
    },
    {
      "epoch": 2.025714285714286,
      "grad_norm": 0.08599168807268143,
      "learning_rate": 6.496190476190477e-06,
      "loss": 1.6276,
      "step": 14180
    },
    {
      "epoch": 2.027142857142857,
      "grad_norm": 0.05341481417417526,
      "learning_rate": 6.486666666666667e-06,
      "loss": 0.0,
      "step": 14190
    },
    {
      "epoch": 2.0285714285714285,
      "grad_norm": 0.14270149171352386,
      "learning_rate": 6.477142857142857e-06,
      "loss": 6.0271,
      "step": 14200
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.28994354605674744,
      "learning_rate": 6.467619047619048e-06,
      "loss": 0.0,
      "step": 14210
    },
    {
      "epoch": 2.0314285714285716,
      "grad_norm": 0.0669870525598526,
      "learning_rate": 6.458095238095239e-06,
      "loss": 0.5029,
      "step": 14220
    },
    {
      "epoch": 2.032857142857143,
      "grad_norm": 0.10732413083314896,
      "learning_rate": 6.448571428571429e-06,
      "loss": 0.0,
      "step": 14230
    },
    {
      "epoch": 2.0342857142857143,
      "grad_norm": 0.10219500213861465,
      "learning_rate": 6.439047619047619e-06,
      "loss": 3.449,
      "step": 14240
    },
    {
      "epoch": 2.0357142857142856,
      "grad_norm": 0.15604828298091888,
      "learning_rate": 6.42952380952381e-06,
      "loss": 0.0,
      "step": 14250
    },
    {
      "epoch": 2.0371428571428574,
      "grad_norm": 0.10480683296918869,
      "learning_rate": 6.42e-06,
      "loss": 0.0003,
      "step": 14260
    },
    {
      "epoch": 2.0385714285714287,
      "grad_norm": 0.05462351068854332,
      "learning_rate": 6.410476190476191e-06,
      "loss": 0.0,
      "step": 14270
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.06909993290901184,
      "learning_rate": 6.400952380952381e-06,
      "loss": 0.4158,
      "step": 14280
    },
    {
      "epoch": 2.0414285714285714,
      "grad_norm": 0.2598775029182434,
      "learning_rate": 6.3914285714285715e-06,
      "loss": 0.2053,
      "step": 14290
    },
    {
      "epoch": 2.0428571428571427,
      "grad_norm": 0.3589019477367401,
      "learning_rate": 6.381904761904762e-06,
      "loss": 1.9562,
      "step": 14300
    },
    {
      "epoch": 2.0442857142857145,
      "grad_norm": 0.10039866715669632,
      "learning_rate": 6.372380952380954e-06,
      "loss": 0.0317,
      "step": 14310
    },
    {
      "epoch": 2.045714285714286,
      "grad_norm": 0.0980788841843605,
      "learning_rate": 6.3628571428571426e-06,
      "loss": 0.0011,
      "step": 14320
    },
    {
      "epoch": 2.047142857142857,
      "grad_norm": 2882071.5,
      "learning_rate": 6.353333333333333e-06,
      "loss": 1.6932,
      "step": 14330
    },
    {
      "epoch": 2.0485714285714285,
      "grad_norm": 0.12254133075475693,
      "learning_rate": 6.343809523809525e-06,
      "loss": 1.6705,
      "step": 14340
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2750368118286133,
      "learning_rate": 6.334285714285715e-06,
      "loss": 3.287,
      "step": 14350
    },
    {
      "epoch": 2.0514285714285716,
      "grad_norm": 0.10952204465866089,
      "learning_rate": 6.324761904761906e-06,
      "loss": 1.4412,
      "step": 14360
    },
    {
      "epoch": 2.052857142857143,
      "grad_norm": 0.06369184702634811,
      "learning_rate": 6.315238095238096e-06,
      "loss": 0.0,
      "step": 14370
    },
    {
      "epoch": 2.0542857142857143,
      "grad_norm": 0.040040429681539536,
      "learning_rate": 6.3057142857142864e-06,
      "loss": 3.1104,
      "step": 14380
    },
    {
      "epoch": 2.0557142857142856,
      "grad_norm": 0.26102545857429504,
      "learning_rate": 6.296190476190477e-06,
      "loss": 3.1006,
      "step": 14390
    },
    {
      "epoch": 2.057142857142857,
      "grad_norm": 0.04162650555372238,
      "learning_rate": 6.286666666666668e-06,
      "loss": 3.101,
      "step": 14400
    },
    {
      "epoch": 2.0585714285714287,
      "grad_norm": 582514496.0,
      "learning_rate": 6.2771428571428575e-06,
      "loss": 3.6604,
      "step": 14410
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.2758941352367401,
      "learning_rate": 6.267619047619048e-06,
      "loss": 3.3641,
      "step": 14420
    },
    {
      "epoch": 2.0614285714285714,
      "grad_norm": 0.13902081549167633,
      "learning_rate": 6.258095238095239e-06,
      "loss": 0.0,
      "step": 14430
    },
    {
      "epoch": 2.0628571428571427,
      "grad_norm": 0.04980390518903732,
      "learning_rate": 6.2485714285714295e-06,
      "loss": 1.7259,
      "step": 14440
    },
    {
      "epoch": 2.064285714285714,
      "grad_norm": 0.1408321112394333,
      "learning_rate": 6.239047619047619e-06,
      "loss": 1.6839,
      "step": 14450
    },
    {
      "epoch": 2.065714285714286,
      "grad_norm": 0.1423533856868744,
      "learning_rate": 6.22952380952381e-06,
      "loss": 1.5018,
      "step": 14460
    },
    {
      "epoch": 2.067142857142857,
      "grad_norm": 0.057106222957372665,
      "learning_rate": 6.220000000000001e-06,
      "loss": 0.0,
      "step": 14470
    },
    {
      "epoch": 2.0685714285714285,
      "grad_norm": 0.10628800839185715,
      "learning_rate": 6.210476190476191e-06,
      "loss": 1.7663,
      "step": 14480
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.12140599638223648,
      "learning_rate": 6.200952380952382e-06,
      "loss": 1.9693,
      "step": 14490
    },
    {
      "epoch": 2.0714285714285716,
      "grad_norm": 0.1266903132200241,
      "learning_rate": 6.191428571428572e-06,
      "loss": 1.7249,
      "step": 14500
    },
    {
      "epoch": 2.072857142857143,
      "grad_norm": 0.05372342839837074,
      "learning_rate": 6.181904761904762e-06,
      "loss": 0.0,
      "step": 14510
    },
    {
      "epoch": 2.0742857142857143,
      "grad_norm": 0.06472672522068024,
      "learning_rate": 6.172380952380953e-06,
      "loss": 1.0732,
      "step": 14520
    },
    {
      "epoch": 2.0757142857142856,
      "grad_norm": 0.0637349933385849,
      "learning_rate": 6.162857142857144e-06,
      "loss": 0.0,
      "step": 14530
    },
    {
      "epoch": 2.077142857142857,
      "grad_norm": 0.07501973956823349,
      "learning_rate": 6.1533333333333334e-06,
      "loss": 3.4036,
      "step": 14540
    },
    {
      "epoch": 2.0785714285714287,
      "grad_norm": 0.047080907970666885,
      "learning_rate": 6.143809523809524e-06,
      "loss": 1.7605,
      "step": 14550
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.07821271568536758,
      "learning_rate": 6.134285714285715e-06,
      "loss": 3.4728,
      "step": 14560
    },
    {
      "epoch": 2.0814285714285714,
      "grad_norm": 0.05036992207169533,
      "learning_rate": 6.124761904761905e-06,
      "loss": 1.189,
      "step": 14570
    },
    {
      "epoch": 2.0828571428571427,
      "grad_norm": 6975678.5,
      "learning_rate": 6.115238095238095e-06,
      "loss": 1.6578,
      "step": 14580
    },
    {
      "epoch": 2.084285714285714,
      "grad_norm": 2996753.25,
      "learning_rate": 6.105714285714286e-06,
      "loss": 2.5826,
      "step": 14590
    },
    {
      "epoch": 2.085714285714286,
      "grad_norm": 0.06502128392457962,
      "learning_rate": 6.0961904761904765e-06,
      "loss": 0.0,
      "step": 14600
    },
    {
      "epoch": 2.087142857142857,
      "grad_norm": 2732043.5,
      "learning_rate": 6.086666666666667e-06,
      "loss": 3.5428,
      "step": 14610
    },
    {
      "epoch": 2.0885714285714285,
      "grad_norm": 0.09846863150596619,
      "learning_rate": 6.077142857142858e-06,
      "loss": 5.0597,
      "step": 14620
    },
    {
      "epoch": 2.09,
      "grad_norm": 33426.67578125,
      "learning_rate": 6.0676190476190476e-06,
      "loss": 0.0003,
      "step": 14630
    },
    {
      "epoch": 2.0914285714285716,
      "grad_norm": 8181629.0,
      "learning_rate": 6.058095238095238e-06,
      "loss": 3.5509,
      "step": 14640
    },
    {
      "epoch": 2.092857142857143,
      "grad_norm": 3.0912938117980957,
      "learning_rate": 6.048571428571429e-06,
      "loss": 1.6229,
      "step": 14650
    },
    {
      "epoch": 2.0942857142857143,
      "grad_norm": 0.12012362480163574,
      "learning_rate": 6.03904761904762e-06,
      "loss": 1.6435,
      "step": 14660
    },
    {
      "epoch": 2.0957142857142856,
      "grad_norm": 63434696.0,
      "learning_rate": 6.029523809523809e-06,
      "loss": 1.937,
      "step": 14670
    },
    {
      "epoch": 2.097142857142857,
      "grad_norm": 0.12049625813961029,
      "learning_rate": 6.02e-06,
      "loss": 5.1093,
      "step": 14680
    },
    {
      "epoch": 2.0985714285714288,
      "grad_norm": 0.07205535471439362,
      "learning_rate": 6.010476190476191e-06,
      "loss": 1.0516,
      "step": 14690
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.0772104561328888,
      "learning_rate": 6.000952380952382e-06,
      "loss": 0.0,
      "step": 14700
    },
    {
      "epoch": 2.1014285714285714,
      "grad_norm": 0.09244561940431595,
      "learning_rate": 5.991428571428571e-06,
      "loss": 1.6621,
      "step": 14710
    },
    {
      "epoch": 2.1028571428571428,
      "grad_norm": 777.942138671875,
      "learning_rate": 5.981904761904762e-06,
      "loss": 0.0,
      "step": 14720
    },
    {
      "epoch": 2.104285714285714,
      "grad_norm": 0.08648849278688431,
      "learning_rate": 5.972380952380953e-06,
      "loss": 0.0,
      "step": 14730
    },
    {
      "epoch": 2.105714285714286,
      "grad_norm": 0.11325788497924805,
      "learning_rate": 5.962857142857144e-06,
      "loss": 4.6851,
      "step": 14740
    },
    {
      "epoch": 2.107142857142857,
      "grad_norm": 0.07822078466415405,
      "learning_rate": 5.9533333333333345e-06,
      "loss": 1.6816,
      "step": 14750
    },
    {
      "epoch": 2.1085714285714285,
      "grad_norm": 0.09899558126926422,
      "learning_rate": 5.943809523809524e-06,
      "loss": 1.699,
      "step": 14760
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.08648914843797684,
      "learning_rate": 5.934285714285715e-06,
      "loss": 0.0,
      "step": 14770
    },
    {
      "epoch": 2.111428571428571,
      "grad_norm": 0.09733191877603531,
      "learning_rate": 5.924761904761906e-06,
      "loss": 1.7335,
      "step": 14780
    },
    {
      "epoch": 2.112857142857143,
      "grad_norm": 0.3498435914516449,
      "learning_rate": 5.915238095238096e-06,
      "loss": 1.0782,
      "step": 14790
    },
    {
      "epoch": 2.1142857142857143,
      "grad_norm": 2727630.0,
      "learning_rate": 5.905714285714286e-06,
      "loss": 3.3522,
      "step": 14800
    },
    {
      "epoch": 2.1157142857142857,
      "grad_norm": 0.07942131161689758,
      "learning_rate": 5.896190476190477e-06,
      "loss": 3.4571,
      "step": 14810
    },
    {
      "epoch": 2.117142857142857,
      "grad_norm": 0.11040477454662323,
      "learning_rate": 5.886666666666667e-06,
      "loss": 0.0,
      "step": 14820
    },
    {
      "epoch": 2.1185714285714288,
      "grad_norm": 3373920.0,
      "learning_rate": 5.877142857142858e-06,
      "loss": 4.948,
      "step": 14830
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.08390995860099792,
      "learning_rate": 5.867619047619048e-06,
      "loss": 3.3615,
      "step": 14840
    },
    {
      "epoch": 2.1214285714285714,
      "grad_norm": 0.08998879790306091,
      "learning_rate": 5.8580952380952384e-06,
      "loss": 3.3586,
      "step": 14850
    },
    {
      "epoch": 2.1228571428571428,
      "grad_norm": 1.4008159637451172,
      "learning_rate": 5.848571428571429e-06,
      "loss": 0.0,
      "step": 14860
    },
    {
      "epoch": 2.124285714285714,
      "grad_norm": 0.10807734727859497,
      "learning_rate": 5.83904761904762e-06,
      "loss": 1.668,
      "step": 14870
    },
    {
      "epoch": 2.125714285714286,
      "grad_norm": 0.07150576263666153,
      "learning_rate": 5.82952380952381e-06,
      "loss": 4.5241,
      "step": 14880
    },
    {
      "epoch": 2.1271428571428572,
      "grad_norm": 0.0659545436501503,
      "learning_rate": 5.82e-06,
      "loss": 1.7098,
      "step": 14890
    },
    {
      "epoch": 2.1285714285714286,
      "grad_norm": 0.30614548921585083,
      "learning_rate": 5.810476190476191e-06,
      "loss": 4.4622,
      "step": 14900
    },
    {
      "epoch": 2.13,
      "grad_norm": 2909250.25,
      "learning_rate": 5.8009523809523815e-06,
      "loss": 3.2135,
      "step": 14910
    },
    {
      "epoch": 2.1314285714285712,
      "grad_norm": 3443043.5,
      "learning_rate": 5.791428571428572e-06,
      "loss": 3.3063,
      "step": 14920
    },
    {
      "epoch": 2.132857142857143,
      "grad_norm": 0.0480378121137619,
      "learning_rate": 5.781904761904762e-06,
      "loss": 0.0,
      "step": 14930
    },
    {
      "epoch": 2.1342857142857143,
      "grad_norm": 0.09969612956047058,
      "learning_rate": 5.7723809523809526e-06,
      "loss": 0.0,
      "step": 14940
    },
    {
      "epoch": 2.1357142857142857,
      "grad_norm": 0.26433810591697693,
      "learning_rate": 5.762857142857143e-06,
      "loss": 3.3871,
      "step": 14950
    },
    {
      "epoch": 2.137142857142857,
      "grad_norm": 0.10193784534931183,
      "learning_rate": 5.753333333333334e-06,
      "loss": 1.6385,
      "step": 14960
    },
    {
      "epoch": 2.138571428571429,
      "grad_norm": 0.30335015058517456,
      "learning_rate": 5.7438095238095245e-06,
      "loss": 3.0222,
      "step": 14970
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.10467278212308884,
      "learning_rate": 5.734285714285714e-06,
      "loss": 3.2594,
      "step": 14980
    },
    {
      "epoch": 2.1414285714285715,
      "grad_norm": 0.08522117882966995,
      "learning_rate": 5.724761904761905e-06,
      "loss": 1.6325,
      "step": 14990
    },
    {
      "epoch": 2.142857142857143,
      "grad_norm": 0.11148612946271896,
      "learning_rate": 5.715238095238096e-06,
      "loss": 1.689,
      "step": 15000
    },
    {
      "epoch": 2.144285714285714,
      "grad_norm": 2809561.5,
      "learning_rate": 5.705714285714286e-06,
      "loss": 3.3394,
      "step": 15010
    },
    {
      "epoch": 2.145714285714286,
      "grad_norm": 0.3571983277797699,
      "learning_rate": 5.696190476190476e-06,
      "loss": 4.7706,
      "step": 15020
    },
    {
      "epoch": 2.1471428571428572,
      "grad_norm": 2886222.0,
      "learning_rate": 5.686666666666667e-06,
      "loss": 5.7614,
      "step": 15030
    },
    {
      "epoch": 2.1485714285714286,
      "grad_norm": 0.09598926454782486,
      "learning_rate": 5.677142857142857e-06,
      "loss": 0.0,
      "step": 15040
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.09624530375003815,
      "learning_rate": 5.667619047619049e-06,
      "loss": 2.4403,
      "step": 15050
    },
    {
      "epoch": 2.1514285714285712,
      "grad_norm": 0.06205407530069351,
      "learning_rate": 5.658095238095238e-06,
      "loss": 0.0,
      "step": 15060
    },
    {
      "epoch": 2.152857142857143,
      "grad_norm": 0.07353714853525162,
      "learning_rate": 5.6485714285714285e-06,
      "loss": 1.9402,
      "step": 15070
    },
    {
      "epoch": 2.1542857142857144,
      "grad_norm": 0.2658030092716217,
      "learning_rate": 5.63904761904762e-06,
      "loss": 0.0,
      "step": 15080
    },
    {
      "epoch": 2.1557142857142857,
      "grad_norm": 0.10514462739229202,
      "learning_rate": 5.629523809523811e-06,
      "loss": 5.2361,
      "step": 15090
    },
    {
      "epoch": 2.157142857142857,
      "grad_norm": 0.6446129083633423,
      "learning_rate": 5.620000000000001e-06,
      "loss": 0.0023,
      "step": 15100
    },
    {
      "epoch": 2.1585714285714284,
      "grad_norm": 0.03986475244164467,
      "learning_rate": 5.610476190476191e-06,
      "loss": 3.284,
      "step": 15110
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.060252316296100616,
      "learning_rate": 5.600952380952382e-06,
      "loss": 3.1824,
      "step": 15120
    },
    {
      "epoch": 2.1614285714285715,
      "grad_norm": 0.35835617780685425,
      "learning_rate": 5.591428571428572e-06,
      "loss": 1.6602,
      "step": 15130
    },
    {
      "epoch": 2.162857142857143,
      "grad_norm": 0.39134544134140015,
      "learning_rate": 5.581904761904763e-06,
      "loss": 0.0,
      "step": 15140
    },
    {
      "epoch": 2.164285714285714,
      "grad_norm": 0.07935457676649094,
      "learning_rate": 5.572380952380953e-06,
      "loss": 1.6676,
      "step": 15150
    },
    {
      "epoch": 2.1657142857142855,
      "grad_norm": 2769437.0,
      "learning_rate": 5.5628571428571434e-06,
      "loss": 1.7182,
      "step": 15160
    },
    {
      "epoch": 2.1671428571428573,
      "grad_norm": 0.10725737363100052,
      "learning_rate": 5.553333333333334e-06,
      "loss": 0.0,
      "step": 15170
    },
    {
      "epoch": 2.1685714285714286,
      "grad_norm": 0.06957859545946121,
      "learning_rate": 5.543809523809525e-06,
      "loss": 0.0,
      "step": 15180
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.0691806897521019,
      "learning_rate": 5.5342857142857145e-06,
      "loss": 2.3536,
      "step": 15190
    },
    {
      "epoch": 2.1714285714285713,
      "grad_norm": 3254016.5,
      "learning_rate": 5.524761904761905e-06,
      "loss": 1.6772,
      "step": 15200
    },
    {
      "epoch": 2.172857142857143,
      "grad_norm": 0.07849116623401642,
      "learning_rate": 5.515238095238096e-06,
      "loss": 0.0,
      "step": 15210
    },
    {
      "epoch": 2.1742857142857144,
      "grad_norm": 0.115604467689991,
      "learning_rate": 5.5057142857142865e-06,
      "loss": 1.7155,
      "step": 15220
    },
    {
      "epoch": 2.1757142857142857,
      "grad_norm": 0.09978499263525009,
      "learning_rate": 5.496190476190477e-06,
      "loss": 1.6274,
      "step": 15230
    },
    {
      "epoch": 2.177142857142857,
      "grad_norm": 0.27082136273384094,
      "learning_rate": 5.486666666666667e-06,
      "loss": 0.0,
      "step": 15240
    },
    {
      "epoch": 2.1785714285714284,
      "grad_norm": 0.2672935426235199,
      "learning_rate": 5.477142857142858e-06,
      "loss": 0.0,
      "step": 15250
    },
    {
      "epoch": 2.18,
      "grad_norm": 4605138.5,
      "learning_rate": 5.467619047619048e-06,
      "loss": 4.9695,
      "step": 15260
    },
    {
      "epoch": 2.1814285714285715,
      "grad_norm": 0.05150757357478142,
      "learning_rate": 5.458095238095239e-06,
      "loss": 3.3244,
      "step": 15270
    },
    {
      "epoch": 2.182857142857143,
      "grad_norm": 1.3933089971542358,
      "learning_rate": 5.448571428571429e-06,
      "loss": 3.2789,
      "step": 15280
    },
    {
      "epoch": 2.184285714285714,
      "grad_norm": 0.10463321954011917,
      "learning_rate": 5.439047619047619e-06,
      "loss": 1.6474,
      "step": 15290
    },
    {
      "epoch": 2.185714285714286,
      "grad_norm": 3051765.75,
      "learning_rate": 5.42952380952381e-06,
      "loss": 1.6964,
      "step": 15300
    },
    {
      "epoch": 2.1871428571428573,
      "grad_norm": 4116065.75,
      "learning_rate": 5.420000000000001e-06,
      "loss": 4.9331,
      "step": 15310
    },
    {
      "epoch": 2.1885714285714286,
      "grad_norm": 0.2887164056301117,
      "learning_rate": 5.4104761904761904e-06,
      "loss": 3.2353,
      "step": 15320
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.3334593772888184,
      "learning_rate": 5.400952380952381e-06,
      "loss": 3.3316,
      "step": 15330
    },
    {
      "epoch": 2.1914285714285713,
      "grad_norm": 0.07983963936567307,
      "learning_rate": 5.391428571428572e-06,
      "loss": 6.4405,
      "step": 15340
    },
    {
      "epoch": 2.192857142857143,
      "grad_norm": 0.06880481541156769,
      "learning_rate": 5.381904761904762e-06,
      "loss": 3.0001,
      "step": 15350
    },
    {
      "epoch": 2.1942857142857144,
      "grad_norm": 0.048788074404001236,
      "learning_rate": 5.372380952380953e-06,
      "loss": 1.7142,
      "step": 15360
    },
    {
      "epoch": 2.1957142857142857,
      "grad_norm": 0.10688213258981705,
      "learning_rate": 5.362857142857143e-06,
      "loss": 3.0561,
      "step": 15370
    },
    {
      "epoch": 2.197142857142857,
      "grad_norm": 0.34766247868537903,
      "learning_rate": 5.3533333333333335e-06,
      "loss": 0.0,
      "step": 15380
    },
    {
      "epoch": 2.1985714285714284,
      "grad_norm": 0.10766433924436569,
      "learning_rate": 5.343809523809524e-06,
      "loss": 3.2443,
      "step": 15390
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.08358312398195267,
      "learning_rate": 5.334285714285715e-06,
      "loss": 0.0,
      "step": 15400
    },
    {
      "epoch": 2.2014285714285715,
      "grad_norm": 0.04238459840416908,
      "learning_rate": 5.3247619047619046e-06,
      "loss": 1.6392,
      "step": 15410
    },
    {
      "epoch": 2.202857142857143,
      "grad_norm": 0.3380556106567383,
      "learning_rate": 5.315238095238095e-06,
      "loss": 0.0,
      "step": 15420
    },
    {
      "epoch": 2.204285714285714,
      "grad_norm": 0.33011671900749207,
      "learning_rate": 5.305714285714286e-06,
      "loss": 6.549,
      "step": 15430
    },
    {
      "epoch": 2.2057142857142855,
      "grad_norm": 0.289965957403183,
      "learning_rate": 5.296190476190477e-06,
      "loss": 3.0869,
      "step": 15440
    },
    {
      "epoch": 2.2071428571428573,
      "grad_norm": 0.0967579260468483,
      "learning_rate": 5.286666666666666e-06,
      "loss": 3.3883,
      "step": 15450
    },
    {
      "epoch": 2.2085714285714286,
      "grad_norm": 0.058047693222761154,
      "learning_rate": 5.277142857142857e-06,
      "loss": 0.0,
      "step": 15460
    },
    {
      "epoch": 2.21,
      "grad_norm": 2839421.75,
      "learning_rate": 5.2676190476190485e-06,
      "loss": 1.7079,
      "step": 15470
    },
    {
      "epoch": 2.2114285714285713,
      "grad_norm": 0.26879990100860596,
      "learning_rate": 5.258095238095239e-06,
      "loss": 1.6764,
      "step": 15480
    },
    {
      "epoch": 2.2128571428571426,
      "grad_norm": 0.3258875012397766,
      "learning_rate": 5.24857142857143e-06,
      "loss": 1.6501,
      "step": 15490
    },
    {
      "epoch": 2.2142857142857144,
      "grad_norm": 0.27874577045440674,
      "learning_rate": 5.2390476190476195e-06,
      "loss": 1.7241,
      "step": 15500
    },
    {
      "epoch": 2.2157142857142857,
      "grad_norm": 0.11473386734724045,
      "learning_rate": 5.22952380952381e-06,
      "loss": 3.3746,
      "step": 15510
    },
    {
      "epoch": 2.217142857142857,
      "grad_norm": 0.2812933921813965,
      "learning_rate": 5.220000000000001e-06,
      "loss": 1.6789,
      "step": 15520
    },
    {
      "epoch": 2.2185714285714284,
      "grad_norm": 0.38786086440086365,
      "learning_rate": 5.2104761904761915e-06,
      "loss": 3.3021,
      "step": 15530
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.12524664402008057,
      "learning_rate": 5.200952380952381e-06,
      "loss": 0.0,
      "step": 15540
    },
    {
      "epoch": 2.2214285714285715,
      "grad_norm": 0.08837586641311646,
      "learning_rate": 5.191428571428572e-06,
      "loss": 1.5956,
      "step": 15550
    },
    {
      "epoch": 2.222857142857143,
      "grad_norm": 0.2992364764213562,
      "learning_rate": 5.181904761904763e-06,
      "loss": 3.2263,
      "step": 15560
    },
    {
      "epoch": 2.224285714285714,
      "grad_norm": 1.0611920356750488,
      "learning_rate": 5.172380952380953e-06,
      "loss": 1.6667,
      "step": 15570
    },
    {
      "epoch": 2.2257142857142855,
      "grad_norm": 0.2899409830570221,
      "learning_rate": 5.162857142857143e-06,
      "loss": 4.8366,
      "step": 15580
    },
    {
      "epoch": 2.2271428571428573,
      "grad_norm": 0.08265358209609985,
      "learning_rate": 5.153333333333334e-06,
      "loss": 0.0,
      "step": 15590
    },
    {
      "epoch": 2.2285714285714286,
      "grad_norm": 0.06919978559017181,
      "learning_rate": 5.143809523809524e-06,
      "loss": 1.7305,
      "step": 15600
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.09931566566228867,
      "learning_rate": 5.134285714285715e-06,
      "loss": 0.0,
      "step": 15610
    },
    {
      "epoch": 2.2314285714285713,
      "grad_norm": 0.10318169742822647,
      "learning_rate": 5.124761904761906e-06,
      "loss": 3.2865,
      "step": 15620
    },
    {
      "epoch": 2.2328571428571427,
      "grad_norm": 0.07147489488124847,
      "learning_rate": 5.1152380952380954e-06,
      "loss": 5.0999,
      "step": 15630
    },
    {
      "epoch": 2.2342857142857144,
      "grad_norm": 0.08596616238355637,
      "learning_rate": 5.105714285714286e-06,
      "loss": 1.565,
      "step": 15640
    },
    {
      "epoch": 2.2357142857142858,
      "grad_norm": 0.06587905436754227,
      "learning_rate": 5.096190476190477e-06,
      "loss": 0.0,
      "step": 15650
    },
    {
      "epoch": 2.237142857142857,
      "grad_norm": 2814173.25,
      "learning_rate": 5.086666666666667e-06,
      "loss": 1.6645,
      "step": 15660
    },
    {
      "epoch": 2.2385714285714284,
      "grad_norm": 2956855.5,
      "learning_rate": 5.077142857142857e-06,
      "loss": 3.4178,
      "step": 15670
    },
    {
      "epoch": 2.24,
      "grad_norm": 2986248.25,
      "learning_rate": 5.067619047619048e-06,
      "loss": 1.7212,
      "step": 15680
    },
    {
      "epoch": 2.2414285714285715,
      "grad_norm": 0.07910823076963425,
      "learning_rate": 5.0580952380952385e-06,
      "loss": 1.5883,
      "step": 15690
    },
    {
      "epoch": 2.242857142857143,
      "grad_norm": 2781626.25,
      "learning_rate": 5.048571428571429e-06,
      "loss": 7.9868,
      "step": 15700
    },
    {
      "epoch": 2.244285714285714,
      "grad_norm": 0.26110953092575073,
      "learning_rate": 5.039047619047619e-06,
      "loss": 0.0,
      "step": 15710
    },
    {
      "epoch": 2.2457142857142856,
      "grad_norm": 0.05997714400291443,
      "learning_rate": 5.0295238095238096e-06,
      "loss": 4.8797,
      "step": 15720
    },
    {
      "epoch": 2.2471428571428573,
      "grad_norm": 0.3195594251155853,
      "learning_rate": 5.02e-06,
      "loss": 1.6772,
      "step": 15730
    },
    {
      "epoch": 2.2485714285714287,
      "grad_norm": 3957361.5,
      "learning_rate": 5.010476190476191e-06,
      "loss": 3.0183,
      "step": 15740
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.0997503325343132,
      "learning_rate": 5.0009523809523815e-06,
      "loss": 0.0,
      "step": 15750
    },
    {
      "epoch": 2.2514285714285713,
      "grad_norm": 0.10416605323553085,
      "learning_rate": 4.991428571428572e-06,
      "loss": 1.6339,
      "step": 15760
    },
    {
      "epoch": 2.2528571428571427,
      "grad_norm": 0.31879639625549316,
      "learning_rate": 4.981904761904762e-06,
      "loss": 3.0586,
      "step": 15770
    },
    {
      "epoch": 2.2542857142857144,
      "grad_norm": 0.06325098872184753,
      "learning_rate": 4.972380952380953e-06,
      "loss": 3.3486,
      "step": 15780
    },
    {
      "epoch": 2.255714285714286,
      "grad_norm": 0.059192996472120285,
      "learning_rate": 4.962857142857143e-06,
      "loss": 0.0,
      "step": 15790
    },
    {
      "epoch": 2.257142857142857,
      "grad_norm": 0.10434688627719879,
      "learning_rate": 4.953333333333334e-06,
      "loss": 1.6832,
      "step": 15800
    },
    {
      "epoch": 2.2585714285714285,
      "grad_norm": 2846855.0,
      "learning_rate": 4.943809523809524e-06,
      "loss": 3.2497,
      "step": 15810
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.09246858954429626,
      "learning_rate": 4.934285714285715e-06,
      "loss": 0.0,
      "step": 15820
    },
    {
      "epoch": 2.2614285714285716,
      "grad_norm": 0.31386512517929077,
      "learning_rate": 4.924761904761905e-06,
      "loss": 1.6747,
      "step": 15830
    },
    {
      "epoch": 2.262857142857143,
      "grad_norm": 0.06386987864971161,
      "learning_rate": 4.915238095238096e-06,
      "loss": 3.2771,
      "step": 15840
    },
    {
      "epoch": 2.2642857142857142,
      "grad_norm": 0.056886982172727585,
      "learning_rate": 4.9057142857142855e-06,
      "loss": 3.1885,
      "step": 15850
    },
    {
      "epoch": 2.2657142857142856,
      "grad_norm": 0.06914041191339493,
      "learning_rate": 4.896190476190477e-06,
      "loss": 0.0,
      "step": 15860
    },
    {
      "epoch": 2.267142857142857,
      "grad_norm": 0.26501938700675964,
      "learning_rate": 4.886666666666668e-06,
      "loss": 1.7135,
      "step": 15870
    },
    {
      "epoch": 2.2685714285714287,
      "grad_norm": 0.07405591011047363,
      "learning_rate": 4.877142857142857e-06,
      "loss": 2.9408,
      "step": 15880
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.2812320291996002,
      "learning_rate": 4.867619047619048e-06,
      "loss": 3.222,
      "step": 15890
    },
    {
      "epoch": 2.2714285714285714,
      "grad_norm": 0.0888664722442627,
      "learning_rate": 4.858095238095239e-06,
      "loss": 1.6959,
      "step": 15900
    },
    {
      "epoch": 2.2728571428571427,
      "grad_norm": 0.060566458851099014,
      "learning_rate": 4.848571428571429e-06,
      "loss": 0.0,
      "step": 15910
    },
    {
      "epoch": 2.2742857142857145,
      "grad_norm": 0.1080077663064003,
      "learning_rate": 4.839047619047619e-06,
      "loss": 3.3798,
      "step": 15920
    },
    {
      "epoch": 2.275714285714286,
      "grad_norm": 0.0831255242228508,
      "learning_rate": 4.82952380952381e-06,
      "loss": 1.6895,
      "step": 15930
    },
    {
      "epoch": 2.277142857142857,
      "grad_norm": 0.11022043228149414,
      "learning_rate": 4.8200000000000004e-06,
      "loss": 1.684,
      "step": 15940
    },
    {
      "epoch": 2.2785714285714285,
      "grad_norm": 0.09711848199367523,
      "learning_rate": 4.810476190476191e-06,
      "loss": 1.6093,
      "step": 15950
    },
    {
      "epoch": 2.2800000000000002,
      "grad_norm": 0.05898979678750038,
      "learning_rate": 4.800952380952381e-06,
      "loss": 1.566,
      "step": 15960
    },
    {
      "epoch": 2.2814285714285716,
      "grad_norm": 0.27049946784973145,
      "learning_rate": 4.7914285714285715e-06,
      "loss": 0.0,
      "step": 15970
    },
    {
      "epoch": 2.282857142857143,
      "grad_norm": 0.09208765625953674,
      "learning_rate": 4.781904761904762e-06,
      "loss": 0.0,
      "step": 15980
    },
    {
      "epoch": 2.2842857142857143,
      "grad_norm": 2881014.5,
      "learning_rate": 4.772380952380953e-06,
      "loss": 1.6906,
      "step": 15990
    },
    {
      "epoch": 2.2857142857142856,
      "grad_norm": 0.2896793484687805,
      "learning_rate": 4.7628571428571435e-06,
      "loss": 3.4643,
      "step": 16000
    },
    {
      "epoch": 2.2871428571428574,
      "grad_norm": 0.0979427918791771,
      "learning_rate": 4.753333333333333e-06,
      "loss": 0.0,
      "step": 16010
    },
    {
      "epoch": 2.2885714285714287,
      "grad_norm": 0.09446549415588379,
      "learning_rate": 4.743809523809525e-06,
      "loss": 3.1635,
      "step": 16020
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.07518912106752396,
      "learning_rate": 4.734285714285715e-06,
      "loss": 1.6742,
      "step": 16030
    },
    {
      "epoch": 2.2914285714285714,
      "grad_norm": 32258104.0,
      "learning_rate": 4.724761904761905e-06,
      "loss": 6.6608,
      "step": 16040
    },
    {
      "epoch": 2.2928571428571427,
      "grad_norm": 0.0604076161980629,
      "learning_rate": 4.715238095238096e-06,
      "loss": 1.6567,
      "step": 16050
    },
    {
      "epoch": 2.2942857142857145,
      "grad_norm": 0.098187655210495,
      "learning_rate": 4.7057142857142865e-06,
      "loss": 0.0,
      "step": 16060
    },
    {
      "epoch": 2.295714285714286,
      "grad_norm": 0.0853106826543808,
      "learning_rate": 4.696190476190476e-06,
      "loss": 1.6818,
      "step": 16070
    },
    {
      "epoch": 2.297142857142857,
      "grad_norm": 0.2672719359397888,
      "learning_rate": 4.686666666666667e-06,
      "loss": 0.0,
      "step": 16080
    },
    {
      "epoch": 2.2985714285714285,
      "grad_norm": 0.06797667592763901,
      "learning_rate": 4.677142857142858e-06,
      "loss": 0.0,
      "step": 16090
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2009556293487549,
      "learning_rate": 4.667619047619048e-06,
      "loss": 1.5836,
      "step": 16100
    },
    {
      "epoch": 2.3014285714285716,
      "grad_norm": 2915770.0,
      "learning_rate": 4.658095238095238e-06,
      "loss": 1.7077,
      "step": 16110
    },
    {
      "epoch": 2.302857142857143,
      "grad_norm": 2803063.0,
      "learning_rate": 4.648571428571429e-06,
      "loss": 3.347,
      "step": 16120
    },
    {
      "epoch": 2.3042857142857143,
      "grad_norm": 0.07251477241516113,
      "learning_rate": 4.639047619047619e-06,
      "loss": 1.558,
      "step": 16130
    },
    {
      "epoch": 2.3057142857142856,
      "grad_norm": 0.029625555500388145,
      "learning_rate": 4.62952380952381e-06,
      "loss": 0.0,
      "step": 16140
    },
    {
      "epoch": 2.307142857142857,
      "grad_norm": 0.08514898270368576,
      "learning_rate": 4.620000000000001e-06,
      "loss": 1.6979,
      "step": 16150
    },
    {
      "epoch": 2.3085714285714287,
      "grad_norm": 2893571.5,
      "learning_rate": 4.6104761904761905e-06,
      "loss": 3.2457,
      "step": 16160
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.06713303923606873,
      "learning_rate": 4.600952380952381e-06,
      "loss": 0.0,
      "step": 16170
    },
    {
      "epoch": 2.3114285714285714,
      "grad_norm": 0.09767282754182816,
      "learning_rate": 4.591428571428572e-06,
      "loss": 1.6745,
      "step": 16180
    },
    {
      "epoch": 2.3128571428571427,
      "grad_norm": 0.19637973606586456,
      "learning_rate": 4.581904761904762e-06,
      "loss": 1.6163,
      "step": 16190
    },
    {
      "epoch": 2.314285714285714,
      "grad_norm": 0.45996084809303284,
      "learning_rate": 4.572380952380952e-06,
      "loss": 4.7693,
      "step": 16200
    },
    {
      "epoch": 2.315714285714286,
      "grad_norm": 0.09207892417907715,
      "learning_rate": 4.562857142857144e-06,
      "loss": 0.0,
      "step": 16210
    },
    {
      "epoch": 2.317142857142857,
      "grad_norm": 0.050029706209897995,
      "learning_rate": 4.5533333333333335e-06,
      "loss": 3.2145,
      "step": 16220
    },
    {
      "epoch": 2.3185714285714285,
      "grad_norm": 0.06667905300855637,
      "learning_rate": 4.543809523809524e-06,
      "loss": 3.2048,
      "step": 16230
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.07480446994304657,
      "learning_rate": 4.534285714285715e-06,
      "loss": 0.0,
      "step": 16240
    },
    {
      "epoch": 2.3214285714285716,
      "grad_norm": 0.09020038694143295,
      "learning_rate": 4.5247619047619054e-06,
      "loss": 1.6794,
      "step": 16250
    },
    {
      "epoch": 2.322857142857143,
      "grad_norm": 0.0968518927693367,
      "learning_rate": 4.515238095238096e-06,
      "loss": 1.6825,
      "step": 16260
    },
    {
      "epoch": 2.3242857142857143,
      "grad_norm": 0.460802286863327,
      "learning_rate": 4.505714285714286e-06,
      "loss": 3.2959,
      "step": 16270
    },
    {
      "epoch": 2.3257142857142856,
      "grad_norm": 0.084690622985363,
      "learning_rate": 4.4961904761904765e-06,
      "loss": 0.0,
      "step": 16280
    },
    {
      "epoch": 2.327142857142857,
      "grad_norm": 0.10669111460447311,
      "learning_rate": 4.486666666666667e-06,
      "loss": 0.0,
      "step": 16290
    },
    {
      "epoch": 2.3285714285714287,
      "grad_norm": 0.09522966295480728,
      "learning_rate": 4.477142857142858e-06,
      "loss": 0.0,
      "step": 16300
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.05118352919816971,
      "learning_rate": 4.467619047619048e-06,
      "loss": 0.0,
      "step": 16310
    },
    {
      "epoch": 2.3314285714285714,
      "grad_norm": 0.06413181126117706,
      "learning_rate": 4.458095238095238e-06,
      "loss": 0.0,
      "step": 16320
    },
    {
      "epoch": 2.3328571428571427,
      "grad_norm": 0.03973051905632019,
      "learning_rate": 4.448571428571429e-06,
      "loss": 0.0,
      "step": 16330
    },
    {
      "epoch": 2.3342857142857145,
      "grad_norm": 0.06312943994998932,
      "learning_rate": 4.43904761904762e-06,
      "loss": 1.7322,
      "step": 16340
    },
    {
      "epoch": 2.335714285714286,
      "grad_norm": 0.08308935165405273,
      "learning_rate": 4.429523809523809e-06,
      "loss": 1.5113,
      "step": 16350
    },
    {
      "epoch": 2.337142857142857,
      "grad_norm": 0.059555403888225555,
      "learning_rate": 4.42e-06,
      "loss": 1.7049,
      "step": 16360
    },
    {
      "epoch": 2.3385714285714285,
      "grad_norm": 0.1178765520453453,
      "learning_rate": 4.410476190476191e-06,
      "loss": 1.7013,
      "step": 16370
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.1017795205116272,
      "learning_rate": 4.400952380952381e-06,
      "loss": 3.2502,
      "step": 16380
    },
    {
      "epoch": 2.3414285714285716,
      "grad_norm": 0.03671667352318764,
      "learning_rate": 4.391428571428572e-06,
      "loss": 0.0,
      "step": 16390
    },
    {
      "epoch": 2.342857142857143,
      "grad_norm": 0.08476561307907104,
      "learning_rate": 4.381904761904763e-06,
      "loss": 0.0,
      "step": 16400
    },
    {
      "epoch": 2.3442857142857143,
      "grad_norm": 0.0755600780248642,
      "learning_rate": 4.372380952380953e-06,
      "loss": 2.8738,
      "step": 16410
    },
    {
      "epoch": 2.3457142857142856,
      "grad_norm": 0.09751175343990326,
      "learning_rate": 4.362857142857143e-06,
      "loss": 3.1448,
      "step": 16420
    },
    {
      "epoch": 2.347142857142857,
      "grad_norm": 0.5000346302986145,
      "learning_rate": 4.353333333333334e-06,
      "loss": 3.3774,
      "step": 16430
    },
    {
      "epoch": 2.3485714285714288,
      "grad_norm": 0.0718575194478035,
      "learning_rate": 4.343809523809524e-06,
      "loss": 1.6773,
      "step": 16440
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.0735817700624466,
      "learning_rate": 4.334285714285715e-06,
      "loss": 1.6646,
      "step": 16450
    },
    {
      "epoch": 2.3514285714285714,
      "grad_norm": 0.372886061668396,
      "learning_rate": 4.324761904761905e-06,
      "loss": 0.0,
      "step": 16460
    },
    {
      "epoch": 2.3528571428571428,
      "grad_norm": 0.05335100740194321,
      "learning_rate": 4.3152380952380955e-06,
      "loss": 0.0,
      "step": 16470
    },
    {
      "epoch": 2.354285714285714,
      "grad_norm": 0.039554230868816376,
      "learning_rate": 4.305714285714286e-06,
      "loss": 3.3617,
      "step": 16480
    },
    {
      "epoch": 2.355714285714286,
      "grad_norm": 0.07735743373632431,
      "learning_rate": 4.296190476190477e-06,
      "loss": 1.6736,
      "step": 16490
    },
    {
      "epoch": 2.357142857142857,
      "grad_norm": 0.3676041066646576,
      "learning_rate": 4.2866666666666666e-06,
      "loss": 3.2293,
      "step": 16500
    },
    {
      "epoch": 2.3585714285714285,
      "grad_norm": 0.48214417695999146,
      "learning_rate": 4.277142857142857e-06,
      "loss": 1.3984,
      "step": 16510
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.1248529925942421,
      "learning_rate": 4.267619047619048e-06,
      "loss": 0.0,
      "step": 16520
    },
    {
      "epoch": 2.361428571428571,
      "grad_norm": 3022314.75,
      "learning_rate": 4.2580952380952385e-06,
      "loss": 1.9332,
      "step": 16530
    },
    {
      "epoch": 2.362857142857143,
      "grad_norm": 0.09354697167873383,
      "learning_rate": 4.248571428571429e-06,
      "loss": 1.6778,
      "step": 16540
    },
    {
      "epoch": 2.3642857142857143,
      "grad_norm": 0.06921490281820297,
      "learning_rate": 4.239047619047619e-06,
      "loss": 0.0,
      "step": 16550
    },
    {
      "epoch": 2.3657142857142857,
      "grad_norm": 0.08803007751703262,
      "learning_rate": 4.2295238095238105e-06,
      "loss": 1.6873,
      "step": 16560
    },
    {
      "epoch": 2.367142857142857,
      "grad_norm": 9311239.0,
      "learning_rate": 4.22e-06,
      "loss": 5.0899,
      "step": 16570
    },
    {
      "epoch": 2.3685714285714283,
      "grad_norm": 0.10647895932197571,
      "learning_rate": 4.210476190476191e-06,
      "loss": 0.0,
      "step": 16580
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.07748497277498245,
      "learning_rate": 4.200952380952381e-06,
      "loss": 0.0,
      "step": 16590
    },
    {
      "epoch": 2.3714285714285714,
      "grad_norm": 0.06016518548130989,
      "learning_rate": 4.191428571428572e-06,
      "loss": 3.3496,
      "step": 16600
    },
    {
      "epoch": 2.3728571428571428,
      "grad_norm": 0.0766996294260025,
      "learning_rate": 4.181904761904762e-06,
      "loss": 1.4268,
      "step": 16610
    },
    {
      "epoch": 2.374285714285714,
      "grad_norm": 0.057005781680345535,
      "learning_rate": 4.172380952380953e-06,
      "loss": 1.6779,
      "step": 16620
    },
    {
      "epoch": 2.375714285714286,
      "grad_norm": 0.05474896728992462,
      "learning_rate": 4.162857142857143e-06,
      "loss": 3.3418,
      "step": 16630
    },
    {
      "epoch": 2.3771428571428572,
      "grad_norm": 0.05456317216157913,
      "learning_rate": 4.153333333333334e-06,
      "loss": 1.7341,
      "step": 16640
    },
    {
      "epoch": 2.3785714285714286,
      "grad_norm": 0.05032799392938614,
      "learning_rate": 4.143809523809525e-06,
      "loss": 0.0,
      "step": 16650
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.05897330865263939,
      "learning_rate": 4.134285714285714e-06,
      "loss": 0.0,
      "step": 16660
    },
    {
      "epoch": 2.3814285714285717,
      "grad_norm": 0.0531931109726429,
      "learning_rate": 4.124761904761905e-06,
      "loss": 0.0,
      "step": 16670
    },
    {
      "epoch": 2.382857142857143,
      "grad_norm": 0.0720493420958519,
      "learning_rate": 4.115238095238096e-06,
      "loss": 0.0,
      "step": 16680
    },
    {
      "epoch": 2.3842857142857143,
      "grad_norm": 0.10720352083444595,
      "learning_rate": 4.105714285714286e-06,
      "loss": 0.0,
      "step": 16690
    },
    {
      "epoch": 2.3857142857142857,
      "grad_norm": 0.07564036548137665,
      "learning_rate": 4.096190476190476e-06,
      "loss": 1.7073,
      "step": 16700
    },
    {
      "epoch": 2.387142857142857,
      "grad_norm": 2711401.25,
      "learning_rate": 4.086666666666667e-06,
      "loss": 5.1172,
      "step": 16710
    },
    {
      "epoch": 2.388571428571429,
      "grad_norm": 0.26577529311180115,
      "learning_rate": 4.0771428571428574e-06,
      "loss": 1.6937,
      "step": 16720
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.06236695498228073,
      "learning_rate": 4.067619047619048e-06,
      "loss": 3.0377,
      "step": 16730
    },
    {
      "epoch": 2.3914285714285715,
      "grad_norm": 0.30098363757133484,
      "learning_rate": 4.058095238095238e-06,
      "loss": 4.9001,
      "step": 16740
    },
    {
      "epoch": 2.392857142857143,
      "grad_norm": 0.06779752671718597,
      "learning_rate": 4.0485714285714285e-06,
      "loss": 3.2565,
      "step": 16750
    },
    {
      "epoch": 2.394285714285714,
      "grad_norm": 3853934.75,
      "learning_rate": 4.039047619047619e-06,
      "loss": 3.0586,
      "step": 16760
    },
    {
      "epoch": 2.395714285714286,
      "grad_norm": 0.06110743060708046,
      "learning_rate": 4.02952380952381e-06,
      "loss": 0.0,
      "step": 16770
    },
    {
      "epoch": 2.3971428571428572,
      "grad_norm": 3014570.5,
      "learning_rate": 4.0200000000000005e-06,
      "loss": 1.599,
      "step": 16780
    },
    {
      "epoch": 2.3985714285714286,
      "grad_norm": 42.65125274658203,
      "learning_rate": 4.010476190476191e-06,
      "loss": 0.0,
      "step": 16790
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.04913210868835449,
      "learning_rate": 4.000952380952382e-06,
      "loss": 0.0,
      "step": 16800
    },
    {
      "epoch": 2.4014285714285712,
      "grad_norm": 0.08522026985883713,
      "learning_rate": 3.9914285714285716e-06,
      "loss": 1.6843,
      "step": 16810
    },
    {
      "epoch": 2.402857142857143,
      "grad_norm": 0.0804784744977951,
      "learning_rate": 3.981904761904762e-06,
      "loss": 1.5534,
      "step": 16820
    },
    {
      "epoch": 2.4042857142857144,
      "grad_norm": 0.047262631356716156,
      "learning_rate": 3.972380952380953e-06,
      "loss": 1.3748,
      "step": 16830
    },
    {
      "epoch": 2.4057142857142857,
      "grad_norm": 9009383.0,
      "learning_rate": 3.9628571428571435e-06,
      "loss": 3.0147,
      "step": 16840
    },
    {
      "epoch": 2.407142857142857,
      "grad_norm": 0.03871653601527214,
      "learning_rate": 3.953333333333333e-06,
      "loss": 1.7016,
      "step": 16850
    },
    {
      "epoch": 2.4085714285714284,
      "grad_norm": 0.09893841296434402,
      "learning_rate": 3.943809523809524e-06,
      "loss": 3.4248,
      "step": 16860
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.06427066773176193,
      "learning_rate": 3.934285714285715e-06,
      "loss": 1.6812,
      "step": 16870
    },
    {
      "epoch": 2.4114285714285715,
      "grad_norm": 7763845.0,
      "learning_rate": 3.924761904761905e-06,
      "loss": 1.6614,
      "step": 16880
    },
    {
      "epoch": 2.412857142857143,
      "grad_norm": 0.041593197733163834,
      "learning_rate": 3.915238095238095e-06,
      "loss": 0.0,
      "step": 16890
    },
    {
      "epoch": 2.414285714285714,
      "grad_norm": 3173323.0,
      "learning_rate": 3.905714285714286e-06,
      "loss": 1.6808,
      "step": 16900
    },
    {
      "epoch": 2.4157142857142855,
      "grad_norm": 0.05881905555725098,
      "learning_rate": 3.896190476190476e-06,
      "loss": 8.3901,
      "step": 16910
    },
    {
      "epoch": 2.4171428571428573,
      "grad_norm": 0.05989515036344528,
      "learning_rate": 3.886666666666667e-06,
      "loss": 0.0,
      "step": 16920
    },
    {
      "epoch": 2.4185714285714286,
      "grad_norm": 0.07776078581809998,
      "learning_rate": 3.877142857142858e-06,
      "loss": 6.7483,
      "step": 16930
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.09942237287759781,
      "learning_rate": 3.8676190476190475e-06,
      "loss": 3.209,
      "step": 16940
    },
    {
      "epoch": 2.4214285714285713,
      "grad_norm": 0.07308793812990189,
      "learning_rate": 3.858095238095239e-06,
      "loss": 0.0,
      "step": 16950
    },
    {
      "epoch": 2.422857142857143,
      "grad_norm": 0.34540876746177673,
      "learning_rate": 3.848571428571429e-06,
      "loss": 0.0,
      "step": 16960
    },
    {
      "epoch": 2.4242857142857144,
      "grad_norm": 1.9131780862808228,
      "learning_rate": 3.839047619047619e-06,
      "loss": 1.6503,
      "step": 16970
    },
    {
      "epoch": 2.4257142857142857,
      "grad_norm": 0.0556781031191349,
      "learning_rate": 3.82952380952381e-06,
      "loss": 3.4382,
      "step": 16980
    },
    {
      "epoch": 2.427142857142857,
      "grad_norm": 0.298822820186615,
      "learning_rate": 3.820000000000001e-06,
      "loss": 1.6455,
      "step": 16990
    },
    {
      "epoch": 2.4285714285714284,
      "grad_norm": 0.09113685041666031,
      "learning_rate": 3.8104761904761905e-06,
      "loss": 0.0,
      "step": 17000
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.1046040877699852,
      "learning_rate": 3.800952380952381e-06,
      "loss": 5.9733,
      "step": 17010
    },
    {
      "epoch": 2.4314285714285715,
      "grad_norm": 0.07390925288200378,
      "learning_rate": 3.7914285714285722e-06,
      "loss": 0.0,
      "step": 17020
    },
    {
      "epoch": 2.432857142857143,
      "grad_norm": 0.06506616622209549,
      "learning_rate": 3.781904761904762e-06,
      "loss": 0.0,
      "step": 17030
    },
    {
      "epoch": 2.434285714285714,
      "grad_norm": 0.08145973831415176,
      "learning_rate": 3.772380952380953e-06,
      "loss": 3.3614,
      "step": 17040
    },
    {
      "epoch": 2.435714285714286,
      "grad_norm": 3784933.5,
      "learning_rate": 3.762857142857143e-06,
      "loss": 5.7476,
      "step": 17050
    },
    {
      "epoch": 2.4371428571428573,
      "grad_norm": 0.30044957995414734,
      "learning_rate": 3.753333333333334e-06,
      "loss": 4.964,
      "step": 17060
    },
    {
      "epoch": 2.4385714285714286,
      "grad_norm": 22.729780197143555,
      "learning_rate": 3.743809523809524e-06,
      "loss": 0.0,
      "step": 17070
    },
    {
      "epoch": 2.44,
      "grad_norm": 6178321.0,
      "learning_rate": 3.734285714285715e-06,
      "loss": 1.6828,
      "step": 17080
    },
    {
      "epoch": 2.4414285714285713,
      "grad_norm": 0.06872838735580444,
      "learning_rate": 3.724761904761905e-06,
      "loss": 0.0,
      "step": 17090
    },
    {
      "epoch": 2.442857142857143,
      "grad_norm": 0.06820473819971085,
      "learning_rate": 3.7152380952380957e-06,
      "loss": 0.0,
      "step": 17100
    },
    {
      "epoch": 2.4442857142857144,
      "grad_norm": 0.05943625792860985,
      "learning_rate": 3.705714285714286e-06,
      "loss": 0.0,
      "step": 17110
    },
    {
      "epoch": 2.4457142857142857,
      "grad_norm": 0.06819573789834976,
      "learning_rate": 3.6961904761904766e-06,
      "loss": 1.5138,
      "step": 17120
    },
    {
      "epoch": 2.447142857142857,
      "grad_norm": 10693608.0,
      "learning_rate": 3.686666666666667e-06,
      "loss": 3.223,
      "step": 17130
    },
    {
      "epoch": 2.4485714285714284,
      "grad_norm": 0.07415792346000671,
      "learning_rate": 3.6771428571428575e-06,
      "loss": 1.6909,
      "step": 17140
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.03616120666265488,
      "learning_rate": 3.667619047619048e-06,
      "loss": 1.6739,
      "step": 17150
    },
    {
      "epoch": 2.4514285714285715,
      "grad_norm": 0.06760001927614212,
      "learning_rate": 3.6580952380952383e-06,
      "loss": 0.0,
      "step": 17160
    },
    {
      "epoch": 2.452857142857143,
      "grad_norm": 11761485.0,
      "learning_rate": 3.648571428571429e-06,
      "loss": 1.4806,
      "step": 17170
    },
    {
      "epoch": 2.454285714285714,
      "grad_norm": 0.05320197716355324,
      "learning_rate": 3.639047619047619e-06,
      "loss": 0.0,
      "step": 17180
    },
    {
      "epoch": 2.4557142857142855,
      "grad_norm": 53536896.0,
      "learning_rate": 3.62952380952381e-06,
      "loss": 1.6409,
      "step": 17190
    },
    {
      "epoch": 2.4571428571428573,
      "grad_norm": 0.05776548758149147,
      "learning_rate": 3.62e-06,
      "loss": 1.7264,
      "step": 17200
    },
    {
      "epoch": 2.4585714285714286,
      "grad_norm": 0.0441688634455204,
      "learning_rate": 3.6104761904761907e-06,
      "loss": 1.6985,
      "step": 17210
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.06203274056315422,
      "learning_rate": 3.600952380952381e-06,
      "loss": 1.7199,
      "step": 17220
    },
    {
      "epoch": 2.4614285714285713,
      "grad_norm": 0.09387984871864319,
      "learning_rate": 3.591428571428572e-06,
      "loss": 0.0,
      "step": 17230
    },
    {
      "epoch": 2.4628571428571426,
      "grad_norm": 0.05327225476503372,
      "learning_rate": 3.581904761904762e-06,
      "loss": 0.0,
      "step": 17240
    },
    {
      "epoch": 2.4642857142857144,
      "grad_norm": 3231185.75,
      "learning_rate": 3.572380952380953e-06,
      "loss": 5.1918,
      "step": 17250
    },
    {
      "epoch": 2.4657142857142857,
      "grad_norm": 0.07702817022800446,
      "learning_rate": 3.562857142857143e-06,
      "loss": 1.6719,
      "step": 17260
    },
    {
      "epoch": 2.467142857142857,
      "grad_norm": 0.10606199502944946,
      "learning_rate": 3.5533333333333338e-06,
      "loss": 0.0,
      "step": 17270
    },
    {
      "epoch": 2.4685714285714284,
      "grad_norm": 0.3135564923286438,
      "learning_rate": 3.5438095238095244e-06,
      "loss": 1.6533,
      "step": 17280
    },
    {
      "epoch": 2.4699999999999998,
      "grad_norm": 0.08659876883029938,
      "learning_rate": 3.5342857142857146e-06,
      "loss": 3.3562,
      "step": 17290
    },
    {
      "epoch": 2.4714285714285715,
      "grad_norm": 0.08222103118896484,
      "learning_rate": 3.5247619047619053e-06,
      "loss": 3.4089,
      "step": 17300
    },
    {
      "epoch": 2.472857142857143,
      "grad_norm": 0.08834532648324966,
      "learning_rate": 3.5152380952380955e-06,
      "loss": 2.991,
      "step": 17310
    },
    {
      "epoch": 2.474285714285714,
      "grad_norm": 64592196.0,
      "learning_rate": 3.505714285714286e-06,
      "loss": 1.4342,
      "step": 17320
    },
    {
      "epoch": 2.4757142857142855,
      "grad_norm": 0.07137088477611542,
      "learning_rate": 3.4961904761904764e-06,
      "loss": 0.0,
      "step": 17330
    },
    {
      "epoch": 2.4771428571428573,
      "grad_norm": 0.07036120444536209,
      "learning_rate": 3.486666666666667e-06,
      "loss": 3.1073,
      "step": 17340
    },
    {
      "epoch": 2.4785714285714286,
      "grad_norm": 0.6350951194763184,
      "learning_rate": 3.4771428571428572e-06,
      "loss": 4.984,
      "step": 17350
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.07385493069887161,
      "learning_rate": 3.467619047619048e-06,
      "loss": 1.6688,
      "step": 17360
    },
    {
      "epoch": 2.4814285714285713,
      "grad_norm": 2903372.0,
      "learning_rate": 3.458095238095238e-06,
      "loss": 3.374,
      "step": 17370
    },
    {
      "epoch": 2.482857142857143,
      "grad_norm": 0.08956115692853928,
      "learning_rate": 3.4485714285714288e-06,
      "loss": 3.3739,
      "step": 17380
    },
    {
      "epoch": 2.4842857142857144,
      "grad_norm": 0.04341549798846245,
      "learning_rate": 3.439047619047619e-06,
      "loss": 3.272,
      "step": 17390
    },
    {
      "epoch": 2.4857142857142858,
      "grad_norm": 0.04856400936841965,
      "learning_rate": 3.4295238095238096e-06,
      "loss": 0.0,
      "step": 17400
    },
    {
      "epoch": 2.487142857142857,
      "grad_norm": 0.2765163779258728,
      "learning_rate": 3.4200000000000007e-06,
      "loss": 0.0,
      "step": 17410
    },
    {
      "epoch": 2.4885714285714284,
      "grad_norm": 0.08408670127391815,
      "learning_rate": 3.4104761904761905e-06,
      "loss": 1.685,
      "step": 17420
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.0643061101436615,
      "learning_rate": 3.4009523809523816e-06,
      "loss": 3.3407,
      "step": 17430
    },
    {
      "epoch": 2.4914285714285715,
      "grad_norm": 0.04934505373239517,
      "learning_rate": 3.391428571428572e-06,
      "loss": 0.0,
      "step": 17440
    },
    {
      "epoch": 2.492857142857143,
      "grad_norm": 0.4361671209335327,
      "learning_rate": 3.3819047619047625e-06,
      "loss": 1.5902,
      "step": 17450
    },
    {
      "epoch": 2.494285714285714,
      "grad_norm": 0.04744811728596687,
      "learning_rate": 3.3723809523809527e-06,
      "loss": 3.4345,
      "step": 17460
    },
    {
      "epoch": 2.4957142857142856,
      "grad_norm": 2.1954827308654785,
      "learning_rate": 3.3628571428571433e-06,
      "loss": 3.2107,
      "step": 17470
    },
    {
      "epoch": 2.4971428571428573,
      "grad_norm": 0.04777258262038231,
      "learning_rate": 3.3533333333333336e-06,
      "loss": 0.0,
      "step": 17480
    },
    {
      "epoch": 2.4985714285714287,
      "grad_norm": 0.041144609451293945,
      "learning_rate": 3.343809523809524e-06,
      "loss": 1.7617,
      "step": 17490
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.1053144782781601,
      "learning_rate": 3.3342857142857144e-06,
      "loss": 3.1447,
      "step": 17500
    },
    {
      "epoch": 2.5014285714285713,
      "grad_norm": 0.05700564756989479,
      "learning_rate": 3.324761904761905e-06,
      "loss": 1.6158,
      "step": 17510
    },
    {
      "epoch": 2.5028571428571427,
      "grad_norm": 0.0460306815803051,
      "learning_rate": 3.3152380952380953e-06,
      "loss": 0.0,
      "step": 17520
    },
    {
      "epoch": 2.5042857142857144,
      "grad_norm": 0.10478309541940689,
      "learning_rate": 3.305714285714286e-06,
      "loss": 1.5227,
      "step": 17530
    },
    {
      "epoch": 2.505714285714286,
      "grad_norm": 1069625984.0,
      "learning_rate": 3.2961904761904766e-06,
      "loss": 0.5972,
      "step": 17540
    },
    {
      "epoch": 2.507142857142857,
      "grad_norm": 0.0458795391023159,
      "learning_rate": 3.286666666666667e-06,
      "loss": 1.6755,
      "step": 17550
    },
    {
      "epoch": 2.5085714285714285,
      "grad_norm": 0.059270717203617096,
      "learning_rate": 3.2771428571428575e-06,
      "loss": 0.0,
      "step": 17560
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.0470893494784832,
      "learning_rate": 3.2676190476190477e-06,
      "loss": 1.6944,
      "step": 17570
    },
    {
      "epoch": 2.5114285714285716,
      "grad_norm": 0.06325303763151169,
      "learning_rate": 3.2580952380952383e-06,
      "loss": 1.6831,
      "step": 17580
    },
    {
      "epoch": 2.512857142857143,
      "grad_norm": 0.295131653547287,
      "learning_rate": 3.2485714285714286e-06,
      "loss": 0.0,
      "step": 17590
    },
    {
      "epoch": 2.5142857142857142,
      "grad_norm": 0.06939782202243805,
      "learning_rate": 3.2390476190476196e-06,
      "loss": 1.6882,
      "step": 17600
    },
    {
      "epoch": 2.5157142857142856,
      "grad_norm": 0.05838878080248833,
      "learning_rate": 3.2295238095238094e-06,
      "loss": 0.0,
      "step": 17610
    },
    {
      "epoch": 2.517142857142857,
      "grad_norm": 3095949.75,
      "learning_rate": 3.2200000000000005e-06,
      "loss": 3.0871,
      "step": 17620
    },
    {
      "epoch": 2.5185714285714287,
      "grad_norm": 0.09218005836009979,
      "learning_rate": 3.2104761904761907e-06,
      "loss": 1.3364,
      "step": 17630
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.10289866477251053,
      "learning_rate": 3.2009523809523814e-06,
      "loss": 1.7715,
      "step": 17640
    },
    {
      "epoch": 2.5214285714285714,
      "grad_norm": 8925778.0,
      "learning_rate": 3.1914285714285716e-06,
      "loss": 3.3225,
      "step": 17650
    },
    {
      "epoch": 2.522857142857143,
      "grad_norm": 2948789.0,
      "learning_rate": 3.1819047619047623e-06,
      "loss": 3.3065,
      "step": 17660
    },
    {
      "epoch": 2.524285714285714,
      "grad_norm": 0.11285129189491272,
      "learning_rate": 3.172380952380953e-06,
      "loss": 0.0,
      "step": 17670
    },
    {
      "epoch": 2.525714285714286,
      "grad_norm": 0.1009829044342041,
      "learning_rate": 3.162857142857143e-06,
      "loss": 3.2312,
      "step": 17680
    },
    {
      "epoch": 2.527142857142857,
      "grad_norm": 0.07325734943151474,
      "learning_rate": 3.1533333333333338e-06,
      "loss": 0.0,
      "step": 17690
    },
    {
      "epoch": 2.5285714285714285,
      "grad_norm": 0.10704729706048965,
      "learning_rate": 3.143809523809524e-06,
      "loss": 1.6619,
      "step": 17700
    },
    {
      "epoch": 2.5300000000000002,
      "grad_norm": 0.05745457112789154,
      "learning_rate": 3.1342857142857146e-06,
      "loss": 0.0,
      "step": 17710
    },
    {
      "epoch": 2.5314285714285716,
      "grad_norm": 0.06917860358953476,
      "learning_rate": 3.124761904761905e-06,
      "loss": 1.7277,
      "step": 17720
    },
    {
      "epoch": 2.532857142857143,
      "grad_norm": 0.07845360785722733,
      "learning_rate": 3.1152380952380955e-06,
      "loss": 1.6234,
      "step": 17730
    },
    {
      "epoch": 2.5342857142857143,
      "grad_norm": 0.30625686049461365,
      "learning_rate": 3.1057142857142857e-06,
      "loss": 1.5345,
      "step": 17740
    },
    {
      "epoch": 2.5357142857142856,
      "grad_norm": 0.3363185524940491,
      "learning_rate": 3.0961904761904764e-06,
      "loss": 1.515,
      "step": 17750
    },
    {
      "epoch": 2.5371428571428574,
      "grad_norm": 0.45407089591026306,
      "learning_rate": 3.0866666666666666e-06,
      "loss": 4.9699,
      "step": 17760
    },
    {
      "epoch": 2.5385714285714287,
      "grad_norm": 0.06678595393896103,
      "learning_rate": 3.0771428571428573e-06,
      "loss": 5.1696,
      "step": 17770
    },
    {
      "epoch": 2.54,
      "grad_norm": 2969121.75,
      "learning_rate": 3.0676190476190475e-06,
      "loss": 4.9455,
      "step": 17780
    },
    {
      "epoch": 2.5414285714285714,
      "grad_norm": 0.06431320309638977,
      "learning_rate": 3.058095238095238e-06,
      "loss": 1.6221,
      "step": 17790
    },
    {
      "epoch": 2.5428571428571427,
      "grad_norm": 0.3824237883090973,
      "learning_rate": 3.048571428571429e-06,
      "loss": 0.0,
      "step": 17800
    },
    {
      "epoch": 2.5442857142857145,
      "grad_norm": 0.5300050973892212,
      "learning_rate": 3.0390476190476194e-06,
      "loss": 0.0,
      "step": 17810
    },
    {
      "epoch": 2.545714285714286,
      "grad_norm": 0.057281386107206345,
      "learning_rate": 3.02952380952381e-06,
      "loss": 0.0,
      "step": 17820
    },
    {
      "epoch": 2.547142857142857,
      "grad_norm": 0.07546645402908325,
      "learning_rate": 3.0200000000000003e-06,
      "loss": 3.2156,
      "step": 17830
    },
    {
      "epoch": 2.5485714285714285,
      "grad_norm": 0.0575578436255455,
      "learning_rate": 3.010476190476191e-06,
      "loss": 3.4367,
      "step": 17840
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.4490925967693329,
      "learning_rate": 3.000952380952381e-06,
      "loss": 1.6858,
      "step": 17850
    },
    {
      "epoch": 2.5514285714285716,
      "grad_norm": 0.07545772194862366,
      "learning_rate": 2.991428571428572e-06,
      "loss": 0.0,
      "step": 17860
    },
    {
      "epoch": 2.552857142857143,
      "grad_norm": 0.06682203710079193,
      "learning_rate": 2.981904761904762e-06,
      "loss": 0.0,
      "step": 17870
    },
    {
      "epoch": 2.5542857142857143,
      "grad_norm": 0.07506732642650604,
      "learning_rate": 2.9723809523809527e-06,
      "loss": 3.2695,
      "step": 17880
    },
    {
      "epoch": 2.5557142857142856,
      "grad_norm": 0.11093661189079285,
      "learning_rate": 2.962857142857143e-06,
      "loss": 0.0,
      "step": 17890
    },
    {
      "epoch": 2.557142857142857,
      "grad_norm": 0.06397268921136856,
      "learning_rate": 2.9533333333333336e-06,
      "loss": 0.0,
      "step": 17900
    },
    {
      "epoch": 2.5585714285714287,
      "grad_norm": 0.04741330444812775,
      "learning_rate": 2.943809523809524e-06,
      "loss": 1.7285,
      "step": 17910
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.2582855820655823,
      "learning_rate": 2.9342857142857144e-06,
      "loss": 0.0,
      "step": 17920
    },
    {
      "epoch": 2.5614285714285714,
      "grad_norm": 0.05902435630559921,
      "learning_rate": 2.924761904761905e-06,
      "loss": 6.5967,
      "step": 17930
    },
    {
      "epoch": 2.5628571428571427,
      "grad_norm": 0.07338949292898178,
      "learning_rate": 2.9152380952380953e-06,
      "loss": 0.0,
      "step": 17940
    },
    {
      "epoch": 2.564285714285714,
      "grad_norm": 0.06216856464743614,
      "learning_rate": 2.905714285714286e-06,
      "loss": 0.0,
      "step": 17950
    },
    {
      "epoch": 2.565714285714286,
      "grad_norm": 0.30222681164741516,
      "learning_rate": 2.896190476190476e-06,
      "loss": 0.0,
      "step": 17960
    },
    {
      "epoch": 2.567142857142857,
      "grad_norm": 0.35392439365386963,
      "learning_rate": 2.8866666666666673e-06,
      "loss": 2.303,
      "step": 17970
    },
    {
      "epoch": 2.5685714285714285,
      "grad_norm": 187554256.0,
      "learning_rate": 2.877142857142857e-06,
      "loss": 1.487,
      "step": 17980
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.04278832674026489,
      "learning_rate": 2.867619047619048e-06,
      "loss": 3.3872,
      "step": 17990
    },
    {
      "epoch": 2.571428571428571,
      "grad_norm": 0.055504631251096725,
      "learning_rate": 2.858095238095238e-06,
      "loss": 0.528,
      "step": 18000
    },
    {
      "epoch": 2.572857142857143,
      "grad_norm": 0.05604641139507294,
      "learning_rate": 2.848571428571429e-06,
      "loss": 0.0,
      "step": 18010
    },
    {
      "epoch": 2.5742857142857143,
      "grad_norm": 1.506832242012024,
      "learning_rate": 2.8390476190476192e-06,
      "loss": 4.2692,
      "step": 18020
    },
    {
      "epoch": 2.5757142857142856,
      "grad_norm": 0.06202215328812599,
      "learning_rate": 2.82952380952381e-06,
      "loss": 1.7136,
      "step": 18030
    },
    {
      "epoch": 2.5771428571428574,
      "grad_norm": 0.25398728251457214,
      "learning_rate": 2.82e-06,
      "loss": 1.7112,
      "step": 18040
    },
    {
      "epoch": 2.5785714285714287,
      "grad_norm": 0.07933187484741211,
      "learning_rate": 2.8104761904761908e-06,
      "loss": 0.0,
      "step": 18050
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.03971269726753235,
      "learning_rate": 2.8009523809523814e-06,
      "loss": 0.0,
      "step": 18060
    },
    {
      "epoch": 2.5814285714285714,
      "grad_norm": 0.08884910494089127,
      "learning_rate": 2.7914285714285716e-06,
      "loss": 6.7482,
      "step": 18070
    },
    {
      "epoch": 2.5828571428571427,
      "grad_norm": 0.10859435796737671,
      "learning_rate": 2.7819047619047623e-06,
      "loss": 0.0,
      "step": 18080
    },
    {
      "epoch": 2.5842857142857145,
      "grad_norm": 0.10067767649888992,
      "learning_rate": 2.7723809523809525e-06,
      "loss": 3.38,
      "step": 18090
    },
    {
      "epoch": 2.585714285714286,
      "grad_norm": 0.05419214814901352,
      "learning_rate": 2.762857142857143e-06,
      "loss": 4.9543,
      "step": 18100
    },
    {
      "epoch": 2.587142857142857,
      "grad_norm": 0.058135055005550385,
      "learning_rate": 2.7533333333333334e-06,
      "loss": 0.0,
      "step": 18110
    },
    {
      "epoch": 2.5885714285714285,
      "grad_norm": 0.06333422660827637,
      "learning_rate": 2.743809523809524e-06,
      "loss": 1.6264,
      "step": 18120
    },
    {
      "epoch": 2.59,
      "grad_norm": 2779809.5,
      "learning_rate": 2.7342857142857142e-06,
      "loss": 5.0192,
      "step": 18130
    },
    {
      "epoch": 2.5914285714285716,
      "grad_norm": 0.09501920640468597,
      "learning_rate": 2.724761904761905e-06,
      "loss": 4.8441,
      "step": 18140
    },
    {
      "epoch": 2.592857142857143,
      "grad_norm": 0.06551498174667358,
      "learning_rate": 2.715238095238095e-06,
      "loss": 0.0,
      "step": 18150
    },
    {
      "epoch": 2.5942857142857143,
      "grad_norm": 0.06673242151737213,
      "learning_rate": 2.7057142857142858e-06,
      "loss": 0.0,
      "step": 18160
    },
    {
      "epoch": 2.5957142857142856,
      "grad_norm": 0.6227743029594421,
      "learning_rate": 2.696190476190476e-06,
      "loss": 1.6945,
      "step": 18170
    },
    {
      "epoch": 2.597142857142857,
      "grad_norm": 0.40244805812835693,
      "learning_rate": 2.686666666666667e-06,
      "loss": 3.2879,
      "step": 18180
    },
    {
      "epoch": 2.5985714285714288,
      "grad_norm": 0.06687512993812561,
      "learning_rate": 2.6771428571428577e-06,
      "loss": 3.4678,
      "step": 18190
    },
    {
      "epoch": 2.6,
      "grad_norm": 3184673.0,
      "learning_rate": 2.667619047619048e-06,
      "loss": 4.9847,
      "step": 18200
    },
    {
      "epoch": 2.6014285714285714,
      "grad_norm": 0.44304364919662476,
      "learning_rate": 2.6580952380952386e-06,
      "loss": 1.6914,
      "step": 18210
    },
    {
      "epoch": 2.6028571428571428,
      "grad_norm": 0.06762910634279251,
      "learning_rate": 2.648571428571429e-06,
      "loss": 0.0,
      "step": 18220
    },
    {
      "epoch": 2.604285714285714,
      "grad_norm": 0.07258135080337524,
      "learning_rate": 2.6390476190476195e-06,
      "loss": 0.0,
      "step": 18230
    },
    {
      "epoch": 2.605714285714286,
      "grad_norm": 0.2826042175292969,
      "learning_rate": 2.6295238095238097e-06,
      "loss": 0.2615,
      "step": 18240
    },
    {
      "epoch": 2.607142857142857,
      "grad_norm": 0.04950680956244469,
      "learning_rate": 2.6200000000000003e-06,
      "loss": 1.6256,
      "step": 18250
    },
    {
      "epoch": 2.6085714285714285,
      "grad_norm": 0.06226107478141785,
      "learning_rate": 2.6104761904761906e-06,
      "loss": 1.5928,
      "step": 18260
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.31696853041648865,
      "learning_rate": 2.600952380952381e-06,
      "loss": 1.7064,
      "step": 18270
    },
    {
      "epoch": 2.611428571428571,
      "grad_norm": 0.05034045875072479,
      "learning_rate": 2.5914285714285714e-06,
      "loss": 0.0,
      "step": 18280
    },
    {
      "epoch": 2.612857142857143,
      "grad_norm": 0.07165046781301498,
      "learning_rate": 2.581904761904762e-06,
      "loss": 3.1647,
      "step": 18290
    },
    {
      "epoch": 2.6142857142857143,
      "grad_norm": 0.04643610492348671,
      "learning_rate": 2.5723809523809527e-06,
      "loss": 0.0,
      "step": 18300
    },
    {
      "epoch": 2.6157142857142857,
      "grad_norm": 0.305340439081192,
      "learning_rate": 2.562857142857143e-06,
      "loss": 0.0,
      "step": 18310
    },
    {
      "epoch": 2.617142857142857,
      "grad_norm": 0.09750977158546448,
      "learning_rate": 2.5533333333333336e-06,
      "loss": 1.7608,
      "step": 18320
    },
    {
      "epoch": 2.6185714285714283,
      "grad_norm": 4331163.0,
      "learning_rate": 2.543809523809524e-06,
      "loss": 3.2775,
      "step": 18330
    },
    {
      "epoch": 2.62,
      "grad_norm": 2767518.25,
      "learning_rate": 2.534285714285715e-06,
      "loss": 5.0515,
      "step": 18340
    },
    {
      "epoch": 2.6214285714285714,
      "grad_norm": 0.041225604712963104,
      "learning_rate": 2.5247619047619047e-06,
      "loss": 0.0,
      "step": 18350
    },
    {
      "epoch": 2.6228571428571428,
      "grad_norm": 0.06597528606653214,
      "learning_rate": 2.5152380952380958e-06,
      "loss": 3.2896,
      "step": 18360
    },
    {
      "epoch": 2.6242857142857146,
      "grad_norm": 0.08355963230133057,
      "learning_rate": 2.5057142857142856e-06,
      "loss": 1.5996,
      "step": 18370
    },
    {
      "epoch": 2.6257142857142854,
      "grad_norm": 0.31941354274749756,
      "learning_rate": 2.4961904761904766e-06,
      "loss": 1.6779,
      "step": 18380
    },
    {
      "epoch": 2.6271428571428572,
      "grad_norm": 0.0897538810968399,
      "learning_rate": 2.486666666666667e-06,
      "loss": 1.6235,
      "step": 18390
    },
    {
      "epoch": 2.6285714285714286,
      "grad_norm": 5837882.0,
      "learning_rate": 2.4771428571428575e-06,
      "loss": 3.1549,
      "step": 18400
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.12462801486253738,
      "learning_rate": 2.4676190476190477e-06,
      "loss": 3.2939,
      "step": 18410
    },
    {
      "epoch": 2.6314285714285717,
      "grad_norm": 0.06251438707113266,
      "learning_rate": 2.4580952380952384e-06,
      "loss": 1.5813,
      "step": 18420
    },
    {
      "epoch": 2.632857142857143,
      "grad_norm": 3717273.25,
      "learning_rate": 2.4485714285714286e-06,
      "loss": 3.3157,
      "step": 18430
    },
    {
      "epoch": 2.6342857142857143,
      "grad_norm": 0.051019422709941864,
      "learning_rate": 2.4390476190476193e-06,
      "loss": 1.6976,
      "step": 18440
    },
    {
      "epoch": 2.6357142857142857,
      "grad_norm": 0.3727816343307495,
      "learning_rate": 2.4295238095238095e-06,
      "loss": 1.5919,
      "step": 18450
    },
    {
      "epoch": 2.637142857142857,
      "grad_norm": 0.08816228061914444,
      "learning_rate": 2.42e-06,
      "loss": 0.0,
      "step": 18460
    },
    {
      "epoch": 2.638571428571429,
      "grad_norm": 2.2024283409118652,
      "learning_rate": 2.4104761904761908e-06,
      "loss": 0.0,
      "step": 18470
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.05471896752715111,
      "learning_rate": 2.400952380952381e-06,
      "loss": 0.0,
      "step": 18480
    },
    {
      "epoch": 2.6414285714285715,
      "grad_norm": 0.0527111180126667,
      "learning_rate": 2.3914285714285716e-06,
      "loss": 0.0,
      "step": 18490
    },
    {
      "epoch": 2.642857142857143,
      "grad_norm": 0.0596454031765461,
      "learning_rate": 2.3819047619047623e-06,
      "loss": 0.0,
      "step": 18500
    },
    {
      "epoch": 2.644285714285714,
      "grad_norm": 2855112.0,
      "learning_rate": 2.3723809523809525e-06,
      "loss": 1.6587,
      "step": 18510
    },
    {
      "epoch": 2.645714285714286,
      "grad_norm": 0.08926579356193542,
      "learning_rate": 2.362857142857143e-06,
      "loss": 0.0,
      "step": 18520
    },
    {
      "epoch": 2.6471428571428572,
      "grad_norm": 0.07764889299869537,
      "learning_rate": 2.3533333333333334e-06,
      "loss": 1.6279,
      "step": 18530
    },
    {
      "epoch": 2.6485714285714286,
      "grad_norm": 7644933.5,
      "learning_rate": 2.343809523809524e-06,
      "loss": 3.3439,
      "step": 18540
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.07009513676166534,
      "learning_rate": 2.3342857142857147e-06,
      "loss": 0.0,
      "step": 18550
    },
    {
      "epoch": 2.6514285714285712,
      "grad_norm": 0.09816383570432663,
      "learning_rate": 2.324761904761905e-06,
      "loss": 3.356,
      "step": 18560
    },
    {
      "epoch": 2.652857142857143,
      "grad_norm": 0.03593236580491066,
      "learning_rate": 2.3152380952380956e-06,
      "loss": 0.0,
      "step": 18570
    },
    {
      "epoch": 2.6542857142857144,
      "grad_norm": 0.07994630932807922,
      "learning_rate": 2.3057142857142858e-06,
      "loss": 1.6092,
      "step": 18580
    },
    {
      "epoch": 2.6557142857142857,
      "grad_norm": 51458940.0,
      "learning_rate": 2.2961904761904764e-06,
      "loss": 1.4467,
      "step": 18590
    },
    {
      "epoch": 2.657142857142857,
      "grad_norm": 0.362773597240448,
      "learning_rate": 2.2866666666666667e-06,
      "loss": 1.6983,
      "step": 18600
    },
    {
      "epoch": 2.6585714285714284,
      "grad_norm": 0.0809844434261322,
      "learning_rate": 2.2771428571428573e-06,
      "loss": 0.0,
      "step": 18610
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.03732491284608841,
      "learning_rate": 2.267619047619048e-06,
      "loss": 0.0,
      "step": 18620
    },
    {
      "epoch": 2.6614285714285715,
      "grad_norm": 0.08131890743970871,
      "learning_rate": 2.2580952380952386e-06,
      "loss": 5.0519,
      "step": 18630
    },
    {
      "epoch": 2.662857142857143,
      "grad_norm": 0.08663040399551392,
      "learning_rate": 2.248571428571429e-06,
      "loss": 1.7717,
      "step": 18640
    },
    {
      "epoch": 2.664285714285714,
      "grad_norm": 0.07369735836982727,
      "learning_rate": 2.2390476190476195e-06,
      "loss": 1.674,
      "step": 18650
    },
    {
      "epoch": 2.6657142857142855,
      "grad_norm": 4159683.25,
      "learning_rate": 2.2295238095238097e-06,
      "loss": 1.6869,
      "step": 18660
    },
    {
      "epoch": 2.6671428571428573,
      "grad_norm": 0.07900146394968033,
      "learning_rate": 2.2200000000000003e-06,
      "loss": 0.0,
      "step": 18670
    },
    {
      "epoch": 2.6685714285714286,
      "grad_norm": 0.5301446914672852,
      "learning_rate": 2.2104761904761906e-06,
      "loss": 3.3893,
      "step": 18680
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.06650257855653763,
      "learning_rate": 2.2009523809523812e-06,
      "loss": 0.0,
      "step": 18690
    },
    {
      "epoch": 2.6714285714285713,
      "grad_norm": 4927419.5,
      "learning_rate": 2.1914285714285714e-06,
      "loss": 5.0821,
      "step": 18700
    },
    {
      "epoch": 2.6728571428571426,
      "grad_norm": 2924082.5,
      "learning_rate": 2.181904761904762e-06,
      "loss": 1.6976,
      "step": 18710
    },
    {
      "epoch": 2.6742857142857144,
      "grad_norm": 3765552.25,
      "learning_rate": 2.1723809523809523e-06,
      "loss": 1.6182,
      "step": 18720
    },
    {
      "epoch": 2.6757142857142857,
      "grad_norm": 0.07347936183214188,
      "learning_rate": 2.162857142857143e-06,
      "loss": 1.6924,
      "step": 18730
    },
    {
      "epoch": 2.677142857142857,
      "grad_norm": 3322099.25,
      "learning_rate": 2.153333333333333e-06,
      "loss": 1.6699,
      "step": 18740
    },
    {
      "epoch": 2.678571428571429,
      "grad_norm": 0.46365880966186523,
      "learning_rate": 2.1438095238095243e-06,
      "loss": 1.6704,
      "step": 18750
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.06006753072142601,
      "learning_rate": 2.1342857142857145e-06,
      "loss": 0.8392,
      "step": 18760
    },
    {
      "epoch": 2.6814285714285715,
      "grad_norm": 0.0687485858798027,
      "learning_rate": 2.124761904761905e-06,
      "loss": 1.6819,
      "step": 18770
    },
    {
      "epoch": 2.682857142857143,
      "grad_norm": 0.0669621154665947,
      "learning_rate": 2.1152380952380954e-06,
      "loss": 3.3596,
      "step": 18780
    },
    {
      "epoch": 2.684285714285714,
      "grad_norm": 0.48757192492485046,
      "learning_rate": 2.105714285714286e-06,
      "loss": 1.6744,
      "step": 18790
    },
    {
      "epoch": 2.685714285714286,
      "grad_norm": 0.34954512119293213,
      "learning_rate": 2.0961904761904762e-06,
      "loss": 0.0,
      "step": 18800
    },
    {
      "epoch": 2.6871428571428573,
      "grad_norm": 0.5982160568237305,
      "learning_rate": 2.086666666666667e-06,
      "loss": 0.0,
      "step": 18810
    },
    {
      "epoch": 2.6885714285714286,
      "grad_norm": 0.31429266929626465,
      "learning_rate": 2.077142857142857e-06,
      "loss": 3.2661,
      "step": 18820
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.03567748889327049,
      "learning_rate": 2.0676190476190477e-06,
      "loss": 0.0,
      "step": 18830
    },
    {
      "epoch": 2.6914285714285713,
      "grad_norm": 0.022829361259937286,
      "learning_rate": 2.0580952380952384e-06,
      "loss": 0.0,
      "step": 18840
    },
    {
      "epoch": 2.692857142857143,
      "grad_norm": 0.07862319052219391,
      "learning_rate": 2.0485714285714286e-06,
      "loss": 0.0,
      "step": 18850
    },
    {
      "epoch": 2.6942857142857144,
      "grad_norm": 3384276.5,
      "learning_rate": 2.0390476190476193e-06,
      "loss": 3.2666,
      "step": 18860
    },
    {
      "epoch": 2.6957142857142857,
      "grad_norm": 0.10529010742902756,
      "learning_rate": 2.0295238095238095e-06,
      "loss": 1.7289,
      "step": 18870
    },
    {
      "epoch": 2.697142857142857,
      "grad_norm": 0.06484568119049072,
      "learning_rate": 2.02e-06,
      "loss": 0.0,
      "step": 18880
    },
    {
      "epoch": 2.6985714285714284,
      "grad_norm": 0.08819378912448883,
      "learning_rate": 2.010476190476191e-06,
      "loss": 0.0,
      "step": 18890
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.30787426233291626,
      "learning_rate": 2.000952380952381e-06,
      "loss": 0.0,
      "step": 18900
    },
    {
      "epoch": 2.7014285714285715,
      "grad_norm": 0.37152042984962463,
      "learning_rate": 1.9914285714285717e-06,
      "loss": 3.3248,
      "step": 18910
    },
    {
      "epoch": 2.702857142857143,
      "grad_norm": 0.04470881447196007,
      "learning_rate": 1.9819047619047623e-06,
      "loss": 0.0,
      "step": 18920
    },
    {
      "epoch": 2.704285714285714,
      "grad_norm": 0.29043346643447876,
      "learning_rate": 1.9723809523809525e-06,
      "loss": 0.0375,
      "step": 18930
    },
    {
      "epoch": 2.7057142857142855,
      "grad_norm": 0.030312513932585716,
      "learning_rate": 1.962857142857143e-06,
      "loss": 4.9199,
      "step": 18940
    },
    {
      "epoch": 2.7071428571428573,
      "grad_norm": 0.3720119595527649,
      "learning_rate": 1.9533333333333334e-06,
      "loss": 3.3353,
      "step": 18950
    },
    {
      "epoch": 2.7085714285714286,
      "grad_norm": 0.36857789754867554,
      "learning_rate": 1.943809523809524e-06,
      "loss": 0.0,
      "step": 18960
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.1615382730960846,
      "learning_rate": 1.9342857142857143e-06,
      "loss": 0.0,
      "step": 18970
    },
    {
      "epoch": 2.7114285714285713,
      "grad_norm": 0.0592011921107769,
      "learning_rate": 1.924761904761905e-06,
      "loss": 0.0,
      "step": 18980
    },
    {
      "epoch": 2.7128571428571426,
      "grad_norm": 2956193.25,
      "learning_rate": 1.915238095238095e-06,
      "loss": 1.68,
      "step": 18990
    },
    {
      "epoch": 2.7142857142857144,
      "grad_norm": 0.04229748249053955,
      "learning_rate": 1.9057142857142858e-06,
      "loss": 3.1354,
      "step": 19000
    },
    {
      "epoch": 2.7157142857142857,
      "grad_norm": 0.3105272054672241,
      "learning_rate": 1.8961904761904764e-06,
      "loss": 3.3176,
      "step": 19010
    },
    {
      "epoch": 2.717142857142857,
      "grad_norm": 0.0806846022605896,
      "learning_rate": 1.8866666666666669e-06,
      "loss": 0.0,
      "step": 19020
    },
    {
      "epoch": 2.7185714285714284,
      "grad_norm": 0.04646509140729904,
      "learning_rate": 1.8771428571428573e-06,
      "loss": 1.6694,
      "step": 19030
    },
    {
      "epoch": 2.7199999999999998,
      "grad_norm": 0.3144683241844177,
      "learning_rate": 1.8676190476190478e-06,
      "loss": 4.9088,
      "step": 19040
    },
    {
      "epoch": 2.7214285714285715,
      "grad_norm": 0.03867564722895622,
      "learning_rate": 1.8580952380952382e-06,
      "loss": 3.2769,
      "step": 19050
    },
    {
      "epoch": 2.722857142857143,
      "grad_norm": 0.05283064395189285,
      "learning_rate": 1.8485714285714288e-06,
      "loss": 1.691,
      "step": 19060
    },
    {
      "epoch": 2.724285714285714,
      "grad_norm": 0.05092881619930267,
      "learning_rate": 1.8390476190476193e-06,
      "loss": 4.7523,
      "step": 19070
    },
    {
      "epoch": 2.725714285714286,
      "grad_norm": 0.5574268698692322,
      "learning_rate": 1.8295238095238097e-06,
      "loss": 0.0,
      "step": 19080
    },
    {
      "epoch": 2.727142857142857,
      "grad_norm": 5718807.0,
      "learning_rate": 1.8200000000000002e-06,
      "loss": 4.9864,
      "step": 19090
    },
    {
      "epoch": 2.7285714285714286,
      "grad_norm": 0.03753545135259628,
      "learning_rate": 1.8104761904761906e-06,
      "loss": 0.0,
      "step": 19100
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.04379879683256149,
      "learning_rate": 1.800952380952381e-06,
      "loss": 0.0,
      "step": 19110
    },
    {
      "epoch": 2.7314285714285713,
      "grad_norm": 0.07014700025320053,
      "learning_rate": 1.7914285714285715e-06,
      "loss": 0.0,
      "step": 19120
    },
    {
      "epoch": 2.732857142857143,
      "grad_norm": 0.06980794668197632,
      "learning_rate": 1.781904761904762e-06,
      "loss": 1.5485,
      "step": 19130
    },
    {
      "epoch": 2.7342857142857144,
      "grad_norm": 2906816.0,
      "learning_rate": 1.7723809523809525e-06,
      "loss": 3.3933,
      "step": 19140
    },
    {
      "epoch": 2.7357142857142858,
      "grad_norm": 0.5751245617866516,
      "learning_rate": 1.7628571428571432e-06,
      "loss": 0.0427,
      "step": 19150
    },
    {
      "epoch": 2.737142857142857,
      "grad_norm": 0.06994755566120148,
      "learning_rate": 1.7533333333333336e-06,
      "loss": 5.021,
      "step": 19160
    },
    {
      "epoch": 2.7385714285714284,
      "grad_norm": 0.06049433723092079,
      "learning_rate": 1.743809523809524e-06,
      "loss": 0.0,
      "step": 19170
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.10631237924098969,
      "learning_rate": 1.7342857142857145e-06,
      "loss": 0.0,
      "step": 19180
    },
    {
      "epoch": 2.7414285714285715,
      "grad_norm": 0.051487550139427185,
      "learning_rate": 1.724761904761905e-06,
      "loss": 0.0,
      "step": 19190
    },
    {
      "epoch": 2.742857142857143,
      "grad_norm": 0.09701497107744217,
      "learning_rate": 1.7152380952380954e-06,
      "loss": 0.0,
      "step": 19200
    },
    {
      "epoch": 2.744285714285714,
      "grad_norm": 0.05763361230492592,
      "learning_rate": 1.7057142857142858e-06,
      "loss": 1.6916,
      "step": 19210
    },
    {
      "epoch": 2.7457142857142856,
      "grad_norm": 0.368590384721756,
      "learning_rate": 1.6961904761904762e-06,
      "loss": 1.6814,
      "step": 19220
    },
    {
      "epoch": 2.7471428571428573,
      "grad_norm": 0.0977129340171814,
      "learning_rate": 1.6866666666666667e-06,
      "loss": 0.0,
      "step": 19230
    },
    {
      "epoch": 2.7485714285714287,
      "grad_norm": 0.07749486714601517,
      "learning_rate": 1.6771428571428571e-06,
      "loss": 5.1316,
      "step": 19240
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.08423241227865219,
      "learning_rate": 1.6676190476190476e-06,
      "loss": 1.6577,
      "step": 19250
    },
    {
      "epoch": 2.7514285714285713,
      "grad_norm": 0.07428459078073502,
      "learning_rate": 1.6580952380952384e-06,
      "loss": 3.4528,
      "step": 19260
    },
    {
      "epoch": 2.7528571428571427,
      "grad_norm": 0.2737586498260498,
      "learning_rate": 1.6485714285714289e-06,
      "loss": 1.6838,
      "step": 19270
    },
    {
      "epoch": 2.7542857142857144,
      "grad_norm": 0.5424584150314331,
      "learning_rate": 1.6390476190476193e-06,
      "loss": 0.0,
      "step": 19280
    },
    {
      "epoch": 2.755714285714286,
      "grad_norm": 0.0836275964975357,
      "learning_rate": 1.6295238095238097e-06,
      "loss": 0.0,
      "step": 19290
    },
    {
      "epoch": 2.757142857142857,
      "grad_norm": 0.2643708288669586,
      "learning_rate": 1.6200000000000002e-06,
      "loss": 1.6503,
      "step": 19300
    },
    {
      "epoch": 2.7585714285714285,
      "grad_norm": 0.04712117090821266,
      "learning_rate": 1.6104761904761906e-06,
      "loss": 3.4115,
      "step": 19310
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.10295884311199188,
      "learning_rate": 1.600952380952381e-06,
      "loss": 1.6338,
      "step": 19320
    },
    {
      "epoch": 2.7614285714285716,
      "grad_norm": 344901632.0,
      "learning_rate": 1.5914285714285715e-06,
      "loss": 1.5981,
      "step": 19330
    },
    {
      "epoch": 2.762857142857143,
      "grad_norm": 0.04071792587637901,
      "learning_rate": 1.581904761904762e-06,
      "loss": 3.3903,
      "step": 19340
    },
    {
      "epoch": 2.7642857142857142,
      "grad_norm": 0.04672207310795784,
      "learning_rate": 1.5723809523809526e-06,
      "loss": 0.0,
      "step": 19350
    },
    {
      "epoch": 2.7657142857142856,
      "grad_norm": 0.0959116593003273,
      "learning_rate": 1.562857142857143e-06,
      "loss": 3.4152,
      "step": 19360
    },
    {
      "epoch": 2.767142857142857,
      "grad_norm": 11.251460075378418,
      "learning_rate": 1.5533333333333334e-06,
      "loss": 0.0,
      "step": 19370
    },
    {
      "epoch": 2.7685714285714287,
      "grad_norm": 0.0793241634964943,
      "learning_rate": 1.5438095238095239e-06,
      "loss": 0.0,
      "step": 19380
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.05445435643196106,
      "learning_rate": 1.5342857142857145e-06,
      "loss": 1.6332,
      "step": 19390
    },
    {
      "epoch": 2.7714285714285714,
      "grad_norm": 0.05922242999076843,
      "learning_rate": 1.524761904761905e-06,
      "loss": 1.7544,
      "step": 19400
    },
    {
      "epoch": 2.772857142857143,
      "grad_norm": 0.10766976326704025,
      "learning_rate": 1.5152380952380954e-06,
      "loss": 0.0,
      "step": 19410
    },
    {
      "epoch": 2.774285714285714,
      "grad_norm": 0.05500432103872299,
      "learning_rate": 1.5057142857142858e-06,
      "loss": 3.462,
      "step": 19420
    },
    {
      "epoch": 2.775714285714286,
      "grad_norm": 0.05606815218925476,
      "learning_rate": 1.4961904761904763e-06,
      "loss": 3.3468,
      "step": 19430
    },
    {
      "epoch": 2.777142857142857,
      "grad_norm": 0.28866884112358093,
      "learning_rate": 1.486666666666667e-06,
      "loss": 1.6377,
      "step": 19440
    },
    {
      "epoch": 2.7785714285714285,
      "grad_norm": 0.11459916085004807,
      "learning_rate": 1.4771428571428573e-06,
      "loss": 1.6709,
      "step": 19450
    },
    {
      "epoch": 2.7800000000000002,
      "grad_norm": 0.30621570348739624,
      "learning_rate": 1.4676190476190478e-06,
      "loss": 1.5697,
      "step": 19460
    },
    {
      "epoch": 2.7814285714285716,
      "grad_norm": 0.2866382896900177,
      "learning_rate": 1.4580952380952382e-06,
      "loss": 0.0,
      "step": 19470
    },
    {
      "epoch": 2.782857142857143,
      "grad_norm": 0.09171321988105774,
      "learning_rate": 1.4485714285714286e-06,
      "loss": 0.0,
      "step": 19480
    },
    {
      "epoch": 2.7842857142857143,
      "grad_norm": 0.1076500341296196,
      "learning_rate": 1.439047619047619e-06,
      "loss": 1.6196,
      "step": 19490
    },
    {
      "epoch": 2.7857142857142856,
      "grad_norm": 0.304789662361145,
      "learning_rate": 1.4295238095238095e-06,
      "loss": 3.3066,
      "step": 19500
    },
    {
      "epoch": 2.7871428571428574,
      "grad_norm": 0.09715273231267929,
      "learning_rate": 1.42e-06,
      "loss": 0.0,
      "step": 19510
    },
    {
      "epoch": 2.7885714285714287,
      "grad_norm": 0.07387883961200714,
      "learning_rate": 1.4104761904761908e-06,
      "loss": 0.0,
      "step": 19520
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.07756970077753067,
      "learning_rate": 1.4009523809523813e-06,
      "loss": 3.2702,
      "step": 19530
    },
    {
      "epoch": 2.7914285714285714,
      "grad_norm": 0.06942323595285416,
      "learning_rate": 1.3914285714285717e-06,
      "loss": 1.5221,
      "step": 19540
    },
    {
      "epoch": 2.7928571428571427,
      "grad_norm": 0.08325006812810898,
      "learning_rate": 1.3819047619047621e-06,
      "loss": 1.6264,
      "step": 19550
    },
    {
      "epoch": 2.7942857142857145,
      "grad_norm": 8903353.0,
      "learning_rate": 1.3723809523809526e-06,
      "loss": 3.2618,
      "step": 19560
    },
    {
      "epoch": 2.795714285714286,
      "grad_norm": 0.08363477140665054,
      "learning_rate": 1.362857142857143e-06,
      "loss": 1.6586,
      "step": 19570
    },
    {
      "epoch": 2.797142857142857,
      "grad_norm": 0.04810464009642601,
      "learning_rate": 1.3533333333333334e-06,
      "loss": 3.2613,
      "step": 19580
    },
    {
      "epoch": 2.7985714285714285,
      "grad_norm": 0.2820413112640381,
      "learning_rate": 1.3438095238095239e-06,
      "loss": 2.963,
      "step": 19590
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.3559705913066864,
      "learning_rate": 1.3342857142857143e-06,
      "loss": 1.7379,
      "step": 19600
    },
    {
      "epoch": 2.8014285714285716,
      "grad_norm": 0.05616416037082672,
      "learning_rate": 1.3247619047619047e-06,
      "loss": 1.6199,
      "step": 19610
    },
    {
      "epoch": 2.802857142857143,
      "grad_norm": 3.6957359313964844,
      "learning_rate": 1.3152380952380952e-06,
      "loss": 0.0,
      "step": 19620
    },
    {
      "epoch": 2.8042857142857143,
      "grad_norm": 0.06086205318570137,
      "learning_rate": 1.3057142857142856e-06,
      "loss": 1.5457,
      "step": 19630
    },
    {
      "epoch": 2.8057142857142856,
      "grad_norm": 2910319.75,
      "learning_rate": 1.2961904761904763e-06,
      "loss": 3.3165,
      "step": 19640
    },
    {
      "epoch": 2.807142857142857,
      "grad_norm": 0.3254007399082184,
      "learning_rate": 1.286666666666667e-06,
      "loss": 1.6462,
      "step": 19650
    },
    {
      "epoch": 2.8085714285714287,
      "grad_norm": 0.10251333564519882,
      "learning_rate": 1.2771428571428573e-06,
      "loss": 1.6229,
      "step": 19660
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.07069320976734161,
      "learning_rate": 1.2676190476190478e-06,
      "loss": 4.58,
      "step": 19670
    },
    {
      "epoch": 2.8114285714285714,
      "grad_norm": 0.372092068195343,
      "learning_rate": 1.2580952380952382e-06,
      "loss": 0.0,
      "step": 19680
    },
    {
      "epoch": 2.8128571428571427,
      "grad_norm": 0.06108337268233299,
      "learning_rate": 1.2485714285714287e-06,
      "loss": 0.0,
      "step": 19690
    },
    {
      "epoch": 2.814285714285714,
      "grad_norm": 0.10800319164991379,
      "learning_rate": 1.239047619047619e-06,
      "loss": 1.6705,
      "step": 19700
    },
    {
      "epoch": 2.815714285714286,
      "grad_norm": 5016307.0,
      "learning_rate": 1.2295238095238095e-06,
      "loss": 4.9428,
      "step": 19710
    },
    {
      "epoch": 2.817142857142857,
      "grad_norm": 0.06454689055681229,
      "learning_rate": 1.2200000000000002e-06,
      "loss": 1.7717,
      "step": 19720
    },
    {
      "epoch": 2.8185714285714285,
      "grad_norm": 0.3130723834037781,
      "learning_rate": 1.2104761904761906e-06,
      "loss": 1.6282,
      "step": 19730
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.0846286416053772,
      "learning_rate": 1.200952380952381e-06,
      "loss": 0.0,
      "step": 19740
    },
    {
      "epoch": 2.821428571428571,
      "grad_norm": 0.6643995642662048,
      "learning_rate": 1.1914285714285715e-06,
      "loss": 1.7284,
      "step": 19750
    },
    {
      "epoch": 2.822857142857143,
      "grad_norm": 0.08478057384490967,
      "learning_rate": 1.181904761904762e-06,
      "loss": 0.0067,
      "step": 19760
    },
    {
      "epoch": 2.8242857142857143,
      "grad_norm": 0.03623301535844803,
      "learning_rate": 1.1723809523809526e-06,
      "loss": 0.0,
      "step": 19770
    },
    {
      "epoch": 2.8257142857142856,
      "grad_norm": 0.41288086771965027,
      "learning_rate": 1.162857142857143e-06,
      "loss": 3.2985,
      "step": 19780
    },
    {
      "epoch": 2.8271428571428574,
      "grad_norm": 0.08572046458721161,
      "learning_rate": 1.1533333333333334e-06,
      "loss": 1.7063,
      "step": 19790
    },
    {
      "epoch": 2.8285714285714287,
      "grad_norm": 0.03604888543486595,
      "learning_rate": 1.1438095238095239e-06,
      "loss": 1.6568,
      "step": 19800
    },
    {
      "epoch": 2.83,
      "grad_norm": 2972458.75,
      "learning_rate": 1.1342857142857145e-06,
      "loss": 1.6868,
      "step": 19810
    },
    {
      "epoch": 2.8314285714285714,
      "grad_norm": 0.06418868154287338,
      "learning_rate": 1.124761904761905e-06,
      "loss": 0.0,
      "step": 19820
    },
    {
      "epoch": 2.8328571428571427,
      "grad_norm": 0.08114910870790482,
      "learning_rate": 1.1152380952380954e-06,
      "loss": 1.6676,
      "step": 19830
    },
    {
      "epoch": 2.8342857142857145,
      "grad_norm": 0.09499746561050415,
      "learning_rate": 1.1057142857142858e-06,
      "loss": 0.0,
      "step": 19840
    },
    {
      "epoch": 2.835714285714286,
      "grad_norm": 0.07595796138048172,
      "learning_rate": 1.0961904761904763e-06,
      "loss": 0.0,
      "step": 19850
    },
    {
      "epoch": 2.837142857142857,
      "grad_norm": 0.05817638710141182,
      "learning_rate": 1.0866666666666667e-06,
      "loss": 0.0,
      "step": 19860
    },
    {
      "epoch": 2.8385714285714285,
      "grad_norm": 0.04610179737210274,
      "learning_rate": 1.0771428571428574e-06,
      "loss": 0.0002,
      "step": 19870
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.06694912165403366,
      "learning_rate": 1.0676190476190478e-06,
      "loss": 0.0,
      "step": 19880
    },
    {
      "epoch": 2.8414285714285716,
      "grad_norm": 0.05761953443288803,
      "learning_rate": 1.0580952380952382e-06,
      "loss": 1.6382,
      "step": 19890
    },
    {
      "epoch": 2.842857142857143,
      "grad_norm": 0.09622196853160858,
      "learning_rate": 1.0485714285714287e-06,
      "loss": 1.6478,
      "step": 19900
    },
    {
      "epoch": 2.8442857142857143,
      "grad_norm": 0.34698793292045593,
      "learning_rate": 1.039047619047619e-06,
      "loss": 0.0,
      "step": 19910
    },
    {
      "epoch": 2.8457142857142856,
      "grad_norm": 0.4451240599155426,
      "learning_rate": 1.0295238095238095e-06,
      "loss": 1.7185,
      "step": 19920
    },
    {
      "epoch": 2.847142857142857,
      "grad_norm": 0.03811236843466759,
      "learning_rate": 1.02e-06,
      "loss": 1.7116,
      "step": 19930
    },
    {
      "epoch": 2.8485714285714288,
      "grad_norm": 3063245.75,
      "learning_rate": 1.0104761904761906e-06,
      "loss": 6.8122,
      "step": 19940
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.10508978366851807,
      "learning_rate": 1.000952380952381e-06,
      "loss": 1.7301,
      "step": 19950
    },
    {
      "epoch": 2.8514285714285714,
      "grad_norm": 0.04818183556199074,
      "learning_rate": 9.914285714285715e-07,
      "loss": 5.1928,
      "step": 19960
    },
    {
      "epoch": 2.8528571428571428,
      "grad_norm": 0.0655495747923851,
      "learning_rate": 9.81904761904762e-07,
      "loss": 3.4782,
      "step": 19970
    },
    {
      "epoch": 2.854285714285714,
      "grad_norm": 0.10731036216020584,
      "learning_rate": 9.723809523809524e-07,
      "loss": 1.687,
      "step": 19980
    },
    {
      "epoch": 2.855714285714286,
      "grad_norm": 0.09394026547670364,
      "learning_rate": 9.628571428571428e-07,
      "loss": 0.0,
      "step": 19990
    },
    {
      "epoch": 2.857142857142857,
      "grad_norm": 0.07835225760936737,
      "learning_rate": 9.533333333333335e-07,
      "loss": 0.0,
      "step": 20000
    },
    {
      "epoch": 2.8585714285714285,
      "grad_norm": 0.33570340275764465,
      "learning_rate": 9.438095238095239e-07,
      "loss": 1.7089,
      "step": 20010
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.10043105483055115,
      "learning_rate": 9.342857142857144e-07,
      "loss": 1.6181,
      "step": 20020
    },
    {
      "epoch": 2.861428571428571,
      "grad_norm": 0.03710104525089264,
      "learning_rate": 9.247619047619049e-07,
      "loss": 0.0,
      "step": 20030
    },
    {
      "epoch": 2.862857142857143,
      "grad_norm": 0.2892886698246002,
      "learning_rate": 9.152380952380953e-07,
      "loss": 1.6305,
      "step": 20040
    },
    {
      "epoch": 2.8642857142857143,
      "grad_norm": 2849814.0,
      "learning_rate": 9.057142857142857e-07,
      "loss": 2.9799,
      "step": 20050
    },
    {
      "epoch": 2.8657142857142857,
      "grad_norm": 0.0883132740855217,
      "learning_rate": 8.961904761904763e-07,
      "loss": 1.7005,
      "step": 20060
    },
    {
      "epoch": 2.867142857142857,
      "grad_norm": 3923258.5,
      "learning_rate": 8.866666666666668e-07,
      "loss": 5.0559,
      "step": 20070
    },
    {
      "epoch": 2.8685714285714283,
      "grad_norm": 0.07252060621976852,
      "learning_rate": 8.771428571428573e-07,
      "loss": 1.7426,
      "step": 20080
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.28827786445617676,
      "learning_rate": 8.676190476190477e-07,
      "loss": 1.6882,
      "step": 20090
    },
    {
      "epoch": 2.8714285714285714,
      "grad_norm": 0.05990985408425331,
      "learning_rate": 8.580952380952381e-07,
      "loss": 0.0,
      "step": 20100
    },
    {
      "epoch": 2.8728571428571428,
      "grad_norm": 0.10813448578119278,
      "learning_rate": 8.485714285714286e-07,
      "loss": 1.6782,
      "step": 20110
    },
    {
      "epoch": 2.8742857142857146,
      "grad_norm": 0.06161240115761757,
      "learning_rate": 8.39047619047619e-07,
      "loss": 1.7668,
      "step": 20120
    },
    {
      "epoch": 2.8757142857142854,
      "grad_norm": 0.04949702322483063,
      "learning_rate": 8.295238095238097e-07,
      "loss": 0.0,
      "step": 20130
    },
    {
      "epoch": 2.8771428571428572,
      "grad_norm": 0.08252423256635666,
      "learning_rate": 8.200000000000001e-07,
      "loss": 1.6731,
      "step": 20140
    },
    {
      "epoch": 2.8785714285714286,
      "grad_norm": 0.1110827699303627,
      "learning_rate": 8.104761904761905e-07,
      "loss": 0.0,
      "step": 20150
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.06449645012617111,
      "learning_rate": 8.00952380952381e-07,
      "loss": 3.4664,
      "step": 20160
    },
    {
      "epoch": 2.8814285714285717,
      "grad_norm": 0.11554664373397827,
      "learning_rate": 7.914285714285715e-07,
      "loss": 1.6166,
      "step": 20170
    },
    {
      "epoch": 2.882857142857143,
      "grad_norm": 0.5682810544967651,
      "learning_rate": 7.819047619047619e-07,
      "loss": 1.6329,
      "step": 20180
    },
    {
      "epoch": 2.8842857142857143,
      "grad_norm": 0.04324725270271301,
      "learning_rate": 7.723809523809525e-07,
      "loss": 1.6262,
      "step": 20190
    },
    {
      "epoch": 2.8857142857142857,
      "grad_norm": 0.0653606653213501,
      "learning_rate": 7.628571428571429e-07,
      "loss": 1.6636,
      "step": 20200
    },
    {
      "epoch": 2.887142857142857,
      "grad_norm": 4140945.75,
      "learning_rate": 7.533333333333335e-07,
      "loss": 1.6186,
      "step": 20210
    },
    {
      "epoch": 2.888571428571429,
      "grad_norm": 0.053161293268203735,
      "learning_rate": 7.438095238095239e-07,
      "loss": 0.0,
      "step": 20220
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.13042093813419342,
      "learning_rate": 7.342857142857143e-07,
      "loss": 0.0,
      "step": 20230
    },
    {
      "epoch": 2.8914285714285715,
      "grad_norm": 0.061153799295425415,
      "learning_rate": 7.247619047619048e-07,
      "loss": 1.7235,
      "step": 20240
    },
    {
      "epoch": 2.892857142857143,
      "grad_norm": 0.08381418883800507,
      "learning_rate": 7.152380952380952e-07,
      "loss": 0.0,
      "step": 20250
    },
    {
      "epoch": 2.894285714285714,
      "grad_norm": 0.058463796973228455,
      "learning_rate": 7.057142857142859e-07,
      "loss": 0.0,
      "step": 20260
    },
    {
      "epoch": 2.895714285714286,
      "grad_norm": 3278507.75,
      "learning_rate": 6.961904761904763e-07,
      "loss": 3.3445,
      "step": 20270
    },
    {
      "epoch": 2.8971428571428572,
      "grad_norm": 0.07932337373495102,
      "learning_rate": 6.866666666666667e-07,
      "loss": 1.6842,
      "step": 20280
    },
    {
      "epoch": 2.8985714285714286,
      "grad_norm": 0.3178240954875946,
      "learning_rate": 6.771428571428572e-07,
      "loss": 3.3081,
      "step": 20290
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.07937416434288025,
      "learning_rate": 6.676190476190476e-07,
      "loss": 3.1891,
      "step": 20300
    },
    {
      "epoch": 2.9014285714285712,
      "grad_norm": 0.4330761134624481,
      "learning_rate": 6.580952380952381e-07,
      "loss": 3.4482,
      "step": 20310
    },
    {
      "epoch": 2.902857142857143,
      "grad_norm": 0.07600096613168716,
      "learning_rate": 6.485714285714287e-07,
      "loss": 0.0,
      "step": 20320
    },
    {
      "epoch": 2.9042857142857144,
      "grad_norm": 0.35725781321525574,
      "learning_rate": 6.390476190476191e-07,
      "loss": 0.0,
      "step": 20330
    },
    {
      "epoch": 2.9057142857142857,
      "grad_norm": 0.030024435371160507,
      "learning_rate": 6.295238095238096e-07,
      "loss": 1.6339,
      "step": 20340
    },
    {
      "epoch": 2.907142857142857,
      "grad_norm": 0.07587550580501556,
      "learning_rate": 6.200000000000001e-07,
      "loss": 1.6654,
      "step": 20350
    },
    {
      "epoch": 2.9085714285714284,
      "grad_norm": 1189568512.0,
      "learning_rate": 6.104761904761905e-07,
      "loss": 0.9949,
      "step": 20360
    },
    {
      "epoch": 2.91,
      "grad_norm": 3190460.5,
      "learning_rate": 6.00952380952381e-07,
      "loss": 5.0397,
      "step": 20370
    },
    {
      "epoch": 2.9114285714285715,
      "grad_norm": 0.04003727436065674,
      "learning_rate": 5.914285714285715e-07,
      "loss": 3.4066,
      "step": 20380
    },
    {
      "epoch": 2.912857142857143,
      "grad_norm": 0.04700377210974693,
      "learning_rate": 5.81904761904762e-07,
      "loss": 0.0,
      "step": 20390
    },
    {
      "epoch": 2.914285714285714,
      "grad_norm": 152154976.0,
      "learning_rate": 5.723809523809525e-07,
      "loss": 1.4273,
      "step": 20400
    },
    {
      "epoch": 2.9157142857142855,
      "grad_norm": 0.3394452929496765,
      "learning_rate": 5.628571428571429e-07,
      "loss": 3.3824,
      "step": 20410
    },
    {
      "epoch": 2.9171428571428573,
      "grad_norm": 0.4125428795814514,
      "learning_rate": 5.533333333333334e-07,
      "loss": 3.3045,
      "step": 20420
    },
    {
      "epoch": 2.9185714285714286,
      "grad_norm": 0.08059248328208923,
      "learning_rate": 5.438095238095239e-07,
      "loss": 1.4886,
      "step": 20430
    },
    {
      "epoch": 2.92,
      "grad_norm": 2924996.0,
      "learning_rate": 5.342857142857143e-07,
      "loss": 5.0182,
      "step": 20440
    },
    {
      "epoch": 2.9214285714285713,
      "grad_norm": 0.06064916029572487,
      "learning_rate": 5.247619047619048e-07,
      "loss": 3.2401,
      "step": 20450
    },
    {
      "epoch": 2.9228571428571426,
      "grad_norm": 0.05410749465227127,
      "learning_rate": 5.152380952380953e-07,
      "loss": 1.6176,
      "step": 20460
    },
    {
      "epoch": 2.9242857142857144,
      "grad_norm": 0.059241894632577896,
      "learning_rate": 5.057142857142858e-07,
      "loss": 3.3991,
      "step": 20470
    },
    {
      "epoch": 2.9257142857142857,
      "grad_norm": 0.6548760533332825,
      "learning_rate": 4.961904761904762e-07,
      "loss": 1.6967,
      "step": 20480
    },
    {
      "epoch": 2.927142857142857,
      "grad_norm": 0.05896244943141937,
      "learning_rate": 4.866666666666666e-07,
      "loss": 1.6227,
      "step": 20490
    },
    {
      "epoch": 2.928571428571429,
      "grad_norm": 0.4082013666629791,
      "learning_rate": 4.771428571428572e-07,
      "loss": 3.3168,
      "step": 20500
    },
    {
      "epoch": 2.93,
      "grad_norm": 179.40505981445312,
      "learning_rate": 4.6761904761904766e-07,
      "loss": 1.4712,
      "step": 20510
    },
    {
      "epoch": 2.9314285714285715,
      "grad_norm": 0.07828725129365921,
      "learning_rate": 4.580952380952381e-07,
      "loss": 0.0,
      "step": 20520
    },
    {
      "epoch": 2.932857142857143,
      "grad_norm": 0.039937227964401245,
      "learning_rate": 4.4857142857142864e-07,
      "loss": 3.2763,
      "step": 20530
    },
    {
      "epoch": 2.934285714285714,
      "grad_norm": 4529526.5,
      "learning_rate": 4.390476190476191e-07,
      "loss": 3.2535,
      "step": 20540
    },
    {
      "epoch": 2.935714285714286,
      "grad_norm": 0.043278440833091736,
      "learning_rate": 4.295238095238095e-07,
      "loss": 0.0,
      "step": 20550
    },
    {
      "epoch": 2.9371428571428573,
      "grad_norm": 0.03489512950181961,
      "learning_rate": 4.2000000000000006e-07,
      "loss": 1.6889,
      "step": 20560
    },
    {
      "epoch": 2.9385714285714286,
      "grad_norm": 0.06100894883275032,
      "learning_rate": 4.104761904761905e-07,
      "loss": 0.0,
      "step": 20570
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.11726921051740646,
      "learning_rate": 4.00952380952381e-07,
      "loss": 0.0,
      "step": 20580
    },
    {
      "epoch": 2.9414285714285713,
      "grad_norm": 2534755840.0,
      "learning_rate": 3.9142857142857147e-07,
      "loss": 5.3729,
      "step": 20590
    },
    {
      "epoch": 2.942857142857143,
      "grad_norm": 0.05541460961103439,
      "learning_rate": 3.8190476190476196e-07,
      "loss": 0.0,
      "step": 20600
    },
    {
      "epoch": 2.9442857142857144,
      "grad_norm": 0.04822700098156929,
      "learning_rate": 3.723809523809524e-07,
      "loss": 1.6516,
      "step": 20610
    },
    {
      "epoch": 2.9457142857142857,
      "grad_norm": 0.09865011274814606,
      "learning_rate": 3.6285714285714283e-07,
      "loss": 1.6833,
      "step": 20620
    },
    {
      "epoch": 2.947142857142857,
      "grad_norm": 3658465.75,
      "learning_rate": 3.533333333333334e-07,
      "loss": 3.3204,
      "step": 20630
    },
    {
      "epoch": 2.9485714285714284,
      "grad_norm": 0.08282684534788132,
      "learning_rate": 3.438095238095238e-07,
      "loss": 1.5239,
      "step": 20640
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.06590000540018082,
      "learning_rate": 3.342857142857143e-07,
      "loss": 1.7064,
      "step": 20650
    },
    {
      "epoch": 2.9514285714285715,
      "grad_norm": 0.3836714029312134,
      "learning_rate": 3.247619047619048e-07,
      "loss": 0.0,
      "step": 20660
    },
    {
      "epoch": 2.952857142857143,
      "grad_norm": 0.0628795251250267,
      "learning_rate": 3.152380952380953e-07,
      "loss": 1.6637,
      "step": 20670
    },
    {
      "epoch": 2.954285714285714,
      "grad_norm": 0.0495738722383976,
      "learning_rate": 3.0571428571428577e-07,
      "loss": 3.2384,
      "step": 20680
    },
    {
      "epoch": 2.9557142857142855,
      "grad_norm": 1.7205970287322998,
      "learning_rate": 2.961904761904762e-07,
      "loss": 1.7735,
      "step": 20690
    },
    {
      "epoch": 2.9571428571428573,
      "grad_norm": 0.10769304633140564,
      "learning_rate": 2.866666666666667e-07,
      "loss": 0.0,
      "step": 20700
    },
    {
      "epoch": 2.9585714285714286,
      "grad_norm": 3323642.0,
      "learning_rate": 2.771428571428572e-07,
      "loss": 3.1875,
      "step": 20710
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.05278324708342552,
      "learning_rate": 2.676190476190476e-07,
      "loss": 1.6796,
      "step": 20720
    },
    {
      "epoch": 2.9614285714285713,
      "grad_norm": 0.024718152359128,
      "learning_rate": 2.580952380952381e-07,
      "loss": 1.7001,
      "step": 20730
    },
    {
      "epoch": 2.9628571428571426,
      "grad_norm": 0.05231381580233574,
      "learning_rate": 2.485714285714286e-07,
      "loss": 0.0,
      "step": 20740
    },
    {
      "epoch": 2.9642857142857144,
      "grad_norm": 9734006.0,
      "learning_rate": 2.390476190476191e-07,
      "loss": 3.2172,
      "step": 20750
    },
    {
      "epoch": 2.9657142857142857,
      "grad_norm": 8616954.0,
      "learning_rate": 2.2952380952380955e-07,
      "loss": 4.6954,
      "step": 20760
    },
    {
      "epoch": 2.967142857142857,
      "grad_norm": 0.06161266192793846,
      "learning_rate": 2.2e-07,
      "loss": 1.7269,
      "step": 20770
    },
    {
      "epoch": 2.9685714285714284,
      "grad_norm": 0.03718281537294388,
      "learning_rate": 2.104761904761905e-07,
      "loss": 4.6428,
      "step": 20780
    },
    {
      "epoch": 2.9699999999999998,
      "grad_norm": 0.42787036299705505,
      "learning_rate": 2.00952380952381e-07,
      "loss": 1.668,
      "step": 20790
    },
    {
      "epoch": 2.9714285714285715,
      "grad_norm": 0.1056976392865181,
      "learning_rate": 1.9142857142857143e-07,
      "loss": 0.0,
      "step": 20800
    },
    {
      "epoch": 2.972857142857143,
      "grad_norm": 0.060868922621011734,
      "learning_rate": 1.8190476190476192e-07,
      "loss": 1.6661,
      "step": 20810
    },
    {
      "epoch": 2.974285714285714,
      "grad_norm": 3371898.5,
      "learning_rate": 1.7238095238095238e-07,
      "loss": 3.3216,
      "step": 20820
    },
    {
      "epoch": 2.975714285714286,
      "grad_norm": 2720985.25,
      "learning_rate": 1.6285714285714287e-07,
      "loss": 1.7024,
      "step": 20830
    },
    {
      "epoch": 2.977142857142857,
      "grad_norm": 0.4902360737323761,
      "learning_rate": 1.5333333333333333e-07,
      "loss": 0.0,
      "step": 20840
    },
    {
      "epoch": 2.9785714285714286,
      "grad_norm": 0.03242696449160576,
      "learning_rate": 1.4380952380952382e-07,
      "loss": 1.7093,
      "step": 20850
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.0385613813996315,
      "learning_rate": 1.342857142857143e-07,
      "loss": 0.0,
      "step": 20860
    },
    {
      "epoch": 2.9814285714285713,
      "grad_norm": 0.4470408856868744,
      "learning_rate": 1.2476190476190477e-07,
      "loss": 4.8262,
      "step": 20870
    },
    {
      "epoch": 2.982857142857143,
      "grad_norm": 0.07203996926546097,
      "learning_rate": 1.1523809523809523e-07,
      "loss": 0.0,
      "step": 20880
    },
    {
      "epoch": 2.9842857142857144,
      "grad_norm": 0.09154542535543442,
      "learning_rate": 1.0571428571428572e-07,
      "loss": 3.2027,
      "step": 20890
    },
    {
      "epoch": 2.9857142857142858,
      "grad_norm": 27347944.0,
      "learning_rate": 9.61904761904762e-08,
      "loss": 3.3555,
      "step": 20900
    },
    {
      "epoch": 2.987142857142857,
      "grad_norm": 0.06018796190619469,
      "learning_rate": 8.666666666666668e-08,
      "loss": 1.6075,
      "step": 20910
    },
    {
      "epoch": 2.9885714285714284,
      "grad_norm": 0.06509305536746979,
      "learning_rate": 7.714285714285715e-08,
      "loss": 1.6358,
      "step": 20920
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.32964831590652466,
      "learning_rate": 6.761904761904763e-08,
      "loss": 0.0,
      "step": 20930
    },
    {
      "epoch": 2.9914285714285715,
      "grad_norm": 0.08186616748571396,
      "learning_rate": 5.80952380952381e-08,
      "loss": 0.0,
      "step": 20940
    },
    {
      "epoch": 2.992857142857143,
      "grad_norm": 0.10681647062301636,
      "learning_rate": 4.857142857142857e-08,
      "loss": 0.0,
      "step": 20950
    },
    {
      "epoch": 2.994285714285714,
      "grad_norm": 0.09508448094129562,
      "learning_rate": 3.904761904761905e-08,
      "loss": 1.6363,
      "step": 20960
    },
    {
      "epoch": 2.9957142857142856,
      "grad_norm": 0.5847575664520264,
      "learning_rate": 2.9523809523809528e-08,
      "loss": 1.5812,
      "step": 20970
    },
    {
      "epoch": 2.9971428571428573,
      "grad_norm": 0.045893121510744095,
      "learning_rate": 2e-08,
      "loss": 1.6774,
      "step": 20980
    },
    {
      "epoch": 2.9985714285714287,
      "grad_norm": 659.1006469726562,
      "learning_rate": 1.0476190476190478e-08,
      "loss": 1.6483,
      "step": 20990
    },
    {
      "epoch": 3.0,
      "grad_norm": 5907684.5,
      "learning_rate": 9.523809523809524e-10,
      "loss": 3.2041,
      "step": 21000
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8775,
      "eval_f1_macro": 0.876753368482751,
      "eval_f1_weighted": 0.8777510088826727,
      "eval_loss": 2.0223817825317383,
      "eval_runtime": 254.6501,
      "eval_samples_per_second": 23.562,
      "eval_steps_per_second": 11.781,
      "step": 21000
    },
    {
      "epoch": 3.0,
      "step": 21000,
      "total_flos": 1699604650315968.0,
      "train_loss": 2.574909625827158,
      "train_runtime": 2610.9442,
      "train_samples_per_second": 16.086,
      "train_steps_per_second": 8.043
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8775,
      "eval_f1_macro": 0.876753368482751,
      "eval_f1_weighted": 0.8777510088826727,
      "eval_loss": 2.0223817825317383,
      "eval_runtime": 255.147,
      "eval_samples_per_second": 23.516,
      "eval_steps_per_second": 11.758,
      "step": 21000
    }
  ],
  "logging_steps": 10,
  "max_steps": 21000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1699604650315968.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}