trainer_state.json

{
  "best_metric": 0.848235547542572,
  "best_model_checkpoint": "Action_Classification/checkpoint-300",
  "epoch": 15.0,
  "eval_steps": 100,
  "global_step": 4710,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "grad_norm": 6.323575019836426,
      "learning_rate": 0.00019957537154989386,
      "loss": 0.4771,
      "step": 10
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.951721429824829,
      "learning_rate": 0.0001991507430997877,
      "loss": 0.4721,
      "step": 20
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.95866584777832,
      "learning_rate": 0.00019872611464968155,
      "loss": 0.5038,
      "step": 30
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.4051642417907715,
      "learning_rate": 0.00019830148619957538,
      "loss": 0.5058,
      "step": 40
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.13778018951416,
      "learning_rate": 0.00019787685774946923,
      "loss": 0.508,
      "step": 50
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.9513375759124756,
      "learning_rate": 0.00019745222929936306,
      "loss": 0.5479,
      "step": 60
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.7185003757476807,
      "learning_rate": 0.00019702760084925691,
      "loss": 0.4929,
      "step": 70
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.2211806774139404,
      "learning_rate": 0.00019660297239915074,
      "loss": 0.4391,
      "step": 80
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.076002597808838,
      "learning_rate": 0.0001961783439490446,
      "loss": 0.3744,
      "step": 90
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.571686744689941,
      "learning_rate": 0.00019575371549893845,
      "loss": 0.3922,
      "step": 100
    },
    {
      "epoch": 0.32,
      "eval_accuracy": 0.6933333333333334,
      "eval_confusion_matrix": [
        [
          66,
          1,
          9,
          6,
          1,
          5,
          1,
          3,
          7,
          1
        ],
        [
          41,
          96,
          0,
          0,
          8,
          0,
          2,
          1,
          45,
          0
        ],
        [
          2,
          0,
          46,
          1,
          1,
          7,
          4,
          0,
          1,
          3
        ],
        [
          9,
          1,
          4,
          19,
          5,
          3,
          2,
          1,
          4,
          4
        ],
        [
          0,
          2,
          0,
          3,
          84,
          2,
          1,
          0,
          0,
          1
        ],
        [
          4,
          1,
          3,
          2,
          0,
          55,
          3,
          1,
          3,
          0
        ],
        [
          0,
          0,
          1,
          0,
          0,
          1,
          54,
          0,
          3,
          0
        ],
        [
          5,
          1,
          4,
          1,
          0,
          1,
          0,
          70,
          1,
          0
        ],
        [
          5,
          12,
          0,
          1,
          1,
          0,
          14,
          0,
          147,
          0
        ],
        [
          9,
          0,
          1,
          38,
          3,
          1,
          4,
          4,
          2,
          91
        ]
      ],
      "eval_loss": 1.0780633687973022,
      "eval_runtime": 15.488,
      "eval_samples_per_second": 67.794,
      "eval_steps_per_second": 4.261,
      "step": 100
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.284886121749878,
      "learning_rate": 0.00019532908704883228,
      "loss": 0.5509,
      "step": 110
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.23751163482666,
      "learning_rate": 0.00019490445859872614,
      "loss": 0.4004,
      "step": 120
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.910186290740967,
      "learning_rate": 0.00019447983014861996,
      "loss": 0.4871,
      "step": 130
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.422680377960205,
      "learning_rate": 0.00019405520169851382,
      "loss": 0.3791,
      "step": 140
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.270318031311035,
      "learning_rate": 0.00019363057324840765,
      "loss": 0.4276,
      "step": 150
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.393375873565674,
      "learning_rate": 0.0001932059447983015,
      "loss": 0.5457,
      "step": 160
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.6365885734558105,
      "learning_rate": 0.00019278131634819533,
      "loss": 0.4431,
      "step": 170
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.3216943740844727,
      "learning_rate": 0.00019235668789808918,
      "loss": 0.3425,
      "step": 180
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.078379154205322,
      "learning_rate": 0.000191932059447983,
      "loss": 0.4102,
      "step": 190
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2445045709609985,
      "learning_rate": 0.00019150743099787687,
      "loss": 0.439,
      "step": 200
    },
    {
      "epoch": 0.64,
      "eval_accuracy": 0.7561904761904762,
      "eval_confusion_matrix": [
        [
          73,
          3,
          6,
          4,
          0,
          3,
          2,
          3,
          3,
          3
        ],
        [
          30,
          121,
          1,
          0,
          1,
          0,
          8,
          0,
          32,
          0
        ],
        [
          1,
          0,
          47,
          1,
          1,
          9,
          1,
          0,
          1,
          4
        ],
        [
          7,
          0,
          5,
          28,
          5,
          1,
          0,
          1,
          2,
          3
        ],
        [
          0,
          2,
          0,
          1,
          88,
          0,
          1,
          0,
          0,
          1
        ],
        [
          4,
          1,
          5,
          3,
          2,
          51,
          0,
          1,
          2,
          3
        ],
        [
          0,
          1,
          1,
          0,
          0,
          0,
          56,
          0,
          1,
          0
        ],
        [
          4,
          2,
          1,
          0,
          0,
          0,
          1,
          74,
          1,
          0
        ],
        [
          4,
          28,
          0,
          1,
          0,
          0,
          19,
          2,
          125,
          1
        ],
        [
          3,
          0,
          1,
          15,
          1,
          0,
          1,
          1,
          0,
          131
        ]
      ],
      "eval_loss": 0.8592283129692078,
      "eval_runtime": 15.2608,
      "eval_samples_per_second": 68.804,
      "eval_steps_per_second": 4.325,
      "step": 200
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.532979965209961,
      "learning_rate": 0.00019108280254777072,
      "loss": 0.3894,
      "step": 210
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3239492177963257,
      "learning_rate": 0.00019065817409766455,
      "loss": 0.3737,
      "step": 220
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.039944171905518,
      "learning_rate": 0.0001902335456475584,
      "loss": 0.5094,
      "step": 230
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.1441848278045654,
      "learning_rate": 0.00018980891719745223,
      "loss": 0.4595,
      "step": 240
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.0012054443359375,
      "learning_rate": 0.0001893842887473461,
      "loss": 0.4973,
      "step": 250
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.918905735015869,
      "learning_rate": 0.00018895966029723992,
      "loss": 0.5684,
      "step": 260
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.0025856494903564,
      "learning_rate": 0.00018853503184713377,
      "loss": 0.4679,
      "step": 270
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.386856555938721,
      "learning_rate": 0.0001881104033970276,
      "loss": 0.3681,
      "step": 280
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.106396198272705,
      "learning_rate": 0.00018768577494692146,
      "loss": 0.5392,
      "step": 290
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.501603126525879,
      "learning_rate": 0.0001872611464968153,
      "loss": 0.4664,
      "step": 300
    },
    {
      "epoch": 0.96,
      "eval_accuracy": 0.7628571428571429,
      "eval_confusion_matrix": [
        [
          45,
          5,
          20,
          4,
          2,
          6,
          4,
          8,
          3,
          3
        ],
        [
          5,
          154,
          4,
          2,
          1,
          2,
          6,
          1,
          17,
          1
        ],
        [
          0,
          0,
          51,
          1,
          2,
          8,
          1,
          0,
          0,
          2
        ],
        [
          1,
          0,
          8,
          26,
          8,
          5,
          0,
          0,
          1,
          3
        ],
        [
          0,
          1,
          0,
          0,
          89,
          3,
          0,
          0,
          0,
          0
        ],
        [
          0,
          1,
          11,
          3,
          1,
          55,
          0,
          1,
          0,
          0
        ],
        [
          0,
          1,
          1,
          0,
          3,
          3,
          51,
          0,
          0,
          0
        ],
        [
          0,
          0,
          10,
          1,
          0,
          4,
          0,
          68,
          0,
          0
        ],
        [
          0,
          26,
          5,
          0,
          1,
          3,
          16,
          1,
          127,
          1
        ],
        [
          3,
          0,
          2,
          9,
          2,
          1,
          0,
          1,
          0,
          135
        ]
      ],
      "eval_loss": 0.848235547542572,
      "eval_runtime": 15.3566,
      "eval_samples_per_second": 68.375,
      "eval_steps_per_second": 4.298,
      "step": 300
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.331725835800171,
      "learning_rate": 0.00018683651804670914,
      "loss": 0.493,
      "step": 310
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.574372291564941,
      "learning_rate": 0.000186411889596603,
      "loss": 0.3776,
      "step": 320
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.206468105316162,
      "learning_rate": 0.00018598726114649682,
      "loss": 0.4086,
      "step": 330
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.6000657081604,
      "learning_rate": 0.00018556263269639068,
      "loss": 0.4096,
      "step": 340
    },
    {
      "epoch": 1.11,
      "grad_norm": 2.429077625274658,
      "learning_rate": 0.0001851380042462845,
      "loss": 0.3707,
      "step": 350
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4174879789352417,
      "learning_rate": 0.00018471337579617836,
      "loss": 0.3507,
      "step": 360
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.691073894500732,
      "learning_rate": 0.0001842887473460722,
      "loss": 0.3365,
      "step": 370
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.105412483215332,
      "learning_rate": 0.00018386411889596604,
      "loss": 0.3132,
      "step": 380
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.423787236213684,
      "learning_rate": 0.00018343949044585987,
      "loss": 0.2726,
      "step": 390
    },
    {
      "epoch": 1.27,
      "grad_norm": 3.5701773166656494,
      "learning_rate": 0.00018301486199575373,
      "loss": 0.2929,
      "step": 400
    },
    {
      "epoch": 1.27,
      "eval_accuracy": 0.679047619047619,
      "eval_confusion_matrix": [
        [
          65,
          3,
          9,
          7,
          1,
          1,
          2,
          2,
          10,
          0
        ],
        [
          38,
          113,
          1,
          0,
          1,
          0,
          5,
          0,
          35,
          0
        ],
        [
          3,
          0,
          54,
          4,
          1,
          1,
          2,
          0,
          0,
          0
        ],
        [
          8,
          2,
          5,
          31,
          5,
          0,
          0,
          0,
          1,
          0
        ],
        [
          0,
          2,
          6,
          3,
          80,
          0,
          1,
          1,
          0,
          0
        ],
        [
          6,
          2,
          16,
          8,
          1,
          34,
          1,
          1,
          3,
          0
        ],
        [
          1,
          2,
          1,
          0,
          0,
          0,
          55,
          0,
          0,
          0
        ],
        [
          6,
          2,
          6,
          2,
          0,
          0,
          0,
          66,
          1,
          0
        ],
        [
          3,
          24,
          2,
          2,
          0,
          0,
          14,
          0,
          135,
          0
        ],
        [
          9,
          2,
          4,
          56,
          1,
          0,
          1,
          0,
          0,
          80
        ]
      ],
      "eval_loss": 1.1280611753463745,
      "eval_runtime": 15.4285,
      "eval_samples_per_second": 68.056,
      "eval_steps_per_second": 4.278,
      "step": 400
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.247598171234131,
      "learning_rate": 0.00018259023354564758,
      "loss": 0.3167,
      "step": 410
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.9821741580963135,
      "learning_rate": 0.0001821656050955414,
      "loss": 0.3745,
      "step": 420
    },
    {
      "epoch": 1.37,
      "grad_norm": 3.7514052391052246,
      "learning_rate": 0.00018174097664543526,
      "loss": 0.3996,
      "step": 430
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.241259574890137,
      "learning_rate": 0.0001813163481953291,
      "loss": 0.2942,
      "step": 440
    },
    {
      "epoch": 1.43,
      "grad_norm": 3.530998945236206,
      "learning_rate": 0.00018089171974522295,
      "loss": 0.3888,
      "step": 450
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6759068965911865,
      "learning_rate": 0.00018046709129511678,
      "loss": 0.26,
      "step": 460
    },
    {
      "epoch": 1.5,
      "grad_norm": 3.3739516735076904,
      "learning_rate": 0.00018004246284501063,
      "loss": 0.4405,
      "step": 470
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.361608982086182,
      "learning_rate": 0.00017961783439490446,
      "loss": 0.4011,
      "step": 480
    },
    {
      "epoch": 1.56,
      "grad_norm": 2.788034200668335,
      "learning_rate": 0.00017919320594479831,
      "loss": 0.3427,
      "step": 490
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.486495494842529,
      "learning_rate": 0.00017876857749469217,
      "loss": 0.4188,
      "step": 500
    },
    {
      "epoch": 1.59,
      "eval_accuracy": 0.6657142857142857,
      "eval_confusion_matrix": [
        [
          61,
          2,
          11,
          6,
          3,
          5,
          2,
          5,
          4,
          1
        ],
        [
          53,
          85,
          5,
          0,
          8,
          5,
          3,
          6,
          28,
          0
        ],
        [
          0,
          0,
          51,
          2,
          2,
          5,
          1,
          3,
          0,
          1
        ],
        [
          2,
          1,
          4,
          34,
          8,
          0,
          0,
          2,
          1,
          0
        ],
        [
          0,
          1,
          0,
          1,
          89,
          0,
          1,
          0,
          1,
          0
        ],
        [
          1,
          0,
          7,
          4,
          5,
          48,
          1,
          5,
          1,
          0
        ],
        [
          0,
          1,
          1,
          0,
          3,
          0,
          54,
          0,
          0,
          0
        ],
        [
          5,
          1,
          1,
          1,
          0,
          3,
          0,
          72,
          0,
          0
        ],
        [
          11,
          18,
          0,
          0,
          2,
          2,
          17,
          8,
          122,
          0
        ],
        [
          1,
          1,
          2,
          42,
          8,
          1,
          10,
          4,
          1,
          83
        ]
      ],
      "eval_loss": 1.185118556022644,
      "eval_runtime": 15.3843,
      "eval_samples_per_second": 68.251,
      "eval_steps_per_second": 4.29,
      "step": 500
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.058138370513916,
      "learning_rate": 0.000178343949044586,
      "loss": 0.3391,
      "step": 510
    },
    {
      "epoch": 1.66,
      "grad_norm": 3.33333683013916,
      "learning_rate": 0.00017791932059447985,
      "loss": 0.3716,
      "step": 520
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.987284183502197,
      "learning_rate": 0.00017749469214437368,
      "loss": 0.421,
      "step": 530
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.609795570373535,
      "learning_rate": 0.00017707006369426754,
      "loss": 0.3626,
      "step": 540
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.64127254486084,
      "learning_rate": 0.00017664543524416136,
      "loss": 0.3032,
      "step": 550
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.4108357429504395,
      "learning_rate": 0.00017622080679405522,
      "loss": 0.3659,
      "step": 560
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7708959579467773,
      "learning_rate": 0.00017579617834394905,
      "loss": 0.3496,
      "step": 570
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.026036262512207,
      "learning_rate": 0.0001753715498938429,
      "loss": 0.3223,
      "step": 580
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.661407470703125,
      "learning_rate": 0.00017494692144373676,
      "loss": 0.3881,
      "step": 590
    },
    {
      "epoch": 1.91,
      "grad_norm": 3.0189332962036133,
      "learning_rate": 0.00017452229299363059,
      "loss": 0.3668,
      "step": 600
    },
    {
      "epoch": 1.91,
      "eval_accuracy": 0.7466666666666667,
      "eval_confusion_matrix": [
        [
          53,
          11,
          11,
          5,
          0,
          3,
          1,
          4,
          10,
          2
        ],
        [
          3,
          145,
          5,
          0,
          1,
          1,
          4,
          5,
          29,
          0
        ],
        [
          0,
          0,
          53,
          1,
          1,
          5,
          2,
          1,
          1,
          1
        ],
        [
          4,
          0,
          9,
          29,
          5,
          2,
          0,
          0,
          1,
          2
        ],
        [
          0,
          1,
          4,
          3,
          84,
          0,
          0,
          0,
          1,
          0
        ],
        [
          2,
          2,
          12,
          3,
          1,
          45,
          0,
          3,
          3,
          1
        ],
        [
          0,
          1,
          2,
          1,
          1,
          0,
          52,
          0,
          2,
          0
        ],
        [
          1,
          2,
          5,
          1,
          0,
          1,
          0,
          73,
          0,
          0
        ],
        [
          4,
          29,
          2,
          0,
          0,
          0,
          7,
          3,
          135,
          0
        ],
        [
          1,
          0,
          11,
          19,
          1,
          5,
          0,
          1,
          0,
          115
        ]
      ],
      "eval_loss": 0.8554251790046692,
      "eval_runtime": 15.3855,
      "eval_samples_per_second": 68.246,
      "eval_steps_per_second": 4.29,
      "step": 600
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.359609127044678,
      "learning_rate": 0.00017409766454352444,
      "loss": 0.3581,
      "step": 610
    },
    {
      "epoch": 1.97,
      "grad_norm": 3.1502652168273926,
      "learning_rate": 0.00017367303609341827,
      "loss": 0.375,
      "step": 620
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.531296968460083,
      "learning_rate": 0.00017324840764331212,
      "loss": 0.3344,
      "step": 630
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.272879600524902,
      "learning_rate": 0.00017282377919320595,
      "loss": 0.2658,
      "step": 640
    },
    {
      "epoch": 2.07,
      "grad_norm": 6.005029678344727,
      "learning_rate": 0.0001723991507430998,
      "loss": 0.2963,
      "step": 650
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.308213710784912,
      "learning_rate": 0.00017197452229299363,
      "loss": 0.3024,
      "step": 660
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.6260054111480713,
      "learning_rate": 0.0001715498938428875,
      "loss": 0.3372,
      "step": 670
    },
    {
      "epoch": 2.17,
      "grad_norm": 5.7114057540893555,
      "learning_rate": 0.00017112526539278132,
      "loss": 0.3561,
      "step": 680
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.8753015995025635,
      "learning_rate": 0.00017070063694267517,
      "loss": 0.3529,
      "step": 690
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4016531705856323,
      "learning_rate": 0.00017027600849256903,
      "loss": 0.342,
      "step": 700
    },
    {
      "epoch": 2.23,
      "eval_accuracy": 0.7047619047619048,
      "eval_confusion_matrix": [
        [
          58,
          5,
          4,
          4,
          1,
          8,
          1,
          9,
          7,
          3
        ],
        [
          36,
          111,
          0,
          2,
          1,
          4,
          1,
          4,
          34,
          0
        ],
        [
          3,
          2,
          45,
          5,
          1,
          6,
          1,
          1,
          0,
          1
        ],
        [
          6,
          0,
          5,
          35,
          2,
          0,
          0,
          0,
          1,
          3
        ],
        [
          1,
          1,
          2,
          6,
          77,
          3,
          1,
          0,
          2,
          0
        ],
        [
          4,
          0,
          10,
          8,
          1,
          39,
          0,
          7,
          2,
          1
        ],
        [
          1,
          2,
          1,
          0,
          2,
          0,
          50,
          0,
          3,
          0
        ],
        [
          1,
          0,
          4,
          1,
          0,
          0,
          0,
          77,
          0,
          0
        ],
        [
          4,
          29,
          0,
          0,
          0,
          0,
          5,
          2,
          140,
          0
        ],
        [
          5,
          0,
          5,
          27,
          0,
          0,
          1,
          7,
          0,
          108
        ]
      ],
      "eval_loss": 1.0290604829788208,
      "eval_runtime": 15.7598,
      "eval_samples_per_second": 66.625,
      "eval_steps_per_second": 4.188,
      "step": 700
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.78086256980896,
      "learning_rate": 0.00016985138004246286,
      "loss": 0.2987,
      "step": 710
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8382972478866577,
      "learning_rate": 0.0001694267515923567,
      "loss": 0.4537,
      "step": 720
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.91944694519043,
      "learning_rate": 0.00016900212314225054,
      "loss": 0.264,
      "step": 730
    },
    {
      "epoch": 2.36,
      "grad_norm": 3.87353515625,
      "learning_rate": 0.0001685774946921444,
      "loss": 0.3189,
      "step": 740
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.6051061153411865,
      "learning_rate": 0.00016815286624203822,
      "loss": 0.2776,
      "step": 750
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.5838241577148438,
      "learning_rate": 0.00016772823779193208,
      "loss": 0.286,
      "step": 760
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.6773152351379395,
      "learning_rate": 0.0001673036093418259,
      "loss": 0.2957,
      "step": 770
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5790090560913086,
      "learning_rate": 0.00016687898089171976,
      "loss": 0.3021,
      "step": 780
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9358294010162354,
      "learning_rate": 0.00016645435244161362,
      "loss": 0.2431,
      "step": 790
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.2236168384552,
      "learning_rate": 0.00016602972399150744,
      "loss": 0.2984,
      "step": 800
    },
    {
      "epoch": 2.55,
      "eval_accuracy": 0.6961904761904761,
      "eval_confusion_matrix": [
        [
          55,
          3,
          11,
          2,
          0,
          1,
          2,
          11,
          10,
          5
        ],
        [
          44,
          71,
          1,
          0,
          1,
          0,
          4,
          5,
          66,
          1
        ],
        [
          0,
          0,
          49,
          3,
          2,
          3,
          3,
          2,
          1,
          2
        ],
        [
          4,
          0,
          5,
          26,
          7,
          0,
          2,
          0,
          1,
          7
        ],
        [
          0,
          1,
          0,
          0,
          86,
          0,
          3,
          0,
          3,
          0
        ],
        [
          5,
          2,
          12,
          4,
          2,
          39,
          1,
          3,
          1,
          3
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          57,
          0,
          0,
          0
        ],
        [
          0,
          0,
          5,
          0,
          0,
          0,
          0,
          78,
          0,
          0
        ],
        [
          5,
          15,
          3,
          0,
          1,
          0,
          9,
          4,
          143,
          0
        ],
        [
          0,
          0,
          1,
          17,
          1,
          0,
          3,
          3,
          1,
          127
        ]
      ],
      "eval_loss": 1.2206960916519165,
      "eval_runtime": 15.2799,
      "eval_samples_per_second": 68.718,
      "eval_steps_per_second": 4.319,
      "step": 800
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.2855160236358643,
      "learning_rate": 0.0001656050955414013,
      "loss": 0.3444,
      "step": 810
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.5719540119171143,
      "learning_rate": 0.00016518046709129513,
      "loss": 0.2846,
      "step": 820
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.042548179626465,
      "learning_rate": 0.00016475583864118898,
      "loss": 0.3975,
      "step": 830
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.08795166015625,
      "learning_rate": 0.0001643312101910828,
      "loss": 0.2734,
      "step": 840
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.969775676727295,
      "learning_rate": 0.00016390658174097667,
      "loss": 0.2576,
      "step": 850
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.800004720687866,
      "learning_rate": 0.0001634819532908705,
      "loss": 0.2583,
      "step": 860
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.8801493644714355,
      "learning_rate": 0.00016305732484076435,
      "loss": 0.3354,
      "step": 870
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.1636312007904053,
      "learning_rate": 0.0001626326963906582,
      "loss": 0.2068,
      "step": 880
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.3073160648345947,
      "learning_rate": 0.00016220806794055203,
      "loss": 0.356,
      "step": 890
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.645033836364746,
      "learning_rate": 0.0001617834394904459,
      "loss": 0.3542,
      "step": 900
    },
    {
      "epoch": 2.87,
      "eval_accuracy": 0.6657142857142857,
      "eval_confusion_matrix": [
        [
          66,
          2,
          6,
          8,
          0,
          4,
          4,
          4,
          5,
          1
        ],
        [
          38,
          78,
          2,
          0,
          3,
          1,
          23,
          4,
          43,
          1
        ],
        [
          2,
          0,
          50,
          7,
          0,
          5,
          1,
          0,
          0,
          0
        ],
        [
          2,
          0,
          2,
          45,
          3,
          0,
          0,
          0,
          0,
          0
        ],
        [
          0,
          1,
          3,
          6,
          76,
          3,
          4,
          0,
          0,
          0
        ],
        [
          3,
          1,
          10,
          8,
          1,
          47,
          0,
          0,
          1,
          1
        ],
        [
          0,
          0,
          2,
          0,
          1,
          0,
          56,
          0,
          0,
          0
        ],
        [
          2,
          0,
          11,
          5,
          0,
          2,
          4,
          59,
          0,
          0
        ],
        [
          4,
          24,
          2,
          1,
          1,
          1,
          23,
          0,
          124,
          0
        ],
        [
          5,
          0,
          1,
          42,
          3,
          0,
          3,
          1,
          0,
          98
        ]
      ],
      "eval_loss": 1.18353271484375,
      "eval_runtime": 15.3461,
      "eval_samples_per_second": 68.421,
      "eval_steps_per_second": 4.301,
      "step": 900
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6855131387710571,
      "learning_rate": 0.00016135881104033971,
      "loss": 0.2494,
      "step": 910
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0318846702575684,
      "learning_rate": 0.00016093418259023357,
      "loss": 0.2848,
      "step": 920
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.573373794555664,
      "learning_rate": 0.0001605095541401274,
      "loss": 0.2974,
      "step": 930
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9912055730819702,
      "learning_rate": 0.00016008492569002125,
      "loss": 0.1985,
      "step": 940
    },
    {
      "epoch": 3.03,
      "grad_norm": 3.447814702987671,
      "learning_rate": 0.00015966029723991508,
      "loss": 0.2648,
      "step": 950
    },
    {
      "epoch": 3.06,
      "grad_norm": 2.786409616470337,
      "learning_rate": 0.00015923566878980894,
      "loss": 0.2106,
      "step": 960
    },
    {
      "epoch": 3.09,
      "grad_norm": 7.731870174407959,
      "learning_rate": 0.00015881104033970276,
      "loss": 0.2545,
      "step": 970
    },
    {
      "epoch": 3.12,
      "grad_norm": 5.680171966552734,
      "learning_rate": 0.00015838641188959662,
      "loss": 0.3026,
      "step": 980
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.9310322999954224,
      "learning_rate": 0.00015796178343949047,
      "loss": 0.2678,
      "step": 990
    },
    {
      "epoch": 3.18,
      "grad_norm": 4.319097995758057,
      "learning_rate": 0.0001575371549893843,
      "loss": 0.2749,
      "step": 1000
    },
    {
      "epoch": 3.18,
      "eval_accuracy": 0.7285714285714285,
      "eval_confusion_matrix": [
        [
          54,
          12,
          5,
          2,
          3,
          1,
          7,
          1,
          12,
          3
        ],
        [
          13,
          155,
          0,
          0,
          3,
          1,
          2,
          1,
          18,
          0
        ],
        [
          2,
          0,
          53,
          1,
          4,
          1,
          3,
          0,
          0,
          1
        ],
        [
          5,
          1,
          7,
          21,
          8,
          0,
          0,
          0,
          1,
          9
        ],
        [
          0,
          2,
          0,
          1,
          89,
          0,
          0,
          1,
          0,
          0
        ],
        [
          2,
          4,
          16,
          1,
          6,
          34,
          3,
          1,
          4,
          1
        ],
        [
          0,
          2,
          1,
          0,
          2,
          0,
          54,
          0,
          0,
          0
        ],
        [
          1,
          3,
          6,
          1,
          0,
          0,
          0,
          70,
          2,
          0
        ],
        [
          4,
          45,
          0,
          1,
          2,
          0,
          13,
          0,
          115,
          0
        ],
        [
          2,
          1,
          6,
          19,
          4,
          0,
          0,
          1,
          0,
          120
        ]
      ],
      "eval_loss": 0.9242235422134399,
      "eval_runtime": 15.2447,
      "eval_samples_per_second": 68.876,
      "eval_steps_per_second": 4.329,
      "step": 1000
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.9356917142868042,
      "learning_rate": 0.00015711252653927816,
      "loss": 0.3368,
      "step": 1010
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.9739004373550415,
      "learning_rate": 0.00015668789808917199,
      "loss": 0.3223,
      "step": 1020
    },
    {
      "epoch": 3.28,
      "grad_norm": 2.2368195056915283,
      "learning_rate": 0.00015626326963906584,
      "loss": 0.2729,
      "step": 1030
    },
    {
      "epoch": 3.31,
      "grad_norm": 2.3171839714050293,
      "learning_rate": 0.00015583864118895967,
      "loss": 0.2081,
      "step": 1040
    },
    {
      "epoch": 3.34,
      "grad_norm": 4.041688442230225,
      "learning_rate": 0.00015541401273885352,
      "loss": 0.3985,
      "step": 1050
    },
    {
      "epoch": 3.38,
      "grad_norm": 2.9750876426696777,
      "learning_rate": 0.00015498938428874735,
      "loss": 0.3221,
      "step": 1060
    },
    {
      "epoch": 3.41,
      "grad_norm": 2.438103675842285,
      "learning_rate": 0.0001545647558386412,
      "loss": 0.2598,
      "step": 1070
    },
    {
      "epoch": 3.44,
      "grad_norm": 3.9497318267822266,
      "learning_rate": 0.00015414012738853506,
      "loss": 0.2488,
      "step": 1080
    },
    {
      "epoch": 3.47,
      "grad_norm": 4.278964996337891,
      "learning_rate": 0.0001537154989384289,
      "loss": 0.2826,
      "step": 1090
    },
    {
      "epoch": 3.5,
      "grad_norm": 2.7707722187042236,
      "learning_rate": 0.00015329087048832275,
      "loss": 0.2695,
      "step": 1100
    },
    {
      "epoch": 3.5,
      "eval_accuracy": 0.7314285714285714,
      "eval_confusion_matrix": [
        [
          58,
          8,
          9,
          3,
          0,
          3,
          2,
          5,
          10,
          2
        ],
        [
          29,
          130,
          2,
          0,
          0,
          3,
          1,
          4,
          24,
          0
        ],
        [
          1,
          0,
          49,
          3,
          1,
          6,
          2,
          0,
          0,
          3
        ],
        [
          6,
          1,
          5,
          26,
          6,
          1,
          0,
          1,
          1,
          5
        ],
        [
          0,
          1,
          2,
          4,
          79,
          1,
          1,
          1,
          4,
          0
        ],
        [
          2,
          2,
          12,
          4,
          1,
          48,
          1,
          0,
          1,
          1
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          57,
          0,
          0,
          0
        ],
        [
          2,
          2,
          4,
          1,
          0,
          2,
          2,
          67,
          3,
          0
        ],
        [
          1,
          32,
          0,
          1,
          0,
          4,
          12,
          2,
          125,
          3
        ],
        [
          3,
          1,
          2,
          10,
          1,
          5,
          2,
          0,
          0,
          129
        ]
      ],
      "eval_loss": 0.9827544689178467,
      "eval_runtime": 15.624,
      "eval_samples_per_second": 67.204,
      "eval_steps_per_second": 4.224,
      "step": 1100
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.2058473825454712,
      "learning_rate": 0.00015286624203821657,
      "loss": 0.2301,
      "step": 1110
    },
    {
      "epoch": 3.57,
      "grad_norm": 4.785125732421875,
      "learning_rate": 0.00015244161358811043,
      "loss": 0.1879,
      "step": 1120
    },
    {
      "epoch": 3.6,
      "grad_norm": 2.2282400131225586,
      "learning_rate": 0.00015201698513800426,
      "loss": 0.2526,
      "step": 1130
    },
    {
      "epoch": 3.63,
      "grad_norm": 4.476231098175049,
      "learning_rate": 0.0001515923566878981,
      "loss": 0.2623,
      "step": 1140
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.7477174997329712,
      "learning_rate": 0.00015116772823779194,
      "loss": 0.2361,
      "step": 1150
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.9111158847808838,
      "learning_rate": 0.0001507430997876858,
      "loss": 0.2147,
      "step": 1160
    },
    {
      "epoch": 3.73,
      "grad_norm": 2.058471918106079,
      "learning_rate": 0.00015031847133757962,
      "loss": 0.2567,
      "step": 1170
    },
    {
      "epoch": 3.76,
      "grad_norm": 3.280287265777588,
      "learning_rate": 0.00014989384288747345,
      "loss": 0.3297,
      "step": 1180
    },
    {
      "epoch": 3.79,
      "grad_norm": 4.744434356689453,
      "learning_rate": 0.0001494692144373673,
      "loss": 0.2678,
      "step": 1190
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.6539621353149414,
      "learning_rate": 0.00014904458598726113,
      "loss": 0.2343,
      "step": 1200
    },
    {
      "epoch": 3.82,
      "eval_accuracy": 0.7295238095238096,
      "eval_confusion_matrix": [
        [
          58,
          2,
          10,
          5,
          1,
          1,
          7,
          4,
          10,
          2
        ],
        [
          21,
          115,
          0,
          0,
          4,
          0,
          12,
          3,
          38,
          0
        ],
        [
          1,
          0,
          53,
          2,
          2,
          3,
          0,
          1,
          1,
          2
        ],
        [
          2,
          2,
          9,
          22,
          9,
          0,
          1,
          0,
          1,
          6
        ],
        [
          0,
          2,
          1,
          1,
          88,
          0,
          0,
          0,
          1,
          0
        ],
        [
          2,
          1,
          10,
          4,
          2,
          48,
          1,
          2,
          2,
          0
        ],
        [
          0,
          0,
          1,
          0,
          2,
          0,
          56,
          0,
          0,
          0
        ],
        [
          2,
          1,
          7,
          0,
          0,
          0,
          0,
          72,
          1,
          0
        ],
        [
          4,
          22,
          0,
          1,
          0,
          0,
          14,
          0,
          139,
          0
        ],
        [
          7,
          0,
          4,
          20,
          1,
          3,
          1,
          2,
          0,
          115
        ]
      ],
      "eval_loss": 1.0871163606643677,
      "eval_runtime": 15.2438,
      "eval_samples_per_second": 68.881,
      "eval_steps_per_second": 4.33,
      "step": 1200
    },
    {
      "epoch": 3.85,
      "grad_norm": 2.8714191913604736,
      "learning_rate": 0.000148619957537155,
      "loss": 0.2571,
      "step": 1210
    },
    {
      "epoch": 3.89,
      "grad_norm": 3.6686244010925293,
      "learning_rate": 0.00014819532908704882,
      "loss": 0.2531,
      "step": 1220
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.589486837387085,
      "learning_rate": 0.00014777070063694267,
      "loss": 0.207,
      "step": 1230
    },
    {
      "epoch": 3.95,
      "grad_norm": 2.834082841873169,
      "learning_rate": 0.00014734607218683653,
      "loss": 0.2296,
      "step": 1240
    },
    {
      "epoch": 3.98,
      "grad_norm": 4.031200885772705,
      "learning_rate": 0.00014692144373673036,
      "loss": 0.2018,
      "step": 1250
    },
    {
      "epoch": 4.01,
      "grad_norm": 2.4544031620025635,
      "learning_rate": 0.0001464968152866242,
      "loss": 0.2817,
      "step": 1260
    },
    {
      "epoch": 4.04,
      "grad_norm": 1.3553822040557861,
      "learning_rate": 0.00014607218683651804,
      "loss": 0.2868,
      "step": 1270
    },
    {
      "epoch": 4.08,
      "grad_norm": 1.9867699146270752,
      "learning_rate": 0.0001456475583864119,
      "loss": 0.2127,
      "step": 1280
    },
    {
      "epoch": 4.11,
      "grad_norm": 1.536102533340454,
      "learning_rate": 0.00014522292993630572,
      "loss": 0.2743,
      "step": 1290
    },
    {
      "epoch": 4.14,
      "grad_norm": 2.0758562088012695,
      "learning_rate": 0.00014479830148619958,
      "loss": 0.2714,
      "step": 1300
    },
    {
      "epoch": 4.14,
      "eval_accuracy": 0.7314285714285714,
      "eval_confusion_matrix": [
        [
          59,
          6,
          8,
          8,
          1,
          1,
          3,
          4,
          7,
          3
        ],
        [
          23,
          114,
          2,
          1,
          1,
          0,
          5,
          3,
          42,
          2
        ],
        [
          1,
          1,
          54,
          2,
          1,
          2,
          0,
          1,
          0,
          3
        ],
        [
          3,
          1,
          3,
          32,
          4,
          0,
          0,
          0,
          1,
          8
        ],
        [
          0,
          1,
          3,
          5,
          80,
          1,
          1,
          0,
          0,
          2
        ],
        [
          3,
          1,
          11,
          7,
          2,
          43,
          1,
          2,
          1,
          1
        ],
        [
          0,
          0,
          1,
          1,
          0,
          0,
          56,
          0,
          0,
          1
        ],
        [
          1,
          0,
          4,
          0,
          0,
          0,
          0,
          77,
          1,
          0
        ],
        [
          6,
          31,
          2,
          1,
          0,
          0,
          10,
          0,
          130,
          0
        ],
        [
          5,
          0,
          1,
          22,
          0,
          1,
          0,
          1,
          0,
          123
        ]
      ],
      "eval_loss": 1.0719889402389526,
      "eval_runtime": 15.7109,
      "eval_samples_per_second": 66.833,
      "eval_steps_per_second": 4.201,
      "step": 1300
    },
    {
      "epoch": 4.17,
      "grad_norm": 2.6378695964813232,
      "learning_rate": 0.0001443736730360934,
      "loss": 0.1884,
      "step": 1310
    },
    {
      "epoch": 4.2,
      "grad_norm": 2.4962825775146484,
      "learning_rate": 0.00014394904458598726,
      "loss": 0.2421,
      "step": 1320
    },
    {
      "epoch": 4.24,
      "grad_norm": 5.444900035858154,
      "learning_rate": 0.0001435244161358811,
      "loss": 0.1745,
      "step": 1330
    },
    {
      "epoch": 4.27,
      "grad_norm": 3.8945775032043457,
      "learning_rate": 0.00014309978768577494,
      "loss": 0.3019,
      "step": 1340
    },
    {
      "epoch": 4.3,
      "grad_norm": 4.094300270080566,
      "learning_rate": 0.0001426751592356688,
      "loss": 0.2902,
      "step": 1350
    },
    {
      "epoch": 4.33,
      "grad_norm": 2.9124348163604736,
      "learning_rate": 0.00014225053078556263,
      "loss": 0.2889,
      "step": 1360
    },
    {
      "epoch": 4.36,
      "grad_norm": 1.5475099086761475,
      "learning_rate": 0.00014182590233545648,
      "loss": 0.1914,
      "step": 1370
    },
    {
      "epoch": 4.39,
      "grad_norm": 3.8749759197235107,
      "learning_rate": 0.0001414012738853503,
      "loss": 0.2835,
      "step": 1380
    },
    {
      "epoch": 4.43,
      "grad_norm": 3.574009656906128,
      "learning_rate": 0.00014097664543524416,
      "loss": 0.2556,
      "step": 1390
    },
    {
      "epoch": 4.46,
      "grad_norm": 1.3013603687286377,
      "learning_rate": 0.000140552016985138,
      "loss": 0.2287,
      "step": 1400
    },
    {
      "epoch": 4.46,
      "eval_accuracy": 0.7057142857142857,
      "eval_confusion_matrix": [
        [
          52,
          5,
          15,
          8,
          1,
          8,
          0,
          3,
          6,
          2
        ],
        [
          27,
          109,
          1,
          0,
          1,
          6,
          2,
          3,
          43,
          1
        ],
        [
          1,
          0,
          55,
          3,
          0,
          3,
          1,
          0,
          0,
          2
        ],
        [
          2,
          1,
          4,
          34,
          4,
          2,
          0,
          0,
          0,
          5
        ],
        [
          0,
          1,
          2,
          4,
          81,
          2,
          1,
          0,
          2,
          0
        ],
        [
          2,
          2,
          7,
          3,
          1,
          54,
          0,
          0,
          3,
          0
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          56,
          0,
          1,
          0
        ],
        [
          1,
          1,
          12,
          1,
          0,
          3,
          2,
          62,
          1,
          0
        ],
        [
          5,
          30,
          1,
          0,
          0,
          3,
          9,
          0,
          131,
          1
        ],
        [
          4,
          6,
          4,
          28,
          0,
          4,
          0,
          0,
          0,
          107
        ]
      ],
      "eval_loss": 1.112500786781311,
      "eval_runtime": 15.3686,
      "eval_samples_per_second": 68.321,
      "eval_steps_per_second": 4.294,
      "step": 1400
    },
    {
      "epoch": 4.49,
      "grad_norm": 2.4055957794189453,
      "learning_rate": 0.00014012738853503185,
      "loss": 0.1898,
      "step": 1410
    },
    {
      "epoch": 4.52,
      "grad_norm": 1.5158133506774902,
      "learning_rate": 0.00013970276008492568,
      "loss": 0.3715,
      "step": 1420
    },
    {
      "epoch": 4.55,
      "grad_norm": 2.5956978797912598,
      "learning_rate": 0.00013927813163481953,
      "loss": 0.1763,
      "step": 1430
    },
    {
      "epoch": 4.59,
      "grad_norm": 3.058171272277832,
      "learning_rate": 0.00013885350318471339,
      "loss": 0.2033,
      "step": 1440
    },
    {
      "epoch": 4.62,
      "grad_norm": 4.358069896697998,
      "learning_rate": 0.00013842887473460721,
      "loss": 0.1918,
      "step": 1450
    },
    {
      "epoch": 4.65,
      "grad_norm": 3.4236209392547607,
      "learning_rate": 0.00013800424628450107,
      "loss": 0.2298,
      "step": 1460
    },
    {
      "epoch": 4.68,
      "grad_norm": 2.0080718994140625,
      "learning_rate": 0.0001375796178343949,
      "loss": 0.2217,
      "step": 1470
    },
    {
      "epoch": 4.71,
      "grad_norm": 1.7190344333648682,
      "learning_rate": 0.00013715498938428875,
      "loss": 0.1858,
      "step": 1480
    },
    {
      "epoch": 4.75,
      "grad_norm": 1.8413598537445068,
      "learning_rate": 0.00013673036093418258,
      "loss": 0.2372,
      "step": 1490
    },
    {
      "epoch": 4.78,
      "grad_norm": 6.0616865158081055,
      "learning_rate": 0.00013630573248407644,
      "loss": 0.2814,
      "step": 1500
    },
    {
      "epoch": 4.78,
      "eval_accuracy": 0.72,
      "eval_confusion_matrix": [
        [
          71,
          3,
          7,
          5,
          2,
          1,
          1,
          6,
          3,
          1
        ],
        [
          53,
          111,
          1,
          0,
          1,
          0,
          5,
          5,
          17,
          0
        ],
        [
          2,
          0,
          48,
          4,
          1,
          4,
          0,
          4,
          0,
          2
        ],
        [
          6,
          0,
          4,
          31,
          6,
          0,
          0,
          0,
          0,
          5
        ],
        [
          0,
          3,
          1,
          4,
          82,
          0,
          1,
          1,
          1,
          0
        ],
        [
          8,
          0,
          4,
          4,
          1,
          49,
          1,
          4,
          1,
          0
        ],
        [
          2,
          0,
          1,
          1,
          1,
          0,
          52,
          2,
          0,
          0
        ],
        [
          2,
          0,
          1,
          0,
          0,
          0,
          0,
          80,
          0,
          0
        ],
        [
          11,
          35,
          2,
          0,
          0,
          1,
          9,
          5,
          117,
          0
        ],
        [
          10,
          1,
          1,
          21,
          1,
          0,
          2,
          2,
          0,
          115
        ]
      ],
      "eval_loss": 1.1163132190704346,
      "eval_runtime": 15.5932,
      "eval_samples_per_second": 67.337,
      "eval_steps_per_second": 4.233,
      "step": 1500
    },
    {
      "epoch": 4.81,
      "grad_norm": 3.689554214477539,
      "learning_rate": 0.00013588110403397026,
      "loss": 0.179,
      "step": 1510
    },
    {
      "epoch": 4.84,
      "grad_norm": 2.459721803665161,
      "learning_rate": 0.00013545647558386412,
      "loss": 0.1896,
      "step": 1520
    },
    {
      "epoch": 4.87,
      "grad_norm": 2.693918228149414,
      "learning_rate": 0.00013503184713375797,
      "loss": 0.1982,
      "step": 1530
    },
    {
      "epoch": 4.9,
      "grad_norm": 2.6597607135772705,
      "learning_rate": 0.0001346072186836518,
      "loss": 0.2241,
      "step": 1540
    },
    {
      "epoch": 4.94,
      "grad_norm": 4.53094482421875,
      "learning_rate": 0.00013418259023354566,
      "loss": 0.2084,
      "step": 1550
    },
    {
      "epoch": 4.97,
      "grad_norm": 0.6765243411064148,
      "learning_rate": 0.00013375796178343948,
      "loss": 0.1917,
      "step": 1560
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.6149380207061768,
      "learning_rate": 0.00013333333333333334,
      "loss": 0.2102,
      "step": 1570
    },
    {
      "epoch": 5.03,
      "grad_norm": 2.5928609371185303,
      "learning_rate": 0.00013290870488322717,
      "loss": 0.239,
      "step": 1580
    },
    {
      "epoch": 5.06,
      "grad_norm": 4.47957181930542,
      "learning_rate": 0.00013248407643312102,
      "loss": 0.2067,
      "step": 1590
    },
    {
      "epoch": 5.1,
      "grad_norm": 3.5347230434417725,
      "learning_rate": 0.00013205944798301485,
      "loss": 0.2648,
      "step": 1600
    },
    {
      "epoch": 5.1,
      "eval_accuracy": 0.7057142857142857,
      "eval_confusion_matrix": [
        [
          61,
          6,
          3,
          3,
          0,
          3,
          2,
          12,
          8,
          2
        ],
        [
          27,
          131,
          0,
          0,
          4,
          0,
          1,
          9,
          21,
          0
        ],
        [
          2,
          2,
          51,
          2,
          2,
          4,
          0,
          1,
          0,
          1
        ],
        [
          3,
          1,
          4,
          28,
          6,
          3,
          0,
          3,
          1,
          3
        ],
        [
          1,
          3,
          0,
          4,
          82,
          1,
          0,
          1,
          1,
          0
        ],
        [
          1,
          0,
          8,
          4,
          3,
          51,
          0,
          3,
          1,
          1
        ],
        [
          3,
          1,
          1,
          1,
          1,
          0,
          51,
          1,
          0,
          0
        ],
        [
          1,
          0,
          2,
          1,
          0,
          0,
          0,
          79,
          0,
          0
        ],
        [
          9,
          31,
          1,
          0,
          0,
          1,
          11,
          10,
          117,
          0
        ],
        [
          17,
          5,
          4,
          28,
          2,
          2,
          1,
          4,
          0,
          90
        ]
      ],
      "eval_loss": 1.1720563173294067,
      "eval_runtime": 15.166,
      "eval_samples_per_second": 69.234,
      "eval_steps_per_second": 4.352,
      "step": 1600
    },
    {
      "epoch": 5.13,
      "grad_norm": 2.3316214084625244,
      "learning_rate": 0.0001316348195329087,
      "loss": 0.1836,
      "step": 1610
    },
    {
      "epoch": 5.16,
      "grad_norm": 1.4129242897033691,
      "learning_rate": 0.00013121019108280253,
      "loss": 0.2966,
      "step": 1620
    },
    {
      "epoch": 5.19,
      "grad_norm": 3.569037437438965,
      "learning_rate": 0.0001307855626326964,
      "loss": 0.2079,
      "step": 1630
    },
    {
      "epoch": 5.22,
      "grad_norm": 3.183149576187134,
      "learning_rate": 0.00013036093418259024,
      "loss": 0.2142,
      "step": 1640
    },
    {
      "epoch": 5.25,
      "grad_norm": 1.4530551433563232,
      "learning_rate": 0.00012993630573248407,
      "loss": 0.1636,
      "step": 1650
    },
    {
      "epoch": 5.29,
      "grad_norm": 1.5123274326324463,
      "learning_rate": 0.00012951167728237793,
      "loss": 0.1772,
      "step": 1660
    },
    {
      "epoch": 5.32,
      "grad_norm": 3.7511801719665527,
      "learning_rate": 0.00012908704883227176,
      "loss": 0.2243,
      "step": 1670
    },
    {
      "epoch": 5.35,
      "grad_norm": 2.2168476581573486,
      "learning_rate": 0.0001286624203821656,
      "loss": 0.2089,
      "step": 1680
    },
    {
      "epoch": 5.38,
      "grad_norm": 0.5811095237731934,
      "learning_rate": 0.00012823779193205944,
      "loss": 0.1868,
      "step": 1690
    },
    {
      "epoch": 5.41,
      "grad_norm": 2.5435843467712402,
      "learning_rate": 0.0001278131634819533,
      "loss": 0.1857,
      "step": 1700
    },
    {
      "epoch": 5.41,
      "eval_accuracy": 0.7514285714285714,
      "eval_confusion_matrix": [
        [
          57,
          9,
          5,
          2,
          1,
          0,
          4,
          7,
          11,
          4
        ],
        [
          22,
          131,
          0,
          0,
          0,
          0,
          5,
          4,
          30,
          1
        ],
        [
          1,
          0,
          56,
          1,
          1,
          2,
          1,
          0,
          0,
          3
        ],
        [
          3,
          1,
          2,
          28,
          8,
          1,
          1,
          1,
          1,
          6
        ],
        [
          1,
          1,
          0,
          3,
          85,
          0,
          0,
          0,
          3,
          0
        ],
        [
          6,
          2,
          11,
          4,
          3,
          36,
          2,
          5,
          1,
          2
        ],
        [
          0,
          0,
          1,
          0,
          0,
          0,
          58,
          0,
          0,
          0
        ],
        [
          1,
          0,
          2,
          0,
          0,
          0,
          0,
          80,
          0,
          0
        ],
        [
          7,
          32,
          1,
          0,
          0,
          0,
          19,
          3,
          117,
          1
        ],
        [
          6,
          0,
          1,
          3,
          0,
          0,
          1,
          1,
          0,
          141
        ]
      ],
      "eval_loss": 1.0403919219970703,
      "eval_runtime": 15.5401,
      "eval_samples_per_second": 67.567,
      "eval_steps_per_second": 4.247,
      "step": 1700
    },
    {
      "epoch": 5.45,
      "grad_norm": 2.8231258392333984,
      "learning_rate": 0.00012738853503184712,
      "loss": 0.1562,
      "step": 1710
    },
    {
      "epoch": 5.48,
      "grad_norm": 3.1249213218688965,
      "learning_rate": 0.00012696390658174098,
      "loss": 0.1543,
      "step": 1720
    },
    {
      "epoch": 5.51,
      "grad_norm": 0.47615599632263184,
      "learning_rate": 0.00012653927813163483,
      "loss": 0.1578,
      "step": 1730
    },
    {
      "epoch": 5.54,
      "grad_norm": 0.8500877618789673,
      "learning_rate": 0.00012611464968152866,
      "loss": 0.1791,
      "step": 1740
    },
    {
      "epoch": 5.57,
      "grad_norm": 3.530510902404785,
      "learning_rate": 0.00012569002123142252,
      "loss": 0.1701,
      "step": 1750
    },
    {
      "epoch": 5.61,
      "grad_norm": 3.1433258056640625,
      "learning_rate": 0.00012526539278131634,
      "loss": 0.0973,
      "step": 1760
    },
    {
      "epoch": 5.64,
      "grad_norm": 2.9921345710754395,
      "learning_rate": 0.0001248407643312102,
      "loss": 0.2193,
      "step": 1770
    },
    {
      "epoch": 5.67,
      "grad_norm": 3.721479892730713,
      "learning_rate": 0.00012441613588110403,
      "loss": 0.265,
      "step": 1780
    },
    {
      "epoch": 5.7,
      "grad_norm": 0.5069979429244995,
      "learning_rate": 0.00012399150743099788,
      "loss": 0.2341,
      "step": 1790
    },
    {
      "epoch": 5.73,
      "grad_norm": 0.4565420150756836,
      "learning_rate": 0.0001235668789808917,
      "loss": 0.1958,
      "step": 1800
    },
    {
      "epoch": 5.73,
      "eval_accuracy": 0.7238095238095238,
      "eval_confusion_matrix": [
        [
          53,
          7,
          4,
          3,
          2,
          1,
          6,
          9,
          13,
          2
        ],
        [
          16,
          134,
          0,
          0,
          0,
          1,
          10,
          7,
          25,
          0
        ],
        [
          2,
          1,
          54,
          2,
          1,
          1,
          1,
          0,
          0,
          3
        ],
        [
          6,
          0,
          3,
          29,
          8,
          0,
          1,
          1,
          1,
          3
        ],
        [
          0,
          2,
          0,
          2,
          85,
          0,
          2,
          0,
          2,
          0
        ],
        [
          7,
          1,
          9,
          2,
          4,
          43,
          2,
          2,
          1,
          1
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          57,
          0,
          0,
          0
        ],
        [
          0,
          1,
          4,
          0,
          0,
          0,
          1,
          77,
          0,
          0
        ],
        [
          4,
          28,
          2,
          0,
          0,
          0,
          19,
          4,
          123,
          0
        ],
        [
          7,
          0,
          2,
          25,
          1,
          0,
          11,
          2,
          0,
          105
        ]
      ],
      "eval_loss": 1.1391521692276,
      "eval_runtime": 15.287,
      "eval_samples_per_second": 68.686,
      "eval_steps_per_second": 4.317,
      "step": 1800
    },
    {
      "epoch": 5.76,
      "grad_norm": 1.6156843900680542,
      "learning_rate": 0.00012314225053078556,
      "loss": 0.1635,
      "step": 1810
    },
    {
      "epoch": 5.8,
      "grad_norm": 2.8691422939300537,
      "learning_rate": 0.00012271762208067942,
      "loss": 0.2309,
      "step": 1820
    },
    {
      "epoch": 5.83,
      "grad_norm": 3.7324392795562744,
      "learning_rate": 0.00012229299363057325,
      "loss": 0.1626,
      "step": 1830
    },
    {
      "epoch": 5.86,
      "grad_norm": 3.8623502254486084,
      "learning_rate": 0.00012186836518046709,
      "loss": 0.1487,
      "step": 1840
    },
    {
      "epoch": 5.89,
      "grad_norm": 4.658466815948486,
      "learning_rate": 0.00012144373673036093,
      "loss": 0.3255,
      "step": 1850
    },
    {
      "epoch": 5.92,
      "grad_norm": 2.3121321201324463,
      "learning_rate": 0.00012101910828025477,
      "loss": 0.2092,
      "step": 1860
    },
    {
      "epoch": 5.96,
      "grad_norm": 1.5467556715011597,
      "learning_rate": 0.00012059447983014863,
      "loss": 0.175,
      "step": 1870
    },
    {
      "epoch": 5.99,
      "grad_norm": 3.587686777114868,
      "learning_rate": 0.00012016985138004247,
      "loss": 0.1644,
      "step": 1880
    },
    {
      "epoch": 6.02,
      "grad_norm": 3.9866294860839844,
      "learning_rate": 0.00011974522292993631,
      "loss": 0.1916,
      "step": 1890
    },
    {
      "epoch": 6.05,
      "grad_norm": 0.14543366432189941,
      "learning_rate": 0.00011932059447983015,
      "loss": 0.1475,
      "step": 1900
    },
    {
      "epoch": 6.05,
      "eval_accuracy": 0.7238095238095238,
      "eval_confusion_matrix": [
        [
          72,
          6,
          4,
          4,
          0,
          1,
          2,
          0,
          8,
          3
        ],
        [
          52,
          97,
          0,
          0,
          1,
          0,
          1,
          0,
          41,
          1
        ],
        [
          3,
          1,
          52,
          3,
          1,
          1,
          1,
          0,
          0,
          3
        ],
        [
          6,
          1,
          3,
          32,
          2,
          1,
          0,
          1,
          1,
          5
        ],
        [
          0,
          4,
          1,
          3,
          79,
          1,
          0,
          0,
          3,
          2
        ],
        [
          3,
          2,
          12,
          6,
          0,
          43,
          0,
          1,
          2,
          3
        ],
        [
          3,
          0,
          1,
          0,
          1,
          0,
          52,
          0,
          1,
          1
        ],
        [
          8,
          0,
          7,
          2,
          0,
          0,
          0,
          66,
          0,
          0
        ],
        [
          13,
          26,
          1,
          1,
          0,
          1,
          8,
          1,
          129,
          0
        ],
        [
          6,
          0,
          0,
          7,
          0,
          0,
          0,
          1,
          1,
          138
        ]
      ],
      "eval_loss": 1.1925528049468994,
      "eval_runtime": 15.2575,
      "eval_samples_per_second": 68.819,
      "eval_steps_per_second": 4.326,
      "step": 1900
    },
    {
      "epoch": 6.08,
      "grad_norm": 3.591832160949707,
      "learning_rate": 0.000118895966029724,
      "loss": 0.1943,
      "step": 1910
    },
    {
      "epoch": 6.11,
      "grad_norm": 1.7828431129455566,
      "learning_rate": 0.00011847133757961784,
      "loss": 0.1596,
      "step": 1920
    },
    {
      "epoch": 6.15,
      "grad_norm": 3.0881214141845703,
      "learning_rate": 0.00011804670912951168,
      "loss": 0.2402,
      "step": 1930
    },
    {
      "epoch": 6.18,
      "grad_norm": 1.4464250802993774,
      "learning_rate": 0.00011762208067940552,
      "loss": 0.1438,
      "step": 1940
    },
    {
      "epoch": 6.21,
      "grad_norm": 2.3067240715026855,
      "learning_rate": 0.00011719745222929936,
      "loss": 0.194,
      "step": 1950
    },
    {
      "epoch": 6.24,
      "grad_norm": 1.2956265211105347,
      "learning_rate": 0.0001167728237791932,
      "loss": 0.2229,
      "step": 1960
    },
    {
      "epoch": 6.27,
      "grad_norm": 2.0302717685699463,
      "learning_rate": 0.00011634819532908706,
      "loss": 0.1599,
      "step": 1970
    },
    {
      "epoch": 6.31,
      "grad_norm": 3.4628517627716064,
      "learning_rate": 0.0001159235668789809,
      "loss": 0.163,
      "step": 1980
    },
    {
      "epoch": 6.34,
      "grad_norm": 4.217594146728516,
      "learning_rate": 0.00011549893842887474,
      "loss": 0.1439,
      "step": 1990
    },
    {
      "epoch": 6.37,
      "grad_norm": 2.3878486156463623,
      "learning_rate": 0.00011507430997876858,
      "loss": 0.1443,
      "step": 2000
    },
    {
      "epoch": 6.37,
      "eval_accuracy": 0.7152380952380952,
      "eval_confusion_matrix": [
        [
          64,
          3,
          18,
          4,
          1,
          3,
          1,
          3,
          2,
          1
        ],
        [
          26,
          112,
          1,
          0,
          2,
          0,
          5,
          6,
          41,
          0
        ],
        [
          4,
          0,
          54,
          2,
          1,
          1,
          0,
          0,
          0,
          3
        ],
        [
          7,
          1,
          3,
          34,
          3,
          1,
          0,
          1,
          1,
          1
        ],
        [
          0,
          3,
          0,
          3,
          82,
          0,
          2,
          0,
          3,
          0
        ],
        [
          5,
          2,
          11,
          5,
          1,
          44,
          1,
          1,
          1,
          1
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          57,
          0,
          0,
          0
        ],
        [
          1,
          0,
          7,
          0,
          0,
          4,
          0,
          71,
          0,
          0
        ],
        [
          5,
          23,
          5,
          2,
          0,
          0,
          12,
          5,
          128,
          0
        ],
        [
          6,
          3,
          1,
          36,
          0,
          0,
          0,
          2,
          0,
          105
        ]
      ],
      "eval_loss": 1.2271040678024292,
      "eval_runtime": 15.4096,
      "eval_samples_per_second": 68.139,
      "eval_steps_per_second": 4.283,
      "step": 2000
    },
    {
      "epoch": 6.4,
      "grad_norm": 2.9746904373168945,
      "learning_rate": 0.00011464968152866242,
      "loss": 0.1776,
      "step": 2010
    },
    {
      "epoch": 6.43,
      "grad_norm": 4.259703159332275,
      "learning_rate": 0.00011422505307855626,
      "loss": 0.1744,
      "step": 2020
    },
    {
      "epoch": 6.46,
      "grad_norm": 2.885249137878418,
      "learning_rate": 0.0001138004246284501,
      "loss": 0.1597,
      "step": 2030
    },
    {
      "epoch": 6.5,
      "grad_norm": 1.6212122440338135,
      "learning_rate": 0.00011337579617834395,
      "loss": 0.2183,
      "step": 2040
    },
    {
      "epoch": 6.53,
      "grad_norm": 2.243490695953369,
      "learning_rate": 0.00011295116772823779,
      "loss": 0.205,
      "step": 2050
    },
    {
      "epoch": 6.56,
      "grad_norm": 2.2670421600341797,
      "learning_rate": 0.00011252653927813163,
      "loss": 0.1432,
      "step": 2060
    },
    {
      "epoch": 6.59,
      "grad_norm": 2.7547404766082764,
      "learning_rate": 0.00011210191082802549,
      "loss": 0.1276,
      "step": 2070
    },
    {
      "epoch": 6.62,
      "grad_norm": 2.6618921756744385,
      "learning_rate": 0.00011167728237791933,
      "loss": 0.1829,
      "step": 2080
    },
    {
      "epoch": 6.66,
      "grad_norm": 4.215251922607422,
      "learning_rate": 0.00011125265392781317,
      "loss": 0.1323,
      "step": 2090
    },
    {
      "epoch": 6.69,
      "grad_norm": 3.4978456497192383,
      "learning_rate": 0.00011082802547770701,
      "loss": 0.1453,
      "step": 2100
    },
    {
      "epoch": 6.69,
      "eval_accuracy": 0.7390476190476191,
      "eval_confusion_matrix": [
        [
          71,
          4,
          11,
          3,
          0,
          4,
          1,
          3,
          2,
          1
        ],
        [
          26,
          127,
          3,
          0,
          0,
          2,
          4,
          4,
          27,
          0
        ],
        [
          1,
          0,
          53,
          2,
          2,
          4,
          1,
          0,
          0,
          2
        ],
        [
          5,
          2,
          5,
          27,
          6,
          2,
          0,
          1,
          1,
          3
        ],
        [
          1,
          1,
          0,
          1,
          87,
          1,
          1,
          0,
          1,
          0
        ],
        [
          2,
          1,
          6,
          1,
          2,
          58,
          0,
          1,
          1,
          0
        ],
        [
          2,
          3,
          1,
          0,
          2,
          0,
          50,
          0,
          0,
          1
        ],
        [
          4,
          0,
          7,
          0,
          0,
          4,
          0,
          68,
          0,
          0
        ],
        [
          4,
          34,
          4,
          0,
          1,
          3,
          14,
          1,
          119,
          0
        ],
        [
          9,
          1,
          2,
          18,
          2,
          1,
          0,
          3,
          1,
          116
        ]
      ],
      "eval_loss": 1.0546270608901978,
      "eval_runtime": 15.5179,
      "eval_samples_per_second": 67.664,
      "eval_steps_per_second": 4.253,
      "step": 2100
    },
    {
      "epoch": 6.72,
      "grad_norm": 1.68947172164917,
      "learning_rate": 0.00011040339702760085,
      "loss": 0.1613,
      "step": 2110
    },
    {
      "epoch": 6.75,
      "grad_norm": 3.1171562671661377,
      "learning_rate": 0.0001099787685774947,
      "loss": 0.176,
      "step": 2120
    },
    {
      "epoch": 6.78,
      "grad_norm": 2.0694828033447266,
      "learning_rate": 0.00010955414012738854,
      "loss": 0.1529,
      "step": 2130
    },
    {
      "epoch": 6.82,
      "grad_norm": 1.485595703125,
      "learning_rate": 0.00010912951167728238,
      "loss": 0.1829,
      "step": 2140
    },
    {
      "epoch": 6.85,
      "grad_norm": 2.217650890350342,
      "learning_rate": 0.00010870488322717622,
      "loss": 0.2196,
      "step": 2150
    },
    {
      "epoch": 6.88,
      "grad_norm": 1.3161414861679077,
      "learning_rate": 0.00010828025477707007,
      "loss": 0.1002,
      "step": 2160
    },
    {
      "epoch": 6.91,
      "grad_norm": 3.4168200492858887,
      "learning_rate": 0.00010785562632696392,
      "loss": 0.1856,
      "step": 2170
    },
    {
      "epoch": 6.94,
      "grad_norm": 2.1230216026306152,
      "learning_rate": 0.00010743099787685776,
      "loss": 0.1529,
      "step": 2180
    },
    {
      "epoch": 6.97,
      "grad_norm": 2.512233018875122,
      "learning_rate": 0.0001070063694267516,
      "loss": 0.2026,
      "step": 2190
    },
    {
      "epoch": 7.01,
      "grad_norm": 2.9873499870300293,
      "learning_rate": 0.00010658174097664544,
      "loss": 0.2319,
      "step": 2200
    },
    {
      "epoch": 7.01,
      "eval_accuracy": 0.7371428571428571,
      "eval_confusion_matrix": [
        [
          60,
          4,
          9,
          7,
          1,
          4,
          2,
          2,
          10,
          1
        ],
        [
          18,
          127,
          1,
          0,
          2,
          0,
          9,
          2,
          34,
          0
        ],
        [
          3,
          0,
          53,
          3,
          1,
          2,
          0,
          1,
          0,
          2
        ],
        [
          4,
          2,
          2,
          36,
          6,
          0,
          0,
          0,
          1,
          1
        ],
        [
          0,
          4,
          0,
          3,
          83,
          0,
          1,
          1,
          1,
          0
        ],
        [
          2,
          2,
          9,
          6,
          1,
          49,
          1,
          0,
          2,
          0
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          57,
          0,
          0,
          0
        ],
        [
          1,
          0,
          10,
          1,
          0,
          0,
          0,
          71,
          0,
          0
        ],
        [
          5,
          24,
          4,
          0,
          0,
          1,
          15,
          1,
          130,
          0
        ],
        [
          4,
          4,
          2,
          28,
          0,
          0,
          2,
          5,
          0,
          108
        ]
      ],
      "eval_loss": 1.0889815092086792,
      "eval_runtime": 15.3147,
      "eval_samples_per_second": 68.562,
      "eval_steps_per_second": 4.31,
      "step": 2200
    },
    {
      "epoch": 7.04,
      "grad_norm": 2.2253472805023193,
      "learning_rate": 0.00010615711252653928,
      "loss": 0.162,
      "step": 2210
    },
    {
      "epoch": 7.07,
      "grad_norm": 2.6832664012908936,
      "learning_rate": 0.00010573248407643312,
      "loss": 0.1435,
      "step": 2220
    },
    {
      "epoch": 7.1,
      "grad_norm": 2.4758527278900146,
      "learning_rate": 0.00010530785562632696,
      "loss": 0.1725,
      "step": 2230
    },
    {
      "epoch": 7.13,
      "grad_norm": 2.8945116996765137,
      "learning_rate": 0.0001048832271762208,
      "loss": 0.1816,
      "step": 2240
    },
    {
      "epoch": 7.17,
      "grad_norm": 1.7378908395767212,
      "learning_rate": 0.00010445859872611465,
      "loss": 0.1915,
      "step": 2250
    },
    {
      "epoch": 7.2,
      "grad_norm": 3.76171612739563,
      "learning_rate": 0.0001040339702760085,
      "loss": 0.1212,
      "step": 2260
    },
    {
      "epoch": 7.23,
      "grad_norm": 2.1323678493499756,
      "learning_rate": 0.00010360934182590234,
      "loss": 0.1488,
      "step": 2270
    },
    {
      "epoch": 7.26,
      "grad_norm": 3.7445242404937744,
      "learning_rate": 0.00010318471337579619,
      "loss": 0.1953,
      "step": 2280
    },
    {
      "epoch": 7.29,
      "grad_norm": 2.166163682937622,
      "learning_rate": 0.00010276008492569003,
      "loss": 0.1469,
      "step": 2290
    },
    {
      "epoch": 7.32,
      "grad_norm": 3.0045981407165527,
      "learning_rate": 0.00010233545647558387,
      "loss": 0.1499,
      "step": 2300
    },
    {
      "epoch": 7.32,
      "eval_accuracy": 0.7,
      "eval_confusion_matrix": [
        [
          68,
          3,
          3,
          11,
          1,
          1,
          1,
          4,
          7,
          1
        ],
        [
          60,
          82,
          0,
          2,
          6,
          0,
          6,
          4,
          31,
          2
        ],
        [
          2,
          1,
          43,
          6,
          2,
          4,
          2,
          2,
          0,
          3
        ],
        [
          2,
          0,
          2,
          36,
          6,
          0,
          0,
          1,
          1,
          4
        ],
        [
          1,
          2,
          1,
          2,
          83,
          2,
          0,
          1,
          0,
          1
        ],
        [
          4,
          0,
          3,
          11,
          2,
          46,
          0,
          3,
          2,
          1
        ],
        [
          0,
          0,
          0,
          0,
          1,
          1,
          54,
          1,
          1,
          1
        ],
        [
          4,
          0,
          1,
          1,
          0,
          0,
          0,
          76,
          1,
          0
        ],
        [
          9,
          22,
          0,
          1,
          2,
          0,
          16,
          2,
          127,
          1
        ],
        [
          1,
          0,
          1,
          27,
          0,
          0,
          0,
          4,
          0,
          120
        ]
      ],
      "eval_loss": 1.3652293682098389,
      "eval_runtime": 15.3818,
      "eval_samples_per_second": 68.262,
      "eval_steps_per_second": 4.291,
      "step": 2300
    },
    {
      "epoch": 7.36,
      "grad_norm": 2.051771879196167,
      "learning_rate": 0.00010191082802547771,
      "loss": 0.2339,
      "step": 2310
    },
    {
      "epoch": 7.39,
      "grad_norm": 2.841862201690674,
      "learning_rate": 0.00010148619957537155,
      "loss": 0.2014,
      "step": 2320
    },
    {
      "epoch": 7.42,
      "grad_norm": 5.293687343597412,
      "learning_rate": 0.0001010615711252654,
      "loss": 0.1718,
      "step": 2330
    },
    {
      "epoch": 7.45,
      "grad_norm": 0.5550873279571533,
      "learning_rate": 0.00010063694267515924,
      "loss": 0.142,
      "step": 2340
    },
    {
      "epoch": 7.48,
      "grad_norm": 4.287944316864014,
      "learning_rate": 0.00010021231422505308,
      "loss": 0.1815,
      "step": 2350
    },
    {
      "epoch": 7.52,
      "grad_norm": 1.0735574960708618,
      "learning_rate": 9.978768577494693e-05,
      "loss": 0.1861,
      "step": 2360
    },
    {
      "epoch": 7.55,
      "grad_norm": 3.090205430984497,
      "learning_rate": 9.936305732484077e-05,
      "loss": 0.169,
      "step": 2370
    },
    {
      "epoch": 7.58,
      "grad_norm": 2.5464820861816406,
      "learning_rate": 9.893842887473462e-05,
      "loss": 0.2243,
      "step": 2380
    },
    {
      "epoch": 7.61,
      "grad_norm": 4.008731365203857,
      "learning_rate": 9.851380042462846e-05,
      "loss": 0.1971,
      "step": 2390
    },
    {
      "epoch": 7.64,
      "grad_norm": 3.4554805755615234,
      "learning_rate": 9.80891719745223e-05,
      "loss": 0.1467,
      "step": 2400
    },
    {
      "epoch": 7.64,
      "eval_accuracy": 0.6676190476190477,
      "eval_confusion_matrix": [
        [
          59,
          3,
          10,
          7,
          0,
          7,
          1,
          2,
          8,
          3
        ],
        [
          55,
          65,
          1,
          0,
          4,
          4,
          15,
          3,
          46,
          0
        ],
        [
          3,
          1,
          48,
          6,
          2,
          1,
          2,
          0,
          0,
          2
        ],
        [
          1,
          0,
          3,
          34,
          5,
          1,
          0,
          0,
          1,
          7
        ],
        [
          0,
          4,
          1,
          2,
          83,
          2,
          0,
          0,
          0,
          1
        ],
        [
          4,
          2,
          12,
          3,
          1,
          46,
          0,
          0,
          2,
          2
        ],
        [
          1,
          0,
          1,
          0,
          1,
          0,
          56,
          0,
          0,
          0
        ],
        [
          3,
          0,
          5,
          1,
          0,
          2,
          4,
          68,
          0,
          0
        ],
        [
          9,
          18,
          1,
          0,
          3,
          1,
          17,
          1,
          129,
          1
        ],
        [
          2,
          3,
          1,
          32,
          0,
          0,
          0,
          2,
          0,
          113
        ]
      ],
      "eval_loss": 1.4623304605484009,
      "eval_runtime": 15.2878,
      "eval_samples_per_second": 68.682,
      "eval_steps_per_second": 4.317,
      "step": 2400
    },
    {
      "epoch": 7.68,
      "grad_norm": 1.7582330703735352,
      "learning_rate": 9.766454352441614e-05,
      "loss": 0.1449,
      "step": 2410
    },
    {
      "epoch": 7.71,
      "grad_norm": 1.2631497383117676,
      "learning_rate": 9.723991507430998e-05,
      "loss": 0.1599,
      "step": 2420
    },
    {
      "epoch": 7.74,
      "grad_norm": 0.9202260971069336,
      "learning_rate": 9.681528662420382e-05,
      "loss": 0.0866,
      "step": 2430
    },
    {
      "epoch": 7.77,
      "grad_norm": 0.5633217692375183,
      "learning_rate": 9.639065817409766e-05,
      "loss": 0.2013,
      "step": 2440
    },
    {
      "epoch": 7.8,
      "grad_norm": 0.0941554605960846,
      "learning_rate": 9.59660297239915e-05,
      "loss": 0.1761,
      "step": 2450
    },
    {
      "epoch": 7.83,
      "grad_norm": 1.5101308822631836,
      "learning_rate": 9.554140127388536e-05,
      "loss": 0.0721,
      "step": 2460
    },
    {
      "epoch": 7.87,
      "grad_norm": 2.1253440380096436,
      "learning_rate": 9.51167728237792e-05,
      "loss": 0.1284,
      "step": 2470
    },
    {
      "epoch": 7.9,
      "grad_norm": 2.5228655338287354,
      "learning_rate": 9.469214437367304e-05,
      "loss": 0.141,
      "step": 2480
    },
    {
      "epoch": 7.93,
      "grad_norm": 1.9860364198684692,
      "learning_rate": 9.426751592356689e-05,
      "loss": 0.1518,
      "step": 2490
    },
    {
      "epoch": 7.96,
      "grad_norm": 0.04018540680408478,
      "learning_rate": 9.384288747346073e-05,
      "loss": 0.1163,
      "step": 2500
    },
    {
      "epoch": 7.96,
      "eval_accuracy": 0.6819047619047619,
      "eval_confusion_matrix": [
        [
          53,
          2,
          15,
          7,
          0,
          3,
          6,
          4,
          5,
          5
        ],
        [
          62,
          76,
          2,
          0,
          4,
          2,
          19,
          7,
          16,
          5
        ],
        [
          1,
          1,
          52,
          1,
          1,
          2,
          2,
          2,
          0,
          3
        ],
        [
          1,
          0,
          5,
          28,
          6,
          2,
          1,
          2,
          0,
          7
        ],
        [
          0,
          1,
          1,
          2,
          83,
          3,
          1,
          1,
          0,
          1
        ],
        [
          2,
          1,
          13,
          3,
          0,
          44,
          1,
          4,
          0,
          4
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          57,
          0,
          0,
          0
        ],
        [
          1,
          0,
          5,
          0,
          0,
          1,
          1,
          75,
          0,
          0
        ],
        [
          11,
          17,
          1,
          0,
          1,
          1,
          28,
          3,
          116,
          2
        ],
        [
          0,
          3,
          2,
          10,
          0,
          1,
          3,
          2,
          0,
          132
        ]
      ],
      "eval_loss": 1.530079960823059,
      "eval_runtime": 15.6324,
      "eval_samples_per_second": 67.168,
      "eval_steps_per_second": 4.222,
      "step": 2500
    },
    {
      "epoch": 7.99,
      "grad_norm": 2.986940622329712,
      "learning_rate": 9.341825902335457e-05,
      "loss": 0.1317,
      "step": 2510
    },
    {
      "epoch": 8.03,
      "grad_norm": 2.86224627494812,
      "learning_rate": 9.299363057324841e-05,
      "loss": 0.133,
      "step": 2520
    },
    {
      "epoch": 8.06,
      "grad_norm": 1.1539607048034668,
      "learning_rate": 9.256900212314225e-05,
      "loss": 0.1061,
      "step": 2530
    },
    {
      "epoch": 8.09,
      "grad_norm": 1.738471508026123,
      "learning_rate": 9.21443736730361e-05,
      "loss": 0.0456,
      "step": 2540
    },
    {
      "epoch": 8.12,
      "grad_norm": 4.975551605224609,
      "learning_rate": 9.171974522292994e-05,
      "loss": 0.1636,
      "step": 2550
    },
    {
      "epoch": 8.15,
      "grad_norm": 2.86753511428833,
      "learning_rate": 9.129511677282379e-05,
      "loss": 0.2237,
      "step": 2560
    },
    {
      "epoch": 8.18,
      "grad_norm": 2.0240707397460938,
      "learning_rate": 9.087048832271763e-05,
      "loss": 0.1529,
      "step": 2570
    },
    {
      "epoch": 8.22,
      "grad_norm": 3.7712082862854004,
      "learning_rate": 9.044585987261147e-05,
      "loss": 0.1449,
      "step": 2580
    },
    {
      "epoch": 8.25,
      "grad_norm": 2.0401346683502197,
      "learning_rate": 9.002123142250532e-05,
      "loss": 0.1584,
      "step": 2590
    },
    {
      "epoch": 8.28,
      "grad_norm": 1.6185662746429443,
      "learning_rate": 8.959660297239916e-05,
      "loss": 0.1087,
      "step": 2600
    },
    {
      "epoch": 8.28,
      "eval_accuracy": 0.7323809523809524,
      "eval_confusion_matrix": [
        [
          62,
          6,
          5,
          6,
          0,
          0,
          2,
          2,
          12,
          5
        ],
        [
          32,
          102,
          0,
          0,
          2,
          1,
          12,
          3,
          41,
          0
        ],
        [
          3,
          2,
          45,
          4,
          1,
          3,
          3,
          0,
          0,
          4
        ],
        [
          5,
          0,
          3,
          29,
          3,
          0,
          0,
          0,
          3,
          9
        ],
        [
          1,
          5,
          0,
          4,
          73,
          2,
          1,
          1,
          4,
          2
        ],
        [
          5,
          3,
          3,
          6,
          1,
          43,
          0,
          3,
          3,
          5
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          57,
          0,
          0,
          0
        ],
        [
          1,
          0,
          5,
          1,
          0,
          0,
          1,
          72,
          3,
          0
        ],
        [
          3,
          21,
          0,
          1,
          1,
          0,
          9,
          1,
          142,
          2
        ],
        [
          1,
          0,
          1,
          6,
          0,
          0,
          0,
          1,
          0,
          144
        ]
      ],
      "eval_loss": 1.2230817079544067,
      "eval_runtime": 15.3373,
      "eval_samples_per_second": 68.46,
      "eval_steps_per_second": 4.303,
      "step": 2600
    },
    {
      "epoch": 8.31,
      "grad_norm": 1.1602622270584106,
      "learning_rate": 8.9171974522293e-05,
      "loss": 0.0938,
      "step": 2610
    },
    {
      "epoch": 8.34,
      "grad_norm": 3.745410680770874,
      "learning_rate": 8.874734607218684e-05,
      "loss": 0.1808,
      "step": 2620
    },
    {
      "epoch": 8.38,
      "grad_norm": 7.07642936706543,
      "learning_rate": 8.832271762208068e-05,
      "loss": 0.2102,
      "step": 2630
    },
    {
      "epoch": 8.41,
      "grad_norm": 2.742769479751587,
      "learning_rate": 8.789808917197452e-05,
      "loss": 0.1217,
      "step": 2640
    },
    {
      "epoch": 8.44,
      "grad_norm": 3.350234031677246,
      "learning_rate": 8.747346072186838e-05,
      "loss": 0.1369,
      "step": 2650
    },
    {
      "epoch": 8.47,
      "grad_norm": 2.5689895153045654,
      "learning_rate": 8.704883227176222e-05,
      "loss": 0.1338,
      "step": 2660
    },
    {
      "epoch": 8.5,
      "grad_norm": 1.8818074464797974,
      "learning_rate": 8.662420382165606e-05,
      "loss": 0.1802,
      "step": 2670
    },
    {
      "epoch": 8.54,
      "grad_norm": 2.009812831878662,
      "learning_rate": 8.61995753715499e-05,
      "loss": 0.094,
      "step": 2680
    },
    {
      "epoch": 8.57,
      "grad_norm": 1.86770761013031,
      "learning_rate": 8.577494692144374e-05,
      "loss": 0.1355,
      "step": 2690
    },
    {
      "epoch": 8.6,
      "grad_norm": 2.059654474258423,
      "learning_rate": 8.535031847133759e-05,
      "loss": 0.1783,
      "step": 2700
    },
    {
      "epoch": 8.6,
      "eval_accuracy": 0.7390476190476191,
      "eval_confusion_matrix": [
        [
          53,
          5,
          17,
          5,
          0,
          3,
          2,
          4,
          7,
          4
        ],
        [
          23,
          127,
          1,
          0,
          4,
          2,
          2,
          3,
          31,
          0
        ],
        [
          0,
          1,
          56,
          2,
          1,
          2,
          0,
          0,
          0,
          3
        ],
        [
          1,
          0,
          7,
          34,
          3,
          1,
          0,
          0,
          0,
          6
        ],
        [
          1,
          2,
          3,
          6,
          75,
          1,
          1,
          0,
          2,
          2
        ],
        [
          2,
          1,
          18,
          5,
          1,
          40,
          0,
          2,
          2,
          1
        ],
        [
          2,
          0,
          1,
          0,
          1,
          0,
          54,
          0,
          0,
          1
        ],
        [
          1,
          0,
          9,
          1,
          0,
          0,
          0,
          71,
          0,
          1
        ],
        [
          6,
          27,
          4,
          0,
          0,
          1,
          12,
          0,
          130,
          0
        ],
        [
          1,
          2,
          2,
          11,
          0,
          0,
          0,
          1,
          0,
          136
        ]
      ],
      "eval_loss": 1.1571182012557983,
      "eval_runtime": 15.4706,
      "eval_samples_per_second": 67.871,
      "eval_steps_per_second": 4.266,
      "step": 2700
    },
    {
      "epoch": 8.63,
      "grad_norm": 3.0963752269744873,
      "learning_rate": 8.492569002123143e-05,
      "loss": 0.1029,
      "step": 2710
    },
    {
      "epoch": 8.66,
      "grad_norm": 0.827233076095581,
      "learning_rate": 8.450106157112527e-05,
      "loss": 0.0846,
      "step": 2720
    },
    {
      "epoch": 8.69,
      "grad_norm": 1.6526960134506226,
      "learning_rate": 8.407643312101911e-05,
      "loss": 0.0648,
      "step": 2730
    },
    {
      "epoch": 8.73,
      "grad_norm": 0.8661704063415527,
      "learning_rate": 8.365180467091295e-05,
      "loss": 0.1123,
      "step": 2740
    },
    {
      "epoch": 8.76,
      "grad_norm": 2.4067697525024414,
      "learning_rate": 8.322717622080681e-05,
      "loss": 0.1395,
      "step": 2750
    },
    {
      "epoch": 8.79,
      "grad_norm": 2.14558482170105,
      "learning_rate": 8.280254777070065e-05,
      "loss": 0.098,
      "step": 2760
    },
    {
      "epoch": 8.82,
      "grad_norm": 2.7629847526550293,
      "learning_rate": 8.237791932059449e-05,
      "loss": 0.1616,
      "step": 2770
    },
    {
      "epoch": 8.85,
      "grad_norm": 0.8335350751876831,
      "learning_rate": 8.195329087048833e-05,
      "loss": 0.1403,
      "step": 2780
    },
    {
      "epoch": 8.89,
      "grad_norm": 0.8157089352607727,
      "learning_rate": 8.152866242038217e-05,
      "loss": 0.1176,
      "step": 2790
    },
    {
      "epoch": 8.92,
      "grad_norm": 11.032809257507324,
      "learning_rate": 8.110403397027602e-05,
      "loss": 0.1733,
      "step": 2800
    },
    {
      "epoch": 8.92,
      "eval_accuracy": 0.719047619047619,
      "eval_confusion_matrix": [
        [
          51,
          5,
          13,
          8,
          0,
          4,
          4,
          5,
          8,
          2
        ],
        [
          29,
          116,
          6,
          0,
          0,
          4,
          10,
          2,
          26,
          0
        ],
        [
          1,
          0,
          49,
          1,
          1,
          8,
          2,
          0,
          0,
          3
        ],
        [
          0,
          0,
          5,
          34,
          4,
          3,
          0,
          0,
          0,
          6
        ],
        [
          1,
          3,
          4,
          2,
          76,
          4,
          2,
          0,
          1,
          0
        ],
        [
          1,
          0,
          8,
          4,
          0,
          52,
          0,
          3,
          2,
          2
        ],
        [
          0,
          0,
          1,
          0,
          0,
          0,
          58,
          0,
          0,
          0
        ],
        [
          1,
          0,
          9,
          0,
          0,
          2,
          0,
          71,
          0,
          0
        ],
        [
          3,
          26,
          5,
          0,
          0,
          4,
          19,
          3,
          118,
          2
        ],
        [
          1,
          2,
          2,
          11,
          0,
          0,
          5,
          2,
          0,
          130
        ]
      ],
      "eval_loss": 1.3043608665466309,
      "eval_runtime": 15.138,
      "eval_samples_per_second": 69.362,
      "eval_steps_per_second": 4.36,
      "step": 2800
    },
    {
      "epoch": 8.95,
      "grad_norm": 3.626187324523926,
      "learning_rate": 8.067940552016986e-05,
      "loss": 0.1916,
      "step": 2810
    },
    {
      "epoch": 8.98,
      "grad_norm": 3.0848803520202637,
      "learning_rate": 8.02547770700637e-05,
      "loss": 0.1663,
      "step": 2820
    },
    {
      "epoch": 9.01,
      "grad_norm": 1.9697245359420776,
      "learning_rate": 7.983014861995754e-05,
      "loss": 0.1136,
      "step": 2830
    },
    {
      "epoch": 9.04,
      "grad_norm": 0.600030243396759,
      "learning_rate": 7.940552016985138e-05,
      "loss": 0.1415,
      "step": 2840
    },
    {
      "epoch": 9.08,
      "grad_norm": 5.171400547027588,
      "learning_rate": 7.898089171974524e-05,
      "loss": 0.1881,
      "step": 2850
    },
    {
      "epoch": 9.11,
      "grad_norm": 2.708814859390259,
      "learning_rate": 7.855626326963908e-05,
      "loss": 0.1517,
      "step": 2860
    },
    {
      "epoch": 9.14,
      "grad_norm": 0.7085170149803162,
      "learning_rate": 7.813163481953292e-05,
      "loss": 0.0796,
      "step": 2870
    },
    {
      "epoch": 9.17,
      "grad_norm": 2.6039786338806152,
      "learning_rate": 7.770700636942676e-05,
      "loss": 0.0846,
      "step": 2880
    },
    {
      "epoch": 9.2,
      "grad_norm": 4.559110641479492,
      "learning_rate": 7.72823779193206e-05,
      "loss": 0.1696,
      "step": 2890
    },
    {
      "epoch": 9.24,
      "grad_norm": 3.545804262161255,
      "learning_rate": 7.685774946921445e-05,
      "loss": 0.1275,
      "step": 2900
    },
    {
      "epoch": 9.24,
      "eval_accuracy": 0.7266666666666667,
      "eval_confusion_matrix": [
        [
          66,
          6,
          8,
          5,
          0,
          4,
          3,
          3,
          4,
          1
        ],
        [
          53,
          111,
          0,
          0,
          4,
          1,
          4,
          3,
          17,
          0
        ],
        [
          3,
          1,
          48,
          3,
          1,
          5,
          2,
          0,
          0,
          2
        ],
        [
          5,
          1,
          3,
          27,
          5,
          2,
          0,
          2,
          1,
          6
        ],
        [
          1,
          2,
          0,
          1,
          85,
          0,
          1,
          0,
          2,
          1
        ],
        [
          5,
          0,
          5,
          6,
          0,
          50,
          2,
          3,
          1,
          0
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          57,
          0,
          0,
          0
        ],
        [
          3,
          0,
          3,
          0,
          0,
          2,
          1,
          74,
          0,
          0
        ],
        [
          13,
          34,
          0,
          1,
          0,
          1,
          11,
          1,
          119,
          0
        ],
        [
          7,
          0,
          1,
          14,
          0,
          0,
          3,
          2,
          0,
          126
        ]
      ],
      "eval_loss": 1.2415543794631958,
      "eval_runtime": 15.3407,
      "eval_samples_per_second": 68.445,
      "eval_steps_per_second": 4.302,
      "step": 2900
    },
    {
      "epoch": 9.27,
      "grad_norm": 0.5964432954788208,
      "learning_rate": 7.643312101910829e-05,
      "loss": 0.0986,
      "step": 2910
    },
    {
      "epoch": 9.3,
      "grad_norm": 1.4419245719909668,
      "learning_rate": 7.600849256900213e-05,
      "loss": 0.0997,
      "step": 2920
    },
    {
      "epoch": 9.33,
      "grad_norm": 1.5141104459762573,
      "learning_rate": 7.558386411889597e-05,
      "loss": 0.1625,
      "step": 2930
    },
    {
      "epoch": 9.36,
      "grad_norm": 0.5192936658859253,
      "learning_rate": 7.515923566878981e-05,
      "loss": 0.1522,
      "step": 2940
    },
    {
      "epoch": 9.39,
      "grad_norm": 2.7084801197052,
      "learning_rate": 7.473460721868365e-05,
      "loss": 0.1412,
      "step": 2950
    },
    {
      "epoch": 9.43,
      "grad_norm": 3.396594762802124,
      "learning_rate": 7.43099787685775e-05,
      "loss": 0.123,
      "step": 2960
    },
    {
      "epoch": 9.46,
      "grad_norm": 0.13223083317279816,
      "learning_rate": 7.388535031847134e-05,
      "loss": 0.1273,
      "step": 2970
    },
    {
      "epoch": 9.49,
      "grad_norm": 1.5881577730178833,
      "learning_rate": 7.346072186836518e-05,
      "loss": 0.1309,
      "step": 2980
    },
    {
      "epoch": 9.52,
      "grad_norm": 2.229868173599243,
      "learning_rate": 7.303609341825902e-05,
      "loss": 0.1302,
      "step": 2990
    },
    {
      "epoch": 9.55,
      "grad_norm": 4.873483657836914,
      "learning_rate": 7.261146496815286e-05,
      "loss": 0.1231,
      "step": 3000
    },
    {
      "epoch": 9.55,
      "eval_accuracy": 0.7123809523809523,
      "eval_confusion_matrix": [
        [
          73,
          3,
          7,
          5,
          0,
          1,
          4,
          2,
          4,
          1
        ],
        [
          84,
          81,
          0,
          0,
          3,
          1,
          3,
          1,
          20,
          0
        ],
        [
          2,
          1,
          51,
          2,
          1,
          5,
          0,
          0,
          0,
          3
        ],
        [
          5,
          0,
          3,
          28,
          6,
          1,
          0,
          0,
          1,
          8
        ],
        [
          1,
          1,
          0,
          1,
          86,
          0,
          1,
          0,
          2,
          1
        ],
        [
          9,
          0,
          6,
          4,
          1,
          46,
          1,
          3,
          2,
          0
        ],
        [
          2,
          0,
          1,
          0,
          1,
          0,
          54,
          0,
          0,
          1
        ],
        [
          10,
          0,
          1,
          0,
          0,
          0,
          1,
          71,
          0,
          0
        ],
        [
          21,
          23,
          1,
          0,
          0,
          2,
          12,
          2,
          119,
          0
        ],
        [
          7,
          0,
          1,
          4,
          0,
          0,
          0,
          2,
          0,
          139
        ]
      ],
      "eval_loss": 1.4283809661865234,
      "eval_runtime": 15.3421,
      "eval_samples_per_second": 68.439,
      "eval_steps_per_second": 4.302,
      "step": 3000
    },
    {
      "epoch": 9.59,
      "grad_norm": 1.5915924310684204,
      "learning_rate": 7.21868365180467e-05,
      "loss": 0.1309,
      "step": 3010
    },
    {
      "epoch": 9.62,
      "grad_norm": 0.040684912353754044,
      "learning_rate": 7.176220806794054e-05,
      "loss": 0.0861,
      "step": 3020
    },
    {
      "epoch": 9.65,
      "grad_norm": 2.1614465713500977,
      "learning_rate": 7.13375796178344e-05,
      "loss": 0.1997,
      "step": 3030
    },
    {
      "epoch": 9.68,
      "grad_norm": 2.8283462524414062,
      "learning_rate": 7.091295116772824e-05,
      "loss": 0.2087,
      "step": 3040
    },
    {
      "epoch": 9.71,
      "grad_norm": 2.3698461055755615,
      "learning_rate": 7.048832271762208e-05,
      "loss": 0.2046,
      "step": 3050
    },
    {
      "epoch": 9.75,
      "grad_norm": 0.0435897558927536,
      "learning_rate": 7.006369426751592e-05,
      "loss": 0.1335,
      "step": 3060
    },
    {
      "epoch": 9.78,
      "grad_norm": 0.3226892948150635,
      "learning_rate": 6.963906581740977e-05,
      "loss": 0.1027,
      "step": 3070
    },
    {
      "epoch": 9.81,
      "grad_norm": 3.1752257347106934,
      "learning_rate": 6.921443736730361e-05,
      "loss": 0.156,
      "step": 3080
    },
    {
      "epoch": 9.84,
      "grad_norm": 1.864380121231079,
      "learning_rate": 6.878980891719745e-05,
      "loss": 0.1441,
      "step": 3090
    },
    {
      "epoch": 9.87,
      "grad_norm": 2.183727502822876,
      "learning_rate": 6.836518046709129e-05,
      "loss": 0.1828,
      "step": 3100
    },
    {
      "epoch": 9.87,
      "eval_accuracy": 0.7523809523809524,
      "eval_confusion_matrix": [
        [
          66,
          2,
          13,
          7,
          0,
          0,
          2,
          1,
          7,
          2
        ],
        [
          38,
          115,
          1,
          0,
          4,
          0,
          4,
          2,
          28,
          1
        ],
        [
          1,
          0,
          52,
          2,
          1,
          4,
          2,
          0,
          0,
          3
        ],
        [
          3,
          0,
          4,
          35,
          5,
          0,
          0,
          0,
          1,
          4
        ],
        [
          0,
          1,
          1,
          5,
          83,
          0,
          1,
          0,
          1,
          1
        ],
        [
          4,
          1,
          12,
          6,
          3,
          41,
          1,
          2,
          2,
          0
        ],
        [
          0,
          0,
          1,
          0,
          0,
          0,
          58,
          0,
          0,
          0
        ],
        [
          5,
          0,
          5,
          0,
          0,
          0,
          1,
          72,
          0,
          0
        ],
        [
          11,
          24,
          1,
          0,
          0,
          0,
          10,
          1,
          132,
          1
        ],
        [
          3,
          0,
          2,
          9,
          0,
          0,
          1,
          2,
          0,
          136
        ]
      ],
      "eval_loss": 1.2049111127853394,
      "eval_runtime": 15.2555,
      "eval_samples_per_second": 68.828,
      "eval_steps_per_second": 4.326,
      "step": 3100
    },
    {
      "epoch": 9.9,
      "grad_norm": 4.488446235656738,
      "learning_rate": 6.794055201698513e-05,
      "loss": 0.1018,
      "step": 3110
    },
    {
      "epoch": 9.94,
      "grad_norm": 3.176164388656616,
      "learning_rate": 6.751592356687899e-05,
      "loss": 0.1049,
      "step": 3120
    },
    {
      "epoch": 9.97,
      "grad_norm": 4.369916915893555,
      "learning_rate": 6.709129511677283e-05,
      "loss": 0.1179,
      "step": 3130
    },
    {
      "epoch": 10.0,
      "grad_norm": 2.099869728088379,
      "learning_rate": 6.666666666666667e-05,
      "loss": 0.1081,
      "step": 3140
    },
    {
      "epoch": 10.03,
      "grad_norm": 4.553194522857666,
      "learning_rate": 6.624203821656051e-05,
      "loss": 0.1152,
      "step": 3150
    },
    {
      "epoch": 10.06,
      "grad_norm": 0.31203553080558777,
      "learning_rate": 6.581740976645435e-05,
      "loss": 0.0666,
      "step": 3160
    },
    {
      "epoch": 10.1,
      "grad_norm": 1.5009262561798096,
      "learning_rate": 6.53927813163482e-05,
      "loss": 0.1208,
      "step": 3170
    },
    {
      "epoch": 10.13,
      "grad_norm": 0.2197587788105011,
      "learning_rate": 6.496815286624204e-05,
      "loss": 0.0826,
      "step": 3180
    },
    {
      "epoch": 10.16,
      "grad_norm": 0.05214032158255577,
      "learning_rate": 6.454352441613588e-05,
      "loss": 0.0983,
      "step": 3190
    },
    {
      "epoch": 10.19,
      "grad_norm": 3.6048381328582764,
      "learning_rate": 6.411889596602972e-05,
      "loss": 0.083,
      "step": 3200
    },
    {
      "epoch": 10.19,
      "eval_accuracy": 0.7238095238095238,
      "eval_confusion_matrix": [
        [
          57,
          5,
          16,
          5,
          1,
          1,
          7,
          2,
          3,
          3
        ],
        [
          30,
          127,
          0,
          0,
          1,
          2,
          11,
          3,
          18,
          1
        ],
        [
          0,
          0,
          52,
          3,
          1,
          5,
          2,
          0,
          0,
          2
        ],
        [
          4,
          0,
          5,
          30,
          5,
          1,
          0,
          0,
          0,
          7
        ],
        [
          1,
          1,
          0,
          4,
          84,
          0,
          1,
          0,
          1,
          1
        ],
        [
          3,
          1,
          9,
          4,
          1,
          48,
          2,
          1,
          2,
          1
        ],
        [
          0,
          0,
          1,
          0,
          0,
          0,
          57,
          0,
          0,
          1
        ],
        [
          4,
          0,
          7,
          0,
          0,
          0,
          3,
          69,
          0,
          0
        ],
        [
          9,
          27,
          1,
          0,
          0,
          1,
          32,
          0,
          109,
          1
        ],
        [
          2,
          1,
          2,
          16,
          0,
          0,
          3,
          2,
          0,
          127
        ]
      ],
      "eval_loss": 1.2484331130981445,
      "eval_runtime": 15.4412,
      "eval_samples_per_second": 68.0,
      "eval_steps_per_second": 4.274,
      "step": 3200
    },
    {
      "epoch": 10.22,
      "grad_norm": 2.7771308422088623,
      "learning_rate": 6.369426751592356e-05,
      "loss": 0.1365,
      "step": 3210
    },
    {
      "epoch": 10.25,
      "grad_norm": 0.18859456479549408,
      "learning_rate": 6.326963906581742e-05,
      "loss": 0.0684,
      "step": 3220
    },
    {
      "epoch": 10.29,
      "grad_norm": 1.620737910270691,
      "learning_rate": 6.284501061571126e-05,
      "loss": 0.1069,
      "step": 3230
    },
    {
      "epoch": 10.32,
      "grad_norm": 0.8589503765106201,
      "learning_rate": 6.24203821656051e-05,
      "loss": 0.172,
      "step": 3240
    },
    {
      "epoch": 10.35,
      "grad_norm": 1.9255318641662598,
      "learning_rate": 6.199575371549894e-05,
      "loss": 0.0901,
      "step": 3250
    },
    {
      "epoch": 10.38,
      "grad_norm": 3.53298282623291,
      "learning_rate": 6.157112526539278e-05,
      "loss": 0.0516,
      "step": 3260
    },
    {
      "epoch": 10.41,
      "grad_norm": 1.6446423530578613,
      "learning_rate": 6.114649681528662e-05,
      "loss": 0.15,
      "step": 3270
    },
    {
      "epoch": 10.45,
      "grad_norm": 2.324852466583252,
      "learning_rate": 6.0721868365180465e-05,
      "loss": 0.1349,
      "step": 3280
    },
    {
      "epoch": 10.48,
      "grad_norm": 1.3299524784088135,
      "learning_rate": 6.0297239915074314e-05,
      "loss": 0.0912,
      "step": 3290
    },
    {
      "epoch": 10.51,
      "grad_norm": 0.16438178718090057,
      "learning_rate": 5.9872611464968155e-05,
      "loss": 0.1256,
      "step": 3300
    },
    {
      "epoch": 10.51,
      "eval_accuracy": 0.7228571428571429,
      "eval_confusion_matrix": [
        [
          64,
          4,
          8,
          4,
          1,
          7,
          5,
          2,
          2,
          3
        ],
        [
          43,
          119,
          0,
          0,
          2,
          1,
          10,
          3,
          14,
          1
        ],
        [
          0,
          0,
          49,
          3,
          1,
          7,
          2,
          0,
          0,
          3
        ],
        [
          4,
          0,
          8,
          27,
          5,
          1,
          0,
          0,
          0,
          7
        ],
        [
          2,
          1,
          2,
          2,
          81,
          3,
          1,
          0,
          1,
          0
        ],
        [
          2,
          2,
          10,
          3,
          0,
          50,
          1,
          0,
          2,
          2
        ],
        [
          0,
          1,
          1,
          0,
          1,
          0,
          55,
          0,
          0,
          1
        ],
        [
          2,
          0,
          3,
          0,
          0,
          0,
          1,
          77,
          0,
          0
        ],
        [
          11,
          37,
          0,
          0,
          0,
          3,
          25,
          2,
          102,
          0
        ],
        [
          1,
          1,
          2,
          11,
          0,
          0,
          1,
          2,
          0,
          135
        ]
      ],
      "eval_loss": 1.2746260166168213,
      "eval_runtime": 15.3791,
      "eval_samples_per_second": 68.274,
      "eval_steps_per_second": 4.292,
      "step": 3300
    },
    {
      "epoch": 10.54,
      "grad_norm": 1.9269123077392578,
      "learning_rate": 5.9447983014862e-05,
      "loss": 0.0969,
      "step": 3310
    },
    {
      "epoch": 10.57,
      "grad_norm": 1.9249624013900757,
      "learning_rate": 5.902335456475584e-05,
      "loss": 0.0927,
      "step": 3320
    },
    {
      "epoch": 10.61,
      "grad_norm": 1.8576029539108276,
      "learning_rate": 5.859872611464968e-05,
      "loss": 0.0572,
      "step": 3330
    },
    {
      "epoch": 10.64,
      "grad_norm": 3.594461441040039,
      "learning_rate": 5.817409766454353e-05,
      "loss": 0.0982,
      "step": 3340
    },
    {
      "epoch": 10.67,
      "grad_norm": 1.7123650312423706,
      "learning_rate": 5.774946921443737e-05,
      "loss": 0.083,
      "step": 3350
    },
    {
      "epoch": 10.7,
      "grad_norm": 0.2372170090675354,
      "learning_rate": 5.732484076433121e-05,
      "loss": 0.0989,
      "step": 3360
    },
    {
      "epoch": 10.73,
      "grad_norm": 4.383459091186523,
      "learning_rate": 5.690021231422505e-05,
      "loss": 0.1262,
      "step": 3370
    },
    {
      "epoch": 10.76,
      "grad_norm": 2.293396472930908,
      "learning_rate": 5.6475583864118895e-05,
      "loss": 0.1413,
      "step": 3380
    },
    {
      "epoch": 10.8,
      "grad_norm": 3.0922889709472656,
      "learning_rate": 5.605095541401274e-05,
      "loss": 0.1755,
      "step": 3390
    },
    {
      "epoch": 10.83,
      "grad_norm": 3.0807735919952393,
      "learning_rate": 5.5626326963906585e-05,
      "loss": 0.1067,
      "step": 3400
    },
    {
      "epoch": 10.83,
      "eval_accuracy": 0.7380952380952381,
      "eval_confusion_matrix": [
        [
          55,
          3,
          11,
          9,
          1,
          2,
          7,
          2,
          6,
          4
        ],
        [
          35,
          122,
          0,
          0,
          1,
          1,
          6,
          2,
          25,
          1
        ],
        [
          2,
          1,
          50,
          2,
          1,
          3,
          4,
          0,
          0,
          2
        ],
        [
          2,
          0,
          4,
          37,
          4,
          0,
          0,
          0,
          0,
          5
        ],
        [
          0,
          1,
          1,
          4,
          82,
          2,
          1,
          0,
          1,
          1
        ],
        [
          2,
          1,
          13,
          6,
          0,
          44,
          1,
          0,
          1,
          4
        ],
        [
          0,
          1,
          1,
          0,
          0,
          0,
          56,
          0,
          0,
          1
        ],
        [
          1,
          0,
          3,
          1,
          0,
          0,
          1,
          76,
          0,
          1
        ],
        [
          9,
          36,
          0,
          0,
          0,
          1,
          20,
          2,
          112,
          0
        ],
        [
          0,
          0,
          2,
          10,
          0,
          0,
          0,
          0,
          0,
          141
        ]
      ],
      "eval_loss": 1.190523624420166,
      "eval_runtime": 15.9133,
      "eval_samples_per_second": 65.982,
      "eval_steps_per_second": 4.147,
      "step": 3400
    },
    {
      "epoch": 10.86,
      "grad_norm": 6.171202182769775,
      "learning_rate": 5.5201698513800426e-05,
      "loss": 0.1262,
      "step": 3410
    },
    {
      "epoch": 10.89,
      "grad_norm": 0.7859517931938171,
      "learning_rate": 5.477707006369427e-05,
      "loss": 0.1045,
      "step": 3420
    },
    {
      "epoch": 10.92,
      "grad_norm": 2.1993987560272217,
      "learning_rate": 5.435244161358811e-05,
      "loss": 0.1104,
      "step": 3430
    },
    {
      "epoch": 10.96,
      "grad_norm": 1.4276065826416016,
      "learning_rate": 5.392781316348196e-05,
      "loss": 0.098,
      "step": 3440
    },
    {
      "epoch": 10.99,
      "grad_norm": 3.335463523864746,
      "learning_rate": 5.35031847133758e-05,
      "loss": 0.0561,
      "step": 3450
    },
    {
      "epoch": 11.02,
      "grad_norm": 1.2473598718643188,
      "learning_rate": 5.307855626326964e-05,
      "loss": 0.0946,
      "step": 3460
    },
    {
      "epoch": 11.05,
      "grad_norm": 2.4468166828155518,
      "learning_rate": 5.265392781316348e-05,
      "loss": 0.1373,
      "step": 3470
    },
    {
      "epoch": 11.08,
      "grad_norm": 4.396722793579102,
      "learning_rate": 5.2229299363057324e-05,
      "loss": 0.0794,
      "step": 3480
    },
    {
      "epoch": 11.11,
      "grad_norm": 1.3597296476364136,
      "learning_rate": 5.180467091295117e-05,
      "loss": 0.1506,
      "step": 3490
    },
    {
      "epoch": 11.15,
      "grad_norm": 1.7938604354858398,
      "learning_rate": 5.1380042462845014e-05,
      "loss": 0.092,
      "step": 3500
    },
    {
      "epoch": 11.15,
      "eval_accuracy": 0.7476190476190476,
      "eval_confusion_matrix": [
        [
          65,
          3,
          8,
          2,
          0,
          4,
          5,
          2,
          9,
          2
        ],
        [
          27,
          108,
          0,
          0,
          1,
          1,
          8,
          2,
          46,
          0
        ],
        [
          4,
          0,
          49,
          2,
          1,
          4,
          3,
          0,
          0,
          2
        ],
        [
          2,
          0,
          3,
          37,
          4,
          0,
          0,
          0,
          1,
          5
        ],
        [
          0,
          1,
          1,
          3,
          83,
          2,
          1,
          0,
          1,
          1
        ],
        [
          6,
          2,
          8,
          6,
          0,
          45,
          1,
          0,
          1,
          3
        ],
        [
          0,
          0,
          1,
          0,
          0,
          0,
          56,
          0,
          1,
          1
        ],
        [
          3,
          0,
          2,
          1,
          0,
          1,
          3,
          72,
          0,
          1
        ],
        [
          10,
          22,
          0,
          0,
          0,
          1,
          12,
          1,
          134,
          0
        ],
        [
          1,
          0,
          1,
          12,
          0,
          1,
          2,
          0,
          0,
          136
        ]
      ],
      "eval_loss": 1.1174523830413818,
      "eval_runtime": 15.7765,
      "eval_samples_per_second": 66.554,
      "eval_steps_per_second": 4.183,
      "step": 3500
    },
    {
      "epoch": 11.18,
      "grad_norm": 0.015160035341978073,
      "learning_rate": 5.0955414012738855e-05,
      "loss": 0.08,
      "step": 3510
    },
    {
      "epoch": 11.21,
      "grad_norm": 2.8019070625305176,
      "learning_rate": 5.05307855626327e-05,
      "loss": 0.1266,
      "step": 3520
    },
    {
      "epoch": 11.24,
      "grad_norm": 0.25064635276794434,
      "learning_rate": 5.010615711252654e-05,
      "loss": 0.1271,
      "step": 3530
    },
    {
      "epoch": 11.27,
      "grad_norm": 2.228379726409912,
      "learning_rate": 4.968152866242039e-05,
      "loss": 0.0815,
      "step": 3540
    },
    {
      "epoch": 11.31,
      "grad_norm": 1.0253998041152954,
      "learning_rate": 4.925690021231423e-05,
      "loss": 0.0373,
      "step": 3550
    },
    {
      "epoch": 11.34,
      "grad_norm": 2.2330989837646484,
      "learning_rate": 4.883227176220807e-05,
      "loss": 0.1606,
      "step": 3560
    },
    {
      "epoch": 11.37,
      "grad_norm": 2.266019821166992,
      "learning_rate": 4.840764331210191e-05,
      "loss": 0.0989,
      "step": 3570
    },
    {
      "epoch": 11.4,
      "grad_norm": 0.6040272116661072,
      "learning_rate": 4.798301486199575e-05,
      "loss": 0.0714,
      "step": 3580
    },
    {
      "epoch": 11.43,
      "grad_norm": 2.601849317550659,
      "learning_rate": 4.75583864118896e-05,
      "loss": 0.1037,
      "step": 3590
    },
    {
      "epoch": 11.46,
      "grad_norm": 2.4073190689086914,
      "learning_rate": 4.713375796178344e-05,
      "loss": 0.153,
      "step": 3600
    },
    {
      "epoch": 11.46,
      "eval_accuracy": 0.7361904761904762,
      "eval_confusion_matrix": [
        [
          75,
          4,
          9,
          1,
          0,
          1,
          2,
          3,
          4,
          1
        ],
        [
          51,
          111,
          0,
          1,
          1,
          1,
          1,
          3,
          24,
          0
        ],
        [
          2,
          1,
          52,
          2,
          1,
          3,
          0,
          2,
          0,
          2
        ],
        [
          4,
          1,
          6,
          30,
          4,
          0,
          0,
          1,
          0,
          6
        ],
        [
          1,
          4,
          1,
          4,
          80,
          0,
          1,
          1,
          0,
          1
        ],
        [
          6,
          4,
          8,
          6,
          1,
          39,
          0,
          1,
          2,
          5
        ],
        [
          2,
          1,
          1,
          1,
          1,
          0,
          52,
          0,
          0,
          1
        ],
        [
          2,
          0,
          1,
          0,
          0,
          0,
          0,
          80,
          0,
          0
        ],
        [
          15,
          33,
          0,
          0,
          1,
          1,
          11,
          3,
          114,
          2
        ],
        [
          1,
          0,
          1,
          7,
          2,
          0,
          1,
          1,
          0,
          140
        ]
      ],
      "eval_loss": 1.243424654006958,
      "eval_runtime": 15.7389,
      "eval_samples_per_second": 66.714,
      "eval_steps_per_second": 4.193,
      "step": 3600
    },
    {
      "epoch": 11.5,
      "grad_norm": 3.7230005264282227,
      "learning_rate": 4.6709129511677285e-05,
      "loss": 0.0819,
      "step": 3610
    },
    {
      "epoch": 11.53,
      "grad_norm": 2.161355495452881,
      "learning_rate": 4.6284501061571126e-05,
      "loss": 0.0946,
      "step": 3620
    },
    {
      "epoch": 11.56,
      "grad_norm": 0.5289556384086609,
      "learning_rate": 4.585987261146497e-05,
      "loss": 0.0634,
      "step": 3630
    },
    {
      "epoch": 11.59,
      "grad_norm": 0.37803375720977783,
      "learning_rate": 4.5435244161358816e-05,
      "loss": 0.0931,
      "step": 3640
    },
    {
      "epoch": 11.62,
      "grad_norm": 1.356149673461914,
      "learning_rate": 4.501061571125266e-05,
      "loss": 0.0854,
      "step": 3650
    },
    {
      "epoch": 11.66,
      "grad_norm": 1.575438380241394,
      "learning_rate": 4.45859872611465e-05,
      "loss": 0.08,
      "step": 3660
    },
    {
      "epoch": 11.69,
      "grad_norm": 1.5214145183563232,
      "learning_rate": 4.416135881104034e-05,
      "loss": 0.0578,
      "step": 3670
    },
    {
      "epoch": 11.72,
      "grad_norm": 0.6308757662773132,
      "learning_rate": 4.373673036093419e-05,
      "loss": 0.0849,
      "step": 3680
    },
    {
      "epoch": 11.75,
      "grad_norm": 2.426156520843506,
      "learning_rate": 4.331210191082803e-05,
      "loss": 0.1025,
      "step": 3690
    },
    {
      "epoch": 11.78,
      "grad_norm": 2.7888565063476562,
      "learning_rate": 4.288747346072187e-05,
      "loss": 0.1065,
      "step": 3700
    },
    {
      "epoch": 11.78,
      "eval_accuracy": 0.7371428571428571,
      "eval_confusion_matrix": [
        [
          69,
          2,
          10,
          2,
          0,
          2,
          6,
          3,
          5,
          1
        ],
        [
          44,
          109,
          2,
          0,
          0,
          2,
          5,
          3,
          28,
          0
        ],
        [
          2,
          1,
          50,
          2,
          1,
          4,
          2,
          1,
          0,
          2
        ],
        [
          4,
          0,
          4,
          32,
          4,
          2,
          0,
          1,
          1,
          4
        ],
        [
          1,
          2,
          2,
          6,
          76,
          2,
          1,
          1,
          2,
          0
        ],
        [
          2,
          1,
          11,
          4,
          0,
          47,
          2,
          1,
          2,
          2
        ],
        [
          0,
          0,
          1,
          0,
          0,
          0,
          58,
          0,
          0,
          0
        ],
        [
          2,
          0,
          4,
          0,
          0,
          0,
          1,
          76,
          0,
          0
        ],
        [
          9,
          27,
          0,
          0,
          0,
          1,
          13,
          1,
          128,
          1
        ],
        [
          2,
          2,
          2,
          13,
          0,
          0,
          4,
          1,
          0,
          129
        ]
      ],
      "eval_loss": 1.232663631439209,
      "eval_runtime": 15.4779,
      "eval_samples_per_second": 67.839,
      "eval_steps_per_second": 4.264,
      "step": 3700
    },
    {
      "epoch": 11.82,
      "grad_norm": 0.030465861782431602,
      "learning_rate": 4.2462845010615714e-05,
      "loss": 0.0873,
      "step": 3710
    },
    {
      "epoch": 11.85,
      "grad_norm": 1.4719470739364624,
      "learning_rate": 4.2038216560509556e-05,
      "loss": 0.1074,
      "step": 3720
    },
    {
      "epoch": 11.88,
      "grad_norm": 0.23886482417583466,
      "learning_rate": 4.1613588110403404e-05,
      "loss": 0.0592,
      "step": 3730
    },
    {
      "epoch": 11.91,
      "grad_norm": 2.1231284141540527,
      "learning_rate": 4.1188959660297246e-05,
      "loss": 0.0642,
      "step": 3740
    },
    {
      "epoch": 11.94,
      "grad_norm": 0.41407281160354614,
      "learning_rate": 4.076433121019109e-05,
      "loss": 0.1265,
      "step": 3750
    },
    {
      "epoch": 11.97,
      "grad_norm": 3.6939876079559326,
      "learning_rate": 4.033970276008493e-05,
      "loss": 0.1025,
      "step": 3760
    },
    {
      "epoch": 12.01,
      "grad_norm": 0.5653472542762756,
      "learning_rate": 3.991507430997877e-05,
      "loss": 0.111,
      "step": 3770
    },
    {
      "epoch": 12.04,
      "grad_norm": 1.5738434791564941,
      "learning_rate": 3.949044585987262e-05,
      "loss": 0.1009,
      "step": 3780
    },
    {
      "epoch": 12.07,
      "grad_norm": 0.18574613332748413,
      "learning_rate": 3.906581740976646e-05,
      "loss": 0.0578,
      "step": 3790
    },
    {
      "epoch": 12.1,
      "grad_norm": 2.1222317218780518,
      "learning_rate": 3.86411889596603e-05,
      "loss": 0.0875,
      "step": 3800
    },
    {
      "epoch": 12.1,
      "eval_accuracy": 0.7457142857142857,
      "eval_confusion_matrix": [
        [
          67,
          3,
          10,
          5,
          0,
          3,
          1,
          3,
          6,
          2
        ],
        [
          40,
          110,
          0,
          0,
          1,
          1,
          4,
          3,
          34,
          0
        ],
        [
          1,
          1,
          51,
          2,
          1,
          3,
          1,
          2,
          0,
          3
        ],
        [
          3,
          0,
          4,
          35,
          4,
          1,
          0,
          1,
          0,
          4
        ],
        [
          0,
          2,
          1,
          5,
          78,
          3,
          1,
          1,
          1,
          1
        ],
        [
          1,
          2,
          12,
          4,
          0,
          45,
          2,
          4,
          1,
          1
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          56,
          0,
          0,
          1
        ],
        [
          2,
          0,
          2,
          0,
          0,
          0,
          1,
          78,
          0,
          0
        ],
        [
          9,
          26,
          0,
          0,
          1,
          1,
          12,
          1,
          129,
          1
        ],
        [
          0,
          0,
          2,
          13,
          0,
          0,
          3,
          1,
          0,
          134
        ]
      ],
      "eval_loss": 1.2356998920440674,
      "eval_runtime": 15.3228,
      "eval_samples_per_second": 68.525,
      "eval_steps_per_second": 4.307,
      "step": 3800
    },
    {
      "epoch": 12.13,
      "grad_norm": 6.8635478019714355,
      "learning_rate": 3.821656050955414e-05,
      "loss": 0.0778,
      "step": 3810
    },
    {
      "epoch": 12.17,
      "grad_norm": 3.9780235290527344,
      "learning_rate": 3.7791932059447985e-05,
      "loss": 0.103,
      "step": 3820
    },
    {
      "epoch": 12.2,
      "grad_norm": 1.2772787809371948,
      "learning_rate": 3.7367303609341826e-05,
      "loss": 0.0413,
      "step": 3830
    },
    {
      "epoch": 12.23,
      "grad_norm": 2.644360065460205,
      "learning_rate": 3.694267515923567e-05,
      "loss": 0.0451,
      "step": 3840
    },
    {
      "epoch": 12.26,
      "grad_norm": 3.3227038383483887,
      "learning_rate": 3.651804670912951e-05,
      "loss": 0.0924,
      "step": 3850
    },
    {
      "epoch": 12.29,
      "grad_norm": 0.4245937764644623,
      "learning_rate": 3.609341825902335e-05,
      "loss": 0.0422,
      "step": 3860
    },
    {
      "epoch": 12.32,
      "grad_norm": 1.7972066402435303,
      "learning_rate": 3.56687898089172e-05,
      "loss": 0.0574,
      "step": 3870
    },
    {
      "epoch": 12.36,
      "grad_norm": 4.524425983428955,
      "learning_rate": 3.524416135881104e-05,
      "loss": 0.0768,
      "step": 3880
    },
    {
      "epoch": 12.39,
      "grad_norm": 3.626488447189331,
      "learning_rate": 3.481953290870488e-05,
      "loss": 0.081,
      "step": 3890
    },
    {
      "epoch": 12.42,
      "grad_norm": 1.6742682456970215,
      "learning_rate": 3.4394904458598724e-05,
      "loss": 0.0714,
      "step": 3900
    },
    {
      "epoch": 12.42,
      "eval_accuracy": 0.7304761904761905,
      "eval_confusion_matrix": [
        [
          77,
          3,
          7,
          3,
          0,
          1,
          2,
          2,
          4,
          1
        ],
        [
          58,
          103,
          0,
          0,
          0,
          1,
          4,
          1,
          26,
          0
        ],
        [
          4,
          1,
          51,
          2,
          1,
          3,
          1,
          0,
          0,
          2
        ],
        [
          4,
          0,
          4,
          33,
          6,
          0,
          0,
          0,
          0,
          5
        ],
        [
          3,
          1,
          4,
          3,
          77,
          2,
          1,
          0,
          1,
          1
        ],
        [
          5,
          2,
          14,
          6,
          0,
          44,
          0,
          0,
          0,
          1
        ],
        [
          2,
          1,
          1,
          0,
          1,
          0,
          53,
          0,
          0,
          1
        ],
        [
          7,
          0,
          7,
          0,
          0,
          0,
          0,
          69,
          0,
          0
        ],
        [
          14,
          28,
          0,
          0,
          0,
          1,
          12,
          1,
          124,
          0
        ],
        [
          2,
          0,
          1,
          11,
          0,
          3,
          0,
          0,
          0,
          136
        ]
      ],
      "eval_loss": 1.2995976209640503,
      "eval_runtime": 15.7291,
      "eval_samples_per_second": 66.755,
      "eval_steps_per_second": 4.196,
      "step": 3900
    },
    {
      "epoch": 12.45,
      "grad_norm": 0.18191276490688324,
      "learning_rate": 3.3970276008492566e-05,
      "loss": 0.1241,
      "step": 3910
    },
    {
      "epoch": 12.48,
      "grad_norm": 1.7220439910888672,
      "learning_rate": 3.3545647558386414e-05,
      "loss": 0.117,
      "step": 3920
    },
    {
      "epoch": 12.52,
      "grad_norm": 2.8817391395568848,
      "learning_rate": 3.3121019108280256e-05,
      "loss": 0.0711,
      "step": 3930
    },
    {
      "epoch": 12.55,
      "grad_norm": 2.5080385208129883,
      "learning_rate": 3.26963906581741e-05,
      "loss": 0.1241,
      "step": 3940
    },
    {
      "epoch": 12.58,
      "grad_norm": 1.9212385416030884,
      "learning_rate": 3.227176220806794e-05,
      "loss": 0.057,
      "step": 3950
    },
    {
      "epoch": 12.61,
      "grad_norm": 0.9736161231994629,
      "learning_rate": 3.184713375796178e-05,
      "loss": 0.0674,
      "step": 3960
    },
    {
      "epoch": 12.64,
      "grad_norm": 3.3597800731658936,
      "learning_rate": 3.142250530785563e-05,
      "loss": 0.0848,
      "step": 3970
    },
    {
      "epoch": 12.68,
      "grad_norm": 0.008619729429483414,
      "learning_rate": 3.099787685774947e-05,
      "loss": 0.0811,
      "step": 3980
    },
    {
      "epoch": 12.71,
      "grad_norm": 1.4590424299240112,
      "learning_rate": 3.057324840764331e-05,
      "loss": 0.0564,
      "step": 3990
    },
    {
      "epoch": 12.74,
      "grad_norm": 0.1266852766275406,
      "learning_rate": 3.0148619957537157e-05,
      "loss": 0.1433,
      "step": 4000
    },
    {
      "epoch": 12.74,
      "eval_accuracy": 0.7409523809523809,
      "eval_confusion_matrix": [
        [
          74,
          4,
          9,
          3,
          0,
          2,
          1,
          2,
          4,
          1
        ],
        [
          42,
          114,
          0,
          0,
          1,
          1,
          3,
          3,
          29,
          0
        ],
        [
          4,
          0,
          49,
          2,
          1,
          5,
          2,
          0,
          0,
          2
        ],
        [
          4,
          0,
          4,
          31,
          6,
          2,
          0,
          0,
          0,
          5
        ],
        [
          0,
          1,
          2,
          2,
          86,
          0,
          1,
          0,
          0,
          1
        ],
        [
          7,
          2,
          9,
          6,
          0,
          46,
          1,
          0,
          0,
          1
        ],
        [
          2,
          1,
          1,
          0,
          1,
          0,
          53,
          0,
          0,
          1
        ],
        [
          3,
          0,
          5,
          0,
          0,
          0,
          0,
          75,
          0,
          0
        ],
        [
          9,
          34,
          0,
          0,
          1,
          2,
          11,
          1,
          120,
          2
        ],
        [
          3,
          0,
          1,
          16,
          0,
          1,
          2,
          0,
          0,
          130
        ]
      ],
      "eval_loss": 1.2166801691055298,
      "eval_runtime": 15.2638,
      "eval_samples_per_second": 68.79,
      "eval_steps_per_second": 4.324,
      "step": 4000
    },
    {
      "epoch": 12.77,
      "grad_norm": 0.010001400485634804,
      "learning_rate": 2.9723991507431e-05,
      "loss": 0.1069,
      "step": 4010
    },
    {
      "epoch": 12.8,
      "grad_norm": 0.3851306438446045,
      "learning_rate": 2.929936305732484e-05,
      "loss": 0.0686,
      "step": 4020
    },
    {
      "epoch": 12.83,
      "grad_norm": 2.4934566020965576,
      "learning_rate": 2.8874734607218685e-05,
      "loss": 0.1435,
      "step": 4030
    },
    {
      "epoch": 12.87,
      "grad_norm": 1.281803846359253,
      "learning_rate": 2.8450106157112527e-05,
      "loss": 0.0722,
      "step": 4040
    },
    {
      "epoch": 12.9,
      "grad_norm": 0.7258008122444153,
      "learning_rate": 2.802547770700637e-05,
      "loss": 0.1327,
      "step": 4050
    },
    {
      "epoch": 12.93,
      "grad_norm": 0.012724624946713448,
      "learning_rate": 2.7600849256900213e-05,
      "loss": 0.0961,
      "step": 4060
    },
    {
      "epoch": 12.96,
      "grad_norm": 3.1563878059387207,
      "learning_rate": 2.7176220806794055e-05,
      "loss": 0.0766,
      "step": 4070
    },
    {
      "epoch": 12.99,
      "grad_norm": 2.661426544189453,
      "learning_rate": 2.67515923566879e-05,
      "loss": 0.0778,
      "step": 4080
    },
    {
      "epoch": 13.03,
      "grad_norm": 4.320444107055664,
      "learning_rate": 2.632696390658174e-05,
      "loss": 0.0692,
      "step": 4090
    },
    {
      "epoch": 13.06,
      "grad_norm": 2.4924447536468506,
      "learning_rate": 2.5902335456475586e-05,
      "loss": 0.0765,
      "step": 4100
    },
    {
      "epoch": 13.06,
      "eval_accuracy": 0.7380952380952381,
      "eval_confusion_matrix": [
        [
          69,
          4,
          10,
          3,
          0,
          4,
          4,
          2,
          3,
          1
        ],
        [
          48,
          105,
          0,
          0,
          2,
          1,
          6,
          3,
          28,
          0
        ],
        [
          3,
          0,
          50,
          1,
          1,
          6,
          2,
          0,
          0,
          2
        ],
        [
          4,
          0,
          5,
          31,
          6,
          1,
          0,
          0,
          0,
          5
        ],
        [
          0,
          1,
          1,
          4,
          82,
          2,
          1,
          1,
          0,
          1
        ],
        [
          3,
          2,
          10,
          5,
          0,
          47,
          2,
          2,
          0,
          1
        ],
        [
          0,
          1,
          1,
          0,
          1,
          0,
          55,
          0,
          0,
          1
        ],
        [
          2,
          0,
          3,
          0,
          0,
          0,
          0,
          78,
          0,
          0
        ],
        [
          10,
          30,
          0,
          0,
          1,
          2,
          16,
          1,
          118,
          2
        ],
        [
          1,
          0,
          1,
          8,
          0,
          2,
          0,
          1,
          0,
          140
        ]
      ],
      "eval_loss": 1.2837210893630981,
      "eval_runtime": 15.5402,
      "eval_samples_per_second": 67.567,
      "eval_steps_per_second": 4.247,
      "step": 4100
    },
    {
      "epoch": 13.09,
      "grad_norm": 2.188967704772949,
      "learning_rate": 2.5477707006369428e-05,
      "loss": 0.0632,
      "step": 4110
    },
    {
      "epoch": 13.12,
      "grad_norm": 1.1061598062515259,
      "learning_rate": 2.505307855626327e-05,
      "loss": 0.0618,
      "step": 4120
    },
    {
      "epoch": 13.15,
      "grad_norm": 1.8072304725646973,
      "learning_rate": 2.4628450106157114e-05,
      "loss": 0.0436,
      "step": 4130
    },
    {
      "epoch": 13.18,
      "grad_norm": 0.3257448971271515,
      "learning_rate": 2.4203821656050956e-05,
      "loss": 0.0804,
      "step": 4140
    },
    {
      "epoch": 13.22,
      "grad_norm": 0.5817819237709045,
      "learning_rate": 2.37791932059448e-05,
      "loss": 0.034,
      "step": 4150
    },
    {
      "epoch": 13.25,
      "grad_norm": 2.3578078746795654,
      "learning_rate": 2.3354564755838642e-05,
      "loss": 0.069,
      "step": 4160
    },
    {
      "epoch": 13.28,
      "grad_norm": 0.23316432535648346,
      "learning_rate": 2.2929936305732484e-05,
      "loss": 0.0593,
      "step": 4170
    },
    {
      "epoch": 13.31,
      "grad_norm": 1.1364542245864868,
      "learning_rate": 2.250530785562633e-05,
      "loss": 0.0801,
      "step": 4180
    },
    {
      "epoch": 13.34,
      "grad_norm": 3.3843605518341064,
      "learning_rate": 2.208067940552017e-05,
      "loss": 0.1013,
      "step": 4190
    },
    {
      "epoch": 13.38,
      "grad_norm": 0.20773310959339142,
      "learning_rate": 2.1656050955414015e-05,
      "loss": 0.0753,
      "step": 4200
    },
    {
      "epoch": 13.38,
      "eval_accuracy": 0.7371428571428571,
      "eval_confusion_matrix": [
        [
          72,
          4,
          9,
          2,
          0,
          3,
          3,
          2,
          4,
          1
        ],
        [
          46,
          110,
          0,
          0,
          1,
          1,
          5,
          3,
          27,
          0
        ],
        [
          3,
          0,
          51,
          1,
          1,
          6,
          1,
          0,
          0,
          2
        ],
        [
          4,
          0,
          5,
          30,
          6,
          1,
          0,
          0,
          0,
          6
        ],
        [
          0,
          1,
          2,
          3,
          80,
          2,
          1,
          1,
          2,
          1
        ],
        [
          7,
          2,
          8,
          4,
          0,
          49,
          0,
          0,
          1,
          1
        ],
        [
          0,
          1,
          1,
          0,
          1,
          0,
          55,
          0,
          0,
          1
        ],
        [
          2,
          0,
          4,
          0,
          0,
          1,
          0,
          76,
          0,
          0
        ],
        [
          9,
          34,
          1,
          0,
          1,
          1,
          12,
          1,
          120,
          1
        ],
        [
          7,
          1,
          1,
          9,
          0,
          3,
          0,
          1,
          0,
          131
        ]
      ],
      "eval_loss": 1.2866138219833374,
      "eval_runtime": 15.4187,
      "eval_samples_per_second": 68.099,
      "eval_steps_per_second": 4.281,
      "step": 4200
    },
    {
      "epoch": 13.41,
      "grad_norm": 1.7785210609436035,
      "learning_rate": 2.1231422505307857e-05,
      "loss": 0.0957,
      "step": 4210
    },
    {
      "epoch": 13.44,
      "grad_norm": 4.4371185302734375,
      "learning_rate": 2.0806794055201702e-05,
      "loss": 0.0679,
      "step": 4220
    },
    {
      "epoch": 13.47,
      "grad_norm": 3.352201223373413,
      "learning_rate": 2.0382165605095544e-05,
      "loss": 0.0673,
      "step": 4230
    },
    {
      "epoch": 13.5,
      "grad_norm": 1.5773141384124756,
      "learning_rate": 1.9957537154989385e-05,
      "loss": 0.1248,
      "step": 4240
    },
    {
      "epoch": 13.54,
      "grad_norm": 1.759545922279358,
      "learning_rate": 1.953290870488323e-05,
      "loss": 0.0741,
      "step": 4250
    },
    {
      "epoch": 13.57,
      "grad_norm": 2.9278829097747803,
      "learning_rate": 1.910828025477707e-05,
      "loss": 0.0693,
      "step": 4260
    },
    {
      "epoch": 13.6,
      "grad_norm": 0.01691564917564392,
      "learning_rate": 1.8683651804670913e-05,
      "loss": 0.095,
      "step": 4270
    },
    {
      "epoch": 13.63,
      "grad_norm": 3.3381359577178955,
      "learning_rate": 1.8259023354564755e-05,
      "loss": 0.0699,
      "step": 4280
    },
    {
      "epoch": 13.66,
      "grad_norm": 0.33229127526283264,
      "learning_rate": 1.78343949044586e-05,
      "loss": 0.0591,
      "step": 4290
    },
    {
      "epoch": 13.69,
      "grad_norm": 0.5169569849967957,
      "learning_rate": 1.740976645435244e-05,
      "loss": 0.0766,
      "step": 4300
    },
    {
      "epoch": 13.69,
      "eval_accuracy": 0.7323809523809524,
      "eval_confusion_matrix": [
        [
          68,
          5,
          9,
          5,
          0,
          2,
          3,
          3,
          3,
          2
        ],
        [
          53,
          106,
          0,
          0,
          1,
          1,
          6,
          4,
          22,
          0
        ],
        [
          2,
          0,
          54,
          1,
          1,
          4,
          1,
          0,
          0,
          2
        ],
        [
          3,
          0,
          5,
          34,
          4,
          1,
          0,
          0,
          0,
          5
        ],
        [
          0,
          2,
          1,
          4,
          79,
          2,
          2,
          1,
          1,
          1
        ],
        [
          4,
          2,
          10,
          4,
          0,
          47,
          1,
          2,
          1,
          1
        ],
        [
          0,
          1,
          1,
          0,
          1,
          0,
          55,
          0,
          0,
          1
        ],
        [
          2,
          0,
          3,
          0,
          0,
          0,
          1,
          77,
          0,
          0
        ],
        [
          10,
          37,
          0,
          0,
          1,
          1,
          13,
          1,
          115,
          2
        ],
        [
          5,
          0,
          1,
          11,
          0,
          1,
          0,
          1,
          0,
          134
        ]
      ],
      "eval_loss": 1.3334447145462036,
      "eval_runtime": 15.6561,
      "eval_samples_per_second": 67.067,
      "eval_steps_per_second": 4.216,
      "step": 4300
    },
    {
      "epoch": 13.73,
      "grad_norm": 3.5023386478424072,
      "learning_rate": 1.6985138004246283e-05,
      "loss": 0.0602,
      "step": 4310
    },
    {
      "epoch": 13.76,
      "grad_norm": 1.4914608001708984,
      "learning_rate": 1.6560509554140128e-05,
      "loss": 0.0985,
      "step": 4320
    },
    {
      "epoch": 13.79,
      "grad_norm": 3.574211359024048,
      "learning_rate": 1.613588110403397e-05,
      "loss": 0.1329,
      "step": 4330
    },
    {
      "epoch": 13.82,
      "grad_norm": 1.6108194589614868,
      "learning_rate": 1.5711252653927814e-05,
      "loss": 0.0571,
      "step": 4340
    },
    {
      "epoch": 13.85,
      "grad_norm": 1.1470590829849243,
      "learning_rate": 1.5286624203821656e-05,
      "loss": 0.0615,
      "step": 4350
    },
    {
      "epoch": 13.89,
      "grad_norm": 1.373115062713623,
      "learning_rate": 1.48619957537155e-05,
      "loss": 0.0951,
      "step": 4360
    },
    {
      "epoch": 13.92,
      "grad_norm": 0.7028450965881348,
      "learning_rate": 1.4437367303609342e-05,
      "loss": 0.0556,
      "step": 4370
    },
    {
      "epoch": 13.95,
      "grad_norm": 1.0304226875305176,
      "learning_rate": 1.4012738853503186e-05,
      "loss": 0.0641,
      "step": 4380
    },
    {
      "epoch": 13.98,
      "grad_norm": 2.3109376430511475,
      "learning_rate": 1.3588110403397027e-05,
      "loss": 0.0375,
      "step": 4390
    },
    {
      "epoch": 14.01,
      "grad_norm": 0.2413637489080429,
      "learning_rate": 1.316348195329087e-05,
      "loss": 0.0699,
      "step": 4400
    },
    {
      "epoch": 14.01,
      "eval_accuracy": 0.7276190476190476,
      "eval_confusion_matrix": [
        [
          66,
          3,
          8,
          5,
          0,
          2,
          3,
          5,
          6,
          2
        ],
        [
          59,
          94,
          0,
          0,
          1,
          1,
          5,
          5,
          28,
          0
        ],
        [
          2,
          0,
          53,
          1,
          1,
          4,
          1,
          1,
          0,
          2
        ],
        [
          4,
          0,
          5,
          33,
          5,
          0,
          0,
          1,
          0,
          4
        ],
        [
          0,
          1,
          2,
          4,
          79,
          2,
          2,
          1,
          1,
          1
        ],
        [
          3,
          1,
          13,
          5,
          0,
          44,
          1,
          3,
          1,
          1
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          56,
          0,
          0,
          1
        ],
        [
          2,
          0,
          1,
          0,
          0,
          0,
          1,
          79,
          0,
          0
        ],
        [
          10,
          27,
          0,
          0,
          1,
          1,
          12,
          2,
          125,
          2
        ],
        [
          5,
          0,
          1,
          11,
          0,
          0,
          0,
          1,
          0,
          135
        ]
      ],
      "eval_loss": 1.390450119972229,
      "eval_runtime": 15.2795,
      "eval_samples_per_second": 68.72,
      "eval_steps_per_second": 4.32,
      "step": 4400
    },
    {
      "epoch": 14.04,
      "grad_norm": 0.03319675475358963,
      "learning_rate": 1.2738853503184714e-05,
      "loss": 0.0788,
      "step": 4410
    },
    {
      "epoch": 14.08,
      "grad_norm": 0.9733434319496155,
      "learning_rate": 1.2314225053078557e-05,
      "loss": 0.1165,
      "step": 4420
    },
    {
      "epoch": 14.11,
      "grad_norm": 0.15556958317756653,
      "learning_rate": 1.18895966029724e-05,
      "loss": 0.0719,
      "step": 4430
    },
    {
      "epoch": 14.14,
      "grad_norm": 0.46565964818000793,
      "learning_rate": 1.1464968152866242e-05,
      "loss": 0.0699,
      "step": 4440
    },
    {
      "epoch": 14.17,
      "grad_norm": 0.6882303357124329,
      "learning_rate": 1.1040339702760085e-05,
      "loss": 0.1229,
      "step": 4450
    },
    {
      "epoch": 14.2,
      "grad_norm": 0.026163959875702858,
      "learning_rate": 1.0615711252653929e-05,
      "loss": 0.0653,
      "step": 4460
    },
    {
      "epoch": 14.24,
      "grad_norm": 2.7348830699920654,
      "learning_rate": 1.0191082802547772e-05,
      "loss": 0.0818,
      "step": 4470
    },
    {
      "epoch": 14.27,
      "grad_norm": 0.20675204694271088,
      "learning_rate": 9.766454352441615e-06,
      "loss": 0.0578,
      "step": 4480
    },
    {
      "epoch": 14.3,
      "grad_norm": 2.2523086071014404,
      "learning_rate": 9.341825902335457e-06,
      "loss": 0.1141,
      "step": 4490
    },
    {
      "epoch": 14.33,
      "grad_norm": 0.6537735462188721,
      "learning_rate": 8.9171974522293e-06,
      "loss": 0.1218,
      "step": 4500
    },
    {
      "epoch": 14.33,
      "eval_accuracy": 0.7323809523809524,
      "eval_confusion_matrix": [
        [
          68,
          3,
          8,
          4,
          0,
          1,
          3,
          4,
          7,
          2
        ],
        [
          58,
          92,
          0,
          0,
          1,
          1,
          5,
          4,
          32,
          0
        ],
        [
          1,
          0,
          54,
          1,
          1,
          4,
          1,
          0,
          0,
          3
        ],
        [
          4,
          0,
          5,
          33,
          5,
          0,
          0,
          0,
          0,
          5
        ],
        [
          0,
          1,
          2,
          4,
          80,
          3,
          2,
          1,
          0,
          0
        ],
        [
          2,
          1,
          13,
          4,
          0,
          44,
          2,
          3,
          1,
          2
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          56,
          0,
          0,
          1
        ],
        [
          2,
          0,
          1,
          0,
          0,
          0,
          1,
          79,
          0,
          0
        ],
        [
          9,
          26,
          0,
          0,
          1,
          1,
          15,
          1,
          126,
          1
        ],
        [
          4,
          0,
          1,
          10,
          0,
          0,
          0,
          1,
          0,
          137
        ]
      ],
      "eval_loss": 1.3635234832763672,
      "eval_runtime": 15.2732,
      "eval_samples_per_second": 68.748,
      "eval_steps_per_second": 4.321,
      "step": 4500
    },
    {
      "epoch": 14.36,
      "grad_norm": 4.527649402618408,
      "learning_rate": 8.492569002123141e-06,
      "loss": 0.0613,
      "step": 4510
    },
    {
      "epoch": 14.39,
      "grad_norm": 1.7878096103668213,
      "learning_rate": 8.067940552016985e-06,
      "loss": 0.0759,
      "step": 4520
    },
    {
      "epoch": 14.43,
      "grad_norm": 3.677371025085449,
      "learning_rate": 7.643312101910828e-06,
      "loss": 0.0909,
      "step": 4530
    },
    {
      "epoch": 14.46,
      "grad_norm": 0.7390831708908081,
      "learning_rate": 7.218683651804671e-06,
      "loss": 0.0565,
      "step": 4540
    },
    {
      "epoch": 14.49,
      "grad_norm": 0.6755375862121582,
      "learning_rate": 6.794055201698514e-06,
      "loss": 0.052,
      "step": 4550
    },
    {
      "epoch": 14.52,
      "grad_norm": 0.20097573101520538,
      "learning_rate": 6.369426751592357e-06,
      "loss": 0.0881,
      "step": 4560
    },
    {
      "epoch": 14.55,
      "grad_norm": 0.3152300715446472,
      "learning_rate": 5.9447983014862e-06,
      "loss": 0.0575,
      "step": 4570
    },
    {
      "epoch": 14.59,
      "grad_norm": 3.195133924484253,
      "learning_rate": 5.520169851380043e-06,
      "loss": 0.1314,
      "step": 4580
    },
    {
      "epoch": 14.62,
      "grad_norm": 3.097222328186035,
      "learning_rate": 5.095541401273886e-06,
      "loss": 0.078,
      "step": 4590
    },
    {
      "epoch": 14.65,
      "grad_norm": 0.018327347934246063,
      "learning_rate": 4.670912951167728e-06,
      "loss": 0.0648,
      "step": 4600
    },
    {
      "epoch": 14.65,
      "eval_accuracy": 0.7342857142857143,
      "eval_confusion_matrix": [
        [
          66,
          3,
          9,
          5,
          0,
          2,
          3,
          3,
          7,
          2
        ],
        [
          52,
          95,
          0,
          0,
          2,
          1,
          6,
          4,
          33,
          0
        ],
        [
          3,
          1,
          52,
          1,
          1,
          3,
          2,
          0,
          0,
          2
        ],
        [
          4,
          0,
          5,
          33,
          5,
          0,
          0,
          0,
          0,
          5
        ],
        [
          0,
          2,
          1,
          3,
          83,
          1,
          2,
          1,
          0,
          0
        ],
        [
          2,
          2,
          13,
          4,
          0,
          44,
          2,
          2,
          1,
          2
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          56,
          0,
          0,
          1
        ],
        [
          2,
          0,
          2,
          0,
          0,
          0,
          1,
          78,
          0,
          0
        ],
        [
          9,
          26,
          0,
          0,
          1,
          1,
          14,
          1,
          127,
          1
        ],
        [
          3,
          0,
          1,
          10,
          0,
          0,
          1,
          1,
          0,
          137
        ]
      ],
      "eval_loss": 1.3204517364501953,
      "eval_runtime": 15.2766,
      "eval_samples_per_second": 68.733,
      "eval_steps_per_second": 4.32,
      "step": 4600
    },
    {
      "epoch": 14.68,
      "grad_norm": 0.03679550439119339,
      "learning_rate": 4.246284501061571e-06,
      "loss": 0.0597,
      "step": 4610
    },
    {
      "epoch": 14.71,
      "grad_norm": 3.308004140853882,
      "learning_rate": 3.821656050955414e-06,
      "loss": 0.0358,
      "step": 4620
    },
    {
      "epoch": 14.75,
      "grad_norm": 3.7965080738067627,
      "learning_rate": 3.397027600849257e-06,
      "loss": 0.0974,
      "step": 4630
    },
    {
      "epoch": 14.78,
      "grad_norm": 1.77022385597229,
      "learning_rate": 2.9723991507431e-06,
      "loss": 0.1215,
      "step": 4640
    },
    {
      "epoch": 14.81,
      "grad_norm": 1.951418399810791,
      "learning_rate": 2.547770700636943e-06,
      "loss": 0.0692,
      "step": 4650
    },
    {
      "epoch": 14.84,
      "grad_norm": 2.4297380447387695,
      "learning_rate": 2.1231422505307854e-06,
      "loss": 0.0927,
      "step": 4660
    },
    {
      "epoch": 14.87,
      "grad_norm": 0.13824594020843506,
      "learning_rate": 1.6985138004246284e-06,
      "loss": 0.0704,
      "step": 4670
    },
    {
      "epoch": 14.9,
      "grad_norm": 2.2139551639556885,
      "learning_rate": 1.2738853503184715e-06,
      "loss": 0.1461,
      "step": 4680
    },
    {
      "epoch": 14.94,
      "grad_norm": 0.4319517910480499,
      "learning_rate": 8.492569002123142e-07,
      "loss": 0.0511,
      "step": 4690
    },
    {
      "epoch": 14.97,
      "grad_norm": 0.4486578702926636,
      "learning_rate": 4.246284501061571e-07,
      "loss": 0.0917,
      "step": 4700
    },
    {
      "epoch": 14.97,
      "eval_accuracy": 0.7342857142857143,
      "eval_confusion_matrix": [
        [
          66,
          3,
          9,
          5,
          0,
          2,
          3,
          3,
          7,
          2
        ],
        [
          52,
          98,
          0,
          0,
          2,
          1,
          6,
          4,
          30,
          0
        ],
        [
          3,
          1,
          52,
          1,
          1,
          3,
          2,
          0,
          0,
          2
        ],
        [
          4,
          0,
          5,
          33,
          5,
          0,
          0,
          0,
          0,
          5
        ],
        [
          0,
          2,
          1,
          3,
          83,
          1,
          2,
          1,
          0,
          0
        ],
        [
          2,
          2,
          13,
          4,
          0,
          44,
          2,
          2,
          1,
          2
        ],
        [
          0,
          0,
          1,
          0,
          1,
          0,
          56,
          0,
          0,
          1
        ],
        [
          2,
          0,
          2,
          0,
          0,
          0,
          1,
          78,
          0,
          0
        ],
        [
          9,
          28,
          0,
          0,
          1,
          1,
          15,
          1,
          124,
          1
        ],
        [
          3,
          0,
          1,
          10,
          0,
          1,
          0,
          1,
          0,
          137
        ]
      ],
      "eval_loss": 1.3111604452133179,
      "eval_runtime": 15.4688,
      "eval_samples_per_second": 67.878,
      "eval_steps_per_second": 4.267,
      "step": 4700
    },
    {
      "epoch": 15.0,
      "grad_norm": 2.132920980453491,
      "learning_rate": 0.0,
      "loss": 0.0933,
      "step": 4710
    },
    {
      "epoch": 15.0,
      "step": 4710,
      "total_flos": 1.1677717704563528e+19,
      "train_loss": 0.19079542797209098,
      "train_runtime": 4656.9657,
      "train_samples_per_second": 32.323,
      "train_steps_per_second": 1.011
    }
  ],
  "logging_steps": 10,
  "max_steps": 4710,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 100,
  "total_flos": 1.1677717704563528e+19,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}