{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 50.0,
  "eval_steps": 100,
  "global_step": 5600,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "grad_norm": Infinity,
      "learning_rate": 0.0,
      "loss": 8.857,
      "step": 1
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5771467685699463,
      "learning_rate": 6e-07,
      "loss": 9.0055,
      "step": 2
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5344083309173584,
      "learning_rate": 1.2e-06,
      "loss": 8.8467,
      "step": 3
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5785155296325684,
      "learning_rate": 1.8e-06,
      "loss": 8.9835,
      "step": 4
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5720224380493164,
      "learning_rate": 2.4e-06,
      "loss": 9.0011,
      "step": 5
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5483667850494385,
      "learning_rate": 2.9999999999999997e-06,
      "loss": 8.8629,
      "step": 6
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.550671100616455,
      "learning_rate": 3.6e-06,
      "loss": 9.07,
      "step": 7
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.605531692504883,
      "learning_rate": 4.2e-06,
      "loss": 9.0519,
      "step": 8
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6086299419403076,
      "learning_rate": 4.8e-06,
      "loss": 9.0498,
      "step": 9
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5434601306915283,
      "learning_rate": 5.399999999999999e-06,
      "loss": 8.6956,
      "step": 10
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.583211898803711,
      "learning_rate": 5.999999999999999e-06,
      "loss": 8.8429,
      "step": 11
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.576791524887085,
      "learning_rate": 6.599999999999999e-06,
      "loss": 8.7839,
      "step": 12
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.724433660507202,
      "learning_rate": 7.2e-06,
      "loss": 9.0123,
      "step": 13
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.7472896575927734,
      "learning_rate": 7.799999999999998e-06,
      "loss": 8.7864,
      "step": 14
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.798316717147827,
      "learning_rate": 8.4e-06,
      "loss": 8.8184,
      "step": 15
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.821938991546631,
      "learning_rate": 8.999999999999999e-06,
      "loss": 8.7918,
      "step": 16
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.066065788269043,
      "learning_rate": 9.6e-06,
      "loss": 8.7545,
      "step": 17
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.990442991256714,
      "learning_rate": 1.02e-05,
      "loss": 8.4884,
      "step": 18
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.1646595001220703,
      "learning_rate": 1.0799999999999998e-05,
      "loss": 8.7851,
      "step": 19
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.23149037361145,
      "learning_rate": 1.14e-05,
      "loss": 8.4444,
      "step": 20
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.3869054317474365,
      "learning_rate": 1.1999999999999999e-05,
      "loss": 8.5387,
      "step": 21
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.4858381748199463,
      "learning_rate": 1.26e-05,
      "loss": 8.5025,
      "step": 22
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.6255927085876465,
      "learning_rate": 1.3199999999999997e-05,
      "loss": 8.3918,
      "step": 23
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.784517765045166,
      "learning_rate": 1.3799999999999998e-05,
      "loss": 8.4957,
      "step": 24
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.8559398651123047,
      "learning_rate": 1.44e-05,
      "loss": 8.443,
      "step": 25
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.133277416229248,
      "learning_rate": 1.4999999999999999e-05,
      "loss": 8.2685,
      "step": 26
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.246262550354004,
      "learning_rate": 1.5599999999999996e-05,
      "loss": 8.0652,
      "step": 27
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.553827285766602,
      "learning_rate": 1.6199999999999997e-05,
      "loss": 7.9432,
      "step": 28
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.786936283111572,
      "learning_rate": 1.68e-05,
      "loss": 7.8762,
      "step": 29
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.037590980529785,
      "learning_rate": 1.74e-05,
      "loss": 7.8238,
      "step": 30
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.190773010253906,
      "learning_rate": 1.7999999999999997e-05,
      "loss": 7.8229,
      "step": 31
    },
    {
      "epoch": 0.29,
      "grad_norm": Infinity,
      "learning_rate": 1.7999999999999997e-05,
      "loss": 8.1275,
      "step": 32
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.320779323577881,
      "learning_rate": 1.8599999999999998e-05,
      "loss": 7.5935,
      "step": 33
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.777307987213135,
      "learning_rate": 1.92e-05,
      "loss": 7.6314,
      "step": 34
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.084140300750732,
      "learning_rate": 1.98e-05,
      "loss": 7.4086,
      "step": 35
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.326981544494629,
      "learning_rate": 2.04e-05,
      "loss": 7.1818,
      "step": 36
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.600687503814697,
      "learning_rate": 2.1e-05,
      "loss": 6.999,
      "step": 37
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.262928009033203,
      "learning_rate": 2.1599999999999996e-05,
      "loss": 7.098,
      "step": 38
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.6612653732299805,
      "learning_rate": 2.2199999999999998e-05,
      "loss": 6.7866,
      "step": 39
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.096963882446289,
      "learning_rate": 2.28e-05,
      "loss": 6.6355,
      "step": 40
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.546558380126953,
      "learning_rate": 2.34e-05,
      "loss": 6.4434,
      "step": 41
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.764568328857422,
      "learning_rate": 2.3999999999999997e-05,
      "loss": 6.1754,
      "step": 42
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.918572425842285,
      "learning_rate": 2.4599999999999998e-05,
      "loss": 5.8441,
      "step": 43
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.128335952758789,
      "learning_rate": 2.52e-05,
      "loss": 6.056,
      "step": 44
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.886902809143066,
      "learning_rate": 2.5799999999999997e-05,
      "loss": 5.5803,
      "step": 45
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.779302597045898,
      "learning_rate": 2.6399999999999995e-05,
      "loss": 5.3041,
      "step": 46
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.12820053100586,
      "learning_rate": 2.6999999999999996e-05,
      "loss": 5.1445,
      "step": 47
    },
    {
      "epoch": 0.43,
      "grad_norm": 10.151182174682617,
      "learning_rate": 2.7599999999999997e-05,
      "loss": 4.955,
      "step": 48
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.030814170837402,
      "learning_rate": 2.8199999999999998e-05,
      "loss": 4.7517,
      "step": 49
    },
    {
      "epoch": 0.45,
      "grad_norm": 10.629888534545898,
      "learning_rate": 2.88e-05,
      "loss": 4.7841,
      "step": 50
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.391251564025879,
      "learning_rate": 2.94e-05,
      "loss": 4.3701,
      "step": 51
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.37604808807373,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 4.2677,
      "step": 52
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.82244873046875,
      "learning_rate": 3.06e-05,
      "loss": 4.1149,
      "step": 53
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.434287071228027,
      "learning_rate": 3.119999999999999e-05,
      "loss": 4.0036,
      "step": 54
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.634397029876709,
      "learning_rate": 3.1799999999999994e-05,
      "loss": 3.8544,
      "step": 55
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.66248893737793,
      "learning_rate": 3.2399999999999995e-05,
      "loss": 3.7289,
      "step": 56
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.404279708862305,
      "learning_rate": 3.2999999999999996e-05,
      "loss": 3.677,
      "step": 57
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.844570159912109,
      "learning_rate": 3.36e-05,
      "loss": 3.5791,
      "step": 58
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.26457405090332,
      "learning_rate": 3.42e-05,
      "loss": 3.5138,
      "step": 59
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.4815754890441895,
      "learning_rate": 3.48e-05,
      "loss": 3.4393,
      "step": 60
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.832083225250244,
      "learning_rate": 3.539999999999999e-05,
      "loss": 3.382,
      "step": 61
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.4209747314453125,
      "learning_rate": 3.5999999999999994e-05,
      "loss": 3.3589,
      "step": 62
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.663290500640869,
      "learning_rate": 3.6599999999999995e-05,
      "loss": 3.3078,
      "step": 63
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2917840480804443,
      "learning_rate": 3.7199999999999996e-05,
      "loss": 3.2546,
      "step": 64
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.849010944366455,
      "learning_rate": 3.78e-05,
      "loss": 3.2213,
      "step": 65
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.583945870399475,
      "learning_rate": 3.84e-05,
      "loss": 3.19,
      "step": 66
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6001330614089966,
      "learning_rate": 3.9e-05,
      "loss": 3.1823,
      "step": 67
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4142842292785645,
      "learning_rate": 3.96e-05,
      "loss": 3.1601,
      "step": 68
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3812508583068848,
      "learning_rate": 4.02e-05,
      "loss": 3.1669,
      "step": 69
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1935653686523438,
      "learning_rate": 4.08e-05,
      "loss": 3.0948,
      "step": 70
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2233781814575195,
      "learning_rate": 4.14e-05,
      "loss": 3.0698,
      "step": 71
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3483402729034424,
      "learning_rate": 4.2e-05,
      "loss": 3.0577,
      "step": 72
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1394308805465698,
      "learning_rate": 4.259999999999999e-05,
      "loss": 3.0351,
      "step": 73
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8430189490318298,
      "learning_rate": 4.319999999999999e-05,
      "loss": 3.0126,
      "step": 74
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0437767505645752,
      "learning_rate": 4.3799999999999994e-05,
      "loss": 3.0426,
      "step": 75
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6937194466590881,
      "learning_rate": 4.4399999999999995e-05,
      "loss": 2.9787,
      "step": 76
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6742655634880066,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 2.9759,
      "step": 77
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7938650846481323,
      "learning_rate": 4.56e-05,
      "loss": 2.9632,
      "step": 78
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8932818174362183,
      "learning_rate": 4.62e-05,
      "loss": 2.9627,
      "step": 79
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6115772724151611,
      "learning_rate": 4.68e-05,
      "loss": 2.9523,
      "step": 80
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7806911468505859,
      "learning_rate": 4.7399999999999993e-05,
      "loss": 2.9645,
      "step": 81
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5309816598892212,
      "learning_rate": 4.7999999999999994e-05,
      "loss": 2.9546,
      "step": 82
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.42638012766838074,
      "learning_rate": 4.8599999999999995e-05,
      "loss": 2.9237,
      "step": 83
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4000401198863983,
      "learning_rate": 4.9199999999999997e-05,
      "loss": 2.9201,
      "step": 84
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4595623314380646,
      "learning_rate": 4.98e-05,
      "loss": 2.9188,
      "step": 85
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.44960692524909973,
      "learning_rate": 5.04e-05,
      "loss": 2.9148,
      "step": 86
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3653245270252228,
      "learning_rate": 5.1e-05,
      "loss": 2.9165,
      "step": 87
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6640657782554626,
      "learning_rate": 5.1599999999999994e-05,
      "loss": 2.931,
      "step": 88
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5979864597320557,
      "learning_rate": 5.2199999999999995e-05,
      "loss": 2.8992,
      "step": 89
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.34149181842803955,
      "learning_rate": 5.279999999999999e-05,
      "loss": 2.8991,
      "step": 90
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.28454726934432983,
      "learning_rate": 5.339999999999999e-05,
      "loss": 2.898,
      "step": 91
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2736092805862427,
      "learning_rate": 5.399999999999999e-05,
      "loss": 2.9005,
      "step": 92
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.25253403186798096,
      "learning_rate": 5.459999999999999e-05,
      "loss": 2.9024,
      "step": 93
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6596384644508362,
      "learning_rate": 5.519999999999999e-05,
      "loss": 2.9311,
      "step": 94
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5612274408340454,
      "learning_rate": 5.5799999999999994e-05,
      "loss": 2.8869,
      "step": 95
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2838483452796936,
      "learning_rate": 5.6399999999999995e-05,
      "loss": 2.8794,
      "step": 96
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.22255806624889374,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 2.8877,
      "step": 97
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2908608317375183,
      "learning_rate": 5.76e-05,
      "loss": 2.8888,
      "step": 98
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20237217843532562,
      "learning_rate": 5.82e-05,
      "loss": 2.8899,
      "step": 99
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19447995722293854,
      "learning_rate": 5.88e-05,
      "loss": 2.911,
      "step": 100
    },
    {
      "epoch": 0.89,
      "eval_cer": 1.0,
      "eval_loss": 2.9202330112457275,
      "eval_runtime": 25.3685,
      "eval_samples_per_second": 104.145,
      "eval_steps_per_second": 1.656,
      "eval_wer": 1.0,
      "step": 100
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5298409461975098,
      "learning_rate": 5.94e-05,
      "loss": 2.8743,
      "step": 101
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0151143074035645,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 2.8804,
      "step": 102
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2283215969800949,
      "learning_rate": 6.0599999999999996e-05,
      "loss": 2.8721,
      "step": 103
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39570146799087524,
      "learning_rate": 6.12e-05,
      "loss": 2.8835,
      "step": 104
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.49264252185821533,
      "learning_rate": 6.18e-05,
      "loss": 2.8768,
      "step": 105
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.40177270770072937,
      "learning_rate": 6.239999999999999e-05,
      "loss": 2.9014,
      "step": 106
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3801201581954956,
      "learning_rate": 6.299999999999999e-05,
      "loss": 2.885,
      "step": 107
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6969913244247437,
      "learning_rate": 6.359999999999999e-05,
      "loss": 2.8762,
      "step": 108
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6147791147232056,
      "learning_rate": 6.419999999999999e-05,
      "loss": 2.8752,
      "step": 109
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3819006085395813,
      "learning_rate": 6.479999999999999e-05,
      "loss": 2.8739,
      "step": 110
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.48288494348526,
      "learning_rate": 6.539999999999999e-05,
      "loss": 2.8775,
      "step": 111
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5739099383354187,
      "learning_rate": 6.599999999999999e-05,
      "loss": 2.8859,
      "step": 112
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.25912415981292725,
      "learning_rate": 6.659999999999999e-05,
      "loss": 2.8655,
      "step": 113
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.3443286120891571,
      "learning_rate": 6.72e-05,
      "loss": 2.8648,
      "step": 114
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.823746919631958,
      "learning_rate": 6.78e-05,
      "loss": 2.869,
      "step": 115
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7289327383041382,
      "learning_rate": 6.84e-05,
      "loss": 2.872,
      "step": 116
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.13738706707954407,
      "learning_rate": 6.9e-05,
      "loss": 2.8722,
      "step": 117
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.5169634819030762,
      "learning_rate": 6.96e-05,
      "loss": 2.8835,
      "step": 118
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.4314047694206238,
      "learning_rate": 7.02e-05,
      "loss": 2.8764,
      "step": 119
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.29641491174697876,
      "learning_rate": 7.079999999999999e-05,
      "loss": 2.8572,
      "step": 120
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.41689497232437134,
      "learning_rate": 7.139999999999999e-05,
      "loss": 2.8577,
      "step": 121
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.45255449414253235,
      "learning_rate": 7.199999999999999e-05,
      "loss": 2.8569,
      "step": 122
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.15350182354450226,
      "learning_rate": 7.259999999999999e-05,
      "loss": 2.8582,
      "step": 123
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.44321611523628235,
      "learning_rate": 7.319999999999999e-05,
      "loss": 2.865,
      "step": 124
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.5034608840942383,
      "learning_rate": 7.379999999999999e-05,
      "loss": 2.8892,
      "step": 125
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.33695104718208313,
      "learning_rate": 7.439999999999999e-05,
      "loss": 2.8518,
      "step": 126
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.1340596228837967,
      "learning_rate": 7.5e-05,
      "loss": 2.8534,
      "step": 127
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.15238897502422333,
      "learning_rate": 7.56e-05,
      "loss": 2.8507,
      "step": 128
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.5069476366043091,
      "learning_rate": 7.62e-05,
      "loss": 2.8568,
      "step": 129
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.38025808334350586,
      "learning_rate": 7.68e-05,
      "loss": 2.8556,
      "step": 130
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.5071201920509338,
      "learning_rate": 7.74e-05,
      "loss": 2.881,
      "step": 131
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.10134594887495041,
      "learning_rate": 7.8e-05,
      "loss": 2.8519,
      "step": 132
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.41321516036987305,
      "learning_rate": 7.86e-05,
      "loss": 2.851,
      "step": 133
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7913742065429688,
      "learning_rate": 7.92e-05,
      "loss": 2.848,
      "step": 134
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.20887543261051178,
      "learning_rate": 7.98e-05,
      "loss": 2.853,
      "step": 135
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.6060431599617004,
      "learning_rate": 8.04e-05,
      "loss": 2.8567,
      "step": 136
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.3716076612472534,
      "learning_rate": 8.1e-05,
      "loss": 2.8746,
      "step": 137
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8390268087387085,
      "learning_rate": 8.16e-05,
      "loss": 2.853,
      "step": 138
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.12195856869220734,
      "learning_rate": 8.22e-05,
      "loss": 2.8403,
      "step": 139
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.6726837158203125,
      "learning_rate": 8.28e-05,
      "loss": 2.8468,
      "step": 140
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.3162238299846649,
      "learning_rate": 8.34e-05,
      "loss": 2.8498,
      "step": 141
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.5609211921691895,
      "learning_rate": 8.4e-05,
      "loss": 2.8494,
      "step": 142
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.4808315932750702,
      "learning_rate": 8.459999999999998e-05,
      "loss": 2.8579,
      "step": 143
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.10037563741207123,
      "learning_rate": 8.519999999999998e-05,
      "loss": 2.8612,
      "step": 144
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.5827155709266663,
      "learning_rate": 8.579999999999998e-05,
      "loss": 2.8418,
      "step": 145
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.16088470816612244,
      "learning_rate": 8.639999999999999e-05,
      "loss": 2.8429,
      "step": 146
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.6953414678573608,
      "learning_rate": 8.699999999999999e-05,
      "loss": 2.8449,
      "step": 147
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.28255772590637207,
      "learning_rate": 8.759999999999999e-05,
      "loss": 2.8393,
      "step": 148
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.5030548572540283,
      "learning_rate": 8.819999999999999e-05,
      "loss": 2.8519,
      "step": 149
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.5152975916862488,
      "learning_rate": 8.879999999999999e-05,
      "loss": 2.8604,
      "step": 150
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9333630800247192,
      "learning_rate": 8.939999999999999e-05,
      "loss": 2.8423,
      "step": 151
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.4193758964538574,
      "learning_rate": 8.999999999999999e-05,
      "loss": 2.8411,
      "step": 152
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8759561777114868,
      "learning_rate": 9.059999999999999e-05,
      "loss": 2.8471,
      "step": 153
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.1738540530204773,
      "learning_rate": 9.12e-05,
      "loss": 2.843,
      "step": 154
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7106860280036926,
      "learning_rate": 9.18e-05,
      "loss": 2.8416,
      "step": 155
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7692212462425232,
      "learning_rate": 9.24e-05,
      "loss": 2.8608,
      "step": 156
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.10494925081729889,
      "learning_rate": 9.3e-05,
      "loss": 2.8365,
      "step": 157
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.651297926902771,
      "learning_rate": 9.36e-05,
      "loss": 2.8454,
      "step": 158
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.5796992778778076,
      "learning_rate": 9.419999999999999e-05,
      "loss": 2.8359,
      "step": 159
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.2963010370731354,
      "learning_rate": 9.479999999999999e-05,
      "loss": 2.8367,
      "step": 160
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7967988848686218,
      "learning_rate": 9.539999999999999e-05,
      "loss": 2.8444,
      "step": 161
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.2917342483997345,
      "learning_rate": 9.599999999999999e-05,
      "loss": 2.8668,
      "step": 162
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.21879711747169495,
      "learning_rate": 9.659999999999999e-05,
      "loss": 2.8316,
      "step": 163
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.09713856130838394,
      "learning_rate": 9.719999999999999e-05,
      "loss": 2.8323,
      "step": 164
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.16893808543682098,
      "learning_rate": 9.779999999999999e-05,
      "loss": 2.8323,
      "step": 165
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.21297118067741394,
      "learning_rate": 9.839999999999999e-05,
      "loss": 2.8347,
      "step": 166
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.09424562007188797,
      "learning_rate": 9.9e-05,
      "loss": 2.8343,
      "step": 167
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.13788117468357086,
      "learning_rate": 9.96e-05,
      "loss": 2.8413,
      "step": 168
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.44669249653816223,
      "learning_rate": 0.0001002,
      "loss": 2.8367,
      "step": 169
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.1038396805524826,
      "learning_rate": 0.0001008,
      "loss": 2.8268,
      "step": 170
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.32863888144493103,
      "learning_rate": 0.0001014,
      "loss": 2.8248,
      "step": 171
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.2267109751701355,
      "learning_rate": 0.000102,
      "loss": 2.8296,
      "step": 172
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.2983206510543823,
      "learning_rate": 0.0001026,
      "loss": 2.8313,
      "step": 173
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.20100681483745575,
      "learning_rate": 0.00010319999999999999,
      "loss": 2.8315,
      "step": 174
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.17761147022247314,
      "learning_rate": 0.00010379999999999999,
      "loss": 2.8338,
      "step": 175
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.14317427575588226,
      "learning_rate": 0.00010439999999999999,
      "loss": 2.8239,
      "step": 176
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.12552930414676666,
      "learning_rate": 0.00010499999999999999,
      "loss": 2.8211,
      "step": 177
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.12438877671957016,
      "learning_rate": 0.00010559999999999998,
      "loss": 2.8225,
      "step": 178
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.12741345167160034,
      "learning_rate": 0.00010619999999999998,
      "loss": 2.8194,
      "step": 179
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.13339456915855408,
      "learning_rate": 0.00010679999999999998,
      "loss": 2.8235,
      "step": 180
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.3932964503765106,
      "learning_rate": 0.00010739999999999998,
      "loss": 2.8305,
      "step": 181
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.3690793514251709,
      "learning_rate": 0.00010799999999999998,
      "loss": 2.8183,
      "step": 182
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.5288318991661072,
      "learning_rate": 0.00010859999999999998,
      "loss": 2.8143,
      "step": 183
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.4029388725757599,
      "learning_rate": 0.00010919999999999998,
      "loss": 2.8117,
      "step": 184
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.6303567290306091,
      "learning_rate": 0.00010979999999999999,
      "loss": 2.8071,
      "step": 185
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.6404778957366943,
      "learning_rate": 0.00011039999999999999,
      "loss": 2.802,
      "step": 186
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1146502494812012,
      "learning_rate": 0.00011099999999999999,
      "loss": 2.8075,
      "step": 187
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7937551736831665,
      "learning_rate": 0.00011159999999999999,
      "loss": 2.7986,
      "step": 188
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.46796396374702454,
      "learning_rate": 0.00011219999999999999,
      "loss": 2.7816,
      "step": 189
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.5845624208450317,
      "learning_rate": 0.00011279999999999999,
      "loss": 2.7793,
      "step": 190
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.4756370484828949,
      "learning_rate": 0.00011339999999999999,
      "loss": 2.7683,
      "step": 191
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1186546087265015,
      "learning_rate": 0.00011399999999999999,
      "loss": 2.7664,
      "step": 192
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1542567014694214,
      "learning_rate": 0.0001146,
      "loss": 2.7696,
      "step": 193
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6869224309921265,
      "learning_rate": 0.0001152,
      "loss": 2.7689,
      "step": 194
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7766700983047485,
      "learning_rate": 0.0001158,
      "loss": 2.7333,
      "step": 195
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.721091091632843,
      "learning_rate": 0.0001164,
      "loss": 2.7311,
      "step": 196
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.5262578725814819,
      "learning_rate": 0.000117,
      "loss": 2.7136,
      "step": 197
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.6685904264450073,
      "learning_rate": 0.0001176,
      "loss": 2.6956,
      "step": 198
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.45331716537475586,
      "learning_rate": 0.0001182,
      "loss": 2.6695,
      "step": 199
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7517866492271423,
      "learning_rate": 0.0001188,
      "loss": 2.6638,
      "step": 200
    },
    {
      "epoch": 1.79,
      "eval_cer": 1.0,
      "eval_loss": 2.6310031414031982,
      "eval_runtime": 21.5145,
      "eval_samples_per_second": 122.801,
      "eval_steps_per_second": 1.952,
      "eval_wer": 1.0,
      "step": 200
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.4016678035259247,
      "learning_rate": 0.0001194,
      "loss": 2.6273,
      "step": 201
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.6813556551933289,
      "learning_rate": 0.00011999999999999999,
      "loss": 2.6077,
      "step": 202
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.5532939434051514,
      "learning_rate": 0.00012059999999999999,
      "loss": 2.583,
      "step": 203
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6075173616409302,
      "learning_rate": 0.00012119999999999999,
      "loss": 2.5695,
      "step": 204
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8583607077598572,
      "learning_rate": 0.00012179999999999999,
      "loss": 2.5337,
      "step": 205
    },
    {
      "epoch": 1.84,
      "grad_norm": 2.8024051189422607,
      "learning_rate": 0.0001224,
      "loss": 2.6199,
      "step": 206
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9846538305282593,
      "learning_rate": 0.00012299999999999998,
      "loss": 2.4776,
      "step": 207
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.8579438924789429,
      "learning_rate": 0.0001236,
      "loss": 2.5176,
      "step": 208
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7917272448539734,
      "learning_rate": 0.00012419999999999998,
      "loss": 2.4321,
      "step": 209
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.09323787689209,
      "learning_rate": 0.00012479999999999997,
      "loss": 2.4381,
      "step": 210
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4129819869995117,
      "learning_rate": 0.00012539999999999999,
      "loss": 2.3959,
      "step": 211
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.6987126469612122,
      "learning_rate": 0.00012599999999999997,
      "loss": 2.3383,
      "step": 212
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3300609588623047,
      "learning_rate": 0.0001266,
      "loss": 2.3031,
      "step": 213
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7649120092391968,
      "learning_rate": 0.00012719999999999997,
      "loss": 2.2646,
      "step": 214
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8899486064910889,
      "learning_rate": 0.0001278,
      "loss": 2.2044,
      "step": 215
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0937632322311401,
      "learning_rate": 0.00012839999999999998,
      "loss": 2.1425,
      "step": 216
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.6736981868743896,
      "learning_rate": 0.000129,
      "loss": 2.1169,
      "step": 217
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.6996066570281982,
      "learning_rate": 0.00012959999999999998,
      "loss": 2.0965,
      "step": 218
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8732956051826477,
      "learning_rate": 0.0001302,
      "loss": 2.0275,
      "step": 219
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.6698563098907471,
      "learning_rate": 0.00013079999999999998,
      "loss": 1.9618,
      "step": 220
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7226149439811707,
      "learning_rate": 0.0001314,
      "loss": 1.9156,
      "step": 221
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8683639764785767,
      "learning_rate": 0.00013199999999999998,
      "loss": 1.8771,
      "step": 222
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.6471112370491028,
      "learning_rate": 0.0001326,
      "loss": 1.833,
      "step": 223
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.917619526386261,
      "learning_rate": 0.00013319999999999999,
      "loss": 1.7896,
      "step": 224
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.6950381398200989,
      "learning_rate": 0.0001338,
      "loss": 1.7237,
      "step": 225
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.6106230616569519,
      "learning_rate": 0.0001344,
      "loss": 1.6922,
      "step": 226
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.7221829295158386,
      "learning_rate": 0.000135,
      "loss": 1.6354,
      "step": 227
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.6865891218185425,
      "learning_rate": 0.0001356,
      "loss": 1.5929,
      "step": 228
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.6008340716362,
      "learning_rate": 0.0001362,
      "loss": 1.5578,
      "step": 229
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.544625461101532,
      "learning_rate": 0.0001368,
      "loss": 1.5526,
      "step": 230
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.6607970595359802,
      "learning_rate": 0.0001374,
      "loss": 1.4816,
      "step": 231
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.5788168907165527,
      "learning_rate": 0.000138,
      "loss": 1.4238,
      "step": 232
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.6171405911445618,
      "learning_rate": 0.0001386,
      "loss": 1.3814,
      "step": 233
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.5263203382492065,
      "learning_rate": 0.0001392,
      "loss": 1.3584,
      "step": 234
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.7827364802360535,
      "learning_rate": 0.00013979999999999998,
      "loss": 1.3101,
      "step": 235
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8221073746681213,
      "learning_rate": 0.0001404,
      "loss": 1.3303,
      "step": 236
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6871161460876465,
      "learning_rate": 0.00014099999999999998,
      "loss": 1.3629,
      "step": 237
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.5879709720611572,
      "learning_rate": 0.00014159999999999997,
      "loss": 1.2105,
      "step": 238
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.0625423192977905,
      "learning_rate": 0.0001422,
      "loss": 1.2261,
      "step": 239
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.9403135180473328,
      "learning_rate": 0.00014279999999999997,
      "loss": 1.1616,
      "step": 240
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.7435916662216187,
      "learning_rate": 0.0001434,
      "loss": 1.133,
      "step": 241
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.6294271349906921,
      "learning_rate": 0.00014399999999999998,
      "loss": 1.1262,
      "step": 242
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.5731372237205505,
      "learning_rate": 0.0001446,
      "loss": 1.1516,
      "step": 243
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.9239470958709717,
      "learning_rate": 0.00014519999999999998,
      "loss": 1.0654,
      "step": 244
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.5397454500198364,
      "learning_rate": 0.0001458,
      "loss": 1.0191,
      "step": 245
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.6285163760185242,
      "learning_rate": 0.00014639999999999998,
      "loss": 0.9882,
      "step": 246
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.47989338636398315,
      "learning_rate": 0.000147,
      "loss": 0.9555,
      "step": 247
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.5626522302627563,
      "learning_rate": 0.00014759999999999998,
      "loss": 0.9482,
      "step": 248
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.5631589889526367,
      "learning_rate": 0.0001482,
      "loss": 1.0041,
      "step": 249
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.4852822721004486,
      "learning_rate": 0.00014879999999999998,
      "loss": 0.8956,
      "step": 250
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.5496442317962646,
      "learning_rate": 0.0001494,
      "loss": 0.8809,
      "step": 251
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.5785245299339294,
      "learning_rate": 0.00015,
      "loss": 0.8292,
      "step": 252
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.44672396779060364,
      "learning_rate": 0.00015059999999999997,
      "loss": 0.8298,
      "step": 253
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.419998437166214,
      "learning_rate": 0.0001512,
      "loss": 0.8039,
      "step": 254
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.4560867249965668,
      "learning_rate": 0.00015179999999999998,
      "loss": 0.8389,
      "step": 255
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.49839121103286743,
      "learning_rate": 0.0001524,
      "loss": 0.8024,
      "step": 256
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.41201063990592957,
      "learning_rate": 0.00015299999999999998,
      "loss": 0.7551,
      "step": 257
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.3268738090991974,
      "learning_rate": 0.0001536,
      "loss": 0.7296,
      "step": 258
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.46522414684295654,
      "learning_rate": 0.00015419999999999998,
      "loss": 0.7201,
      "step": 259
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.26850831508636475,
      "learning_rate": 0.0001548,
      "loss": 0.6957,
      "step": 260
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.38073208928108215,
      "learning_rate": 0.00015539999999999998,
      "loss": 0.7161,
      "step": 261
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.3288465738296509,
      "learning_rate": 0.000156,
      "loss": 0.7478,
      "step": 262
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.535457193851471,
      "learning_rate": 0.00015659999999999998,
      "loss": 0.6699,
      "step": 263
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.46445024013519287,
      "learning_rate": 0.0001572,
      "loss": 0.6405,
      "step": 264
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.37084606289863586,
      "learning_rate": 0.0001578,
      "loss": 0.6309,
      "step": 265
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.38821446895599365,
      "learning_rate": 0.0001584,
      "loss": 0.6272,
      "step": 266
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.2973250150680542,
      "learning_rate": 0.000159,
      "loss": 0.6242,
      "step": 267
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.4151294529438019,
      "learning_rate": 0.0001596,
      "loss": 0.6912,
      "step": 268
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.3166775107383728,
      "learning_rate": 0.0001602,
      "loss": 0.5899,
      "step": 269
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.426556795835495,
      "learning_rate": 0.0001608,
      "loss": 0.5855,
      "step": 270
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.27777695655822754,
      "learning_rate": 0.0001614,
      "loss": 0.5672,
      "step": 271
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.32402169704437256,
      "learning_rate": 0.000162,
      "loss": 0.5713,
      "step": 272
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.350260466337204,
      "learning_rate": 0.0001626,
      "loss": 0.5935,
      "step": 273
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.3984275162220001,
      "learning_rate": 0.0001632,
      "loss": 0.6266,
      "step": 274
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.39264658093452454,
      "learning_rate": 0.0001638,
      "loss": 0.5461,
      "step": 275
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.27602052688598633,
      "learning_rate": 0.0001644,
      "loss": 0.5409,
      "step": 276
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.3373797833919525,
      "learning_rate": 0.000165,
      "loss": 0.5357,
      "step": 277
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.2839423418045044,
      "learning_rate": 0.0001656,
      "loss": 0.5114,
      "step": 278
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.25080233812332153,
      "learning_rate": 0.0001662,
      "loss": 0.509,
      "step": 279
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.32631421089172363,
      "learning_rate": 0.0001668,
      "loss": 0.5674,
      "step": 280
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.3136104643344879,
      "learning_rate": 0.0001674,
      "loss": 0.5381,
      "step": 281
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.25540921092033386,
      "learning_rate": 0.000168,
      "loss": 0.4858,
      "step": 282
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.1932198703289032,
      "learning_rate": 0.0001686,
      "loss": 0.4748,
      "step": 283
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.2537154257297516,
      "learning_rate": 0.00016919999999999997,
      "loss": 0.488,
      "step": 284
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.27697309851646423,
      "learning_rate": 0.00016979999999999998,
      "loss": 0.4721,
      "step": 285
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.3219203054904938,
      "learning_rate": 0.00017039999999999997,
      "loss": 0.4909,
      "step": 286
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.29123619198799133,
      "learning_rate": 0.00017099999999999998,
      "loss": 0.5271,
      "step": 287
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.308193176984787,
      "learning_rate": 0.00017159999999999997,
      "loss": 0.4632,
      "step": 288
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.18822936713695526,
      "learning_rate": 0.00017219999999999998,
      "loss": 0.4514,
      "step": 289
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.25762876868247986,
      "learning_rate": 0.00017279999999999997,
      "loss": 0.453,
      "step": 290
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.22614410519599915,
      "learning_rate": 0.00017339999999999996,
      "loss": 0.4392,
      "step": 291
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.23655934631824493,
      "learning_rate": 0.00017399999999999997,
      "loss": 0.4501,
      "step": 292
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.308523029088974,
      "learning_rate": 0.00017459999999999996,
      "loss": 0.5553,
      "step": 293
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.21490143239498138,
      "learning_rate": 0.00017519999999999998,
      "loss": 0.4264,
      "step": 294
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.23168258368968964,
      "learning_rate": 0.00017579999999999996,
      "loss": 0.4151,
      "step": 295
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.22771677374839783,
      "learning_rate": 0.00017639999999999998,
      "loss": 0.4113,
      "step": 296
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.2682896852493286,
      "learning_rate": 0.00017699999999999997,
      "loss": 0.4089,
      "step": 297
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.24218544363975525,
      "learning_rate": 0.00017759999999999998,
      "loss": 0.4204,
      "step": 298
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.32331299781799316,
      "learning_rate": 0.00017819999999999997,
      "loss": 0.4884,
      "step": 299
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.22354701161384583,
      "learning_rate": 0.00017879999999999998,
      "loss": 0.3898,
      "step": 300
    },
    {
      "epoch": 2.68,
      "eval_cer": 0.09684936088968066,
      "eval_loss": 0.38924866914749146,
      "eval_runtime": 22.3523,
      "eval_samples_per_second": 118.198,
      "eval_steps_per_second": 1.879,
      "eval_wer": 0.33661245537485124,
      "step": 300
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.1929866522550583,
      "learning_rate": 0.00017939999999999997,
      "loss": 0.3921,
      "step": 301
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.2592866122722626,
      "learning_rate": 0.00017999999999999998,
      "loss": 0.3905,
      "step": 302
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.2864495515823364,
      "learning_rate": 0.00018059999999999997,
      "loss": 0.3885,
      "step": 303
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.21913693845272064,
      "learning_rate": 0.00018119999999999999,
      "loss": 0.4009,
      "step": 304
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.3071497082710266,
      "learning_rate": 0.00018179999999999997,
      "loss": 0.4195,
      "step": 305
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.22289547324180603,
      "learning_rate": 0.0001824,
      "loss": 0.4112,
      "step": 306
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.31046903133392334,
      "learning_rate": 0.00018299999999999998,
      "loss": 0.3764,
      "step": 307
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.2204066812992096,
      "learning_rate": 0.0001836,
      "loss": 0.3691,
      "step": 308
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.20296648144721985,
      "learning_rate": 0.00018419999999999998,
      "loss": 0.3492,
      "step": 309
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.2594054937362671,
      "learning_rate": 0.0001848,
      "loss": 0.3495,
      "step": 310
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.21860192716121674,
      "learning_rate": 0.00018539999999999998,
      "loss": 0.4076,
      "step": 311
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.36586835980415344,
      "learning_rate": 0.000186,
      "loss": 0.4302,
      "step": 312
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.26571759581565857,
      "learning_rate": 0.00018659999999999998,
      "loss": 0.3518,
      "step": 313
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.23198798298835754,
      "learning_rate": 0.0001872,
      "loss": 0.3443,
      "step": 314
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.2801920175552368,
      "learning_rate": 0.00018779999999999998,
      "loss": 0.3414,
      "step": 315
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.16064494848251343,
      "learning_rate": 0.00018839999999999997,
      "loss": 0.3418,
      "step": 316
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.34374213218688965,
      "learning_rate": 0.00018899999999999999,
      "loss": 0.3635,
      "step": 317
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.26433858275413513,
      "learning_rate": 0.00018959999999999997,
      "loss": 0.4407,
      "step": 318
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.40656784176826477,
      "learning_rate": 0.0001902,
      "loss": 0.3366,
      "step": 319
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.3449411690235138,
      "learning_rate": 0.00019079999999999998,
      "loss": 0.337,
      "step": 320
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.2536275386810303,
      "learning_rate": 0.0001914,
      "loss": 0.3441,
      "step": 321
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.28934991359710693,
      "learning_rate": 0.00019199999999999998,
      "loss": 0.3376,
      "step": 322
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.28276798129081726,
      "learning_rate": 0.0001926,
      "loss": 0.3569,
      "step": 323
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.3971540927886963,
      "learning_rate": 0.00019319999999999998,
      "loss": 0.4098,
      "step": 324
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.5473229289054871,
      "learning_rate": 0.0001938,
      "loss": 0.3406,
      "step": 325
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.36361709237098694,
      "learning_rate": 0.00019439999999999998,
      "loss": 0.3197,
      "step": 326
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.35563981533050537,
      "learning_rate": 0.000195,
      "loss": 0.3355,
      "step": 327
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.3679071366786957,
      "learning_rate": 0.00019559999999999998,
      "loss": 0.3351,
      "step": 328
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.2461666613817215,
      "learning_rate": 0.0001962,
      "loss": 0.3145,
      "step": 329
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.4335785210132599,
      "learning_rate": 0.00019679999999999999,
      "loss": 0.3834,
      "step": 330
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.23562943935394287,
      "learning_rate": 0.0001974,
      "loss": 0.359,
      "step": 331
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.25963151454925537,
      "learning_rate": 0.000198,
      "loss": 0.3103,
      "step": 332
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.17004214227199554,
      "learning_rate": 0.0001986,
      "loss": 0.3078,
      "step": 333
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.31410500407218933,
      "learning_rate": 0.0001992,
      "loss": 0.3279,
      "step": 334
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.18988707661628723,
      "learning_rate": 0.0001998,
      "loss": 0.3153,
      "step": 335
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.3162887394428253,
      "learning_rate": 0.0002004,
      "loss": 0.3523,
      "step": 336
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.2362556755542755,
      "learning_rate": 0.000201,
      "loss": 0.2923,
      "step": 337
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.2226056605577469,
      "learning_rate": 0.0002016,
      "loss": 0.2763,
      "step": 338
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.18942689895629883,
      "learning_rate": 0.0002022,
      "loss": 0.2864,
      "step": 339
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.18445442616939545,
      "learning_rate": 0.0002028,
      "loss": 0.2725,
      "step": 340
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.16619300842285156,
      "learning_rate": 0.00020339999999999998,
      "loss": 0.2787,
      "step": 341
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.2150353491306305,
      "learning_rate": 0.000204,
      "loss": 0.3169,
      "step": 342
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.1720004379749298,
      "learning_rate": 0.00020459999999999999,
      "loss": 0.2789,
      "step": 343
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.18419624865055084,
      "learning_rate": 0.0002052,
      "loss": 0.2567,
      "step": 344
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.1442434936761856,
      "learning_rate": 0.0002058,
      "loss": 0.2555,
      "step": 345
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.20136016607284546,
      "learning_rate": 0.00020639999999999998,
      "loss": 0.2687,
      "step": 346
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.19720318913459778,
      "learning_rate": 0.00020699999999999996,
      "loss": 0.2695,
      "step": 347
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.21785709261894226,
      "learning_rate": 0.00020759999999999998,
      "loss": 0.2665,
      "step": 348
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.258958637714386,
      "learning_rate": 0.00020819999999999996,
      "loss": 0.3171,
      "step": 349
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.18629953265190125,
      "learning_rate": 0.00020879999999999998,
      "loss": 0.2549,
      "step": 350
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.19203972816467285,
      "learning_rate": 0.00020939999999999997,
      "loss": 0.2583,
      "step": 351
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.1483580619096756,
      "learning_rate": 0.00020999999999999998,
      "loss": 0.2569,
      "step": 352
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.2105817049741745,
      "learning_rate": 0.00021059999999999997,
      "loss": 0.2571,
      "step": 353
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.18942958116531372,
      "learning_rate": 0.00021119999999999996,
      "loss": 0.2649,
      "step": 354
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.2832517623901367,
      "learning_rate": 0.00021179999999999997,
      "loss": 0.3367,
      "step": 355
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.2398645132780075,
      "learning_rate": 0.00021239999999999996,
      "loss": 0.2547,
      "step": 356
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.2130969762802124,
      "learning_rate": 0.00021299999999999997,
      "loss": 0.244,
      "step": 357
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.18089188635349274,
      "learning_rate": 0.00021359999999999996,
      "loss": 0.2409,
      "step": 358
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.19771325588226318,
      "learning_rate": 0.00021419999999999998,
      "loss": 0.244,
      "step": 359
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.22420358657836914,
      "learning_rate": 0.00021479999999999996,
      "loss": 0.2496,
      "step": 360
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.2420346587896347,
      "learning_rate": 0.00021539999999999998,
      "loss": 0.2755,
      "step": 361
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.21282508969306946,
      "learning_rate": 0.00021599999999999996,
      "loss": 0.2572,
      "step": 362
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.16155952215194702,
      "learning_rate": 0.00021659999999999998,
      "loss": 0.2462,
      "step": 363
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.20106732845306396,
      "learning_rate": 0.00021719999999999997,
      "loss": 0.2489,
      "step": 364
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.15768885612487793,
      "learning_rate": 0.00021779999999999998,
      "loss": 0.2322,
      "step": 365
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.1964852511882782,
      "learning_rate": 0.00021839999999999997,
      "loss": 0.2312,
      "step": 366
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.20234230160713196,
      "learning_rate": 0.00021899999999999998,
      "loss": 0.27,
      "step": 367
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.18610413372516632,
      "learning_rate": 0.00021959999999999997,
      "loss": 0.264,
      "step": 368
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.1503794938325882,
      "learning_rate": 0.00022019999999999999,
      "loss": 0.2441,
      "step": 369
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.1564638763666153,
      "learning_rate": 0.00022079999999999997,
      "loss": 0.2397,
      "step": 370
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.1418813318014145,
      "learning_rate": 0.0002214,
      "loss": 0.2311,
      "step": 371
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.13767391443252563,
      "learning_rate": 0.00022199999999999998,
      "loss": 0.2355,
      "step": 372
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.17412413656711578,
      "learning_rate": 0.0002226,
      "loss": 0.2443,
      "step": 373
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.19369201362133026,
      "learning_rate": 0.00022319999999999998,
      "loss": 0.269,
      "step": 374
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.15970537066459656,
      "learning_rate": 0.0002238,
      "loss": 0.2268,
      "step": 375
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.14835906028747559,
      "learning_rate": 0.00022439999999999998,
      "loss": 0.2322,
      "step": 376
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.16173814237117767,
      "learning_rate": 0.000225,
      "loss": 0.2218,
      "step": 377
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.16744820773601532,
      "learning_rate": 0.00022559999999999998,
      "loss": 0.2378,
      "step": 378
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.1666780412197113,
      "learning_rate": 0.00022619999999999997,
      "loss": 0.2364,
      "step": 379
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.24702821671962738,
      "learning_rate": 0.00022679999999999998,
      "loss": 0.2624,
      "step": 380
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.1768096536397934,
      "learning_rate": 0.00022739999999999997,
      "loss": 0.2365,
      "step": 381
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.16908930242061615,
      "learning_rate": 0.00022799999999999999,
      "loss": 0.22,
      "step": 382
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.2275974303483963,
      "learning_rate": 0.00022859999999999997,
      "loss": 0.2147,
      "step": 383
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.17168572545051575,
      "learning_rate": 0.0002292,
      "loss": 0.2287,
      "step": 384
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.24577546119689941,
      "learning_rate": 0.00022979999999999997,
      "loss": 0.2518,
      "step": 385
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.2721084654331207,
      "learning_rate": 0.0002304,
      "loss": 0.2791,
      "step": 386
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.16132959723472595,
      "learning_rate": 0.00023099999999999998,
      "loss": 0.2062,
      "step": 387
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.17369233071804047,
      "learning_rate": 0.0002316,
      "loss": 0.2196,
      "step": 388
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.14189645648002625,
      "learning_rate": 0.00023219999999999998,
      "loss": 0.2201,
      "step": 389
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.1626410335302353,
      "learning_rate": 0.0002328,
      "loss": 0.2223,
      "step": 390
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.18501843512058258,
      "learning_rate": 0.00023339999999999998,
      "loss": 0.2265,
      "step": 391
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.21089443564414978,
      "learning_rate": 0.000234,
      "loss": 0.2517,
      "step": 392
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.25956982374191284,
      "learning_rate": 0.00023459999999999998,
      "loss": 0.2559,
      "step": 393
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.13847751915454865,
      "learning_rate": 0.0002352,
      "loss": 0.2118,
      "step": 394
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.18859167397022247,
      "learning_rate": 0.00023579999999999999,
      "loss": 0.2209,
      "step": 395
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.15580683946609497,
      "learning_rate": 0.0002364,
      "loss": 0.2025,
      "step": 396
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.1728525608778,
      "learning_rate": 0.000237,
      "loss": 0.2298,
      "step": 397
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.22779783606529236,
      "learning_rate": 0.0002376,
      "loss": 0.2489,
      "step": 398
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.23927363753318787,
      "learning_rate": 0.0002382,
      "loss": 0.2566,
      "step": 399
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.19308319687843323,
      "learning_rate": 0.0002388,
      "loss": 0.2156,
      "step": 400
    },
    {
      "epoch": 3.57,
      "eval_cer": 0.059137905762704224,
      "eval_loss": 0.22499564290046692,
      "eval_runtime": 21.7981,
      "eval_samples_per_second": 121.203,
      "eval_steps_per_second": 1.927,
      "eval_wer": 0.2090241967473225,
      "step": 400
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.15958182513713837,
      "learning_rate": 0.0002394,
      "loss": 0.2196,
      "step": 401
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.17798563838005066,
      "learning_rate": 0.00023999999999999998,
      "loss": 0.2133,
      "step": 402
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.1666220873594284,
      "learning_rate": 0.0002406,
      "loss": 0.2002,
      "step": 403
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.1777678281068802,
      "learning_rate": 0.00024119999999999998,
      "loss": 0.2215,
      "step": 404
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.27465999126434326,
      "learning_rate": 0.0002418,
      "loss": 0.2546,
      "step": 405
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.2463161200284958,
      "learning_rate": 0.00024239999999999998,
      "loss": 0.2158,
      "step": 406
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.1615164428949356,
      "learning_rate": 0.000243,
      "loss": 0.2107,
      "step": 407
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.22111576795578003,
      "learning_rate": 0.00024359999999999999,
      "loss": 0.2044,
      "step": 408
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.14725008606910706,
      "learning_rate": 0.00024419999999999997,
      "loss": 0.2099,
      "step": 409
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.20693030953407288,
      "learning_rate": 0.0002448,
      "loss": 0.2324,
      "step": 410
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.2718674838542938,
      "learning_rate": 0.00024539999999999995,
      "loss": 0.2529,
      "step": 411
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.2779533565044403,
      "learning_rate": 0.00024599999999999996,
      "loss": 0.2105,
      "step": 412
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.1879548728466034,
      "learning_rate": 0.0002466,
      "loss": 0.2089,
      "step": 413
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.22634433209896088,
      "learning_rate": 0.0002472,
      "loss": 0.202,
      "step": 414
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.17272689938545227,
      "learning_rate": 0.00024779999999999995,
      "loss": 0.1992,
      "step": 415
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.18425069749355316,
      "learning_rate": 0.00024839999999999997,
      "loss": 0.2011,
      "step": 416
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.2612092196941376,
      "learning_rate": 0.000249,
      "loss": 0.238,
      "step": 417
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.25745177268981934,
      "learning_rate": 0.00024959999999999994,
      "loss": 0.2373,
      "step": 418
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.19050335884094238,
      "learning_rate": 0.00025019999999999996,
      "loss": 0.2077,
      "step": 419
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.1341092884540558,
      "learning_rate": 0.00025079999999999997,
      "loss": 0.195,
      "step": 420
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.20959509909152985,
      "learning_rate": 0.0002514,
      "loss": 0.2142,
      "step": 421
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.13756853342056274,
      "learning_rate": 0.00025199999999999995,
      "loss": 0.2017,
      "step": 422
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.18742893636226654,
      "learning_rate": 0.00025259999999999996,
      "loss": 0.2258,
      "step": 423
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.25547361373901367,
      "learning_rate": 0.0002532,
      "loss": 0.2334,
      "step": 424
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.19071930646896362,
      "learning_rate": 0.0002538,
      "loss": 0.2154,
      "step": 425
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.166627898812294,
      "learning_rate": 0.00025439999999999995,
      "loss": 0.1889,
      "step": 426
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.13609127700328827,
      "learning_rate": 0.00025499999999999996,
      "loss": 0.1815,
      "step": 427
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.17250128090381622,
      "learning_rate": 0.0002556,
      "loss": 0.1946,
      "step": 428
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.14866022765636444,
      "learning_rate": 0.0002562,
      "loss": 0.2075,
      "step": 429
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.2631528377532959,
      "learning_rate": 0.00025679999999999995,
      "loss": 0.2687,
      "step": 430
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.17119698226451874,
      "learning_rate": 0.00025739999999999997,
      "loss": 0.1888,
      "step": 431
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.14718417823314667,
      "learning_rate": 0.000258,
      "loss": 0.1934,
      "step": 432
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.126860111951828,
      "learning_rate": 0.0002586,
      "loss": 0.186,
      "step": 433
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.14548039436340332,
      "learning_rate": 0.00025919999999999996,
      "loss": 0.1956,
      "step": 434
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.15526461601257324,
      "learning_rate": 0.00025979999999999997,
      "loss": 0.1991,
      "step": 435
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.1961231231689453,
      "learning_rate": 0.0002604,
      "loss": 0.2342,
      "step": 436
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.12892864644527435,
      "learning_rate": 0.000261,
      "loss": 0.1939,
      "step": 437
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.1452789306640625,
      "learning_rate": 0.00026159999999999996,
      "loss": 0.189,
      "step": 438
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.12380103021860123,
      "learning_rate": 0.0002622,
      "loss": 0.1954,
      "step": 439
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.1424674391746521,
      "learning_rate": 0.0002628,
      "loss": 0.2032,
      "step": 440
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.12863987684249878,
      "learning_rate": 0.00026339999999999995,
      "loss": 0.1796,
      "step": 441
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.19804953038692474,
      "learning_rate": 0.00026399999999999997,
      "loss": 0.2079,
      "step": 442
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.20673587918281555,
      "learning_rate": 0.0002646,
      "loss": 0.228,
      "step": 443
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.1997382789850235,
      "learning_rate": 0.0002652,
      "loss": 0.1879,
      "step": 444
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.13489511609077454,
      "learning_rate": 0.00026579999999999996,
      "loss": 0.1888,
      "step": 445
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.20545919239521027,
      "learning_rate": 0.00026639999999999997,
      "loss": 0.1823,
      "step": 446
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.1583697646856308,
      "learning_rate": 0.000267,
      "loss": 0.1908,
      "step": 447
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.24280981719493866,
      "learning_rate": 0.0002676,
      "loss": 0.2184,
      "step": 448
    },
    {
      "epoch": 4.01,
      "grad_norm": 0.13582612574100494,
      "learning_rate": 0.00026819999999999996,
      "loss": 0.1597,
      "step": 449
    },
    {
      "epoch": 4.02,
      "grad_norm": 0.17924658954143524,
      "learning_rate": 0.0002688,
      "loss": 0.1655,
      "step": 450
    },
    {
      "epoch": 4.03,
      "grad_norm": 0.14039453864097595,
      "learning_rate": 0.0002694,
      "loss": 0.1585,
      "step": 451
    },
    {
      "epoch": 4.04,
      "grad_norm": 0.16404248774051666,
      "learning_rate": 0.00027,
      "loss": 0.156,
      "step": 452
    },
    {
      "epoch": 4.04,
      "grad_norm": 0.14572083950042725,
      "learning_rate": 0.00027059999999999996,
      "loss": 0.1527,
      "step": 453
    },
    {
      "epoch": 4.05,
      "grad_norm": 0.21694505214691162,
      "learning_rate": 0.0002712,
      "loss": 0.1679,
      "step": 454
    },
    {
      "epoch": 4.06,
      "grad_norm": 0.20453929901123047,
      "learning_rate": 0.0002718,
      "loss": 0.1739,
      "step": 455
    },
    {
      "epoch": 4.07,
      "grad_norm": 0.16913868486881256,
      "learning_rate": 0.0002724,
      "loss": 0.1656,
      "step": 456
    },
    {
      "epoch": 4.08,
      "grad_norm": 0.14071783423423767,
      "learning_rate": 0.00027299999999999997,
      "loss": 0.1531,
      "step": 457
    },
    {
      "epoch": 4.09,
      "grad_norm": 0.1545068472623825,
      "learning_rate": 0.0002736,
      "loss": 0.157,
      "step": 458
    },
    {
      "epoch": 4.1,
      "grad_norm": 0.1283985674381256,
      "learning_rate": 0.0002742,
      "loss": 0.1633,
      "step": 459
    },
    {
      "epoch": 4.11,
      "grad_norm": 0.17630359530448914,
      "learning_rate": 0.0002748,
      "loss": 0.1709,
      "step": 460
    },
    {
      "epoch": 4.12,
      "grad_norm": 0.19642595946788788,
      "learning_rate": 0.00027539999999999997,
      "loss": 0.1938,
      "step": 461
    },
    {
      "epoch": 4.12,
      "grad_norm": 0.1480659544467926,
      "learning_rate": 0.000276,
      "loss": 0.1502,
      "step": 462
    },
    {
      "epoch": 4.13,
      "grad_norm": 0.14618328213691711,
      "learning_rate": 0.0002766,
      "loss": 0.1557,
      "step": 463
    },
    {
      "epoch": 4.14,
      "grad_norm": 0.1548188030719757,
      "learning_rate": 0.0002772,
      "loss": 0.1548,
      "step": 464
    },
    {
      "epoch": 4.15,
      "grad_norm": 0.14934171736240387,
      "learning_rate": 0.0002778,
      "loss": 0.1491,
      "step": 465
    },
    {
      "epoch": 4.16,
      "grad_norm": 0.16205880045890808,
      "learning_rate": 0.0002784,
      "loss": 0.1549,
      "step": 466
    },
    {
      "epoch": 4.17,
      "grad_norm": 0.19758005440235138,
      "learning_rate": 0.000279,
      "loss": 0.1907,
      "step": 467
    },
    {
      "epoch": 4.18,
      "grad_norm": 0.14266394078731537,
      "learning_rate": 0.00027959999999999997,
      "loss": 0.1456,
      "step": 468
    },
    {
      "epoch": 4.19,
      "grad_norm": 0.18047606945037842,
      "learning_rate": 0.0002802,
      "loss": 0.1656,
      "step": 469
    },
    {
      "epoch": 4.2,
      "grad_norm": 0.1618647575378418,
      "learning_rate": 0.0002808,
      "loss": 0.164,
      "step": 470
    },
    {
      "epoch": 4.21,
      "grad_norm": 0.1894494742155075,
      "learning_rate": 0.00028139999999999996,
      "loss": 0.1477,
      "step": 471
    },
    {
      "epoch": 4.21,
      "grad_norm": 0.15157738327980042,
      "learning_rate": 0.00028199999999999997,
      "loss": 0.142,
      "step": 472
    },
    {
      "epoch": 4.22,
      "grad_norm": 0.20689848065376282,
      "learning_rate": 0.0002826,
      "loss": 0.1747,
      "step": 473
    },
    {
      "epoch": 4.23,
      "grad_norm": 0.15236985683441162,
      "learning_rate": 0.00028319999999999994,
      "loss": 0.1523,
      "step": 474
    },
    {
      "epoch": 4.24,
      "grad_norm": 0.1496431976556778,
      "learning_rate": 0.00028379999999999996,
      "loss": 0.1457,
      "step": 475
    },
    {
      "epoch": 4.25,
      "grad_norm": 0.12969504296779633,
      "learning_rate": 0.0002844,
      "loss": 0.1522,
      "step": 476
    },
    {
      "epoch": 4.26,
      "grad_norm": 0.15846145153045654,
      "learning_rate": 0.000285,
      "loss": 0.1397,
      "step": 477
    },
    {
      "epoch": 4.27,
      "grad_norm": 0.13522788882255554,
      "learning_rate": 0.00028559999999999995,
      "loss": 0.1503,
      "step": 478
    },
    {
      "epoch": 4.28,
      "grad_norm": 0.17433424293994904,
      "learning_rate": 0.00028619999999999996,
      "loss": 0.1656,
      "step": 479
    },
    {
      "epoch": 4.29,
      "grad_norm": 0.19288386404514313,
      "learning_rate": 0.0002868,
      "loss": 0.1653,
      "step": 480
    },
    {
      "epoch": 4.29,
      "grad_norm": 0.16040611267089844,
      "learning_rate": 0.00028739999999999994,
      "loss": 0.1561,
      "step": 481
    },
    {
      "epoch": 4.3,
      "grad_norm": 0.14547991752624512,
      "learning_rate": 0.00028799999999999995,
      "loss": 0.1482,
      "step": 482
    },
    {
      "epoch": 4.31,
      "grad_norm": 0.1573590487241745,
      "learning_rate": 0.00028859999999999997,
      "loss": 0.1459,
      "step": 483
    },
    {
      "epoch": 4.32,
      "grad_norm": 0.13309909403324127,
      "learning_rate": 0.0002892,
      "loss": 0.15,
      "step": 484
    },
    {
      "epoch": 4.33,
      "grad_norm": 0.16878025233745575,
      "learning_rate": 0.00028979999999999994,
      "loss": 0.1532,
      "step": 485
    },
    {
      "epoch": 4.34,
      "grad_norm": 0.19783174991607666,
      "learning_rate": 0.00029039999999999996,
      "loss": 0.1755,
      "step": 486
    },
    {
      "epoch": 4.35,
      "grad_norm": 0.24179938435554504,
      "learning_rate": 0.00029099999999999997,
      "loss": 0.1467,
      "step": 487
    },
    {
      "epoch": 4.36,
      "grad_norm": 0.13137803971767426,
      "learning_rate": 0.0002916,
      "loss": 0.149,
      "step": 488
    },
    {
      "epoch": 4.37,
      "grad_norm": 0.19716697931289673,
      "learning_rate": 0.00029219999999999995,
      "loss": 0.1416,
      "step": 489
    },
    {
      "epoch": 4.38,
      "grad_norm": 0.13461431860923767,
      "learning_rate": 0.00029279999999999996,
      "loss": 0.1491,
      "step": 490
    },
    {
      "epoch": 4.38,
      "grad_norm": 0.20108233392238617,
      "learning_rate": 0.0002934,
      "loss": 0.1454,
      "step": 491
    },
    {
      "epoch": 4.39,
      "grad_norm": 0.22257226705551147,
      "learning_rate": 0.000294,
      "loss": 0.1836,
      "step": 492
    },
    {
      "epoch": 4.4,
      "grad_norm": 0.13674646615982056,
      "learning_rate": 0.00029459999999999995,
      "loss": 0.1431,
      "step": 493
    },
    {
      "epoch": 4.41,
      "grad_norm": 0.18464460968971252,
      "learning_rate": 0.00029519999999999997,
      "loss": 0.1584,
      "step": 494
    },
    {
      "epoch": 4.42,
      "grad_norm": 0.12371817231178284,
      "learning_rate": 0.0002958,
      "loss": 0.148,
      "step": 495
    },
    {
      "epoch": 4.43,
      "grad_norm": 0.18389379978179932,
      "learning_rate": 0.0002964,
      "loss": 0.1516,
      "step": 496
    },
    {
      "epoch": 4.44,
      "grad_norm": 0.16786426305770874,
      "learning_rate": 0.00029699999999999996,
      "loss": 0.1612,
      "step": 497
    },
    {
      "epoch": 4.45,
      "grad_norm": 0.22654010355472565,
      "learning_rate": 0.00029759999999999997,
      "loss": 0.1589,
      "step": 498
    },
    {
      "epoch": 4.46,
      "grad_norm": 0.18563547730445862,
      "learning_rate": 0.0002982,
      "loss": 0.1576,
      "step": 499
    },
    {
      "epoch": 4.46,
      "grad_norm": 0.13945861160755157,
      "learning_rate": 0.0002988,
      "loss": 0.1517,
      "step": 500
    },
    {
      "epoch": 4.46,
      "eval_cer": 0.047389250318521606,
      "eval_loss": 0.18336819112300873,
      "eval_runtime": 21.8812,
      "eval_samples_per_second": 120.743,
      "eval_steps_per_second": 1.919,
      "eval_wer": 0.16951606505355019,
      "step": 500
    },
    {
      "epoch": 4.47,
      "grad_norm": 0.1721625030040741,
      "learning_rate": 0.00029939999999999996,
      "loss": 0.139,
      "step": 501
    },
    {
      "epoch": 4.48,
      "grad_norm": 0.14510883390903473,
      "learning_rate": 0.0003,
      "loss": 0.1509,
      "step": 502
    },
    {
      "epoch": 4.49,
      "grad_norm": 0.14989228546619415,
      "learning_rate": 0.0002999411764705882,
      "loss": 0.1467,
      "step": 503
    },
    {
      "epoch": 4.5,
      "grad_norm": 0.17836323380470276,
      "learning_rate": 0.00029988235294117646,
      "loss": 0.153,
      "step": 504
    },
    {
      "epoch": 4.51,
      "grad_norm": 0.19929830729961395,
      "learning_rate": 0.00029982352941176465,
      "loss": 0.1533,
      "step": 505
    },
    {
      "epoch": 4.52,
      "grad_norm": 0.14675268530845642,
      "learning_rate": 0.00029976470588235295,
      "loss": 0.14,
      "step": 506
    },
    {
      "epoch": 4.53,
      "grad_norm": 0.18037748336791992,
      "learning_rate": 0.00029970588235294114,
      "loss": 0.1436,
      "step": 507
    },
    {
      "epoch": 4.54,
      "grad_norm": 0.1641611009836197,
      "learning_rate": 0.0002996470588235294,
      "loss": 0.1404,
      "step": 508
    },
    {
      "epoch": 4.54,
      "grad_norm": 0.1628531962633133,
      "learning_rate": 0.0002995882352941176,
      "loss": 0.1502,
      "step": 509
    },
    {
      "epoch": 4.55,
      "grad_norm": 0.19253024458885193,
      "learning_rate": 0.00029952941176470587,
      "loss": 0.1695,
      "step": 510
    },
    {
      "epoch": 4.56,
      "grad_norm": 0.21144042909145355,
      "learning_rate": 0.0002994705882352941,
      "loss": 0.1739,
      "step": 511
    },
    {
      "epoch": 4.57,
      "grad_norm": 0.11775948852300644,
      "learning_rate": 0.00029941176470588235,
      "loss": 0.1422,
      "step": 512
    },
    {
      "epoch": 4.58,
      "grad_norm": 0.1344527155160904,
      "learning_rate": 0.00029935294117647054,
      "loss": 0.1527,
      "step": 513
    },
    {
      "epoch": 4.59,
      "grad_norm": 0.12983818352222443,
      "learning_rate": 0.0002992941176470588,
      "loss": 0.1489,
      "step": 514
    },
    {
      "epoch": 4.6,
      "grad_norm": 0.15714521706104279,
      "learning_rate": 0.00029923529411764703,
      "loss": 0.1448,
      "step": 515
    },
    {
      "epoch": 4.61,
      "grad_norm": 0.13811440765857697,
      "learning_rate": 0.0002991764705882353,
      "loss": 0.1434,
      "step": 516
    },
    {
      "epoch": 4.62,
      "grad_norm": 0.21657413244247437,
      "learning_rate": 0.0002991176470588235,
      "loss": 0.1848,
      "step": 517
    },
    {
      "epoch": 4.62,
      "grad_norm": 0.14271414279937744,
      "learning_rate": 0.0002990588235294117,
      "loss": 0.1323,
      "step": 518
    },
    {
      "epoch": 4.63,
      "grad_norm": 0.13417980074882507,
      "learning_rate": 0.000299,
      "loss": 0.1459,
      "step": 519
    },
    {
      "epoch": 4.64,
      "grad_norm": 0.13681019842624664,
      "learning_rate": 0.0002989411764705882,
      "loss": 0.1434,
      "step": 520
    },
    {
      "epoch": 4.65,
      "grad_norm": 0.1443687230348587,
      "learning_rate": 0.00029888235294117644,
      "loss": 0.1456,
      "step": 521
    },
    {
      "epoch": 4.66,
      "grad_norm": 0.13649730384349823,
      "learning_rate": 0.0002988235294117647,
      "loss": 0.1482,
      "step": 522
    },
    {
      "epoch": 4.67,
      "grad_norm": 0.2059570699930191,
      "learning_rate": 0.0002987647058823529,
      "loss": 0.1747,
      "step": 523
    },
    {
      "epoch": 4.68,
      "grad_norm": 0.13787458837032318,
      "learning_rate": 0.00029870588235294117,
      "loss": 0.1427,
      "step": 524
    },
    {
      "epoch": 4.69,
      "grad_norm": 0.13346923887729645,
      "learning_rate": 0.0002986470588235294,
      "loss": 0.141,
      "step": 525
    },
    {
      "epoch": 4.7,
      "grad_norm": 0.11142217367887497,
      "learning_rate": 0.0002985882352941176,
      "loss": 0.1331,
      "step": 526
    },
    {
      "epoch": 4.71,
      "grad_norm": 0.14373567700386047,
      "learning_rate": 0.00029852941176470584,
      "loss": 0.1341,
      "step": 527
    },
    {
      "epoch": 4.71,
      "grad_norm": 0.11874829977750778,
      "learning_rate": 0.0002984705882352941,
      "loss": 0.1452,
      "step": 528
    },
    {
      "epoch": 4.72,
      "grad_norm": 0.18138618767261505,
      "learning_rate": 0.00029841176470588233,
      "loss": 0.1531,
      "step": 529
    },
    {
      "epoch": 4.73,
      "grad_norm": 0.15628069639205933,
      "learning_rate": 0.0002983529411764706,
      "loss": 0.1584,
      "step": 530
    },
    {
      "epoch": 4.74,
      "grad_norm": 0.1584649682044983,
      "learning_rate": 0.00029829411764705876,
      "loss": 0.1381,
      "step": 531
    },
    {
      "epoch": 4.75,
      "grad_norm": 0.11128696799278259,
      "learning_rate": 0.00029823529411764706,
      "loss": 0.1366,
      "step": 532
    },
    {
      "epoch": 4.76,
      "grad_norm": 0.13757440447807312,
      "learning_rate": 0.00029817647058823525,
      "loss": 0.1492,
      "step": 533
    },
    {
      "epoch": 4.77,
      "grad_norm": 0.14631593227386475,
      "learning_rate": 0.0002981176470588235,
      "loss": 0.1423,
      "step": 534
    },
    {
      "epoch": 4.78,
      "grad_norm": 0.1369020640850067,
      "learning_rate": 0.00029805882352941174,
      "loss": 0.1432,
      "step": 535
    },
    {
      "epoch": 4.79,
      "grad_norm": 0.24188145995140076,
      "learning_rate": 0.000298,
      "loss": 0.1641,
      "step": 536
    },
    {
      "epoch": 4.79,
      "grad_norm": 0.13066911697387695,
      "learning_rate": 0.0002979411764705882,
      "loss": 0.1384,
      "step": 537
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.13668803870677948,
      "learning_rate": 0.0002978823529411764,
      "loss": 0.1372,
      "step": 538
    },
    {
      "epoch": 4.81,
      "grad_norm": 0.11504946649074554,
      "learning_rate": 0.0002978235294117647,
      "loss": 0.1339,
      "step": 539
    },
    {
      "epoch": 4.82,
      "grad_norm": 0.16001489758491516,
      "learning_rate": 0.0002977647058823529,
      "loss": 0.1445,
      "step": 540
    },
    {
      "epoch": 4.83,
      "grad_norm": 0.12840960919857025,
      "learning_rate": 0.00029770588235294114,
      "loss": 0.1342,
      "step": 541
    },
    {
      "epoch": 4.84,
      "grad_norm": 0.19975897669792175,
      "learning_rate": 0.0002976470588235294,
      "loss": 0.1696,
      "step": 542
    },
    {
      "epoch": 4.85,
      "grad_norm": 0.13238808512687683,
      "learning_rate": 0.00029758823529411763,
      "loss": 0.1426,
      "step": 543
    },
    {
      "epoch": 4.86,
      "grad_norm": 0.1489226073026657,
      "learning_rate": 0.00029752941176470587,
      "loss": 0.1349,
      "step": 544
    },
    {
      "epoch": 4.87,
      "grad_norm": 0.10776083916425705,
      "learning_rate": 0.0002974705882352941,
      "loss": 0.1351,
      "step": 545
    },
    {
      "epoch": 4.88,
      "grad_norm": 0.12893898785114288,
      "learning_rate": 0.0002974117647058823,
      "loss": 0.1235,
      "step": 546
    },
    {
      "epoch": 4.88,
      "grad_norm": 0.14186742901802063,
      "learning_rate": 0.00029735294117647055,
      "loss": 0.1424,
      "step": 547
    },
    {
      "epoch": 4.89,
      "grad_norm": 0.18122699856758118,
      "learning_rate": 0.0002972941176470588,
      "loss": 0.1575,
      "step": 548
    },
    {
      "epoch": 4.9,
      "grad_norm": 0.14512194693088531,
      "learning_rate": 0.00029723529411764704,
      "loss": 0.1374,
      "step": 549
    },
    {
      "epoch": 4.91,
      "grad_norm": 0.12667463719844818,
      "learning_rate": 0.0002971764705882353,
      "loss": 0.1422,
      "step": 550
    },
    {
      "epoch": 4.92,
      "grad_norm": 0.10744798183441162,
      "learning_rate": 0.00029711764705882347,
      "loss": 0.1231,
      "step": 551
    },
    {
      "epoch": 4.93,
      "grad_norm": 0.12515398859977722,
      "learning_rate": 0.00029705882352941177,
      "loss": 0.1458,
      "step": 552
    },
    {
      "epoch": 4.94,
      "grad_norm": 0.13184143602848053,
      "learning_rate": 0.00029699999999999996,
      "loss": 0.1361,
      "step": 553
    },
    {
      "epoch": 4.95,
      "grad_norm": 0.17081902921199799,
      "learning_rate": 0.0002969411764705882,
      "loss": 0.1616,
      "step": 554
    },
    {
      "epoch": 4.96,
      "grad_norm": 0.17962642014026642,
      "learning_rate": 0.00029688235294117644,
      "loss": 0.1613,
      "step": 555
    },
    {
      "epoch": 4.96,
      "grad_norm": 0.14994260668754578,
      "learning_rate": 0.0002968235294117647,
      "loss": 0.1378,
      "step": 556
    },
    {
      "epoch": 4.97,
      "grad_norm": 0.13085141777992249,
      "learning_rate": 0.00029676470588235293,
      "loss": 0.1262,
      "step": 557
    },
    {
      "epoch": 4.98,
      "grad_norm": 0.11866625398397446,
      "learning_rate": 0.00029670588235294117,
      "loss": 0.1397,
      "step": 558
    },
    {
      "epoch": 4.99,
      "grad_norm": 0.14264865219593048,
      "learning_rate": 0.00029664705882352936,
      "loss": 0.1436,
      "step": 559
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.16405989229679108,
      "learning_rate": 0.0002965882352941176,
      "loss": 0.1387,
      "step": 560
    },
    {
      "epoch": 5.01,
      "grad_norm": 0.11050975322723389,
      "learning_rate": 0.00029652941176470585,
      "loss": 0.1179,
      "step": 561
    },
    {
      "epoch": 5.02,
      "grad_norm": 0.10286074876785278,
      "learning_rate": 0.0002964705882352941,
      "loss": 0.1072,
      "step": 562
    },
    {
      "epoch": 5.03,
      "grad_norm": 0.11145926266908646,
      "learning_rate": 0.00029641176470588234,
      "loss": 0.1139,
      "step": 563
    },
    {
      "epoch": 5.04,
      "grad_norm": 0.1139025166630745,
      "learning_rate": 0.0002963529411764705,
      "loss": 0.1105,
      "step": 564
    },
    {
      "epoch": 5.04,
      "grad_norm": 0.11843300610780716,
      "learning_rate": 0.0002962941176470588,
      "loss": 0.1057,
      "step": 565
    },
    {
      "epoch": 5.05,
      "grad_norm": 0.13360877335071564,
      "learning_rate": 0.000296235294117647,
      "loss": 0.1109,
      "step": 566
    },
    {
      "epoch": 5.06,
      "grad_norm": 0.16410833597183228,
      "learning_rate": 0.0002961764705882353,
      "loss": 0.1193,
      "step": 567
    },
    {
      "epoch": 5.07,
      "grad_norm": 0.1047048345208168,
      "learning_rate": 0.0002961176470588235,
      "loss": 0.1105,
      "step": 568
    },
    {
      "epoch": 5.08,
      "grad_norm": 0.11158055067062378,
      "learning_rate": 0.00029605882352941174,
      "loss": 0.1051,
      "step": 569
    },
    {
      "epoch": 5.09,
      "grad_norm": 0.10403604805469513,
      "learning_rate": 0.000296,
      "loss": 0.1106,
      "step": 570
    },
    {
      "epoch": 5.1,
      "grad_norm": 0.11520044505596161,
      "learning_rate": 0.00029594117647058823,
      "loss": 0.11,
      "step": 571
    },
    {
      "epoch": 5.11,
      "grad_norm": 0.1380784809589386,
      "learning_rate": 0.00029588235294117647,
      "loss": 0.1066,
      "step": 572
    },
    {
      "epoch": 5.12,
      "grad_norm": 0.1517052948474884,
      "learning_rate": 0.00029582352941176466,
      "loss": 0.1411,
      "step": 573
    },
    {
      "epoch": 5.12,
      "grad_norm": 0.1428508311510086,
      "learning_rate": 0.0002957647058823529,
      "loss": 0.1069,
      "step": 574
    },
    {
      "epoch": 5.13,
      "grad_norm": 0.11712061613798141,
      "learning_rate": 0.00029570588235294115,
      "loss": 0.1083,
      "step": 575
    },
    {
      "epoch": 5.14,
      "grad_norm": 0.12171980738639832,
      "learning_rate": 0.0002956470588235294,
      "loss": 0.1052,
      "step": 576
    },
    {
      "epoch": 5.15,
      "grad_norm": 0.14088501036167145,
      "learning_rate": 0.00029558823529411763,
      "loss": 0.1142,
      "step": 577
    },
    {
      "epoch": 5.16,
      "grad_norm": 0.11284144222736359,
      "learning_rate": 0.0002955294117647059,
      "loss": 0.0995,
      "step": 578
    },
    {
      "epoch": 5.17,
      "grad_norm": 0.1761561781167984,
      "learning_rate": 0.00029547058823529407,
      "loss": 0.1189,
      "step": 579
    },
    {
      "epoch": 5.18,
      "grad_norm": 0.1454172134399414,
      "learning_rate": 0.0002954117647058823,
      "loss": 0.109,
      "step": 580
    },
    {
      "epoch": 5.19,
      "grad_norm": 0.10470675677061081,
      "learning_rate": 0.00029535294117647055,
      "loss": 0.1041,
      "step": 581
    },
    {
      "epoch": 5.2,
      "grad_norm": 0.13161127269268036,
      "learning_rate": 0.0002952941176470588,
      "loss": 0.1081,
      "step": 582
    },
    {
      "epoch": 5.21,
      "grad_norm": 0.1211889311671257,
      "learning_rate": 0.00029523529411764704,
      "loss": 0.1009,
      "step": 583
    },
    {
      "epoch": 5.21,
      "grad_norm": 0.10698932409286499,
      "learning_rate": 0.0002951764705882353,
      "loss": 0.0989,
      "step": 584
    },
    {
      "epoch": 5.22,
      "grad_norm": 0.1922842413187027,
      "learning_rate": 0.00029511764705882353,
      "loss": 0.1321,
      "step": 585
    },
    {
      "epoch": 5.23,
      "grad_norm": 0.14545674622058868,
      "learning_rate": 0.0002950588235294117,
      "loss": 0.1151,
      "step": 586
    },
    {
      "epoch": 5.24,
      "grad_norm": 0.10841576009988785,
      "learning_rate": 0.00029499999999999996,
      "loss": 0.0999,
      "step": 587
    },
    {
      "epoch": 5.25,
      "grad_norm": 0.12169073522090912,
      "learning_rate": 0.0002949411764705882,
      "loss": 0.1161,
      "step": 588
    },
    {
      "epoch": 5.26,
      "grad_norm": 0.1342030167579651,
      "learning_rate": 0.00029488235294117645,
      "loss": 0.1053,
      "step": 589
    },
    {
      "epoch": 5.27,
      "grad_norm": 0.10973326116800308,
      "learning_rate": 0.0002948235294117647,
      "loss": 0.1066,
      "step": 590
    },
    {
      "epoch": 5.28,
      "grad_norm": 0.1585254669189453,
      "learning_rate": 0.00029476470588235293,
      "loss": 0.1154,
      "step": 591
    },
    {
      "epoch": 5.29,
      "grad_norm": 0.14881926774978638,
      "learning_rate": 0.0002947058823529411,
      "loss": 0.1177,
      "step": 592
    },
    {
      "epoch": 5.29,
      "grad_norm": 0.1331123262643814,
      "learning_rate": 0.00029464705882352937,
      "loss": 0.1101,
      "step": 593
    },
    {
      "epoch": 5.3,
      "grad_norm": 0.11872667074203491,
      "learning_rate": 0.0002945882352941176,
      "loss": 0.0973,
      "step": 594
    },
    {
      "epoch": 5.31,
      "grad_norm": 0.11906508356332779,
      "learning_rate": 0.00029452941176470585,
      "loss": 0.1039,
      "step": 595
    },
    {
      "epoch": 5.32,
      "grad_norm": 0.1454726755619049,
      "learning_rate": 0.0002944705882352941,
      "loss": 0.1054,
      "step": 596
    },
    {
      "epoch": 5.33,
      "grad_norm": 0.13531070947647095,
      "learning_rate": 0.00029441176470588234,
      "loss": 0.1064,
      "step": 597
    },
    {
      "epoch": 5.34,
      "grad_norm": 0.19511525332927704,
      "learning_rate": 0.0002943529411764706,
      "loss": 0.1233,
      "step": 598
    },
    {
      "epoch": 5.35,
      "grad_norm": 0.11403120309114456,
      "learning_rate": 0.0002942941176470588,
      "loss": 0.1138,
      "step": 599
    },
    {
      "epoch": 5.36,
      "grad_norm": 0.11977498233318329,
      "learning_rate": 0.00029423529411764707,
      "loss": 0.1059,
      "step": 600
    },
    {
      "epoch": 5.36,
      "eval_cer": 0.042773975563097345,
      "eval_loss": 0.1667700558900833,
      "eval_runtime": 21.8271,
      "eval_samples_per_second": 121.042,
      "eval_steps_per_second": 1.924,
      "eval_wer": 0.15021816739389132,
      "step": 600
    },
    {
      "epoch": 5.37,
      "grad_norm": 0.11358664929866791,
      "learning_rate": 0.00029417647058823526,
      "loss": 0.1036,
      "step": 601
    },
    {
      "epoch": 5.38,
      "grad_norm": 0.11147243529558182,
      "learning_rate": 0.0002941176470588235,
      "loss": 0.1082,
      "step": 602
    },
    {
      "epoch": 5.38,
      "grad_norm": 0.1350492686033249,
      "learning_rate": 0.00029405882352941175,
      "loss": 0.1051,
      "step": 603
    },
    {
      "epoch": 5.39,
      "grad_norm": 0.1623750776052475,
      "learning_rate": 0.000294,
      "loss": 0.1249,
      "step": 604
    },
    {
      "epoch": 5.4,
      "grad_norm": 0.13666796684265137,
      "learning_rate": 0.00029394117647058823,
      "loss": 0.1104,
      "step": 605
    },
    {
      "epoch": 5.41,
      "grad_norm": 0.10063891112804413,
      "learning_rate": 0.0002938823529411764,
      "loss": 0.111,
      "step": 606
    },
    {
      "epoch": 5.42,
      "grad_norm": 0.11101557314395905,
      "learning_rate": 0.00029382352941176467,
      "loss": 0.1014,
      "step": 607
    },
    {
      "epoch": 5.43,
      "grad_norm": 0.11853890120983124,
      "learning_rate": 0.0002937647058823529,
      "loss": 0.1067,
      "step": 608
    },
    {
      "epoch": 5.44,
      "grad_norm": 0.12706832587718964,
      "learning_rate": 0.00029370588235294115,
      "loss": 0.1073,
      "step": 609
    },
    {
      "epoch": 5.45,
      "grad_norm": 0.16450826823711395,
      "learning_rate": 0.0002936470588235294,
      "loss": 0.1272,
      "step": 610
    },
    {
      "epoch": 5.46,
      "grad_norm": 0.10771495848894119,
      "learning_rate": 0.00029358823529411764,
      "loss": 0.111,
      "step": 611
    },
    {
      "epoch": 5.46,
      "grad_norm": 0.10839465260505676,
      "learning_rate": 0.00029352941176470583,
      "loss": 0.099,
      "step": 612
    },
    {
      "epoch": 5.47,
      "grad_norm": 0.12419650703668594,
      "learning_rate": 0.00029347058823529413,
      "loss": 0.1043,
      "step": 613
    },
    {
      "epoch": 5.48,
      "grad_norm": 0.10811444371938705,
      "learning_rate": 0.0002934117647058823,
      "loss": 0.1082,
      "step": 614
    },
    {
      "epoch": 5.49,
      "grad_norm": 0.10220162570476532,
      "learning_rate": 0.00029335294117647056,
      "loss": 0.099,
      "step": 615
    },
    {
      "epoch": 5.5,
      "grad_norm": 0.14317558705806732,
      "learning_rate": 0.0002932941176470588,
      "loss": 0.1037,
      "step": 616
    },
    {
      "epoch": 5.51,
      "grad_norm": 0.1705309897661209,
      "learning_rate": 0.00029323529411764705,
      "loss": 0.127,
      "step": 617
    },
    {
      "epoch": 5.52,
      "grad_norm": 0.0989428237080574,
      "learning_rate": 0.0002931764705882353,
      "loss": 0.1117,
      "step": 618
    },
    {
      "epoch": 5.53,
      "grad_norm": 0.12371882796287537,
      "learning_rate": 0.0002931176470588235,
      "loss": 0.1131,
      "step": 619
    },
    {
      "epoch": 5.54,
      "grad_norm": 0.11418908834457397,
      "learning_rate": 0.0002930588235294117,
      "loss": 0.1058,
      "step": 620
    },
    {
      "epoch": 5.54,
      "grad_norm": 0.1010790467262268,
      "learning_rate": 0.00029299999999999997,
      "loss": 0.102,
      "step": 621
    },
    {
      "epoch": 5.55,
      "grad_norm": 0.14377234876155853,
      "learning_rate": 0.0002929411764705882,
      "loss": 0.1122,
      "step": 622
    },
    {
      "epoch": 5.56,
      "grad_norm": 0.1575341671705246,
      "learning_rate": 0.00029288235294117645,
      "loss": 0.1126,
      "step": 623
    },
    {
      "epoch": 5.57,
      "grad_norm": 0.12532857060432434,
      "learning_rate": 0.0002928235294117647,
      "loss": 0.1055,
      "step": 624
    },
    {
      "epoch": 5.58,
      "grad_norm": 0.1194140687584877,
      "learning_rate": 0.0002927647058823529,
      "loss": 0.0976,
      "step": 625
    },
    {
      "epoch": 5.59,
      "grad_norm": 0.11946146935224533,
      "learning_rate": 0.0002927058823529412,
      "loss": 0.1045,
      "step": 626
    },
    {
      "epoch": 5.6,
      "grad_norm": 0.10743141919374466,
      "learning_rate": 0.00029264705882352937,
      "loss": 0.1,
      "step": 627
    },
    {
      "epoch": 5.61,
      "grad_norm": 0.12224756926298141,
      "learning_rate": 0.0002925882352941176,
      "loss": 0.1013,
      "step": 628
    },
    {
      "epoch": 5.62,
      "grad_norm": 0.18434454500675201,
      "learning_rate": 0.00029252941176470586,
      "loss": 0.1182,
      "step": 629
    },
    {
      "epoch": 5.62,
      "grad_norm": 0.10739492624998093,
      "learning_rate": 0.0002924705882352941,
      "loss": 0.1031,
      "step": 630
    },
    {
      "epoch": 5.63,
      "grad_norm": 0.121010921895504,
      "learning_rate": 0.00029241176470588235,
      "loss": 0.1054,
      "step": 631
    },
    {
      "epoch": 5.64,
      "grad_norm": 0.11945497244596481,
      "learning_rate": 0.00029235294117647054,
      "loss": 0.0955,
      "step": 632
    },
    {
      "epoch": 5.65,
      "grad_norm": 0.11026103794574738,
      "learning_rate": 0.00029229411764705883,
      "loss": 0.1097,
      "step": 633
    },
    {
      "epoch": 5.66,
      "grad_norm": 0.11731443554162979,
      "learning_rate": 0.000292235294117647,
      "loss": 0.1008,
      "step": 634
    },
    {
      "epoch": 5.67,
      "grad_norm": 0.19039209187030792,
      "learning_rate": 0.00029217647058823527,
      "loss": 0.135,
      "step": 635
    },
    {
      "epoch": 5.68,
      "grad_norm": 0.12411346286535263,
      "learning_rate": 0.0002921176470588235,
      "loss": 0.1054,
      "step": 636
    },
    {
      "epoch": 5.69,
      "grad_norm": 0.12534594535827637,
      "learning_rate": 0.00029205882352941175,
      "loss": 0.1025,
      "step": 637
    },
    {
      "epoch": 5.7,
      "grad_norm": 0.11824847012758255,
      "learning_rate": 0.000292,
      "loss": 0.1018,
      "step": 638
    },
    {
      "epoch": 5.71,
      "grad_norm": 0.11353108286857605,
      "learning_rate": 0.0002919411764705882,
      "loss": 0.1052,
      "step": 639
    },
    {
      "epoch": 5.71,
      "grad_norm": 0.12157388031482697,
      "learning_rate": 0.00029188235294117643,
      "loss": 0.0997,
      "step": 640
    },
    {
      "epoch": 5.72,
      "grad_norm": 0.13171547651290894,
      "learning_rate": 0.00029182352941176467,
      "loss": 0.1024,
      "step": 641
    },
    {
      "epoch": 5.73,
      "grad_norm": 0.18709394335746765,
      "learning_rate": 0.0002917647058823529,
      "loss": 0.1241,
      "step": 642
    },
    {
      "epoch": 5.74,
      "grad_norm": 0.12204131484031677,
      "learning_rate": 0.00029170588235294116,
      "loss": 0.1061,
      "step": 643
    },
    {
      "epoch": 5.75,
      "grad_norm": 0.1004999428987503,
      "learning_rate": 0.0002916470588235294,
      "loss": 0.1122,
      "step": 644
    },
    {
      "epoch": 5.76,
      "grad_norm": 0.12732595205307007,
      "learning_rate": 0.0002915882352941176,
      "loss": 0.1086,
      "step": 645
    },
    {
      "epoch": 5.77,
      "grad_norm": 0.12090475857257843,
      "learning_rate": 0.0002915294117647059,
      "loss": 0.112,
      "step": 646
    },
    {
      "epoch": 5.78,
      "grad_norm": 0.11316008865833282,
      "learning_rate": 0.0002914705882352941,
      "loss": 0.1057,
      "step": 647
    },
    {
      "epoch": 5.79,
      "grad_norm": 0.1507110595703125,
      "learning_rate": 0.0002914117647058823,
      "loss": 0.1152,
      "step": 648
    },
    {
      "epoch": 5.79,
      "grad_norm": 0.12285605818033218,
      "learning_rate": 0.00029135294117647057,
      "loss": 0.1077,
      "step": 649
    },
    {
      "epoch": 5.8,
      "grad_norm": 0.1024705171585083,
      "learning_rate": 0.0002912941176470588,
      "loss": 0.0996,
      "step": 650
    },
    {
      "epoch": 5.81,
      "grad_norm": 0.1180039569735527,
      "learning_rate": 0.00029123529411764705,
      "loss": 0.0967,
      "step": 651
    },
    {
      "epoch": 5.82,
      "grad_norm": 0.11817004531621933,
      "learning_rate": 0.00029117647058823524,
      "loss": 0.1015,
      "step": 652
    },
    {
      "epoch": 5.83,
      "grad_norm": 0.12755826115608215,
      "learning_rate": 0.0002911176470588235,
      "loss": 0.0987,
      "step": 653
    },
    {
      "epoch": 5.84,
      "grad_norm": 0.20057161152362823,
      "learning_rate": 0.00029105882352941173,
      "loss": 0.1245,
      "step": 654
    },
    {
      "epoch": 5.85,
      "grad_norm": 0.11148959398269653,
      "learning_rate": 0.00029099999999999997,
      "loss": 0.0975,
      "step": 655
    },
    {
      "epoch": 5.86,
      "grad_norm": 0.10137537866830826,
      "learning_rate": 0.0002909411764705882,
      "loss": 0.1051,
      "step": 656
    },
    {
      "epoch": 5.87,
      "grad_norm": 0.10529523342847824,
      "learning_rate": 0.00029088235294117646,
      "loss": 0.1043,
      "step": 657
    },
    {
      "epoch": 5.88,
      "grad_norm": 0.11081627756357193,
      "learning_rate": 0.00029082352941176465,
      "loss": 0.0981,
      "step": 658
    },
    {
      "epoch": 5.88,
      "grad_norm": 0.11853086948394775,
      "learning_rate": 0.00029076470588235295,
      "loss": 0.1061,
      "step": 659
    },
    {
      "epoch": 5.89,
      "grad_norm": 0.18606580793857574,
      "learning_rate": 0.00029070588235294113,
      "loss": 0.1187,
      "step": 660
    },
    {
      "epoch": 5.9,
      "grad_norm": 0.11597760021686554,
      "learning_rate": 0.0002906470588235294,
      "loss": 0.1109,
      "step": 661
    },
    {
      "epoch": 5.91,
      "grad_norm": 0.10947950929403305,
      "learning_rate": 0.0002905882352941176,
      "loss": 0.1065,
      "step": 662
    },
    {
      "epoch": 5.92,
      "grad_norm": 0.1000477746129036,
      "learning_rate": 0.00029052941176470587,
      "loss": 0.0936,
      "step": 663
    },
    {
      "epoch": 5.93,
      "grad_norm": 0.11304661631584167,
      "learning_rate": 0.0002904705882352941,
      "loss": 0.1028,
      "step": 664
    },
    {
      "epoch": 5.94,
      "grad_norm": 0.10919023305177689,
      "learning_rate": 0.0002904117647058823,
      "loss": 0.099,
      "step": 665
    },
    {
      "epoch": 5.95,
      "grad_norm": 0.1719236522912979,
      "learning_rate": 0.0002903529411764706,
      "loss": 0.1139,
      "step": 666
    },
    {
      "epoch": 5.96,
      "grad_norm": 0.1882009357213974,
      "learning_rate": 0.0002902941176470588,
      "loss": 0.1242,
      "step": 667
    },
    {
      "epoch": 5.96,
      "grad_norm": 0.11197476089000702,
      "learning_rate": 0.00029023529411764703,
      "loss": 0.1034,
      "step": 668
    },
    {
      "epoch": 5.97,
      "grad_norm": 0.1102653369307518,
      "learning_rate": 0.00029017647058823527,
      "loss": 0.0969,
      "step": 669
    },
    {
      "epoch": 5.98,
      "grad_norm": 0.10863769054412842,
      "learning_rate": 0.0002901176470588235,
      "loss": 0.0934,
      "step": 670
    },
    {
      "epoch": 5.99,
      "grad_norm": 0.12643195688724518,
      "learning_rate": 0.00029005882352941176,
      "loss": 0.0985,
      "step": 671
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.14720340073108673,
      "learning_rate": 0.00029,
      "loss": 0.0892,
      "step": 672
    },
    {
      "epoch": 6.01,
      "grad_norm": 0.11089643836021423,
      "learning_rate": 0.0002899411764705882,
      "loss": 0.0855,
      "step": 673
    },
    {
      "epoch": 6.02,
      "grad_norm": 0.09804321080446243,
      "learning_rate": 0.00028988235294117643,
      "loss": 0.0903,
      "step": 674
    },
    {
      "epoch": 6.03,
      "grad_norm": 0.11876717209815979,
      "learning_rate": 0.0002898235294117647,
      "loss": 0.0864,
      "step": 675
    },
    {
      "epoch": 6.04,
      "grad_norm": 0.12075202167034149,
      "learning_rate": 0.0002897647058823529,
      "loss": 0.0887,
      "step": 676
    },
    {
      "epoch": 6.04,
      "grad_norm": 0.10439233481884003,
      "learning_rate": 0.00028970588235294116,
      "loss": 0.0878,
      "step": 677
    },
    {
      "epoch": 6.05,
      "grad_norm": 0.1481756716966629,
      "learning_rate": 0.00028964705882352935,
      "loss": 0.0866,
      "step": 678
    },
    {
      "epoch": 6.06,
      "grad_norm": 0.15272961556911469,
      "learning_rate": 0.00028958823529411765,
      "loss": 0.0986,
      "step": 679
    },
    {
      "epoch": 6.07,
      "grad_norm": 0.10657694935798645,
      "learning_rate": 0.00028952941176470584,
      "loss": 0.0892,
      "step": 680
    },
    {
      "epoch": 6.08,
      "grad_norm": 0.11053676903247833,
      "learning_rate": 0.0002894705882352941,
      "loss": 0.0812,
      "step": 681
    },
    {
      "epoch": 6.09,
      "grad_norm": 0.10250803083181381,
      "learning_rate": 0.00028941176470588233,
      "loss": 0.0771,
      "step": 682
    },
    {
      "epoch": 6.1,
      "grad_norm": 0.09995327144861221,
      "learning_rate": 0.00028935294117647057,
      "loss": 0.078,
      "step": 683
    },
    {
      "epoch": 6.11,
      "grad_norm": 0.11875566095113754,
      "learning_rate": 0.0002892941176470588,
      "loss": 0.0762,
      "step": 684
    },
    {
      "epoch": 6.12,
      "grad_norm": 0.17253772914409637,
      "learning_rate": 0.00028923529411764706,
      "loss": 0.0989,
      "step": 685
    },
    {
      "epoch": 6.12,
      "grad_norm": 0.11323681473731995,
      "learning_rate": 0.00028917647058823525,
      "loss": 0.082,
      "step": 686
    },
    {
      "epoch": 6.13,
      "grad_norm": 0.11157402396202087,
      "learning_rate": 0.0002891176470588235,
      "loss": 0.0789,
      "step": 687
    },
    {
      "epoch": 6.14,
      "grad_norm": 0.10261103510856628,
      "learning_rate": 0.00028905882352941173,
      "loss": 0.082,
      "step": 688
    },
    {
      "epoch": 6.15,
      "grad_norm": 0.10514463484287262,
      "learning_rate": 0.000289,
      "loss": 0.0778,
      "step": 689
    },
    {
      "epoch": 6.16,
      "grad_norm": 0.14041176438331604,
      "learning_rate": 0.0002889411764705882,
      "loss": 0.0798,
      "step": 690
    },
    {
      "epoch": 6.17,
      "grad_norm": 0.19121907651424408,
      "learning_rate": 0.0002888823529411764,
      "loss": 0.1093,
      "step": 691
    },
    {
      "epoch": 6.18,
      "grad_norm": 0.10961633175611496,
      "learning_rate": 0.0002888235294117647,
      "loss": 0.0888,
      "step": 692
    },
    {
      "epoch": 6.19,
      "grad_norm": 0.09609069675207138,
      "learning_rate": 0.0002887647058823529,
      "loss": 0.0781,
      "step": 693
    },
    {
      "epoch": 6.2,
      "grad_norm": 0.11091334372758865,
      "learning_rate": 0.00028870588235294114,
      "loss": 0.0908,
      "step": 694
    },
    {
      "epoch": 6.21,
      "grad_norm": 0.11728901416063309,
      "learning_rate": 0.0002886470588235294,
      "loss": 0.0898,
      "step": 695
    },
    {
      "epoch": 6.21,
      "grad_norm": 0.11376498639583588,
      "learning_rate": 0.00028858823529411763,
      "loss": 0.0848,
      "step": 696
    },
    {
      "epoch": 6.22,
      "grad_norm": 0.18690404295921326,
      "learning_rate": 0.00028852941176470587,
      "loss": 0.1101,
      "step": 697
    },
    {
      "epoch": 6.23,
      "grad_norm": 0.10487819463014603,
      "learning_rate": 0.00028847058823529406,
      "loss": 0.0858,
      "step": 698
    },
    {
      "epoch": 6.24,
      "grad_norm": 0.09847170114517212,
      "learning_rate": 0.00028841176470588236,
      "loss": 0.0828,
      "step": 699
    },
    {
      "epoch": 6.25,
      "grad_norm": 0.09805063903331757,
      "learning_rate": 0.00028835294117647055,
      "loss": 0.0825,
      "step": 700
    },
    {
      "epoch": 6.25,
      "eval_cer": 0.03928996929414108,
      "eval_loss": 0.16620230674743652,
      "eval_runtime": 22.7244,
      "eval_samples_per_second": 116.263,
      "eval_steps_per_second": 1.848,
      "eval_wer": 0.14061880206267355,
      "step": 700
    },
    {
      "epoch": 6.26,
      "grad_norm": 0.11333134770393372,
      "learning_rate": 0.0002882941176470588,
      "loss": 0.0823,
      "step": 701
    },
    {
      "epoch": 6.27,
      "grad_norm": 0.09311158955097198,
      "learning_rate": 0.00028823529411764703,
      "loss": 0.0745,
      "step": 702
    },
    {
      "epoch": 6.28,
      "grad_norm": 0.11538677662611008,
      "learning_rate": 0.0002881764705882353,
      "loss": 0.0822,
      "step": 703
    },
    {
      "epoch": 6.29,
      "grad_norm": 0.15968160331249237,
      "learning_rate": 0.0002881176470588235,
      "loss": 0.0889,
      "step": 704
    },
    {
      "epoch": 6.29,
      "grad_norm": 0.12699197232723236,
      "learning_rate": 0.00028805882352941176,
      "loss": 0.0884,
      "step": 705
    },
    {
      "epoch": 6.3,
      "grad_norm": 0.09725473076105118,
      "learning_rate": 0.00028799999999999995,
      "loss": 0.0818,
      "step": 706
    },
    {
      "epoch": 6.31,
      "grad_norm": 0.11126311868429184,
      "learning_rate": 0.0002879411764705882,
      "loss": 0.0866,
      "step": 707
    },
    {
      "epoch": 6.32,
      "grad_norm": 0.12113364040851593,
      "learning_rate": 0.00028788235294117644,
      "loss": 0.0846,
      "step": 708
    },
    {
      "epoch": 6.33,
      "grad_norm": 0.11363382637500763,
      "learning_rate": 0.0002878235294117647,
      "loss": 0.0838,
      "step": 709
    },
    {
      "epoch": 6.34,
      "grad_norm": 0.17675095796585083,
      "learning_rate": 0.0002877647058823529,
      "loss": 0.1035,
      "step": 710
    },
    {
      "epoch": 6.35,
      "grad_norm": 0.12177148461341858,
      "learning_rate": 0.0002877058823529411,
      "loss": 0.0893,
      "step": 711
    },
    {
      "epoch": 6.36,
      "grad_norm": 0.10355890542268753,
      "learning_rate": 0.0002876470588235294,
      "loss": 0.0812,
      "step": 712
    },
    {
      "epoch": 6.37,
      "grad_norm": 0.11242563277482986,
      "learning_rate": 0.0002875882352941176,
      "loss": 0.0747,
      "step": 713
    },
    {
      "epoch": 6.38,
      "grad_norm": 0.12086278945207596,
      "learning_rate": 0.00028752941176470585,
      "loss": 0.0832,
      "step": 714
    },
    {
      "epoch": 6.38,
      "grad_norm": 0.10405004024505615,
      "learning_rate": 0.0002874705882352941,
      "loss": 0.0785,
      "step": 715
    },
    {
      "epoch": 6.39,
      "grad_norm": 0.17571356892585754,
      "learning_rate": 0.00028741176470588233,
      "loss": 0.0951,
      "step": 716
    },
    {
      "epoch": 6.4,
      "grad_norm": 0.10275229066610336,
      "learning_rate": 0.0002873529411764706,
      "loss": 0.0838,
      "step": 717
    },
    {
      "epoch": 6.41,
      "grad_norm": 0.09858614951372147,
      "learning_rate": 0.0002872941176470588,
      "loss": 0.0854,
      "step": 718
    },
    {
      "epoch": 6.42,
      "grad_norm": 0.11851891875267029,
      "learning_rate": 0.000287235294117647,
      "loss": 0.0809,
      "step": 719
    },
    {
      "epoch": 6.43,
      "grad_norm": 0.10504165291786194,
      "learning_rate": 0.00028717647058823525,
      "loss": 0.083,
      "step": 720
    },
    {
      "epoch": 6.44,
      "grad_norm": 0.10159049928188324,
      "learning_rate": 0.0002871176470588235,
      "loss": 0.0781,
      "step": 721
    },
    {
      "epoch": 6.45,
      "grad_norm": 0.17435893416404724,
      "learning_rate": 0.00028705882352941174,
      "loss": 0.0963,
      "step": 722
    },
    {
      "epoch": 6.46,
      "grad_norm": 0.11468575149774551,
      "learning_rate": 0.000287,
      "loss": 0.0885,
      "step": 723
    },
    {
      "epoch": 6.46,
      "grad_norm": 0.09264139831066132,
      "learning_rate": 0.00028694117647058817,
      "loss": 0.0801,
      "step": 724
    },
    {
      "epoch": 6.47,
      "grad_norm": 0.12143049389123917,
      "learning_rate": 0.00028688235294117647,
      "loss": 0.0918,
      "step": 725
    },
    {
      "epoch": 6.48,
      "grad_norm": 0.15513074398040771,
      "learning_rate": 0.00028682352941176466,
      "loss": 0.0837,
      "step": 726
    },
    {
      "epoch": 6.49,
      "grad_norm": 0.09233376383781433,
      "learning_rate": 0.00028676470588235296,
      "loss": 0.0744,
      "step": 727
    },
    {
      "epoch": 6.5,
      "grad_norm": 0.14203836023807526,
      "learning_rate": 0.00028670588235294115,
      "loss": 0.0887,
      "step": 728
    },
    {
      "epoch": 6.51,
      "grad_norm": 0.14789164066314697,
      "learning_rate": 0.0002866470588235294,
      "loss": 0.0991,
      "step": 729
    },
    {
      "epoch": 6.52,
      "grad_norm": 0.10407066345214844,
      "learning_rate": 0.00028658823529411763,
      "loss": 0.0868,
      "step": 730
    },
    {
      "epoch": 6.53,
      "grad_norm": 0.10366200655698776,
      "learning_rate": 0.0002865294117647059,
      "loss": 0.0813,
      "step": 731
    },
    {
      "epoch": 6.54,
      "grad_norm": 0.11755690723657608,
      "learning_rate": 0.0002864705882352941,
      "loss": 0.0829,
      "step": 732
    },
    {
      "epoch": 6.54,
      "grad_norm": 0.0985301211476326,
      "learning_rate": 0.0002864117647058823,
      "loss": 0.0759,
      "step": 733
    },
    {
      "epoch": 6.55,
      "grad_norm": 0.10445389896631241,
      "learning_rate": 0.00028635294117647055,
      "loss": 0.074,
      "step": 734
    },
    {
      "epoch": 6.56,
      "grad_norm": 0.19053040444850922,
      "learning_rate": 0.0002862941176470588,
      "loss": 0.1166,
      "step": 735
    },
    {
      "epoch": 6.57,
      "grad_norm": 0.10317280888557434,
      "learning_rate": 0.00028623529411764704,
      "loss": 0.0845,
      "step": 736
    },
    {
      "epoch": 6.58,
      "grad_norm": 0.0945534035563469,
      "learning_rate": 0.0002861764705882353,
      "loss": 0.0769,
      "step": 737
    },
    {
      "epoch": 6.59,
      "grad_norm": 0.10750401020050049,
      "learning_rate": 0.0002861176470588235,
      "loss": 0.0849,
      "step": 738
    },
    {
      "epoch": 6.6,
      "grad_norm": 0.11884390562772751,
      "learning_rate": 0.0002860588235294117,
      "loss": 0.0834,
      "step": 739
    },
    {
      "epoch": 6.61,
      "grad_norm": 0.10446945577859879,
      "learning_rate": 0.00028599999999999996,
      "loss": 0.0803,
      "step": 740
    },
    {
      "epoch": 6.62,
      "grad_norm": 0.170147106051445,
      "learning_rate": 0.0002859411764705882,
      "loss": 0.106,
      "step": 741
    },
    {
      "epoch": 6.62,
      "grad_norm": 0.12705525755882263,
      "learning_rate": 0.00028588235294117645,
      "loss": 0.0899,
      "step": 742
    },
    {
      "epoch": 6.63,
      "grad_norm": 0.09435348957777023,
      "learning_rate": 0.0002858235294117647,
      "loss": 0.0855,
      "step": 743
    },
    {
      "epoch": 6.64,
      "grad_norm": 0.10736219584941864,
      "learning_rate": 0.00028576470588235293,
      "loss": 0.0775,
      "step": 744
    },
    {
      "epoch": 6.65,
      "grad_norm": 0.11375342309474945,
      "learning_rate": 0.0002857058823529412,
      "loss": 0.0836,
      "step": 745
    },
    {
      "epoch": 6.66,
      "grad_norm": 0.11772213876247406,
      "learning_rate": 0.00028564705882352937,
      "loss": 0.0828,
      "step": 746
    },
    {
      "epoch": 6.67,
      "grad_norm": 0.18420763313770294,
      "learning_rate": 0.0002855882352941176,
      "loss": 0.0961,
      "step": 747
    },
    {
      "epoch": 6.68,
      "grad_norm": 0.10926060378551483,
      "learning_rate": 0.00028552941176470585,
      "loss": 0.0871,
      "step": 748
    },
    {
      "epoch": 6.69,
      "grad_norm": 0.12082089483737946,
      "learning_rate": 0.0002854705882352941,
      "loss": 0.0888,
      "step": 749
    },
    {
      "epoch": 6.7,
      "grad_norm": 0.11248403042554855,
      "learning_rate": 0.00028541176470588234,
      "loss": 0.0875,
      "step": 750
    },
    {
      "epoch": 6.71,
      "grad_norm": 0.10935381054878235,
      "learning_rate": 0.0002853529411764706,
      "loss": 0.0829,
      "step": 751
    },
    {
      "epoch": 6.71,
      "grad_norm": 0.12610478699207306,
      "learning_rate": 0.00028529411764705877,
      "loss": 0.0799,
      "step": 752
    },
    {
      "epoch": 6.72,
      "grad_norm": 0.13737954199314117,
      "learning_rate": 0.000285235294117647,
      "loss": 0.0953,
      "step": 753
    },
    {
      "epoch": 6.73,
      "grad_norm": 0.16050100326538086,
      "learning_rate": 0.00028517647058823526,
      "loss": 0.1001,
      "step": 754
    },
    {
      "epoch": 6.74,
      "grad_norm": 0.12358574569225311,
      "learning_rate": 0.0002851176470588235,
      "loss": 0.0817,
      "step": 755
    },
    {
      "epoch": 6.75,
      "grad_norm": 0.1142040342092514,
      "learning_rate": 0.00028505882352941175,
      "loss": 0.0835,
      "step": 756
    },
    {
      "epoch": 6.76,
      "grad_norm": 0.10749552398920059,
      "learning_rate": 0.000285,
      "loss": 0.0834,
      "step": 757
    },
    {
      "epoch": 6.77,
      "grad_norm": 0.10637697577476501,
      "learning_rate": 0.00028494117647058823,
      "loss": 0.0774,
      "step": 758
    },
    {
      "epoch": 6.78,
      "grad_norm": 0.12128923088312149,
      "learning_rate": 0.0002848823529411764,
      "loss": 0.0817,
      "step": 759
    },
    {
      "epoch": 6.79,
      "grad_norm": 0.15798382461071014,
      "learning_rate": 0.0002848235294117647,
      "loss": 0.1014,
      "step": 760
    },
    {
      "epoch": 6.79,
      "grad_norm": 0.11042901873588562,
      "learning_rate": 0.0002847647058823529,
      "loss": 0.0789,
      "step": 761
    },
    {
      "epoch": 6.8,
      "grad_norm": 0.10593892633914948,
      "learning_rate": 0.00028470588235294115,
      "loss": 0.0795,
      "step": 762
    },
    {
      "epoch": 6.81,
      "grad_norm": 0.10690458863973618,
      "learning_rate": 0.0002846470588235294,
      "loss": 0.0757,
      "step": 763
    },
    {
      "epoch": 6.82,
      "grad_norm": 0.1336659938097,
      "learning_rate": 0.00028458823529411764,
      "loss": 0.0826,
      "step": 764
    },
    {
      "epoch": 6.83,
      "grad_norm": 0.10572272539138794,
      "learning_rate": 0.0002845294117647059,
      "loss": 0.0831,
      "step": 765
    },
    {
      "epoch": 6.84,
      "grad_norm": 0.14707894623279572,
      "learning_rate": 0.00028447058823529407,
      "loss": 0.0923,
      "step": 766
    },
    {
      "epoch": 6.85,
      "grad_norm": 0.11748572438955307,
      "learning_rate": 0.0002844117647058823,
      "loss": 0.0855,
      "step": 767
    },
    {
      "epoch": 6.86,
      "grad_norm": 0.10177357494831085,
      "learning_rate": 0.00028435294117647056,
      "loss": 0.0794,
      "step": 768
    },
    {
      "epoch": 6.87,
      "grad_norm": 0.11586737632751465,
      "learning_rate": 0.0002842941176470588,
      "loss": 0.0819,
      "step": 769
    },
    {
      "epoch": 6.88,
      "grad_norm": 0.10180850327014923,
      "learning_rate": 0.00028423529411764704,
      "loss": 0.0766,
      "step": 770
    },
    {
      "epoch": 6.88,
      "grad_norm": 0.10513108968734741,
      "learning_rate": 0.0002841764705882353,
      "loss": 0.0759,
      "step": 771
    },
    {
      "epoch": 6.89,
      "grad_norm": 0.1731770634651184,
      "learning_rate": 0.0002841176470588235,
      "loss": 0.1055,
      "step": 772
    },
    {
      "epoch": 6.9,
      "grad_norm": 0.11565803736448288,
      "learning_rate": 0.0002840588235294118,
      "loss": 0.0882,
      "step": 773
    },
    {
      "epoch": 6.91,
      "grad_norm": 0.09913301467895508,
      "learning_rate": 0.00028399999999999996,
      "loss": 0.0874,
      "step": 774
    },
    {
      "epoch": 6.92,
      "grad_norm": 0.10817453265190125,
      "learning_rate": 0.0002839411764705882,
      "loss": 0.0811,
      "step": 775
    },
    {
      "epoch": 6.93,
      "grad_norm": 0.09647570550441742,
      "learning_rate": 0.00028388235294117645,
      "loss": 0.079,
      "step": 776
    },
    {
      "epoch": 6.94,
      "grad_norm": 0.10882366448640823,
      "learning_rate": 0.0002838235294117647,
      "loss": 0.0809,
      "step": 777
    },
    {
      "epoch": 6.95,
      "grad_norm": 0.11249500513076782,
      "learning_rate": 0.00028376470588235294,
      "loss": 0.0757,
      "step": 778
    },
    {
      "epoch": 6.96,
      "grad_norm": 0.19009627401828766,
      "learning_rate": 0.00028370588235294113,
      "loss": 0.1112,
      "step": 779
    },
    {
      "epoch": 6.96,
      "grad_norm": 0.10359571129083633,
      "learning_rate": 0.00028364705882352937,
      "loss": 0.0851,
      "step": 780
    },
    {
      "epoch": 6.97,
      "grad_norm": 0.12161215394735336,
      "learning_rate": 0.0002835882352941176,
      "loss": 0.0765,
      "step": 781
    },
    {
      "epoch": 6.98,
      "grad_norm": 0.09756043553352356,
      "learning_rate": 0.00028352941176470586,
      "loss": 0.0805,
      "step": 782
    },
    {
      "epoch": 6.99,
      "grad_norm": 0.11786554753780365,
      "learning_rate": 0.0002834705882352941,
      "loss": 0.0743,
      "step": 783
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.1479356288909912,
      "learning_rate": 0.00028341176470588234,
      "loss": 0.0802,
      "step": 784
    },
    {
      "epoch": 7.01,
      "grad_norm": 0.08490903675556183,
      "learning_rate": 0.00028335294117647053,
      "loss": 0.0692,
      "step": 785
    },
    {
      "epoch": 7.02,
      "grad_norm": 0.08878425508737564,
      "learning_rate": 0.00028329411764705883,
      "loss": 0.0697,
      "step": 786
    },
    {
      "epoch": 7.03,
      "grad_norm": 0.0996265783905983,
      "learning_rate": 0.000283235294117647,
      "loss": 0.0697,
      "step": 787
    },
    {
      "epoch": 7.04,
      "grad_norm": 0.09276427328586578,
      "learning_rate": 0.00028317647058823526,
      "loss": 0.0656,
      "step": 788
    },
    {
      "epoch": 7.04,
      "grad_norm": 0.08711417764425278,
      "learning_rate": 0.0002831176470588235,
      "loss": 0.0622,
      "step": 789
    },
    {
      "epoch": 7.05,
      "grad_norm": 0.11598426848649979,
      "learning_rate": 0.00028305882352941175,
      "loss": 0.0689,
      "step": 790
    },
    {
      "epoch": 7.06,
      "grad_norm": 0.13975971937179565,
      "learning_rate": 0.000283,
      "loss": 0.0851,
      "step": 791
    },
    {
      "epoch": 7.07,
      "grad_norm": 0.09264056384563446,
      "learning_rate": 0.0002829411764705882,
      "loss": 0.0723,
      "step": 792
    },
    {
      "epoch": 7.08,
      "grad_norm": 0.09951192885637283,
      "learning_rate": 0.0002828823529411765,
      "loss": 0.0676,
      "step": 793
    },
    {
      "epoch": 7.09,
      "grad_norm": 0.12242364138364792,
      "learning_rate": 0.00028282352941176467,
      "loss": 0.0749,
      "step": 794
    },
    {
      "epoch": 7.1,
      "grad_norm": 0.09080631285905838,
      "learning_rate": 0.0002827647058823529,
      "loss": 0.0653,
      "step": 795
    },
    {
      "epoch": 7.11,
      "grad_norm": 0.110308937728405,
      "learning_rate": 0.00028270588235294116,
      "loss": 0.0672,
      "step": 796
    },
    {
      "epoch": 7.12,
      "grad_norm": 0.18142534792423248,
      "learning_rate": 0.0002826470588235294,
      "loss": 0.0875,
      "step": 797
    },
    {
      "epoch": 7.12,
      "grad_norm": 0.09989775717258453,
      "learning_rate": 0.00028258823529411764,
      "loss": 0.0675,
      "step": 798
    },
    {
      "epoch": 7.13,
      "grad_norm": 0.09028211981058121,
      "learning_rate": 0.00028252941176470583,
      "loss": 0.0658,
      "step": 799
    },
    {
      "epoch": 7.14,
      "grad_norm": 0.13056732714176178,
      "learning_rate": 0.0002824705882352941,
      "loss": 0.0679,
      "step": 800
    },
    {
      "epoch": 7.14,
      "eval_cer": 0.03930124439533512,
      "eval_loss": 0.17467264831066132,
      "eval_runtime": 21.9956,
      "eval_samples_per_second": 120.115,
      "eval_steps_per_second": 1.909,
      "eval_wer": 0.135739785799286,
      "step": 800
    },
    {
      "epoch": 7.15,
      "grad_norm": 0.11192794889211655,
      "learning_rate": 0.0002824117647058823,
      "loss": 0.0643,
      "step": 801
    },
    {
      "epoch": 7.16,
      "grad_norm": 0.11463401466608047,
      "learning_rate": 0.00028235294117647056,
      "loss": 0.0716,
      "step": 802
    },
    {
      "epoch": 7.17,
      "grad_norm": 0.1752871721982956,
      "learning_rate": 0.0002822941176470588,
      "loss": 0.0864,
      "step": 803
    },
    {
      "epoch": 7.18,
      "grad_norm": 0.16184575855731964,
      "learning_rate": 0.00028223529411764705,
      "loss": 0.0736,
      "step": 804
    },
    {
      "epoch": 7.19,
      "grad_norm": 0.08955268561840057,
      "learning_rate": 0.00028217647058823524,
      "loss": 0.065,
      "step": 805
    },
    {
      "epoch": 7.2,
      "grad_norm": 0.13877011835575104,
      "learning_rate": 0.00028211764705882354,
      "loss": 0.0702,
      "step": 806
    },
    {
      "epoch": 7.21,
      "grad_norm": 0.13071797788143158,
      "learning_rate": 0.0002820588235294117,
      "loss": 0.0765,
      "step": 807
    },
    {
      "epoch": 7.21,
      "grad_norm": 0.09239855408668518,
      "learning_rate": 0.00028199999999999997,
      "loss": 0.0658,
      "step": 808
    },
    {
      "epoch": 7.22,
      "grad_norm": 0.18327827751636505,
      "learning_rate": 0.0002819411764705882,
      "loss": 0.0807,
      "step": 809
    },
    {
      "epoch": 7.23,
      "grad_norm": 0.19865559041500092,
      "learning_rate": 0.00028188235294117646,
      "loss": 0.0764,
      "step": 810
    },
    {
      "epoch": 7.24,
      "grad_norm": 0.09919494390487671,
      "learning_rate": 0.0002818235294117647,
      "loss": 0.0692,
      "step": 811
    },
    {
      "epoch": 7.25,
      "grad_norm": 0.12102092057466507,
      "learning_rate": 0.0002817647058823529,
      "loss": 0.0686,
      "step": 812
    },
    {
      "epoch": 7.26,
      "grad_norm": 0.1632012128829956,
      "learning_rate": 0.00028170588235294113,
      "loss": 0.0767,
      "step": 813
    },
    {
      "epoch": 7.27,
      "grad_norm": 0.10684625804424286,
      "learning_rate": 0.0002816470588235294,
      "loss": 0.0663,
      "step": 814
    },
    {
      "epoch": 7.28,
      "grad_norm": 0.13295963406562805,
      "learning_rate": 0.0002815882352941176,
      "loss": 0.0681,
      "step": 815
    },
    {
      "epoch": 7.29,
      "grad_norm": 0.2165580689907074,
      "learning_rate": 0.00028152941176470586,
      "loss": 0.0869,
      "step": 816
    },
    {
      "epoch": 7.29,
      "grad_norm": 0.13848912715911865,
      "learning_rate": 0.0002814705882352941,
      "loss": 0.0725,
      "step": 817
    },
    {
      "epoch": 7.3,
      "grad_norm": 0.08974460512399673,
      "learning_rate": 0.0002814117647058823,
      "loss": 0.0726,
      "step": 818
    },
    {
      "epoch": 7.31,
      "grad_norm": 0.13465768098831177,
      "learning_rate": 0.0002813529411764706,
      "loss": 0.0731,
      "step": 819
    },
    {
      "epoch": 7.32,
      "grad_norm": 0.1530093252658844,
      "learning_rate": 0.0002812941176470588,
      "loss": 0.0736,
      "step": 820
    },
    {
      "epoch": 7.33,
      "grad_norm": 0.11127042770385742,
      "learning_rate": 0.000281235294117647,
      "loss": 0.0712,
      "step": 821
    },
    {
      "epoch": 7.34,
      "grad_norm": 0.17609044909477234,
      "learning_rate": 0.00028117647058823527,
      "loss": 0.0815,
      "step": 822
    },
    {
      "epoch": 7.35,
      "grad_norm": 0.16929319500923157,
      "learning_rate": 0.0002811176470588235,
      "loss": 0.0699,
      "step": 823
    },
    {
      "epoch": 7.36,
      "grad_norm": 0.12846098840236664,
      "learning_rate": 0.00028105882352941176,
      "loss": 0.08,
      "step": 824
    },
    {
      "epoch": 7.37,
      "grad_norm": 0.1075296625494957,
      "learning_rate": 0.00028099999999999995,
      "loss": 0.0641,
      "step": 825
    },
    {
      "epoch": 7.38,
      "grad_norm": 0.13904526829719543,
      "learning_rate": 0.00028094117647058824,
      "loss": 0.0667,
      "step": 826
    },
    {
      "epoch": 7.38,
      "grad_norm": 0.147836834192276,
      "learning_rate": 0.00028088235294117643,
      "loss": 0.0727,
      "step": 827
    },
    {
      "epoch": 7.39,
      "grad_norm": 0.15775331854820251,
      "learning_rate": 0.0002808235294117647,
      "loss": 0.0784,
      "step": 828
    },
    {
      "epoch": 7.4,
      "grad_norm": 0.1532425731420517,
      "learning_rate": 0.0002807647058823529,
      "loss": 0.076,
      "step": 829
    },
    {
      "epoch": 7.41,
      "grad_norm": 0.14072157442569733,
      "learning_rate": 0.00028070588235294116,
      "loss": 0.0751,
      "step": 830
    },
    {
      "epoch": 7.42,
      "grad_norm": 0.09857498109340668,
      "learning_rate": 0.0002806470588235294,
      "loss": 0.066,
      "step": 831
    },
    {
      "epoch": 7.43,
      "grad_norm": 0.12894302606582642,
      "learning_rate": 0.00028058823529411765,
      "loss": 0.0678,
      "step": 832
    },
    {
      "epoch": 7.44,
      "grad_norm": 0.14168338477611542,
      "learning_rate": 0.00028052941176470584,
      "loss": 0.0708,
      "step": 833
    },
    {
      "epoch": 7.45,
      "grad_norm": 0.18657492101192474,
      "learning_rate": 0.0002804705882352941,
      "loss": 0.0736,
      "step": 834
    },
    {
      "epoch": 7.46,
      "grad_norm": 0.12351266294717789,
      "learning_rate": 0.0002804117647058823,
      "loss": 0.0687,
      "step": 835
    },
    {
      "epoch": 7.46,
      "grad_norm": 0.11583027243614197,
      "learning_rate": 0.00028035294117647057,
      "loss": 0.0714,
      "step": 836
    },
    {
      "epoch": 7.47,
      "grad_norm": 0.10590634495019913,
      "learning_rate": 0.0002802941176470588,
      "loss": 0.0646,
      "step": 837
    },
    {
      "epoch": 7.48,
      "grad_norm": 0.09078478068113327,
      "learning_rate": 0.000280235294117647,
      "loss": 0.0587,
      "step": 838
    },
    {
      "epoch": 7.49,
      "grad_norm": 0.10528270900249481,
      "learning_rate": 0.0002801764705882353,
      "loss": 0.0645,
      "step": 839
    },
    {
      "epoch": 7.5,
      "grad_norm": 0.14622372388839722,
      "learning_rate": 0.0002801176470588235,
      "loss": 0.0824,
      "step": 840
    },
    {
      "epoch": 7.51,
      "grad_norm": 0.1434820294380188,
      "learning_rate": 0.00028005882352941173,
      "loss": 0.0822,
      "step": 841
    },
    {
      "epoch": 7.52,
      "grad_norm": 0.10534597933292389,
      "learning_rate": 0.00028,
      "loss": 0.0691,
      "step": 842
    },
    {
      "epoch": 7.53,
      "grad_norm": 0.10406238585710526,
      "learning_rate": 0.0002799411764705882,
      "loss": 0.0737,
      "step": 843
    },
    {
      "epoch": 7.54,
      "grad_norm": 0.09948386251926422,
      "learning_rate": 0.00027988235294117646,
      "loss": 0.0674,
      "step": 844
    },
    {
      "epoch": 7.54,
      "grad_norm": 0.11599912494421005,
      "learning_rate": 0.0002798235294117647,
      "loss": 0.0692,
      "step": 845
    },
    {
      "epoch": 7.55,
      "grad_norm": 0.1225421354174614,
      "learning_rate": 0.0002797647058823529,
      "loss": 0.0661,
      "step": 846
    },
    {
      "epoch": 7.56,
      "grad_norm": 0.1687183827161789,
      "learning_rate": 0.00027970588235294114,
      "loss": 0.0936,
      "step": 847
    },
    {
      "epoch": 7.57,
      "grad_norm": 0.10220091044902802,
      "learning_rate": 0.0002796470588235294,
      "loss": 0.0719,
      "step": 848
    },
    {
      "epoch": 7.58,
      "grad_norm": 0.09554179757833481,
      "learning_rate": 0.0002795882352941176,
      "loss": 0.0711,
      "step": 849
    },
    {
      "epoch": 7.59,
      "grad_norm": 0.0986250564455986,
      "learning_rate": 0.00027952941176470587,
      "loss": 0.0739,
      "step": 850
    },
    {
      "epoch": 7.6,
      "grad_norm": 0.10000172257423401,
      "learning_rate": 0.00027947058823529406,
      "loss": 0.0702,
      "step": 851
    },
    {
      "epoch": 7.61,
      "grad_norm": 0.10283897817134857,
      "learning_rate": 0.00027941176470588236,
      "loss": 0.0747,
      "step": 852
    },
    {
      "epoch": 7.62,
      "grad_norm": 0.17079313099384308,
      "learning_rate": 0.00027935294117647054,
      "loss": 0.0862,
      "step": 853
    },
    {
      "epoch": 7.62,
      "grad_norm": 0.0937611535191536,
      "learning_rate": 0.0002792941176470588,
      "loss": 0.0649,
      "step": 854
    },
    {
      "epoch": 7.63,
      "grad_norm": 0.09779685735702515,
      "learning_rate": 0.00027923529411764703,
      "loss": 0.0751,
      "step": 855
    },
    {
      "epoch": 7.64,
      "grad_norm": 0.0809277668595314,
      "learning_rate": 0.0002791764705882353,
      "loss": 0.0582,
      "step": 856
    },
    {
      "epoch": 7.65,
      "grad_norm": 0.10247834771871567,
      "learning_rate": 0.0002791176470588235,
      "loss": 0.0718,
      "step": 857
    },
    {
      "epoch": 7.66,
      "grad_norm": 0.10511092841625214,
      "learning_rate": 0.0002790588235294117,
      "loss": 0.0716,
      "step": 858
    },
    {
      "epoch": 7.67,
      "grad_norm": 0.1589038223028183,
      "learning_rate": 0.000279,
      "loss": 0.0808,
      "step": 859
    },
    {
      "epoch": 7.68,
      "grad_norm": 0.10465843230485916,
      "learning_rate": 0.0002789411764705882,
      "loss": 0.078,
      "step": 860
    },
    {
      "epoch": 7.69,
      "grad_norm": 0.10627088695764542,
      "learning_rate": 0.00027888235294117644,
      "loss": 0.0671,
      "step": 861
    },
    {
      "epoch": 7.7,
      "grad_norm": 0.09457018971443176,
      "learning_rate": 0.0002788235294117647,
      "loss": 0.0698,
      "step": 862
    },
    {
      "epoch": 7.71,
      "grad_norm": 0.08960054069757462,
      "learning_rate": 0.0002787647058823529,
      "loss": 0.0665,
      "step": 863
    },
    {
      "epoch": 7.71,
      "grad_norm": 0.09914160519838333,
      "learning_rate": 0.00027870588235294117,
      "loss": 0.0717,
      "step": 864
    },
    {
      "epoch": 7.72,
      "grad_norm": 0.11835897713899612,
      "learning_rate": 0.0002786470588235294,
      "loss": 0.0762,
      "step": 865
    },
    {
      "epoch": 7.73,
      "grad_norm": 0.15103641152381897,
      "learning_rate": 0.0002785882352941176,
      "loss": 0.0858,
      "step": 866
    },
    {
      "epoch": 7.74,
      "grad_norm": 0.08144506067037582,
      "learning_rate": 0.00027852941176470584,
      "loss": 0.0662,
      "step": 867
    },
    {
      "epoch": 7.75,
      "grad_norm": 0.09567223489284515,
      "learning_rate": 0.0002784705882352941,
      "loss": 0.0704,
      "step": 868
    },
    {
      "epoch": 7.76,
      "grad_norm": 0.1075381338596344,
      "learning_rate": 0.00027841176470588233,
      "loss": 0.0699,
      "step": 869
    },
    {
      "epoch": 7.77,
      "grad_norm": 0.09838805347681046,
      "learning_rate": 0.0002783529411764706,
      "loss": 0.0615,
      "step": 870
    },
    {
      "epoch": 7.78,
      "grad_norm": 0.10799439996480942,
      "learning_rate": 0.00027829411764705876,
      "loss": 0.0621,
      "step": 871
    },
    {
      "epoch": 7.79,
      "grad_norm": 0.13640907406806946,
      "learning_rate": 0.00027823529411764706,
      "loss": 0.0845,
      "step": 872
    },
    {
      "epoch": 7.79,
      "grad_norm": 0.09985334426164627,
      "learning_rate": 0.00027817647058823525,
      "loss": 0.0702,
      "step": 873
    },
    {
      "epoch": 7.8,
      "grad_norm": 0.10088913887739182,
      "learning_rate": 0.0002781176470588235,
      "loss": 0.0711,
      "step": 874
    },
    {
      "epoch": 7.81,
      "grad_norm": 0.08682433515787125,
      "learning_rate": 0.00027805882352941174,
      "loss": 0.0684,
      "step": 875
    },
    {
      "epoch": 7.82,
      "grad_norm": 0.10389798134565353,
      "learning_rate": 0.000278,
      "loss": 0.0716,
      "step": 876
    },
    {
      "epoch": 7.83,
      "grad_norm": 0.11450890451669693,
      "learning_rate": 0.0002779411764705882,
      "loss": 0.0739,
      "step": 877
    },
    {
      "epoch": 7.84,
      "grad_norm": 0.13690023124217987,
      "learning_rate": 0.00027788235294117647,
      "loss": 0.072,
      "step": 878
    },
    {
      "epoch": 7.85,
      "grad_norm": 0.08232972025871277,
      "learning_rate": 0.00027782352941176466,
      "loss": 0.0654,
      "step": 879
    },
    {
      "epoch": 7.86,
      "grad_norm": 0.09728033095598221,
      "learning_rate": 0.0002777647058823529,
      "loss": 0.0717,
      "step": 880
    },
    {
      "epoch": 7.87,
      "grad_norm": 0.08549140393733978,
      "learning_rate": 0.00027770588235294114,
      "loss": 0.0682,
      "step": 881
    },
    {
      "epoch": 7.88,
      "grad_norm": 0.09016639739274979,
      "learning_rate": 0.0002776470588235294,
      "loss": 0.0588,
      "step": 882
    },
    {
      "epoch": 7.88,
      "grad_norm": 0.10060504823923111,
      "learning_rate": 0.00027758823529411763,
      "loss": 0.0636,
      "step": 883
    },
    {
      "epoch": 7.89,
      "grad_norm": 0.1495007872581482,
      "learning_rate": 0.0002775294117647058,
      "loss": 0.0793,
      "step": 884
    },
    {
      "epoch": 7.9,
      "grad_norm": 0.09040667861700058,
      "learning_rate": 0.0002774705882352941,
      "loss": 0.0675,
      "step": 885
    },
    {
      "epoch": 7.91,
      "grad_norm": 0.09728195518255234,
      "learning_rate": 0.0002774117647058823,
      "loss": 0.0691,
      "step": 886
    },
    {
      "epoch": 7.92,
      "grad_norm": 0.10015254467725754,
      "learning_rate": 0.0002773529411764706,
      "loss": 0.0631,
      "step": 887
    },
    {
      "epoch": 7.93,
      "grad_norm": 0.08846605569124222,
      "learning_rate": 0.0002772941176470588,
      "loss": 0.0653,
      "step": 888
    },
    {
      "epoch": 7.94,
      "grad_norm": 0.10071244090795517,
      "learning_rate": 0.00027723529411764704,
      "loss": 0.0629,
      "step": 889
    },
    {
      "epoch": 7.95,
      "grad_norm": 0.11886356770992279,
      "learning_rate": 0.0002771764705882353,
      "loss": 0.0725,
      "step": 890
    },
    {
      "epoch": 7.96,
      "grad_norm": 0.13261528313159943,
      "learning_rate": 0.0002771176470588235,
      "loss": 0.0849,
      "step": 891
    },
    {
      "epoch": 7.96,
      "grad_norm": 0.09412933886051178,
      "learning_rate": 0.00027705882352941177,
      "loss": 0.0683,
      "step": 892
    },
    {
      "epoch": 7.97,
      "grad_norm": 0.0950956642627716,
      "learning_rate": 0.00027699999999999996,
      "loss": 0.0742,
      "step": 893
    },
    {
      "epoch": 7.98,
      "grad_norm": 0.08235663175582886,
      "learning_rate": 0.0002769411764705882,
      "loss": 0.0664,
      "step": 894
    },
    {
      "epoch": 7.99,
      "grad_norm": 0.10337395966053009,
      "learning_rate": 0.00027688235294117644,
      "loss": 0.0696,
      "step": 895
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.13153767585754395,
      "learning_rate": 0.0002768235294117647,
      "loss": 0.0798,
      "step": 896
    },
    {
      "epoch": 8.01,
      "grad_norm": 0.09291201084852219,
      "learning_rate": 0.00027676470588235293,
      "loss": 0.0643,
      "step": 897
    },
    {
      "epoch": 8.02,
      "grad_norm": 0.07855793088674545,
      "learning_rate": 0.0002767058823529412,
      "loss": 0.0522,
      "step": 898
    },
    {
      "epoch": 8.03,
      "grad_norm": 0.08614505082368851,
      "learning_rate": 0.00027664705882352936,
      "loss": 0.0538,
      "step": 899
    },
    {
      "epoch": 8.04,
      "grad_norm": 0.09544532001018524,
      "learning_rate": 0.0002765882352941176,
      "loss": 0.0602,
      "step": 900
    },
    {
      "epoch": 8.04,
      "eval_cer": 0.038962991359514115,
      "eval_loss": 0.17666253447532654,
      "eval_runtime": 21.8806,
      "eval_samples_per_second": 120.746,
      "eval_steps_per_second": 1.92,
      "eval_wer": 0.13335977786592623,
      "step": 900
    },
    {
      "epoch": 8.04,
      "grad_norm": 0.09941945225000381,
      "learning_rate": 0.00027652941176470585,
      "loss": 0.0578,
      "step": 901
    },
    {
      "epoch": 8.05,
      "grad_norm": 0.11371108889579773,
      "learning_rate": 0.0002764705882352941,
      "loss": 0.0526,
      "step": 902
    },
    {
      "epoch": 8.06,
      "grad_norm": 0.14651525020599365,
      "learning_rate": 0.00027641176470588234,
      "loss": 0.0683,
      "step": 903
    },
    {
      "epoch": 8.07,
      "grad_norm": 0.08412709087133408,
      "learning_rate": 0.0002763529411764706,
      "loss": 0.056,
      "step": 904
    },
    {
      "epoch": 8.08,
      "grad_norm": 0.08637087792158127,
      "learning_rate": 0.0002762941176470588,
      "loss": 0.0573,
      "step": 905
    },
    {
      "epoch": 8.09,
      "grad_norm": 0.1048191487789154,
      "learning_rate": 0.000276235294117647,
      "loss": 0.0633,
      "step": 906
    },
    {
      "epoch": 8.1,
      "grad_norm": 0.08025553077459335,
      "learning_rate": 0.00027617647058823526,
      "loss": 0.0521,
      "step": 907
    },
    {
      "epoch": 8.11,
      "grad_norm": 0.10991553962230682,
      "learning_rate": 0.0002761176470588235,
      "loss": 0.059,
      "step": 908
    },
    {
      "epoch": 8.12,
      "grad_norm": 0.14864669740200043,
      "learning_rate": 0.00027605882352941174,
      "loss": 0.0785,
      "step": 909
    },
    {
      "epoch": 8.12,
      "grad_norm": 0.09984641522169113,
      "learning_rate": 0.000276,
      "loss": 0.0609,
      "step": 910
    },
    {
      "epoch": 8.13,
      "grad_norm": 0.0754564106464386,
      "learning_rate": 0.00027594117647058823,
      "loss": 0.0508,
      "step": 911
    },
    {
      "epoch": 8.14,
      "grad_norm": 0.10399307310581207,
      "learning_rate": 0.0002758823529411764,
      "loss": 0.0604,
      "step": 912
    },
    {
      "epoch": 8.15,
      "grad_norm": 0.09703277051448822,
      "learning_rate": 0.00027582352941176466,
      "loss": 0.0608,
      "step": 913
    },
    {
      "epoch": 8.16,
      "grad_norm": 0.0977257639169693,
      "learning_rate": 0.0002757647058823529,
      "loss": 0.0494,
      "step": 914
    },
    {
      "epoch": 8.17,
      "grad_norm": 0.1973361223936081,
      "learning_rate": 0.00027570588235294115,
      "loss": 0.0769,
      "step": 915
    },
    {
      "epoch": 8.18,
      "grad_norm": 0.09270834177732468,
      "learning_rate": 0.0002756470588235294,
      "loss": 0.061,
      "step": 916
    },
    {
      "epoch": 8.19,
      "grad_norm": 0.08389316499233246,
      "learning_rate": 0.00027558823529411764,
      "loss": 0.0582,
      "step": 917
    },
    {
      "epoch": 8.2,
      "grad_norm": 0.0841669887304306,
      "learning_rate": 0.0002755294117647059,
      "loss": 0.0593,
      "step": 918
    },
    {
      "epoch": 8.21,
      "grad_norm": 0.0851660966873169,
      "learning_rate": 0.00027547058823529407,
      "loss": 0.0519,
      "step": 919
    },
    {
      "epoch": 8.21,
      "grad_norm": 0.08716004341840744,
      "learning_rate": 0.00027541176470588237,
      "loss": 0.0559,
      "step": 920
    },
    {
      "epoch": 8.22,
      "grad_norm": 0.1524549275636673,
      "learning_rate": 0.00027535294117647056,
      "loss": 0.0747,
      "step": 921
    },
    {
      "epoch": 8.23,
      "grad_norm": 0.09230168163776398,
      "learning_rate": 0.0002752941176470588,
      "loss": 0.065,
      "step": 922
    },
    {
      "epoch": 8.24,
      "grad_norm": 0.0995868593454361,
      "learning_rate": 0.00027523529411764704,
      "loss": 0.0605,
      "step": 923
    },
    {
      "epoch": 8.25,
      "grad_norm": 0.09459994733333588,
      "learning_rate": 0.0002751764705882353,
      "loss": 0.0565,
      "step": 924
    },
    {
      "epoch": 8.26,
      "grad_norm": 0.09015697985887527,
      "learning_rate": 0.00027511764705882353,
      "loss": 0.0564,
      "step": 925
    },
    {
      "epoch": 8.27,
      "grad_norm": 0.0839724987745285,
      "learning_rate": 0.0002750588235294117,
      "loss": 0.0576,
      "step": 926
    },
    {
      "epoch": 8.28,
      "grad_norm": 0.12400028854608536,
      "learning_rate": 0.00027499999999999996,
      "loss": 0.066,
      "step": 927
    },
    {
      "epoch": 8.29,
      "grad_norm": 0.14414748549461365,
      "learning_rate": 0.0002749411764705882,
      "loss": 0.0691,
      "step": 928
    },
    {
      "epoch": 8.29,
      "grad_norm": 0.08903595060110092,
      "learning_rate": 0.00027488235294117645,
      "loss": 0.0622,
      "step": 929
    },
    {
      "epoch": 8.3,
      "grad_norm": 0.09169638156890869,
      "learning_rate": 0.0002748235294117647,
      "loss": 0.0618,
      "step": 930
    },
    {
      "epoch": 8.31,
      "grad_norm": 0.0896272212266922,
      "learning_rate": 0.00027476470588235294,
      "loss": 0.0555,
      "step": 931
    },
    {
      "epoch": 8.32,
      "grad_norm": 0.08792594820261002,
      "learning_rate": 0.0002747058823529411,
      "loss": 0.0546,
      "step": 932
    },
    {
      "epoch": 8.33,
      "grad_norm": 0.10805311053991318,
      "learning_rate": 0.0002746470588235294,
      "loss": 0.0553,
      "step": 933
    },
    {
      "epoch": 8.34,
      "grad_norm": 0.16416336596012115,
      "learning_rate": 0.0002745882352941176,
      "loss": 0.0748,
      "step": 934
    },
    {
      "epoch": 8.35,
      "grad_norm": 0.08754893392324448,
      "learning_rate": 0.00027452941176470586,
      "loss": 0.0604,
      "step": 935
    },
    {
      "epoch": 8.36,
      "grad_norm": 0.09177000820636749,
      "learning_rate": 0.0002744705882352941,
      "loss": 0.0576,
      "step": 936
    },
    {
      "epoch": 8.37,
      "grad_norm": 0.09111658483743668,
      "learning_rate": 0.00027441176470588234,
      "loss": 0.0607,
      "step": 937
    },
    {
      "epoch": 8.38,
      "grad_norm": 0.0870172306895256,
      "learning_rate": 0.0002743529411764706,
      "loss": 0.056,
      "step": 938
    },
    {
      "epoch": 8.38,
      "grad_norm": 0.09112705290317535,
      "learning_rate": 0.0002742941176470588,
      "loss": 0.059,
      "step": 939
    },
    {
      "epoch": 8.39,
      "grad_norm": 0.1845976859331131,
      "learning_rate": 0.000274235294117647,
      "loss": 0.0888,
      "step": 940
    },
    {
      "epoch": 8.4,
      "grad_norm": 0.09225645661354065,
      "learning_rate": 0.00027417647058823526,
      "loss": 0.0555,
      "step": 941
    },
    {
      "epoch": 8.41,
      "grad_norm": 0.08703172951936722,
      "learning_rate": 0.0002741176470588235,
      "loss": 0.0626,
      "step": 942
    },
    {
      "epoch": 8.42,
      "grad_norm": 0.09380820393562317,
      "learning_rate": 0.00027405882352941175,
      "loss": 0.0588,
      "step": 943
    },
    {
      "epoch": 8.43,
      "grad_norm": 0.09199957549571991,
      "learning_rate": 0.000274,
      "loss": 0.0533,
      "step": 944
    },
    {
      "epoch": 8.44,
      "grad_norm": 0.09230124950408936,
      "learning_rate": 0.0002739411764705882,
      "loss": 0.0618,
      "step": 945
    },
    {
      "epoch": 8.45,
      "grad_norm": 0.1362978219985962,
      "learning_rate": 0.0002738823529411765,
      "loss": 0.0731,
      "step": 946
    },
    {
      "epoch": 8.46,
      "grad_norm": 0.10372074693441391,
      "learning_rate": 0.00027382352941176467,
      "loss": 0.0559,
      "step": 947
    },
    {
      "epoch": 8.46,
      "grad_norm": 0.09024462848901749,
      "learning_rate": 0.0002737647058823529,
      "loss": 0.0576,
      "step": 948
    },
    {
      "epoch": 8.47,
      "grad_norm": 0.09534650295972824,
      "learning_rate": 0.00027370588235294116,
      "loss": 0.0597,
      "step": 949
    },
    {
      "epoch": 8.48,
      "grad_norm": 0.09491518139839172,
      "learning_rate": 0.0002736470588235294,
      "loss": 0.0578,
      "step": 950
    },
    {
      "epoch": 8.49,
      "grad_norm": 0.08020088076591492,
      "learning_rate": 0.00027358823529411764,
      "loss": 0.0509,
      "step": 951
    },
    {
      "epoch": 8.5,
      "grad_norm": 0.11732397228479385,
      "learning_rate": 0.00027352941176470583,
      "loss": 0.0602,
      "step": 952
    },
    {
      "epoch": 8.51,
      "grad_norm": 0.11543024331331253,
      "learning_rate": 0.00027347058823529413,
      "loss": 0.0705,
      "step": 953
    },
    {
      "epoch": 8.52,
      "grad_norm": 0.09199339896440506,
      "learning_rate": 0.0002734117647058823,
      "loss": 0.0627,
      "step": 954
    },
    {
      "epoch": 8.53,
      "grad_norm": 0.08132177591323853,
      "learning_rate": 0.00027335294117647056,
      "loss": 0.0564,
      "step": 955
    },
    {
      "epoch": 8.54,
      "grad_norm": 0.08557543158531189,
      "learning_rate": 0.0002732941176470588,
      "loss": 0.0588,
      "step": 956
    },
    {
      "epoch": 8.54,
      "grad_norm": 0.08863170444965363,
      "learning_rate": 0.00027323529411764705,
      "loss": 0.0569,
      "step": 957
    },
    {
      "epoch": 8.55,
      "grad_norm": 0.09457981586456299,
      "learning_rate": 0.0002731764705882353,
      "loss": 0.063,
      "step": 958
    },
    {
      "epoch": 8.56,
      "grad_norm": 0.13630041480064392,
      "learning_rate": 0.0002731176470588235,
      "loss": 0.0724,
      "step": 959
    },
    {
      "epoch": 8.57,
      "grad_norm": 0.0962732583284378,
      "learning_rate": 0.0002730588235294117,
      "loss": 0.0612,
      "step": 960
    },
    {
      "epoch": 8.58,
      "grad_norm": 0.0848195031285286,
      "learning_rate": 0.00027299999999999997,
      "loss": 0.0565,
      "step": 961
    },
    {
      "epoch": 8.59,
      "grad_norm": 0.08632772415876389,
      "learning_rate": 0.0002729411764705882,
      "loss": 0.0576,
      "step": 962
    },
    {
      "epoch": 8.6,
      "grad_norm": 0.09068741649389267,
      "learning_rate": 0.00027288235294117645,
      "loss": 0.0539,
      "step": 963
    },
    {
      "epoch": 8.61,
      "grad_norm": 0.08290354162454605,
      "learning_rate": 0.0002728235294117647,
      "loss": 0.0525,
      "step": 964
    },
    {
      "epoch": 8.62,
      "grad_norm": 0.14676202833652496,
      "learning_rate": 0.0002727647058823529,
      "loss": 0.081,
      "step": 965
    },
    {
      "epoch": 8.62,
      "grad_norm": 0.09038417041301727,
      "learning_rate": 0.0002727058823529412,
      "loss": 0.0514,
      "step": 966
    },
    {
      "epoch": 8.63,
      "grad_norm": 0.09433845430612564,
      "learning_rate": 0.0002726470588235294,
      "loss": 0.059,
      "step": 967
    },
    {
      "epoch": 8.64,
      "grad_norm": 0.0914558544754982,
      "learning_rate": 0.0002725882352941176,
      "loss": 0.0634,
      "step": 968
    },
    {
      "epoch": 8.65,
      "grad_norm": 0.09283778071403503,
      "learning_rate": 0.00027252941176470586,
      "loss": 0.0632,
      "step": 969
    },
    {
      "epoch": 8.66,
      "grad_norm": 0.08448383957147598,
      "learning_rate": 0.0002724705882352941,
      "loss": 0.0542,
      "step": 970
    },
    {
      "epoch": 8.67,
      "grad_norm": 0.16179527342319489,
      "learning_rate": 0.00027241176470588235,
      "loss": 0.0779,
      "step": 971
    },
    {
      "epoch": 8.68,
      "grad_norm": 0.10197985172271729,
      "learning_rate": 0.00027235294117647054,
      "loss": 0.0581,
      "step": 972
    },
    {
      "epoch": 8.69,
      "grad_norm": 0.09279035031795502,
      "learning_rate": 0.0002722941176470588,
      "loss": 0.0582,
      "step": 973
    },
    {
      "epoch": 8.7,
      "grad_norm": 0.08126161992549896,
      "learning_rate": 0.000272235294117647,
      "loss": 0.0582,
      "step": 974
    },
    {
      "epoch": 8.71,
      "grad_norm": 0.09172060340642929,
      "learning_rate": 0.00027217647058823527,
      "loss": 0.0556,
      "step": 975
    },
    {
      "epoch": 8.71,
      "grad_norm": 0.08291953802108765,
      "learning_rate": 0.0002721176470588235,
      "loss": 0.0515,
      "step": 976
    },
    {
      "epoch": 8.72,
      "grad_norm": 0.11062249541282654,
      "learning_rate": 0.00027205882352941175,
      "loss": 0.0574,
      "step": 977
    },
    {
      "epoch": 8.73,
      "grad_norm": 0.13899770379066467,
      "learning_rate": 0.00027199999999999994,
      "loss": 0.0749,
      "step": 978
    },
    {
      "epoch": 8.74,
      "grad_norm": 0.09075085073709488,
      "learning_rate": 0.00027194117647058824,
      "loss": 0.0565,
      "step": 979
    },
    {
      "epoch": 8.75,
      "grad_norm": 0.08652327954769135,
      "learning_rate": 0.00027188235294117643,
      "loss": 0.0562,
      "step": 980
    },
    {
      "epoch": 8.76,
      "grad_norm": 0.08490996062755585,
      "learning_rate": 0.0002718235294117647,
      "loss": 0.0544,
      "step": 981
    },
    {
      "epoch": 8.77,
      "grad_norm": 0.08791162073612213,
      "learning_rate": 0.0002717647058823529,
      "loss": 0.0486,
      "step": 982
    },
    {
      "epoch": 8.78,
      "grad_norm": 0.10936279594898224,
      "learning_rate": 0.00027170588235294116,
      "loss": 0.0601,
      "step": 983
    },
    {
      "epoch": 8.79,
      "grad_norm": 0.1718919575214386,
      "learning_rate": 0.0002716470588235294,
      "loss": 0.0714,
      "step": 984
    },
    {
      "epoch": 8.79,
      "grad_norm": 0.08547305315732956,
      "learning_rate": 0.0002715882352941176,
      "loss": 0.0567,
      "step": 985
    },
    {
      "epoch": 8.8,
      "grad_norm": 0.08878172934055328,
      "learning_rate": 0.0002715294117647059,
      "loss": 0.0605,
      "step": 986
    },
    {
      "epoch": 8.81,
      "grad_norm": 0.08798754215240479,
      "learning_rate": 0.0002714705882352941,
      "loss": 0.062,
      "step": 987
    },
    {
      "epoch": 8.82,
      "grad_norm": 0.08759734779596329,
      "learning_rate": 0.0002714117647058823,
      "loss": 0.0579,
      "step": 988
    },
    {
      "epoch": 8.83,
      "grad_norm": 0.10001529008150101,
      "learning_rate": 0.00027135294117647057,
      "loss": 0.066,
      "step": 989
    },
    {
      "epoch": 8.84,
      "grad_norm": 0.16819550096988678,
      "learning_rate": 0.0002712941176470588,
      "loss": 0.0866,
      "step": 990
    },
    {
      "epoch": 8.85,
      "grad_norm": 0.07452823966741562,
      "learning_rate": 0.00027123529411764705,
      "loss": 0.0551,
      "step": 991
    },
    {
      "epoch": 8.86,
      "grad_norm": 0.09405795484781265,
      "learning_rate": 0.0002711764705882353,
      "loss": 0.0615,
      "step": 992
    },
    {
      "epoch": 8.87,
      "grad_norm": 0.0971001610159874,
      "learning_rate": 0.0002711176470588235,
      "loss": 0.0579,
      "step": 993
    },
    {
      "epoch": 8.88,
      "grad_norm": 0.09090657532215118,
      "learning_rate": 0.00027105882352941173,
      "loss": 0.0551,
      "step": 994
    },
    {
      "epoch": 8.88,
      "grad_norm": 0.09210459142923355,
      "learning_rate": 0.000271,
      "loss": 0.0565,
      "step": 995
    },
    {
      "epoch": 8.89,
      "grad_norm": 0.14499081671237946,
      "learning_rate": 0.0002709411764705882,
      "loss": 0.0641,
      "step": 996
    },
    {
      "epoch": 8.9,
      "grad_norm": 0.08520587533712387,
      "learning_rate": 0.00027088235294117646,
      "loss": 0.0622,
      "step": 997
    },
    {
      "epoch": 8.91,
      "grad_norm": 0.08412635326385498,
      "learning_rate": 0.00027082352941176465,
      "loss": 0.0589,
      "step": 998
    },
    {
      "epoch": 8.92,
      "grad_norm": 0.09470722079277039,
      "learning_rate": 0.00027076470588235295,
      "loss": 0.0557,
      "step": 999
    },
    {
      "epoch": 8.93,
      "grad_norm": 0.09419083595275879,
      "learning_rate": 0.00027070588235294114,
      "loss": 0.0587,
      "step": 1000
    },
    {
      "epoch": 8.93,
      "eval_cer": 0.037576153912648035,
      "eval_loss": 0.1708151251077652,
      "eval_runtime": 22.2458,
      "eval_samples_per_second": 118.764,
      "eval_steps_per_second": 1.888,
      "eval_wer": 0.12915509718365728,
      "step": 1000
    },
    {
      "epoch": 8.94,
      "grad_norm": 0.08441416174173355,
      "learning_rate": 0.0002706470588235294,
      "loss": 0.056,
      "step": 1001
    },
    {
      "epoch": 8.95,
      "grad_norm": 0.14337719976902008,
      "learning_rate": 0.0002705882352941176,
      "loss": 0.0599,
      "step": 1002
    },
    {
      "epoch": 8.96,
      "grad_norm": 0.13183283805847168,
      "learning_rate": 0.00027052941176470587,
      "loss": 0.0725,
      "step": 1003
    },
    {
      "epoch": 8.96,
      "grad_norm": 0.08453375846147537,
      "learning_rate": 0.0002704705882352941,
      "loss": 0.0656,
      "step": 1004
    },
    {
      "epoch": 8.97,
      "grad_norm": 0.09020174294710159,
      "learning_rate": 0.00027041176470588235,
      "loss": 0.0578,
      "step": 1005
    },
    {
      "epoch": 8.98,
      "grad_norm": 0.09032034873962402,
      "learning_rate": 0.00027035294117647054,
      "loss": 0.0605,
      "step": 1006
    },
    {
      "epoch": 8.99,
      "grad_norm": 0.10323207825422287,
      "learning_rate": 0.0002702941176470588,
      "loss": 0.0608,
      "step": 1007
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.10484860837459564,
      "learning_rate": 0.00027023529411764703,
      "loss": 0.0621,
      "step": 1008
    },
    {
      "epoch": 9.01,
      "grad_norm": 0.08602391928434372,
      "learning_rate": 0.0002701764705882353,
      "loss": 0.0532,
      "step": 1009
    },
    {
      "epoch": 9.02,
      "grad_norm": 0.08654514700174332,
      "learning_rate": 0.0002701176470588235,
      "loss": 0.0499,
      "step": 1010
    },
    {
      "epoch": 9.03,
      "grad_norm": 0.07744330167770386,
      "learning_rate": 0.0002700588235294117,
      "loss": 0.0555,
      "step": 1011
    },
    {
      "epoch": 9.04,
      "grad_norm": 0.07674050331115723,
      "learning_rate": 0.00027,
      "loss": 0.0471,
      "step": 1012
    },
    {
      "epoch": 9.04,
      "grad_norm": 0.0944066122174263,
      "learning_rate": 0.0002699411764705882,
      "loss": 0.0498,
      "step": 1013
    },
    {
      "epoch": 9.05,
      "grad_norm": 0.11238271743059158,
      "learning_rate": 0.00026988235294117644,
      "loss": 0.0537,
      "step": 1014
    },
    {
      "epoch": 9.06,
      "grad_norm": 0.13990385830402374,
      "learning_rate": 0.0002698235294117647,
      "loss": 0.0631,
      "step": 1015
    },
    {
      "epoch": 9.07,
      "grad_norm": 0.09597953408956528,
      "learning_rate": 0.0002697647058823529,
      "loss": 0.0525,
      "step": 1016
    },
    {
      "epoch": 9.08,
      "grad_norm": 0.08831837773323059,
      "learning_rate": 0.00026970588235294117,
      "loss": 0.0474,
      "step": 1017
    },
    {
      "epoch": 9.09,
      "grad_norm": 0.09950924664735794,
      "learning_rate": 0.0002696470588235294,
      "loss": 0.0556,
      "step": 1018
    },
    {
      "epoch": 9.1,
      "grad_norm": 0.09094338119029999,
      "learning_rate": 0.00026958823529411765,
      "loss": 0.0532,
      "step": 1019
    },
    {
      "epoch": 9.11,
      "grad_norm": 0.09362645447254181,
      "learning_rate": 0.00026952941176470584,
      "loss": 0.0489,
      "step": 1020
    },
    {
      "epoch": 9.12,
      "grad_norm": 0.1337110996246338,
      "learning_rate": 0.0002694705882352941,
      "loss": 0.0753,
      "step": 1021
    },
    {
      "epoch": 9.12,
      "grad_norm": 0.09633689373731613,
      "learning_rate": 0.00026941176470588233,
      "loss": 0.0542,
      "step": 1022
    },
    {
      "epoch": 9.13,
      "grad_norm": 0.07622726261615753,
      "learning_rate": 0.00026935294117647057,
      "loss": 0.0465,
      "step": 1023
    },
    {
      "epoch": 9.14,
      "grad_norm": 0.08054569363594055,
      "learning_rate": 0.0002692941176470588,
      "loss": 0.0549,
      "step": 1024
    },
    {
      "epoch": 9.15,
      "grad_norm": 0.09806215763092041,
      "learning_rate": 0.00026923529411764706,
      "loss": 0.0507,
      "step": 1025
    },
    {
      "epoch": 9.16,
      "grad_norm": 0.10202576220035553,
      "learning_rate": 0.00026917647058823525,
      "loss": 0.0531,
      "step": 1026
    },
    {
      "epoch": 9.17,
      "grad_norm": 0.15359221398830414,
      "learning_rate": 0.0002691176470588235,
      "loss": 0.0692,
      "step": 1027
    },
    {
      "epoch": 9.18,
      "grad_norm": 0.11079756170511246,
      "learning_rate": 0.00026905882352941174,
      "loss": 0.0526,
      "step": 1028
    },
    {
      "epoch": 9.19,
      "grad_norm": 0.08722373843193054,
      "learning_rate": 0.000269,
      "loss": 0.0507,
      "step": 1029
    },
    {
      "epoch": 9.2,
      "grad_norm": 0.09620029479265213,
      "learning_rate": 0.0002689411764705882,
      "loss": 0.0533,
      "step": 1030
    },
    {
      "epoch": 9.21,
      "grad_norm": 0.12046362459659576,
      "learning_rate": 0.0002688823529411764,
      "loss": 0.0549,
      "step": 1031
    },
    {
      "epoch": 9.21,
      "grad_norm": 0.10825692117214203,
      "learning_rate": 0.0002688235294117647,
      "loss": 0.0513,
      "step": 1032
    },
    {
      "epoch": 9.22,
      "grad_norm": 0.14859241247177124,
      "learning_rate": 0.0002687647058823529,
      "loss": 0.0663,
      "step": 1033
    },
    {
      "epoch": 9.23,
      "grad_norm": 0.13949048519134521,
      "learning_rate": 0.00026870588235294114,
      "loss": 0.0593,
      "step": 1034
    },
    {
      "epoch": 9.24,
      "grad_norm": 0.08228960633277893,
      "learning_rate": 0.0002686470588235294,
      "loss": 0.0503,
      "step": 1035
    },
    {
      "epoch": 9.25,
      "grad_norm": 0.07926836609840393,
      "learning_rate": 0.00026858823529411763,
      "loss": 0.0546,
      "step": 1036
    },
    {
      "epoch": 9.26,
      "grad_norm": 0.09826277196407318,
      "learning_rate": 0.00026852941176470587,
      "loss": 0.0516,
      "step": 1037
    },
    {
      "epoch": 9.27,
      "grad_norm": 0.09743984788656235,
      "learning_rate": 0.0002684705882352941,
      "loss": 0.0514,
      "step": 1038
    },
    {
      "epoch": 9.28,
      "grad_norm": 0.12605629861354828,
      "learning_rate": 0.0002684117647058823,
      "loss": 0.0618,
      "step": 1039
    },
    {
      "epoch": 9.29,
      "grad_norm": 0.14350754022598267,
      "learning_rate": 0.00026835294117647055,
      "loss": 0.0692,
      "step": 1040
    },
    {
      "epoch": 9.29,
      "grad_norm": 0.10322259366512299,
      "learning_rate": 0.0002682941176470588,
      "loss": 0.0553,
      "step": 1041
    },
    {
      "epoch": 9.3,
      "grad_norm": 0.10587611794471741,
      "learning_rate": 0.00026823529411764704,
      "loss": 0.0534,
      "step": 1042
    },
    {
      "epoch": 9.31,
      "grad_norm": 0.09072357416152954,
      "learning_rate": 0.0002681764705882353,
      "loss": 0.0497,
      "step": 1043
    },
    {
      "epoch": 9.32,
      "grad_norm": 0.0994672104716301,
      "learning_rate": 0.00026811764705882347,
      "loss": 0.0501,
      "step": 1044
    },
    {
      "epoch": 9.33,
      "grad_norm": 0.1119416207075119,
      "learning_rate": 0.00026805882352941177,
      "loss": 0.0542,
      "step": 1045
    },
    {
      "epoch": 9.34,
      "grad_norm": 0.1785297393798828,
      "learning_rate": 0.00026799999999999995,
      "loss": 0.0716,
      "step": 1046
    },
    {
      "epoch": 9.35,
      "grad_norm": 0.08592676371335983,
      "learning_rate": 0.00026794117647058825,
      "loss": 0.0537,
      "step": 1047
    },
    {
      "epoch": 9.36,
      "grad_norm": 0.08539824932813644,
      "learning_rate": 0.00026788235294117644,
      "loss": 0.0523,
      "step": 1048
    },
    {
      "epoch": 9.37,
      "grad_norm": 0.08961360156536102,
      "learning_rate": 0.0002678235294117647,
      "loss": 0.05,
      "step": 1049
    },
    {
      "epoch": 9.38,
      "grad_norm": 0.10039302706718445,
      "learning_rate": 0.00026776470588235293,
      "loss": 0.0521,
      "step": 1050
    },
    {
      "epoch": 9.38,
      "grad_norm": 0.09265781193971634,
      "learning_rate": 0.00026770588235294117,
      "loss": 0.0495,
      "step": 1051
    },
    {
      "epoch": 9.39,
      "grad_norm": 0.11948375403881073,
      "learning_rate": 0.0002676470588235294,
      "loss": 0.0617,
      "step": 1052
    },
    {
      "epoch": 9.4,
      "grad_norm": 0.10758794844150543,
      "learning_rate": 0.0002675882352941176,
      "loss": 0.0583,
      "step": 1053
    },
    {
      "epoch": 9.41,
      "grad_norm": 0.10403919965028763,
      "learning_rate": 0.00026752941176470585,
      "loss": 0.0504,
      "step": 1054
    },
    {
      "epoch": 9.42,
      "grad_norm": 0.08936557173728943,
      "learning_rate": 0.0002674705882352941,
      "loss": 0.0488,
      "step": 1055
    },
    {
      "epoch": 9.43,
      "grad_norm": 0.09061574190855026,
      "learning_rate": 0.00026741176470588233,
      "loss": 0.0477,
      "step": 1056
    },
    {
      "epoch": 9.44,
      "grad_norm": 0.1119331419467926,
      "learning_rate": 0.0002673529411764706,
      "loss": 0.0562,
      "step": 1057
    },
    {
      "epoch": 9.45,
      "grad_norm": 0.19698235392570496,
      "learning_rate": 0.0002672941176470588,
      "loss": 0.078,
      "step": 1058
    },
    {
      "epoch": 9.46,
      "grad_norm": 0.07684513926506042,
      "learning_rate": 0.000267235294117647,
      "loss": 0.0527,
      "step": 1059
    },
    {
      "epoch": 9.46,
      "grad_norm": 0.08601843565702438,
      "learning_rate": 0.00026717647058823525,
      "loss": 0.0489,
      "step": 1060
    },
    {
      "epoch": 9.47,
      "grad_norm": 0.09158769994974136,
      "learning_rate": 0.0002671176470588235,
      "loss": 0.051,
      "step": 1061
    },
    {
      "epoch": 9.48,
      "grad_norm": 0.08711161464452744,
      "learning_rate": 0.00026705882352941174,
      "loss": 0.0511,
      "step": 1062
    },
    {
      "epoch": 9.49,
      "grad_norm": 0.09953027963638306,
      "learning_rate": 0.000267,
      "loss": 0.0522,
      "step": 1063
    },
    {
      "epoch": 9.5,
      "grad_norm": 0.12099027633666992,
      "learning_rate": 0.00026694117647058823,
      "loss": 0.0565,
      "step": 1064
    },
    {
      "epoch": 9.51,
      "grad_norm": 0.11144962161779404,
      "learning_rate": 0.00026688235294117647,
      "loss": 0.0611,
      "step": 1065
    },
    {
      "epoch": 9.52,
      "grad_norm": 0.08424019068479538,
      "learning_rate": 0.00026682352941176466,
      "loss": 0.0531,
      "step": 1066
    },
    {
      "epoch": 9.53,
      "grad_norm": 0.10561881214380264,
      "learning_rate": 0.0002667647058823529,
      "loss": 0.0562,
      "step": 1067
    },
    {
      "epoch": 9.54,
      "grad_norm": 0.07985714077949524,
      "learning_rate": 0.00026670588235294115,
      "loss": 0.0502,
      "step": 1068
    },
    {
      "epoch": 9.54,
      "grad_norm": 0.09215716272592545,
      "learning_rate": 0.0002666470588235294,
      "loss": 0.058,
      "step": 1069
    },
    {
      "epoch": 9.55,
      "grad_norm": 0.0968819111585617,
      "learning_rate": 0.00026658823529411763,
      "loss": 0.0486,
      "step": 1070
    },
    {
      "epoch": 9.56,
      "grad_norm": 0.14727649092674255,
      "learning_rate": 0.0002665294117647059,
      "loss": 0.0675,
      "step": 1071
    },
    {
      "epoch": 9.57,
      "grad_norm": 0.08814878016710281,
      "learning_rate": 0.00026647058823529407,
      "loss": 0.0487,
      "step": 1072
    },
    {
      "epoch": 9.58,
      "grad_norm": 0.09155187755823135,
      "learning_rate": 0.0002664117647058823,
      "loss": 0.0536,
      "step": 1073
    },
    {
      "epoch": 9.59,
      "grad_norm": 0.08814623206853867,
      "learning_rate": 0.00026635294117647055,
      "loss": 0.051,
      "step": 1074
    },
    {
      "epoch": 9.6,
      "grad_norm": 0.08754730969667435,
      "learning_rate": 0.0002662941176470588,
      "loss": 0.048,
      "step": 1075
    },
    {
      "epoch": 9.61,
      "grad_norm": 0.1041862815618515,
      "learning_rate": 0.00026623529411764704,
      "loss": 0.0523,
      "step": 1076
    },
    {
      "epoch": 9.62,
      "grad_norm": 0.1607969105243683,
      "learning_rate": 0.0002661764705882353,
      "loss": 0.0684,
      "step": 1077
    },
    {
      "epoch": 9.62,
      "grad_norm": 0.08164142072200775,
      "learning_rate": 0.00026611764705882353,
      "loss": 0.053,
      "step": 1078
    },
    {
      "epoch": 9.63,
      "grad_norm": 0.1007528230547905,
      "learning_rate": 0.0002660588235294117,
      "loss": 0.0553,
      "step": 1079
    },
    {
      "epoch": 9.64,
      "grad_norm": 0.0847950130701065,
      "learning_rate": 0.000266,
      "loss": 0.0493,
      "step": 1080
    },
    {
      "epoch": 9.65,
      "grad_norm": 0.08259861171245575,
      "learning_rate": 0.0002659411764705882,
      "loss": 0.0518,
      "step": 1081
    },
    {
      "epoch": 9.66,
      "grad_norm": 0.09309901297092438,
      "learning_rate": 0.00026588235294117645,
      "loss": 0.0542,
      "step": 1082
    },
    {
      "epoch": 9.67,
      "grad_norm": 0.13553769886493683,
      "learning_rate": 0.0002658235294117647,
      "loss": 0.0612,
      "step": 1083
    },
    {
      "epoch": 9.68,
      "grad_norm": 0.08014171570539474,
      "learning_rate": 0.00026576470588235293,
      "loss": 0.0534,
      "step": 1084
    },
    {
      "epoch": 9.69,
      "grad_norm": 0.08530393987894058,
      "learning_rate": 0.0002657058823529412,
      "loss": 0.0506,
      "step": 1085
    },
    {
      "epoch": 9.7,
      "grad_norm": 0.08911828696727753,
      "learning_rate": 0.00026564705882352937,
      "loss": 0.0534,
      "step": 1086
    },
    {
      "epoch": 9.71,
      "grad_norm": 0.0889921635389328,
      "learning_rate": 0.0002655882352941176,
      "loss": 0.0578,
      "step": 1087
    },
    {
      "epoch": 9.71,
      "grad_norm": 0.09011814743280411,
      "learning_rate": 0.00026552941176470585,
      "loss": 0.046,
      "step": 1088
    },
    {
      "epoch": 9.72,
      "grad_norm": 0.11802621185779572,
      "learning_rate": 0.0002654705882352941,
      "loss": 0.0532,
      "step": 1089
    },
    {
      "epoch": 9.73,
      "grad_norm": 0.10768546909093857,
      "learning_rate": 0.00026541176470588234,
      "loss": 0.0606,
      "step": 1090
    },
    {
      "epoch": 9.74,
      "grad_norm": 0.0835331603884697,
      "learning_rate": 0.0002653529411764706,
      "loss": 0.0553,
      "step": 1091
    },
    {
      "epoch": 9.75,
      "grad_norm": 0.08594707399606705,
      "learning_rate": 0.0002652941176470588,
      "loss": 0.0506,
      "step": 1092
    },
    {
      "epoch": 9.76,
      "grad_norm": 0.07250364124774933,
      "learning_rate": 0.00026523529411764707,
      "loss": 0.0477,
      "step": 1093
    },
    {
      "epoch": 9.77,
      "grad_norm": 0.09176357835531235,
      "learning_rate": 0.00026517647058823526,
      "loss": 0.0534,
      "step": 1094
    },
    {
      "epoch": 9.78,
      "grad_norm": 0.10806913673877716,
      "learning_rate": 0.0002651176470588235,
      "loss": 0.0558,
      "step": 1095
    },
    {
      "epoch": 9.79,
      "grad_norm": 0.1502862423658371,
      "learning_rate": 0.00026505882352941175,
      "loss": 0.0693,
      "step": 1096
    },
    {
      "epoch": 9.79,
      "grad_norm": 0.07961053401231766,
      "learning_rate": 0.000265,
      "loss": 0.0505,
      "step": 1097
    },
    {
      "epoch": 9.8,
      "grad_norm": 0.08200050890445709,
      "learning_rate": 0.00026494117647058823,
      "loss": 0.0571,
      "step": 1098
    },
    {
      "epoch": 9.81,
      "grad_norm": 0.07281451672315598,
      "learning_rate": 0.0002648823529411764,
      "loss": 0.0523,
      "step": 1099
    },
    {
      "epoch": 9.82,
      "grad_norm": 0.0782695934176445,
      "learning_rate": 0.00026482352941176467,
      "loss": 0.0517,
      "step": 1100
    },
    {
      "epoch": 9.82,
      "eval_cer": 0.037222867408568325,
      "eval_loss": 0.16769397258758545,
      "eval_runtime": 22.665,
      "eval_samples_per_second": 116.567,
      "eval_steps_per_second": 1.853,
      "eval_wer": 0.12554541848472828,
      "step": 1100
    },
    {
      "epoch": 9.83,
      "grad_norm": 0.08382805436849594,
      "learning_rate": 0.0002647647058823529,
      "loss": 0.0457,
      "step": 1101
    },
    {
      "epoch": 9.84,
      "grad_norm": 0.14687584340572357,
      "learning_rate": 0.00026470588235294115,
      "loss": 0.064,
      "step": 1102
    },
    {
      "epoch": 9.85,
      "grad_norm": 0.08547946810722351,
      "learning_rate": 0.0002646470588235294,
      "loss": 0.0513,
      "step": 1103
    },
    {
      "epoch": 9.86,
      "grad_norm": 0.08478090167045593,
      "learning_rate": 0.00026458823529411764,
      "loss": 0.0551,
      "step": 1104
    },
    {
      "epoch": 9.87,
      "grad_norm": 0.08604072779417038,
      "learning_rate": 0.00026452941176470583,
      "loss": 0.0534,
      "step": 1105
    },
    {
      "epoch": 9.88,
      "grad_norm": 0.08539281785488129,
      "learning_rate": 0.0002644705882352941,
      "loss": 0.0553,
      "step": 1106
    },
    {
      "epoch": 9.88,
      "grad_norm": 0.08788000047206879,
      "learning_rate": 0.0002644117647058823,
      "loss": 0.0427,
      "step": 1107
    },
    {
      "epoch": 9.89,
      "grad_norm": 0.13512328267097473,
      "learning_rate": 0.00026435294117647056,
      "loss": 0.0618,
      "step": 1108
    },
    {
      "epoch": 9.9,
      "grad_norm": 0.07933490723371506,
      "learning_rate": 0.0002642941176470588,
      "loss": 0.0589,
      "step": 1109
    },
    {
      "epoch": 9.91,
      "grad_norm": 0.0664326548576355,
      "learning_rate": 0.00026423529411764705,
      "loss": 0.0473,
      "step": 1110
    },
    {
      "epoch": 9.92,
      "grad_norm": 0.07653143256902695,
      "learning_rate": 0.0002641764705882353,
      "loss": 0.0494,
      "step": 1111
    },
    {
      "epoch": 9.93,
      "grad_norm": 0.08729689568281174,
      "learning_rate": 0.0002641176470588235,
      "loss": 0.0507,
      "step": 1112
    },
    {
      "epoch": 9.94,
      "grad_norm": 0.0986635610461235,
      "learning_rate": 0.0002640588235294118,
      "loss": 0.0612,
      "step": 1113
    },
    {
      "epoch": 9.95,
      "grad_norm": 0.11171715706586838,
      "learning_rate": 0.00026399999999999997,
      "loss": 0.0602,
      "step": 1114
    },
    {
      "epoch": 9.96,
      "grad_norm": 0.12350744754076004,
      "learning_rate": 0.0002639411764705882,
      "loss": 0.0624,
      "step": 1115
    },
    {
      "epoch": 9.96,
      "grad_norm": 0.07649248093366623,
      "learning_rate": 0.00026388235294117645,
      "loss": 0.0545,
      "step": 1116
    },
    {
      "epoch": 9.97,
      "grad_norm": 0.0775519609451294,
      "learning_rate": 0.0002638235294117647,
      "loss": 0.0448,
      "step": 1117
    },
    {
      "epoch": 9.98,
      "grad_norm": 0.08271569758653641,
      "learning_rate": 0.00026376470588235294,
      "loss": 0.0562,
      "step": 1118
    },
    {
      "epoch": 9.99,
      "grad_norm": 0.10066267102956772,
      "learning_rate": 0.00026370588235294113,
      "loss": 0.05,
      "step": 1119
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.11251965910196304,
      "learning_rate": 0.00026364705882352937,
      "loss": 0.0498,
      "step": 1120
    },
    {
      "epoch": 10.01,
      "grad_norm": 0.08379779756069183,
      "learning_rate": 0.0002635882352941176,
      "loss": 0.0485,
      "step": 1121
    },
    {
      "epoch": 10.02,
      "grad_norm": 0.08651068806648254,
      "learning_rate": 0.00026352941176470586,
      "loss": 0.0461,
      "step": 1122
    },
    {
      "epoch": 10.03,
      "grad_norm": 0.08629054576158524,
      "learning_rate": 0.0002634705882352941,
      "loss": 0.0416,
      "step": 1123
    },
    {
      "epoch": 10.04,
      "grad_norm": 0.08721557259559631,
      "learning_rate": 0.00026341176470588235,
      "loss": 0.0491,
      "step": 1124
    },
    {
      "epoch": 10.04,
      "grad_norm": 0.0717180147767067,
      "learning_rate": 0.00026335294117647054,
      "loss": 0.0443,
      "step": 1125
    },
    {
      "epoch": 10.05,
      "grad_norm": 0.131173238158226,
      "learning_rate": 0.00026329411764705883,
      "loss": 0.0584,
      "step": 1126
    },
    {
      "epoch": 10.06,
      "grad_norm": 0.12688426673412323,
      "learning_rate": 0.000263235294117647,
      "loss": 0.0564,
      "step": 1127
    },
    {
      "epoch": 10.07,
      "grad_norm": 0.0769183412194252,
      "learning_rate": 0.00026317647058823527,
      "loss": 0.0487,
      "step": 1128
    },
    {
      "epoch": 10.08,
      "grad_norm": 0.08034687489271164,
      "learning_rate": 0.0002631176470588235,
      "loss": 0.0427,
      "step": 1129
    },
    {
      "epoch": 10.09,
      "grad_norm": 0.09149681776762009,
      "learning_rate": 0.00026305882352941175,
      "loss": 0.0442,
      "step": 1130
    },
    {
      "epoch": 10.1,
      "grad_norm": 0.07682140916585922,
      "learning_rate": 0.000263,
      "loss": 0.0405,
      "step": 1131
    },
    {
      "epoch": 10.11,
      "grad_norm": 0.09561823308467865,
      "learning_rate": 0.0002629411764705882,
      "loss": 0.0408,
      "step": 1132
    },
    {
      "epoch": 10.12,
      "grad_norm": 0.13112470507621765,
      "learning_rate": 0.00026288235294117643,
      "loss": 0.0586,
      "step": 1133
    },
    {
      "epoch": 10.12,
      "grad_norm": 0.081934355199337,
      "learning_rate": 0.00026282352941176467,
      "loss": 0.0436,
      "step": 1134
    },
    {
      "epoch": 10.13,
      "grad_norm": 0.08208665251731873,
      "learning_rate": 0.0002627647058823529,
      "loss": 0.0418,
      "step": 1135
    },
    {
      "epoch": 10.14,
      "grad_norm": 0.08310721814632416,
      "learning_rate": 0.00026270588235294116,
      "loss": 0.0474,
      "step": 1136
    },
    {
      "epoch": 10.15,
      "grad_norm": 0.07716703414916992,
      "learning_rate": 0.0002626470588235294,
      "loss": 0.042,
      "step": 1137
    },
    {
      "epoch": 10.16,
      "grad_norm": 0.08447854965925217,
      "learning_rate": 0.0002625882352941176,
      "loss": 0.0465,
      "step": 1138
    },
    {
      "epoch": 10.17,
      "grad_norm": 0.1743331402540207,
      "learning_rate": 0.0002625294117647059,
      "loss": 0.0682,
      "step": 1139
    },
    {
      "epoch": 10.18,
      "grad_norm": 0.08664188534021378,
      "learning_rate": 0.0002624705882352941,
      "loss": 0.0433,
      "step": 1140
    },
    {
      "epoch": 10.19,
      "grad_norm": 0.0768757164478302,
      "learning_rate": 0.0002624117647058823,
      "loss": 0.047,
      "step": 1141
    },
    {
      "epoch": 10.2,
      "grad_norm": 0.07780515402555466,
      "learning_rate": 0.00026235294117647057,
      "loss": 0.045,
      "step": 1142
    },
    {
      "epoch": 10.21,
      "grad_norm": 0.09276898950338364,
      "learning_rate": 0.0002622941176470588,
      "loss": 0.0437,
      "step": 1143
    },
    {
      "epoch": 10.21,
      "grad_norm": 0.08791891485452652,
      "learning_rate": 0.00026223529411764705,
      "loss": 0.0421,
      "step": 1144
    },
    {
      "epoch": 10.22,
      "grad_norm": 0.1696498543024063,
      "learning_rate": 0.00026217647058823524,
      "loss": 0.0631,
      "step": 1145
    },
    {
      "epoch": 10.23,
      "grad_norm": 0.08368546515703201,
      "learning_rate": 0.00026211764705882354,
      "loss": 0.0446,
      "step": 1146
    },
    {
      "epoch": 10.24,
      "grad_norm": 0.0758795365691185,
      "learning_rate": 0.00026205882352941173,
      "loss": 0.0454,
      "step": 1147
    },
    {
      "epoch": 10.25,
      "grad_norm": 0.08984597027301788,
      "learning_rate": 0.00026199999999999997,
      "loss": 0.0521,
      "step": 1148
    },
    {
      "epoch": 10.26,
      "grad_norm": 0.08602012693881989,
      "learning_rate": 0.0002619411764705882,
      "loss": 0.0505,
      "step": 1149
    },
    {
      "epoch": 10.27,
      "grad_norm": 0.08585749566555023,
      "learning_rate": 0.00026188235294117646,
      "loss": 0.0436,
      "step": 1150
    },
    {
      "epoch": 10.28,
      "grad_norm": 0.11137910187244415,
      "learning_rate": 0.0002618235294117647,
      "loss": 0.0528,
      "step": 1151
    },
    {
      "epoch": 10.29,
      "grad_norm": 0.13748031854629517,
      "learning_rate": 0.00026176470588235295,
      "loss": 0.0627,
      "step": 1152
    },
    {
      "epoch": 10.29,
      "grad_norm": 0.08785070478916168,
      "learning_rate": 0.00026170588235294113,
      "loss": 0.0528,
      "step": 1153
    },
    {
      "epoch": 10.3,
      "grad_norm": 0.08275540918111801,
      "learning_rate": 0.0002616470588235294,
      "loss": 0.0464,
      "step": 1154
    },
    {
      "epoch": 10.31,
      "grad_norm": 0.080641008913517,
      "learning_rate": 0.0002615882352941176,
      "loss": 0.0436,
      "step": 1155
    },
    {
      "epoch": 10.32,
      "grad_norm": 0.07808531075716019,
      "learning_rate": 0.00026152941176470586,
      "loss": 0.0428,
      "step": 1156
    },
    {
      "epoch": 10.33,
      "grad_norm": 0.0999738797545433,
      "learning_rate": 0.0002614705882352941,
      "loss": 0.0483,
      "step": 1157
    },
    {
      "epoch": 10.34,
      "grad_norm": 0.13058152794837952,
      "learning_rate": 0.0002614117647058823,
      "loss": 0.0681,
      "step": 1158
    },
    {
      "epoch": 10.35,
      "grad_norm": 0.07405345141887665,
      "learning_rate": 0.0002613529411764706,
      "loss": 0.0453,
      "step": 1159
    },
    {
      "epoch": 10.36,
      "grad_norm": 0.07965939491987228,
      "learning_rate": 0.0002612941176470588,
      "loss": 0.0475,
      "step": 1160
    },
    {
      "epoch": 10.37,
      "grad_norm": 0.08219412714242935,
      "learning_rate": 0.00026123529411764703,
      "loss": 0.0457,
      "step": 1161
    },
    {
      "epoch": 10.38,
      "grad_norm": 0.07677296549081802,
      "learning_rate": 0.00026117647058823527,
      "loss": 0.0396,
      "step": 1162
    },
    {
      "epoch": 10.38,
      "grad_norm": 0.09775242209434509,
      "learning_rate": 0.0002611176470588235,
      "loss": 0.0557,
      "step": 1163
    },
    {
      "epoch": 10.39,
      "grad_norm": 0.12974615395069122,
      "learning_rate": 0.00026105882352941176,
      "loss": 0.056,
      "step": 1164
    },
    {
      "epoch": 10.4,
      "grad_norm": 0.07911017537117004,
      "learning_rate": 0.000261,
      "loss": 0.0476,
      "step": 1165
    },
    {
      "epoch": 10.41,
      "grad_norm": 0.08935990184545517,
      "learning_rate": 0.0002609411764705882,
      "loss": 0.0459,
      "step": 1166
    },
    {
      "epoch": 10.42,
      "grad_norm": 0.09498455375432968,
      "learning_rate": 0.00026088235294117643,
      "loss": 0.047,
      "step": 1167
    },
    {
      "epoch": 10.43,
      "grad_norm": 0.07545097917318344,
      "learning_rate": 0.0002608235294117647,
      "loss": 0.0451,
      "step": 1168
    },
    {
      "epoch": 10.44,
      "grad_norm": 0.08683010190725327,
      "learning_rate": 0.0002607647058823529,
      "loss": 0.0483,
      "step": 1169
    },
    {
      "epoch": 10.45,
      "grad_norm": 0.17454423010349274,
      "learning_rate": 0.00026070588235294116,
      "loss": 0.0663,
      "step": 1170
    },
    {
      "epoch": 10.46,
      "grad_norm": 0.08547163009643555,
      "learning_rate": 0.00026064705882352935,
      "loss": 0.0475,
      "step": 1171
    },
    {
      "epoch": 10.46,
      "grad_norm": 0.09390612691640854,
      "learning_rate": 0.00026058823529411765,
      "loss": 0.0544,
      "step": 1172
    },
    {
      "epoch": 10.47,
      "grad_norm": 0.08010204136371613,
      "learning_rate": 0.00026052941176470584,
      "loss": 0.0416,
      "step": 1173
    },
    {
      "epoch": 10.48,
      "grad_norm": 0.0884295329451561,
      "learning_rate": 0.0002604705882352941,
      "loss": 0.0484,
      "step": 1174
    },
    {
      "epoch": 10.49,
      "grad_norm": 0.10123898088932037,
      "learning_rate": 0.00026041176470588233,
      "loss": 0.0426,
      "step": 1175
    },
    {
      "epoch": 10.5,
      "grad_norm": 0.11035466194152832,
      "learning_rate": 0.00026035294117647057,
      "loss": 0.0535,
      "step": 1176
    },
    {
      "epoch": 10.51,
      "grad_norm": 0.10585707426071167,
      "learning_rate": 0.0002602941176470588,
      "loss": 0.0598,
      "step": 1177
    },
    {
      "epoch": 10.52,
      "grad_norm": 0.08312973380088806,
      "learning_rate": 0.00026023529411764706,
      "loss": 0.0439,
      "step": 1178
    },
    {
      "epoch": 10.53,
      "grad_norm": 0.07807246595621109,
      "learning_rate": 0.0002601764705882353,
      "loss": 0.041,
      "step": 1179
    },
    {
      "epoch": 10.54,
      "grad_norm": 0.07961349189281464,
      "learning_rate": 0.0002601176470588235,
      "loss": 0.0417,
      "step": 1180
    },
    {
      "epoch": 10.54,
      "grad_norm": 0.08511442691087723,
      "learning_rate": 0.00026005882352941173,
      "loss": 0.049,
      "step": 1181
    },
    {
      "epoch": 10.55,
      "grad_norm": 0.09853603690862656,
      "learning_rate": 0.00026,
      "loss": 0.0516,
      "step": 1182
    },
    {
      "epoch": 10.56,
      "grad_norm": 0.1274108737707138,
      "learning_rate": 0.0002599411764705882,
      "loss": 0.0561,
      "step": 1183
    },
    {
      "epoch": 10.57,
      "grad_norm": 0.0804334208369255,
      "learning_rate": 0.00025988235294117646,
      "loss": 0.0442,
      "step": 1184
    },
    {
      "epoch": 10.58,
      "grad_norm": 0.08435020595788956,
      "learning_rate": 0.0002598235294117647,
      "loss": 0.0426,
      "step": 1185
    },
    {
      "epoch": 10.59,
      "grad_norm": 0.08223892748355865,
      "learning_rate": 0.0002597647058823529,
      "loss": 0.0491,
      "step": 1186
    },
    {
      "epoch": 10.6,
      "grad_norm": 0.07661667466163635,
      "learning_rate": 0.00025970588235294114,
      "loss": 0.0485,
      "step": 1187
    },
    {
      "epoch": 10.61,
      "grad_norm": 0.09493746608495712,
      "learning_rate": 0.0002596470588235294,
      "loss": 0.0517,
      "step": 1188
    },
    {
      "epoch": 10.62,
      "grad_norm": 0.13816070556640625,
      "learning_rate": 0.0002595882352941176,
      "loss": 0.0591,
      "step": 1189
    },
    {
      "epoch": 10.62,
      "grad_norm": 0.07286964356899261,
      "learning_rate": 0.00025952941176470587,
      "loss": 0.0465,
      "step": 1190
    },
    {
      "epoch": 10.63,
      "grad_norm": 0.07412020862102509,
      "learning_rate": 0.00025947058823529406,
      "loss": 0.049,
      "step": 1191
    },
    {
      "epoch": 10.64,
      "grad_norm": 0.0823870301246643,
      "learning_rate": 0.00025941176470588236,
      "loss": 0.0504,
      "step": 1192
    },
    {
      "epoch": 10.65,
      "grad_norm": 0.07828681915998459,
      "learning_rate": 0.00025935294117647055,
      "loss": 0.0495,
      "step": 1193
    },
    {
      "epoch": 10.66,
      "grad_norm": 0.08328825980424881,
      "learning_rate": 0.0002592941176470588,
      "loss": 0.0439,
      "step": 1194
    },
    {
      "epoch": 10.67,
      "grad_norm": 0.1659298837184906,
      "learning_rate": 0.00025923529411764703,
      "loss": 0.0686,
      "step": 1195
    },
    {
      "epoch": 10.68,
      "grad_norm": 0.08040114492177963,
      "learning_rate": 0.0002591764705882353,
      "loss": 0.0522,
      "step": 1196
    },
    {
      "epoch": 10.69,
      "grad_norm": 0.07643823325634003,
      "learning_rate": 0.0002591176470588235,
      "loss": 0.0459,
      "step": 1197
    },
    {
      "epoch": 10.7,
      "grad_norm": 0.08668006211519241,
      "learning_rate": 0.00025905882352941176,
      "loss": 0.0501,
      "step": 1198
    },
    {
      "epoch": 10.71,
      "grad_norm": 0.07934591919183731,
      "learning_rate": 0.00025899999999999995,
      "loss": 0.0508,
      "step": 1199
    },
    {
      "epoch": 10.71,
      "grad_norm": 0.08404715359210968,
      "learning_rate": 0.0002589411764705882,
      "loss": 0.0413,
      "step": 1200
    },
    {
      "epoch": 10.71,
      "eval_cer": 0.036076565453841615,
      "eval_loss": 0.1771070957183838,
      "eval_runtime": 22.5753,
      "eval_samples_per_second": 117.031,
      "eval_steps_per_second": 1.86,
      "eval_wer": 0.12336374454581515,
      "step": 1200
    },
    {
      "epoch": 10.72,
      "grad_norm": 0.12253329157829285,
      "learning_rate": 0.00025888235294117644,
      "loss": 0.063,
      "step": 1201
    },
    {
      "epoch": 10.73,
      "grad_norm": 0.16100212931632996,
      "learning_rate": 0.0002588235294117647,
      "loss": 0.0538,
      "step": 1202
    },
    {
      "epoch": 10.74,
      "grad_norm": 0.08247417956590652,
      "learning_rate": 0.0002587647058823529,
      "loss": 0.0509,
      "step": 1203
    },
    {
      "epoch": 10.75,
      "grad_norm": 0.07974733412265778,
      "learning_rate": 0.0002587058823529411,
      "loss": 0.0464,
      "step": 1204
    },
    {
      "epoch": 10.76,
      "grad_norm": 0.07784251123666763,
      "learning_rate": 0.0002586470588235294,
      "loss": 0.044,
      "step": 1205
    },
    {
      "epoch": 10.77,
      "grad_norm": 0.08174080401659012,
      "learning_rate": 0.0002585882352941176,
      "loss": 0.0428,
      "step": 1206
    },
    {
      "epoch": 10.78,
      "grad_norm": 0.10989406704902649,
      "learning_rate": 0.0002585294117647059,
      "loss": 0.0502,
      "step": 1207
    },
    {
      "epoch": 10.79,
      "grad_norm": 0.14650195837020874,
      "learning_rate": 0.0002584705882352941,
      "loss": 0.0724,
      "step": 1208
    },
    {
      "epoch": 10.79,
      "grad_norm": 0.07386109977960587,
      "learning_rate": 0.00025841176470588233,
      "loss": 0.0496,
      "step": 1209
    },
    {
      "epoch": 10.8,
      "grad_norm": 0.08304432779550552,
      "learning_rate": 0.0002583529411764706,
      "loss": 0.0527,
      "step": 1210
    },
    {
      "epoch": 10.81,
      "grad_norm": 0.08041701465845108,
      "learning_rate": 0.0002582941176470588,
      "loss": 0.0457,
      "step": 1211
    },
    {
      "epoch": 10.82,
      "grad_norm": 0.08432194590568542,
      "learning_rate": 0.00025823529411764706,
      "loss": 0.0459,
      "step": 1212
    },
    {
      "epoch": 10.83,
      "grad_norm": 0.07892563939094543,
      "learning_rate": 0.00025817647058823525,
      "loss": 0.0399,
      "step": 1213
    },
    {
      "epoch": 10.84,
      "grad_norm": 0.15423566102981567,
      "learning_rate": 0.0002581176470588235,
      "loss": 0.0747,
      "step": 1214
    },
    {
      "epoch": 10.85,
      "grad_norm": 0.0816345140337944,
      "learning_rate": 0.00025805882352941174,
      "loss": 0.0502,
      "step": 1215
    },
    {
      "epoch": 10.86,
      "grad_norm": 0.07691647857427597,
      "learning_rate": 0.000258,
      "loss": 0.0472,
      "step": 1216
    },
    {
      "epoch": 10.87,
      "grad_norm": 0.08117283135652542,
      "learning_rate": 0.0002579411764705882,
      "loss": 0.0437,
      "step": 1217
    },
    {
      "epoch": 10.88,
      "grad_norm": 0.09006890654563904,
      "learning_rate": 0.00025788235294117647,
      "loss": 0.0438,
      "step": 1218
    },
    {
      "epoch": 10.88,
      "grad_norm": 0.08083365112543106,
      "learning_rate": 0.00025782352941176466,
      "loss": 0.0518,
      "step": 1219
    },
    {
      "epoch": 10.89,
      "grad_norm": 0.11542165279388428,
      "learning_rate": 0.0002577647058823529,
      "loss": 0.0543,
      "step": 1220
    },
    {
      "epoch": 10.9,
      "grad_norm": 0.10748816281557083,
      "learning_rate": 0.00025770588235294115,
      "loss": 0.0562,
      "step": 1221
    },
    {
      "epoch": 10.91,
      "grad_norm": 0.08316796272993088,
      "learning_rate": 0.0002576470588235294,
      "loss": 0.049,
      "step": 1222
    },
    {
      "epoch": 10.92,
      "grad_norm": 0.09037583321332932,
      "learning_rate": 0.00025758823529411763,
      "loss": 0.0502,
      "step": 1223
    },
    {
      "epoch": 10.93,
      "grad_norm": 0.10377881675958633,
      "learning_rate": 0.0002575294117647059,
      "loss": 0.0494,
      "step": 1224
    },
    {
      "epoch": 10.94,
      "grad_norm": 0.08607975393533707,
      "learning_rate": 0.0002574705882352941,
      "loss": 0.0505,
      "step": 1225
    },
    {
      "epoch": 10.95,
      "grad_norm": 0.10056962072849274,
      "learning_rate": 0.0002574117647058823,
      "loss": 0.0499,
      "step": 1226
    },
    {
      "epoch": 10.96,
      "grad_norm": 0.1339590847492218,
      "learning_rate": 0.00025735294117647055,
      "loss": 0.0637,
      "step": 1227
    },
    {
      "epoch": 10.96,
      "grad_norm": 0.07969089597463608,
      "learning_rate": 0.0002572941176470588,
      "loss": 0.0464,
      "step": 1228
    },
    {
      "epoch": 10.97,
      "grad_norm": 0.08407177776098251,
      "learning_rate": 0.00025723529411764704,
      "loss": 0.0479,
      "step": 1229
    },
    {
      "epoch": 10.98,
      "grad_norm": 0.12283366918563843,
      "learning_rate": 0.0002571764705882353,
      "loss": 0.0509,
      "step": 1230
    },
    {
      "epoch": 10.99,
      "grad_norm": 0.12030702829360962,
      "learning_rate": 0.0002571176470588235,
      "loss": 0.0492,
      "step": 1231
    },
    {
      "epoch": 11.0,
      "grad_norm": 0.09268844872713089,
      "learning_rate": 0.0002570588235294117,
      "loss": 0.0451,
      "step": 1232
    },
    {
      "epoch": 11.01,
      "grad_norm": 0.1249970942735672,
      "learning_rate": 0.00025699999999999996,
      "loss": 0.045,
      "step": 1233
    },
    {
      "epoch": 11.02,
      "grad_norm": 0.10564065724611282,
      "learning_rate": 0.0002569411764705882,
      "loss": 0.0414,
      "step": 1234
    },
    {
      "epoch": 11.03,
      "grad_norm": 0.06608693301677704,
      "learning_rate": 0.00025688235294117645,
      "loss": 0.0396,
      "step": 1235
    },
    {
      "epoch": 11.04,
      "grad_norm": 0.09664716571569443,
      "learning_rate": 0.0002568235294117647,
      "loss": 0.0399,
      "step": 1236
    },
    {
      "epoch": 11.04,
      "grad_norm": 0.12408290803432465,
      "learning_rate": 0.00025676470588235293,
      "loss": 0.0468,
      "step": 1237
    },
    {
      "epoch": 11.05,
      "grad_norm": 0.11983111500740051,
      "learning_rate": 0.0002567058823529412,
      "loss": 0.0457,
      "step": 1238
    },
    {
      "epoch": 11.06,
      "grad_norm": 0.10642583668231964,
      "learning_rate": 0.00025664705882352936,
      "loss": 0.0505,
      "step": 1239
    },
    {
      "epoch": 11.07,
      "grad_norm": 0.10309722274541855,
      "learning_rate": 0.00025658823529411766,
      "loss": 0.0462,
      "step": 1240
    },
    {
      "epoch": 11.08,
      "grad_norm": 0.11561737209558487,
      "learning_rate": 0.00025652941176470585,
      "loss": 0.048,
      "step": 1241
    },
    {
      "epoch": 11.09,
      "grad_norm": 0.0772339478135109,
      "learning_rate": 0.0002564705882352941,
      "loss": 0.0423,
      "step": 1242
    },
    {
      "epoch": 11.1,
      "grad_norm": 0.08576243370771408,
      "learning_rate": 0.00025641176470588234,
      "loss": 0.0464,
      "step": 1243
    },
    {
      "epoch": 11.11,
      "grad_norm": 0.10407901555299759,
      "learning_rate": 0.0002563529411764706,
      "loss": 0.0444,
      "step": 1244
    },
    {
      "epoch": 11.12,
      "grad_norm": 0.22960324585437775,
      "learning_rate": 0.0002562941176470588,
      "loss": 0.0631,
      "step": 1245
    },
    {
      "epoch": 11.12,
      "grad_norm": 0.06789670139551163,
      "learning_rate": 0.000256235294117647,
      "loss": 0.0453,
      "step": 1246
    },
    {
      "epoch": 11.13,
      "grad_norm": 0.08458477258682251,
      "learning_rate": 0.00025617647058823526,
      "loss": 0.0435,
      "step": 1247
    },
    {
      "epoch": 11.14,
      "grad_norm": 0.09180407226085663,
      "learning_rate": 0.0002561176470588235,
      "loss": 0.0473,
      "step": 1248
    },
    {
      "epoch": 11.15,
      "grad_norm": 0.07602664083242416,
      "learning_rate": 0.00025605882352941174,
      "loss": 0.0395,
      "step": 1249
    },
    {
      "epoch": 11.16,
      "grad_norm": 0.09161707758903503,
      "learning_rate": 0.000256,
      "loss": 0.0413,
      "step": 1250
    },
    {
      "epoch": 11.17,
      "grad_norm": 0.28046804666519165,
      "learning_rate": 0.00025594117647058823,
      "loss": 0.0724,
      "step": 1251
    },
    {
      "epoch": 11.18,
      "grad_norm": 0.07246720790863037,
      "learning_rate": 0.0002558823529411764,
      "loss": 0.0433,
      "step": 1252
    },
    {
      "epoch": 11.19,
      "grad_norm": 0.07940445095300674,
      "learning_rate": 0.0002558235294117647,
      "loss": 0.0445,
      "step": 1253
    },
    {
      "epoch": 11.2,
      "grad_norm": 0.07803329825401306,
      "learning_rate": 0.0002557647058823529,
      "loss": 0.0375,
      "step": 1254
    },
    {
      "epoch": 11.21,
      "grad_norm": 0.07922541350126266,
      "learning_rate": 0.00025570588235294115,
      "loss": 0.0487,
      "step": 1255
    },
    {
      "epoch": 11.21,
      "grad_norm": 0.09134754538536072,
      "learning_rate": 0.0002556470588235294,
      "loss": 0.047,
      "step": 1256
    },
    {
      "epoch": 11.22,
      "grad_norm": 0.1504497528076172,
      "learning_rate": 0.00025558823529411764,
      "loss": 0.0599,
      "step": 1257
    },
    {
      "epoch": 11.23,
      "grad_norm": 0.07694713026285172,
      "learning_rate": 0.0002555294117647059,
      "loss": 0.0469,
      "step": 1258
    },
    {
      "epoch": 11.24,
      "grad_norm": 0.07742314040660858,
      "learning_rate": 0.00025547058823529407,
      "loss": 0.0426,
      "step": 1259
    },
    {
      "epoch": 11.25,
      "grad_norm": 0.07481490820646286,
      "learning_rate": 0.0002554117647058823,
      "loss": 0.043,
      "step": 1260
    },
    {
      "epoch": 11.26,
      "grad_norm": 0.08957730233669281,
      "learning_rate": 0.00025535294117647056,
      "loss": 0.0443,
      "step": 1261
    },
    {
      "epoch": 11.27,
      "grad_norm": 0.0939517617225647,
      "learning_rate": 0.0002552941176470588,
      "loss": 0.0516,
      "step": 1262
    },
    {
      "epoch": 11.28,
      "grad_norm": 0.12942364811897278,
      "learning_rate": 0.00025523529411764704,
      "loss": 0.0492,
      "step": 1263
    },
    {
      "epoch": 11.29,
      "grad_norm": 0.16845741868019104,
      "learning_rate": 0.0002551764705882353,
      "loss": 0.0649,
      "step": 1264
    },
    {
      "epoch": 11.29,
      "grad_norm": 0.08106888830661774,
      "learning_rate": 0.0002551176470588235,
      "loss": 0.0453,
      "step": 1265
    },
    {
      "epoch": 11.3,
      "grad_norm": 0.08626700192689896,
      "learning_rate": 0.0002550588235294118,
      "loss": 0.0469,
      "step": 1266
    },
    {
      "epoch": 11.31,
      "grad_norm": 0.07476877421140671,
      "learning_rate": 0.00025499999999999996,
      "loss": 0.04,
      "step": 1267
    },
    {
      "epoch": 11.32,
      "grad_norm": 0.06678514182567596,
      "learning_rate": 0.0002549411764705882,
      "loss": 0.0379,
      "step": 1268
    },
    {
      "epoch": 11.33,
      "grad_norm": 0.09826838225126266,
      "learning_rate": 0.00025488235294117645,
      "loss": 0.0526,
      "step": 1269
    },
    {
      "epoch": 11.34,
      "grad_norm": 0.12593460083007812,
      "learning_rate": 0.0002548235294117647,
      "loss": 0.0571,
      "step": 1270
    },
    {
      "epoch": 11.35,
      "grad_norm": 0.08111362159252167,
      "learning_rate": 0.00025476470588235294,
      "loss": 0.0439,
      "step": 1271
    },
    {
      "epoch": 11.36,
      "grad_norm": 0.08590488880872726,
      "learning_rate": 0.0002547058823529411,
      "loss": 0.0481,
      "step": 1272
    },
    {
      "epoch": 11.37,
      "grad_norm": 0.07635786384344101,
      "learning_rate": 0.0002546470588235294,
      "loss": 0.0449,
      "step": 1273
    },
    {
      "epoch": 11.38,
      "grad_norm": 0.07801787555217743,
      "learning_rate": 0.0002545882352941176,
      "loss": 0.0419,
      "step": 1274
    },
    {
      "epoch": 11.38,
      "grad_norm": 0.07785259187221527,
      "learning_rate": 0.00025452941176470586,
      "loss": 0.0401,
      "step": 1275
    },
    {
      "epoch": 11.39,
      "grad_norm": 0.14835655689239502,
      "learning_rate": 0.0002544705882352941,
      "loss": 0.0606,
      "step": 1276
    },
    {
      "epoch": 11.4,
      "grad_norm": 0.06783640384674072,
      "learning_rate": 0.00025441176470588234,
      "loss": 0.0444,
      "step": 1277
    },
    {
      "epoch": 11.41,
      "grad_norm": 0.07653181999921799,
      "learning_rate": 0.0002543529411764706,
      "loss": 0.0432,
      "step": 1278
    },
    {
      "epoch": 11.42,
      "grad_norm": 0.08609187602996826,
      "learning_rate": 0.0002542941176470588,
      "loss": 0.0405,
      "step": 1279
    },
    {
      "epoch": 11.43,
      "grad_norm": 0.08816090226173401,
      "learning_rate": 0.000254235294117647,
      "loss": 0.0442,
      "step": 1280
    },
    {
      "epoch": 11.44,
      "grad_norm": 0.0865228921175003,
      "learning_rate": 0.00025417647058823526,
      "loss": 0.0409,
      "step": 1281
    },
    {
      "epoch": 11.45,
      "grad_norm": 0.16418685019016266,
      "learning_rate": 0.0002541176470588235,
      "loss": 0.0623,
      "step": 1282
    },
    {
      "epoch": 11.46,
      "grad_norm": 0.07710708677768707,
      "learning_rate": 0.00025405882352941175,
      "loss": 0.0429,
      "step": 1283
    },
    {
      "epoch": 11.46,
      "grad_norm": 0.0758427307009697,
      "learning_rate": 0.000254,
      "loss": 0.0457,
      "step": 1284
    },
    {
      "epoch": 11.47,
      "grad_norm": 0.07040686160326004,
      "learning_rate": 0.0002539411764705882,
      "loss": 0.0446,
      "step": 1285
    },
    {
      "epoch": 11.48,
      "grad_norm": 0.07687915861606598,
      "learning_rate": 0.0002538823529411765,
      "loss": 0.0491,
      "step": 1286
    },
    {
      "epoch": 11.49,
      "grad_norm": 0.07504872232675552,
      "learning_rate": 0.00025382352941176467,
      "loss": 0.0455,
      "step": 1287
    },
    {
      "epoch": 11.5,
      "grad_norm": 0.10116589814424515,
      "learning_rate": 0.0002537647058823529,
      "loss": 0.0481,
      "step": 1288
    },
    {
      "epoch": 11.51,
      "grad_norm": 0.1314215064048767,
      "learning_rate": 0.00025370588235294116,
      "loss": 0.0669,
      "step": 1289
    },
    {
      "epoch": 11.52,
      "grad_norm": 0.0785100981593132,
      "learning_rate": 0.0002536470588235294,
      "loss": 0.0442,
      "step": 1290
    },
    {
      "epoch": 11.53,
      "grad_norm": 0.07649140805006027,
      "learning_rate": 0.00025358823529411764,
      "loss": 0.0462,
      "step": 1291
    },
    {
      "epoch": 11.54,
      "grad_norm": 0.07404986768960953,
      "learning_rate": 0.00025352941176470583,
      "loss": 0.0394,
      "step": 1292
    },
    {
      "epoch": 11.54,
      "grad_norm": 0.08178161829710007,
      "learning_rate": 0.0002534705882352941,
      "loss": 0.0442,
      "step": 1293
    },
    {
      "epoch": 11.55,
      "grad_norm": 0.11859782785177231,
      "learning_rate": 0.0002534117647058823,
      "loss": 0.0451,
      "step": 1294
    },
    {
      "epoch": 11.56,
      "grad_norm": 0.15042392909526825,
      "learning_rate": 0.00025335294117647056,
      "loss": 0.0684,
      "step": 1295
    },
    {
      "epoch": 11.57,
      "grad_norm": 0.08013623207807541,
      "learning_rate": 0.0002532941176470588,
      "loss": 0.0392,
      "step": 1296
    },
    {
      "epoch": 11.58,
      "grad_norm": 0.08386145532131195,
      "learning_rate": 0.00025323529411764705,
      "loss": 0.0463,
      "step": 1297
    },
    {
      "epoch": 11.59,
      "grad_norm": 0.08066698908805847,
      "learning_rate": 0.00025317647058823524,
      "loss": 0.0444,
      "step": 1298
    },
    {
      "epoch": 11.6,
      "grad_norm": 0.08640114963054657,
      "learning_rate": 0.00025311764705882354,
      "loss": 0.0486,
      "step": 1299
    },
    {
      "epoch": 11.61,
      "grad_norm": 0.09660367667675018,
      "learning_rate": 0.0002530588235294117,
      "loss": 0.0418,
      "step": 1300
    },
    {
      "epoch": 11.61,
      "eval_cer": 0.03582475486050821,
      "eval_loss": 0.17305758595466614,
      "eval_runtime": 22.1946,
      "eval_samples_per_second": 119.038,
      "eval_steps_per_second": 1.892,
      "eval_wer": 0.12292740975803253,
      "step": 1300
    },
    {
      "epoch": 11.62,
      "grad_norm": 0.11291833966970444,
      "learning_rate": 0.00025299999999999997,
      "loss": 0.0451,
      "step": 1301
    },
    {
      "epoch": 11.62,
      "grad_norm": 0.08570745587348938,
      "learning_rate": 0.0002529411764705882,
      "loss": 0.0484,
      "step": 1302
    },
    {
      "epoch": 11.63,
      "grad_norm": 0.08498316258192062,
      "learning_rate": 0.00025288235294117646,
      "loss": 0.0454,
      "step": 1303
    },
    {
      "epoch": 11.64,
      "grad_norm": 0.08248021453619003,
      "learning_rate": 0.0002528235294117647,
      "loss": 0.0424,
      "step": 1304
    },
    {
      "epoch": 11.65,
      "grad_norm": 0.08129890263080597,
      "learning_rate": 0.0002527647058823529,
      "loss": 0.0428,
      "step": 1305
    },
    {
      "epoch": 11.66,
      "grad_norm": 0.07866654545068741,
      "learning_rate": 0.0002527058823529412,
      "loss": 0.0365,
      "step": 1306
    },
    {
      "epoch": 11.67,
      "grad_norm": 0.14105282723903656,
      "learning_rate": 0.0002526470588235294,
      "loss": 0.064,
      "step": 1307
    },
    {
      "epoch": 11.68,
      "grad_norm": 0.07599008828401566,
      "learning_rate": 0.0002525882352941176,
      "loss": 0.0411,
      "step": 1308
    },
    {
      "epoch": 11.69,
      "grad_norm": 0.0781768336892128,
      "learning_rate": 0.00025252941176470586,
      "loss": 0.0431,
      "step": 1309
    },
    {
      "epoch": 11.7,
      "grad_norm": 0.07411547005176544,
      "learning_rate": 0.0002524705882352941,
      "loss": 0.0436,
      "step": 1310
    },
    {
      "epoch": 11.71,
      "grad_norm": 0.09701592475175858,
      "learning_rate": 0.00025241176470588235,
      "loss": 0.0467,
      "step": 1311
    },
    {
      "epoch": 11.71,
      "grad_norm": 0.0785735622048378,
      "learning_rate": 0.0002523529411764706,
      "loss": 0.0419,
      "step": 1312
    },
    {
      "epoch": 11.72,
      "grad_norm": 0.13147878646850586,
      "learning_rate": 0.0002522941176470588,
      "loss": 0.0587,
      "step": 1313
    },
    {
      "epoch": 11.73,
      "grad_norm": 0.11371983587741852,
      "learning_rate": 0.000252235294117647,
      "loss": 0.0559,
      "step": 1314
    },
    {
      "epoch": 11.74,
      "grad_norm": 0.07586345821619034,
      "learning_rate": 0.00025217647058823527,
      "loss": 0.0484,
      "step": 1315
    },
    {
      "epoch": 11.75,
      "grad_norm": 0.07440981268882751,
      "learning_rate": 0.0002521176470588235,
      "loss": 0.0448,
      "step": 1316
    },
    {
      "epoch": 11.76,
      "grad_norm": 0.07169202715158463,
      "learning_rate": 0.00025205882352941176,
      "loss": 0.04,
      "step": 1317
    },
    {
      "epoch": 11.77,
      "grad_norm": 0.08172278851270676,
      "learning_rate": 0.00025199999999999995,
      "loss": 0.04,
      "step": 1318
    },
    {
      "epoch": 11.78,
      "grad_norm": 0.0898115262389183,
      "learning_rate": 0.00025194117647058824,
      "loss": 0.0399,
      "step": 1319
    },
    {
      "epoch": 11.79,
      "grad_norm": 0.10978999733924866,
      "learning_rate": 0.00025188235294117643,
      "loss": 0.0505,
      "step": 1320
    },
    {
      "epoch": 11.79,
      "grad_norm": 0.07988785952329636,
      "learning_rate": 0.0002518235294117647,
      "loss": 0.0439,
      "step": 1321
    },
    {
      "epoch": 11.8,
      "grad_norm": 0.078738734126091,
      "learning_rate": 0.0002517647058823529,
      "loss": 0.0447,
      "step": 1322
    },
    {
      "epoch": 11.81,
      "grad_norm": 0.08227694779634476,
      "learning_rate": 0.00025170588235294116,
      "loss": 0.0452,
      "step": 1323
    },
    {
      "epoch": 11.82,
      "grad_norm": 0.06910045444965363,
      "learning_rate": 0.0002516470588235294,
      "loss": 0.0354,
      "step": 1324
    },
    {
      "epoch": 11.83,
      "grad_norm": 0.07851975411176682,
      "learning_rate": 0.00025158823529411765,
      "loss": 0.0404,
      "step": 1325
    },
    {
      "epoch": 11.84,
      "grad_norm": 0.13973665237426758,
      "learning_rate": 0.00025152941176470584,
      "loss": 0.0622,
      "step": 1326
    },
    {
      "epoch": 11.85,
      "grad_norm": 0.07589449733495712,
      "learning_rate": 0.0002514705882352941,
      "loss": 0.0482,
      "step": 1327
    },
    {
      "epoch": 11.86,
      "grad_norm": 0.07309146970510483,
      "learning_rate": 0.0002514117647058823,
      "loss": 0.0426,
      "step": 1328
    },
    {
      "epoch": 11.87,
      "grad_norm": 0.08154064416885376,
      "learning_rate": 0.00025135294117647057,
      "loss": 0.043,
      "step": 1329
    },
    {
      "epoch": 11.88,
      "grad_norm": 0.0723952054977417,
      "learning_rate": 0.0002512941176470588,
      "loss": 0.0375,
      "step": 1330
    },
    {
      "epoch": 11.88,
      "grad_norm": 0.07835827022790909,
      "learning_rate": 0.000251235294117647,
      "loss": 0.041,
      "step": 1331
    },
    {
      "epoch": 11.89,
      "grad_norm": 0.13197559118270874,
      "learning_rate": 0.0002511764705882353,
      "loss": 0.056,
      "step": 1332
    },
    {
      "epoch": 11.9,
      "grad_norm": 0.08319175988435745,
      "learning_rate": 0.0002511176470588235,
      "loss": 0.0418,
      "step": 1333
    },
    {
      "epoch": 11.91,
      "grad_norm": 0.07897458225488663,
      "learning_rate": 0.00025105882352941173,
      "loss": 0.0479,
      "step": 1334
    },
    {
      "epoch": 11.92,
      "grad_norm": 0.0702613964676857,
      "learning_rate": 0.000251,
      "loss": 0.0446,
      "step": 1335
    },
    {
      "epoch": 11.93,
      "grad_norm": 0.07538347691297531,
      "learning_rate": 0.0002509411764705882,
      "loss": 0.0517,
      "step": 1336
    },
    {
      "epoch": 11.94,
      "grad_norm": 0.0804985836148262,
      "learning_rate": 0.00025088235294117646,
      "loss": 0.0441,
      "step": 1337
    },
    {
      "epoch": 11.95,
      "grad_norm": 0.09818108379840851,
      "learning_rate": 0.0002508235294117647,
      "loss": 0.0455,
      "step": 1338
    },
    {
      "epoch": 11.96,
      "grad_norm": 0.11883387714624405,
      "learning_rate": 0.00025076470588235295,
      "loss": 0.0519,
      "step": 1339
    },
    {
      "epoch": 11.96,
      "grad_norm": 0.07313095778226852,
      "learning_rate": 0.00025070588235294114,
      "loss": 0.0445,
      "step": 1340
    },
    {
      "epoch": 11.97,
      "grad_norm": 0.07245111465454102,
      "learning_rate": 0.0002506470588235294,
      "loss": 0.0427,
      "step": 1341
    },
    {
      "epoch": 11.98,
      "grad_norm": 0.07658955454826355,
      "learning_rate": 0.0002505882352941176,
      "loss": 0.0456,
      "step": 1342
    },
    {
      "epoch": 11.99,
      "grad_norm": 0.08219477534294128,
      "learning_rate": 0.00025052941176470587,
      "loss": 0.0358,
      "step": 1343
    },
    {
      "epoch": 12.0,
      "grad_norm": 0.10890087485313416,
      "learning_rate": 0.0002504705882352941,
      "loss": 0.0458,
      "step": 1344
    },
    {
      "epoch": 12.01,
      "grad_norm": 0.06425008177757263,
      "learning_rate": 0.00025041176470588236,
      "loss": 0.0399,
      "step": 1345
    },
    {
      "epoch": 12.02,
      "grad_norm": 0.06577488034963608,
      "learning_rate": 0.00025035294117647054,
      "loss": 0.0372,
      "step": 1346
    },
    {
      "epoch": 12.03,
      "grad_norm": 0.07996486872434616,
      "learning_rate": 0.0002502941176470588,
      "loss": 0.0445,
      "step": 1347
    },
    {
      "epoch": 12.04,
      "grad_norm": 0.0695604607462883,
      "learning_rate": 0.00025023529411764703,
      "loss": 0.0398,
      "step": 1348
    },
    {
      "epoch": 12.04,
      "grad_norm": 0.06891915202140808,
      "learning_rate": 0.0002501764705882353,
      "loss": 0.0373,
      "step": 1349
    },
    {
      "epoch": 12.05,
      "grad_norm": 0.10478998720645905,
      "learning_rate": 0.0002501176470588235,
      "loss": 0.0454,
      "step": 1350
    },
    {
      "epoch": 12.06,
      "grad_norm": 0.12049918621778488,
      "learning_rate": 0.0002500588235294117,
      "loss": 0.0539,
      "step": 1351
    },
    {
      "epoch": 12.07,
      "grad_norm": 0.07435008138418198,
      "learning_rate": 0.00025,
      "loss": 0.0379,
      "step": 1352
    },
    {
      "epoch": 12.08,
      "grad_norm": 0.08182618021965027,
      "learning_rate": 0.0002499411764705882,
      "loss": 0.0429,
      "step": 1353
    },
    {
      "epoch": 12.09,
      "grad_norm": 0.06991595774888992,
      "learning_rate": 0.00024988235294117644,
      "loss": 0.0384,
      "step": 1354
    },
    {
      "epoch": 12.1,
      "grad_norm": 0.07907950133085251,
      "learning_rate": 0.0002498235294117647,
      "loss": 0.0405,
      "step": 1355
    },
    {
      "epoch": 12.11,
      "grad_norm": 0.0810050368309021,
      "learning_rate": 0.0002497647058823529,
      "loss": 0.0402,
      "step": 1356
    },
    {
      "epoch": 12.12,
      "grad_norm": 0.12017211318016052,
      "learning_rate": 0.00024970588235294117,
      "loss": 0.049,
      "step": 1357
    },
    {
      "epoch": 12.12,
      "grad_norm": 0.06470083445310593,
      "learning_rate": 0.0002496470588235294,
      "loss": 0.0407,
      "step": 1358
    },
    {
      "epoch": 12.13,
      "grad_norm": 0.06676702201366425,
      "learning_rate": 0.0002495882352941176,
      "loss": 0.0375,
      "step": 1359
    },
    {
      "epoch": 12.14,
      "grad_norm": 0.07182049006223679,
      "learning_rate": 0.00024952941176470584,
      "loss": 0.0355,
      "step": 1360
    },
    {
      "epoch": 12.15,
      "grad_norm": 0.06126086786389351,
      "learning_rate": 0.0002494705882352941,
      "loss": 0.0364,
      "step": 1361
    },
    {
      "epoch": 12.16,
      "grad_norm": 0.08764153718948364,
      "learning_rate": 0.00024941176470588233,
      "loss": 0.0487,
      "step": 1362
    },
    {
      "epoch": 12.17,
      "grad_norm": 0.13072869181632996,
      "learning_rate": 0.0002493529411764706,
      "loss": 0.0502,
      "step": 1363
    },
    {
      "epoch": 12.18,
      "grad_norm": 0.07012687623500824,
      "learning_rate": 0.00024929411764705876,
      "loss": 0.0398,
      "step": 1364
    },
    {
      "epoch": 12.19,
      "grad_norm": 0.07666277885437012,
      "learning_rate": 0.00024923529411764706,
      "loss": 0.0376,
      "step": 1365
    },
    {
      "epoch": 12.2,
      "grad_norm": 0.07735677808523178,
      "learning_rate": 0.00024917647058823525,
      "loss": 0.04,
      "step": 1366
    },
    {
      "epoch": 12.21,
      "grad_norm": 0.07179579883813858,
      "learning_rate": 0.00024911764705882355,
      "loss": 0.0383,
      "step": 1367
    },
    {
      "epoch": 12.21,
      "grad_norm": 0.07943487912416458,
      "learning_rate": 0.00024905882352941174,
      "loss": 0.0393,
      "step": 1368
    },
    {
      "epoch": 12.22,
      "grad_norm": 0.1343044489622116,
      "learning_rate": 0.000249,
      "loss": 0.0517,
      "step": 1369
    },
    {
      "epoch": 12.23,
      "grad_norm": 0.0660211443901062,
      "learning_rate": 0.0002489411764705882,
      "loss": 0.0405,
      "step": 1370
    },
    {
      "epoch": 12.24,
      "grad_norm": 0.06342874467372894,
      "learning_rate": 0.00024888235294117647,
      "loss": 0.0351,
      "step": 1371
    },
    {
      "epoch": 12.25,
      "grad_norm": 0.06716018170118332,
      "learning_rate": 0.0002488235294117647,
      "loss": 0.0352,
      "step": 1372
    },
    {
      "epoch": 12.26,
      "grad_norm": 0.08210413157939911,
      "learning_rate": 0.0002487647058823529,
      "loss": 0.0403,
      "step": 1373
    },
    {
      "epoch": 12.27,
      "grad_norm": 0.07900650054216385,
      "learning_rate": 0.00024870588235294114,
      "loss": 0.0394,
      "step": 1374
    },
    {
      "epoch": 12.28,
      "grad_norm": 0.0946178138256073,
      "learning_rate": 0.0002486470588235294,
      "loss": 0.042,
      "step": 1375
    },
    {
      "epoch": 12.29,
      "grad_norm": 0.10450688749551773,
      "learning_rate": 0.00024858823529411763,
      "loss": 0.0461,
      "step": 1376
    },
    {
      "epoch": 12.29,
      "grad_norm": 0.07046644389629364,
      "learning_rate": 0.0002485294117647059,
      "loss": 0.0387,
      "step": 1377
    },
    {
      "epoch": 12.3,
      "grad_norm": 0.06566546112298965,
      "learning_rate": 0.0002484705882352941,
      "loss": 0.0379,
      "step": 1378
    },
    {
      "epoch": 12.31,
      "grad_norm": 0.07577628642320633,
      "learning_rate": 0.0002484117647058823,
      "loss": 0.0381,
      "step": 1379
    },
    {
      "epoch": 12.32,
      "grad_norm": 0.0703827366232872,
      "learning_rate": 0.00024835294117647055,
      "loss": 0.0389,
      "step": 1380
    },
    {
      "epoch": 12.33,
      "grad_norm": 0.08395031094551086,
      "learning_rate": 0.0002482941176470588,
      "loss": 0.0414,
      "step": 1381
    },
    {
      "epoch": 12.34,
      "grad_norm": 0.1367170810699463,
      "learning_rate": 0.00024823529411764704,
      "loss": 0.0547,
      "step": 1382
    },
    {
      "epoch": 12.35,
      "grad_norm": 0.07022585719823837,
      "learning_rate": 0.0002481764705882353,
      "loss": 0.0436,
      "step": 1383
    },
    {
      "epoch": 12.36,
      "grad_norm": 0.07005397975444794,
      "learning_rate": 0.0002481176470588235,
      "loss": 0.0379,
      "step": 1384
    },
    {
      "epoch": 12.37,
      "grad_norm": 0.07170049101114273,
      "learning_rate": 0.00024805882352941177,
      "loss": 0.0361,
      "step": 1385
    },
    {
      "epoch": 12.38,
      "grad_norm": 0.07410118728876114,
      "learning_rate": 0.00024799999999999996,
      "loss": 0.039,
      "step": 1386
    },
    {
      "epoch": 12.38,
      "grad_norm": 0.0882721021771431,
      "learning_rate": 0.0002479411764705882,
      "loss": 0.0349,
      "step": 1387
    },
    {
      "epoch": 12.39,
      "grad_norm": 0.12550528347492218,
      "learning_rate": 0.00024788235294117644,
      "loss": 0.0518,
      "step": 1388
    },
    {
      "epoch": 12.4,
      "grad_norm": 0.0727790966629982,
      "learning_rate": 0.0002478235294117647,
      "loss": 0.0394,
      "step": 1389
    },
    {
      "epoch": 12.41,
      "grad_norm": 0.07850344479084015,
      "learning_rate": 0.00024776470588235293,
      "loss": 0.0386,
      "step": 1390
    },
    {
      "epoch": 12.42,
      "grad_norm": 0.08455193042755127,
      "learning_rate": 0.0002477058823529412,
      "loss": 0.0442,
      "step": 1391
    },
    {
      "epoch": 12.43,
      "grad_norm": 0.07608269900083542,
      "learning_rate": 0.00024764705882352936,
      "loss": 0.0422,
      "step": 1392
    },
    {
      "epoch": 12.44,
      "grad_norm": 0.07037638127803802,
      "learning_rate": 0.0002475882352941176,
      "loss": 0.035,
      "step": 1393
    },
    {
      "epoch": 12.45,
      "grad_norm": 0.1343112289905548,
      "learning_rate": 0.00024752941176470585,
      "loss": 0.0566,
      "step": 1394
    },
    {
      "epoch": 12.46,
      "grad_norm": 0.07384752482175827,
      "learning_rate": 0.0002474705882352941,
      "loss": 0.043,
      "step": 1395
    },
    {
      "epoch": 12.46,
      "grad_norm": 0.0817333459854126,
      "learning_rate": 0.00024741176470588234,
      "loss": 0.0449,
      "step": 1396
    },
    {
      "epoch": 12.47,
      "grad_norm": 0.07389001548290253,
      "learning_rate": 0.0002473529411764706,
      "loss": 0.0395,
      "step": 1397
    },
    {
      "epoch": 12.48,
      "grad_norm": 0.08193815499544144,
      "learning_rate": 0.0002472941176470588,
      "loss": 0.0419,
      "step": 1398
    },
    {
      "epoch": 12.49,
      "grad_norm": 0.06868834048509598,
      "learning_rate": 0.000247235294117647,
      "loss": 0.0352,
      "step": 1399
    },
    {
      "epoch": 12.5,
      "grad_norm": 0.10745295137166977,
      "learning_rate": 0.0002471764705882353,
      "loss": 0.0424,
      "step": 1400
    },
    {
      "epoch": 12.5,
      "eval_cer": 0.03482127085423925,
      "eval_loss": 0.17963390052318573,
      "eval_runtime": 22.1408,
      "eval_samples_per_second": 119.327,
      "eval_steps_per_second": 1.897,
      "eval_wer": 0.11909956366521222,
      "step": 1400
    },
    {
      "epoch": 12.51,
      "grad_norm": 0.11358443647623062,
      "learning_rate": 0.0002471176470588235,
      "loss": 0.0515,
      "step": 1401
    },
    {
      "epoch": 12.52,
      "grad_norm": 0.0776226669549942,
      "learning_rate": 0.00024705882352941174,
      "loss": 0.0415,
      "step": 1402
    },
    {
      "epoch": 12.53,
      "grad_norm": 0.08852792531251907,
      "learning_rate": 0.000247,
      "loss": 0.0367,
      "step": 1403
    },
    {
      "epoch": 12.54,
      "grad_norm": 0.07338748872280121,
      "learning_rate": 0.00024694117647058823,
      "loss": 0.0406,
      "step": 1404
    },
    {
      "epoch": 12.54,
      "grad_norm": 0.0752982348203659,
      "learning_rate": 0.0002468823529411765,
      "loss": 0.0399,
      "step": 1405
    },
    {
      "epoch": 12.55,
      "grad_norm": 0.10211584717035294,
      "learning_rate": 0.00024682352941176466,
      "loss": 0.0399,
      "step": 1406
    },
    {
      "epoch": 12.56,
      "grad_norm": 0.1165146678686142,
      "learning_rate": 0.0002467647058823529,
      "loss": 0.0546,
      "step": 1407
    },
    {
      "epoch": 12.57,
      "grad_norm": 0.07215819507837296,
      "learning_rate": 0.00024670588235294115,
      "loss": 0.0393,
      "step": 1408
    },
    {
      "epoch": 12.58,
      "grad_norm": 0.07633297890424728,
      "learning_rate": 0.0002466470588235294,
      "loss": 0.0428,
      "step": 1409
    },
    {
      "epoch": 12.59,
      "grad_norm": 0.07824792712926865,
      "learning_rate": 0.00024658823529411764,
      "loss": 0.0397,
      "step": 1410
    },
    {
      "epoch": 12.6,
      "grad_norm": 0.07324550300836563,
      "learning_rate": 0.0002465294117647059,
      "loss": 0.043,
      "step": 1411
    },
    {
      "epoch": 12.61,
      "grad_norm": 0.08240360021591187,
      "learning_rate": 0.00024647058823529407,
      "loss": 0.039,
      "step": 1412
    },
    {
      "epoch": 12.62,
      "grad_norm": 0.15403307974338531,
      "learning_rate": 0.00024641176470588237,
      "loss": 0.0539,
      "step": 1413
    },
    {
      "epoch": 12.62,
      "grad_norm": 0.07089138776063919,
      "learning_rate": 0.00024635294117647056,
      "loss": 0.0394,
      "step": 1414
    },
    {
      "epoch": 12.63,
      "grad_norm": 0.06545587629079819,
      "learning_rate": 0.0002462941176470588,
      "loss": 0.0354,
      "step": 1415
    },
    {
      "epoch": 12.64,
      "grad_norm": 0.07407490164041519,
      "learning_rate": 0.00024623529411764704,
      "loss": 0.0414,
      "step": 1416
    },
    {
      "epoch": 12.65,
      "grad_norm": 0.07717980444431305,
      "learning_rate": 0.0002461764705882353,
      "loss": 0.0404,
      "step": 1417
    },
    {
      "epoch": 12.66,
      "grad_norm": 0.07711130380630493,
      "learning_rate": 0.00024611764705882353,
      "loss": 0.0374,
      "step": 1418
    },
    {
      "epoch": 12.67,
      "grad_norm": 0.12361445277929306,
      "learning_rate": 0.0002460588235294117,
      "loss": 0.0503,
      "step": 1419
    },
    {
      "epoch": 12.68,
      "grad_norm": 0.07513695955276489,
      "learning_rate": 0.00024599999999999996,
      "loss": 0.0405,
      "step": 1420
    },
    {
      "epoch": 12.69,
      "grad_norm": 0.07495496422052383,
      "learning_rate": 0.0002459411764705882,
      "loss": 0.0428,
      "step": 1421
    },
    {
      "epoch": 12.7,
      "grad_norm": 0.07376106828451157,
      "learning_rate": 0.00024588235294117645,
      "loss": 0.0372,
      "step": 1422
    },
    {
      "epoch": 12.71,
      "grad_norm": 0.0759698674082756,
      "learning_rate": 0.0002458235294117647,
      "loss": 0.0369,
      "step": 1423
    },
    {
      "epoch": 12.71,
      "grad_norm": 0.06762997061014175,
      "learning_rate": 0.00024576470588235294,
      "loss": 0.0354,
      "step": 1424
    },
    {
      "epoch": 12.72,
      "grad_norm": 0.09362104535102844,
      "learning_rate": 0.0002457058823529411,
      "loss": 0.0424,
      "step": 1425
    },
    {
      "epoch": 12.73,
      "grad_norm": 0.12256449460983276,
      "learning_rate": 0.0002456470588235294,
      "loss": 0.0449,
      "step": 1426
    },
    {
      "epoch": 12.74,
      "grad_norm": 0.07384248822927475,
      "learning_rate": 0.0002455882352941176,
      "loss": 0.0453,
      "step": 1427
    },
    {
      "epoch": 12.75,
      "grad_norm": 0.07044611871242523,
      "learning_rate": 0.00024552941176470586,
      "loss": 0.0397,
      "step": 1428
    },
    {
      "epoch": 12.76,
      "grad_norm": 0.0751928761601448,
      "learning_rate": 0.0002454705882352941,
      "loss": 0.0391,
      "step": 1429
    },
    {
      "epoch": 12.77,
      "grad_norm": 0.07614405453205109,
      "learning_rate": 0.00024541176470588234,
      "loss": 0.0372,
      "step": 1430
    },
    {
      "epoch": 12.78,
      "grad_norm": 0.08169575035572052,
      "learning_rate": 0.0002453529411764706,
      "loss": 0.043,
      "step": 1431
    },
    {
      "epoch": 12.79,
      "grad_norm": 0.16052187979221344,
      "learning_rate": 0.0002452941176470588,
      "loss": 0.0561,
      "step": 1432
    },
    {
      "epoch": 12.79,
      "grad_norm": 0.08795813471078873,
      "learning_rate": 0.00024523529411764707,
      "loss": 0.0384,
      "step": 1433
    },
    {
      "epoch": 12.8,
      "grad_norm": 0.07420073449611664,
      "learning_rate": 0.00024517647058823526,
      "loss": 0.0398,
      "step": 1434
    },
    {
      "epoch": 12.81,
      "grad_norm": 0.07348296046257019,
      "learning_rate": 0.0002451176470588235,
      "loss": 0.0405,
      "step": 1435
    },
    {
      "epoch": 12.82,
      "grad_norm": 0.07852587848901749,
      "learning_rate": 0.00024505882352941175,
      "loss": 0.0398,
      "step": 1436
    },
    {
      "epoch": 12.83,
      "grad_norm": 0.08845631778240204,
      "learning_rate": 0.000245,
      "loss": 0.0409,
      "step": 1437
    },
    {
      "epoch": 12.84,
      "grad_norm": 0.13187581300735474,
      "learning_rate": 0.00024494117647058824,
      "loss": 0.0515,
      "step": 1438
    },
    {
      "epoch": 12.85,
      "grad_norm": 0.07846330851316452,
      "learning_rate": 0.0002448823529411764,
      "loss": 0.041,
      "step": 1439
    },
    {
      "epoch": 12.86,
      "grad_norm": 0.07889274507761002,
      "learning_rate": 0.00024482352941176467,
      "loss": 0.0415,
      "step": 1440
    },
    {
      "epoch": 12.87,
      "grad_norm": 0.0869123712182045,
      "learning_rate": 0.0002447647058823529,
      "loss": 0.0413,
      "step": 1441
    },
    {
      "epoch": 12.88,
      "grad_norm": 0.0797288566827774,
      "learning_rate": 0.00024470588235294116,
      "loss": 0.0452,
      "step": 1442
    },
    {
      "epoch": 12.88,
      "grad_norm": 0.08006476610898972,
      "learning_rate": 0.0002446470588235294,
      "loss": 0.0405,
      "step": 1443
    },
    {
      "epoch": 12.89,
      "grad_norm": 0.1381281018257141,
      "learning_rate": 0.00024458823529411764,
      "loss": 0.0633,
      "step": 1444
    },
    {
      "epoch": 12.9,
      "grad_norm": 0.07537613064050674,
      "learning_rate": 0.00024452941176470583,
      "loss": 0.0411,
      "step": 1445
    },
    {
      "epoch": 12.91,
      "grad_norm": 0.075586698949337,
      "learning_rate": 0.00024447058823529413,
      "loss": 0.0396,
      "step": 1446
    },
    {
      "epoch": 12.92,
      "grad_norm": 0.07694211602210999,
      "learning_rate": 0.0002444117647058823,
      "loss": 0.0417,
      "step": 1447
    },
    {
      "epoch": 12.93,
      "grad_norm": 0.07583875954151154,
      "learning_rate": 0.00024435294117647056,
      "loss": 0.0391,
      "step": 1448
    },
    {
      "epoch": 12.94,
      "grad_norm": 0.08428283780813217,
      "learning_rate": 0.0002442941176470588,
      "loss": 0.039,
      "step": 1449
    },
    {
      "epoch": 12.95,
      "grad_norm": 0.10015322268009186,
      "learning_rate": 0.00024423529411764705,
      "loss": 0.0438,
      "step": 1450
    },
    {
      "epoch": 12.96,
      "grad_norm": 0.09561532735824585,
      "learning_rate": 0.0002441764705882353,
      "loss": 0.0471,
      "step": 1451
    },
    {
      "epoch": 12.96,
      "grad_norm": 0.07340362668037415,
      "learning_rate": 0.0002441176470588235,
      "loss": 0.0399,
      "step": 1452
    },
    {
      "epoch": 12.97,
      "grad_norm": 0.07676450163125992,
      "learning_rate": 0.00024405882352941172,
      "loss": 0.0374,
      "step": 1453
    },
    {
      "epoch": 12.98,
      "grad_norm": 0.09453106671571732,
      "learning_rate": 0.000244,
      "loss": 0.0432,
      "step": 1454
    },
    {
      "epoch": 12.99,
      "grad_norm": 0.09403645992279053,
      "learning_rate": 0.0002439411764705882,
      "loss": 0.0406,
      "step": 1455
    },
    {
      "epoch": 13.0,
      "grad_norm": 0.14890922605991364,
      "learning_rate": 0.00024388235294117645,
      "loss": 0.0434,
      "step": 1456
    },
    {
      "epoch": 13.01,
      "grad_norm": 0.07152199000120163,
      "learning_rate": 0.00024382352941176467,
      "loss": 0.0424,
      "step": 1457
    },
    {
      "epoch": 13.02,
      "grad_norm": 0.08089304715394974,
      "learning_rate": 0.00024376470588235291,
      "loss": 0.0423,
      "step": 1458
    },
    {
      "epoch": 13.03,
      "grad_norm": 0.06755346804857254,
      "learning_rate": 0.00024370588235294116,
      "loss": 0.036,
      "step": 1459
    },
    {
      "epoch": 13.04,
      "grad_norm": 0.0716506764292717,
      "learning_rate": 0.00024364705882352937,
      "loss": 0.0376,
      "step": 1460
    },
    {
      "epoch": 13.04,
      "grad_norm": 0.07225367426872253,
      "learning_rate": 0.00024358823529411764,
      "loss": 0.0424,
      "step": 1461
    },
    {
      "epoch": 13.05,
      "grad_norm": 0.09959781169891357,
      "learning_rate": 0.00024352941176470586,
      "loss": 0.0487,
      "step": 1462
    },
    {
      "epoch": 13.06,
      "grad_norm": 0.10509227961301804,
      "learning_rate": 0.00024347058823529408,
      "loss": 0.0452,
      "step": 1463
    },
    {
      "epoch": 13.07,
      "grad_norm": 0.06608708202838898,
      "learning_rate": 0.00024341176470588235,
      "loss": 0.0387,
      "step": 1464
    },
    {
      "epoch": 13.08,
      "grad_norm": 0.06264564394950867,
      "learning_rate": 0.00024335294117647056,
      "loss": 0.0387,
      "step": 1465
    },
    {
      "epoch": 13.09,
      "grad_norm": 0.07046733051538467,
      "learning_rate": 0.0002432941176470588,
      "loss": 0.0348,
      "step": 1466
    },
    {
      "epoch": 13.1,
      "grad_norm": 0.0691027045249939,
      "learning_rate": 0.00024323529411764705,
      "loss": 0.0363,
      "step": 1467
    },
    {
      "epoch": 13.11,
      "grad_norm": 0.07697401940822601,
      "learning_rate": 0.00024317647058823527,
      "loss": 0.0355,
      "step": 1468
    },
    {
      "epoch": 13.12,
      "grad_norm": 0.12853693962097168,
      "learning_rate": 0.0002431176470588235,
      "loss": 0.0493,
      "step": 1469
    },
    {
      "epoch": 13.12,
      "grad_norm": 0.06530895829200745,
      "learning_rate": 0.00024305882352941173,
      "loss": 0.0376,
      "step": 1470
    },
    {
      "epoch": 13.13,
      "grad_norm": 0.06249760091304779,
      "learning_rate": 0.000243,
      "loss": 0.0357,
      "step": 1471
    },
    {
      "epoch": 13.14,
      "grad_norm": 0.073604054749012,
      "learning_rate": 0.00024294117647058821,
      "loss": 0.0384,
      "step": 1472
    },
    {
      "epoch": 13.15,
      "grad_norm": 0.0683189257979393,
      "learning_rate": 0.00024288235294117643,
      "loss": 0.0387,
      "step": 1473
    },
    {
      "epoch": 13.16,
      "grad_norm": 0.08956163376569748,
      "learning_rate": 0.0002428235294117647,
      "loss": 0.0411,
      "step": 1474
    },
    {
      "epoch": 13.17,
      "grad_norm": 0.13329704105854034,
      "learning_rate": 0.00024276470588235292,
      "loss": 0.0559,
      "step": 1475
    },
    {
      "epoch": 13.18,
      "grad_norm": 0.06661229580640793,
      "learning_rate": 0.00024270588235294113,
      "loss": 0.0371,
      "step": 1476
    },
    {
      "epoch": 13.19,
      "grad_norm": 0.07215193659067154,
      "learning_rate": 0.0002426470588235294,
      "loss": 0.0422,
      "step": 1477
    },
    {
      "epoch": 13.2,
      "grad_norm": 0.06695840507745743,
      "learning_rate": 0.00024258823529411762,
      "loss": 0.0355,
      "step": 1478
    },
    {
      "epoch": 13.21,
      "grad_norm": 0.07014728337526321,
      "learning_rate": 0.00024252941176470586,
      "loss": 0.0336,
      "step": 1479
    },
    {
      "epoch": 13.21,
      "grad_norm": 0.06828466057777405,
      "learning_rate": 0.0002424705882352941,
      "loss": 0.0311,
      "step": 1480
    },
    {
      "epoch": 13.22,
      "grad_norm": 0.13612179458141327,
      "learning_rate": 0.00024241176470588232,
      "loss": 0.0482,
      "step": 1481
    },
    {
      "epoch": 13.23,
      "grad_norm": 0.0706053376197815,
      "learning_rate": 0.00024235294117647057,
      "loss": 0.0381,
      "step": 1482
    },
    {
      "epoch": 13.24,
      "grad_norm": 0.08331095427274704,
      "learning_rate": 0.00024229411764705878,
      "loss": 0.0379,
      "step": 1483
    },
    {
      "epoch": 13.25,
      "grad_norm": 0.06359073519706726,
      "learning_rate": 0.00024223529411764705,
      "loss": 0.0337,
      "step": 1484
    },
    {
      "epoch": 13.26,
      "grad_norm": 0.07826108485460281,
      "learning_rate": 0.00024217647058823527,
      "loss": 0.0356,
      "step": 1485
    },
    {
      "epoch": 13.27,
      "grad_norm": 0.07564547657966614,
      "learning_rate": 0.0002421176470588235,
      "loss": 0.0373,
      "step": 1486
    },
    {
      "epoch": 13.28,
      "grad_norm": 0.10281079262495041,
      "learning_rate": 0.00024205882352941176,
      "loss": 0.0389,
      "step": 1487
    },
    {
      "epoch": 13.29,
      "grad_norm": 0.10683814436197281,
      "learning_rate": 0.00024199999999999997,
      "loss": 0.0477,
      "step": 1488
    },
    {
      "epoch": 13.29,
      "grad_norm": 0.06714615225791931,
      "learning_rate": 0.00024194117647058822,
      "loss": 0.0399,
      "step": 1489
    },
    {
      "epoch": 13.3,
      "grad_norm": 0.06968620419502258,
      "learning_rate": 0.00024188235294117646,
      "loss": 0.0363,
      "step": 1490
    },
    {
      "epoch": 13.31,
      "grad_norm": 0.06511169672012329,
      "learning_rate": 0.00024182352941176468,
      "loss": 0.0415,
      "step": 1491
    },
    {
      "epoch": 13.32,
      "grad_norm": 0.06858762353658676,
      "learning_rate": 0.00024176470588235292,
      "loss": 0.0379,
      "step": 1492
    },
    {
      "epoch": 13.33,
      "grad_norm": 0.08811775594949722,
      "learning_rate": 0.00024170588235294114,
      "loss": 0.0421,
      "step": 1493
    },
    {
      "epoch": 13.34,
      "grad_norm": 0.12302280217409134,
      "learning_rate": 0.0002416470588235294,
      "loss": 0.0596,
      "step": 1494
    },
    {
      "epoch": 13.35,
      "grad_norm": 0.07077848166227341,
      "learning_rate": 0.00024158823529411762,
      "loss": 0.0381,
      "step": 1495
    },
    {
      "epoch": 13.36,
      "grad_norm": 0.08123049885034561,
      "learning_rate": 0.00024152941176470584,
      "loss": 0.0375,
      "step": 1496
    },
    {
      "epoch": 13.37,
      "grad_norm": 0.07264683395624161,
      "learning_rate": 0.0002414705882352941,
      "loss": 0.04,
      "step": 1497
    },
    {
      "epoch": 13.38,
      "grad_norm": 0.07483851909637451,
      "learning_rate": 0.00024141176470588233,
      "loss": 0.04,
      "step": 1498
    },
    {
      "epoch": 13.38,
      "grad_norm": 0.07415857166051865,
      "learning_rate": 0.00024135294117647057,
      "loss": 0.034,
      "step": 1499
    },
    {
      "epoch": 13.39,
      "grad_norm": 0.10655782371759415,
      "learning_rate": 0.0002412941176470588,
      "loss": 0.0469,
      "step": 1500
    },
    {
      "epoch": 13.39,
      "eval_cer": 0.03583227159463756,
      "eval_loss": 0.18480569124221802,
      "eval_runtime": 21.8993,
      "eval_samples_per_second": 120.643,
      "eval_steps_per_second": 1.918,
      "eval_wer": 0.12072590241967474,
      "step": 1500
    },
    {
      "epoch": 13.4,
      "grad_norm": 0.07293380796909332,
      "learning_rate": 0.00024123529411764703,
      "loss": 0.038,
      "step": 1501
    },
    {
      "epoch": 13.41,
      "grad_norm": 0.07702398300170898,
      "learning_rate": 0.00024117647058823527,
      "loss": 0.0401,
      "step": 1502
    },
    {
      "epoch": 13.42,
      "grad_norm": 0.07244368642568588,
      "learning_rate": 0.00024111764705882352,
      "loss": 0.0404,
      "step": 1503
    },
    {
      "epoch": 13.43,
      "grad_norm": 0.06752533465623856,
      "learning_rate": 0.00024105882352941176,
      "loss": 0.035,
      "step": 1504
    },
    {
      "epoch": 13.44,
      "grad_norm": 0.06490998715162277,
      "learning_rate": 0.00024099999999999998,
      "loss": 0.0363,
      "step": 1505
    },
    {
      "epoch": 13.45,
      "grad_norm": 0.11970426887273788,
      "learning_rate": 0.0002409411764705882,
      "loss": 0.0499,
      "step": 1506
    },
    {
      "epoch": 13.46,
      "grad_norm": 0.06805020570755005,
      "learning_rate": 0.00024088235294117646,
      "loss": 0.0369,
      "step": 1507
    },
    {
      "epoch": 13.46,
      "grad_norm": 0.0638478472828865,
      "learning_rate": 0.00024082352941176468,
      "loss": 0.0382,
      "step": 1508
    },
    {
      "epoch": 13.47,
      "grad_norm": 0.05978140980005264,
      "learning_rate": 0.0002407647058823529,
      "loss": 0.0308,
      "step": 1509
    },
    {
      "epoch": 13.48,
      "grad_norm": 0.06912688165903091,
      "learning_rate": 0.00024070588235294117,
      "loss": 0.0346,
      "step": 1510
    },
    {
      "epoch": 13.49,
      "grad_norm": 0.08909628540277481,
      "learning_rate": 0.00024064705882352938,
      "loss": 0.0399,
      "step": 1511
    },
    {
      "epoch": 13.5,
      "grad_norm": 0.10677731782197952,
      "learning_rate": 0.00024058823529411763,
      "loss": 0.0439,
      "step": 1512
    },
    {
      "epoch": 13.51,
      "grad_norm": 0.10795870423316956,
      "learning_rate": 0.00024052941176470587,
      "loss": 0.0493,
      "step": 1513
    },
    {
      "epoch": 13.52,
      "grad_norm": 0.06797435134649277,
      "learning_rate": 0.00024047058823529409,
      "loss": 0.0397,
      "step": 1514
    },
    {
      "epoch": 13.53,
      "grad_norm": 0.07106486707925797,
      "learning_rate": 0.00024041176470588233,
      "loss": 0.0352,
      "step": 1515
    },
    {
      "epoch": 13.54,
      "grad_norm": 0.07276222109794617,
      "learning_rate": 0.00024035294117647057,
      "loss": 0.0387,
      "step": 1516
    },
    {
      "epoch": 13.54,
      "grad_norm": 0.07053562998771667,
      "learning_rate": 0.00024029411764705882,
      "loss": 0.037,
      "step": 1517
    },
    {
      "epoch": 13.55,
      "grad_norm": 0.08380800485610962,
      "learning_rate": 0.00024023529411764703,
      "loss": 0.0375,
      "step": 1518
    },
    {
      "epoch": 13.56,
      "grad_norm": 0.11110502481460571,
      "learning_rate": 0.00024017647058823525,
      "loss": 0.0526,
      "step": 1519
    },
    {
      "epoch": 13.57,
      "grad_norm": 0.06970023363828659,
      "learning_rate": 0.00024011764705882352,
      "loss": 0.0419,
      "step": 1520
    },
    {
      "epoch": 13.58,
      "grad_norm": 0.06434200704097748,
      "learning_rate": 0.00024005882352941174,
      "loss": 0.0392,
      "step": 1521
    },
    {
      "epoch": 13.59,
      "grad_norm": 0.07295121997594833,
      "learning_rate": 0.00023999999999999998,
      "loss": 0.0367,
      "step": 1522
    },
    {
      "epoch": 13.6,
      "grad_norm": 0.07384601980447769,
      "learning_rate": 0.00023994117647058822,
      "loss": 0.0338,
      "step": 1523
    },
    {
      "epoch": 13.61,
      "grad_norm": 0.0841047614812851,
      "learning_rate": 0.00023988235294117644,
      "loss": 0.0347,
      "step": 1524
    },
    {
      "epoch": 13.62,
      "grad_norm": 0.153300479054451,
      "learning_rate": 0.00023982352941176468,
      "loss": 0.0684,
      "step": 1525
    },
    {
      "epoch": 13.62,
      "grad_norm": 0.06421291083097458,
      "learning_rate": 0.00023976470588235293,
      "loss": 0.0375,
      "step": 1526
    },
    {
      "epoch": 13.63,
      "grad_norm": 0.06903710216283798,
      "learning_rate": 0.00023970588235294117,
      "loss": 0.0423,
      "step": 1527
    },
    {
      "epoch": 13.64,
      "grad_norm": 0.06346481293439865,
      "learning_rate": 0.00023964705882352939,
      "loss": 0.0369,
      "step": 1528
    },
    {
      "epoch": 13.65,
      "grad_norm": 0.06991536915302277,
      "learning_rate": 0.0002395882352941176,
      "loss": 0.0377,
      "step": 1529
    },
    {
      "epoch": 13.66,
      "grad_norm": 0.07881798595190048,
      "learning_rate": 0.00023952941176470587,
      "loss": 0.0424,
      "step": 1530
    },
    {
      "epoch": 13.67,
      "grad_norm": 0.14213302731513977,
      "learning_rate": 0.0002394705882352941,
      "loss": 0.0601,
      "step": 1531
    },
    {
      "epoch": 13.68,
      "grad_norm": 0.06558680534362793,
      "learning_rate": 0.00023941176470588236,
      "loss": 0.0355,
      "step": 1532
    },
    {
      "epoch": 13.69,
      "grad_norm": 0.06903904676437378,
      "learning_rate": 0.00023935294117647058,
      "loss": 0.036,
      "step": 1533
    },
    {
      "epoch": 13.7,
      "grad_norm": 0.07653883099555969,
      "learning_rate": 0.0002392941176470588,
      "loss": 0.0415,
      "step": 1534
    },
    {
      "epoch": 13.71,
      "grad_norm": 0.07556813955307007,
      "learning_rate": 0.00023923529411764704,
      "loss": 0.0326,
      "step": 1535
    },
    {
      "epoch": 13.71,
      "grad_norm": 0.08494540303945541,
      "learning_rate": 0.00023917647058823528,
      "loss": 0.0404,
      "step": 1536
    },
    {
      "epoch": 13.72,
      "grad_norm": 0.0914946123957634,
      "learning_rate": 0.00023911764705882352,
      "loss": 0.0392,
      "step": 1537
    },
    {
      "epoch": 13.73,
      "grad_norm": 0.09990125149488449,
      "learning_rate": 0.00023905882352941174,
      "loss": 0.0477,
      "step": 1538
    },
    {
      "epoch": 13.74,
      "grad_norm": 0.06531285494565964,
      "learning_rate": 0.00023899999999999998,
      "loss": 0.0368,
      "step": 1539
    },
    {
      "epoch": 13.75,
      "grad_norm": 0.0672503188252449,
      "learning_rate": 0.00023894117647058823,
      "loss": 0.0355,
      "step": 1540
    },
    {
      "epoch": 13.76,
      "grad_norm": 0.06878861039876938,
      "learning_rate": 0.00023888235294117644,
      "loss": 0.0333,
      "step": 1541
    },
    {
      "epoch": 13.77,
      "grad_norm": 0.07407777011394501,
      "learning_rate": 0.00023882352941176466,
      "loss": 0.0361,
      "step": 1542
    },
    {
      "epoch": 13.78,
      "grad_norm": 0.09538985043764114,
      "learning_rate": 0.00023876470588235293,
      "loss": 0.0405,
      "step": 1543
    },
    {
      "epoch": 13.79,
      "grad_norm": 0.1270383596420288,
      "learning_rate": 0.00023870588235294114,
      "loss": 0.051,
      "step": 1544
    },
    {
      "epoch": 13.79,
      "grad_norm": 0.06746938079595566,
      "learning_rate": 0.00023864705882352942,
      "loss": 0.0371,
      "step": 1545
    },
    {
      "epoch": 13.8,
      "grad_norm": 0.0679408609867096,
      "learning_rate": 0.00023858823529411763,
      "loss": 0.0384,
      "step": 1546
    },
    {
      "epoch": 13.81,
      "grad_norm": 0.06483151018619537,
      "learning_rate": 0.00023852941176470585,
      "loss": 0.0349,
      "step": 1547
    },
    {
      "epoch": 13.82,
      "grad_norm": 0.07324409484863281,
      "learning_rate": 0.0002384705882352941,
      "loss": 0.0382,
      "step": 1548
    },
    {
      "epoch": 13.83,
      "grad_norm": 0.08959255367517471,
      "learning_rate": 0.00023841176470588233,
      "loss": 0.0379,
      "step": 1549
    },
    {
      "epoch": 13.84,
      "grad_norm": 0.1187070682644844,
      "learning_rate": 0.00023835294117647058,
      "loss": 0.0512,
      "step": 1550
    },
    {
      "epoch": 13.85,
      "grad_norm": 0.07361667603254318,
      "learning_rate": 0.0002382941176470588,
      "loss": 0.0415,
      "step": 1551
    },
    {
      "epoch": 13.86,
      "grad_norm": 0.07351993024349213,
      "learning_rate": 0.000238235294117647,
      "loss": 0.0367,
      "step": 1552
    },
    {
      "epoch": 13.87,
      "grad_norm": 0.0712309330701828,
      "learning_rate": 0.00023817647058823528,
      "loss": 0.0398,
      "step": 1553
    },
    {
      "epoch": 13.88,
      "grad_norm": 0.06710576266050339,
      "learning_rate": 0.0002381176470588235,
      "loss": 0.0391,
      "step": 1554
    },
    {
      "epoch": 13.88,
      "grad_norm": 0.08506819605827332,
      "learning_rate": 0.00023805882352941177,
      "loss": 0.0373,
      "step": 1555
    },
    {
      "epoch": 13.89,
      "grad_norm": 0.16108375787734985,
      "learning_rate": 0.00023799999999999998,
      "loss": 0.0627,
      "step": 1556
    },
    {
      "epoch": 13.9,
      "grad_norm": 0.06580577790737152,
      "learning_rate": 0.0002379411764705882,
      "loss": 0.0333,
      "step": 1557
    },
    {
      "epoch": 13.91,
      "grad_norm": 0.07547849416732788,
      "learning_rate": 0.00023788235294117644,
      "loss": 0.0396,
      "step": 1558
    },
    {
      "epoch": 13.92,
      "grad_norm": 0.07966912537813187,
      "learning_rate": 0.0002378235294117647,
      "loss": 0.0391,
      "step": 1559
    },
    {
      "epoch": 13.93,
      "grad_norm": 0.06370997428894043,
      "learning_rate": 0.00023776470588235293,
      "loss": 0.0319,
      "step": 1560
    },
    {
      "epoch": 13.94,
      "grad_norm": 0.078092560172081,
      "learning_rate": 0.00023770588235294115,
      "loss": 0.0402,
      "step": 1561
    },
    {
      "epoch": 13.95,
      "grad_norm": 0.08986909687519073,
      "learning_rate": 0.0002376470588235294,
      "loss": 0.0412,
      "step": 1562
    },
    {
      "epoch": 13.96,
      "grad_norm": 0.10101500153541565,
      "learning_rate": 0.00023758823529411763,
      "loss": 0.0501,
      "step": 1563
    },
    {
      "epoch": 13.96,
      "grad_norm": 0.07079905271530151,
      "learning_rate": 0.00023752941176470585,
      "loss": 0.0386,
      "step": 1564
    },
    {
      "epoch": 13.97,
      "grad_norm": 0.07269637286663055,
      "learning_rate": 0.00023747058823529412,
      "loss": 0.0366,
      "step": 1565
    },
    {
      "epoch": 13.98,
      "grad_norm": 0.07832439988851547,
      "learning_rate": 0.00023741176470588234,
      "loss": 0.0407,
      "step": 1566
    },
    {
      "epoch": 13.99,
      "grad_norm": 0.09018362313508987,
      "learning_rate": 0.00023735294117647055,
      "loss": 0.0439,
      "step": 1567
    },
    {
      "epoch": 14.0,
      "grad_norm": 0.13287939131259918,
      "learning_rate": 0.00023729411764705882,
      "loss": 0.0475,
      "step": 1568
    },
    {
      "epoch": 14.01,
      "grad_norm": 0.06339030712842941,
      "learning_rate": 0.00023723529411764704,
      "loss": 0.0317,
      "step": 1569
    },
    {
      "epoch": 14.02,
      "grad_norm": 0.06684370338916779,
      "learning_rate": 0.00023717647058823526,
      "loss": 0.038,
      "step": 1570
    },
    {
      "epoch": 14.03,
      "grad_norm": 0.06527197360992432,
      "learning_rate": 0.0002371176470588235,
      "loss": 0.0366,
      "step": 1571
    },
    {
      "epoch": 14.04,
      "grad_norm": 0.06343524903059006,
      "learning_rate": 0.00023705882352941174,
      "loss": 0.0365,
      "step": 1572
    },
    {
      "epoch": 14.04,
      "grad_norm": 0.05973381549119949,
      "learning_rate": 0.000237,
      "loss": 0.0328,
      "step": 1573
    },
    {
      "epoch": 14.05,
      "grad_norm": 0.08552396297454834,
      "learning_rate": 0.0002369411764705882,
      "loss": 0.0356,
      "step": 1574
    },
    {
      "epoch": 14.06,
      "grad_norm": 0.12069303542375565,
      "learning_rate": 0.00023688235294117645,
      "loss": 0.0433,
      "step": 1575
    },
    {
      "epoch": 14.07,
      "grad_norm": 0.06969575583934784,
      "learning_rate": 0.0002368235294117647,
      "loss": 0.0355,
      "step": 1576
    },
    {
      "epoch": 14.08,
      "grad_norm": 0.07012692093849182,
      "learning_rate": 0.0002367647058823529,
      "loss": 0.0387,
      "step": 1577
    },
    {
      "epoch": 14.09,
      "grad_norm": 0.06612176448106766,
      "learning_rate": 0.00023670588235294118,
      "loss": 0.033,
      "step": 1578
    },
    {
      "epoch": 14.1,
      "grad_norm": 0.06898734718561172,
      "learning_rate": 0.0002366470588235294,
      "loss": 0.0353,
      "step": 1579
    },
    {
      "epoch": 14.11,
      "grad_norm": 0.0772644579410553,
      "learning_rate": 0.0002365882352941176,
      "loss": 0.0318,
      "step": 1580
    },
    {
      "epoch": 14.12,
      "grad_norm": 0.10841916501522064,
      "learning_rate": 0.00023652941176470588,
      "loss": 0.0451,
      "step": 1581
    },
    {
      "epoch": 14.12,
      "grad_norm": 0.06888675689697266,
      "learning_rate": 0.0002364705882352941,
      "loss": 0.0362,
      "step": 1582
    },
    {
      "epoch": 14.13,
      "grad_norm": 0.06781990081071854,
      "learning_rate": 0.00023641176470588234,
      "loss": 0.0349,
      "step": 1583
    },
    {
      "epoch": 14.14,
      "grad_norm": 0.060208071023225784,
      "learning_rate": 0.00023635294117647056,
      "loss": 0.0331,
      "step": 1584
    },
    {
      "epoch": 14.15,
      "grad_norm": 0.0651528388261795,
      "learning_rate": 0.0002362941176470588,
      "loss": 0.0359,
      "step": 1585
    },
    {
      "epoch": 14.16,
      "grad_norm": 0.0784277617931366,
      "learning_rate": 0.00023623529411764704,
      "loss": 0.0386,
      "step": 1586
    },
    {
      "epoch": 14.17,
      "grad_norm": 0.1297052502632141,
      "learning_rate": 0.00023617647058823526,
      "loss": 0.0465,
      "step": 1587
    },
    {
      "epoch": 14.18,
      "grad_norm": 0.06815207749605179,
      "learning_rate": 0.00023611764705882353,
      "loss": 0.0353,
      "step": 1588
    },
    {
      "epoch": 14.19,
      "grad_norm": 0.06945068389177322,
      "learning_rate": 0.00023605882352941175,
      "loss": 0.0339,
      "step": 1589
    },
    {
      "epoch": 14.2,
      "grad_norm": 0.07188992947340012,
      "learning_rate": 0.00023599999999999996,
      "loss": 0.0353,
      "step": 1590
    },
    {
      "epoch": 14.21,
      "grad_norm": 0.07547110319137573,
      "learning_rate": 0.00023594117647058823,
      "loss": 0.0381,
      "step": 1591
    },
    {
      "epoch": 14.21,
      "grad_norm": 0.07905665785074234,
      "learning_rate": 0.00023588235294117645,
      "loss": 0.0355,
      "step": 1592
    },
    {
      "epoch": 14.22,
      "grad_norm": 0.14025074243545532,
      "learning_rate": 0.0002358235294117647,
      "loss": 0.0597,
      "step": 1593
    },
    {
      "epoch": 14.23,
      "grad_norm": 0.06963734328746796,
      "learning_rate": 0.0002357647058823529,
      "loss": 0.0364,
      "step": 1594
    },
    {
      "epoch": 14.24,
      "grad_norm": 0.06071014329791069,
      "learning_rate": 0.00023570588235294115,
      "loss": 0.0358,
      "step": 1595
    },
    {
      "epoch": 14.25,
      "grad_norm": 0.06164901331067085,
      "learning_rate": 0.0002356470588235294,
      "loss": 0.032,
      "step": 1596
    },
    {
      "epoch": 14.26,
      "grad_norm": 0.06867073476314545,
      "learning_rate": 0.0002355882352941176,
      "loss": 0.0362,
      "step": 1597
    },
    {
      "epoch": 14.27,
      "grad_norm": 0.07247691601514816,
      "learning_rate": 0.00023552941176470588,
      "loss": 0.0364,
      "step": 1598
    },
    {
      "epoch": 14.28,
      "grad_norm": 0.10158004611730576,
      "learning_rate": 0.0002354705882352941,
      "loss": 0.0448,
      "step": 1599
    },
    {
      "epoch": 14.29,
      "grad_norm": 0.10288799554109573,
      "learning_rate": 0.00023541176470588232,
      "loss": 0.0414,
      "step": 1600
    },
    {
      "epoch": 14.29,
      "eval_cer": 0.03673427969016022,
      "eval_loss": 0.18632982671260834,
      "eval_runtime": 22.0305,
      "eval_samples_per_second": 119.925,
      "eval_steps_per_second": 1.906,
      "eval_wer": 0.12126140420468068,
      "step": 1600
    },
    {
      "epoch": 14.29,
      "grad_norm": 0.07562468200922012,
      "learning_rate": 0.0002353529411764706,
      "loss": 0.037,
      "step": 1601
    },
    {
      "epoch": 14.3,
      "grad_norm": 0.0639897957444191,
      "learning_rate": 0.0002352941176470588,
      "loss": 0.0337,
      "step": 1602
    },
    {
      "epoch": 14.31,
      "grad_norm": 0.07470469176769257,
      "learning_rate": 0.00023523529411764702,
      "loss": 0.0376,
      "step": 1603
    },
    {
      "epoch": 14.32,
      "grad_norm": 0.07621467858552933,
      "learning_rate": 0.0002351764705882353,
      "loss": 0.0343,
      "step": 1604
    },
    {
      "epoch": 14.33,
      "grad_norm": 0.09280934184789658,
      "learning_rate": 0.0002351176470588235,
      "loss": 0.0375,
      "step": 1605
    },
    {
      "epoch": 14.34,
      "grad_norm": 0.12475255876779556,
      "learning_rate": 0.00023505882352941175,
      "loss": 0.0496,
      "step": 1606
    },
    {
      "epoch": 14.35,
      "grad_norm": 0.0690474882721901,
      "learning_rate": 0.00023499999999999997,
      "loss": 0.036,
      "step": 1607
    },
    {
      "epoch": 14.36,
      "grad_norm": 0.07725930213928223,
      "learning_rate": 0.0002349411764705882,
      "loss": 0.0397,
      "step": 1608
    },
    {
      "epoch": 14.37,
      "grad_norm": 0.0686090812087059,
      "learning_rate": 0.00023488235294117645,
      "loss": 0.036,
      "step": 1609
    },
    {
      "epoch": 14.38,
      "grad_norm": 0.06346847862005234,
      "learning_rate": 0.00023482352941176467,
      "loss": 0.0362,
      "step": 1610
    },
    {
      "epoch": 14.38,
      "grad_norm": 0.0800243616104126,
      "learning_rate": 0.00023476470588235294,
      "loss": 0.032,
      "step": 1611
    },
    {
      "epoch": 14.39,
      "grad_norm": 0.12318249046802521,
      "learning_rate": 0.00023470588235294116,
      "loss": 0.0464,
      "step": 1612
    },
    {
      "epoch": 14.4,
      "grad_norm": 0.06889480352401733,
      "learning_rate": 0.00023464705882352937,
      "loss": 0.0375,
      "step": 1613
    },
    {
      "epoch": 14.41,
      "grad_norm": 0.07333625853061676,
      "learning_rate": 0.00023458823529411764,
      "loss": 0.0377,
      "step": 1614
    },
    {
      "epoch": 14.42,
      "grad_norm": 0.0626189261674881,
      "learning_rate": 0.00023452941176470586,
      "loss": 0.0328,
      "step": 1615
    },
    {
      "epoch": 14.43,
      "grad_norm": 0.08424117416143417,
      "learning_rate": 0.0002344705882352941,
      "loss": 0.0424,
      "step": 1616
    },
    {
      "epoch": 14.44,
      "grad_norm": 0.0901605561375618,
      "learning_rate": 0.00023441176470588232,
      "loss": 0.0408,
      "step": 1617
    },
    {
      "epoch": 14.45,
      "grad_norm": 0.10998565703630447,
      "learning_rate": 0.00023435294117647056,
      "loss": 0.0472,
      "step": 1618
    },
    {
      "epoch": 14.46,
      "grad_norm": 0.06640747934579849,
      "learning_rate": 0.0002342941176470588,
      "loss": 0.0403,
      "step": 1619
    },
    {
      "epoch": 14.46,
      "grad_norm": 0.07842729985713959,
      "learning_rate": 0.00023423529411764702,
      "loss": 0.0376,
      "step": 1620
    },
    {
      "epoch": 14.47,
      "grad_norm": 0.06519323587417603,
      "learning_rate": 0.0002341764705882353,
      "loss": 0.0377,
      "step": 1621
    },
    {
      "epoch": 14.48,
      "grad_norm": 0.06695389747619629,
      "learning_rate": 0.0002341176470588235,
      "loss": 0.0344,
      "step": 1622
    },
    {
      "epoch": 14.49,
      "grad_norm": 0.06938100606203079,
      "learning_rate": 0.00023405882352941173,
      "loss": 0.0341,
      "step": 1623
    },
    {
      "epoch": 14.5,
      "grad_norm": 0.11317170411348343,
      "learning_rate": 0.000234,
      "loss": 0.0471,
      "step": 1624
    },
    {
      "epoch": 14.51,
      "grad_norm": 0.09052317589521408,
      "learning_rate": 0.0002339411764705882,
      "loss": 0.0425,
      "step": 1625
    },
    {
      "epoch": 14.52,
      "grad_norm": 0.06292369216680527,
      "learning_rate": 0.00023388235294117646,
      "loss": 0.0324,
      "step": 1626
    },
    {
      "epoch": 14.53,
      "grad_norm": 0.06844203919172287,
      "learning_rate": 0.0002338235294117647,
      "loss": 0.0318,
      "step": 1627
    },
    {
      "epoch": 14.54,
      "grad_norm": 0.07209029793739319,
      "learning_rate": 0.00023376470588235292,
      "loss": 0.0321,
      "step": 1628
    },
    {
      "epoch": 14.54,
      "grad_norm": 0.07459453493356705,
      "learning_rate": 0.00023370588235294116,
      "loss": 0.0373,
      "step": 1629
    },
    {
      "epoch": 14.55,
      "grad_norm": 0.07792117446660995,
      "learning_rate": 0.00023364705882352938,
      "loss": 0.0334,
      "step": 1630
    },
    {
      "epoch": 14.56,
      "grad_norm": 0.10604255646467209,
      "learning_rate": 0.00023358823529411765,
      "loss": 0.0535,
      "step": 1631
    },
    {
      "epoch": 14.57,
      "grad_norm": 0.06339162588119507,
      "learning_rate": 0.00023352941176470586,
      "loss": 0.0334,
      "step": 1632
    },
    {
      "epoch": 14.58,
      "grad_norm": 0.06921108067035675,
      "learning_rate": 0.00023347058823529408,
      "loss": 0.0372,
      "step": 1633
    },
    {
      "epoch": 14.59,
      "grad_norm": 0.058756157755851746,
      "learning_rate": 0.00023341176470588235,
      "loss": 0.0315,
      "step": 1634
    },
    {
      "epoch": 14.6,
      "grad_norm": 0.07588847726583481,
      "learning_rate": 0.00023335294117647057,
      "loss": 0.0353,
      "step": 1635
    },
    {
      "epoch": 14.61,
      "grad_norm": 0.08001239597797394,
      "learning_rate": 0.00023329411764705878,
      "loss": 0.0318,
      "step": 1636
    },
    {
      "epoch": 14.62,
      "grad_norm": 0.11674462258815765,
      "learning_rate": 0.00023323529411764705,
      "loss": 0.0436,
      "step": 1637
    },
    {
      "epoch": 14.62,
      "grad_norm": 0.07083316892385483,
      "learning_rate": 0.00023317647058823527,
      "loss": 0.0394,
      "step": 1638
    },
    {
      "epoch": 14.63,
      "grad_norm": 0.07271532714366913,
      "learning_rate": 0.0002331176470588235,
      "loss": 0.0364,
      "step": 1639
    },
    {
      "epoch": 14.64,
      "grad_norm": 0.07413634657859802,
      "learning_rate": 0.00023305882352941176,
      "loss": 0.0378,
      "step": 1640
    },
    {
      "epoch": 14.65,
      "grad_norm": 0.07597372680902481,
      "learning_rate": 0.00023299999999999997,
      "loss": 0.0334,
      "step": 1641
    },
    {
      "epoch": 14.66,
      "grad_norm": 0.06561561673879623,
      "learning_rate": 0.00023294117647058821,
      "loss": 0.0337,
      "step": 1642
    },
    {
      "epoch": 14.67,
      "grad_norm": 0.12736088037490845,
      "learning_rate": 0.00023288235294117643,
      "loss": 0.0436,
      "step": 1643
    },
    {
      "epoch": 14.68,
      "grad_norm": 0.06355338543653488,
      "learning_rate": 0.0002328235294117647,
      "loss": 0.0367,
      "step": 1644
    },
    {
      "epoch": 14.69,
      "grad_norm": 0.06789432466030121,
      "learning_rate": 0.00023276470588235292,
      "loss": 0.0348,
      "step": 1645
    },
    {
      "epoch": 14.7,
      "grad_norm": 0.0628594234585762,
      "learning_rate": 0.00023270588235294113,
      "loss": 0.0351,
      "step": 1646
    },
    {
      "epoch": 14.71,
      "grad_norm": 0.0669761672616005,
      "learning_rate": 0.0002326470588235294,
      "loss": 0.0358,
      "step": 1647
    },
    {
      "epoch": 14.71,
      "grad_norm": 0.0717419907450676,
      "learning_rate": 0.00023258823529411762,
      "loss": 0.032,
      "step": 1648
    },
    {
      "epoch": 14.72,
      "grad_norm": 0.08335498720407486,
      "learning_rate": 0.00023252941176470586,
      "loss": 0.0329,
      "step": 1649
    },
    {
      "epoch": 14.73,
      "grad_norm": 0.13398894667625427,
      "learning_rate": 0.0002324705882352941,
      "loss": 0.0498,
      "step": 1650
    },
    {
      "epoch": 14.74,
      "grad_norm": 0.06331564486026764,
      "learning_rate": 0.00023241176470588232,
      "loss": 0.0348,
      "step": 1651
    },
    {
      "epoch": 14.75,
      "grad_norm": 0.07277001440525055,
      "learning_rate": 0.00023235294117647057,
      "loss": 0.039,
      "step": 1652
    },
    {
      "epoch": 14.76,
      "grad_norm": 0.07185853272676468,
      "learning_rate": 0.00023229411764705878,
      "loss": 0.0409,
      "step": 1653
    },
    {
      "epoch": 14.77,
      "grad_norm": 0.06759529560804367,
      "learning_rate": 0.00023223529411764705,
      "loss": 0.0334,
      "step": 1654
    },
    {
      "epoch": 14.78,
      "grad_norm": 0.0870920717716217,
      "learning_rate": 0.00023217647058823527,
      "loss": 0.0356,
      "step": 1655
    },
    {
      "epoch": 14.79,
      "grad_norm": 0.13036197423934937,
      "learning_rate": 0.0002321176470588235,
      "loss": 0.0565,
      "step": 1656
    },
    {
      "epoch": 14.79,
      "grad_norm": 0.05877867713570595,
      "learning_rate": 0.00023205882352941176,
      "loss": 0.0314,
      "step": 1657
    },
    {
      "epoch": 14.8,
      "grad_norm": 0.06166622415184975,
      "learning_rate": 0.00023199999999999997,
      "loss": 0.038,
      "step": 1658
    },
    {
      "epoch": 14.81,
      "grad_norm": 0.06303553283214569,
      "learning_rate": 0.00023194117647058822,
      "loss": 0.0352,
      "step": 1659
    },
    {
      "epoch": 14.82,
      "grad_norm": 0.07221624255180359,
      "learning_rate": 0.00023188235294117646,
      "loss": 0.0357,
      "step": 1660
    },
    {
      "epoch": 14.83,
      "grad_norm": 0.07816413044929504,
      "learning_rate": 0.00023182352941176468,
      "loss": 0.0322,
      "step": 1661
    },
    {
      "epoch": 14.84,
      "grad_norm": 0.13044504821300507,
      "learning_rate": 0.00023176470588235292,
      "loss": 0.0458,
      "step": 1662
    },
    {
      "epoch": 14.85,
      "grad_norm": 0.06987060606479645,
      "learning_rate": 0.00023170588235294116,
      "loss": 0.0382,
      "step": 1663
    },
    {
      "epoch": 14.86,
      "grad_norm": 0.0673326849937439,
      "learning_rate": 0.0002316470588235294,
      "loss": 0.0356,
      "step": 1664
    },
    {
      "epoch": 14.87,
      "grad_norm": 0.08185671269893646,
      "learning_rate": 0.00023158823529411762,
      "loss": 0.0375,
      "step": 1665
    },
    {
      "epoch": 14.88,
      "grad_norm": 0.07717020064592361,
      "learning_rate": 0.00023152941176470584,
      "loss": 0.0369,
      "step": 1666
    },
    {
      "epoch": 14.88,
      "grad_norm": 0.06798449158668518,
      "learning_rate": 0.0002314705882352941,
      "loss": 0.0332,
      "step": 1667
    },
    {
      "epoch": 14.89,
      "grad_norm": 0.1054532453417778,
      "learning_rate": 0.00023141176470588233,
      "loss": 0.0397,
      "step": 1668
    },
    {
      "epoch": 14.9,
      "grad_norm": 0.06646037101745605,
      "learning_rate": 0.00023135294117647054,
      "loss": 0.0363,
      "step": 1669
    },
    {
      "epoch": 14.91,
      "grad_norm": 0.0664471909403801,
      "learning_rate": 0.00023129411764705881,
      "loss": 0.0381,
      "step": 1670
    },
    {
      "epoch": 14.92,
      "grad_norm": 0.07188865542411804,
      "learning_rate": 0.00023123529411764703,
      "loss": 0.0356,
      "step": 1671
    },
    {
      "epoch": 14.93,
      "grad_norm": 0.06646988540887833,
      "learning_rate": 0.00023117647058823527,
      "loss": 0.037,
      "step": 1672
    },
    {
      "epoch": 14.94,
      "grad_norm": 0.07141093164682388,
      "learning_rate": 0.00023111764705882352,
      "loss": 0.0335,
      "step": 1673
    },
    {
      "epoch": 14.95,
      "grad_norm": 0.09563656896352768,
      "learning_rate": 0.00023105882352941173,
      "loss": 0.0436,
      "step": 1674
    },
    {
      "epoch": 14.96,
      "grad_norm": 0.11324489116668701,
      "learning_rate": 0.00023099999999999998,
      "loss": 0.0485,
      "step": 1675
    },
    {
      "epoch": 14.96,
      "grad_norm": 0.07574322819709778,
      "learning_rate": 0.00023094117647058822,
      "loss": 0.0392,
      "step": 1676
    },
    {
      "epoch": 14.97,
      "grad_norm": 0.06677575409412384,
      "learning_rate": 0.00023088235294117646,
      "loss": 0.0346,
      "step": 1677
    },
    {
      "epoch": 14.98,
      "grad_norm": 0.07006186246871948,
      "learning_rate": 0.00023082352941176468,
      "loss": 0.033,
      "step": 1678
    },
    {
      "epoch": 14.99,
      "grad_norm": 0.08206184208393097,
      "learning_rate": 0.0002307647058823529,
      "loss": 0.0361,
      "step": 1679
    },
    {
      "epoch": 15.0,
      "grad_norm": 0.10202016681432724,
      "learning_rate": 0.00023070588235294117,
      "loss": 0.0422,
      "step": 1680
    },
    {
      "epoch": 15.01,
      "grad_norm": 0.06775103509426117,
      "learning_rate": 0.00023064705882352938,
      "loss": 0.0414,
      "step": 1681
    },
    {
      "epoch": 15.02,
      "grad_norm": 0.059331733733415604,
      "learning_rate": 0.00023058823529411763,
      "loss": 0.0315,
      "step": 1682
    },
    {
      "epoch": 15.03,
      "grad_norm": 0.06496758759021759,
      "learning_rate": 0.00023052941176470587,
      "loss": 0.0319,
      "step": 1683
    },
    {
      "epoch": 15.04,
      "grad_norm": 0.06576515734195709,
      "learning_rate": 0.0002304705882352941,
      "loss": 0.0325,
      "step": 1684
    },
    {
      "epoch": 15.04,
      "grad_norm": 0.07281138747930527,
      "learning_rate": 0.00023041176470588233,
      "loss": 0.0342,
      "step": 1685
    },
    {
      "epoch": 15.05,
      "grad_norm": 0.10825911909341812,
      "learning_rate": 0.00023035294117647057,
      "loss": 0.0447,
      "step": 1686
    },
    {
      "epoch": 15.06,
      "grad_norm": 0.11847017705440521,
      "learning_rate": 0.00023029411764705882,
      "loss": 0.0387,
      "step": 1687
    },
    {
      "epoch": 15.07,
      "grad_norm": 0.06485091149806976,
      "learning_rate": 0.00023023529411764703,
      "loss": 0.0318,
      "step": 1688
    },
    {
      "epoch": 15.08,
      "grad_norm": 0.0644741952419281,
      "learning_rate": 0.00023017647058823525,
      "loss": 0.0358,
      "step": 1689
    },
    {
      "epoch": 15.09,
      "grad_norm": 0.06077628210186958,
      "learning_rate": 0.00023011764705882352,
      "loss": 0.0286,
      "step": 1690
    },
    {
      "epoch": 15.1,
      "grad_norm": 0.06924694031476974,
      "learning_rate": 0.00023005882352941174,
      "loss": 0.0362,
      "step": 1691
    },
    {
      "epoch": 15.11,
      "grad_norm": 0.0772029235959053,
      "learning_rate": 0.00023,
      "loss": 0.0324,
      "step": 1692
    },
    {
      "epoch": 15.12,
      "grad_norm": 0.10378186404705048,
      "learning_rate": 0.00022994117647058822,
      "loss": 0.0486,
      "step": 1693
    },
    {
      "epoch": 15.12,
      "grad_norm": 0.0626181811094284,
      "learning_rate": 0.00022988235294117644,
      "loss": 0.0342,
      "step": 1694
    },
    {
      "epoch": 15.13,
      "grad_norm": 0.06330876797437668,
      "learning_rate": 0.00022982352941176468,
      "loss": 0.034,
      "step": 1695
    },
    {
      "epoch": 15.14,
      "grad_norm": 0.059350576251745224,
      "learning_rate": 0.00022976470588235293,
      "loss": 0.0304,
      "step": 1696
    },
    {
      "epoch": 15.15,
      "grad_norm": 0.06034482270479202,
      "learning_rate": 0.00022970588235294117,
      "loss": 0.0312,
      "step": 1697
    },
    {
      "epoch": 15.16,
      "grad_norm": 0.07836218178272247,
      "learning_rate": 0.00022964705882352939,
      "loss": 0.0391,
      "step": 1698
    },
    {
      "epoch": 15.17,
      "grad_norm": 0.12190468609333038,
      "learning_rate": 0.00022958823529411763,
      "loss": 0.0511,
      "step": 1699
    },
    {
      "epoch": 15.18,
      "grad_norm": 0.06269179284572601,
      "learning_rate": 0.00022952941176470587,
      "loss": 0.0338,
      "step": 1700
    },
    {
      "epoch": 15.18,
      "eval_cer": 0.03474610351294569,
      "eval_loss": 0.1889059841632843,
      "eval_runtime": 22.7906,
      "eval_samples_per_second": 115.925,
      "eval_steps_per_second": 1.843,
      "eval_wer": 0.11771122570408568,
      "step": 1700
    },
    {
      "epoch": 15.19,
      "grad_norm": 0.06443001329898834,
      "learning_rate": 0.0002294705882352941,
      "loss": 0.0341,
      "step": 1701
    },
    {
      "epoch": 15.2,
      "grad_norm": 0.06785762310028076,
      "learning_rate": 0.0002294117647058823,
      "loss": 0.0317,
      "step": 1702
    },
    {
      "epoch": 15.21,
      "grad_norm": 0.0721161812543869,
      "learning_rate": 0.00022935294117647058,
      "loss": 0.0297,
      "step": 1703
    },
    {
      "epoch": 15.21,
      "grad_norm": 0.08249220252037048,
      "learning_rate": 0.0002292941176470588,
      "loss": 0.0326,
      "step": 1704
    },
    {
      "epoch": 15.22,
      "grad_norm": 0.1352924406528473,
      "learning_rate": 0.00022923529411764706,
      "loss": 0.053,
      "step": 1705
    },
    {
      "epoch": 15.23,
      "grad_norm": 0.06891375035047531,
      "learning_rate": 0.00022917647058823528,
      "loss": 0.0354,
      "step": 1706
    },
    {
      "epoch": 15.24,
      "grad_norm": 0.06851013004779816,
      "learning_rate": 0.0002291176470588235,
      "loss": 0.0357,
      "step": 1707
    },
    {
      "epoch": 15.25,
      "grad_norm": 0.06332221627235413,
      "learning_rate": 0.00022905882352941174,
      "loss": 0.032,
      "step": 1708
    },
    {
      "epoch": 15.26,
      "grad_norm": 0.06636039167642593,
      "learning_rate": 0.00022899999999999998,
      "loss": 0.0337,
      "step": 1709
    },
    {
      "epoch": 15.27,
      "grad_norm": 0.0747261643409729,
      "learning_rate": 0.00022894117647058823,
      "loss": 0.0349,
      "step": 1710
    },
    {
      "epoch": 15.28,
      "grad_norm": 0.09190411120653152,
      "learning_rate": 0.00022888235294117644,
      "loss": 0.036,
      "step": 1711
    },
    {
      "epoch": 15.29,
      "grad_norm": 0.11545003950595856,
      "learning_rate": 0.00022882352941176466,
      "loss": 0.0504,
      "step": 1712
    },
    {
      "epoch": 15.29,
      "grad_norm": 0.06128847226500511,
      "learning_rate": 0.00022876470588235293,
      "loss": 0.0334,
      "step": 1713
    },
    {
      "epoch": 15.3,
      "grad_norm": 0.06289771944284439,
      "learning_rate": 0.00022870588235294115,
      "loss": 0.0327,
      "step": 1714
    },
    {
      "epoch": 15.31,
      "grad_norm": 0.06811051070690155,
      "learning_rate": 0.00022864705882352942,
      "loss": 0.0327,
      "step": 1715
    },
    {
      "epoch": 15.32,
      "grad_norm": 0.07467852532863617,
      "learning_rate": 0.00022858823529411763,
      "loss": 0.0316,
      "step": 1716
    },
    {
      "epoch": 15.33,
      "grad_norm": 0.07979732006788254,
      "learning_rate": 0.00022852941176470585,
      "loss": 0.0342,
      "step": 1717
    },
    {
      "epoch": 15.34,
      "grad_norm": 0.11657968163490295,
      "learning_rate": 0.0002284705882352941,
      "loss": 0.043,
      "step": 1718
    },
    {
      "epoch": 15.35,
      "grad_norm": 0.06508392840623856,
      "learning_rate": 0.00022841176470588234,
      "loss": 0.0334,
      "step": 1719
    },
    {
      "epoch": 15.36,
      "grad_norm": 0.07240886241197586,
      "learning_rate": 0.00022835294117647058,
      "loss": 0.0364,
      "step": 1720
    },
    {
      "epoch": 15.37,
      "grad_norm": 0.07049129158258438,
      "learning_rate": 0.0002282941176470588,
      "loss": 0.0333,
      "step": 1721
    },
    {
      "epoch": 15.38,
      "grad_norm": 0.07092287391424179,
      "learning_rate": 0.00022823529411764704,
      "loss": 0.0347,
      "step": 1722
    },
    {
      "epoch": 15.38,
      "grad_norm": 0.0840243473649025,
      "learning_rate": 0.00022817647058823528,
      "loss": 0.0387,
      "step": 1723
    },
    {
      "epoch": 15.39,
      "grad_norm": 0.1114409863948822,
      "learning_rate": 0.0002281176470588235,
      "loss": 0.0451,
      "step": 1724
    },
    {
      "epoch": 15.4,
      "grad_norm": 0.06822095811367035,
      "learning_rate": 0.00022805882352941177,
      "loss": 0.0366,
      "step": 1725
    },
    {
      "epoch": 15.41,
      "grad_norm": 0.06745601445436478,
      "learning_rate": 0.00022799999999999999,
      "loss": 0.0322,
      "step": 1726
    },
    {
      "epoch": 15.42,
      "grad_norm": 0.06213675066828728,
      "learning_rate": 0.0002279411764705882,
      "loss": 0.0301,
      "step": 1727
    },
    {
      "epoch": 15.43,
      "grad_norm": 0.06839065253734589,
      "learning_rate": 0.00022788235294117647,
      "loss": 0.0324,
      "step": 1728
    },
    {
      "epoch": 15.44,
      "grad_norm": 0.07828672975301743,
      "learning_rate": 0.0002278235294117647,
      "loss": 0.0367,
      "step": 1729
    },
    {
      "epoch": 15.45,
      "grad_norm": 0.11986172199249268,
      "learning_rate": 0.0002277647058823529,
      "loss": 0.044,
      "step": 1730
    },
    {
      "epoch": 15.46,
      "grad_norm": 0.06630124896764755,
      "learning_rate": 0.00022770588235294115,
      "loss": 0.0362,
      "step": 1731
    },
    {
      "epoch": 15.46,
      "grad_norm": 0.07336899638175964,
      "learning_rate": 0.0002276470588235294,
      "loss": 0.0379,
      "step": 1732
    },
    {
      "epoch": 15.47,
      "grad_norm": 0.0715327188372612,
      "learning_rate": 0.00022758823529411764,
      "loss": 0.0394,
      "step": 1733
    },
    {
      "epoch": 15.48,
      "grad_norm": 0.07718680799007416,
      "learning_rate": 0.00022752941176470585,
      "loss": 0.0346,
      "step": 1734
    },
    {
      "epoch": 15.49,
      "grad_norm": 0.07004740089178085,
      "learning_rate": 0.0002274705882352941,
      "loss": 0.0297,
      "step": 1735
    },
    {
      "epoch": 15.5,
      "grad_norm": 0.08288314938545227,
      "learning_rate": 0.00022741176470588234,
      "loss": 0.0292,
      "step": 1736
    },
    {
      "epoch": 15.51,
      "grad_norm": 0.08451540023088455,
      "learning_rate": 0.00022735294117647055,
      "loss": 0.0353,
      "step": 1737
    },
    {
      "epoch": 15.52,
      "grad_norm": 0.06629110872745514,
      "learning_rate": 0.00022729411764705883,
      "loss": 0.0311,
      "step": 1738
    },
    {
      "epoch": 15.53,
      "grad_norm": 0.06104528158903122,
      "learning_rate": 0.00022723529411764704,
      "loss": 0.0318,
      "step": 1739
    },
    {
      "epoch": 15.54,
      "grad_norm": 0.07883122563362122,
      "learning_rate": 0.00022717647058823526,
      "loss": 0.0357,
      "step": 1740
    },
    {
      "epoch": 15.54,
      "grad_norm": 0.06872067600488663,
      "learning_rate": 0.00022711764705882353,
      "loss": 0.0367,
      "step": 1741
    },
    {
      "epoch": 15.55,
      "grad_norm": 0.08901038765907288,
      "learning_rate": 0.00022705882352941174,
      "loss": 0.0344,
      "step": 1742
    },
    {
      "epoch": 15.56,
      "grad_norm": 0.14025089144706726,
      "learning_rate": 0.000227,
      "loss": 0.048,
      "step": 1743
    },
    {
      "epoch": 15.57,
      "grad_norm": 0.06312990188598633,
      "learning_rate": 0.0002269411764705882,
      "loss": 0.0316,
      "step": 1744
    },
    {
      "epoch": 15.58,
      "grad_norm": 0.06658343970775604,
      "learning_rate": 0.00022688235294117645,
      "loss": 0.0355,
      "step": 1745
    },
    {
      "epoch": 15.59,
      "grad_norm": 0.07009308785200119,
      "learning_rate": 0.0002268235294117647,
      "loss": 0.0325,
      "step": 1746
    },
    {
      "epoch": 15.6,
      "grad_norm": 0.06212510168552399,
      "learning_rate": 0.0002267647058823529,
      "loss": 0.0336,
      "step": 1747
    },
    {
      "epoch": 15.61,
      "grad_norm": 0.08150997757911682,
      "learning_rate": 0.00022670588235294118,
      "loss": 0.0371,
      "step": 1748
    },
    {
      "epoch": 15.62,
      "grad_norm": 0.10127957910299301,
      "learning_rate": 0.0002266470588235294,
      "loss": 0.0421,
      "step": 1749
    },
    {
      "epoch": 15.62,
      "grad_norm": 0.06718476861715317,
      "learning_rate": 0.0002265882352941176,
      "loss": 0.0366,
      "step": 1750
    },
    {
      "epoch": 15.63,
      "grad_norm": 0.0708250179886818,
      "learning_rate": 0.00022652941176470588,
      "loss": 0.0364,
      "step": 1751
    },
    {
      "epoch": 15.64,
      "grad_norm": 0.060382526367902756,
      "learning_rate": 0.0002264705882352941,
      "loss": 0.0319,
      "step": 1752
    },
    {
      "epoch": 15.65,
      "grad_norm": 0.07351882755756378,
      "learning_rate": 0.00022641176470588234,
      "loss": 0.0334,
      "step": 1753
    },
    {
      "epoch": 15.66,
      "grad_norm": 0.08200761675834656,
      "learning_rate": 0.00022635294117647056,
      "loss": 0.0326,
      "step": 1754
    },
    {
      "epoch": 15.67,
      "grad_norm": 0.13429179787635803,
      "learning_rate": 0.0002262941176470588,
      "loss": 0.0466,
      "step": 1755
    },
    {
      "epoch": 15.68,
      "grad_norm": 0.06309697031974792,
      "learning_rate": 0.00022623529411764704,
      "loss": 0.033,
      "step": 1756
    },
    {
      "epoch": 15.69,
      "grad_norm": 0.0706702172756195,
      "learning_rate": 0.00022617647058823526,
      "loss": 0.0344,
      "step": 1757
    },
    {
      "epoch": 15.7,
      "grad_norm": 0.07139620929956436,
      "learning_rate": 0.00022611764705882353,
      "loss": 0.0353,
      "step": 1758
    },
    {
      "epoch": 15.71,
      "grad_norm": 0.06745688617229462,
      "learning_rate": 0.00022605882352941175,
      "loss": 0.0311,
      "step": 1759
    },
    {
      "epoch": 15.71,
      "grad_norm": 0.08064214885234833,
      "learning_rate": 0.00022599999999999996,
      "loss": 0.034,
      "step": 1760
    },
    {
      "epoch": 15.72,
      "grad_norm": 0.10125376284122467,
      "learning_rate": 0.00022594117647058823,
      "loss": 0.0384,
      "step": 1761
    },
    {
      "epoch": 15.73,
      "grad_norm": 0.13965125381946564,
      "learning_rate": 0.00022588235294117645,
      "loss": 0.0414,
      "step": 1762
    },
    {
      "epoch": 15.74,
      "grad_norm": 0.06126793846487999,
      "learning_rate": 0.00022582352941176467,
      "loss": 0.0335,
      "step": 1763
    },
    {
      "epoch": 15.75,
      "grad_norm": 0.06566361337900162,
      "learning_rate": 0.00022576470588235294,
      "loss": 0.0339,
      "step": 1764
    },
    {
      "epoch": 15.76,
      "grad_norm": 0.0652477964758873,
      "learning_rate": 0.00022570588235294115,
      "loss": 0.0319,
      "step": 1765
    },
    {
      "epoch": 15.77,
      "grad_norm": 0.07817530632019043,
      "learning_rate": 0.0002256470588235294,
      "loss": 0.0336,
      "step": 1766
    },
    {
      "epoch": 15.78,
      "grad_norm": 0.09044196456670761,
      "learning_rate": 0.00022558823529411761,
      "loss": 0.0346,
      "step": 1767
    },
    {
      "epoch": 15.79,
      "grad_norm": 0.12936484813690186,
      "learning_rate": 0.00022552941176470586,
      "loss": 0.0455,
      "step": 1768
    },
    {
      "epoch": 15.79,
      "grad_norm": 0.06324309855699539,
      "learning_rate": 0.0002254705882352941,
      "loss": 0.0333,
      "step": 1769
    },
    {
      "epoch": 15.8,
      "grad_norm": 0.07502604275941849,
      "learning_rate": 0.00022541176470588232,
      "loss": 0.033,
      "step": 1770
    },
    {
      "epoch": 15.81,
      "grad_norm": 0.07541787624359131,
      "learning_rate": 0.0002253529411764706,
      "loss": 0.0348,
      "step": 1771
    },
    {
      "epoch": 15.82,
      "grad_norm": 0.06684520840644836,
      "learning_rate": 0.0002252941176470588,
      "loss": 0.0362,
      "step": 1772
    },
    {
      "epoch": 15.83,
      "grad_norm": 0.07394609600305557,
      "learning_rate": 0.00022523529411764702,
      "loss": 0.0328,
      "step": 1773
    },
    {
      "epoch": 15.84,
      "grad_norm": 0.14768072962760925,
      "learning_rate": 0.0002251764705882353,
      "loss": 0.056,
      "step": 1774
    },
    {
      "epoch": 15.85,
      "grad_norm": 0.06404601037502289,
      "learning_rate": 0.0002251176470588235,
      "loss": 0.0345,
      "step": 1775
    },
    {
      "epoch": 15.86,
      "grad_norm": 0.06520042568445206,
      "learning_rate": 0.00022505882352941175,
      "loss": 0.0322,
      "step": 1776
    },
    {
      "epoch": 15.87,
      "grad_norm": 0.07000089436769485,
      "learning_rate": 0.000225,
      "loss": 0.0339,
      "step": 1777
    },
    {
      "epoch": 15.88,
      "grad_norm": 0.07399477809667587,
      "learning_rate": 0.0002249411764705882,
      "loss": 0.0353,
      "step": 1778
    },
    {
      "epoch": 15.88,
      "grad_norm": 0.07349999994039536,
      "learning_rate": 0.00022488235294117645,
      "loss": 0.0355,
      "step": 1779
    },
    {
      "epoch": 15.89,
      "grad_norm": 0.12206725031137466,
      "learning_rate": 0.00022482352941176467,
      "loss": 0.0452,
      "step": 1780
    },
    {
      "epoch": 15.9,
      "grad_norm": 0.0694836750626564,
      "learning_rate": 0.00022476470588235294,
      "loss": 0.0354,
      "step": 1781
    },
    {
      "epoch": 15.91,
      "grad_norm": 0.06848176568746567,
      "learning_rate": 0.00022470588235294116,
      "loss": 0.0336,
      "step": 1782
    },
    {
      "epoch": 15.92,
      "grad_norm": 0.0721769779920578,
      "learning_rate": 0.00022464705882352937,
      "loss": 0.0336,
      "step": 1783
    },
    {
      "epoch": 15.93,
      "grad_norm": 0.07481008768081665,
      "learning_rate": 0.00022458823529411764,
      "loss": 0.0362,
      "step": 1784
    },
    {
      "epoch": 15.94,
      "grad_norm": 0.06670886278152466,
      "learning_rate": 0.00022452941176470586,
      "loss": 0.0306,
      "step": 1785
    },
    {
      "epoch": 15.95,
      "grad_norm": 0.10526525974273682,
      "learning_rate": 0.0002244705882352941,
      "loss": 0.0347,
      "step": 1786
    },
    {
      "epoch": 15.96,
      "grad_norm": 0.09913452714681625,
      "learning_rate": 0.00022441176470588235,
      "loss": 0.0457,
      "step": 1787
    },
    {
      "epoch": 15.96,
      "grad_norm": 0.07062456011772156,
      "learning_rate": 0.00022435294117647056,
      "loss": 0.0342,
      "step": 1788
    },
    {
      "epoch": 15.97,
      "grad_norm": 0.07000609487295151,
      "learning_rate": 0.0002242941176470588,
      "loss": 0.033,
      "step": 1789
    },
    {
      "epoch": 15.98,
      "grad_norm": 0.07293125241994858,
      "learning_rate": 0.00022423529411764702,
      "loss": 0.033,
      "step": 1790
    },
    {
      "epoch": 15.99,
      "grad_norm": 0.08094222098588943,
      "learning_rate": 0.0002241764705882353,
      "loss": 0.0377,
      "step": 1791
    },
    {
      "epoch": 16.0,
      "grad_norm": 0.09451135993003845,
      "learning_rate": 0.0002241176470588235,
      "loss": 0.0416,
      "step": 1792
    },
    {
      "epoch": 16.01,
      "grad_norm": 0.0639321580529213,
      "learning_rate": 0.00022405882352941173,
      "loss": 0.0316,
      "step": 1793
    },
    {
      "epoch": 16.02,
      "grad_norm": 0.06530879437923431,
      "learning_rate": 0.000224,
      "loss": 0.0307,
      "step": 1794
    },
    {
      "epoch": 16.03,
      "grad_norm": 0.058440230786800385,
      "learning_rate": 0.0002239411764705882,
      "loss": 0.0277,
      "step": 1795
    },
    {
      "epoch": 16.04,
      "grad_norm": 0.08349408954381943,
      "learning_rate": 0.00022388235294117643,
      "loss": 0.0305,
      "step": 1796
    },
    {
      "epoch": 16.04,
      "grad_norm": 0.08139226585626602,
      "learning_rate": 0.0002238235294117647,
      "loss": 0.0345,
      "step": 1797
    },
    {
      "epoch": 16.05,
      "grad_norm": 0.1024152860045433,
      "learning_rate": 0.00022376470588235292,
      "loss": 0.0389,
      "step": 1798
    },
    {
      "epoch": 16.06,
      "grad_norm": 0.10898366570472717,
      "learning_rate": 0.00022370588235294116,
      "loss": 0.0399,
      "step": 1799
    },
    {
      "epoch": 16.07,
      "grad_norm": 0.06472934037446976,
      "learning_rate": 0.0002236470588235294,
      "loss": 0.0334,
      "step": 1800
    },
    {
      "epoch": 16.07,
      "eval_cer": 0.035982606277224674,
      "eval_loss": 0.19003820419311523,
      "eval_runtime": 22.2377,
      "eval_samples_per_second": 118.807,
      "eval_steps_per_second": 1.889,
      "eval_wer": 0.11878222927409758,
      "step": 1800
    },
    {
      "epoch": 16.08,
      "grad_norm": 0.05682501941919327,
      "learning_rate": 0.00022358823529411762,
      "loss": 0.0294,
      "step": 1801
    },
    {
      "epoch": 16.09,
      "grad_norm": 0.06381959468126297,
      "learning_rate": 0.00022352941176470586,
      "loss": 0.0325,
      "step": 1802
    },
    {
      "epoch": 16.1,
      "grad_norm": 0.07066192477941513,
      "learning_rate": 0.00022347058823529408,
      "loss": 0.0312,
      "step": 1803
    },
    {
      "epoch": 16.11,
      "grad_norm": 0.07962921261787415,
      "learning_rate": 0.00022341176470588235,
      "loss": 0.033,
      "step": 1804
    },
    {
      "epoch": 16.12,
      "grad_norm": 0.12375257909297943,
      "learning_rate": 0.00022335294117647057,
      "loss": 0.0473,
      "step": 1805
    },
    {
      "epoch": 16.12,
      "grad_norm": 0.06608864665031433,
      "learning_rate": 0.00022329411764705878,
      "loss": 0.0314,
      "step": 1806
    },
    {
      "epoch": 16.13,
      "grad_norm": 0.06043463572859764,
      "learning_rate": 0.00022323529411764705,
      "loss": 0.0268,
      "step": 1807
    },
    {
      "epoch": 16.14,
      "grad_norm": 0.06614497303962708,
      "learning_rate": 0.00022317647058823527,
      "loss": 0.0318,
      "step": 1808
    },
    {
      "epoch": 16.15,
      "grad_norm": 0.06933463364839554,
      "learning_rate": 0.0002231176470588235,
      "loss": 0.0318,
      "step": 1809
    },
    {
      "epoch": 16.16,
      "grad_norm": 0.07098463922739029,
      "learning_rate": 0.00022305882352941176,
      "loss": 0.0277,
      "step": 1810
    },
    {
      "epoch": 16.17,
      "grad_norm": 0.13402563333511353,
      "learning_rate": 0.00022299999999999997,
      "loss": 0.0474,
      "step": 1811
    },
    {
      "epoch": 16.18,
      "grad_norm": 0.06053219735622406,
      "learning_rate": 0.00022294117647058822,
      "loss": 0.0311,
      "step": 1812
    },
    {
      "epoch": 16.19,
      "grad_norm": 0.06736651808023453,
      "learning_rate": 0.00022288235294117643,
      "loss": 0.0298,
      "step": 1813
    },
    {
      "epoch": 16.2,
      "grad_norm": 0.06711181253194809,
      "learning_rate": 0.0002228235294117647,
      "loss": 0.0284,
      "step": 1814
    },
    {
      "epoch": 16.21,
      "grad_norm": 0.07631996273994446,
      "learning_rate": 0.00022276470588235292,
      "loss": 0.0356,
      "step": 1815
    },
    {
      "epoch": 16.21,
      "grad_norm": 0.06564201414585114,
      "learning_rate": 0.00022270588235294114,
      "loss": 0.0341,
      "step": 1816
    },
    {
      "epoch": 16.22,
      "grad_norm": 0.11746109277009964,
      "learning_rate": 0.0002226470588235294,
      "loss": 0.0473,
      "step": 1817
    },
    {
      "epoch": 16.23,
      "grad_norm": 0.06227483972907066,
      "learning_rate": 0.00022258823529411762,
      "loss": 0.034,
      "step": 1818
    },
    {
      "epoch": 16.24,
      "grad_norm": 0.06249179318547249,
      "learning_rate": 0.00022252941176470587,
      "loss": 0.0332,
      "step": 1819
    },
    {
      "epoch": 16.25,
      "grad_norm": 0.05764083191752434,
      "learning_rate": 0.0002224705882352941,
      "loss": 0.0282,
      "step": 1820
    },
    {
      "epoch": 16.26,
      "grad_norm": 0.057898227125406265,
      "learning_rate": 0.00022241176470588233,
      "loss": 0.0273,
      "step": 1821
    },
    {
      "epoch": 16.27,
      "grad_norm": 0.07494675368070602,
      "learning_rate": 0.00022235294117647057,
      "loss": 0.0317,
      "step": 1822
    },
    {
      "epoch": 16.28,
      "grad_norm": 0.09450692683458328,
      "learning_rate": 0.0002222941176470588,
      "loss": 0.0351,
      "step": 1823
    },
    {
      "epoch": 16.29,
      "grad_norm": 0.1166757121682167,
      "learning_rate": 0.00022223529411764706,
      "loss": 0.0378,
      "step": 1824
    },
    {
      "epoch": 16.29,
      "grad_norm": 0.06373901665210724,
      "learning_rate": 0.00022217647058823527,
      "loss": 0.0367,
      "step": 1825
    },
    {
      "epoch": 16.3,
      "grad_norm": 0.058000992983579636,
      "learning_rate": 0.0002221176470588235,
      "loss": 0.0297,
      "step": 1826
    },
    {
      "epoch": 16.31,
      "grad_norm": 0.07125075161457062,
      "learning_rate": 0.00022205882352941176,
      "loss": 0.0366,
      "step": 1827
    },
    {
      "epoch": 16.32,
      "grad_norm": 0.07212217152118683,
      "learning_rate": 0.00022199999999999998,
      "loss": 0.0317,
      "step": 1828
    },
    {
      "epoch": 16.33,
      "grad_norm": 0.08258946239948273,
      "learning_rate": 0.0002219411764705882,
      "loss": 0.0325,
      "step": 1829
    },
    {
      "epoch": 16.34,
      "grad_norm": 0.11259468644857407,
      "learning_rate": 0.00022188235294117646,
      "loss": 0.0407,
      "step": 1830
    },
    {
      "epoch": 16.35,
      "grad_norm": 0.05695272982120514,
      "learning_rate": 0.00022182352941176468,
      "loss": 0.0334,
      "step": 1831
    },
    {
      "epoch": 16.36,
      "grad_norm": 0.05781324952840805,
      "learning_rate": 0.00022176470588235292,
      "loss": 0.035,
      "step": 1832
    },
    {
      "epoch": 16.37,
      "grad_norm": 0.07321536540985107,
      "learning_rate": 0.00022170588235294117,
      "loss": 0.0376,
      "step": 1833
    },
    {
      "epoch": 16.38,
      "grad_norm": 0.06028003245592117,
      "learning_rate": 0.00022164705882352938,
      "loss": 0.0263,
      "step": 1834
    },
    {
      "epoch": 16.38,
      "grad_norm": 0.07853641360998154,
      "learning_rate": 0.00022158823529411762,
      "loss": 0.0323,
      "step": 1835
    },
    {
      "epoch": 16.39,
      "grad_norm": 0.12827949225902557,
      "learning_rate": 0.00022152941176470587,
      "loss": 0.045,
      "step": 1836
    },
    {
      "epoch": 16.4,
      "grad_norm": 0.05804058164358139,
      "learning_rate": 0.0002214705882352941,
      "loss": 0.0291,
      "step": 1837
    },
    {
      "epoch": 16.41,
      "grad_norm": 0.06798772513866425,
      "learning_rate": 0.00022141176470588233,
      "loss": 0.0359,
      "step": 1838
    },
    {
      "epoch": 16.42,
      "grad_norm": 0.07073916494846344,
      "learning_rate": 0.00022135294117647054,
      "loss": 0.0343,
      "step": 1839
    },
    {
      "epoch": 16.43,
      "grad_norm": 0.06816134601831436,
      "learning_rate": 0.00022129411764705881,
      "loss": 0.0334,
      "step": 1840
    },
    {
      "epoch": 16.44,
      "grad_norm": 0.0676409974694252,
      "learning_rate": 0.00022123529411764703,
      "loss": 0.0316,
      "step": 1841
    },
    {
      "epoch": 16.45,
      "grad_norm": 0.17641691863536835,
      "learning_rate": 0.0002211764705882353,
      "loss": 0.0475,
      "step": 1842
    },
    {
      "epoch": 16.46,
      "grad_norm": 0.07010889798402786,
      "learning_rate": 0.00022111764705882352,
      "loss": 0.0329,
      "step": 1843
    },
    {
      "epoch": 16.46,
      "grad_norm": 0.06269314140081406,
      "learning_rate": 0.00022105882352941173,
      "loss": 0.0323,
      "step": 1844
    },
    {
      "epoch": 16.47,
      "grad_norm": 0.06245392560958862,
      "learning_rate": 0.00022099999999999998,
      "loss": 0.0346,
      "step": 1845
    },
    {
      "epoch": 16.48,
      "grad_norm": 0.06329552084207535,
      "learning_rate": 0.00022094117647058822,
      "loss": 0.0295,
      "step": 1846
    },
    {
      "epoch": 16.49,
      "grad_norm": 0.07149852812290192,
      "learning_rate": 0.00022088235294117646,
      "loss": 0.033,
      "step": 1847
    },
    {
      "epoch": 16.5,
      "grad_norm": 0.1021699607372284,
      "learning_rate": 0.00022082352941176468,
      "loss": 0.0383,
      "step": 1848
    },
    {
      "epoch": 16.51,
      "grad_norm": 0.08884425461292267,
      "learning_rate": 0.0002207647058823529,
      "loss": 0.0373,
      "step": 1849
    },
    {
      "epoch": 16.52,
      "grad_norm": 0.06852715462446213,
      "learning_rate": 0.00022070588235294117,
      "loss": 0.0335,
      "step": 1850
    },
    {
      "epoch": 16.53,
      "grad_norm": 0.06676394492387772,
      "learning_rate": 0.00022064705882352938,
      "loss": 0.0345,
      "step": 1851
    },
    {
      "epoch": 16.54,
      "grad_norm": 0.06098376214504242,
      "learning_rate": 0.00022058823529411765,
      "loss": 0.0304,
      "step": 1852
    },
    {
      "epoch": 16.54,
      "grad_norm": 0.06835677474737167,
      "learning_rate": 0.00022052941176470587,
      "loss": 0.0301,
      "step": 1853
    },
    {
      "epoch": 16.55,
      "grad_norm": 0.07412328571081161,
      "learning_rate": 0.0002204705882352941,
      "loss": 0.0339,
      "step": 1854
    },
    {
      "epoch": 16.56,
      "grad_norm": 0.1628560721874237,
      "learning_rate": 0.00022041176470588233,
      "loss": 0.0498,
      "step": 1855
    },
    {
      "epoch": 16.57,
      "grad_norm": 0.07015007734298706,
      "learning_rate": 0.00022035294117647057,
      "loss": 0.0363,
      "step": 1856
    },
    {
      "epoch": 16.58,
      "grad_norm": 0.05822065472602844,
      "learning_rate": 0.0002202941176470588,
      "loss": 0.0312,
      "step": 1857
    },
    {
      "epoch": 16.59,
      "grad_norm": 0.06740330904722214,
      "learning_rate": 0.00022023529411764703,
      "loss": 0.0381,
      "step": 1858
    },
    {
      "epoch": 16.6,
      "grad_norm": 0.06438733637332916,
      "learning_rate": 0.00022017647058823528,
      "loss": 0.0305,
      "step": 1859
    },
    {
      "epoch": 16.61,
      "grad_norm": 0.07315194606781006,
      "learning_rate": 0.00022011764705882352,
      "loss": 0.0314,
      "step": 1860
    },
    {
      "epoch": 16.62,
      "grad_norm": 0.13501623272895813,
      "learning_rate": 0.00022005882352941174,
      "loss": 0.0506,
      "step": 1861
    },
    {
      "epoch": 16.62,
      "grad_norm": 0.0613023079931736,
      "learning_rate": 0.00021999999999999995,
      "loss": 0.0321,
      "step": 1862
    },
    {
      "epoch": 16.63,
      "grad_norm": 0.06343629956245422,
      "learning_rate": 0.00021994117647058822,
      "loss": 0.0318,
      "step": 1863
    },
    {
      "epoch": 16.64,
      "grad_norm": 0.061639804393053055,
      "learning_rate": 0.00021988235294117644,
      "loss": 0.0339,
      "step": 1864
    },
    {
      "epoch": 16.65,
      "grad_norm": 0.06862615793943405,
      "learning_rate": 0.0002198235294117647,
      "loss": 0.0346,
      "step": 1865
    },
    {
      "epoch": 16.66,
      "grad_norm": 0.06971698999404907,
      "learning_rate": 0.00021976470588235293,
      "loss": 0.0347,
      "step": 1866
    },
    {
      "epoch": 16.67,
      "grad_norm": 0.13029135763645172,
      "learning_rate": 0.00021970588235294114,
      "loss": 0.0438,
      "step": 1867
    },
    {
      "epoch": 16.68,
      "grad_norm": 0.06730790436267853,
      "learning_rate": 0.0002196470588235294,
      "loss": 0.0307,
      "step": 1868
    },
    {
      "epoch": 16.69,
      "grad_norm": 0.06171699985861778,
      "learning_rate": 0.00021958823529411763,
      "loss": 0.0326,
      "step": 1869
    },
    {
      "epoch": 16.7,
      "grad_norm": 0.06088051572442055,
      "learning_rate": 0.00021952941176470587,
      "loss": 0.0306,
      "step": 1870
    },
    {
      "epoch": 16.71,
      "grad_norm": 0.06121975556015968,
      "learning_rate": 0.0002194705882352941,
      "loss": 0.0289,
      "step": 1871
    },
    {
      "epoch": 16.71,
      "grad_norm": 0.07202765345573425,
      "learning_rate": 0.00021941176470588233,
      "loss": 0.0348,
      "step": 1872
    },
    {
      "epoch": 16.72,
      "grad_norm": 0.09348710626363754,
      "learning_rate": 0.00021935294117647058,
      "loss": 0.0331,
      "step": 1873
    },
    {
      "epoch": 16.73,
      "grad_norm": 0.10430654883384705,
      "learning_rate": 0.0002192941176470588,
      "loss": 0.0424,
      "step": 1874
    },
    {
      "epoch": 16.74,
      "grad_norm": 0.06490690261125565,
      "learning_rate": 0.00021923529411764706,
      "loss": 0.0302,
      "step": 1875
    },
    {
      "epoch": 16.75,
      "grad_norm": 0.059396229684352875,
      "learning_rate": 0.00021917647058823528,
      "loss": 0.0298,
      "step": 1876
    },
    {
      "epoch": 16.76,
      "grad_norm": 0.061630092561244965,
      "learning_rate": 0.0002191176470588235,
      "loss": 0.0279,
      "step": 1877
    },
    {
      "epoch": 16.77,
      "grad_norm": 0.06576143205165863,
      "learning_rate": 0.00021905882352941174,
      "loss": 0.029,
      "step": 1878
    },
    {
      "epoch": 16.78,
      "grad_norm": 0.08609016239643097,
      "learning_rate": 0.00021899999999999998,
      "loss": 0.0292,
      "step": 1879
    },
    {
      "epoch": 16.79,
      "grad_norm": 0.1195787563920021,
      "learning_rate": 0.00021894117647058823,
      "loss": 0.0422,
      "step": 1880
    },
    {
      "epoch": 16.79,
      "grad_norm": 0.06603191047906876,
      "learning_rate": 0.00021888235294117644,
      "loss": 0.0322,
      "step": 1881
    },
    {
      "epoch": 16.8,
      "grad_norm": 0.07025889307260513,
      "learning_rate": 0.0002188235294117647,
      "loss": 0.0341,
      "step": 1882
    },
    {
      "epoch": 16.81,
      "grad_norm": 0.06338518857955933,
      "learning_rate": 0.00021876470588235293,
      "loss": 0.0307,
      "step": 1883
    },
    {
      "epoch": 16.82,
      "grad_norm": 0.06663335859775543,
      "learning_rate": 0.00021870588235294115,
      "loss": 0.0313,
      "step": 1884
    },
    {
      "epoch": 16.83,
      "grad_norm": 0.07321462780237198,
      "learning_rate": 0.00021864705882352942,
      "loss": 0.0313,
      "step": 1885
    },
    {
      "epoch": 16.84,
      "grad_norm": 0.11784330010414124,
      "learning_rate": 0.00021858823529411763,
      "loss": 0.0417,
      "step": 1886
    },
    {
      "epoch": 16.85,
      "grad_norm": 0.07092329859733582,
      "learning_rate": 0.00021852941176470585,
      "loss": 0.0345,
      "step": 1887
    },
    {
      "epoch": 16.86,
      "grad_norm": 0.07185757905244827,
      "learning_rate": 0.00021847058823529412,
      "loss": 0.0355,
      "step": 1888
    },
    {
      "epoch": 16.87,
      "grad_norm": 0.05557725206017494,
      "learning_rate": 0.00021841176470588234,
      "loss": 0.0278,
      "step": 1889
    },
    {
      "epoch": 16.88,
      "grad_norm": 0.07115907222032547,
      "learning_rate": 0.00021835294117647055,
      "loss": 0.0378,
      "step": 1890
    },
    {
      "epoch": 16.88,
      "grad_norm": 0.06429091095924377,
      "learning_rate": 0.0002182941176470588,
      "loss": 0.0304,
      "step": 1891
    },
    {
      "epoch": 16.89,
      "grad_norm": 0.10283501446247101,
      "learning_rate": 0.00021823529411764704,
      "loss": 0.0436,
      "step": 1892
    },
    {
      "epoch": 16.9,
      "grad_norm": 0.06136798858642578,
      "learning_rate": 0.00021817647058823528,
      "loss": 0.0357,
      "step": 1893
    },
    {
      "epoch": 16.91,
      "grad_norm": 0.05844808369874954,
      "learning_rate": 0.0002181176470588235,
      "loss": 0.0299,
      "step": 1894
    },
    {
      "epoch": 16.92,
      "grad_norm": 0.07017340511083603,
      "learning_rate": 0.00021805882352941174,
      "loss": 0.0343,
      "step": 1895
    },
    {
      "epoch": 16.93,
      "grad_norm": 0.07959800958633423,
      "learning_rate": 0.00021799999999999999,
      "loss": 0.0317,
      "step": 1896
    },
    {
      "epoch": 16.94,
      "grad_norm": 0.06927252560853958,
      "learning_rate": 0.0002179411764705882,
      "loss": 0.0307,
      "step": 1897
    },
    {
      "epoch": 16.95,
      "grad_norm": 0.07652056962251663,
      "learning_rate": 0.00021788235294117647,
      "loss": 0.0279,
      "step": 1898
    },
    {
      "epoch": 16.96,
      "grad_norm": 0.08846669644117355,
      "learning_rate": 0.0002178235294117647,
      "loss": 0.039,
      "step": 1899
    },
    {
      "epoch": 16.96,
      "grad_norm": 0.062216632068157196,
      "learning_rate": 0.0002177647058823529,
      "loss": 0.0315,
      "step": 1900
    },
    {
      "epoch": 16.96,
      "eval_cer": 0.034584493729164556,
      "eval_loss": 0.19013793766498566,
      "eval_runtime": 22.6106,
      "eval_samples_per_second": 116.848,
      "eval_steps_per_second": 1.858,
      "eval_wer": 0.11576755255850853,
      "step": 1900
    },
    {
      "epoch": 16.97,
      "grad_norm": 0.06767335534095764,
      "learning_rate": 0.00021770588235294118,
      "loss": 0.0341,
      "step": 1901
    },
    {
      "epoch": 16.98,
      "grad_norm": 0.06550324708223343,
      "learning_rate": 0.0002176470588235294,
      "loss": 0.0291,
      "step": 1902
    },
    {
      "epoch": 16.99,
      "grad_norm": 0.08068516105413437,
      "learning_rate": 0.00021758823529411764,
      "loss": 0.0304,
      "step": 1903
    },
    {
      "epoch": 17.0,
      "grad_norm": 0.08599540591239929,
      "learning_rate": 0.00021752941176470585,
      "loss": 0.0407,
      "step": 1904
    },
    {
      "epoch": 17.01,
      "grad_norm": 0.0673895999789238,
      "learning_rate": 0.0002174705882352941,
      "loss": 0.0321,
      "step": 1905
    },
    {
      "epoch": 17.02,
      "grad_norm": 0.06604141741991043,
      "learning_rate": 0.00021741176470588234,
      "loss": 0.0345,
      "step": 1906
    },
    {
      "epoch": 17.03,
      "grad_norm": 0.062167853116989136,
      "learning_rate": 0.00021735294117647056,
      "loss": 0.0302,
      "step": 1907
    },
    {
      "epoch": 17.04,
      "grad_norm": 0.0598611906170845,
      "learning_rate": 0.00021729411764705883,
      "loss": 0.0314,
      "step": 1908
    },
    {
      "epoch": 17.04,
      "grad_norm": 0.06386779248714447,
      "learning_rate": 0.00021723529411764704,
      "loss": 0.0278,
      "step": 1909
    },
    {
      "epoch": 17.05,
      "grad_norm": 0.0866820365190506,
      "learning_rate": 0.00021717647058823526,
      "loss": 0.0293,
      "step": 1910
    },
    {
      "epoch": 17.06,
      "grad_norm": 0.09303127974271774,
      "learning_rate": 0.00021711764705882353,
      "loss": 0.0362,
      "step": 1911
    },
    {
      "epoch": 17.07,
      "grad_norm": 0.060847409069538116,
      "learning_rate": 0.00021705882352941175,
      "loss": 0.0329,
      "step": 1912
    },
    {
      "epoch": 17.08,
      "grad_norm": 0.05897834151983261,
      "learning_rate": 0.000217,
      "loss": 0.0247,
      "step": 1913
    },
    {
      "epoch": 17.09,
      "grad_norm": 0.0682649239897728,
      "learning_rate": 0.0002169411764705882,
      "loss": 0.0302,
      "step": 1914
    },
    {
      "epoch": 17.1,
      "grad_norm": 0.07063417881727219,
      "learning_rate": 0.00021688235294117645,
      "loss": 0.0336,
      "step": 1915
    },
    {
      "epoch": 17.11,
      "grad_norm": 0.09338242560625076,
      "learning_rate": 0.0002168235294117647,
      "loss": 0.0362,
      "step": 1916
    },
    {
      "epoch": 17.12,
      "grad_norm": 0.14794890582561493,
      "learning_rate": 0.0002167647058823529,
      "loss": 0.0447,
      "step": 1917
    },
    {
      "epoch": 17.12,
      "grad_norm": 0.06314072012901306,
      "learning_rate": 0.00021670588235294118,
      "loss": 0.0321,
      "step": 1918
    },
    {
      "epoch": 17.13,
      "grad_norm": 0.061584699898958206,
      "learning_rate": 0.0002166470588235294,
      "loss": 0.032,
      "step": 1919
    },
    {
      "epoch": 17.14,
      "grad_norm": 0.06706103682518005,
      "learning_rate": 0.0002165882352941176,
      "loss": 0.03,
      "step": 1920
    },
    {
      "epoch": 17.15,
      "grad_norm": 0.06886307895183563,
      "learning_rate": 0.00021652941176470588,
      "loss": 0.0331,
      "step": 1921
    },
    {
      "epoch": 17.16,
      "grad_norm": 0.06668057292699814,
      "learning_rate": 0.0002164705882352941,
      "loss": 0.0306,
      "step": 1922
    },
    {
      "epoch": 17.17,
      "grad_norm": 0.1145925298333168,
      "learning_rate": 0.00021641176470588232,
      "loss": 0.0411,
      "step": 1923
    },
    {
      "epoch": 17.18,
      "grad_norm": 0.06388647109270096,
      "learning_rate": 0.00021635294117647059,
      "loss": 0.0343,
      "step": 1924
    },
    {
      "epoch": 17.19,
      "grad_norm": 0.059622038155794144,
      "learning_rate": 0.0002162941176470588,
      "loss": 0.0265,
      "step": 1925
    },
    {
      "epoch": 17.2,
      "grad_norm": 0.07900681346654892,
      "learning_rate": 0.00021623529411764705,
      "loss": 0.0359,
      "step": 1926
    },
    {
      "epoch": 17.21,
      "grad_norm": 0.06877920031547546,
      "learning_rate": 0.00021617647058823526,
      "loss": 0.0271,
      "step": 1927
    },
    {
      "epoch": 17.21,
      "grad_norm": 0.08047014474868774,
      "learning_rate": 0.0002161176470588235,
      "loss": 0.0274,
      "step": 1928
    },
    {
      "epoch": 17.22,
      "grad_norm": 0.1356843262910843,
      "learning_rate": 0.00021605882352941175,
      "loss": 0.0462,
      "step": 1929
    },
    {
      "epoch": 17.23,
      "grad_norm": 0.06882345676422119,
      "learning_rate": 0.00021599999999999996,
      "loss": 0.0321,
      "step": 1930
    },
    {
      "epoch": 17.24,
      "grad_norm": 0.06174418702721596,
      "learning_rate": 0.00021594117647058824,
      "loss": 0.0308,
      "step": 1931
    },
    {
      "epoch": 17.25,
      "grad_norm": 0.06057072803378105,
      "learning_rate": 0.00021588235294117645,
      "loss": 0.0329,
      "step": 1932
    },
    {
      "epoch": 17.26,
      "grad_norm": 0.06363937258720398,
      "learning_rate": 0.00021582352941176467,
      "loss": 0.0329,
      "step": 1933
    },
    {
      "epoch": 17.27,
      "grad_norm": 0.06881600618362427,
      "learning_rate": 0.00021576470588235294,
      "loss": 0.0333,
      "step": 1934
    },
    {
      "epoch": 17.28,
      "grad_norm": 0.07912091165781021,
      "learning_rate": 0.00021570588235294115,
      "loss": 0.0361,
      "step": 1935
    },
    {
      "epoch": 17.29,
      "grad_norm": 0.08154988288879395,
      "learning_rate": 0.0002156470588235294,
      "loss": 0.036,
      "step": 1936
    },
    {
      "epoch": 17.29,
      "grad_norm": 0.06133905425667763,
      "learning_rate": 0.00021558823529411764,
      "loss": 0.0329,
      "step": 1937
    },
    {
      "epoch": 17.3,
      "grad_norm": 0.0610046423971653,
      "learning_rate": 0.00021552941176470586,
      "loss": 0.0307,
      "step": 1938
    },
    {
      "epoch": 17.31,
      "grad_norm": 0.054651498794555664,
      "learning_rate": 0.0002154705882352941,
      "loss": 0.0284,
      "step": 1939
    },
    {
      "epoch": 17.32,
      "grad_norm": 0.06436837464570999,
      "learning_rate": 0.00021541176470588232,
      "loss": 0.0329,
      "step": 1940
    },
    {
      "epoch": 17.33,
      "grad_norm": 0.07338185608386993,
      "learning_rate": 0.0002153529411764706,
      "loss": 0.0322,
      "step": 1941
    },
    {
      "epoch": 17.34,
      "grad_norm": 0.15808679163455963,
      "learning_rate": 0.0002152941176470588,
      "loss": 0.0529,
      "step": 1942
    },
    {
      "epoch": 17.35,
      "grad_norm": 0.0614892840385437,
      "learning_rate": 0.00021523529411764702,
      "loss": 0.0334,
      "step": 1943
    },
    {
      "epoch": 17.36,
      "grad_norm": 0.05659007653594017,
      "learning_rate": 0.0002151764705882353,
      "loss": 0.0287,
      "step": 1944
    },
    {
      "epoch": 17.37,
      "grad_norm": 0.06882838159799576,
      "learning_rate": 0.0002151176470588235,
      "loss": 0.0292,
      "step": 1945
    },
    {
      "epoch": 17.38,
      "grad_norm": 0.06403681635856628,
      "learning_rate": 0.00021505882352941175,
      "loss": 0.0338,
      "step": 1946
    },
    {
      "epoch": 17.38,
      "grad_norm": 0.07217351347208023,
      "learning_rate": 0.000215,
      "loss": 0.0263,
      "step": 1947
    },
    {
      "epoch": 17.39,
      "grad_norm": 0.12260174751281738,
      "learning_rate": 0.0002149411764705882,
      "loss": 0.05,
      "step": 1948
    },
    {
      "epoch": 17.4,
      "grad_norm": 0.06135178729891777,
      "learning_rate": 0.00021488235294117645,
      "loss": 0.0327,
      "step": 1949
    },
    {
      "epoch": 17.41,
      "grad_norm": 0.06472061574459076,
      "learning_rate": 0.00021482352941176467,
      "loss": 0.0362,
      "step": 1950
    },
    {
      "epoch": 17.42,
      "grad_norm": 0.06496337801218033,
      "learning_rate": 0.00021476470588235294,
      "loss": 0.0244,
      "step": 1951
    },
    {
      "epoch": 17.43,
      "grad_norm": 0.06565911322832108,
      "learning_rate": 0.00021470588235294116,
      "loss": 0.0305,
      "step": 1952
    },
    {
      "epoch": 17.44,
      "grad_norm": 0.0685436949133873,
      "learning_rate": 0.00021464705882352937,
      "loss": 0.0283,
      "step": 1953
    },
    {
      "epoch": 17.45,
      "grad_norm": 0.1360541582107544,
      "learning_rate": 0.00021458823529411764,
      "loss": 0.0429,
      "step": 1954
    },
    {
      "epoch": 17.46,
      "grad_norm": 0.060156237334012985,
      "learning_rate": 0.00021452941176470586,
      "loss": 0.0355,
      "step": 1955
    },
    {
      "epoch": 17.46,
      "grad_norm": 0.06217309832572937,
      "learning_rate": 0.00021447058823529408,
      "loss": 0.0334,
      "step": 1956
    },
    {
      "epoch": 17.47,
      "grad_norm": 0.06349260360002518,
      "learning_rate": 0.00021441176470588235,
      "loss": 0.0322,
      "step": 1957
    },
    {
      "epoch": 17.48,
      "grad_norm": 0.06153681501746178,
      "learning_rate": 0.00021435294117647056,
      "loss": 0.0285,
      "step": 1958
    },
    {
      "epoch": 17.49,
      "grad_norm": 0.05936651676893234,
      "learning_rate": 0.0002142941176470588,
      "loss": 0.0297,
      "step": 1959
    },
    {
      "epoch": 17.5,
      "grad_norm": 0.08301527798175812,
      "learning_rate": 0.00021423529411764705,
      "loss": 0.0289,
      "step": 1960
    },
    {
      "epoch": 17.51,
      "grad_norm": 0.1044265478849411,
      "learning_rate": 0.00021417647058823527,
      "loss": 0.0403,
      "step": 1961
    },
    {
      "epoch": 17.52,
      "grad_norm": 0.056587181985378265,
      "learning_rate": 0.0002141176470588235,
      "loss": 0.0296,
      "step": 1962
    },
    {
      "epoch": 17.53,
      "grad_norm": 0.06559578329324722,
      "learning_rate": 0.00021405882352941173,
      "loss": 0.0336,
      "step": 1963
    },
    {
      "epoch": 17.54,
      "grad_norm": 0.060637760907411575,
      "learning_rate": 0.000214,
      "loss": 0.033,
      "step": 1964
    },
    {
      "epoch": 17.54,
      "grad_norm": 0.07044398784637451,
      "learning_rate": 0.00021394117647058821,
      "loss": 0.0361,
      "step": 1965
    },
    {
      "epoch": 17.55,
      "grad_norm": 0.07613010704517365,
      "learning_rate": 0.00021388235294117643,
      "loss": 0.0266,
      "step": 1966
    },
    {
      "epoch": 17.56,
      "grad_norm": 0.18630178272724152,
      "learning_rate": 0.0002138235294117647,
      "loss": 0.0453,
      "step": 1967
    },
    {
      "epoch": 17.57,
      "grad_norm": 0.07019291073083878,
      "learning_rate": 0.00021376470588235292,
      "loss": 0.0319,
      "step": 1968
    },
    {
      "epoch": 17.58,
      "grad_norm": 0.07487376779317856,
      "learning_rate": 0.00021370588235294116,
      "loss": 0.0349,
      "step": 1969
    },
    {
      "epoch": 17.59,
      "grad_norm": 0.06258773803710938,
      "learning_rate": 0.0002136470588235294,
      "loss": 0.0267,
      "step": 1970
    },
    {
      "epoch": 17.6,
      "grad_norm": 0.06805012375116348,
      "learning_rate": 0.00021358823529411762,
      "loss": 0.0336,
      "step": 1971
    },
    {
      "epoch": 17.61,
      "grad_norm": 0.07135777920484543,
      "learning_rate": 0.00021352941176470586,
      "loss": 0.0312,
      "step": 1972
    },
    {
      "epoch": 17.62,
      "grad_norm": 0.12469202280044556,
      "learning_rate": 0.00021347058823529408,
      "loss": 0.0427,
      "step": 1973
    },
    {
      "epoch": 17.62,
      "grad_norm": 0.05374273657798767,
      "learning_rate": 0.00021341176470588235,
      "loss": 0.0306,
      "step": 1974
    },
    {
      "epoch": 17.63,
      "grad_norm": 0.05935364589095116,
      "learning_rate": 0.00021335294117647057,
      "loss": 0.0287,
      "step": 1975
    },
    {
      "epoch": 17.64,
      "grad_norm": 0.05376312881708145,
      "learning_rate": 0.00021329411764705878,
      "loss": 0.0275,
      "step": 1976
    },
    {
      "epoch": 17.65,
      "grad_norm": 0.06635450571775436,
      "learning_rate": 0.00021323529411764705,
      "loss": 0.0329,
      "step": 1977
    },
    {
      "epoch": 17.66,
      "grad_norm": 0.060057323426008224,
      "learning_rate": 0.00021317647058823527,
      "loss": 0.027,
      "step": 1978
    },
    {
      "epoch": 17.67,
      "grad_norm": 0.11194895207881927,
      "learning_rate": 0.0002131176470588235,
      "loss": 0.0441,
      "step": 1979
    },
    {
      "epoch": 17.68,
      "grad_norm": 0.06371347606182098,
      "learning_rate": 0.00021305882352941176,
      "loss": 0.0367,
      "step": 1980
    },
    {
      "epoch": 17.69,
      "grad_norm": 0.06562800705432892,
      "learning_rate": 0.00021299999999999997,
      "loss": 0.0305,
      "step": 1981
    },
    {
      "epoch": 17.7,
      "grad_norm": 0.05915699526667595,
      "learning_rate": 0.00021294117647058822,
      "loss": 0.0298,
      "step": 1982
    },
    {
      "epoch": 17.71,
      "grad_norm": 0.06269513815641403,
      "learning_rate": 0.00021288235294117646,
      "loss": 0.028,
      "step": 1983
    },
    {
      "epoch": 17.71,
      "grad_norm": 0.07125990092754364,
      "learning_rate": 0.0002128235294117647,
      "loss": 0.0318,
      "step": 1984
    },
    {
      "epoch": 17.72,
      "grad_norm": 0.11740262806415558,
      "learning_rate": 0.00021276470588235292,
      "loss": 0.0415,
      "step": 1985
    },
    {
      "epoch": 17.73,
      "grad_norm": 0.11232439428567886,
      "learning_rate": 0.00021270588235294114,
      "loss": 0.0418,
      "step": 1986
    },
    {
      "epoch": 17.74,
      "grad_norm": 0.061971794813871384,
      "learning_rate": 0.0002126470588235294,
      "loss": 0.0313,
      "step": 1987
    },
    {
      "epoch": 17.75,
      "grad_norm": 0.06482931226491928,
      "learning_rate": 0.00021258823529411762,
      "loss": 0.0286,
      "step": 1988
    },
    {
      "epoch": 17.76,
      "grad_norm": 0.06639612466096878,
      "learning_rate": 0.00021252941176470584,
      "loss": 0.03,
      "step": 1989
    },
    {
      "epoch": 17.77,
      "grad_norm": 0.06994175165891647,
      "learning_rate": 0.0002124705882352941,
      "loss": 0.0346,
      "step": 1990
    },
    {
      "epoch": 17.78,
      "grad_norm": 0.07315286248922348,
      "learning_rate": 0.00021241176470588233,
      "loss": 0.0284,
      "step": 1991
    },
    {
      "epoch": 17.79,
      "grad_norm": 0.12527918815612793,
      "learning_rate": 0.00021235294117647057,
      "loss": 0.0406,
      "step": 1992
    },
    {
      "epoch": 17.79,
      "grad_norm": 0.07032306492328644,
      "learning_rate": 0.0002122941176470588,
      "loss": 0.0318,
      "step": 1993
    },
    {
      "epoch": 17.8,
      "grad_norm": 0.07027847319841385,
      "learning_rate": 0.00021223529411764703,
      "loss": 0.0357,
      "step": 1994
    },
    {
      "epoch": 17.81,
      "grad_norm": 0.06513828784227371,
      "learning_rate": 0.00021217647058823527,
      "loss": 0.0287,
      "step": 1995
    },
    {
      "epoch": 17.82,
      "grad_norm": 0.07004145532846451,
      "learning_rate": 0.00021211764705882352,
      "loss": 0.0361,
      "step": 1996
    },
    {
      "epoch": 17.83,
      "grad_norm": 0.06596864014863968,
      "learning_rate": 0.00021205882352941176,
      "loss": 0.0299,
      "step": 1997
    },
    {
      "epoch": 17.84,
      "grad_norm": 0.14162759482860565,
      "learning_rate": 0.00021199999999999998,
      "loss": 0.0493,
      "step": 1998
    },
    {
      "epoch": 17.85,
      "grad_norm": 0.05906546860933304,
      "learning_rate": 0.0002119411764705882,
      "loss": 0.0337,
      "step": 1999
    },
    {
      "epoch": 17.86,
      "grad_norm": 0.05804269760847092,
      "learning_rate": 0.00021188235294117646,
      "loss": 0.0317,
      "step": 2000
    },
    {
      "epoch": 17.86,
      "eval_cer": 0.03409214764369177,
      "eval_loss": 0.1789909154176712,
      "eval_runtime": 22.8941,
      "eval_samples_per_second": 115.401,
      "eval_steps_per_second": 1.835,
      "eval_wer": 0.11342721142403808,
      "step": 2000
    },
    {
      "epoch": 17.87,
      "grad_norm": 0.05982355773448944,
      "learning_rate": 0.00021182352941176468,
      "loss": 0.0304,
      "step": 2001
    },
    {
      "epoch": 17.88,
      "grad_norm": 0.061808496713638306,
      "learning_rate": 0.00021176470588235295,
      "loss": 0.0344,
      "step": 2002
    },
    {
      "epoch": 17.88,
      "grad_norm": 0.0662093535065651,
      "learning_rate": 0.00021170588235294117,
      "loss": 0.0319,
      "step": 2003
    },
    {
      "epoch": 17.89,
      "grad_norm": 0.1215112954378128,
      "learning_rate": 0.00021164705882352938,
      "loss": 0.0425,
      "step": 2004
    },
    {
      "epoch": 17.9,
      "grad_norm": 0.057328928261995316,
      "learning_rate": 0.00021158823529411763,
      "loss": 0.0328,
      "step": 2005
    },
    {
      "epoch": 17.91,
      "grad_norm": 0.06580270081758499,
      "learning_rate": 0.00021152941176470587,
      "loss": 0.0344,
      "step": 2006
    },
    {
      "epoch": 17.92,
      "grad_norm": 0.06612606346607208,
      "learning_rate": 0.0002114705882352941,
      "loss": 0.0304,
      "step": 2007
    },
    {
      "epoch": 17.93,
      "grad_norm": 0.06731640547513962,
      "learning_rate": 0.00021141176470588233,
      "loss": 0.0297,
      "step": 2008
    },
    {
      "epoch": 17.94,
      "grad_norm": 0.06695950776338577,
      "learning_rate": 0.00021135294117647055,
      "loss": 0.029,
      "step": 2009
    },
    {
      "epoch": 17.95,
      "grad_norm": 0.0974825844168663,
      "learning_rate": 0.00021129411764705882,
      "loss": 0.0375,
      "step": 2010
    },
    {
      "epoch": 17.96,
      "grad_norm": 0.09003818035125732,
      "learning_rate": 0.00021123529411764703,
      "loss": 0.0384,
      "step": 2011
    },
    {
      "epoch": 17.96,
      "grad_norm": 0.05922992527484894,
      "learning_rate": 0.0002111764705882353,
      "loss": 0.0315,
      "step": 2012
    },
    {
      "epoch": 17.97,
      "grad_norm": 0.05970041826367378,
      "learning_rate": 0.00021111764705882352,
      "loss": 0.0301,
      "step": 2013
    },
    {
      "epoch": 17.98,
      "grad_norm": 0.07018531113862991,
      "learning_rate": 0.00021105882352941174,
      "loss": 0.0293,
      "step": 2014
    },
    {
      "epoch": 17.99,
      "grad_norm": 0.07534600049257278,
      "learning_rate": 0.00021099999999999998,
      "loss": 0.0329,
      "step": 2015
    },
    {
      "epoch": 18.0,
      "grad_norm": 0.08939068019390106,
      "learning_rate": 0.00021094117647058822,
      "loss": 0.0344,
      "step": 2016
    },
    {
      "epoch": 18.01,
      "grad_norm": 0.05164641514420509,
      "learning_rate": 0.00021088235294117644,
      "loss": 0.0298,
      "step": 2017
    },
    {
      "epoch": 18.02,
      "grad_norm": 0.064960777759552,
      "learning_rate": 0.00021082352941176468,
      "loss": 0.0287,
      "step": 2018
    },
    {
      "epoch": 18.03,
      "grad_norm": 0.06120393052697182,
      "learning_rate": 0.00021076470588235293,
      "loss": 0.0317,
      "step": 2019
    },
    {
      "epoch": 18.04,
      "grad_norm": 0.07327283173799515,
      "learning_rate": 0.00021070588235294117,
      "loss": 0.031,
      "step": 2020
    },
    {
      "epoch": 18.04,
      "grad_norm": 0.06303515285253525,
      "learning_rate": 0.00021064705882352939,
      "loss": 0.0263,
      "step": 2021
    },
    {
      "epoch": 18.05,
      "grad_norm": 0.09693946689367294,
      "learning_rate": 0.0002105882352941176,
      "loss": 0.0358,
      "step": 2022
    },
    {
      "epoch": 18.06,
      "grad_norm": 0.09954741597175598,
      "learning_rate": 0.00021052941176470587,
      "loss": 0.0397,
      "step": 2023
    },
    {
      "epoch": 18.07,
      "grad_norm": 0.06713460385799408,
      "learning_rate": 0.0002104705882352941,
      "loss": 0.0334,
      "step": 2024
    },
    {
      "epoch": 18.08,
      "grad_norm": 0.0598839707672596,
      "learning_rate": 0.00021041176470588236,
      "loss": 0.0303,
      "step": 2025
    },
    {
      "epoch": 18.09,
      "grad_norm": 0.05972038954496384,
      "learning_rate": 0.00021035294117647058,
      "loss": 0.0299,
      "step": 2026
    },
    {
      "epoch": 18.1,
      "grad_norm": 0.06614918261766434,
      "learning_rate": 0.0002102941176470588,
      "loss": 0.0302,
      "step": 2027
    },
    {
      "epoch": 18.11,
      "grad_norm": 0.07636161893606186,
      "learning_rate": 0.00021023529411764703,
      "loss": 0.0351,
      "step": 2028
    },
    {
      "epoch": 18.12,
      "grad_norm": 0.09338352084159851,
      "learning_rate": 0.00021017647058823528,
      "loss": 0.0341,
      "step": 2029
    },
    {
      "epoch": 18.12,
      "grad_norm": 0.06121871992945671,
      "learning_rate": 0.00021011764705882352,
      "loss": 0.0321,
      "step": 2030
    },
    {
      "epoch": 18.13,
      "grad_norm": 0.05937650427222252,
      "learning_rate": 0.00021005882352941174,
      "loss": 0.0279,
      "step": 2031
    },
    {
      "epoch": 18.14,
      "grad_norm": 0.06002294272184372,
      "learning_rate": 0.00020999999999999998,
      "loss": 0.0262,
      "step": 2032
    },
    {
      "epoch": 18.15,
      "grad_norm": 0.06456071138381958,
      "learning_rate": 0.00020994117647058823,
      "loss": 0.0267,
      "step": 2033
    },
    {
      "epoch": 18.16,
      "grad_norm": 0.0755121260881424,
      "learning_rate": 0.00020988235294117644,
      "loss": 0.0326,
      "step": 2034
    },
    {
      "epoch": 18.17,
      "grad_norm": 0.12616093456745148,
      "learning_rate": 0.0002098235294117647,
      "loss": 0.0461,
      "step": 2035
    },
    {
      "epoch": 18.18,
      "grad_norm": 0.05936598777770996,
      "learning_rate": 0.00020976470588235293,
      "loss": 0.0288,
      "step": 2036
    },
    {
      "epoch": 18.19,
      "grad_norm": 0.06453947722911835,
      "learning_rate": 0.00020970588235294114,
      "loss": 0.0314,
      "step": 2037
    },
    {
      "epoch": 18.2,
      "grad_norm": 0.06521280854940414,
      "learning_rate": 0.0002096470588235294,
      "loss": 0.0315,
      "step": 2038
    },
    {
      "epoch": 18.21,
      "grad_norm": 0.05086246132850647,
      "learning_rate": 0.00020958823529411763,
      "loss": 0.0255,
      "step": 2039
    },
    {
      "epoch": 18.21,
      "grad_norm": 0.06328276544809341,
      "learning_rate": 0.00020952941176470587,
      "loss": 0.0277,
      "step": 2040
    },
    {
      "epoch": 18.22,
      "grad_norm": 0.15216706693172455,
      "learning_rate": 0.0002094705882352941,
      "loss": 0.0422,
      "step": 2041
    },
    {
      "epoch": 18.23,
      "grad_norm": 0.0609402172267437,
      "learning_rate": 0.00020941176470588233,
      "loss": 0.0301,
      "step": 2042
    },
    {
      "epoch": 18.24,
      "grad_norm": 0.06806506961584091,
      "learning_rate": 0.00020935294117647058,
      "loss": 0.029,
      "step": 2043
    },
    {
      "epoch": 18.25,
      "grad_norm": 0.06164085119962692,
      "learning_rate": 0.0002092941176470588,
      "loss": 0.0311,
      "step": 2044
    },
    {
      "epoch": 18.26,
      "grad_norm": 0.06438139081001282,
      "learning_rate": 0.00020923529411764706,
      "loss": 0.027,
      "step": 2045
    },
    {
      "epoch": 18.27,
      "grad_norm": 0.06687790155410767,
      "learning_rate": 0.00020917647058823528,
      "loss": 0.0283,
      "step": 2046
    },
    {
      "epoch": 18.28,
      "grad_norm": 0.09623175859451294,
      "learning_rate": 0.0002091176470588235,
      "loss": 0.0284,
      "step": 2047
    },
    {
      "epoch": 18.29,
      "grad_norm": 0.09724742919206619,
      "learning_rate": 0.00020905882352941177,
      "loss": 0.0385,
      "step": 2048
    },
    {
      "epoch": 18.29,
      "grad_norm": 0.06700415909290314,
      "learning_rate": 0.00020899999999999998,
      "loss": 0.0281,
      "step": 2049
    },
    {
      "epoch": 18.3,
      "grad_norm": 0.05784018710255623,
      "learning_rate": 0.0002089411764705882,
      "loss": 0.0276,
      "step": 2050
    },
    {
      "epoch": 18.31,
      "grad_norm": 0.06372254341840744,
      "learning_rate": 0.00020888235294117644,
      "loss": 0.0308,
      "step": 2051
    },
    {
      "epoch": 18.32,
      "grad_norm": 0.057946398854255676,
      "learning_rate": 0.0002088235294117647,
      "loss": 0.0318,
      "step": 2052
    },
    {
      "epoch": 18.33,
      "grad_norm": 0.08636609464883804,
      "learning_rate": 0.00020876470588235293,
      "loss": 0.032,
      "step": 2053
    },
    {
      "epoch": 18.34,
      "grad_norm": 0.12471861392259598,
      "learning_rate": 0.00020870588235294115,
      "loss": 0.0416,
      "step": 2054
    },
    {
      "epoch": 18.35,
      "grad_norm": 0.05486654117703438,
      "learning_rate": 0.0002086470588235294,
      "loss": 0.0322,
      "step": 2055
    },
    {
      "epoch": 18.36,
      "grad_norm": 0.05568893253803253,
      "learning_rate": 0.00020858823529411763,
      "loss": 0.0301,
      "step": 2056
    },
    {
      "epoch": 18.37,
      "grad_norm": 0.06621300429105759,
      "learning_rate": 0.00020852941176470585,
      "loss": 0.0311,
      "step": 2057
    },
    {
      "epoch": 18.38,
      "grad_norm": 0.05489505082368851,
      "learning_rate": 0.00020847058823529412,
      "loss": 0.0258,
      "step": 2058
    },
    {
      "epoch": 18.38,
      "grad_norm": 0.06559371203184128,
      "learning_rate": 0.00020841176470588234,
      "loss": 0.0278,
      "step": 2059
    },
    {
      "epoch": 18.39,
      "grad_norm": 0.1418338567018509,
      "learning_rate": 0.00020835294117647055,
      "loss": 0.0419,
      "step": 2060
    },
    {
      "epoch": 18.4,
      "grad_norm": 0.06461971998214722,
      "learning_rate": 0.00020829411764705882,
      "loss": 0.0308,
      "step": 2061
    },
    {
      "epoch": 18.41,
      "grad_norm": 0.05940650403499603,
      "learning_rate": 0.00020823529411764704,
      "loss": 0.0273,
      "step": 2062
    },
    {
      "epoch": 18.42,
      "grad_norm": 0.06172783672809601,
      "learning_rate": 0.00020817647058823528,
      "loss": 0.0297,
      "step": 2063
    },
    {
      "epoch": 18.43,
      "grad_norm": 0.06961867213249207,
      "learning_rate": 0.0002081176470588235,
      "loss": 0.0302,
      "step": 2064
    },
    {
      "epoch": 18.44,
      "grad_norm": 0.06662897765636444,
      "learning_rate": 0.00020805882352941174,
      "loss": 0.0279,
      "step": 2065
    },
    {
      "epoch": 18.45,
      "grad_norm": 0.12071380764245987,
      "learning_rate": 0.000208,
      "loss": 0.0431,
      "step": 2066
    },
    {
      "epoch": 18.46,
      "grad_norm": 0.05877776816487312,
      "learning_rate": 0.0002079411764705882,
      "loss": 0.0299,
      "step": 2067
    },
    {
      "epoch": 18.46,
      "grad_norm": 0.06061911582946777,
      "learning_rate": 0.00020788235294117647,
      "loss": 0.0298,
      "step": 2068
    },
    {
      "epoch": 18.47,
      "grad_norm": 0.05871176719665527,
      "learning_rate": 0.0002078235294117647,
      "loss": 0.0281,
      "step": 2069
    },
    {
      "epoch": 18.48,
      "grad_norm": 0.06598810106515884,
      "learning_rate": 0.0002077647058823529,
      "loss": 0.0315,
      "step": 2070
    },
    {
      "epoch": 18.49,
      "grad_norm": 0.06471865624189377,
      "learning_rate": 0.00020770588235294118,
      "loss": 0.0291,
      "step": 2071
    },
    {
      "epoch": 18.5,
      "grad_norm": 0.08320900797843933,
      "learning_rate": 0.0002076470588235294,
      "loss": 0.0324,
      "step": 2072
    },
    {
      "epoch": 18.51,
      "grad_norm": 0.11166468262672424,
      "learning_rate": 0.00020758823529411764,
      "loss": 0.0378,
      "step": 2073
    },
    {
      "epoch": 18.52,
      "grad_norm": 0.06702893227338791,
      "learning_rate": 0.00020752941176470585,
      "loss": 0.0308,
      "step": 2074
    },
    {
      "epoch": 18.53,
      "grad_norm": 0.062195952981710434,
      "learning_rate": 0.0002074705882352941,
      "loss": 0.0286,
      "step": 2075
    },
    {
      "epoch": 18.54,
      "grad_norm": 0.058582693338394165,
      "learning_rate": 0.00020741176470588234,
      "loss": 0.0271,
      "step": 2076
    },
    {
      "epoch": 18.54,
      "grad_norm": 0.06748209148645401,
      "learning_rate": 0.00020735294117647056,
      "loss": 0.0327,
      "step": 2077
    },
    {
      "epoch": 18.55,
      "grad_norm": 0.09048274159431458,
      "learning_rate": 0.00020729411764705883,
      "loss": 0.0285,
      "step": 2078
    },
    {
      "epoch": 18.56,
      "grad_norm": 0.13230320811271667,
      "learning_rate": 0.00020723529411764704,
      "loss": 0.0429,
      "step": 2079
    },
    {
      "epoch": 18.57,
      "grad_norm": 0.05814466252923012,
      "learning_rate": 0.00020717647058823526,
      "loss": 0.0254,
      "step": 2080
    },
    {
      "epoch": 18.58,
      "grad_norm": 0.06753221154212952,
      "learning_rate": 0.00020711764705882353,
      "loss": 0.0336,
      "step": 2081
    },
    {
      "epoch": 18.59,
      "grad_norm": 0.06285896897315979,
      "learning_rate": 0.00020705882352941175,
      "loss": 0.0269,
      "step": 2082
    },
    {
      "epoch": 18.6,
      "grad_norm": 0.06398814171552658,
      "learning_rate": 0.00020699999999999996,
      "loss": 0.0268,
      "step": 2083
    },
    {
      "epoch": 18.61,
      "grad_norm": 0.07129596918821335,
      "learning_rate": 0.00020694117647058823,
      "loss": 0.0334,
      "step": 2084
    },
    {
      "epoch": 18.62,
      "grad_norm": 0.14242440462112427,
      "learning_rate": 0.00020688235294117645,
      "loss": 0.0466,
      "step": 2085
    },
    {
      "epoch": 18.62,
      "grad_norm": 0.05649041011929512,
      "learning_rate": 0.0002068235294117647,
      "loss": 0.0289,
      "step": 2086
    },
    {
      "epoch": 18.63,
      "grad_norm": 0.054231468588113785,
      "learning_rate": 0.0002067647058823529,
      "loss": 0.0301,
      "step": 2087
    },
    {
      "epoch": 18.64,
      "grad_norm": 0.05468952655792236,
      "learning_rate": 0.00020670588235294115,
      "loss": 0.0279,
      "step": 2088
    },
    {
      "epoch": 18.65,
      "grad_norm": 0.05843213573098183,
      "learning_rate": 0.0002066470588235294,
      "loss": 0.0335,
      "step": 2089
    },
    {
      "epoch": 18.66,
      "grad_norm": 0.06207503005862236,
      "learning_rate": 0.0002065882352941176,
      "loss": 0.0306,
      "step": 2090
    },
    {
      "epoch": 18.67,
      "grad_norm": 0.10590814799070358,
      "learning_rate": 0.00020652941176470588,
      "loss": 0.0416,
      "step": 2091
    },
    {
      "epoch": 18.68,
      "grad_norm": 0.06286563724279404,
      "learning_rate": 0.0002064705882352941,
      "loss": 0.0336,
      "step": 2092
    },
    {
      "epoch": 18.69,
      "grad_norm": 0.06435458362102509,
      "learning_rate": 0.00020641176470588232,
      "loss": 0.0303,
      "step": 2093
    },
    {
      "epoch": 18.7,
      "grad_norm": 0.06846845149993896,
      "learning_rate": 0.00020635294117647059,
      "loss": 0.0312,
      "step": 2094
    },
    {
      "epoch": 18.71,
      "grad_norm": 0.06863770633935928,
      "learning_rate": 0.0002062941176470588,
      "loss": 0.033,
      "step": 2095
    },
    {
      "epoch": 18.71,
      "grad_norm": 0.07404244691133499,
      "learning_rate": 0.00020623529411764705,
      "loss": 0.0334,
      "step": 2096
    },
    {
      "epoch": 18.72,
      "grad_norm": 0.09194135665893555,
      "learning_rate": 0.0002061764705882353,
      "loss": 0.0327,
      "step": 2097
    },
    {
      "epoch": 18.73,
      "grad_norm": 0.13283534348011017,
      "learning_rate": 0.0002061176470588235,
      "loss": 0.038,
      "step": 2098
    },
    {
      "epoch": 18.74,
      "grad_norm": 0.05884513258934021,
      "learning_rate": 0.00020605882352941175,
      "loss": 0.036,
      "step": 2099
    },
    {
      "epoch": 18.75,
      "grad_norm": 0.06227260082960129,
      "learning_rate": 0.00020599999999999997,
      "loss": 0.0264,
      "step": 2100
    },
    {
      "epoch": 18.75,
      "eval_cer": 0.03562556140608029,
      "eval_loss": 0.18643657863140106,
      "eval_runtime": 22.248,
      "eval_samples_per_second": 118.752,
      "eval_steps_per_second": 1.888,
      "eval_wer": 0.11590638635462118,
      "step": 2100
    },
    {
      "epoch": 18.76,
      "grad_norm": 0.06634752452373505,
      "learning_rate": 0.00020594117647058824,
      "loss": 0.0276,
      "step": 2101
    },
    {
      "epoch": 18.77,
      "grad_norm": 0.06490685045719147,
      "learning_rate": 0.00020588235294117645,
      "loss": 0.0318,
      "step": 2102
    },
    {
      "epoch": 18.78,
      "grad_norm": 0.07782765477895737,
      "learning_rate": 0.00020582352941176467,
      "loss": 0.0306,
      "step": 2103
    },
    {
      "epoch": 18.79,
      "grad_norm": 0.09754545241594315,
      "learning_rate": 0.00020576470588235294,
      "loss": 0.0403,
      "step": 2104
    },
    {
      "epoch": 18.79,
      "grad_norm": 0.0597204752266407,
      "learning_rate": 0.00020570588235294116,
      "loss": 0.0276,
      "step": 2105
    },
    {
      "epoch": 18.8,
      "grad_norm": 0.06698883324861526,
      "learning_rate": 0.0002056470588235294,
      "loss": 0.0318,
      "step": 2106
    },
    {
      "epoch": 18.81,
      "grad_norm": 0.06619041413068771,
      "learning_rate": 0.00020558823529411764,
      "loss": 0.0334,
      "step": 2107
    },
    {
      "epoch": 18.82,
      "grad_norm": 0.05624597892165184,
      "learning_rate": 0.00020552941176470586,
      "loss": 0.0262,
      "step": 2108
    },
    {
      "epoch": 18.83,
      "grad_norm": 0.0718187689781189,
      "learning_rate": 0.0002054705882352941,
      "loss": 0.0314,
      "step": 2109
    },
    {
      "epoch": 18.84,
      "grad_norm": 0.11686796694993973,
      "learning_rate": 0.00020541176470588232,
      "loss": 0.0407,
      "step": 2110
    },
    {
      "epoch": 18.85,
      "grad_norm": 0.05923014134168625,
      "learning_rate": 0.0002053529411764706,
      "loss": 0.0247,
      "step": 2111
    },
    {
      "epoch": 18.86,
      "grad_norm": 0.06864576041698456,
      "learning_rate": 0.0002052941176470588,
      "loss": 0.0346,
      "step": 2112
    },
    {
      "epoch": 18.87,
      "grad_norm": 0.06370431184768677,
      "learning_rate": 0.00020523529411764702,
      "loss": 0.0285,
      "step": 2113
    },
    {
      "epoch": 18.88,
      "grad_norm": 0.062278784811496735,
      "learning_rate": 0.0002051764705882353,
      "loss": 0.0273,
      "step": 2114
    },
    {
      "epoch": 18.88,
      "grad_norm": 0.07636075466871262,
      "learning_rate": 0.0002051176470588235,
      "loss": 0.0324,
      "step": 2115
    },
    {
      "epoch": 18.89,
      "grad_norm": 0.12746940553188324,
      "learning_rate": 0.00020505882352941173,
      "loss": 0.0415,
      "step": 2116
    },
    {
      "epoch": 18.9,
      "grad_norm": 0.05222547799348831,
      "learning_rate": 0.000205,
      "loss": 0.0291,
      "step": 2117
    },
    {
      "epoch": 18.91,
      "grad_norm": 0.060084518045186996,
      "learning_rate": 0.0002049411764705882,
      "loss": 0.0288,
      "step": 2118
    },
    {
      "epoch": 18.92,
      "grad_norm": 0.06343322992324829,
      "learning_rate": 0.00020488235294117646,
      "loss": 0.0294,
      "step": 2119
    },
    {
      "epoch": 18.93,
      "grad_norm": 0.06317029148340225,
      "learning_rate": 0.0002048235294117647,
      "loss": 0.0296,
      "step": 2120
    },
    {
      "epoch": 18.94,
      "grad_norm": 0.07388322055339813,
      "learning_rate": 0.00020476470588235292,
      "loss": 0.0288,
      "step": 2121
    },
    {
      "epoch": 18.95,
      "grad_norm": 0.09308870881795883,
      "learning_rate": 0.00020470588235294116,
      "loss": 0.0365,
      "step": 2122
    },
    {
      "epoch": 18.96,
      "grad_norm": 0.09080292284488678,
      "learning_rate": 0.00020464705882352937,
      "loss": 0.0431,
      "step": 2123
    },
    {
      "epoch": 18.96,
      "grad_norm": 0.06618691235780716,
      "learning_rate": 0.00020458823529411765,
      "loss": 0.0355,
      "step": 2124
    },
    {
      "epoch": 18.97,
      "grad_norm": 0.06465675681829453,
      "learning_rate": 0.00020452941176470586,
      "loss": 0.0332,
      "step": 2125
    },
    {
      "epoch": 18.98,
      "grad_norm": 0.06030743569135666,
      "learning_rate": 0.00020447058823529408,
      "loss": 0.0273,
      "step": 2126
    },
    {
      "epoch": 18.99,
      "grad_norm": 0.08355317264795303,
      "learning_rate": 0.00020441176470588235,
      "loss": 0.0343,
      "step": 2127
    },
    {
      "epoch": 19.0,
      "grad_norm": 0.09227706491947174,
      "learning_rate": 0.00020435294117647056,
      "loss": 0.0341,
      "step": 2128
    },
    {
      "epoch": 19.01,
      "grad_norm": 0.0566699281334877,
      "learning_rate": 0.0002042941176470588,
      "loss": 0.0255,
      "step": 2129
    },
    {
      "epoch": 19.02,
      "grad_norm": 0.05342443659901619,
      "learning_rate": 0.00020423529411764705,
      "loss": 0.0287,
      "step": 2130
    },
    {
      "epoch": 19.03,
      "grad_norm": 0.05447328835725784,
      "learning_rate": 0.00020417647058823527,
      "loss": 0.0291,
      "step": 2131
    },
    {
      "epoch": 19.04,
      "grad_norm": 0.054897114634513855,
      "learning_rate": 0.0002041176470588235,
      "loss": 0.026,
      "step": 2132
    },
    {
      "epoch": 19.04,
      "grad_norm": 0.06621937453746796,
      "learning_rate": 0.00020405882352941173,
      "loss": 0.0251,
      "step": 2133
    },
    {
      "epoch": 19.05,
      "grad_norm": 0.08687276393175125,
      "learning_rate": 0.000204,
      "loss": 0.033,
      "step": 2134
    },
    {
      "epoch": 19.06,
      "grad_norm": 0.1125141829252243,
      "learning_rate": 0.00020394117647058821,
      "loss": 0.04,
      "step": 2135
    },
    {
      "epoch": 19.07,
      "grad_norm": 0.06082363426685333,
      "learning_rate": 0.00020388235294117643,
      "loss": 0.0315,
      "step": 2136
    },
    {
      "epoch": 19.08,
      "grad_norm": 0.0553433857858181,
      "learning_rate": 0.0002038235294117647,
      "loss": 0.0268,
      "step": 2137
    },
    {
      "epoch": 19.09,
      "grad_norm": 0.06114237755537033,
      "learning_rate": 0.00020376470588235292,
      "loss": 0.0279,
      "step": 2138
    },
    {
      "epoch": 19.1,
      "grad_norm": 0.05804065987467766,
      "learning_rate": 0.00020370588235294116,
      "loss": 0.0272,
      "step": 2139
    },
    {
      "epoch": 19.11,
      "grad_norm": 0.09285765141248703,
      "learning_rate": 0.0002036470588235294,
      "loss": 0.0291,
      "step": 2140
    },
    {
      "epoch": 19.12,
      "grad_norm": 0.1081705391407013,
      "learning_rate": 0.00020358823529411762,
      "loss": 0.038,
      "step": 2141
    },
    {
      "epoch": 19.12,
      "grad_norm": 0.05302184820175171,
      "learning_rate": 0.00020352941176470586,
      "loss": 0.028,
      "step": 2142
    },
    {
      "epoch": 19.13,
      "grad_norm": 0.06427904963493347,
      "learning_rate": 0.0002034705882352941,
      "loss": 0.0266,
      "step": 2143
    },
    {
      "epoch": 19.14,
      "grad_norm": 0.05788515880703926,
      "learning_rate": 0.00020341176470588232,
      "loss": 0.025,
      "step": 2144
    },
    {
      "epoch": 19.15,
      "grad_norm": 0.0727313756942749,
      "learning_rate": 0.00020335294117647057,
      "loss": 0.0305,
      "step": 2145
    },
    {
      "epoch": 19.16,
      "grad_norm": 0.06664348393678665,
      "learning_rate": 0.00020329411764705878,
      "loss": 0.026,
      "step": 2146
    },
    {
      "epoch": 19.17,
      "grad_norm": 0.11840501427650452,
      "learning_rate": 0.00020323529411764705,
      "loss": 0.0388,
      "step": 2147
    },
    {
      "epoch": 19.18,
      "grad_norm": 0.0626482218503952,
      "learning_rate": 0.00020317647058823527,
      "loss": 0.0313,
      "step": 2148
    },
    {
      "epoch": 19.19,
      "grad_norm": 0.06950737535953522,
      "learning_rate": 0.0002031176470588235,
      "loss": 0.0351,
      "step": 2149
    },
    {
      "epoch": 19.2,
      "grad_norm": 0.061929699033498764,
      "learning_rate": 0.00020305882352941176,
      "loss": 0.0329,
      "step": 2150
    },
    {
      "epoch": 19.21,
      "grad_norm": 0.05309795215725899,
      "learning_rate": 0.00020299999999999997,
      "loss": 0.0233,
      "step": 2151
    },
    {
      "epoch": 19.21,
      "grad_norm": 0.062375832349061966,
      "learning_rate": 0.00020294117647058822,
      "loss": 0.0257,
      "step": 2152
    },
    {
      "epoch": 19.22,
      "grad_norm": 0.13026969134807587,
      "learning_rate": 0.00020288235294117646,
      "loss": 0.0431,
      "step": 2153
    },
    {
      "epoch": 19.23,
      "grad_norm": 0.06232600286602974,
      "learning_rate": 0.00020282352941176468,
      "loss": 0.0308,
      "step": 2154
    },
    {
      "epoch": 19.24,
      "grad_norm": 0.0655936449766159,
      "learning_rate": 0.00020276470588235292,
      "loss": 0.0299,
      "step": 2155
    },
    {
      "epoch": 19.25,
      "grad_norm": 0.06494201719760895,
      "learning_rate": 0.00020270588235294116,
      "loss": 0.0309,
      "step": 2156
    },
    {
      "epoch": 19.26,
      "grad_norm": 0.06123069301247597,
      "learning_rate": 0.0002026470588235294,
      "loss": 0.0267,
      "step": 2157
    },
    {
      "epoch": 19.27,
      "grad_norm": 0.06938119232654572,
      "learning_rate": 0.00020258823529411762,
      "loss": 0.0304,
      "step": 2158
    },
    {
      "epoch": 19.28,
      "grad_norm": 0.10094217211008072,
      "learning_rate": 0.00020252941176470584,
      "loss": 0.035,
      "step": 2159
    },
    {
      "epoch": 19.29,
      "grad_norm": 0.08669798076152802,
      "learning_rate": 0.0002024705882352941,
      "loss": 0.035,
      "step": 2160
    },
    {
      "epoch": 19.29,
      "grad_norm": 0.0575607568025589,
      "learning_rate": 0.00020241176470588233,
      "loss": 0.0307,
      "step": 2161
    },
    {
      "epoch": 19.3,
      "grad_norm": 0.061026837676763535,
      "learning_rate": 0.0002023529411764706,
      "loss": 0.0301,
      "step": 2162
    },
    {
      "epoch": 19.31,
      "grad_norm": 0.05477208271622658,
      "learning_rate": 0.00020229411764705881,
      "loss": 0.0278,
      "step": 2163
    },
    {
      "epoch": 19.32,
      "grad_norm": 0.05956774577498436,
      "learning_rate": 0.00020223529411764703,
      "loss": 0.0275,
      "step": 2164
    },
    {
      "epoch": 19.33,
      "grad_norm": 0.06744513660669327,
      "learning_rate": 0.00020217647058823527,
      "loss": 0.0251,
      "step": 2165
    },
    {
      "epoch": 19.34,
      "grad_norm": 0.10411574691534042,
      "learning_rate": 0.00020211764705882352,
      "loss": 0.0408,
      "step": 2166
    },
    {
      "epoch": 19.35,
      "grad_norm": 0.05916345492005348,
      "learning_rate": 0.00020205882352941176,
      "loss": 0.0285,
      "step": 2167
    },
    {
      "epoch": 19.36,
      "grad_norm": 0.07293649762868881,
      "learning_rate": 0.00020199999999999998,
      "loss": 0.0284,
      "step": 2168
    },
    {
      "epoch": 19.37,
      "grad_norm": 0.06648380309343338,
      "learning_rate": 0.0002019411764705882,
      "loss": 0.0273,
      "step": 2169
    },
    {
      "epoch": 19.38,
      "grad_norm": 0.06677570194005966,
      "learning_rate": 0.00020188235294117646,
      "loss": 0.03,
      "step": 2170
    },
    {
      "epoch": 19.38,
      "grad_norm": 0.06720144301652908,
      "learning_rate": 0.00020182352941176468,
      "loss": 0.0263,
      "step": 2171
    },
    {
      "epoch": 19.39,
      "grad_norm": 0.1155744194984436,
      "learning_rate": 0.00020176470588235295,
      "loss": 0.039,
      "step": 2172
    },
    {
      "epoch": 19.4,
      "grad_norm": 0.051371246576309204,
      "learning_rate": 0.00020170588235294117,
      "loss": 0.0254,
      "step": 2173
    },
    {
      "epoch": 19.41,
      "grad_norm": 0.05222965031862259,
      "learning_rate": 0.00020164705882352938,
      "loss": 0.0283,
      "step": 2174
    },
    {
      "epoch": 19.42,
      "grad_norm": 0.05005746707320213,
      "learning_rate": 0.00020158823529411763,
      "loss": 0.0248,
      "step": 2175
    },
    {
      "epoch": 19.43,
      "grad_norm": 0.06565649062395096,
      "learning_rate": 0.00020152941176470587,
      "loss": 0.026,
      "step": 2176
    },
    {
      "epoch": 19.44,
      "grad_norm": 0.06322000920772552,
      "learning_rate": 0.00020147058823529409,
      "loss": 0.0277,
      "step": 2177
    },
    {
      "epoch": 19.45,
      "grad_norm": 0.12081991881132126,
      "learning_rate": 0.00020141176470588233,
      "loss": 0.0389,
      "step": 2178
    },
    {
      "epoch": 19.46,
      "grad_norm": 0.06125609204173088,
      "learning_rate": 0.00020135294117647057,
      "loss": 0.0308,
      "step": 2179
    },
    {
      "epoch": 19.46,
      "grad_norm": 0.07139113545417786,
      "learning_rate": 0.00020129411764705882,
      "loss": 0.0323,
      "step": 2180
    },
    {
      "epoch": 19.47,
      "grad_norm": 0.06434722244739532,
      "learning_rate": 0.00020123529411764703,
      "loss": 0.0322,
      "step": 2181
    },
    {
      "epoch": 19.48,
      "grad_norm": 0.0564647801220417,
      "learning_rate": 0.00020117647058823525,
      "loss": 0.0222,
      "step": 2182
    },
    {
      "epoch": 19.49,
      "grad_norm": 0.062009118497371674,
      "learning_rate": 0.00020111764705882352,
      "loss": 0.0251,
      "step": 2183
    },
    {
      "epoch": 19.5,
      "grad_norm": 0.09379425644874573,
      "learning_rate": 0.00020105882352941174,
      "loss": 0.0345,
      "step": 2184
    },
    {
      "epoch": 19.51,
      "grad_norm": 0.11802651733160019,
      "learning_rate": 0.000201,
      "loss": 0.0434,
      "step": 2185
    },
    {
      "epoch": 19.52,
      "grad_norm": 0.059047456830739975,
      "learning_rate": 0.00020094117647058822,
      "loss": 0.0288,
      "step": 2186
    },
    {
      "epoch": 19.53,
      "grad_norm": 0.05848240479826927,
      "learning_rate": 0.00020088235294117644,
      "loss": 0.0303,
      "step": 2187
    },
    {
      "epoch": 19.54,
      "grad_norm": 0.0600242018699646,
      "learning_rate": 0.00020082352941176468,
      "loss": 0.0277,
      "step": 2188
    },
    {
      "epoch": 19.54,
      "grad_norm": 0.06942985206842422,
      "learning_rate": 0.00020076470588235293,
      "loss": 0.0295,
      "step": 2189
    },
    {
      "epoch": 19.55,
      "grad_norm": 0.07458772510290146,
      "learning_rate": 0.00020070588235294117,
      "loss": 0.03,
      "step": 2190
    },
    {
      "epoch": 19.56,
      "grad_norm": 0.11516823619604111,
      "learning_rate": 0.00020064705882352939,
      "loss": 0.0408,
      "step": 2191
    },
    {
      "epoch": 19.57,
      "grad_norm": 0.060609959065914154,
      "learning_rate": 0.00020058823529411763,
      "loss": 0.0283,
      "step": 2192
    },
    {
      "epoch": 19.58,
      "grad_norm": 0.06131531670689583,
      "learning_rate": 0.00020052941176470587,
      "loss": 0.0283,
      "step": 2193
    },
    {
      "epoch": 19.59,
      "grad_norm": 0.05881686136126518,
      "learning_rate": 0.0002004705882352941,
      "loss": 0.0298,
      "step": 2194
    },
    {
      "epoch": 19.6,
      "grad_norm": 0.06249463930726051,
      "learning_rate": 0.00020041176470588236,
      "loss": 0.0294,
      "step": 2195
    },
    {
      "epoch": 19.61,
      "grad_norm": 0.08682270348072052,
      "learning_rate": 0.00020035294117647058,
      "loss": 0.0338,
      "step": 2196
    },
    {
      "epoch": 19.62,
      "grad_norm": 0.12249086797237396,
      "learning_rate": 0.0002002941176470588,
      "loss": 0.042,
      "step": 2197
    },
    {
      "epoch": 19.62,
      "grad_norm": 0.05887043476104736,
      "learning_rate": 0.00020023529411764704,
      "loss": 0.0289,
      "step": 2198
    },
    {
      "epoch": 19.63,
      "grad_norm": 0.05861738324165344,
      "learning_rate": 0.00020017647058823528,
      "loss": 0.0292,
      "step": 2199
    },
    {
      "epoch": 19.64,
      "grad_norm": 0.059787504374980927,
      "learning_rate": 0.00020011764705882352,
      "loss": 0.0271,
      "step": 2200
    },
    {
      "epoch": 19.64,
      "eval_cer": 0.03406208070717435,
      "eval_loss": 0.1861434280872345,
      "eval_runtime": 22.5624,
      "eval_samples_per_second": 117.098,
      "eval_steps_per_second": 1.862,
      "eval_wer": 0.1149940499801666,
      "step": 2200
    },
    {
      "epoch": 19.65,
      "grad_norm": 0.05699005350470543,
      "learning_rate": 0.00020005882352941174,
      "loss": 0.0298,
      "step": 2201
    },
    {
      "epoch": 19.66,
      "grad_norm": 0.0681491419672966,
      "learning_rate": 0.00019999999999999998,
      "loss": 0.0313,
      "step": 2202
    },
    {
      "epoch": 19.67,
      "grad_norm": 0.1209702342748642,
      "learning_rate": 0.00019994117647058823,
      "loss": 0.0457,
      "step": 2203
    },
    {
      "epoch": 19.68,
      "grad_norm": 0.06152784079313278,
      "learning_rate": 0.00019988235294117644,
      "loss": 0.0297,
      "step": 2204
    },
    {
      "epoch": 19.69,
      "grad_norm": 0.06058143824338913,
      "learning_rate": 0.0001998235294117647,
      "loss": 0.0318,
      "step": 2205
    },
    {
      "epoch": 19.7,
      "grad_norm": 0.06092606484889984,
      "learning_rate": 0.00019976470588235293,
      "loss": 0.0283,
      "step": 2206
    },
    {
      "epoch": 19.71,
      "grad_norm": 0.0646490603685379,
      "learning_rate": 0.00019970588235294115,
      "loss": 0.0325,
      "step": 2207
    },
    {
      "epoch": 19.71,
      "grad_norm": 0.067710280418396,
      "learning_rate": 0.00019964705882352942,
      "loss": 0.0323,
      "step": 2208
    },
    {
      "epoch": 19.72,
      "grad_norm": 0.08690015226602554,
      "learning_rate": 0.00019958823529411763,
      "loss": 0.0349,
      "step": 2209
    },
    {
      "epoch": 19.73,
      "grad_norm": 0.07700332999229431,
      "learning_rate": 0.00019952941176470585,
      "loss": 0.0357,
      "step": 2210
    },
    {
      "epoch": 19.74,
      "grad_norm": 0.056973446160554886,
      "learning_rate": 0.0001994705882352941,
      "loss": 0.0298,
      "step": 2211
    },
    {
      "epoch": 19.75,
      "grad_norm": 0.06385958194732666,
      "learning_rate": 0.00019941176470588234,
      "loss": 0.0312,
      "step": 2212
    },
    {
      "epoch": 19.76,
      "grad_norm": 0.06139913946390152,
      "learning_rate": 0.00019935294117647058,
      "loss": 0.0285,
      "step": 2213
    },
    {
      "epoch": 19.77,
      "grad_norm": 0.06697844713926315,
      "learning_rate": 0.0001992941176470588,
      "loss": 0.0312,
      "step": 2214
    },
    {
      "epoch": 19.78,
      "grad_norm": 0.07495854794979095,
      "learning_rate": 0.00019923529411764704,
      "loss": 0.026,
      "step": 2215
    },
    {
      "epoch": 19.79,
      "grad_norm": 0.1093396246433258,
      "learning_rate": 0.00019917647058823528,
      "loss": 0.0416,
      "step": 2216
    },
    {
      "epoch": 19.79,
      "grad_norm": 0.059817247092723846,
      "learning_rate": 0.0001991176470588235,
      "loss": 0.0301,
      "step": 2217
    },
    {
      "epoch": 19.8,
      "grad_norm": 0.06090405583381653,
      "learning_rate": 0.00019905882352941177,
      "loss": 0.0303,
      "step": 2218
    },
    {
      "epoch": 19.81,
      "grad_norm": 0.053513459861278534,
      "learning_rate": 0.00019899999999999999,
      "loss": 0.0275,
      "step": 2219
    },
    {
      "epoch": 19.82,
      "grad_norm": 0.0619007870554924,
      "learning_rate": 0.0001989411764705882,
      "loss": 0.0246,
      "step": 2220
    },
    {
      "epoch": 19.83,
      "grad_norm": 0.07230164110660553,
      "learning_rate": 0.00019888235294117647,
      "loss": 0.0303,
      "step": 2221
    },
    {
      "epoch": 19.84,
      "grad_norm": 0.11419408023357391,
      "learning_rate": 0.0001988235294117647,
      "loss": 0.0474,
      "step": 2222
    },
    {
      "epoch": 19.85,
      "grad_norm": 0.05775037035346031,
      "learning_rate": 0.00019876470588235293,
      "loss": 0.0241,
      "step": 2223
    },
    {
      "epoch": 19.86,
      "grad_norm": 0.06071503460407257,
      "learning_rate": 0.00019870588235294115,
      "loss": 0.0293,
      "step": 2224
    },
    {
      "epoch": 19.87,
      "grad_norm": 0.060517024248838425,
      "learning_rate": 0.0001986470588235294,
      "loss": 0.0305,
      "step": 2225
    },
    {
      "epoch": 19.88,
      "grad_norm": 0.05888737365603447,
      "learning_rate": 0.00019858823529411764,
      "loss": 0.0321,
      "step": 2226
    },
    {
      "epoch": 19.88,
      "grad_norm": 0.06835459172725677,
      "learning_rate": 0.00019852941176470585,
      "loss": 0.0297,
      "step": 2227
    },
    {
      "epoch": 19.89,
      "grad_norm": 0.1049518883228302,
      "learning_rate": 0.00019847058823529412,
      "loss": 0.0396,
      "step": 2228
    },
    {
      "epoch": 19.9,
      "grad_norm": 0.05264618247747421,
      "learning_rate": 0.00019841176470588234,
      "loss": 0.027,
      "step": 2229
    },
    {
      "epoch": 19.91,
      "grad_norm": 0.0566607229411602,
      "learning_rate": 0.00019835294117647055,
      "loss": 0.028,
      "step": 2230
    },
    {
      "epoch": 19.92,
      "grad_norm": 0.0578635036945343,
      "learning_rate": 0.00019829411764705883,
      "loss": 0.0277,
      "step": 2231
    },
    {
      "epoch": 19.93,
      "grad_norm": 0.055492985993623734,
      "learning_rate": 0.00019823529411764704,
      "loss": 0.0258,
      "step": 2232
    },
    {
      "epoch": 19.94,
      "grad_norm": 0.057883501052856445,
      "learning_rate": 0.00019817647058823528,
      "loss": 0.0267,
      "step": 2233
    },
    {
      "epoch": 19.95,
      "grad_norm": 0.11310465633869171,
      "learning_rate": 0.0001981176470588235,
      "loss": 0.0353,
      "step": 2234
    },
    {
      "epoch": 19.96,
      "grad_norm": 0.11274304986000061,
      "learning_rate": 0.00019805882352941174,
      "loss": 0.0385,
      "step": 2235
    },
    {
      "epoch": 19.96,
      "grad_norm": 0.05661230534315109,
      "learning_rate": 0.000198,
      "loss": 0.0258,
      "step": 2236
    },
    {
      "epoch": 19.97,
      "grad_norm": 0.07197310775518417,
      "learning_rate": 0.0001979411764705882,
      "loss": 0.0298,
      "step": 2237
    },
    {
      "epoch": 19.98,
      "grad_norm": 0.0674099549651146,
      "learning_rate": 0.00019788235294117647,
      "loss": 0.029,
      "step": 2238
    },
    {
      "epoch": 19.99,
      "grad_norm": 0.07113310694694519,
      "learning_rate": 0.0001978235294117647,
      "loss": 0.0283,
      "step": 2239
    },
    {
      "epoch": 20.0,
      "grad_norm": 0.08611607551574707,
      "learning_rate": 0.0001977647058823529,
      "loss": 0.0375,
      "step": 2240
    },
    {
      "epoch": 20.01,
      "grad_norm": 0.051717352122068405,
      "learning_rate": 0.00019770588235294118,
      "loss": 0.0291,
      "step": 2241
    },
    {
      "epoch": 20.02,
      "grad_norm": 0.055984143167734146,
      "learning_rate": 0.0001976470588235294,
      "loss": 0.0277,
      "step": 2242
    },
    {
      "epoch": 20.03,
      "grad_norm": 0.060987718403339386,
      "learning_rate": 0.0001975882352941176,
      "loss": 0.0313,
      "step": 2243
    },
    {
      "epoch": 20.04,
      "grad_norm": 0.05700723081827164,
      "learning_rate": 0.00019752941176470588,
      "loss": 0.0244,
      "step": 2244
    },
    {
      "epoch": 20.04,
      "grad_norm": 0.05862624570727348,
      "learning_rate": 0.0001974705882352941,
      "loss": 0.0255,
      "step": 2245
    },
    {
      "epoch": 20.05,
      "grad_norm": 0.07748149335384369,
      "learning_rate": 0.00019741176470588234,
      "loss": 0.0266,
      "step": 2246
    },
    {
      "epoch": 20.06,
      "grad_norm": 0.08739695698022842,
      "learning_rate": 0.00019735294117647056,
      "loss": 0.0287,
      "step": 2247
    },
    {
      "epoch": 20.07,
      "grad_norm": 0.05724698305130005,
      "learning_rate": 0.0001972941176470588,
      "loss": 0.0273,
      "step": 2248
    },
    {
      "epoch": 20.08,
      "grad_norm": 0.06167592108249664,
      "learning_rate": 0.00019723529411764704,
      "loss": 0.0297,
      "step": 2249
    },
    {
      "epoch": 20.09,
      "grad_norm": 0.0602608323097229,
      "learning_rate": 0.00019717647058823526,
      "loss": 0.0298,
      "step": 2250
    },
    {
      "epoch": 20.1,
      "grad_norm": 0.06602107733488083,
      "learning_rate": 0.00019711764705882353,
      "loss": 0.0276,
      "step": 2251
    },
    {
      "epoch": 20.11,
      "grad_norm": 0.06687210500240326,
      "learning_rate": 0.00019705882352941175,
      "loss": 0.0302,
      "step": 2252
    },
    {
      "epoch": 20.12,
      "grad_norm": 0.09839525073766708,
      "learning_rate": 0.00019699999999999996,
      "loss": 0.0394,
      "step": 2253
    },
    {
      "epoch": 20.12,
      "grad_norm": 0.049129754304885864,
      "learning_rate": 0.00019694117647058823,
      "loss": 0.0267,
      "step": 2254
    },
    {
      "epoch": 20.13,
      "grad_norm": 0.053939610719680786,
      "learning_rate": 0.00019688235294117645,
      "loss": 0.0288,
      "step": 2255
    },
    {
      "epoch": 20.14,
      "grad_norm": 0.05686270073056221,
      "learning_rate": 0.0001968235294117647,
      "loss": 0.0293,
      "step": 2256
    },
    {
      "epoch": 20.15,
      "grad_norm": 0.05865943804383278,
      "learning_rate": 0.00019676470588235294,
      "loss": 0.0247,
      "step": 2257
    },
    {
      "epoch": 20.16,
      "grad_norm": 0.06347531825304031,
      "learning_rate": 0.00019670588235294115,
      "loss": 0.0242,
      "step": 2258
    },
    {
      "epoch": 20.17,
      "grad_norm": 0.1037096381187439,
      "learning_rate": 0.0001966470588235294,
      "loss": 0.0299,
      "step": 2259
    },
    {
      "epoch": 20.18,
      "grad_norm": 0.05274982750415802,
      "learning_rate": 0.00019658823529411761,
      "loss": 0.0272,
      "step": 2260
    },
    {
      "epoch": 20.19,
      "grad_norm": 0.0705869272351265,
      "learning_rate": 0.00019652941176470588,
      "loss": 0.0325,
      "step": 2261
    },
    {
      "epoch": 20.2,
      "grad_norm": 0.06054059416055679,
      "learning_rate": 0.0001964705882352941,
      "loss": 0.0269,
      "step": 2262
    },
    {
      "epoch": 20.21,
      "grad_norm": 0.060479119420051575,
      "learning_rate": 0.00019641176470588232,
      "loss": 0.0269,
      "step": 2263
    },
    {
      "epoch": 20.21,
      "grad_norm": 0.06396442651748657,
      "learning_rate": 0.0001963529411764706,
      "loss": 0.0232,
      "step": 2264
    },
    {
      "epoch": 20.22,
      "grad_norm": 0.0979032889008522,
      "learning_rate": 0.0001962941176470588,
      "loss": 0.0325,
      "step": 2265
    },
    {
      "epoch": 20.23,
      "grad_norm": 0.059267591685056686,
      "learning_rate": 0.00019623529411764705,
      "loss": 0.0236,
      "step": 2266
    },
    {
      "epoch": 20.24,
      "grad_norm": 0.05923108384013176,
      "learning_rate": 0.0001961764705882353,
      "loss": 0.0271,
      "step": 2267
    },
    {
      "epoch": 20.25,
      "grad_norm": 0.06391739845275879,
      "learning_rate": 0.0001961176470588235,
      "loss": 0.0288,
      "step": 2268
    },
    {
      "epoch": 20.26,
      "grad_norm": 0.06083698943257332,
      "learning_rate": 0.00019605882352941175,
      "loss": 0.0285,
      "step": 2269
    },
    {
      "epoch": 20.27,
      "grad_norm": 0.05461670830845833,
      "learning_rate": 0.00019599999999999997,
      "loss": 0.0268,
      "step": 2270
    },
    {
      "epoch": 20.28,
      "grad_norm": 0.08133354783058167,
      "learning_rate": 0.00019594117647058824,
      "loss": 0.031,
      "step": 2271
    },
    {
      "epoch": 20.29,
      "grad_norm": 0.06891635805368423,
      "learning_rate": 0.00019588235294117645,
      "loss": 0.0288,
      "step": 2272
    },
    {
      "epoch": 20.29,
      "grad_norm": 0.054957352578639984,
      "learning_rate": 0.00019582352941176467,
      "loss": 0.0266,
      "step": 2273
    },
    {
      "epoch": 20.3,
      "grad_norm": 0.06640379130840302,
      "learning_rate": 0.00019576470588235294,
      "loss": 0.0284,
      "step": 2274
    },
    {
      "epoch": 20.31,
      "grad_norm": 0.056157227605581284,
      "learning_rate": 0.00019570588235294116,
      "loss": 0.0291,
      "step": 2275
    },
    {
      "epoch": 20.32,
      "grad_norm": 0.06232301518321037,
      "learning_rate": 0.00019564705882352937,
      "loss": 0.0251,
      "step": 2276
    },
    {
      "epoch": 20.33,
      "grad_norm": 0.06556544452905655,
      "learning_rate": 0.00019558823529411764,
      "loss": 0.0208,
      "step": 2277
    },
    {
      "epoch": 20.34,
      "grad_norm": 0.08982174098491669,
      "learning_rate": 0.00019552941176470586,
      "loss": 0.0355,
      "step": 2278
    },
    {
      "epoch": 20.35,
      "grad_norm": 0.05531982332468033,
      "learning_rate": 0.0001954705882352941,
      "loss": 0.0288,
      "step": 2279
    },
    {
      "epoch": 20.36,
      "grad_norm": 0.057537741959095,
      "learning_rate": 0.00019541176470588235,
      "loss": 0.0292,
      "step": 2280
    },
    {
      "epoch": 20.37,
      "grad_norm": 0.06393026560544968,
      "learning_rate": 0.00019535294117647056,
      "loss": 0.0259,
      "step": 2281
    },
    {
      "epoch": 20.38,
      "grad_norm": 0.06747432053089142,
      "learning_rate": 0.0001952941176470588,
      "loss": 0.0285,
      "step": 2282
    },
    {
      "epoch": 20.38,
      "grad_norm": 0.06794365495443344,
      "learning_rate": 0.00019523529411764702,
      "loss": 0.0258,
      "step": 2283
    },
    {
      "epoch": 20.39,
      "grad_norm": 0.10379933565855026,
      "learning_rate": 0.0001951764705882353,
      "loss": 0.0363,
      "step": 2284
    },
    {
      "epoch": 20.4,
      "grad_norm": 0.058310963213443756,
      "learning_rate": 0.0001951176470588235,
      "loss": 0.0255,
      "step": 2285
    },
    {
      "epoch": 20.41,
      "grad_norm": 0.059759609401226044,
      "learning_rate": 0.00019505882352941173,
      "loss": 0.0244,
      "step": 2286
    },
    {
      "epoch": 20.42,
      "grad_norm": 0.05604662001132965,
      "learning_rate": 0.000195,
      "loss": 0.0276,
      "step": 2287
    },
    {
      "epoch": 20.43,
      "grad_norm": 0.05281359329819679,
      "learning_rate": 0.0001949411764705882,
      "loss": 0.0277,
      "step": 2288
    },
    {
      "epoch": 20.44,
      "grad_norm": 0.062450770288705826,
      "learning_rate": 0.00019488235294117646,
      "loss": 0.026,
      "step": 2289
    },
    {
      "epoch": 20.45,
      "grad_norm": 0.1067199781537056,
      "learning_rate": 0.0001948235294117647,
      "loss": 0.0391,
      "step": 2290
    },
    {
      "epoch": 20.46,
      "grad_norm": 0.0533805713057518,
      "learning_rate": 0.00019476470588235292,
      "loss": 0.0297,
      "step": 2291
    },
    {
      "epoch": 20.46,
      "grad_norm": 0.05415339767932892,
      "learning_rate": 0.00019470588235294116,
      "loss": 0.0284,
      "step": 2292
    },
    {
      "epoch": 20.47,
      "grad_norm": 0.059836555272340775,
      "learning_rate": 0.0001946470588235294,
      "loss": 0.0284,
      "step": 2293
    },
    {
      "epoch": 20.48,
      "grad_norm": 0.0606420524418354,
      "learning_rate": 0.00019458823529411765,
      "loss": 0.0243,
      "step": 2294
    },
    {
      "epoch": 20.49,
      "grad_norm": 0.060408059507608414,
      "learning_rate": 0.00019452941176470586,
      "loss": 0.0242,
      "step": 2295
    },
    {
      "epoch": 20.5,
      "grad_norm": 0.09023467451334,
      "learning_rate": 0.00019447058823529408,
      "loss": 0.031,
      "step": 2296
    },
    {
      "epoch": 20.51,
      "grad_norm": 0.1027788519859314,
      "learning_rate": 0.00019441176470588235,
      "loss": 0.0351,
      "step": 2297
    },
    {
      "epoch": 20.52,
      "grad_norm": 0.05779183283448219,
      "learning_rate": 0.00019435294117647057,
      "loss": 0.0261,
      "step": 2298
    },
    {
      "epoch": 20.53,
      "grad_norm": 0.05863555893301964,
      "learning_rate": 0.0001942941176470588,
      "loss": 0.0275,
      "step": 2299
    },
    {
      "epoch": 20.54,
      "grad_norm": 0.058221373707056046,
      "learning_rate": 0.00019423529411764705,
      "loss": 0.0272,
      "step": 2300
    },
    {
      "epoch": 20.54,
      "eval_cer": 0.03385537051861707,
      "eval_loss": 0.19447565078735352,
      "eval_runtime": 22.2173,
      "eval_samples_per_second": 118.916,
      "eval_steps_per_second": 1.89,
      "eval_wer": 0.1129115430384768,
      "step": 2300
    },
    {
      "epoch": 20.54,
      "grad_norm": 0.06473752856254578,
      "learning_rate": 0.00019417647058823527,
      "loss": 0.0288,
      "step": 2301
    },
    {
      "epoch": 20.55,
      "grad_norm": 0.0694224163889885,
      "learning_rate": 0.0001941176470588235,
      "loss": 0.0316,
      "step": 2302
    },
    {
      "epoch": 20.56,
      "grad_norm": 0.11026883870363235,
      "learning_rate": 0.00019405882352941176,
      "loss": 0.0375,
      "step": 2303
    },
    {
      "epoch": 20.57,
      "grad_norm": 0.06142175942659378,
      "learning_rate": 0.00019399999999999997,
      "loss": 0.0287,
      "step": 2304
    },
    {
      "epoch": 20.58,
      "grad_norm": 0.0559944249689579,
      "learning_rate": 0.00019394117647058822,
      "loss": 0.0257,
      "step": 2305
    },
    {
      "epoch": 20.59,
      "grad_norm": 0.061539456248283386,
      "learning_rate": 0.00019388235294117643,
      "loss": 0.0283,
      "step": 2306
    },
    {
      "epoch": 20.6,
      "grad_norm": 0.059694308787584305,
      "learning_rate": 0.0001938235294117647,
      "loss": 0.0289,
      "step": 2307
    },
    {
      "epoch": 20.61,
      "grad_norm": 0.0647382065653801,
      "learning_rate": 0.00019376470588235292,
      "loss": 0.027,
      "step": 2308
    },
    {
      "epoch": 20.62,
      "grad_norm": 0.11706908047199249,
      "learning_rate": 0.00019370588235294114,
      "loss": 0.04,
      "step": 2309
    },
    {
      "epoch": 20.62,
      "grad_norm": 0.06547684967517853,
      "learning_rate": 0.0001936470588235294,
      "loss": 0.0301,
      "step": 2310
    },
    {
      "epoch": 20.63,
      "grad_norm": 0.056935232132673264,
      "learning_rate": 0.00019358823529411762,
      "loss": 0.0267,
      "step": 2311
    },
    {
      "epoch": 20.64,
      "grad_norm": 0.062037091702222824,
      "learning_rate": 0.00019352941176470587,
      "loss": 0.0267,
      "step": 2312
    },
    {
      "epoch": 20.65,
      "grad_norm": 0.06144816055893898,
      "learning_rate": 0.0001934705882352941,
      "loss": 0.0294,
      "step": 2313
    },
    {
      "epoch": 20.66,
      "grad_norm": 0.0712016299366951,
      "learning_rate": 0.00019341176470588233,
      "loss": 0.0307,
      "step": 2314
    },
    {
      "epoch": 20.67,
      "grad_norm": 0.1184035986661911,
      "learning_rate": 0.00019335294117647057,
      "loss": 0.0404,
      "step": 2315
    },
    {
      "epoch": 20.68,
      "grad_norm": 0.057029321789741516,
      "learning_rate": 0.0001932941176470588,
      "loss": 0.0293,
      "step": 2316
    },
    {
      "epoch": 20.69,
      "grad_norm": 0.05398004129528999,
      "learning_rate": 0.00019323529411764706,
      "loss": 0.0237,
      "step": 2317
    },
    {
      "epoch": 20.7,
      "grad_norm": 0.0657753124833107,
      "learning_rate": 0.00019317647058823527,
      "loss": 0.0307,
      "step": 2318
    },
    {
      "epoch": 20.71,
      "grad_norm": 0.06391260772943497,
      "learning_rate": 0.0001931176470588235,
      "loss": 0.031,
      "step": 2319
    },
    {
      "epoch": 20.71,
      "grad_norm": 0.07627767324447632,
      "learning_rate": 0.00019305882352941176,
      "loss": 0.0304,
      "step": 2320
    },
    {
      "epoch": 20.72,
      "grad_norm": 0.07448402047157288,
      "learning_rate": 0.00019299999999999997,
      "loss": 0.028,
      "step": 2321
    },
    {
      "epoch": 20.73,
      "grad_norm": 0.0848042294383049,
      "learning_rate": 0.00019294117647058825,
      "loss": 0.0386,
      "step": 2322
    },
    {
      "epoch": 20.74,
      "grad_norm": 0.04866953194141388,
      "learning_rate": 0.00019288235294117646,
      "loss": 0.0235,
      "step": 2323
    },
    {
      "epoch": 20.75,
      "grad_norm": 0.05988512188196182,
      "learning_rate": 0.00019282352941176468,
      "loss": 0.0285,
      "step": 2324
    },
    {
      "epoch": 20.76,
      "grad_norm": 0.06434763222932816,
      "learning_rate": 0.00019276470588235292,
      "loss": 0.029,
      "step": 2325
    },
    {
      "epoch": 20.77,
      "grad_norm": 0.06563811749219894,
      "learning_rate": 0.00019270588235294117,
      "loss": 0.0282,
      "step": 2326
    },
    {
      "epoch": 20.78,
      "grad_norm": 0.07576802372932434,
      "learning_rate": 0.0001926470588235294,
      "loss": 0.0271,
      "step": 2327
    },
    {
      "epoch": 20.79,
      "grad_norm": 0.0955667719244957,
      "learning_rate": 0.00019258823529411762,
      "loss": 0.0369,
      "step": 2328
    },
    {
      "epoch": 20.79,
      "grad_norm": 0.05960403382778168,
      "learning_rate": 0.00019252941176470584,
      "loss": 0.0289,
      "step": 2329
    },
    {
      "epoch": 20.8,
      "grad_norm": 0.05784255638718605,
      "learning_rate": 0.0001924705882352941,
      "loss": 0.0296,
      "step": 2330
    },
    {
      "epoch": 20.81,
      "grad_norm": 0.05947304144501686,
      "learning_rate": 0.00019241176470588233,
      "loss": 0.0263,
      "step": 2331
    },
    {
      "epoch": 20.82,
      "grad_norm": 0.06548023223876953,
      "learning_rate": 0.0001923529411764706,
      "loss": 0.0295,
      "step": 2332
    },
    {
      "epoch": 20.83,
      "grad_norm": 0.06398402154445648,
      "learning_rate": 0.00019229411764705881,
      "loss": 0.0259,
      "step": 2333
    },
    {
      "epoch": 20.84,
      "grad_norm": 0.1204400360584259,
      "learning_rate": 0.00019223529411764703,
      "loss": 0.0365,
      "step": 2334
    },
    {
      "epoch": 20.85,
      "grad_norm": 0.06722953915596008,
      "learning_rate": 0.00019217647058823527,
      "loss": 0.0322,
      "step": 2335
    },
    {
      "epoch": 20.86,
      "grad_norm": 0.05971822142601013,
      "learning_rate": 0.00019211764705882352,
      "loss": 0.0296,
      "step": 2336
    },
    {
      "epoch": 20.87,
      "grad_norm": 0.06084239110350609,
      "learning_rate": 0.00019205882352941173,
      "loss": 0.0247,
      "step": 2337
    },
    {
      "epoch": 20.88,
      "grad_norm": 0.0638883039355278,
      "learning_rate": 0.00019199999999999998,
      "loss": 0.0297,
      "step": 2338
    },
    {
      "epoch": 20.88,
      "grad_norm": 0.06581132858991623,
      "learning_rate": 0.00019194117647058822,
      "loss": 0.0311,
      "step": 2339
    },
    {
      "epoch": 20.89,
      "grad_norm": 0.09877829253673553,
      "learning_rate": 0.00019188235294117646,
      "loss": 0.0391,
      "step": 2340
    },
    {
      "epoch": 20.9,
      "grad_norm": 0.05891522020101547,
      "learning_rate": 0.00019182352941176468,
      "loss": 0.0281,
      "step": 2341
    },
    {
      "epoch": 20.91,
      "grad_norm": 0.05553098022937775,
      "learning_rate": 0.0001917647058823529,
      "loss": 0.0274,
      "step": 2342
    },
    {
      "epoch": 20.92,
      "grad_norm": 0.060920946300029755,
      "learning_rate": 0.00019170588235294117,
      "loss": 0.0282,
      "step": 2343
    },
    {
      "epoch": 20.93,
      "grad_norm": 0.05553551763296127,
      "learning_rate": 0.00019164705882352938,
      "loss": 0.0272,
      "step": 2344
    },
    {
      "epoch": 20.94,
      "grad_norm": 0.05533089116215706,
      "learning_rate": 0.00019158823529411765,
      "loss": 0.029,
      "step": 2345
    },
    {
      "epoch": 20.95,
      "grad_norm": 0.09240368008613586,
      "learning_rate": 0.00019152941176470587,
      "loss": 0.0299,
      "step": 2346
    },
    {
      "epoch": 20.96,
      "grad_norm": 0.11492682993412018,
      "learning_rate": 0.0001914705882352941,
      "loss": 0.0426,
      "step": 2347
    },
    {
      "epoch": 20.96,
      "grad_norm": 0.05383117124438286,
      "learning_rate": 0.00019141176470588233,
      "loss": 0.0236,
      "step": 2348
    },
    {
      "epoch": 20.97,
      "grad_norm": 0.06221175566315651,
      "learning_rate": 0.00019135294117647057,
      "loss": 0.0287,
      "step": 2349
    },
    {
      "epoch": 20.98,
      "grad_norm": 0.06276100873947144,
      "learning_rate": 0.00019129411764705882,
      "loss": 0.0283,
      "step": 2350
    },
    {
      "epoch": 20.99,
      "grad_norm": 0.07971462607383728,
      "learning_rate": 0.00019123529411764703,
      "loss": 0.0297,
      "step": 2351
    },
    {
      "epoch": 21.0,
      "grad_norm": 0.09926093369722366,
      "learning_rate": 0.00019117647058823528,
      "loss": 0.0368,
      "step": 2352
    },
    {
      "epoch": 21.01,
      "grad_norm": 0.05817103013396263,
      "learning_rate": 0.00019111764705882352,
      "loss": 0.0291,
      "step": 2353
    },
    {
      "epoch": 21.02,
      "grad_norm": 0.05710224062204361,
      "learning_rate": 0.00019105882352941174,
      "loss": 0.0237,
      "step": 2354
    },
    {
      "epoch": 21.03,
      "grad_norm": 0.05489571392536163,
      "learning_rate": 0.000191,
      "loss": 0.0268,
      "step": 2355
    },
    {
      "epoch": 21.04,
      "grad_norm": 0.05414852127432823,
      "learning_rate": 0.00019094117647058822,
      "loss": 0.0247,
      "step": 2356
    },
    {
      "epoch": 21.04,
      "grad_norm": 0.05747884884476662,
      "learning_rate": 0.00019088235294117644,
      "loss": 0.0255,
      "step": 2357
    },
    {
      "epoch": 21.05,
      "grad_norm": 0.07731017470359802,
      "learning_rate": 0.0001908235294117647,
      "loss": 0.0293,
      "step": 2358
    },
    {
      "epoch": 21.06,
      "grad_norm": 0.06619216501712799,
      "learning_rate": 0.00019076470588235293,
      "loss": 0.0303,
      "step": 2359
    },
    {
      "epoch": 21.07,
      "grad_norm": 0.0528254434466362,
      "learning_rate": 0.00019070588235294117,
      "loss": 0.0267,
      "step": 2360
    },
    {
      "epoch": 21.08,
      "grad_norm": 0.054829806089401245,
      "learning_rate": 0.0001906470588235294,
      "loss": 0.023,
      "step": 2361
    },
    {
      "epoch": 21.09,
      "grad_norm": 0.05584556236863136,
      "learning_rate": 0.00019058823529411763,
      "loss": 0.026,
      "step": 2362
    },
    {
      "epoch": 21.1,
      "grad_norm": 0.06575556844472885,
      "learning_rate": 0.00019052941176470587,
      "loss": 0.0289,
      "step": 2363
    },
    {
      "epoch": 21.11,
      "grad_norm": 0.06698581576347351,
      "learning_rate": 0.0001904705882352941,
      "loss": 0.025,
      "step": 2364
    },
    {
      "epoch": 21.12,
      "grad_norm": 0.08821806311607361,
      "learning_rate": 0.00019041176470588236,
      "loss": 0.0381,
      "step": 2365
    },
    {
      "epoch": 21.12,
      "grad_norm": 0.053715500980615616,
      "learning_rate": 0.00019035294117647058,
      "loss": 0.0295,
      "step": 2366
    },
    {
      "epoch": 21.13,
      "grad_norm": 0.061022866517305374,
      "learning_rate": 0.0001902941176470588,
      "loss": 0.0332,
      "step": 2367
    },
    {
      "epoch": 21.14,
      "grad_norm": 0.05637509003281593,
      "learning_rate": 0.00019023529411764706,
      "loss": 0.0253,
      "step": 2368
    },
    {
      "epoch": 21.15,
      "grad_norm": 0.058392640203237534,
      "learning_rate": 0.00019017647058823528,
      "loss": 0.0247,
      "step": 2369
    },
    {
      "epoch": 21.16,
      "grad_norm": 0.0698927491903305,
      "learning_rate": 0.0001901176470588235,
      "loss": 0.0277,
      "step": 2370
    },
    {
      "epoch": 21.17,
      "grad_norm": 0.09259064495563507,
      "learning_rate": 0.00019005882352941174,
      "loss": 0.0378,
      "step": 2371
    },
    {
      "epoch": 21.18,
      "grad_norm": 0.05133192986249924,
      "learning_rate": 0.00018999999999999998,
      "loss": 0.0241,
      "step": 2372
    },
    {
      "epoch": 21.19,
      "grad_norm": 0.05441851168870926,
      "learning_rate": 0.00018994117647058823,
      "loss": 0.0266,
      "step": 2373
    },
    {
      "epoch": 21.2,
      "grad_norm": 0.0640125423669815,
      "learning_rate": 0.00018988235294117644,
      "loss": 0.0279,
      "step": 2374
    },
    {
      "epoch": 21.21,
      "grad_norm": 0.05552051216363907,
      "learning_rate": 0.0001898235294117647,
      "loss": 0.0247,
      "step": 2375
    },
    {
      "epoch": 21.21,
      "grad_norm": 0.06344220042228699,
      "learning_rate": 0.00018976470588235293,
      "loss": 0.0257,
      "step": 2376
    },
    {
      "epoch": 21.22,
      "grad_norm": 0.13118135929107666,
      "learning_rate": 0.00018970588235294115,
      "loss": 0.0351,
      "step": 2377
    },
    {
      "epoch": 21.23,
      "grad_norm": 0.06830672919750214,
      "learning_rate": 0.00018964705882352942,
      "loss": 0.0229,
      "step": 2378
    },
    {
      "epoch": 21.24,
      "grad_norm": 0.050936296582221985,
      "learning_rate": 0.00018958823529411763,
      "loss": 0.0218,
      "step": 2379
    },
    {
      "epoch": 21.25,
      "grad_norm": 0.05940894037485123,
      "learning_rate": 0.00018952941176470585,
      "loss": 0.0279,
      "step": 2380
    },
    {
      "epoch": 21.26,
      "grad_norm": 0.06089415401220322,
      "learning_rate": 0.00018947058823529412,
      "loss": 0.0273,
      "step": 2381
    },
    {
      "epoch": 21.27,
      "grad_norm": 0.06544557958841324,
      "learning_rate": 0.00018941176470588234,
      "loss": 0.0261,
      "step": 2382
    },
    {
      "epoch": 21.28,
      "grad_norm": 0.0979183092713356,
      "learning_rate": 0.00018935294117647058,
      "loss": 0.0313,
      "step": 2383
    },
    {
      "epoch": 21.29,
      "grad_norm": 0.11826761811971664,
      "learning_rate": 0.0001892941176470588,
      "loss": 0.0387,
      "step": 2384
    },
    {
      "epoch": 21.29,
      "grad_norm": 0.0608263723552227,
      "learning_rate": 0.00018923529411764704,
      "loss": 0.0316,
      "step": 2385
    },
    {
      "epoch": 21.3,
      "grad_norm": 0.05365098640322685,
      "learning_rate": 0.00018917647058823528,
      "loss": 0.0222,
      "step": 2386
    },
    {
      "epoch": 21.31,
      "grad_norm": 0.058039531111717224,
      "learning_rate": 0.0001891176470588235,
      "loss": 0.0278,
      "step": 2387
    },
    {
      "epoch": 21.32,
      "grad_norm": 0.06796740740537643,
      "learning_rate": 0.00018905882352941177,
      "loss": 0.0217,
      "step": 2388
    },
    {
      "epoch": 21.33,
      "grad_norm": 0.06995701044797897,
      "learning_rate": 0.00018899999999999999,
      "loss": 0.028,
      "step": 2389
    },
    {
      "epoch": 21.34,
      "grad_norm": 0.09670662879943848,
      "learning_rate": 0.0001889411764705882,
      "loss": 0.0353,
      "step": 2390
    },
    {
      "epoch": 21.35,
      "grad_norm": 0.04985029995441437,
      "learning_rate": 0.00018888235294117647,
      "loss": 0.0257,
      "step": 2391
    },
    {
      "epoch": 21.36,
      "grad_norm": 0.057722825556993484,
      "learning_rate": 0.0001888235294117647,
      "loss": 0.0241,
      "step": 2392
    },
    {
      "epoch": 21.37,
      "grad_norm": 0.06418773531913757,
      "learning_rate": 0.00018876470588235293,
      "loss": 0.0269,
      "step": 2393
    },
    {
      "epoch": 21.38,
      "grad_norm": 0.0579729825258255,
      "learning_rate": 0.00018870588235294115,
      "loss": 0.0238,
      "step": 2394
    },
    {
      "epoch": 21.38,
      "grad_norm": 0.08441362529993057,
      "learning_rate": 0.0001886470588235294,
      "loss": 0.0322,
      "step": 2395
    },
    {
      "epoch": 21.39,
      "grad_norm": 0.1134045273065567,
      "learning_rate": 0.00018858823529411764,
      "loss": 0.0317,
      "step": 2396
    },
    {
      "epoch": 21.4,
      "grad_norm": 0.054769083857536316,
      "learning_rate": 0.00018852941176470585,
      "loss": 0.026,
      "step": 2397
    },
    {
      "epoch": 21.41,
      "grad_norm": 0.06260394304990768,
      "learning_rate": 0.00018847058823529412,
      "loss": 0.0256,
      "step": 2398
    },
    {
      "epoch": 21.42,
      "grad_norm": 0.060272328555583954,
      "learning_rate": 0.00018841176470588234,
      "loss": 0.0268,
      "step": 2399
    },
    {
      "epoch": 21.43,
      "grad_norm": 0.06387102603912354,
      "learning_rate": 0.00018835294117647056,
      "loss": 0.0278,
      "step": 2400
    },
    {
      "epoch": 21.43,
      "eval_cer": 0.03428006599692566,
      "eval_loss": 0.19497886300086975,
      "eval_runtime": 22.2455,
      "eval_samples_per_second": 118.766,
      "eval_steps_per_second": 1.888,
      "eval_wer": 0.11310987703292344,
      "step": 2400
    },
    {
      "epoch": 21.44,
      "grad_norm": 0.056825462728738785,
      "learning_rate": 0.00018829411764705883,
      "loss": 0.0226,
      "step": 2401
    },
    {
      "epoch": 21.45,
      "grad_norm": 0.1038694679737091,
      "learning_rate": 0.00018823529411764704,
      "loss": 0.0443,
      "step": 2402
    },
    {
      "epoch": 21.46,
      "grad_norm": 0.06339828670024872,
      "learning_rate": 0.00018817647058823526,
      "loss": 0.0286,
      "step": 2403
    },
    {
      "epoch": 21.46,
      "grad_norm": 0.05528872832655907,
      "learning_rate": 0.00018811764705882353,
      "loss": 0.0278,
      "step": 2404
    },
    {
      "epoch": 21.47,
      "grad_norm": 0.05746517330408096,
      "learning_rate": 0.00018805882352941175,
      "loss": 0.0255,
      "step": 2405
    },
    {
      "epoch": 21.48,
      "grad_norm": 0.05627716705203056,
      "learning_rate": 0.000188,
      "loss": 0.0273,
      "step": 2406
    },
    {
      "epoch": 21.49,
      "grad_norm": 0.05431535094976425,
      "learning_rate": 0.0001879411764705882,
      "loss": 0.0223,
      "step": 2407
    },
    {
      "epoch": 21.5,
      "grad_norm": 0.0860472247004509,
      "learning_rate": 0.00018788235294117645,
      "loss": 0.0303,
      "step": 2408
    },
    {
      "epoch": 21.51,
      "grad_norm": 0.09086859971284866,
      "learning_rate": 0.0001878235294117647,
      "loss": 0.0292,
      "step": 2409
    },
    {
      "epoch": 21.52,
      "grad_norm": 0.06806120276451111,
      "learning_rate": 0.0001877647058823529,
      "loss": 0.0304,
      "step": 2410
    },
    {
      "epoch": 21.53,
      "grad_norm": 0.05934057757258415,
      "learning_rate": 0.00018770588235294118,
      "loss": 0.0285,
      "step": 2411
    },
    {
      "epoch": 21.54,
      "grad_norm": 0.058098290115594864,
      "learning_rate": 0.0001876470588235294,
      "loss": 0.0246,
      "step": 2412
    },
    {
      "epoch": 21.54,
      "grad_norm": 0.059182148426771164,
      "learning_rate": 0.0001875882352941176,
      "loss": 0.0243,
      "step": 2413
    },
    {
      "epoch": 21.55,
      "grad_norm": 0.08074504882097244,
      "learning_rate": 0.00018752941176470588,
      "loss": 0.0274,
      "step": 2414
    },
    {
      "epoch": 21.56,
      "grad_norm": 0.10785527527332306,
      "learning_rate": 0.0001874705882352941,
      "loss": 0.0407,
      "step": 2415
    },
    {
      "epoch": 21.57,
      "grad_norm": 0.06091910973191261,
      "learning_rate": 0.00018741176470588234,
      "loss": 0.0262,
      "step": 2416
    },
    {
      "epoch": 21.58,
      "grad_norm": 0.05796665698289871,
      "learning_rate": 0.00018735294117647059,
      "loss": 0.0263,
      "step": 2417
    },
    {
      "epoch": 21.59,
      "grad_norm": 0.06581158936023712,
      "learning_rate": 0.0001872941176470588,
      "loss": 0.0297,
      "step": 2418
    },
    {
      "epoch": 21.6,
      "grad_norm": 0.07226616144180298,
      "learning_rate": 0.00018723529411764705,
      "loss": 0.0342,
      "step": 2419
    },
    {
      "epoch": 21.61,
      "grad_norm": 0.06742393225431442,
      "learning_rate": 0.00018717647058823526,
      "loss": 0.0268,
      "step": 2420
    },
    {
      "epoch": 21.62,
      "grad_norm": 0.11267576366662979,
      "learning_rate": 0.00018711764705882353,
      "loss": 0.0432,
      "step": 2421
    },
    {
      "epoch": 21.62,
      "grad_norm": 0.06181510165333748,
      "learning_rate": 0.00018705882352941175,
      "loss": 0.0245,
      "step": 2422
    },
    {
      "epoch": 21.63,
      "grad_norm": 0.058838147670030594,
      "learning_rate": 0.00018699999999999996,
      "loss": 0.0275,
      "step": 2423
    },
    {
      "epoch": 21.64,
      "grad_norm": 0.05783597007393837,
      "learning_rate": 0.00018694117647058824,
      "loss": 0.0237,
      "step": 2424
    },
    {
      "epoch": 21.65,
      "grad_norm": 0.0620993971824646,
      "learning_rate": 0.00018688235294117645,
      "loss": 0.0276,
      "step": 2425
    },
    {
      "epoch": 21.66,
      "grad_norm": 0.07318071275949478,
      "learning_rate": 0.0001868235294117647,
      "loss": 0.0297,
      "step": 2426
    },
    {
      "epoch": 21.67,
      "grad_norm": 0.12050062417984009,
      "learning_rate": 0.00018676470588235294,
      "loss": 0.0359,
      "step": 2427
    },
    {
      "epoch": 21.68,
      "grad_norm": 0.05566311627626419,
      "learning_rate": 0.00018670588235294115,
      "loss": 0.0258,
      "step": 2428
    },
    {
      "epoch": 21.69,
      "grad_norm": 0.059560324996709824,
      "learning_rate": 0.0001866470588235294,
      "loss": 0.0287,
      "step": 2429
    },
    {
      "epoch": 21.7,
      "grad_norm": 0.049338553100824356,
      "learning_rate": 0.00018658823529411761,
      "loss": 0.021,
      "step": 2430
    },
    {
      "epoch": 21.71,
      "grad_norm": 0.06321321427822113,
      "learning_rate": 0.00018652941176470586,
      "loss": 0.0249,
      "step": 2431
    },
    {
      "epoch": 21.71,
      "grad_norm": 0.05936066433787346,
      "learning_rate": 0.0001864705882352941,
      "loss": 0.0247,
      "step": 2432
    },
    {
      "epoch": 21.72,
      "grad_norm": 0.08597270399332047,
      "learning_rate": 0.00018641176470588232,
      "loss": 0.0308,
      "step": 2433
    },
    {
      "epoch": 21.73,
      "grad_norm": 0.07818442583084106,
      "learning_rate": 0.0001863529411764706,
      "loss": 0.0315,
      "step": 2434
    },
    {
      "epoch": 21.74,
      "grad_norm": 0.06305693835020065,
      "learning_rate": 0.0001862941176470588,
      "loss": 0.0296,
      "step": 2435
    },
    {
      "epoch": 21.75,
      "grad_norm": 0.06259511411190033,
      "learning_rate": 0.00018623529411764702,
      "loss": 0.0287,
      "step": 2436
    },
    {
      "epoch": 21.76,
      "grad_norm": 0.06238440424203873,
      "learning_rate": 0.0001861764705882353,
      "loss": 0.0247,
      "step": 2437
    },
    {
      "epoch": 21.77,
      "grad_norm": 0.05943160131573677,
      "learning_rate": 0.0001861176470588235,
      "loss": 0.0282,
      "step": 2438
    },
    {
      "epoch": 21.78,
      "grad_norm": 0.0842471495270729,
      "learning_rate": 0.00018605882352941175,
      "loss": 0.035,
      "step": 2439
    },
    {
      "epoch": 21.79,
      "grad_norm": 0.19098611176013947,
      "learning_rate": 0.000186,
      "loss": 0.0354,
      "step": 2440
    },
    {
      "epoch": 21.79,
      "grad_norm": 0.05623243749141693,
      "learning_rate": 0.0001859411764705882,
      "loss": 0.029,
      "step": 2441
    },
    {
      "epoch": 21.8,
      "grad_norm": 0.05806316062808037,
      "learning_rate": 0.00018588235294117645,
      "loss": 0.0259,
      "step": 2442
    },
    {
      "epoch": 21.81,
      "grad_norm": 0.0664072260260582,
      "learning_rate": 0.00018582352941176467,
      "loss": 0.0312,
      "step": 2443
    },
    {
      "epoch": 21.82,
      "grad_norm": 0.058459002524614334,
      "learning_rate": 0.00018576470588235294,
      "loss": 0.028,
      "step": 2444
    },
    {
      "epoch": 21.83,
      "grad_norm": 0.07315807044506073,
      "learning_rate": 0.00018570588235294116,
      "loss": 0.0258,
      "step": 2445
    },
    {
      "epoch": 21.84,
      "grad_norm": 0.10811853408813477,
      "learning_rate": 0.00018564705882352937,
      "loss": 0.0451,
      "step": 2446
    },
    {
      "epoch": 21.85,
      "grad_norm": 0.04934457316994667,
      "learning_rate": 0.00018558823529411764,
      "loss": 0.0248,
      "step": 2447
    },
    {
      "epoch": 21.86,
      "grad_norm": 0.0537499338388443,
      "learning_rate": 0.00018552941176470586,
      "loss": 0.0274,
      "step": 2448
    },
    {
      "epoch": 21.87,
      "grad_norm": 0.06229821965098381,
      "learning_rate": 0.0001854705882352941,
      "loss": 0.0293,
      "step": 2449
    },
    {
      "epoch": 21.88,
      "grad_norm": 0.07059778273105621,
      "learning_rate": 0.00018541176470588235,
      "loss": 0.03,
      "step": 2450
    },
    {
      "epoch": 21.88,
      "grad_norm": 0.061576519161462784,
      "learning_rate": 0.00018535294117647056,
      "loss": 0.0275,
      "step": 2451
    },
    {
      "epoch": 21.89,
      "grad_norm": 0.09695862233638763,
      "learning_rate": 0.0001852941176470588,
      "loss": 0.0382,
      "step": 2452
    },
    {
      "epoch": 21.9,
      "grad_norm": 0.054285623133182526,
      "learning_rate": 0.00018523529411764705,
      "loss": 0.0254,
      "step": 2453
    },
    {
      "epoch": 21.91,
      "grad_norm": 0.0540817454457283,
      "learning_rate": 0.0001851764705882353,
      "loss": 0.0267,
      "step": 2454
    },
    {
      "epoch": 21.92,
      "grad_norm": 0.061084263026714325,
      "learning_rate": 0.0001851176470588235,
      "loss": 0.0281,
      "step": 2455
    },
    {
      "epoch": 21.93,
      "grad_norm": 0.056585874408483505,
      "learning_rate": 0.00018505882352941173,
      "loss": 0.0264,
      "step": 2456
    },
    {
      "epoch": 21.94,
      "grad_norm": 0.06149120256304741,
      "learning_rate": 0.000185,
      "loss": 0.0251,
      "step": 2457
    },
    {
      "epoch": 21.95,
      "grad_norm": 0.08614380657672882,
      "learning_rate": 0.00018494117647058821,
      "loss": 0.0251,
      "step": 2458
    },
    {
      "epoch": 21.96,
      "grad_norm": 0.1088634803891182,
      "learning_rate": 0.00018488235294117646,
      "loss": 0.0406,
      "step": 2459
    },
    {
      "epoch": 21.96,
      "grad_norm": 0.0585268959403038,
      "learning_rate": 0.0001848235294117647,
      "loss": 0.0262,
      "step": 2460
    },
    {
      "epoch": 21.97,
      "grad_norm": 0.061667658388614655,
      "learning_rate": 0.00018476470588235292,
      "loss": 0.0314,
      "step": 2461
    },
    {
      "epoch": 21.98,
      "grad_norm": 0.06478738784790039,
      "learning_rate": 0.00018470588235294116,
      "loss": 0.0292,
      "step": 2462
    },
    {
      "epoch": 21.99,
      "grad_norm": 0.06945609301328659,
      "learning_rate": 0.0001846470588235294,
      "loss": 0.0259,
      "step": 2463
    },
    {
      "epoch": 22.0,
      "grad_norm": 0.09599378705024719,
      "learning_rate": 0.00018458823529411762,
      "loss": 0.0357,
      "step": 2464
    },
    {
      "epoch": 22.01,
      "grad_norm": 0.045506224036216736,
      "learning_rate": 0.00018452941176470586,
      "loss": 0.0249,
      "step": 2465
    },
    {
      "epoch": 22.02,
      "grad_norm": 0.057261284440755844,
      "learning_rate": 0.00018447058823529408,
      "loss": 0.0231,
      "step": 2466
    },
    {
      "epoch": 22.03,
      "grad_norm": 0.056278035044670105,
      "learning_rate": 0.00018441176470588235,
      "loss": 0.027,
      "step": 2467
    },
    {
      "epoch": 22.04,
      "grad_norm": 0.060056935995817184,
      "learning_rate": 0.00018435294117647057,
      "loss": 0.0322,
      "step": 2468
    },
    {
      "epoch": 22.04,
      "grad_norm": 0.05936797335743904,
      "learning_rate": 0.00018429411764705878,
      "loss": 0.0253,
      "step": 2469
    },
    {
      "epoch": 22.05,
      "grad_norm": 0.10536514967679977,
      "learning_rate": 0.00018423529411764705,
      "loss": 0.0284,
      "step": 2470
    },
    {
      "epoch": 22.06,
      "grad_norm": 0.09906871616840363,
      "learning_rate": 0.00018417647058823527,
      "loss": 0.0334,
      "step": 2471
    },
    {
      "epoch": 22.07,
      "grad_norm": 0.05583423376083374,
      "learning_rate": 0.0001841176470588235,
      "loss": 0.0257,
      "step": 2472
    },
    {
      "epoch": 22.08,
      "grad_norm": 0.05540235713124275,
      "learning_rate": 0.00018405882352941176,
      "loss": 0.0242,
      "step": 2473
    },
    {
      "epoch": 22.09,
      "grad_norm": 0.055019501596689224,
      "learning_rate": 0.00018399999999999997,
      "loss": 0.0264,
      "step": 2474
    },
    {
      "epoch": 22.1,
      "grad_norm": 0.06339872628450394,
      "learning_rate": 0.00018394117647058822,
      "loss": 0.0285,
      "step": 2475
    },
    {
      "epoch": 22.11,
      "grad_norm": 0.07679237425327301,
      "learning_rate": 0.00018388235294117646,
      "loss": 0.029,
      "step": 2476
    },
    {
      "epoch": 22.12,
      "grad_norm": 0.09308271110057831,
      "learning_rate": 0.0001838235294117647,
      "loss": 0.035,
      "step": 2477
    },
    {
      "epoch": 22.12,
      "grad_norm": 0.05326671153306961,
      "learning_rate": 0.00018376470588235292,
      "loss": 0.0282,
      "step": 2478
    },
    {
      "epoch": 22.13,
      "grad_norm": 0.053068701177835464,
      "learning_rate": 0.00018370588235294114,
      "loss": 0.0256,
      "step": 2479
    },
    {
      "epoch": 22.14,
      "grad_norm": 0.057122208178043365,
      "learning_rate": 0.0001836470588235294,
      "loss": 0.0272,
      "step": 2480
    },
    {
      "epoch": 22.15,
      "grad_norm": 0.06364129483699799,
      "learning_rate": 0.00018358823529411762,
      "loss": 0.0278,
      "step": 2481
    },
    {
      "epoch": 22.16,
      "grad_norm": 0.06380900740623474,
      "learning_rate": 0.0001835294117647059,
      "loss": 0.0276,
      "step": 2482
    },
    {
      "epoch": 22.17,
      "grad_norm": 0.11803562939167023,
      "learning_rate": 0.0001834705882352941,
      "loss": 0.0366,
      "step": 2483
    },
    {
      "epoch": 22.18,
      "grad_norm": 0.058370329439640045,
      "learning_rate": 0.00018341176470588233,
      "loss": 0.0266,
      "step": 2484
    },
    {
      "epoch": 22.19,
      "grad_norm": 0.054435014724731445,
      "learning_rate": 0.00018335294117647057,
      "loss": 0.0246,
      "step": 2485
    },
    {
      "epoch": 22.2,
      "grad_norm": 0.057081181555986404,
      "learning_rate": 0.0001832941176470588,
      "loss": 0.0262,
      "step": 2486
    },
    {
      "epoch": 22.21,
      "grad_norm": 0.05508088320493698,
      "learning_rate": 0.00018323529411764706,
      "loss": 0.0256,
      "step": 2487
    },
    {
      "epoch": 22.21,
      "grad_norm": 0.06385696679353714,
      "learning_rate": 0.00018317647058823527,
      "loss": 0.0292,
      "step": 2488
    },
    {
      "epoch": 22.22,
      "grad_norm": 0.13734489679336548,
      "learning_rate": 0.0001831176470588235,
      "loss": 0.0457,
      "step": 2489
    },
    {
      "epoch": 22.23,
      "grad_norm": 0.05183258280158043,
      "learning_rate": 0.00018305882352941176,
      "loss": 0.0255,
      "step": 2490
    },
    {
      "epoch": 22.24,
      "grad_norm": 0.05746317282319069,
      "learning_rate": 0.00018299999999999998,
      "loss": 0.0322,
      "step": 2491
    },
    {
      "epoch": 22.25,
      "grad_norm": 0.05643812194466591,
      "learning_rate": 0.00018294117647058825,
      "loss": 0.0242,
      "step": 2492
    },
    {
      "epoch": 22.26,
      "grad_norm": 0.053931787610054016,
      "learning_rate": 0.00018288235294117646,
      "loss": 0.0261,
      "step": 2493
    },
    {
      "epoch": 22.27,
      "grad_norm": 0.07110514491796494,
      "learning_rate": 0.00018282352941176468,
      "loss": 0.0338,
      "step": 2494
    },
    {
      "epoch": 22.28,
      "grad_norm": 0.08726546913385391,
      "learning_rate": 0.00018276470588235292,
      "loss": 0.0325,
      "step": 2495
    },
    {
      "epoch": 22.29,
      "grad_norm": 0.06899622082710266,
      "learning_rate": 0.00018270588235294117,
      "loss": 0.0313,
      "step": 2496
    },
    {
      "epoch": 22.29,
      "grad_norm": 0.05990142747759819,
      "learning_rate": 0.00018264705882352938,
      "loss": 0.0287,
      "step": 2497
    },
    {
      "epoch": 22.3,
      "grad_norm": 0.055539678782224655,
      "learning_rate": 0.00018258823529411763,
      "loss": 0.0233,
      "step": 2498
    },
    {
      "epoch": 22.31,
      "grad_norm": 0.06318056583404541,
      "learning_rate": 0.00018252941176470587,
      "loss": 0.0279,
      "step": 2499
    },
    {
      "epoch": 22.32,
      "grad_norm": 0.06352417171001434,
      "learning_rate": 0.0001824705882352941,
      "loss": 0.0254,
      "step": 2500
    },
    {
      "epoch": 22.32,
      "eval_cer": 0.03299846282787055,
      "eval_loss": 0.20145094394683838,
      "eval_runtime": 23.1799,
      "eval_samples_per_second": 113.978,
      "eval_steps_per_second": 1.812,
      "eval_wer": 0.10973819912733042,
      "step": 2500
    },
    {
      "epoch": 22.33,
      "grad_norm": 0.08036191761493683,
      "learning_rate": 0.00018241176470588233,
      "loss": 0.0242,
      "step": 2501
    },
    {
      "epoch": 22.34,
      "grad_norm": 0.11871591210365295,
      "learning_rate": 0.00018235294117647055,
      "loss": 0.0309,
      "step": 2502
    },
    {
      "epoch": 22.35,
      "grad_norm": 0.055190663784742355,
      "learning_rate": 0.00018229411764705882,
      "loss": 0.0282,
      "step": 2503
    },
    {
      "epoch": 22.36,
      "grad_norm": 0.06247410550713539,
      "learning_rate": 0.00018223529411764703,
      "loss": 0.0249,
      "step": 2504
    },
    {
      "epoch": 22.37,
      "grad_norm": 0.05551055073738098,
      "learning_rate": 0.0001821764705882353,
      "loss": 0.0262,
      "step": 2505
    },
    {
      "epoch": 22.38,
      "grad_norm": 0.05253708362579346,
      "learning_rate": 0.00018211764705882352,
      "loss": 0.0263,
      "step": 2506
    },
    {
      "epoch": 22.38,
      "grad_norm": 0.06559304893016815,
      "learning_rate": 0.00018205882352941174,
      "loss": 0.0286,
      "step": 2507
    },
    {
      "epoch": 22.39,
      "grad_norm": 0.10570012778043747,
      "learning_rate": 0.00018199999999999998,
      "loss": 0.0368,
      "step": 2508
    },
    {
      "epoch": 22.4,
      "grad_norm": 0.06190425157546997,
      "learning_rate": 0.00018194117647058822,
      "loss": 0.027,
      "step": 2509
    },
    {
      "epoch": 22.41,
      "grad_norm": 0.050871044397354126,
      "learning_rate": 0.00018188235294117647,
      "loss": 0.0248,
      "step": 2510
    },
    {
      "epoch": 22.42,
      "grad_norm": 0.05783969163894653,
      "learning_rate": 0.00018182352941176468,
      "loss": 0.0253,
      "step": 2511
    },
    {
      "epoch": 22.43,
      "grad_norm": 0.057936739176511765,
      "learning_rate": 0.00018176470588235293,
      "loss": 0.025,
      "step": 2512
    },
    {
      "epoch": 22.44,
      "grad_norm": 0.05643759295344353,
      "learning_rate": 0.00018170588235294117,
      "loss": 0.0255,
      "step": 2513
    },
    {
      "epoch": 22.45,
      "grad_norm": 0.10881562530994415,
      "learning_rate": 0.00018164705882352939,
      "loss": 0.0335,
      "step": 2514
    },
    {
      "epoch": 22.46,
      "grad_norm": 0.060932427644729614,
      "learning_rate": 0.00018158823529411766,
      "loss": 0.0291,
      "step": 2515
    },
    {
      "epoch": 22.46,
      "grad_norm": 0.054830025881528854,
      "learning_rate": 0.00018152941176470587,
      "loss": 0.0255,
      "step": 2516
    },
    {
      "epoch": 22.47,
      "grad_norm": 0.05401693284511566,
      "learning_rate": 0.0001814705882352941,
      "loss": 0.0223,
      "step": 2517
    },
    {
      "epoch": 22.48,
      "grad_norm": 0.06534389406442642,
      "learning_rate": 0.00018141176470588236,
      "loss": 0.0272,
      "step": 2518
    },
    {
      "epoch": 22.49,
      "grad_norm": 0.05824942886829376,
      "learning_rate": 0.00018135294117647058,
      "loss": 0.026,
      "step": 2519
    },
    {
      "epoch": 22.5,
      "grad_norm": 0.0738939642906189,
      "learning_rate": 0.00018129411764705882,
      "loss": 0.0286,
      "step": 2520
    },
    {
      "epoch": 22.51,
      "grad_norm": 0.11110525578260422,
      "learning_rate": 0.00018123529411764703,
      "loss": 0.0299,
      "step": 2521
    },
    {
      "epoch": 22.52,
      "grad_norm": 0.0617167092859745,
      "learning_rate": 0.00018117647058823528,
      "loss": 0.0286,
      "step": 2522
    },
    {
      "epoch": 22.53,
      "grad_norm": 0.05880562588572502,
      "learning_rate": 0.00018111764705882352,
      "loss": 0.03,
      "step": 2523
    },
    {
      "epoch": 22.54,
      "grad_norm": 0.05205706134438515,
      "learning_rate": 0.00018105882352941174,
      "loss": 0.0253,
      "step": 2524
    },
    {
      "epoch": 22.54,
      "grad_norm": 0.05533972755074501,
      "learning_rate": 0.000181,
      "loss": 0.0282,
      "step": 2525
    },
    {
      "epoch": 22.55,
      "grad_norm": 0.06595009565353394,
      "learning_rate": 0.00018094117647058822,
      "loss": 0.0265,
      "step": 2526
    },
    {
      "epoch": 22.56,
      "grad_norm": 0.09634100645780563,
      "learning_rate": 0.00018088235294117644,
      "loss": 0.0436,
      "step": 2527
    },
    {
      "epoch": 22.57,
      "grad_norm": 0.045013271272182465,
      "learning_rate": 0.0001808235294117647,
      "loss": 0.0234,
      "step": 2528
    },
    {
      "epoch": 22.58,
      "grad_norm": 0.05480833351612091,
      "learning_rate": 0.00018076470588235293,
      "loss": 0.03,
      "step": 2529
    },
    {
      "epoch": 22.59,
      "grad_norm": 0.058044854551553726,
      "learning_rate": 0.00018070588235294114,
      "loss": 0.0253,
      "step": 2530
    },
    {
      "epoch": 22.6,
      "grad_norm": 0.061460185796022415,
      "learning_rate": 0.0001806470588235294,
      "loss": 0.0275,
      "step": 2531
    },
    {
      "epoch": 22.61,
      "grad_norm": 0.06623505055904388,
      "learning_rate": 0.00018058823529411763,
      "loss": 0.0252,
      "step": 2532
    },
    {
      "epoch": 22.62,
      "grad_norm": 0.11571113765239716,
      "learning_rate": 0.00018052941176470587,
      "loss": 0.043,
      "step": 2533
    },
    {
      "epoch": 22.62,
      "grad_norm": 0.048881854861974716,
      "learning_rate": 0.0001804705882352941,
      "loss": 0.0262,
      "step": 2534
    },
    {
      "epoch": 22.63,
      "grad_norm": 0.0761791467666626,
      "learning_rate": 0.00018041176470588233,
      "loss": 0.0238,
      "step": 2535
    },
    {
      "epoch": 22.64,
      "grad_norm": 0.06024662405252457,
      "learning_rate": 0.00018035294117647058,
      "loss": 0.0212,
      "step": 2536
    },
    {
      "epoch": 22.65,
      "grad_norm": 0.06886886805295944,
      "learning_rate": 0.0001802941176470588,
      "loss": 0.0257,
      "step": 2537
    },
    {
      "epoch": 22.66,
      "grad_norm": 0.06255342811346054,
      "learning_rate": 0.00018023529411764706,
      "loss": 0.0252,
      "step": 2538
    },
    {
      "epoch": 22.67,
      "grad_norm": 0.10939238965511322,
      "learning_rate": 0.00018017647058823528,
      "loss": 0.0438,
      "step": 2539
    },
    {
      "epoch": 22.68,
      "grad_norm": 0.05792367085814476,
      "learning_rate": 0.0001801176470588235,
      "loss": 0.0261,
      "step": 2540
    },
    {
      "epoch": 22.69,
      "grad_norm": 0.05639879032969475,
      "learning_rate": 0.00018005882352941177,
      "loss": 0.0262,
      "step": 2541
    },
    {
      "epoch": 22.7,
      "grad_norm": 0.06334155797958374,
      "learning_rate": 0.00017999999999999998,
      "loss": 0.0247,
      "step": 2542
    },
    {
      "epoch": 22.71,
      "grad_norm": 0.0545668862760067,
      "learning_rate": 0.00017994117647058823,
      "loss": 0.023,
      "step": 2543
    },
    {
      "epoch": 22.71,
      "grad_norm": 0.05709097906947136,
      "learning_rate": 0.00017988235294117644,
      "loss": 0.0208,
      "step": 2544
    },
    {
      "epoch": 22.72,
      "grad_norm": 0.08601279556751251,
      "learning_rate": 0.0001798235294117647,
      "loss": 0.0334,
      "step": 2545
    },
    {
      "epoch": 22.73,
      "grad_norm": 0.15349067747592926,
      "learning_rate": 0.00017976470588235293,
      "loss": 0.0343,
      "step": 2546
    },
    {
      "epoch": 22.74,
      "grad_norm": 0.065592460334301,
      "learning_rate": 0.00017970588235294115,
      "loss": 0.0257,
      "step": 2547
    },
    {
      "epoch": 22.75,
      "grad_norm": 0.05844647437334061,
      "learning_rate": 0.00017964705882352942,
      "loss": 0.0256,
      "step": 2548
    },
    {
      "epoch": 22.76,
      "grad_norm": 0.05912429094314575,
      "learning_rate": 0.00017958823529411763,
      "loss": 0.0249,
      "step": 2549
    },
    {
      "epoch": 22.77,
      "grad_norm": 0.05476592481136322,
      "learning_rate": 0.00017952941176470585,
      "loss": 0.0238,
      "step": 2550
    },
    {
      "epoch": 22.78,
      "grad_norm": 0.06735950708389282,
      "learning_rate": 0.00017947058823529412,
      "loss": 0.0245,
      "step": 2551
    },
    {
      "epoch": 22.79,
      "grad_norm": 0.09796684235334396,
      "learning_rate": 0.00017941176470588234,
      "loss": 0.0369,
      "step": 2552
    },
    {
      "epoch": 22.79,
      "grad_norm": 0.05760080739855766,
      "learning_rate": 0.00017935294117647058,
      "loss": 0.0253,
      "step": 2553
    },
    {
      "epoch": 22.8,
      "grad_norm": 0.05072222277522087,
      "learning_rate": 0.0001792941176470588,
      "loss": 0.0271,
      "step": 2554
    },
    {
      "epoch": 22.81,
      "grad_norm": 0.05979233235120773,
      "learning_rate": 0.00017923529411764704,
      "loss": 0.0307,
      "step": 2555
    },
    {
      "epoch": 22.82,
      "grad_norm": 0.0697891116142273,
      "learning_rate": 0.00017917647058823528,
      "loss": 0.0258,
      "step": 2556
    },
    {
      "epoch": 22.83,
      "grad_norm": 0.06077735498547554,
      "learning_rate": 0.0001791176470588235,
      "loss": 0.0271,
      "step": 2557
    },
    {
      "epoch": 22.84,
      "grad_norm": 0.11549992859363556,
      "learning_rate": 0.00017905882352941174,
      "loss": 0.0383,
      "step": 2558
    },
    {
      "epoch": 22.85,
      "grad_norm": 0.054385095834732056,
      "learning_rate": 0.000179,
      "loss": 0.026,
      "step": 2559
    },
    {
      "epoch": 22.86,
      "grad_norm": 0.06153104826807976,
      "learning_rate": 0.0001789411764705882,
      "loss": 0.0284,
      "step": 2560
    },
    {
      "epoch": 22.87,
      "grad_norm": 0.05704614892601967,
      "learning_rate": 0.00017888235294117647,
      "loss": 0.0236,
      "step": 2561
    },
    {
      "epoch": 22.88,
      "grad_norm": 0.06553328037261963,
      "learning_rate": 0.0001788235294117647,
      "loss": 0.0254,
      "step": 2562
    },
    {
      "epoch": 22.88,
      "grad_norm": 0.06245945021510124,
      "learning_rate": 0.0001787647058823529,
      "loss": 0.0269,
      "step": 2563
    },
    {
      "epoch": 22.89,
      "grad_norm": 0.12217217683792114,
      "learning_rate": 0.00017870588235294118,
      "loss": 0.0344,
      "step": 2564
    },
    {
      "epoch": 22.9,
      "grad_norm": 0.0552246980369091,
      "learning_rate": 0.0001786470588235294,
      "loss": 0.025,
      "step": 2565
    },
    {
      "epoch": 22.91,
      "grad_norm": 0.05491182580590248,
      "learning_rate": 0.00017858823529411764,
      "loss": 0.0272,
      "step": 2566
    },
    {
      "epoch": 22.92,
      "grad_norm": 0.056860268115997314,
      "learning_rate": 0.00017852941176470585,
      "loss": 0.0282,
      "step": 2567
    },
    {
      "epoch": 22.93,
      "grad_norm": 0.05380900576710701,
      "learning_rate": 0.0001784705882352941,
      "loss": 0.0226,
      "step": 2568
    },
    {
      "epoch": 22.94,
      "grad_norm": 0.06300035864114761,
      "learning_rate": 0.00017841176470588234,
      "loss": 0.0272,
      "step": 2569
    },
    {
      "epoch": 22.95,
      "grad_norm": 0.0868477001786232,
      "learning_rate": 0.00017835294117647056,
      "loss": 0.026,
      "step": 2570
    },
    {
      "epoch": 22.96,
      "grad_norm": 0.09526658058166504,
      "learning_rate": 0.00017829411764705883,
      "loss": 0.0296,
      "step": 2571
    },
    {
      "epoch": 22.96,
      "grad_norm": 0.05786629021167755,
      "learning_rate": 0.00017823529411764704,
      "loss": 0.0281,
      "step": 2572
    },
    {
      "epoch": 22.97,
      "grad_norm": 0.06393415480852127,
      "learning_rate": 0.00017817647058823526,
      "loss": 0.0285,
      "step": 2573
    },
    {
      "epoch": 22.98,
      "grad_norm": 0.059029340744018555,
      "learning_rate": 0.00017811764705882353,
      "loss": 0.0214,
      "step": 2574
    },
    {
      "epoch": 22.99,
      "grad_norm": 0.060377780348062515,
      "learning_rate": 0.00017805882352941175,
      "loss": 0.0245,
      "step": 2575
    },
    {
      "epoch": 23.0,
      "grad_norm": 0.07968320697546005,
      "learning_rate": 0.000178,
      "loss": 0.0305,
      "step": 2576
    },
    {
      "epoch": 23.01,
      "grad_norm": 0.055544592440128326,
      "learning_rate": 0.00017794117647058823,
      "loss": 0.0254,
      "step": 2577
    },
    {
      "epoch": 23.02,
      "grad_norm": 0.05564003810286522,
      "learning_rate": 0.00017788235294117645,
      "loss": 0.0275,
      "step": 2578
    },
    {
      "epoch": 23.03,
      "grad_norm": 0.05317571386694908,
      "learning_rate": 0.0001778235294117647,
      "loss": 0.0241,
      "step": 2579
    },
    {
      "epoch": 23.04,
      "grad_norm": 0.05589379370212555,
      "learning_rate": 0.0001777647058823529,
      "loss": 0.0249,
      "step": 2580
    },
    {
      "epoch": 23.04,
      "grad_norm": 0.06784486025571823,
      "learning_rate": 0.00017770588235294118,
      "loss": 0.029,
      "step": 2581
    },
    {
      "epoch": 23.05,
      "grad_norm": 0.07419680804014206,
      "learning_rate": 0.0001776470588235294,
      "loss": 0.0248,
      "step": 2582
    },
    {
      "epoch": 23.06,
      "grad_norm": 0.07389215379953384,
      "learning_rate": 0.0001775882352941176,
      "loss": 0.0284,
      "step": 2583
    },
    {
      "epoch": 23.07,
      "grad_norm": 0.05012824758887291,
      "learning_rate": 0.00017752941176470588,
      "loss": 0.0268,
      "step": 2584
    },
    {
      "epoch": 23.08,
      "grad_norm": 0.05442946404218674,
      "learning_rate": 0.0001774705882352941,
      "loss": 0.0227,
      "step": 2585
    },
    {
      "epoch": 23.09,
      "grad_norm": 0.055399347096681595,
      "learning_rate": 0.00017741176470588234,
      "loss": 0.0294,
      "step": 2586
    },
    {
      "epoch": 23.1,
      "grad_norm": 0.05383411794900894,
      "learning_rate": 0.00017735294117647059,
      "loss": 0.023,
      "step": 2587
    },
    {
      "epoch": 23.11,
      "grad_norm": 0.06131847947835922,
      "learning_rate": 0.0001772941176470588,
      "loss": 0.0241,
      "step": 2588
    },
    {
      "epoch": 23.12,
      "grad_norm": 0.10607028007507324,
      "learning_rate": 0.00017723529411764705,
      "loss": 0.0326,
      "step": 2589
    },
    {
      "epoch": 23.12,
      "grad_norm": 0.05809440463781357,
      "learning_rate": 0.00017717647058823526,
      "loss": 0.0255,
      "step": 2590
    },
    {
      "epoch": 23.13,
      "grad_norm": 0.05491780862212181,
      "learning_rate": 0.0001771176470588235,
      "loss": 0.0238,
      "step": 2591
    },
    {
      "epoch": 23.14,
      "grad_norm": 0.0558369942009449,
      "learning_rate": 0.00017705882352941175,
      "loss": 0.0262,
      "step": 2592
    },
    {
      "epoch": 23.15,
      "grad_norm": 0.05765916779637337,
      "learning_rate": 0.00017699999999999997,
      "loss": 0.0234,
      "step": 2593
    },
    {
      "epoch": 23.16,
      "grad_norm": 0.05445518344640732,
      "learning_rate": 0.00017694117647058824,
      "loss": 0.0236,
      "step": 2594
    },
    {
      "epoch": 23.17,
      "grad_norm": 0.1029948964715004,
      "learning_rate": 0.00017688235294117645,
      "loss": 0.0375,
      "step": 2595
    },
    {
      "epoch": 23.18,
      "grad_norm": 0.05391351133584976,
      "learning_rate": 0.00017682352941176467,
      "loss": 0.0261,
      "step": 2596
    },
    {
      "epoch": 23.19,
      "grad_norm": 0.05429863557219505,
      "learning_rate": 0.00017676470588235294,
      "loss": 0.0277,
      "step": 2597
    },
    {
      "epoch": 23.2,
      "grad_norm": 0.05892942100763321,
      "learning_rate": 0.00017670588235294116,
      "loss": 0.0261,
      "step": 2598
    },
    {
      "epoch": 23.21,
      "grad_norm": 0.05917773395776749,
      "learning_rate": 0.0001766470588235294,
      "loss": 0.026,
      "step": 2599
    },
    {
      "epoch": 23.21,
      "grad_norm": 0.057894766330718994,
      "learning_rate": 0.00017658823529411764,
      "loss": 0.0204,
      "step": 2600
    },
    {
      "epoch": 23.21,
      "eval_cer": 0.032618867754338095,
      "eval_loss": 0.19516734778881073,
      "eval_runtime": 22.3281,
      "eval_samples_per_second": 118.326,
      "eval_steps_per_second": 1.881,
      "eval_wer": 0.10694168980563269,
      "step": 2600
    },
    {
      "epoch": 23.22,
      "grad_norm": 0.1185203269124031,
      "learning_rate": 0.00017652941176470586,
      "loss": 0.0386,
      "step": 2601
    },
    {
      "epoch": 23.23,
      "grad_norm": 0.05277341604232788,
      "learning_rate": 0.0001764705882352941,
      "loss": 0.0236,
      "step": 2602
    },
    {
      "epoch": 23.24,
      "grad_norm": 0.049617115408182144,
      "learning_rate": 0.00017641176470588232,
      "loss": 0.0259,
      "step": 2603
    },
    {
      "epoch": 23.25,
      "grad_norm": 0.05507615953683853,
      "learning_rate": 0.0001763529411764706,
      "loss": 0.0277,
      "step": 2604
    },
    {
      "epoch": 23.26,
      "grad_norm": 0.05926371365785599,
      "learning_rate": 0.0001762941176470588,
      "loss": 0.0277,
      "step": 2605
    },
    {
      "epoch": 23.27,
      "grad_norm": 0.06230008229613304,
      "learning_rate": 0.00017623529411764702,
      "loss": 0.028,
      "step": 2606
    },
    {
      "epoch": 23.28,
      "grad_norm": 0.08549349755048752,
      "learning_rate": 0.0001761764705882353,
      "loss": 0.0283,
      "step": 2607
    },
    {
      "epoch": 23.29,
      "grad_norm": 0.10650275647640228,
      "learning_rate": 0.0001761176470588235,
      "loss": 0.0343,
      "step": 2608
    },
    {
      "epoch": 23.29,
      "grad_norm": 0.05561896041035652,
      "learning_rate": 0.00017605882352941175,
      "loss": 0.0278,
      "step": 2609
    },
    {
      "epoch": 23.3,
      "grad_norm": 0.05234317108988762,
      "learning_rate": 0.000176,
      "loss": 0.0284,
      "step": 2610
    },
    {
      "epoch": 23.31,
      "grad_norm": 0.05738748237490654,
      "learning_rate": 0.0001759411764705882,
      "loss": 0.0246,
      "step": 2611
    },
    {
      "epoch": 23.32,
      "grad_norm": 0.054971832782030106,
      "learning_rate": 0.00017588235294117646,
      "loss": 0.0228,
      "step": 2612
    },
    {
      "epoch": 23.33,
      "grad_norm": 0.06708146631717682,
      "learning_rate": 0.0001758235294117647,
      "loss": 0.0276,
      "step": 2613
    },
    {
      "epoch": 23.34,
      "grad_norm": 0.09368041157722473,
      "learning_rate": 0.00017576470588235294,
      "loss": 0.0365,
      "step": 2614
    },
    {
      "epoch": 23.35,
      "grad_norm": 0.051021311432123184,
      "learning_rate": 0.00017570588235294116,
      "loss": 0.0237,
      "step": 2615
    },
    {
      "epoch": 23.36,
      "grad_norm": 0.05173582211136818,
      "learning_rate": 0.00017564705882352937,
      "loss": 0.0253,
      "step": 2616
    },
    {
      "epoch": 23.37,
      "grad_norm": 0.05599309131503105,
      "learning_rate": 0.00017558823529411765,
      "loss": 0.0244,
      "step": 2617
    },
    {
      "epoch": 23.38,
      "grad_norm": 0.05602375045418739,
      "learning_rate": 0.00017552941176470586,
      "loss": 0.0268,
      "step": 2618
    },
    {
      "epoch": 23.38,
      "grad_norm": 0.06255139410495758,
      "learning_rate": 0.0001754705882352941,
      "loss": 0.0245,
      "step": 2619
    },
    {
      "epoch": 23.39,
      "grad_norm": 0.12780724465847015,
      "learning_rate": 0.00017541176470588235,
      "loss": 0.0375,
      "step": 2620
    },
    {
      "epoch": 23.4,
      "grad_norm": 0.0565602108836174,
      "learning_rate": 0.00017535294117647056,
      "loss": 0.0251,
      "step": 2621
    },
    {
      "epoch": 23.41,
      "grad_norm": 0.05471047759056091,
      "learning_rate": 0.0001752941176470588,
      "loss": 0.0221,
      "step": 2622
    },
    {
      "epoch": 23.42,
      "grad_norm": 0.05079091712832451,
      "learning_rate": 0.00017523529411764705,
      "loss": 0.0248,
      "step": 2623
    },
    {
      "epoch": 23.43,
      "grad_norm": 0.05325259268283844,
      "learning_rate": 0.00017517647058823527,
      "loss": 0.0267,
      "step": 2624
    },
    {
      "epoch": 23.44,
      "grad_norm": 0.0550168938934803,
      "learning_rate": 0.0001751176470588235,
      "loss": 0.0236,
      "step": 2625
    },
    {
      "epoch": 23.45,
      "grad_norm": 0.09860525280237198,
      "learning_rate": 0.00017505882352941173,
      "loss": 0.0362,
      "step": 2626
    },
    {
      "epoch": 23.46,
      "grad_norm": 0.04859638586640358,
      "learning_rate": 0.000175,
      "loss": 0.0257,
      "step": 2627
    },
    {
      "epoch": 23.46,
      "grad_norm": 0.05800432711839676,
      "learning_rate": 0.00017494117647058821,
      "loss": 0.0281,
      "step": 2628
    },
    {
      "epoch": 23.47,
      "grad_norm": 0.05190037563443184,
      "learning_rate": 0.00017488235294117643,
      "loss": 0.0254,
      "step": 2629
    },
    {
      "epoch": 23.48,
      "grad_norm": 0.05299176275730133,
      "learning_rate": 0.0001748235294117647,
      "loss": 0.0249,
      "step": 2630
    },
    {
      "epoch": 23.49,
      "grad_norm": 0.053858134895563126,
      "learning_rate": 0.00017476470588235292,
      "loss": 0.0246,
      "step": 2631
    },
    {
      "epoch": 23.5,
      "grad_norm": 0.08895876258611679,
      "learning_rate": 0.00017470588235294116,
      "loss": 0.0304,
      "step": 2632
    },
    {
      "epoch": 23.51,
      "grad_norm": 0.08598772436380386,
      "learning_rate": 0.0001746470588235294,
      "loss": 0.0268,
      "step": 2633
    },
    {
      "epoch": 23.52,
      "grad_norm": 0.05937615782022476,
      "learning_rate": 0.00017458823529411762,
      "loss": 0.0239,
      "step": 2634
    },
    {
      "epoch": 23.53,
      "grad_norm": 0.057922814041376114,
      "learning_rate": 0.00017452941176470586,
      "loss": 0.0261,
      "step": 2635
    },
    {
      "epoch": 23.54,
      "grad_norm": 0.056914910674095154,
      "learning_rate": 0.0001744705882352941,
      "loss": 0.0247,
      "step": 2636
    },
    {
      "epoch": 23.54,
      "grad_norm": 0.06148446351289749,
      "learning_rate": 0.00017441176470588235,
      "loss": 0.0233,
      "step": 2637
    },
    {
      "epoch": 23.55,
      "grad_norm": 0.06675130128860474,
      "learning_rate": 0.00017435294117647057,
      "loss": 0.0264,
      "step": 2638
    },
    {
      "epoch": 23.56,
      "grad_norm": 0.11668359488248825,
      "learning_rate": 0.00017429411764705878,
      "loss": 0.0378,
      "step": 2639
    },
    {
      "epoch": 23.57,
      "grad_norm": 0.051918674260377884,
      "learning_rate": 0.00017423529411764705,
      "loss": 0.0249,
      "step": 2640
    },
    {
      "epoch": 23.58,
      "grad_norm": 0.053308840841054916,
      "learning_rate": 0.00017417647058823527,
      "loss": 0.023,
      "step": 2641
    },
    {
      "epoch": 23.59,
      "grad_norm": 0.06300992518663406,
      "learning_rate": 0.00017411764705882354,
      "loss": 0.0296,
      "step": 2642
    },
    {
      "epoch": 23.6,
      "grad_norm": 0.05485334247350693,
      "learning_rate": 0.00017405882352941176,
      "loss": 0.0222,
      "step": 2643
    },
    {
      "epoch": 23.61,
      "grad_norm": 0.061029527336359024,
      "learning_rate": 0.00017399999999999997,
      "loss": 0.0218,
      "step": 2644
    },
    {
      "epoch": 23.62,
      "grad_norm": 0.12146846950054169,
      "learning_rate": 0.00017394117647058822,
      "loss": 0.0323,
      "step": 2645
    },
    {
      "epoch": 23.62,
      "grad_norm": 0.05135010927915573,
      "learning_rate": 0.00017388235294117646,
      "loss": 0.0234,
      "step": 2646
    },
    {
      "epoch": 23.63,
      "grad_norm": 0.0580328069627285,
      "learning_rate": 0.0001738235294117647,
      "loss": 0.0282,
      "step": 2647
    },
    {
      "epoch": 23.64,
      "grad_norm": 0.05992249771952629,
      "learning_rate": 0.00017376470588235292,
      "loss": 0.0273,
      "step": 2648
    },
    {
      "epoch": 23.65,
      "grad_norm": 0.05653415247797966,
      "learning_rate": 0.00017370588235294114,
      "loss": 0.0242,
      "step": 2649
    },
    {
      "epoch": 23.66,
      "grad_norm": 0.05682414025068283,
      "learning_rate": 0.0001736470588235294,
      "loss": 0.0262,
      "step": 2650
    },
    {
      "epoch": 23.67,
      "grad_norm": 0.10287006944417953,
      "learning_rate": 0.00017358823529411762,
      "loss": 0.032,
      "step": 2651
    },
    {
      "epoch": 23.68,
      "grad_norm": 0.060233984142541885,
      "learning_rate": 0.0001735294117647059,
      "loss": 0.0282,
      "step": 2652
    },
    {
      "epoch": 23.69,
      "grad_norm": 0.059896353632211685,
      "learning_rate": 0.0001734705882352941,
      "loss": 0.0272,
      "step": 2653
    },
    {
      "epoch": 23.7,
      "grad_norm": 0.05700116604566574,
      "learning_rate": 0.00017341176470588233,
      "loss": 0.0274,
      "step": 2654
    },
    {
      "epoch": 23.71,
      "grad_norm": 0.05264915153384209,
      "learning_rate": 0.00017335294117647057,
      "loss": 0.0199,
      "step": 2655
    },
    {
      "epoch": 23.71,
      "grad_norm": 0.06769115477800369,
      "learning_rate": 0.00017329411764705881,
      "loss": 0.027,
      "step": 2656
    },
    {
      "epoch": 23.72,
      "grad_norm": 0.07319138199090958,
      "learning_rate": 0.00017323529411764703,
      "loss": 0.0248,
      "step": 2657
    },
    {
      "epoch": 23.73,
      "grad_norm": 0.12389098107814789,
      "learning_rate": 0.00017317647058823527,
      "loss": 0.0416,
      "step": 2658
    },
    {
      "epoch": 23.74,
      "grad_norm": 0.05109129846096039,
      "learning_rate": 0.00017311764705882352,
      "loss": 0.0264,
      "step": 2659
    },
    {
      "epoch": 23.75,
      "grad_norm": 0.05779051408171654,
      "learning_rate": 0.00017305882352941176,
      "loss": 0.0271,
      "step": 2660
    },
    {
      "epoch": 23.76,
      "grad_norm": 0.05919601768255234,
      "learning_rate": 0.00017299999999999998,
      "loss": 0.0274,
      "step": 2661
    },
    {
      "epoch": 23.77,
      "grad_norm": 0.05351078137755394,
      "learning_rate": 0.0001729411764705882,
      "loss": 0.0236,
      "step": 2662
    },
    {
      "epoch": 23.78,
      "grad_norm": 0.06535013765096664,
      "learning_rate": 0.00017288235294117646,
      "loss": 0.0235,
      "step": 2663
    },
    {
      "epoch": 23.79,
      "grad_norm": 0.23863983154296875,
      "learning_rate": 0.00017282352941176468,
      "loss": 0.0435,
      "step": 2664
    },
    {
      "epoch": 23.79,
      "grad_norm": 0.05540243908762932,
      "learning_rate": 0.00017276470588235295,
      "loss": 0.0268,
      "step": 2665
    },
    {
      "epoch": 23.8,
      "grad_norm": 0.08166693896055222,
      "learning_rate": 0.00017270588235294117,
      "loss": 0.0284,
      "step": 2666
    },
    {
      "epoch": 23.81,
      "grad_norm": 0.09516628831624985,
      "learning_rate": 0.00017264705882352938,
      "loss": 0.0332,
      "step": 2667
    },
    {
      "epoch": 23.82,
      "grad_norm": 0.061793334782123566,
      "learning_rate": 0.00017258823529411763,
      "loss": 0.0272,
      "step": 2668
    },
    {
      "epoch": 23.83,
      "grad_norm": 0.06748347729444504,
      "learning_rate": 0.00017252941176470587,
      "loss": 0.0236,
      "step": 2669
    },
    {
      "epoch": 23.84,
      "grad_norm": 0.10607940703630447,
      "learning_rate": 0.0001724705882352941,
      "loss": 0.0381,
      "step": 2670
    },
    {
      "epoch": 23.85,
      "grad_norm": 0.06220061331987381,
      "learning_rate": 0.00017241176470588233,
      "loss": 0.0288,
      "step": 2671
    },
    {
      "epoch": 23.86,
      "grad_norm": 0.05508628487586975,
      "learning_rate": 0.00017235294117647057,
      "loss": 0.0244,
      "step": 2672
    },
    {
      "epoch": 23.87,
      "grad_norm": 0.05792519077658653,
      "learning_rate": 0.00017229411764705882,
      "loss": 0.0268,
      "step": 2673
    },
    {
      "epoch": 23.88,
      "grad_norm": 0.06214618682861328,
      "learning_rate": 0.00017223529411764703,
      "loss": 0.0289,
      "step": 2674
    },
    {
      "epoch": 23.88,
      "grad_norm": 0.06868556886911392,
      "learning_rate": 0.0001721764705882353,
      "loss": 0.026,
      "step": 2675
    },
    {
      "epoch": 23.89,
      "grad_norm": 0.1072574034333229,
      "learning_rate": 0.00017211764705882352,
      "loss": 0.0369,
      "step": 2676
    },
    {
      "epoch": 23.9,
      "grad_norm": 0.05920523777604103,
      "learning_rate": 0.00017205882352941174,
      "loss": 0.0287,
      "step": 2677
    },
    {
      "epoch": 23.91,
      "grad_norm": 0.057916879653930664,
      "learning_rate": 0.000172,
      "loss": 0.0277,
      "step": 2678
    },
    {
      "epoch": 23.92,
      "grad_norm": 0.06307437270879745,
      "learning_rate": 0.00017194117647058822,
      "loss": 0.0253,
      "step": 2679
    },
    {
      "epoch": 23.93,
      "grad_norm": 0.06734710931777954,
      "learning_rate": 0.00017188235294117647,
      "loss": 0.0253,
      "step": 2680
    },
    {
      "epoch": 23.94,
      "grad_norm": 0.06740694493055344,
      "learning_rate": 0.00017182352941176468,
      "loss": 0.0249,
      "step": 2681
    },
    {
      "epoch": 23.95,
      "grad_norm": 0.09946932643651962,
      "learning_rate": 0.00017176470588235293,
      "loss": 0.0312,
      "step": 2682
    },
    {
      "epoch": 23.96,
      "grad_norm": 0.10713358968496323,
      "learning_rate": 0.00017170588235294117,
      "loss": 0.0309,
      "step": 2683
    },
    {
      "epoch": 23.96,
      "grad_norm": 0.057803675532341,
      "learning_rate": 0.00017164705882352939,
      "loss": 0.0273,
      "step": 2684
    },
    {
      "epoch": 23.97,
      "grad_norm": 0.06359840929508209,
      "learning_rate": 0.00017158823529411766,
      "loss": 0.0229,
      "step": 2685
    },
    {
      "epoch": 23.98,
      "grad_norm": 0.064893439412117,
      "learning_rate": 0.00017152941176470587,
      "loss": 0.0269,
      "step": 2686
    },
    {
      "epoch": 23.99,
      "grad_norm": 0.06777261942625046,
      "learning_rate": 0.0001714705882352941,
      "loss": 0.0276,
      "step": 2687
    },
    {
      "epoch": 24.0,
      "grad_norm": 0.06847578287124634,
      "learning_rate": 0.00017141176470588236,
      "loss": 0.0293,
      "step": 2688
    },
    {
      "epoch": 24.01,
      "grad_norm": 0.05095682293176651,
      "learning_rate": 0.00017135294117647058,
      "loss": 0.0235,
      "step": 2689
    },
    {
      "epoch": 24.02,
      "grad_norm": 0.047058165073394775,
      "learning_rate": 0.0001712941176470588,
      "loss": 0.0228,
      "step": 2690
    },
    {
      "epoch": 24.03,
      "grad_norm": 0.049135103821754456,
      "learning_rate": 0.00017123529411764704,
      "loss": 0.0221,
      "step": 2691
    },
    {
      "epoch": 24.04,
      "grad_norm": 0.053765006363391876,
      "learning_rate": 0.00017117647058823528,
      "loss": 0.0236,
      "step": 2692
    },
    {
      "epoch": 24.04,
      "grad_norm": 0.05885041877627373,
      "learning_rate": 0.00017111764705882352,
      "loss": 0.0269,
      "step": 2693
    },
    {
      "epoch": 24.05,
      "grad_norm": 0.09634266793727875,
      "learning_rate": 0.00017105882352941174,
      "loss": 0.0309,
      "step": 2694
    },
    {
      "epoch": 24.06,
      "grad_norm": 0.1012498065829277,
      "learning_rate": 0.00017099999999999998,
      "loss": 0.0309,
      "step": 2695
    },
    {
      "epoch": 24.07,
      "grad_norm": 0.046449918299913406,
      "learning_rate": 0.00017094117647058823,
      "loss": 0.0222,
      "step": 2696
    },
    {
      "epoch": 24.08,
      "grad_norm": 0.0512268990278244,
      "learning_rate": 0.00017088235294117644,
      "loss": 0.026,
      "step": 2697
    },
    {
      "epoch": 24.09,
      "grad_norm": 0.05378634110093117,
      "learning_rate": 0.0001708235294117647,
      "loss": 0.0234,
      "step": 2698
    },
    {
      "epoch": 24.1,
      "grad_norm": 0.053176265209913254,
      "learning_rate": 0.00017076470588235293,
      "loss": 0.0225,
      "step": 2699
    },
    {
      "epoch": 24.11,
      "grad_norm": 0.06803590804338455,
      "learning_rate": 0.00017070588235294115,
      "loss": 0.0259,
      "step": 2700
    },
    {
      "epoch": 24.11,
      "eval_cer": 0.033002221194935226,
      "eval_loss": 0.19758771359920502,
      "eval_runtime": 22.4242,
      "eval_samples_per_second": 117.819,
      "eval_steps_per_second": 1.873,
      "eval_wer": 0.1102538675128917,
      "step": 2700
    },
    {
      "epoch": 24.12,
      "grad_norm": 0.08501246571540833,
      "learning_rate": 0.00017064705882352942,
      "loss": 0.0257,
      "step": 2701
    },
    {
      "epoch": 24.12,
      "grad_norm": 0.04914569482207298,
      "learning_rate": 0.00017058823529411763,
      "loss": 0.0237,
      "step": 2702
    },
    {
      "epoch": 24.13,
      "grad_norm": 0.05821319669485092,
      "learning_rate": 0.00017052941176470588,
      "loss": 0.0281,
      "step": 2703
    },
    {
      "epoch": 24.14,
      "grad_norm": 0.05908571183681488,
      "learning_rate": 0.0001704705882352941,
      "loss": 0.0267,
      "step": 2704
    },
    {
      "epoch": 24.15,
      "grad_norm": 0.06436707079410553,
      "learning_rate": 0.00017041176470588234,
      "loss": 0.0244,
      "step": 2705
    },
    {
      "epoch": 24.16,
      "grad_norm": 0.07469526678323746,
      "learning_rate": 0.00017035294117647058,
      "loss": 0.0264,
      "step": 2706
    },
    {
      "epoch": 24.17,
      "grad_norm": 0.12090101093053818,
      "learning_rate": 0.0001702941176470588,
      "loss": 0.0335,
      "step": 2707
    },
    {
      "epoch": 24.18,
      "grad_norm": 0.051356274634599686,
      "learning_rate": 0.00017023529411764707,
      "loss": 0.0258,
      "step": 2708
    },
    {
      "epoch": 24.19,
      "grad_norm": 0.05642694979906082,
      "learning_rate": 0.00017017647058823528,
      "loss": 0.0249,
      "step": 2709
    },
    {
      "epoch": 24.2,
      "grad_norm": 0.051083944737911224,
      "learning_rate": 0.0001701176470588235,
      "loss": 0.0246,
      "step": 2710
    },
    {
      "epoch": 24.21,
      "grad_norm": 0.057927340269088745,
      "learning_rate": 0.00017005882352941177,
      "loss": 0.0266,
      "step": 2711
    },
    {
      "epoch": 24.21,
      "grad_norm": 0.059637777507305145,
      "learning_rate": 0.00016999999999999999,
      "loss": 0.0256,
      "step": 2712
    },
    {
      "epoch": 24.22,
      "grad_norm": 0.09623510390520096,
      "learning_rate": 0.00016994117647058823,
      "loss": 0.0354,
      "step": 2713
    },
    {
      "epoch": 24.23,
      "grad_norm": 0.05712822824716568,
      "learning_rate": 0.00016988235294117644,
      "loss": 0.0275,
      "step": 2714
    },
    {
      "epoch": 24.24,
      "grad_norm": 0.055225055664777756,
      "learning_rate": 0.0001698235294117647,
      "loss": 0.0231,
      "step": 2715
    },
    {
      "epoch": 24.25,
      "grad_norm": 0.06384652107954025,
      "learning_rate": 0.00016976470588235293,
      "loss": 0.0246,
      "step": 2716
    },
    {
      "epoch": 24.26,
      "grad_norm": 0.060066428035497665,
      "learning_rate": 0.00016970588235294115,
      "loss": 0.0257,
      "step": 2717
    },
    {
      "epoch": 24.27,
      "grad_norm": 0.05861593782901764,
      "learning_rate": 0.0001696470588235294,
      "loss": 0.0242,
      "step": 2718
    },
    {
      "epoch": 24.28,
      "grad_norm": 0.10007983446121216,
      "learning_rate": 0.00016958823529411763,
      "loss": 0.0283,
      "step": 2719
    },
    {
      "epoch": 24.29,
      "grad_norm": 0.07153783738613129,
      "learning_rate": 0.00016952941176470585,
      "loss": 0.0279,
      "step": 2720
    },
    {
      "epoch": 24.29,
      "grad_norm": 0.061844177544116974,
      "learning_rate": 0.00016947058823529412,
      "loss": 0.0253,
      "step": 2721
    },
    {
      "epoch": 24.3,
      "grad_norm": 0.05117771402001381,
      "learning_rate": 0.00016941176470588234,
      "loss": 0.0232,
      "step": 2722
    },
    {
      "epoch": 24.31,
      "grad_norm": 0.05482063069939613,
      "learning_rate": 0.00016935294117647055,
      "loss": 0.026,
      "step": 2723
    },
    {
      "epoch": 24.32,
      "grad_norm": 0.06420817226171494,
      "learning_rate": 0.00016929411764705882,
      "loss": 0.0297,
      "step": 2724
    },
    {
      "epoch": 24.33,
      "grad_norm": 0.06492055952548981,
      "learning_rate": 0.00016923529411764704,
      "loss": 0.0222,
      "step": 2725
    },
    {
      "epoch": 24.34,
      "grad_norm": 0.10581710934638977,
      "learning_rate": 0.00016917647058823528,
      "loss": 0.0418,
      "step": 2726
    },
    {
      "epoch": 24.35,
      "grad_norm": 0.05373619124293327,
      "learning_rate": 0.0001691176470588235,
      "loss": 0.0264,
      "step": 2727
    },
    {
      "epoch": 24.36,
      "grad_norm": 0.057972416281700134,
      "learning_rate": 0.00016905882352941174,
      "loss": 0.0324,
      "step": 2728
    },
    {
      "epoch": 24.37,
      "grad_norm": 0.052475400269031525,
      "learning_rate": 0.000169,
      "loss": 0.0233,
      "step": 2729
    },
    {
      "epoch": 24.38,
      "grad_norm": 0.057983409613370895,
      "learning_rate": 0.0001689411764705882,
      "loss": 0.0274,
      "step": 2730
    },
    {
      "epoch": 24.38,
      "grad_norm": 0.06758495420217514,
      "learning_rate": 0.00016888235294117647,
      "loss": 0.0233,
      "step": 2731
    },
    {
      "epoch": 24.39,
      "grad_norm": 0.09746670722961426,
      "learning_rate": 0.0001688235294117647,
      "loss": 0.0322,
      "step": 2732
    },
    {
      "epoch": 24.4,
      "grad_norm": 0.04812254011631012,
      "learning_rate": 0.0001687647058823529,
      "loss": 0.0223,
      "step": 2733
    },
    {
      "epoch": 24.41,
      "grad_norm": 0.06194176897406578,
      "learning_rate": 0.00016870588235294118,
      "loss": 0.0234,
      "step": 2734
    },
    {
      "epoch": 24.42,
      "grad_norm": 0.05478568375110626,
      "learning_rate": 0.0001686470588235294,
      "loss": 0.0213,
      "step": 2735
    },
    {
      "epoch": 24.43,
      "grad_norm": 0.06427139043807983,
      "learning_rate": 0.00016858823529411764,
      "loss": 0.0263,
      "step": 2736
    },
    {
      "epoch": 24.44,
      "grad_norm": 0.06361524015665054,
      "learning_rate": 0.00016852941176470588,
      "loss": 0.0219,
      "step": 2737
    },
    {
      "epoch": 24.45,
      "grad_norm": 0.17025306820869446,
      "learning_rate": 0.0001684705882352941,
      "loss": 0.0242,
      "step": 2738
    },
    {
      "epoch": 24.46,
      "grad_norm": 0.05920528247952461,
      "learning_rate": 0.00016841176470588234,
      "loss": 0.0252,
      "step": 2739
    },
    {
      "epoch": 24.46,
      "grad_norm": 0.06601275503635406,
      "learning_rate": 0.00016835294117647056,
      "loss": 0.0239,
      "step": 2740
    },
    {
      "epoch": 24.47,
      "grad_norm": 0.0645885318517685,
      "learning_rate": 0.00016829411764705883,
      "loss": 0.0269,
      "step": 2741
    },
    {
      "epoch": 24.48,
      "grad_norm": 0.07101695984601974,
      "learning_rate": 0.00016823529411764704,
      "loss": 0.0239,
      "step": 2742
    },
    {
      "epoch": 24.49,
      "grad_norm": 0.05991099402308464,
      "learning_rate": 0.00016817647058823526,
      "loss": 0.0227,
      "step": 2743
    },
    {
      "epoch": 24.5,
      "grad_norm": 0.08743740618228912,
      "learning_rate": 0.00016811764705882353,
      "loss": 0.0278,
      "step": 2744
    },
    {
      "epoch": 24.51,
      "grad_norm": 0.12182283401489258,
      "learning_rate": 0.00016805882352941175,
      "loss": 0.0379,
      "step": 2745
    },
    {
      "epoch": 24.52,
      "grad_norm": 0.05626633018255234,
      "learning_rate": 0.000168,
      "loss": 0.029,
      "step": 2746
    },
    {
      "epoch": 24.53,
      "grad_norm": 0.054795052856206894,
      "learning_rate": 0.00016794117647058823,
      "loss": 0.0267,
      "step": 2747
    },
    {
      "epoch": 24.54,
      "grad_norm": 0.05787048116326332,
      "learning_rate": 0.00016788235294117645,
      "loss": 0.022,
      "step": 2748
    },
    {
      "epoch": 24.54,
      "grad_norm": 0.05009082704782486,
      "learning_rate": 0.0001678235294117647,
      "loss": 0.0199,
      "step": 2749
    },
    {
      "epoch": 24.55,
      "grad_norm": 0.06413988769054413,
      "learning_rate": 0.0001677647058823529,
      "loss": 0.0235,
      "step": 2750
    },
    {
      "epoch": 24.56,
      "grad_norm": 0.09867819398641586,
      "learning_rate": 0.00016770588235294115,
      "loss": 0.0312,
      "step": 2751
    },
    {
      "epoch": 24.57,
      "grad_norm": 0.057203102856874466,
      "learning_rate": 0.0001676470588235294,
      "loss": 0.0283,
      "step": 2752
    },
    {
      "epoch": 24.58,
      "grad_norm": 0.05727151408791542,
      "learning_rate": 0.0001675882352941176,
      "loss": 0.0255,
      "step": 2753
    },
    {
      "epoch": 24.59,
      "grad_norm": 0.05368828773498535,
      "learning_rate": 0.00016752941176470588,
      "loss": 0.0246,
      "step": 2754
    },
    {
      "epoch": 24.6,
      "grad_norm": 0.0596047081053257,
      "learning_rate": 0.0001674705882352941,
      "loss": 0.0254,
      "step": 2755
    },
    {
      "epoch": 24.61,
      "grad_norm": 0.07500813901424408,
      "learning_rate": 0.00016741176470588232,
      "loss": 0.0296,
      "step": 2756
    },
    {
      "epoch": 24.62,
      "grad_norm": 0.10808512568473816,
      "learning_rate": 0.0001673529411764706,
      "loss": 0.0389,
      "step": 2757
    },
    {
      "epoch": 24.62,
      "grad_norm": 0.057034000754356384,
      "learning_rate": 0.0001672941176470588,
      "loss": 0.0234,
      "step": 2758
    },
    {
      "epoch": 24.63,
      "grad_norm": 0.0573807917535305,
      "learning_rate": 0.00016723529411764705,
      "loss": 0.0279,
      "step": 2759
    },
    {
      "epoch": 24.64,
      "grad_norm": 0.05286450684070587,
      "learning_rate": 0.0001671764705882353,
      "loss": 0.0233,
      "step": 2760
    },
    {
      "epoch": 24.65,
      "grad_norm": 0.05330032855272293,
      "learning_rate": 0.0001671176470588235,
      "loss": 0.024,
      "step": 2761
    },
    {
      "epoch": 24.66,
      "grad_norm": 0.06454110890626907,
      "learning_rate": 0.00016705882352941175,
      "loss": 0.0288,
      "step": 2762
    },
    {
      "epoch": 24.67,
      "grad_norm": 0.08926012367010117,
      "learning_rate": 0.00016699999999999997,
      "loss": 0.0324,
      "step": 2763
    },
    {
      "epoch": 24.68,
      "grad_norm": 0.04690074920654297,
      "learning_rate": 0.00016694117647058824,
      "loss": 0.0212,
      "step": 2764
    },
    {
      "epoch": 24.69,
      "grad_norm": 0.051096923649311066,
      "learning_rate": 0.00016688235294117645,
      "loss": 0.0253,
      "step": 2765
    },
    {
      "epoch": 24.7,
      "grad_norm": 0.057431526482105255,
      "learning_rate": 0.00016682352941176467,
      "loss": 0.026,
      "step": 2766
    },
    {
      "epoch": 24.71,
      "grad_norm": 0.057798273861408234,
      "learning_rate": 0.00016676470588235294,
      "loss": 0.0226,
      "step": 2767
    },
    {
      "epoch": 24.71,
      "grad_norm": 0.05295524746179581,
      "learning_rate": 0.00016670588235294116,
      "loss": 0.0195,
      "step": 2768
    },
    {
      "epoch": 24.72,
      "grad_norm": 0.08081305027008057,
      "learning_rate": 0.0001666470588235294,
      "loss": 0.0277,
      "step": 2769
    },
    {
      "epoch": 24.73,
      "grad_norm": 0.10082969069480896,
      "learning_rate": 0.00016658823529411764,
      "loss": 0.0301,
      "step": 2770
    },
    {
      "epoch": 24.74,
      "grad_norm": 0.05353362858295441,
      "learning_rate": 0.00016652941176470586,
      "loss": 0.0237,
      "step": 2771
    },
    {
      "epoch": 24.75,
      "grad_norm": 0.04614794999361038,
      "learning_rate": 0.0001664705882352941,
      "loss": 0.0209,
      "step": 2772
    },
    {
      "epoch": 24.76,
      "grad_norm": 0.05549345910549164,
      "learning_rate": 0.00016641176470588235,
      "loss": 0.0271,
      "step": 2773
    },
    {
      "epoch": 24.77,
      "grad_norm": 0.05876650661230087,
      "learning_rate": 0.0001663529411764706,
      "loss": 0.026,
      "step": 2774
    },
    {
      "epoch": 24.78,
      "grad_norm": 0.06546223908662796,
      "learning_rate": 0.0001662941176470588,
      "loss": 0.0188,
      "step": 2775
    },
    {
      "epoch": 24.79,
      "grad_norm": 0.13608670234680176,
      "learning_rate": 0.00016623529411764702,
      "loss": 0.0448,
      "step": 2776
    },
    {
      "epoch": 24.79,
      "grad_norm": 0.04852099344134331,
      "learning_rate": 0.0001661764705882353,
      "loss": 0.0251,
      "step": 2777
    },
    {
      "epoch": 24.8,
      "grad_norm": 0.05532992631196976,
      "learning_rate": 0.0001661176470588235,
      "loss": 0.0266,
      "step": 2778
    },
    {
      "epoch": 24.81,
      "grad_norm": 0.05641378462314606,
      "learning_rate": 0.00016605882352941175,
      "loss": 0.0217,
      "step": 2779
    },
    {
      "epoch": 24.82,
      "grad_norm": 0.06152420490980148,
      "learning_rate": 0.000166,
      "loss": 0.0263,
      "step": 2780
    },
    {
      "epoch": 24.83,
      "grad_norm": 0.05759526416659355,
      "learning_rate": 0.0001659411764705882,
      "loss": 0.0251,
      "step": 2781
    },
    {
      "epoch": 24.84,
      "grad_norm": 0.09831923246383667,
      "learning_rate": 0.00016588235294117646,
      "loss": 0.0348,
      "step": 2782
    },
    {
      "epoch": 24.85,
      "grad_norm": 0.05034704878926277,
      "learning_rate": 0.0001658235294117647,
      "loss": 0.0229,
      "step": 2783
    },
    {
      "epoch": 24.86,
      "grad_norm": 0.05134845897555351,
      "learning_rate": 0.00016576470588235292,
      "loss": 0.0245,
      "step": 2784
    },
    {
      "epoch": 24.87,
      "grad_norm": 0.05441165342926979,
      "learning_rate": 0.00016570588235294116,
      "loss": 0.0233,
      "step": 2785
    },
    {
      "epoch": 24.88,
      "grad_norm": 0.05475650727748871,
      "learning_rate": 0.00016564705882352938,
      "loss": 0.0255,
      "step": 2786
    },
    {
      "epoch": 24.88,
      "grad_norm": 0.06395895779132843,
      "learning_rate": 0.00016558823529411765,
      "loss": 0.0269,
      "step": 2787
    },
    {
      "epoch": 24.89,
      "grad_norm": 0.11297452449798584,
      "learning_rate": 0.00016552941176470586,
      "loss": 0.0419,
      "step": 2788
    },
    {
      "epoch": 24.9,
      "grad_norm": 0.05215073376893997,
      "learning_rate": 0.00016547058823529408,
      "loss": 0.0244,
      "step": 2789
    },
    {
      "epoch": 24.91,
      "grad_norm": 0.05535958334803581,
      "learning_rate": 0.00016541176470588235,
      "loss": 0.0225,
      "step": 2790
    },
    {
      "epoch": 24.92,
      "grad_norm": 0.05166322737932205,
      "learning_rate": 0.00016535294117647057,
      "loss": 0.0244,
      "step": 2791
    },
    {
      "epoch": 24.93,
      "grad_norm": 0.060888104140758514,
      "learning_rate": 0.0001652941176470588,
      "loss": 0.0324,
      "step": 2792
    },
    {
      "epoch": 24.94,
      "grad_norm": 0.058065325021743774,
      "learning_rate": 0.00016523529411764705,
      "loss": 0.0266,
      "step": 2793
    },
    {
      "epoch": 24.95,
      "grad_norm": 0.07291809469461441,
      "learning_rate": 0.00016517647058823527,
      "loss": 0.0256,
      "step": 2794
    },
    {
      "epoch": 24.96,
      "grad_norm": 0.13164353370666504,
      "learning_rate": 0.0001651176470588235,
      "loss": 0.0385,
      "step": 2795
    },
    {
      "epoch": 24.96,
      "grad_norm": 0.04778966307640076,
      "learning_rate": 0.00016505882352941176,
      "loss": 0.0257,
      "step": 2796
    },
    {
      "epoch": 24.97,
      "grad_norm": 0.05566604807972908,
      "learning_rate": 0.000165,
      "loss": 0.0243,
      "step": 2797
    },
    {
      "epoch": 24.98,
      "grad_norm": 0.05550533905625343,
      "learning_rate": 0.00016494117647058822,
      "loss": 0.0285,
      "step": 2798
    },
    {
      "epoch": 24.99,
      "grad_norm": 0.06879201531410217,
      "learning_rate": 0.00016488235294117643,
      "loss": 0.0257,
      "step": 2799
    },
    {
      "epoch": 25.0,
      "grad_norm": 0.09967371076345444,
      "learning_rate": 0.0001648235294117647,
      "loss": 0.0325,
      "step": 2800
    },
    {
      "epoch": 25.0,
      "eval_cer": 0.03283309467702473,
      "eval_loss": 0.19575656950473785,
      "eval_runtime": 22.3096,
      "eval_samples_per_second": 118.425,
      "eval_steps_per_second": 1.883,
      "eval_wer": 0.10876636255454185,
      "step": 2800
    },
    {
      "epoch": 25.01,
      "grad_norm": 0.05569753795862198,
      "learning_rate": 0.00016476470588235292,
      "loss": 0.025,
      "step": 2801
    },
    {
      "epoch": 25.02,
      "grad_norm": 0.058367833495140076,
      "learning_rate": 0.0001647058823529412,
      "loss": 0.0292,
      "step": 2802
    },
    {
      "epoch": 25.03,
      "grad_norm": 0.0486668162047863,
      "learning_rate": 0.0001646470588235294,
      "loss": 0.0241,
      "step": 2803
    },
    {
      "epoch": 25.04,
      "grad_norm": 0.05004143714904785,
      "learning_rate": 0.00016458823529411762,
      "loss": 0.0217,
      "step": 2804
    },
    {
      "epoch": 25.04,
      "grad_norm": 0.056677915155887604,
      "learning_rate": 0.00016452941176470587,
      "loss": 0.0256,
      "step": 2805
    },
    {
      "epoch": 25.05,
      "grad_norm": 0.06901763379573822,
      "learning_rate": 0.0001644705882352941,
      "loss": 0.0215,
      "step": 2806
    },
    {
      "epoch": 25.06,
      "grad_norm": 0.09663349390029907,
      "learning_rate": 0.00016441176470588235,
      "loss": 0.0296,
      "step": 2807
    },
    {
      "epoch": 25.07,
      "grad_norm": 0.05198526382446289,
      "learning_rate": 0.00016435294117647057,
      "loss": 0.024,
      "step": 2808
    },
    {
      "epoch": 25.08,
      "grad_norm": 0.05520065873861313,
      "learning_rate": 0.0001642941176470588,
      "loss": 0.0251,
      "step": 2809
    },
    {
      "epoch": 25.09,
      "grad_norm": 0.04921083524823189,
      "learning_rate": 0.00016423529411764706,
      "loss": 0.0196,
      "step": 2810
    },
    {
      "epoch": 25.1,
      "grad_norm": 0.05463823676109314,
      "learning_rate": 0.00016417647058823527,
      "loss": 0.0222,
      "step": 2811
    },
    {
      "epoch": 25.11,
      "grad_norm": 0.07825464755296707,
      "learning_rate": 0.00016411764705882354,
      "loss": 0.0275,
      "step": 2812
    },
    {
      "epoch": 25.12,
      "grad_norm": 0.08404690027236938,
      "learning_rate": 0.00016405882352941176,
      "loss": 0.0316,
      "step": 2813
    },
    {
      "epoch": 25.12,
      "grad_norm": 0.05047415941953659,
      "learning_rate": 0.00016399999999999997,
      "loss": 0.0223,
      "step": 2814
    },
    {
      "epoch": 25.13,
      "grad_norm": 0.05363975092768669,
      "learning_rate": 0.00016394117647058822,
      "loss": 0.0271,
      "step": 2815
    },
    {
      "epoch": 25.14,
      "grad_norm": 0.0585324689745903,
      "learning_rate": 0.00016388235294117646,
      "loss": 0.0265,
      "step": 2816
    },
    {
      "epoch": 25.15,
      "grad_norm": 0.05032774433493614,
      "learning_rate": 0.00016382352941176468,
      "loss": 0.0206,
      "step": 2817
    },
    {
      "epoch": 25.16,
      "grad_norm": 0.06968959420919418,
      "learning_rate": 0.00016376470588235292,
      "loss": 0.023,
      "step": 2818
    },
    {
      "epoch": 25.17,
      "grad_norm": 0.08379457145929337,
      "learning_rate": 0.00016370588235294116,
      "loss": 0.0266,
      "step": 2819
    },
    {
      "epoch": 25.18,
      "grad_norm": 0.05781732127070427,
      "learning_rate": 0.0001636470588235294,
      "loss": 0.0286,
      "step": 2820
    },
    {
      "epoch": 25.19,
      "grad_norm": 0.04706644266843796,
      "learning_rate": 0.00016358823529411762,
      "loss": 0.0217,
      "step": 2821
    },
    {
      "epoch": 25.2,
      "grad_norm": 0.0497005395591259,
      "learning_rate": 0.00016352941176470584,
      "loss": 0.0233,
      "step": 2822
    },
    {
      "epoch": 25.21,
      "grad_norm": 0.0531892329454422,
      "learning_rate": 0.0001634705882352941,
      "loss": 0.0236,
      "step": 2823
    },
    {
      "epoch": 25.21,
      "grad_norm": 0.05514576658606529,
      "learning_rate": 0.00016341176470588233,
      "loss": 0.0217,
      "step": 2824
    },
    {
      "epoch": 25.22,
      "grad_norm": 0.10693211853504181,
      "learning_rate": 0.0001633529411764706,
      "loss": 0.0331,
      "step": 2825
    },
    {
      "epoch": 25.23,
      "grad_norm": 0.049358803778886795,
      "learning_rate": 0.00016329411764705881,
      "loss": 0.024,
      "step": 2826
    },
    {
      "epoch": 25.24,
      "grad_norm": 0.050118591636419296,
      "learning_rate": 0.00016323529411764703,
      "loss": 0.0248,
      "step": 2827
    },
    {
      "epoch": 25.25,
      "grad_norm": 0.055307239294052124,
      "learning_rate": 0.00016317647058823527,
      "loss": 0.028,
      "step": 2828
    },
    {
      "epoch": 25.26,
      "grad_norm": 0.05096099153161049,
      "learning_rate": 0.00016311764705882352,
      "loss": 0.0224,
      "step": 2829
    },
    {
      "epoch": 25.27,
      "grad_norm": 0.05580078065395355,
      "learning_rate": 0.00016305882352941176,
      "loss": 0.026,
      "step": 2830
    },
    {
      "epoch": 25.28,
      "grad_norm": 0.0734834298491478,
      "learning_rate": 0.00016299999999999998,
      "loss": 0.0258,
      "step": 2831
    },
    {
      "epoch": 25.29,
      "grad_norm": 0.13371030986309052,
      "learning_rate": 0.00016294117647058822,
      "loss": 0.0291,
      "step": 2832
    },
    {
      "epoch": 25.29,
      "grad_norm": 0.04842384159564972,
      "learning_rate": 0.00016288235294117646,
      "loss": 0.0228,
      "step": 2833
    },
    {
      "epoch": 25.3,
      "grad_norm": 0.0555318146944046,
      "learning_rate": 0.00016282352941176468,
      "loss": 0.0236,
      "step": 2834
    },
    {
      "epoch": 25.31,
      "grad_norm": 0.056293439120054245,
      "learning_rate": 0.00016276470588235295,
      "loss": 0.0225,
      "step": 2835
    },
    {
      "epoch": 25.32,
      "grad_norm": 0.05625167116522789,
      "learning_rate": 0.00016270588235294117,
      "loss": 0.0266,
      "step": 2836
    },
    {
      "epoch": 25.33,
      "grad_norm": 0.08097604662179947,
      "learning_rate": 0.00016264705882352938,
      "loss": 0.0291,
      "step": 2837
    },
    {
      "epoch": 25.34,
      "grad_norm": 0.0822027325630188,
      "learning_rate": 0.00016258823529411765,
      "loss": 0.0229,
      "step": 2838
    },
    {
      "epoch": 25.35,
      "grad_norm": 0.05596937611699104,
      "learning_rate": 0.00016252941176470587,
      "loss": 0.0277,
      "step": 2839
    },
    {
      "epoch": 25.36,
      "grad_norm": 0.05915316939353943,
      "learning_rate": 0.00016247058823529411,
      "loss": 0.0259,
      "step": 2840
    },
    {
      "epoch": 25.37,
      "grad_norm": 0.058485615998506546,
      "learning_rate": 0.00016241176470588233,
      "loss": 0.0241,
      "step": 2841
    },
    {
      "epoch": 25.38,
      "grad_norm": 0.05064799636602402,
      "learning_rate": 0.00016235294117647057,
      "loss": 0.0195,
      "step": 2842
    },
    {
      "epoch": 25.38,
      "grad_norm": 0.06672133505344391,
      "learning_rate": 0.00016229411764705882,
      "loss": 0.0208,
      "step": 2843
    },
    {
      "epoch": 25.39,
      "grad_norm": 0.09813190251588821,
      "learning_rate": 0.00016223529411764703,
      "loss": 0.033,
      "step": 2844
    },
    {
      "epoch": 25.4,
      "grad_norm": 0.05445200577378273,
      "learning_rate": 0.00016217647058823525,
      "loss": 0.0254,
      "step": 2845
    },
    {
      "epoch": 25.41,
      "grad_norm": 0.058646202087402344,
      "learning_rate": 0.00016211764705882352,
      "loss": 0.022,
      "step": 2846
    },
    {
      "epoch": 25.42,
      "grad_norm": 0.053171828389167786,
      "learning_rate": 0.00016205882352941174,
      "loss": 0.0232,
      "step": 2847
    },
    {
      "epoch": 25.43,
      "grad_norm": 0.052805688232183456,
      "learning_rate": 0.000162,
      "loss": 0.0237,
      "step": 2848
    },
    {
      "epoch": 25.44,
      "grad_norm": 0.054522205144166946,
      "learning_rate": 0.00016194117647058822,
      "loss": 0.0237,
      "step": 2849
    },
    {
      "epoch": 25.45,
      "grad_norm": 0.09734851866960526,
      "learning_rate": 0.00016188235294117644,
      "loss": 0.0265,
      "step": 2850
    },
    {
      "epoch": 25.46,
      "grad_norm": 0.05089952424168587,
      "learning_rate": 0.00016182352941176468,
      "loss": 0.023,
      "step": 2851
    },
    {
      "epoch": 25.46,
      "grad_norm": 0.05102134123444557,
      "learning_rate": 0.00016176470588235293,
      "loss": 0.0241,
      "step": 2852
    },
    {
      "epoch": 25.47,
      "grad_norm": 0.05429181456565857,
      "learning_rate": 0.00016170588235294117,
      "loss": 0.0271,
      "step": 2853
    },
    {
      "epoch": 25.48,
      "grad_norm": 0.052763838320970535,
      "learning_rate": 0.0001616470588235294,
      "loss": 0.0244,
      "step": 2854
    },
    {
      "epoch": 25.49,
      "grad_norm": 0.049629341810941696,
      "learning_rate": 0.00016158823529411763,
      "loss": 0.0203,
      "step": 2855
    },
    {
      "epoch": 25.5,
      "grad_norm": 0.07441843301057816,
      "learning_rate": 0.00016152941176470587,
      "loss": 0.0246,
      "step": 2856
    },
    {
      "epoch": 25.51,
      "grad_norm": 0.0944853201508522,
      "learning_rate": 0.0001614705882352941,
      "loss": 0.0338,
      "step": 2857
    },
    {
      "epoch": 25.52,
      "grad_norm": 0.05425774306058884,
      "learning_rate": 0.00016141176470588236,
      "loss": 0.0256,
      "step": 2858
    },
    {
      "epoch": 25.53,
      "grad_norm": 0.0528445765376091,
      "learning_rate": 0.00016135294117647058,
      "loss": 0.0254,
      "step": 2859
    },
    {
      "epoch": 25.54,
      "grad_norm": 0.061194173991680145,
      "learning_rate": 0.0001612941176470588,
      "loss": 0.0232,
      "step": 2860
    },
    {
      "epoch": 25.54,
      "grad_norm": 0.058666814118623734,
      "learning_rate": 0.00016123529411764706,
      "loss": 0.0254,
      "step": 2861
    },
    {
      "epoch": 25.55,
      "grad_norm": 0.06553830951452255,
      "learning_rate": 0.00016117647058823528,
      "loss": 0.0187,
      "step": 2862
    },
    {
      "epoch": 25.56,
      "grad_norm": 0.10020612180233002,
      "learning_rate": 0.00016111764705882352,
      "loss": 0.026,
      "step": 2863
    },
    {
      "epoch": 25.57,
      "grad_norm": 0.07953240722417831,
      "learning_rate": 0.00016105882352941174,
      "loss": 0.0249,
      "step": 2864
    },
    {
      "epoch": 25.58,
      "grad_norm": 0.058467797935009,
      "learning_rate": 0.00016099999999999998,
      "loss": 0.0281,
      "step": 2865
    },
    {
      "epoch": 25.59,
      "grad_norm": 0.05778868496417999,
      "learning_rate": 0.00016094117647058823,
      "loss": 0.0224,
      "step": 2866
    },
    {
      "epoch": 25.6,
      "grad_norm": 0.05610688775777817,
      "learning_rate": 0.00016088235294117644,
      "loss": 0.0235,
      "step": 2867
    },
    {
      "epoch": 25.61,
      "grad_norm": 0.06687482446432114,
      "learning_rate": 0.0001608235294117647,
      "loss": 0.0221,
      "step": 2868
    },
    {
      "epoch": 25.62,
      "grad_norm": 0.08267761021852493,
      "learning_rate": 0.00016076470588235293,
      "loss": 0.0293,
      "step": 2869
    },
    {
      "epoch": 25.62,
      "grad_norm": 0.0476447232067585,
      "learning_rate": 0.00016070588235294115,
      "loss": 0.0216,
      "step": 2870
    },
    {
      "epoch": 25.63,
      "grad_norm": 0.057498421519994736,
      "learning_rate": 0.00016064705882352942,
      "loss": 0.0248,
      "step": 2871
    },
    {
      "epoch": 25.64,
      "grad_norm": 0.06706904619932175,
      "learning_rate": 0.00016058823529411763,
      "loss": 0.0251,
      "step": 2872
    },
    {
      "epoch": 25.65,
      "grad_norm": 0.05606892332434654,
      "learning_rate": 0.00016052941176470588,
      "loss": 0.0217,
      "step": 2873
    },
    {
      "epoch": 25.66,
      "grad_norm": 0.054642338305711746,
      "learning_rate": 0.00016047058823529412,
      "loss": 0.0197,
      "step": 2874
    },
    {
      "epoch": 25.67,
      "grad_norm": 0.12919902801513672,
      "learning_rate": 0.00016041176470588234,
      "loss": 0.0412,
      "step": 2875
    },
    {
      "epoch": 25.68,
      "grad_norm": 0.06097327917814255,
      "learning_rate": 0.00016035294117647058,
      "loss": 0.0192,
      "step": 2876
    },
    {
      "epoch": 25.69,
      "grad_norm": 0.05964032560586929,
      "learning_rate": 0.0001602941176470588,
      "loss": 0.0244,
      "step": 2877
    },
    {
      "epoch": 25.7,
      "grad_norm": 0.060374390333890915,
      "learning_rate": 0.00016023529411764704,
      "loss": 0.0248,
      "step": 2878
    },
    {
      "epoch": 25.71,
      "grad_norm": 0.056740935891866684,
      "learning_rate": 0.00016017647058823528,
      "loss": 0.0256,
      "step": 2879
    },
    {
      "epoch": 25.71,
      "grad_norm": 0.05796075239777565,
      "learning_rate": 0.0001601176470588235,
      "loss": 0.0242,
      "step": 2880
    },
    {
      "epoch": 25.72,
      "grad_norm": 0.07502967864274979,
      "learning_rate": 0.00016005882352941177,
      "loss": 0.0216,
      "step": 2881
    },
    {
      "epoch": 25.73,
      "grad_norm": 0.09179262071847916,
      "learning_rate": 0.00015999999999999999,
      "loss": 0.0313,
      "step": 2882
    },
    {
      "epoch": 25.74,
      "grad_norm": 0.04778487607836723,
      "learning_rate": 0.0001599411764705882,
      "loss": 0.0231,
      "step": 2883
    },
    {
      "epoch": 25.75,
      "grad_norm": 0.05621251463890076,
      "learning_rate": 0.00015988235294117647,
      "loss": 0.0307,
      "step": 2884
    },
    {
      "epoch": 25.76,
      "grad_norm": 0.05809913948178291,
      "learning_rate": 0.0001598235294117647,
      "loss": 0.0268,
      "step": 2885
    },
    {
      "epoch": 25.77,
      "grad_norm": 0.05506697669625282,
      "learning_rate": 0.00015976470588235293,
      "loss": 0.0228,
      "step": 2886
    },
    {
      "epoch": 25.78,
      "grad_norm": 0.09386373311281204,
      "learning_rate": 0.00015970588235294115,
      "loss": 0.0301,
      "step": 2887
    },
    {
      "epoch": 25.79,
      "grad_norm": 0.11229165643453598,
      "learning_rate": 0.0001596470588235294,
      "loss": 0.0364,
      "step": 2888
    },
    {
      "epoch": 25.79,
      "grad_norm": 0.04829331114888191,
      "learning_rate": 0.00015958823529411764,
      "loss": 0.0205,
      "step": 2889
    },
    {
      "epoch": 25.8,
      "grad_norm": 0.04852853715419769,
      "learning_rate": 0.00015952941176470585,
      "loss": 0.0219,
      "step": 2890
    },
    {
      "epoch": 25.81,
      "grad_norm": 0.04522327333688736,
      "learning_rate": 0.00015947058823529412,
      "loss": 0.0181,
      "step": 2891
    },
    {
      "epoch": 25.82,
      "grad_norm": 0.05684347823262215,
      "learning_rate": 0.00015941176470588234,
      "loss": 0.0242,
      "step": 2892
    },
    {
      "epoch": 25.83,
      "grad_norm": 0.07527941465377808,
      "learning_rate": 0.00015935294117647056,
      "loss": 0.024,
      "step": 2893
    },
    {
      "epoch": 25.84,
      "grad_norm": 0.12193810939788818,
      "learning_rate": 0.00015929411764705883,
      "loss": 0.0382,
      "step": 2894
    },
    {
      "epoch": 25.85,
      "grad_norm": 0.05096936970949173,
      "learning_rate": 0.00015923529411764704,
      "loss": 0.0245,
      "step": 2895
    },
    {
      "epoch": 25.86,
      "grad_norm": 0.05811227858066559,
      "learning_rate": 0.00015917647058823529,
      "loss": 0.0289,
      "step": 2896
    },
    {
      "epoch": 25.87,
      "grad_norm": 0.05780656635761261,
      "learning_rate": 0.00015911764705882353,
      "loss": 0.0258,
      "step": 2897
    },
    {
      "epoch": 25.88,
      "grad_norm": 0.05754803493618965,
      "learning_rate": 0.00015905882352941175,
      "loss": 0.0201,
      "step": 2898
    },
    {
      "epoch": 25.88,
      "grad_norm": 0.0601537711918354,
      "learning_rate": 0.000159,
      "loss": 0.0238,
      "step": 2899
    },
    {
      "epoch": 25.89,
      "grad_norm": 0.11887750029563904,
      "learning_rate": 0.0001589411764705882,
      "loss": 0.0359,
      "step": 2900
    },
    {
      "epoch": 25.89,
      "eval_cer": 0.034576976995035195,
      "eval_loss": 0.1907992660999298,
      "eval_runtime": 22.3409,
      "eval_samples_per_second": 118.258,
      "eval_steps_per_second": 1.88,
      "eval_wer": 0.11047203490678302,
      "step": 2900
    },
    {
      "epoch": 25.9,
      "grad_norm": 0.05362885817885399,
      "learning_rate": 0.00015888235294117648,
      "loss": 0.0264,
      "step": 2901
    },
    {
      "epoch": 25.91,
      "grad_norm": 0.05292373523116112,
      "learning_rate": 0.0001588235294117647,
      "loss": 0.0269,
      "step": 2902
    },
    {
      "epoch": 25.92,
      "grad_norm": 0.05303650349378586,
      "learning_rate": 0.0001587647058823529,
      "loss": 0.0225,
      "step": 2903
    },
    {
      "epoch": 25.93,
      "grad_norm": 0.06052716076374054,
      "learning_rate": 0.00015870588235294118,
      "loss": 0.0275,
      "step": 2904
    },
    {
      "epoch": 25.94,
      "grad_norm": 0.055717941373586655,
      "learning_rate": 0.0001586470588235294,
      "loss": 0.0218,
      "step": 2905
    },
    {
      "epoch": 25.95,
      "grad_norm": 0.0733630508184433,
      "learning_rate": 0.00015858823529411764,
      "loss": 0.0241,
      "step": 2906
    },
    {
      "epoch": 25.96,
      "grad_norm": 0.08755376935005188,
      "learning_rate": 0.00015852941176470588,
      "loss": 0.027,
      "step": 2907
    },
    {
      "epoch": 25.96,
      "grad_norm": 0.052491169422864914,
      "learning_rate": 0.0001584705882352941,
      "loss": 0.019,
      "step": 2908
    },
    {
      "epoch": 25.97,
      "grad_norm": 0.0682724192738533,
      "learning_rate": 0.00015841176470588234,
      "loss": 0.0254,
      "step": 2909
    },
    {
      "epoch": 25.98,
      "grad_norm": 0.071016825735569,
      "learning_rate": 0.00015835294117647056,
      "loss": 0.0254,
      "step": 2910
    },
    {
      "epoch": 25.99,
      "grad_norm": 0.06912308186292648,
      "learning_rate": 0.0001582941176470588,
      "loss": 0.0227,
      "step": 2911
    },
    {
      "epoch": 26.0,
      "grad_norm": 0.08627953380346298,
      "learning_rate": 0.00015823529411764704,
      "loss": 0.0278,
      "step": 2912
    },
    {
      "epoch": 26.01,
      "grad_norm": 0.05893798917531967,
      "learning_rate": 0.00015817647058823526,
      "loss": 0.0263,
      "step": 2913
    },
    {
      "epoch": 26.02,
      "grad_norm": 0.05467715486884117,
      "learning_rate": 0.00015811764705882353,
      "loss": 0.0274,
      "step": 2914
    },
    {
      "epoch": 26.03,
      "grad_norm": 0.04888711869716644,
      "learning_rate": 0.00015805882352941175,
      "loss": 0.0245,
      "step": 2915
    },
    {
      "epoch": 26.04,
      "grad_norm": 0.0500195249915123,
      "learning_rate": 0.00015799999999999996,
      "loss": 0.0232,
      "step": 2916
    },
    {
      "epoch": 26.04,
      "grad_norm": 0.056167636066675186,
      "learning_rate": 0.00015794117647058824,
      "loss": 0.0249,
      "step": 2917
    },
    {
      "epoch": 26.05,
      "grad_norm": 0.08539999276399612,
      "learning_rate": 0.00015788235294117645,
      "loss": 0.029,
      "step": 2918
    },
    {
      "epoch": 26.06,
      "grad_norm": 0.07519900053739548,
      "learning_rate": 0.0001578235294117647,
      "loss": 0.0282,
      "step": 2919
    },
    {
      "epoch": 26.07,
      "grad_norm": 0.04647299647331238,
      "learning_rate": 0.00015776470588235294,
      "loss": 0.0249,
      "step": 2920
    },
    {
      "epoch": 26.08,
      "grad_norm": 0.05957910418510437,
      "learning_rate": 0.00015770588235294115,
      "loss": 0.0271,
      "step": 2921
    },
    {
      "epoch": 26.09,
      "grad_norm": 0.053026359528303146,
      "learning_rate": 0.0001576470588235294,
      "loss": 0.0243,
      "step": 2922
    },
    {
      "epoch": 26.1,
      "grad_norm": 0.046042006462812424,
      "learning_rate": 0.00015758823529411761,
      "loss": 0.0214,
      "step": 2923
    },
    {
      "epoch": 26.11,
      "grad_norm": 0.08956961333751678,
      "learning_rate": 0.00015752941176470588,
      "loss": 0.0269,
      "step": 2924
    },
    {
      "epoch": 26.12,
      "grad_norm": 0.11800762265920639,
      "learning_rate": 0.0001574705882352941,
      "loss": 0.0406,
      "step": 2925
    },
    {
      "epoch": 26.12,
      "grad_norm": 0.04754096269607544,
      "learning_rate": 0.00015741176470588232,
      "loss": 0.024,
      "step": 2926
    },
    {
      "epoch": 26.13,
      "grad_norm": 0.052196651697158813,
      "learning_rate": 0.0001573529411764706,
      "loss": 0.027,
      "step": 2927
    },
    {
      "epoch": 26.14,
      "grad_norm": 0.05159645155072212,
      "learning_rate": 0.0001572941176470588,
      "loss": 0.0231,
      "step": 2928
    },
    {
      "epoch": 26.15,
      "grad_norm": 0.0545731857419014,
      "learning_rate": 0.00015723529411764705,
      "loss": 0.0234,
      "step": 2929
    },
    {
      "epoch": 26.16,
      "grad_norm": 0.062391672283411026,
      "learning_rate": 0.0001571764705882353,
      "loss": 0.0246,
      "step": 2930
    },
    {
      "epoch": 26.17,
      "grad_norm": 0.0956978127360344,
      "learning_rate": 0.0001571176470588235,
      "loss": 0.0269,
      "step": 2931
    },
    {
      "epoch": 26.18,
      "grad_norm": 0.05004408210515976,
      "learning_rate": 0.00015705882352941175,
      "loss": 0.0225,
      "step": 2932
    },
    {
      "epoch": 26.19,
      "grad_norm": 0.05430254712700844,
      "learning_rate": 0.000157,
      "loss": 0.0269,
      "step": 2933
    },
    {
      "epoch": 26.2,
      "grad_norm": 0.051544129848480225,
      "learning_rate": 0.00015694117647058824,
      "loss": 0.0221,
      "step": 2934
    },
    {
      "epoch": 26.21,
      "grad_norm": 0.057763636112213135,
      "learning_rate": 0.00015688235294117645,
      "loss": 0.0231,
      "step": 2935
    },
    {
      "epoch": 26.21,
      "grad_norm": 0.060375794768333435,
      "learning_rate": 0.00015682352941176467,
      "loss": 0.0277,
      "step": 2936
    },
    {
      "epoch": 26.22,
      "grad_norm": 0.09702575951814651,
      "learning_rate": 0.00015676470588235294,
      "loss": 0.0319,
      "step": 2937
    },
    {
      "epoch": 26.23,
      "grad_norm": 0.04802504926919937,
      "learning_rate": 0.00015670588235294116,
      "loss": 0.0219,
      "step": 2938
    },
    {
      "epoch": 26.24,
      "grad_norm": 0.048738472163677216,
      "learning_rate": 0.00015664705882352943,
      "loss": 0.0211,
      "step": 2939
    },
    {
      "epoch": 26.25,
      "grad_norm": 0.046131592243909836,
      "learning_rate": 0.00015658823529411764,
      "loss": 0.0224,
      "step": 2940
    },
    {
      "epoch": 26.26,
      "grad_norm": 0.0489065982401371,
      "learning_rate": 0.00015652941176470586,
      "loss": 0.0211,
      "step": 2941
    },
    {
      "epoch": 26.27,
      "grad_norm": 0.05561669170856476,
      "learning_rate": 0.0001564705882352941,
      "loss": 0.0214,
      "step": 2942
    },
    {
      "epoch": 26.28,
      "grad_norm": 0.08418484032154083,
      "learning_rate": 0.00015641176470588235,
      "loss": 0.031,
      "step": 2943
    },
    {
      "epoch": 26.29,
      "grad_norm": 0.08891662210226059,
      "learning_rate": 0.00015635294117647056,
      "loss": 0.0304,
      "step": 2944
    },
    {
      "epoch": 26.29,
      "grad_norm": 0.05543001741170883,
      "learning_rate": 0.0001562941176470588,
      "loss": 0.0257,
      "step": 2945
    },
    {
      "epoch": 26.3,
      "grad_norm": 0.05482387915253639,
      "learning_rate": 0.00015623529411764702,
      "loss": 0.0251,
      "step": 2946
    },
    {
      "epoch": 26.31,
      "grad_norm": 0.05174694210290909,
      "learning_rate": 0.0001561764705882353,
      "loss": 0.0234,
      "step": 2947
    },
    {
      "epoch": 26.32,
      "grad_norm": 0.05397618189454079,
      "learning_rate": 0.0001561176470588235,
      "loss": 0.0237,
      "step": 2948
    },
    {
      "epoch": 26.33,
      "grad_norm": 0.06483880430459976,
      "learning_rate": 0.00015605882352941173,
      "loss": 0.024,
      "step": 2949
    },
    {
      "epoch": 26.34,
      "grad_norm": 0.12109798938035965,
      "learning_rate": 0.000156,
      "loss": 0.0289,
      "step": 2950
    },
    {
      "epoch": 26.35,
      "grad_norm": 0.050347570329904556,
      "learning_rate": 0.0001559411764705882,
      "loss": 0.0257,
      "step": 2951
    },
    {
      "epoch": 26.36,
      "grad_norm": 0.057522788643836975,
      "learning_rate": 0.00015588235294117646,
      "loss": 0.0237,
      "step": 2952
    },
    {
      "epoch": 26.37,
      "grad_norm": 0.049146901816129684,
      "learning_rate": 0.0001558235294117647,
      "loss": 0.0197,
      "step": 2953
    },
    {
      "epoch": 26.38,
      "grad_norm": 0.04698359966278076,
      "learning_rate": 0.00015576470588235292,
      "loss": 0.018,
      "step": 2954
    },
    {
      "epoch": 26.38,
      "grad_norm": 0.07178395241498947,
      "learning_rate": 0.00015570588235294116,
      "loss": 0.024,
      "step": 2955
    },
    {
      "epoch": 26.39,
      "grad_norm": 0.10159935057163239,
      "learning_rate": 0.0001556470588235294,
      "loss": 0.0315,
      "step": 2956
    },
    {
      "epoch": 26.4,
      "grad_norm": 0.05663328990340233,
      "learning_rate": 0.00015558823529411765,
      "loss": 0.0235,
      "step": 2957
    },
    {
      "epoch": 26.41,
      "grad_norm": 0.05715930089354515,
      "learning_rate": 0.00015552941176470586,
      "loss": 0.0254,
      "step": 2958
    },
    {
      "epoch": 26.42,
      "grad_norm": 0.06861431896686554,
      "learning_rate": 0.00015547058823529408,
      "loss": 0.0208,
      "step": 2959
    },
    {
      "epoch": 26.43,
      "grad_norm": 0.05948348343372345,
      "learning_rate": 0.00015541176470588235,
      "loss": 0.0209,
      "step": 2960
    },
    {
      "epoch": 26.44,
      "grad_norm": 0.0590101033449173,
      "learning_rate": 0.00015535294117647057,
      "loss": 0.0242,
      "step": 2961
    },
    {
      "epoch": 26.45,
      "grad_norm": 0.09696076065301895,
      "learning_rate": 0.00015529411764705884,
      "loss": 0.0324,
      "step": 2962
    },
    {
      "epoch": 26.46,
      "grad_norm": 0.04854610934853554,
      "learning_rate": 0.00015523529411764705,
      "loss": 0.024,
      "step": 2963
    },
    {
      "epoch": 26.46,
      "grad_norm": 0.04646662622690201,
      "learning_rate": 0.00015517647058823527,
      "loss": 0.0225,
      "step": 2964
    },
    {
      "epoch": 26.47,
      "grad_norm": 0.05486273020505905,
      "learning_rate": 0.0001551176470588235,
      "loss": 0.0248,
      "step": 2965
    },
    {
      "epoch": 26.48,
      "grad_norm": 0.049627941101789474,
      "learning_rate": 0.00015505882352941176,
      "loss": 0.0208,
      "step": 2966
    },
    {
      "epoch": 26.49,
      "grad_norm": 0.05518784373998642,
      "learning_rate": 0.000155,
      "loss": 0.0248,
      "step": 2967
    },
    {
      "epoch": 26.5,
      "grad_norm": 0.07837014645338058,
      "learning_rate": 0.00015494117647058822,
      "loss": 0.0265,
      "step": 2968
    },
    {
      "epoch": 26.51,
      "grad_norm": 0.11061987280845642,
      "learning_rate": 0.00015488235294117646,
      "loss": 0.0301,
      "step": 2969
    },
    {
      "epoch": 26.52,
      "grad_norm": 0.050842780619859695,
      "learning_rate": 0.0001548235294117647,
      "loss": 0.0212,
      "step": 2970
    },
    {
      "epoch": 26.53,
      "grad_norm": 0.05200947821140289,
      "learning_rate": 0.00015476470588235292,
      "loss": 0.0231,
      "step": 2971
    },
    {
      "epoch": 26.54,
      "grad_norm": 0.0542827807366848,
      "learning_rate": 0.0001547058823529412,
      "loss": 0.0251,
      "step": 2972
    },
    {
      "epoch": 26.54,
      "grad_norm": 0.05743546783924103,
      "learning_rate": 0.0001546470588235294,
      "loss": 0.0236,
      "step": 2973
    },
    {
      "epoch": 26.55,
      "grad_norm": 0.06169158220291138,
      "learning_rate": 0.00015458823529411762,
      "loss": 0.0196,
      "step": 2974
    },
    {
      "epoch": 26.56,
      "grad_norm": 0.08541537076234818,
      "learning_rate": 0.00015452941176470587,
      "loss": 0.03,
      "step": 2975
    },
    {
      "epoch": 26.57,
      "grad_norm": 0.053783103823661804,
      "learning_rate": 0.0001544705882352941,
      "loss": 0.0236,
      "step": 2976
    },
    {
      "epoch": 26.58,
      "grad_norm": 0.05454250052571297,
      "learning_rate": 0.00015441176470588233,
      "loss": 0.023,
      "step": 2977
    },
    {
      "epoch": 26.59,
      "grad_norm": 0.06030486896634102,
      "learning_rate": 0.00015435294117647057,
      "loss": 0.0263,
      "step": 2978
    },
    {
      "epoch": 26.6,
      "grad_norm": 0.05425736680626869,
      "learning_rate": 0.0001542941176470588,
      "loss": 0.024,
      "step": 2979
    },
    {
      "epoch": 26.61,
      "grad_norm": 0.05609467625617981,
      "learning_rate": 0.00015423529411764706,
      "loss": 0.0179,
      "step": 2980
    },
    {
      "epoch": 26.62,
      "grad_norm": 0.19363701343536377,
      "learning_rate": 0.00015417647058823527,
      "loss": 0.031,
      "step": 2981
    },
    {
      "epoch": 26.62,
      "grad_norm": 0.050824277102947235,
      "learning_rate": 0.0001541176470588235,
      "loss": 0.0233,
      "step": 2982
    },
    {
      "epoch": 26.63,
      "grad_norm": 0.05721789225935936,
      "learning_rate": 0.00015405882352941176,
      "loss": 0.0257,
      "step": 2983
    },
    {
      "epoch": 26.64,
      "grad_norm": 0.05407975986599922,
      "learning_rate": 0.00015399999999999998,
      "loss": 0.0224,
      "step": 2984
    },
    {
      "epoch": 26.65,
      "grad_norm": 0.05091584846377373,
      "learning_rate": 0.00015394117647058825,
      "loss": 0.0229,
      "step": 2985
    },
    {
      "epoch": 26.66,
      "grad_norm": 0.057925041764974594,
      "learning_rate": 0.00015388235294117646,
      "loss": 0.0265,
      "step": 2986
    },
    {
      "epoch": 26.67,
      "grad_norm": 0.10425586253404617,
      "learning_rate": 0.00015382352941176468,
      "loss": 0.0312,
      "step": 2987
    },
    {
      "epoch": 26.68,
      "grad_norm": 0.054229021072387695,
      "learning_rate": 0.00015376470588235292,
      "loss": 0.0257,
      "step": 2988
    },
    {
      "epoch": 26.69,
      "grad_norm": 0.05078696832060814,
      "learning_rate": 0.00015370588235294117,
      "loss": 0.0218,
      "step": 2989
    },
    {
      "epoch": 26.7,
      "grad_norm": 0.05847737938165665,
      "learning_rate": 0.0001536470588235294,
      "loss": 0.0263,
      "step": 2990
    },
    {
      "epoch": 26.71,
      "grad_norm": 0.047788508236408234,
      "learning_rate": 0.00015358823529411763,
      "loss": 0.021,
      "step": 2991
    },
    {
      "epoch": 26.71,
      "grad_norm": 0.059393640607595444,
      "learning_rate": 0.00015352941176470587,
      "loss": 0.0226,
      "step": 2992
    },
    {
      "epoch": 26.72,
      "grad_norm": 0.08014336973428726,
      "learning_rate": 0.0001534705882352941,
      "loss": 0.0284,
      "step": 2993
    },
    {
      "epoch": 26.73,
      "grad_norm": 0.07928035408258438,
      "learning_rate": 0.00015341176470588233,
      "loss": 0.026,
      "step": 2994
    },
    {
      "epoch": 26.74,
      "grad_norm": 0.05100613459944725,
      "learning_rate": 0.0001533529411764706,
      "loss": 0.0237,
      "step": 2995
    },
    {
      "epoch": 26.75,
      "grad_norm": 0.05011659488081932,
      "learning_rate": 0.00015329411764705882,
      "loss": 0.0225,
      "step": 2996
    },
    {
      "epoch": 26.76,
      "grad_norm": 0.06029354780912399,
      "learning_rate": 0.00015323529411764703,
      "loss": 0.024,
      "step": 2997
    },
    {
      "epoch": 26.77,
      "grad_norm": 0.05785534158349037,
      "learning_rate": 0.0001531764705882353,
      "loss": 0.023,
      "step": 2998
    },
    {
      "epoch": 26.78,
      "grad_norm": 0.07849323004484177,
      "learning_rate": 0.00015311764705882352,
      "loss": 0.0296,
      "step": 2999
    },
    {
      "epoch": 26.79,
      "grad_norm": 0.08631402999162674,
      "learning_rate": 0.00015305882352941176,
      "loss": 0.0265,
      "step": 3000
    },
    {
      "epoch": 26.79,
      "eval_cer": 0.033731344405482706,
      "eval_loss": 0.19908082485198975,
      "eval_runtime": 22.4013,
      "eval_samples_per_second": 117.94,
      "eval_steps_per_second": 1.875,
      "eval_wer": 0.10959936533121777,
      "step": 3000
    },
    {
      "epoch": 26.79,
      "grad_norm": 0.054585572332143784,
      "learning_rate": 0.00015299999999999998,
      "loss": 0.0245,
      "step": 3001
    },
    {
      "epoch": 26.8,
      "grad_norm": 0.05038483068346977,
      "learning_rate": 0.00015294117647058822,
      "loss": 0.0283,
      "step": 3002
    },
    {
      "epoch": 26.81,
      "grad_norm": 0.05247160419821739,
      "learning_rate": 0.00015288235294117647,
      "loss": 0.0241,
      "step": 3003
    },
    {
      "epoch": 26.82,
      "grad_norm": 0.052994318306446075,
      "learning_rate": 0.00015282352941176468,
      "loss": 0.0229,
      "step": 3004
    },
    {
      "epoch": 26.83,
      "grad_norm": 0.05985628813505173,
      "learning_rate": 0.0001527647058823529,
      "loss": 0.0246,
      "step": 3005
    },
    {
      "epoch": 26.84,
      "grad_norm": 0.11439044028520584,
      "learning_rate": 0.00015270588235294117,
      "loss": 0.0347,
      "step": 3006
    },
    {
      "epoch": 26.85,
      "grad_norm": 0.046517860144376755,
      "learning_rate": 0.00015264705882352938,
      "loss": 0.0224,
      "step": 3007
    },
    {
      "epoch": 26.86,
      "grad_norm": 0.04747984558343887,
      "learning_rate": 0.00015258823529411766,
      "loss": 0.0192,
      "step": 3008
    },
    {
      "epoch": 26.87,
      "grad_norm": 0.05847001448273659,
      "learning_rate": 0.00015252941176470587,
      "loss": 0.0269,
      "step": 3009
    },
    {
      "epoch": 26.88,
      "grad_norm": 0.06655603647232056,
      "learning_rate": 0.0001524705882352941,
      "loss": 0.0245,
      "step": 3010
    },
    {
      "epoch": 26.88,
      "grad_norm": 0.05660974979400635,
      "learning_rate": 0.00015241176470588233,
      "loss": 0.0215,
      "step": 3011
    },
    {
      "epoch": 26.89,
      "grad_norm": 0.1280333697795868,
      "learning_rate": 0.00015235294117647057,
      "loss": 0.0354,
      "step": 3012
    },
    {
      "epoch": 26.9,
      "grad_norm": 0.052807681262493134,
      "learning_rate": 0.00015229411764705882,
      "loss": 0.0254,
      "step": 3013
    },
    {
      "epoch": 26.91,
      "grad_norm": 0.05705934017896652,
      "learning_rate": 0.00015223529411764703,
      "loss": 0.0215,
      "step": 3014
    },
    {
      "epoch": 26.92,
      "grad_norm": 0.05490132048726082,
      "learning_rate": 0.00015217647058823528,
      "loss": 0.0203,
      "step": 3015
    },
    {
      "epoch": 26.93,
      "grad_norm": 0.05354372784495354,
      "learning_rate": 0.00015211764705882352,
      "loss": 0.025,
      "step": 3016
    },
    {
      "epoch": 26.94,
      "grad_norm": 0.06647264212369919,
      "learning_rate": 0.00015205882352941174,
      "loss": 0.0225,
      "step": 3017
    },
    {
      "epoch": 26.95,
      "grad_norm": 0.06453800201416016,
      "learning_rate": 0.000152,
      "loss": 0.0228,
      "step": 3018
    },
    {
      "epoch": 26.96,
      "grad_norm": 0.07514040917158127,
      "learning_rate": 0.00015194117647058822,
      "loss": 0.0257,
      "step": 3019
    },
    {
      "epoch": 26.96,
      "grad_norm": 0.061309799551963806,
      "learning_rate": 0.00015188235294117644,
      "loss": 0.0229,
      "step": 3020
    },
    {
      "epoch": 26.97,
      "grad_norm": 0.05893779918551445,
      "learning_rate": 0.0001518235294117647,
      "loss": 0.0254,
      "step": 3021
    },
    {
      "epoch": 26.98,
      "grad_norm": 0.059200119227170944,
      "learning_rate": 0.00015176470588235293,
      "loss": 0.0198,
      "step": 3022
    },
    {
      "epoch": 26.99,
      "grad_norm": 0.06384547054767609,
      "learning_rate": 0.00015170588235294117,
      "loss": 0.0237,
      "step": 3023
    },
    {
      "epoch": 27.0,
      "grad_norm": 0.08574508130550385,
      "learning_rate": 0.0001516470588235294,
      "loss": 0.0257,
      "step": 3024
    },
    {
      "epoch": 27.01,
      "grad_norm": 0.050044167786836624,
      "learning_rate": 0.00015158823529411763,
      "loss": 0.023,
      "step": 3025
    },
    {
      "epoch": 27.02,
      "grad_norm": 0.05156770721077919,
      "learning_rate": 0.00015152941176470587,
      "loss": 0.0239,
      "step": 3026
    },
    {
      "epoch": 27.03,
      "grad_norm": 0.04528946802020073,
      "learning_rate": 0.0001514705882352941,
      "loss": 0.0198,
      "step": 3027
    },
    {
      "epoch": 27.04,
      "grad_norm": 0.055816467851400375,
      "learning_rate": 0.00015141176470588236,
      "loss": 0.0214,
      "step": 3028
    },
    {
      "epoch": 27.04,
      "grad_norm": 0.04916231334209442,
      "learning_rate": 0.00015135294117647058,
      "loss": 0.0208,
      "step": 3029
    },
    {
      "epoch": 27.05,
      "grad_norm": 0.08857579529285431,
      "learning_rate": 0.0001512941176470588,
      "loss": 0.0247,
      "step": 3030
    },
    {
      "epoch": 27.06,
      "grad_norm": 0.06720104813575745,
      "learning_rate": 0.00015123529411764706,
      "loss": 0.0219,
      "step": 3031
    },
    {
      "epoch": 27.07,
      "grad_norm": 0.055760834366083145,
      "learning_rate": 0.00015117647058823528,
      "loss": 0.0224,
      "step": 3032
    },
    {
      "epoch": 27.08,
      "grad_norm": 0.05123697221279144,
      "learning_rate": 0.00015111764705882352,
      "loss": 0.0233,
      "step": 3033
    },
    {
      "epoch": 27.09,
      "grad_norm": 0.050634220242500305,
      "learning_rate": 0.00015105882352941177,
      "loss": 0.0198,
      "step": 3034
    },
    {
      "epoch": 27.1,
      "grad_norm": 0.058983735740184784,
      "learning_rate": 0.00015099999999999998,
      "loss": 0.023,
      "step": 3035
    },
    {
      "epoch": 27.11,
      "grad_norm": 0.057488929480314255,
      "learning_rate": 0.00015094117647058823,
      "loss": 0.0201,
      "step": 3036
    },
    {
      "epoch": 27.12,
      "grad_norm": 0.09951991587877274,
      "learning_rate": 0.00015088235294117644,
      "loss": 0.0283,
      "step": 3037
    },
    {
      "epoch": 27.12,
      "grad_norm": 0.048263560980558395,
      "learning_rate": 0.0001508235294117647,
      "loss": 0.0226,
      "step": 3038
    },
    {
      "epoch": 27.13,
      "grad_norm": 0.055034998804330826,
      "learning_rate": 0.00015076470588235293,
      "loss": 0.0269,
      "step": 3039
    },
    {
      "epoch": 27.14,
      "grad_norm": 0.05377781391143799,
      "learning_rate": 0.00015070588235294115,
      "loss": 0.0228,
      "step": 3040
    },
    {
      "epoch": 27.15,
      "grad_norm": 0.05008714646100998,
      "learning_rate": 0.00015064705882352942,
      "loss": 0.0217,
      "step": 3041
    },
    {
      "epoch": 27.16,
      "grad_norm": 0.06198955699801445,
      "learning_rate": 0.00015058823529411763,
      "loss": 0.0228,
      "step": 3042
    },
    {
      "epoch": 27.17,
      "grad_norm": 0.09308627992868423,
      "learning_rate": 0.00015052941176470585,
      "loss": 0.0342,
      "step": 3043
    },
    {
      "epoch": 27.18,
      "grad_norm": 0.05396142601966858,
      "learning_rate": 0.00015047058823529412,
      "loss": 0.0231,
      "step": 3044
    },
    {
      "epoch": 27.19,
      "grad_norm": 0.04787048324942589,
      "learning_rate": 0.00015041176470588234,
      "loss": 0.0231,
      "step": 3045
    },
    {
      "epoch": 27.2,
      "grad_norm": 0.05582379549741745,
      "learning_rate": 0.00015035294117647058,
      "loss": 0.0259,
      "step": 3046
    },
    {
      "epoch": 27.21,
      "grad_norm": 0.04990950971841812,
      "learning_rate": 0.0001502941176470588,
      "loss": 0.0218,
      "step": 3047
    },
    {
      "epoch": 27.21,
      "grad_norm": 0.05668466538190842,
      "learning_rate": 0.00015023529411764704,
      "loss": 0.0234,
      "step": 3048
    },
    {
      "epoch": 27.22,
      "grad_norm": 0.1383943408727646,
      "learning_rate": 0.00015017647058823528,
      "loss": 0.037,
      "step": 3049
    },
    {
      "epoch": 27.23,
      "grad_norm": 0.05292258784174919,
      "learning_rate": 0.0001501176470588235,
      "loss": 0.0249,
      "step": 3050
    },
    {
      "epoch": 27.24,
      "grad_norm": 0.050082866102457047,
      "learning_rate": 0.00015005882352941177,
      "loss": 0.025,
      "step": 3051
    },
    {
      "epoch": 27.25,
      "grad_norm": 0.05611185356974602,
      "learning_rate": 0.00015,
      "loss": 0.0263,
      "step": 3052
    },
    {
      "epoch": 27.26,
      "grad_norm": 0.047544948756694794,
      "learning_rate": 0.00014994117647058823,
      "loss": 0.0213,
      "step": 3053
    },
    {
      "epoch": 27.27,
      "grad_norm": 0.054379239678382874,
      "learning_rate": 0.00014988235294117647,
      "loss": 0.0194,
      "step": 3054
    },
    {
      "epoch": 27.28,
      "grad_norm": 0.12211582809686661,
      "learning_rate": 0.0001498235294117647,
      "loss": 0.0293,
      "step": 3055
    },
    {
      "epoch": 27.29,
      "grad_norm": 0.07340998947620392,
      "learning_rate": 0.00014976470588235293,
      "loss": 0.0277,
      "step": 3056
    },
    {
      "epoch": 27.29,
      "grad_norm": 0.04934728890657425,
      "learning_rate": 0.00014970588235294118,
      "loss": 0.0217,
      "step": 3057
    },
    {
      "epoch": 27.3,
      "grad_norm": 0.06107524037361145,
      "learning_rate": 0.0001496470588235294,
      "loss": 0.0266,
      "step": 3058
    },
    {
      "epoch": 27.31,
      "grad_norm": 0.046629469841718674,
      "learning_rate": 0.00014958823529411764,
      "loss": 0.0175,
      "step": 3059
    },
    {
      "epoch": 27.32,
      "grad_norm": 0.05639394000172615,
      "learning_rate": 0.00014952941176470585,
      "loss": 0.0219,
      "step": 3060
    },
    {
      "epoch": 27.33,
      "grad_norm": 0.07006697356700897,
      "learning_rate": 0.0001494705882352941,
      "loss": 0.0248,
      "step": 3061
    },
    {
      "epoch": 27.34,
      "grad_norm": 0.06783292442560196,
      "learning_rate": 0.00014941176470588234,
      "loss": 0.0253,
      "step": 3062
    },
    {
      "epoch": 27.35,
      "grad_norm": 0.059967514127492905,
      "learning_rate": 0.00014935294117647058,
      "loss": 0.023,
      "step": 3063
    },
    {
      "epoch": 27.36,
      "grad_norm": 0.050735682249069214,
      "learning_rate": 0.0001492941176470588,
      "loss": 0.0219,
      "step": 3064
    },
    {
      "epoch": 27.37,
      "grad_norm": 0.056490566581487656,
      "learning_rate": 0.00014923529411764704,
      "loss": 0.023,
      "step": 3065
    },
    {
      "epoch": 27.38,
      "grad_norm": 0.06102607026696205,
      "learning_rate": 0.0001491764705882353,
      "loss": 0.0252,
      "step": 3066
    },
    {
      "epoch": 27.38,
      "grad_norm": 0.0579913966357708,
      "learning_rate": 0.00014911764705882353,
      "loss": 0.0167,
      "step": 3067
    },
    {
      "epoch": 27.39,
      "grad_norm": 0.10275489836931229,
      "learning_rate": 0.00014905882352941175,
      "loss": 0.0318,
      "step": 3068
    },
    {
      "epoch": 27.4,
      "grad_norm": 0.04819761961698532,
      "learning_rate": 0.000149,
      "loss": 0.0224,
      "step": 3069
    },
    {
      "epoch": 27.41,
      "grad_norm": 0.04517188295722008,
      "learning_rate": 0.0001489411764705882,
      "loss": 0.0229,
      "step": 3070
    },
    {
      "epoch": 27.42,
      "grad_norm": 0.051060132682323456,
      "learning_rate": 0.00014888235294117645,
      "loss": 0.02,
      "step": 3071
    },
    {
      "epoch": 27.43,
      "grad_norm": 0.05070464685559273,
      "learning_rate": 0.0001488235294117647,
      "loss": 0.0226,
      "step": 3072
    },
    {
      "epoch": 27.44,
      "grad_norm": 0.05855333060026169,
      "learning_rate": 0.00014876470588235294,
      "loss": 0.0204,
      "step": 3073
    },
    {
      "epoch": 27.45,
      "grad_norm": 0.09937647730112076,
      "learning_rate": 0.00014870588235294115,
      "loss": 0.031,
      "step": 3074
    },
    {
      "epoch": 27.46,
      "grad_norm": 0.047324638813734055,
      "learning_rate": 0.0001486470588235294,
      "loss": 0.0216,
      "step": 3075
    },
    {
      "epoch": 27.46,
      "grad_norm": 0.05366073548793793,
      "learning_rate": 0.00014858823529411764,
      "loss": 0.0219,
      "step": 3076
    },
    {
      "epoch": 27.47,
      "grad_norm": 0.05626094341278076,
      "learning_rate": 0.00014852941176470588,
      "loss": 0.0242,
      "step": 3077
    },
    {
      "epoch": 27.48,
      "grad_norm": 0.06329621374607086,
      "learning_rate": 0.0001484705882352941,
      "loss": 0.0213,
      "step": 3078
    },
    {
      "epoch": 27.49,
      "grad_norm": 0.056740060448646545,
      "learning_rate": 0.00014841176470588234,
      "loss": 0.0208,
      "step": 3079
    },
    {
      "epoch": 27.5,
      "grad_norm": 0.07845351099967957,
      "learning_rate": 0.00014835294117647059,
      "loss": 0.0248,
      "step": 3080
    },
    {
      "epoch": 27.51,
      "grad_norm": 0.07826192677021027,
      "learning_rate": 0.0001482941176470588,
      "loss": 0.026,
      "step": 3081
    },
    {
      "epoch": 27.52,
      "grad_norm": 0.05459087714552879,
      "learning_rate": 0.00014823529411764705,
      "loss": 0.0255,
      "step": 3082
    },
    {
      "epoch": 27.53,
      "grad_norm": 0.05910929664969444,
      "learning_rate": 0.00014817647058823526,
      "loss": 0.0248,
      "step": 3083
    },
    {
      "epoch": 27.54,
      "grad_norm": 0.04926055669784546,
      "learning_rate": 0.0001481176470588235,
      "loss": 0.0221,
      "step": 3084
    },
    {
      "epoch": 27.54,
      "grad_norm": 0.052828963845968246,
      "learning_rate": 0.00014805882352941175,
      "loss": 0.0197,
      "step": 3085
    },
    {
      "epoch": 27.55,
      "grad_norm": 0.06274675577878952,
      "learning_rate": 0.000148,
      "loss": 0.0247,
      "step": 3086
    },
    {
      "epoch": 27.56,
      "grad_norm": 0.06828172504901886,
      "learning_rate": 0.00014794117647058824,
      "loss": 0.028,
      "step": 3087
    },
    {
      "epoch": 27.57,
      "grad_norm": 0.05155690386891365,
      "learning_rate": 0.00014788235294117645,
      "loss": 0.0256,
      "step": 3088
    },
    {
      "epoch": 27.58,
      "grad_norm": 0.047592706978321075,
      "learning_rate": 0.0001478235294117647,
      "loss": 0.021,
      "step": 3089
    },
    {
      "epoch": 27.59,
      "grad_norm": 0.056468334048986435,
      "learning_rate": 0.00014776470588235294,
      "loss": 0.0212,
      "step": 3090
    },
    {
      "epoch": 27.6,
      "grad_norm": 0.05021779611706734,
      "learning_rate": 0.00014770588235294116,
      "loss": 0.0197,
      "step": 3091
    },
    {
      "epoch": 27.61,
      "grad_norm": 0.06249059736728668,
      "learning_rate": 0.0001476470588235294,
      "loss": 0.0225,
      "step": 3092
    },
    {
      "epoch": 27.62,
      "grad_norm": 0.11793142557144165,
      "learning_rate": 0.00014758823529411764,
      "loss": 0.0356,
      "step": 3093
    },
    {
      "epoch": 27.62,
      "grad_norm": 0.053746987134218216,
      "learning_rate": 0.00014752941176470586,
      "loss": 0.021,
      "step": 3094
    },
    {
      "epoch": 27.63,
      "grad_norm": 0.0484970398247242,
      "learning_rate": 0.0001474705882352941,
      "loss": 0.0222,
      "step": 3095
    },
    {
      "epoch": 27.64,
      "grad_norm": 0.05492851883172989,
      "learning_rate": 0.00014741176470588235,
      "loss": 0.0237,
      "step": 3096
    },
    {
      "epoch": 27.65,
      "grad_norm": 0.06060340628027916,
      "learning_rate": 0.00014735294117647056,
      "loss": 0.024,
      "step": 3097
    },
    {
      "epoch": 27.66,
      "grad_norm": 0.06306329369544983,
      "learning_rate": 0.0001472941176470588,
      "loss": 0.0231,
      "step": 3098
    },
    {
      "epoch": 27.67,
      "grad_norm": 0.0970316007733345,
      "learning_rate": 0.00014723529411764705,
      "loss": 0.0279,
      "step": 3099
    },
    {
      "epoch": 27.68,
      "grad_norm": 0.04738640785217285,
      "learning_rate": 0.0001471764705882353,
      "loss": 0.0223,
      "step": 3100
    },
    {
      "epoch": 27.68,
      "eval_cer": 0.034483017818418255,
      "eval_loss": 0.1947711706161499,
      "eval_runtime": 22.3359,
      "eval_samples_per_second": 118.285,
      "eval_steps_per_second": 1.88,
      "eval_wer": 0.11069020230067433,
      "step": 3100
    },
    {
      "epoch": 27.69,
      "grad_norm": 0.050134338438510895,
      "learning_rate": 0.00014711764705882354,
      "loss": 0.0232,
      "step": 3101
    },
    {
      "epoch": 27.7,
      "grad_norm": 0.046567123383283615,
      "learning_rate": 0.00014705882352941175,
      "loss": 0.0198,
      "step": 3102
    },
    {
      "epoch": 27.71,
      "grad_norm": 0.04978331923484802,
      "learning_rate": 0.000147,
      "loss": 0.0177,
      "step": 3103
    },
    {
      "epoch": 27.71,
      "grad_norm": 0.05636028200387955,
      "learning_rate": 0.0001469411764705882,
      "loss": 0.0236,
      "step": 3104
    },
    {
      "epoch": 27.72,
      "grad_norm": 0.09316873550415039,
      "learning_rate": 0.00014688235294117646,
      "loss": 0.0271,
      "step": 3105
    },
    {
      "epoch": 27.73,
      "grad_norm": 0.07818813621997833,
      "learning_rate": 0.0001468235294117647,
      "loss": 0.0263,
      "step": 3106
    },
    {
      "epoch": 27.74,
      "grad_norm": 0.04544926434755325,
      "learning_rate": 0.00014676470588235291,
      "loss": 0.0207,
      "step": 3107
    },
    {
      "epoch": 27.75,
      "grad_norm": 0.0495959110558033,
      "learning_rate": 0.00014670588235294116,
      "loss": 0.0201,
      "step": 3108
    },
    {
      "epoch": 27.76,
      "grad_norm": 0.05512333661317825,
      "learning_rate": 0.0001466470588235294,
      "loss": 0.0243,
      "step": 3109
    },
    {
      "epoch": 27.77,
      "grad_norm": 0.04825921729207039,
      "learning_rate": 0.00014658823529411765,
      "loss": 0.0197,
      "step": 3110
    },
    {
      "epoch": 27.78,
      "grad_norm": 0.0721101313829422,
      "learning_rate": 0.00014652941176470586,
      "loss": 0.026,
      "step": 3111
    },
    {
      "epoch": 27.79,
      "grad_norm": 0.08447795361280441,
      "learning_rate": 0.0001464705882352941,
      "loss": 0.0264,
      "step": 3112
    },
    {
      "epoch": 27.79,
      "grad_norm": 0.054003193974494934,
      "learning_rate": 0.00014641176470588235,
      "loss": 0.0225,
      "step": 3113
    },
    {
      "epoch": 27.8,
      "grad_norm": 0.05464816093444824,
      "learning_rate": 0.0001463529411764706,
      "loss": 0.0245,
      "step": 3114
    },
    {
      "epoch": 27.81,
      "grad_norm": 0.05618251860141754,
      "learning_rate": 0.0001462941176470588,
      "loss": 0.0225,
      "step": 3115
    },
    {
      "epoch": 27.82,
      "grad_norm": 0.06628713756799698,
      "learning_rate": 0.00014623529411764705,
      "loss": 0.0249,
      "step": 3116
    },
    {
      "epoch": 27.83,
      "grad_norm": 0.06224177032709122,
      "learning_rate": 0.00014617647058823527,
      "loss": 0.0215,
      "step": 3117
    },
    {
      "epoch": 27.84,
      "grad_norm": 0.11041702330112457,
      "learning_rate": 0.0001461176470588235,
      "loss": 0.0311,
      "step": 3118
    },
    {
      "epoch": 27.85,
      "grad_norm": 0.04507511481642723,
      "learning_rate": 0.00014605882352941175,
      "loss": 0.0221,
      "step": 3119
    },
    {
      "epoch": 27.86,
      "grad_norm": 0.05189411714673042,
      "learning_rate": 0.000146,
      "loss": 0.0228,
      "step": 3120
    },
    {
      "epoch": 27.87,
      "grad_norm": 0.05701563507318497,
      "learning_rate": 0.00014594117647058821,
      "loss": 0.0246,
      "step": 3121
    },
    {
      "epoch": 27.88,
      "grad_norm": 0.04980058595538139,
      "learning_rate": 0.00014588235294117646,
      "loss": 0.0204,
      "step": 3122
    },
    {
      "epoch": 27.88,
      "grad_norm": 0.06099824979901314,
      "learning_rate": 0.0001458235294117647,
      "loss": 0.0223,
      "step": 3123
    },
    {
      "epoch": 27.89,
      "grad_norm": 0.1298723965883255,
      "learning_rate": 0.00014576470588235294,
      "loss": 0.0336,
      "step": 3124
    },
    {
      "epoch": 27.9,
      "grad_norm": 0.04897035285830498,
      "learning_rate": 0.00014570588235294116,
      "loss": 0.0217,
      "step": 3125
    },
    {
      "epoch": 27.91,
      "grad_norm": 0.04782100021839142,
      "learning_rate": 0.0001456470588235294,
      "loss": 0.0235,
      "step": 3126
    },
    {
      "epoch": 27.92,
      "grad_norm": 0.06138615682721138,
      "learning_rate": 0.00014558823529411762,
      "loss": 0.0274,
      "step": 3127
    },
    {
      "epoch": 27.93,
      "grad_norm": 0.06044183671474457,
      "learning_rate": 0.00014552941176470586,
      "loss": 0.0235,
      "step": 3128
    },
    {
      "epoch": 27.94,
      "grad_norm": 0.05670559033751488,
      "learning_rate": 0.0001454705882352941,
      "loss": 0.0184,
      "step": 3129
    },
    {
      "epoch": 27.95,
      "grad_norm": 0.07986319810152054,
      "learning_rate": 0.00014541176470588232,
      "loss": 0.0225,
      "step": 3130
    },
    {
      "epoch": 27.96,
      "grad_norm": 0.1061500832438469,
      "learning_rate": 0.00014535294117647057,
      "loss": 0.0319,
      "step": 3131
    },
    {
      "epoch": 27.96,
      "grad_norm": 0.049021266400814056,
      "learning_rate": 0.0001452941176470588,
      "loss": 0.0226,
      "step": 3132
    },
    {
      "epoch": 27.97,
      "grad_norm": 0.05631643906235695,
      "learning_rate": 0.00014523529411764705,
      "loss": 0.0264,
      "step": 3133
    },
    {
      "epoch": 27.98,
      "grad_norm": 0.051798731088638306,
      "learning_rate": 0.0001451764705882353,
      "loss": 0.0207,
      "step": 3134
    },
    {
      "epoch": 27.99,
      "grad_norm": 0.06304339319467545,
      "learning_rate": 0.00014511764705882351,
      "loss": 0.0206,
      "step": 3135
    },
    {
      "epoch": 28.0,
      "grad_norm": 0.07847974449396133,
      "learning_rate": 0.00014505882352941176,
      "loss": 0.0228,
      "step": 3136
    },
    {
      "epoch": 28.01,
      "grad_norm": 0.0499185211956501,
      "learning_rate": 0.000145,
      "loss": 0.0236,
      "step": 3137
    },
    {
      "epoch": 28.02,
      "grad_norm": 0.049133725464344025,
      "learning_rate": 0.00014494117647058822,
      "loss": 0.021,
      "step": 3138
    },
    {
      "epoch": 28.03,
      "grad_norm": 0.05714169517159462,
      "learning_rate": 0.00014488235294117646,
      "loss": 0.0273,
      "step": 3139
    },
    {
      "epoch": 28.04,
      "grad_norm": 0.0510253980755806,
      "learning_rate": 0.00014482352941176468,
      "loss": 0.0211,
      "step": 3140
    },
    {
      "epoch": 28.04,
      "grad_norm": 0.05600978434085846,
      "learning_rate": 0.00014476470588235292,
      "loss": 0.0237,
      "step": 3141
    },
    {
      "epoch": 28.05,
      "grad_norm": 0.08738350123167038,
      "learning_rate": 0.00014470588235294116,
      "loss": 0.0289,
      "step": 3142
    },
    {
      "epoch": 28.06,
      "grad_norm": 0.08066991716623306,
      "learning_rate": 0.0001446470588235294,
      "loss": 0.0273,
      "step": 3143
    },
    {
      "epoch": 28.07,
      "grad_norm": 0.04763602092862129,
      "learning_rate": 0.00014458823529411762,
      "loss": 0.0246,
      "step": 3144
    },
    {
      "epoch": 28.08,
      "grad_norm": 0.047240495681762695,
      "learning_rate": 0.00014452941176470587,
      "loss": 0.0214,
      "step": 3145
    },
    {
      "epoch": 28.09,
      "grad_norm": 0.04805517569184303,
      "learning_rate": 0.0001444705882352941,
      "loss": 0.0225,
      "step": 3146
    },
    {
      "epoch": 28.1,
      "grad_norm": 0.049226801842451096,
      "learning_rate": 0.00014441176470588235,
      "loss": 0.0204,
      "step": 3147
    },
    {
      "epoch": 28.11,
      "grad_norm": 0.06287840008735657,
      "learning_rate": 0.00014435294117647057,
      "loss": 0.0222,
      "step": 3148
    },
    {
      "epoch": 28.12,
      "grad_norm": 0.06693531572818756,
      "learning_rate": 0.00014429411764705881,
      "loss": 0.0222,
      "step": 3149
    },
    {
      "epoch": 28.12,
      "grad_norm": 0.04513375088572502,
      "learning_rate": 0.00014423529411764703,
      "loss": 0.0194,
      "step": 3150
    },
    {
      "epoch": 28.13,
      "grad_norm": 0.04871902987360954,
      "learning_rate": 0.00014417647058823527,
      "loss": 0.0219,
      "step": 3151
    },
    {
      "epoch": 28.14,
      "grad_norm": 0.05776379629969597,
      "learning_rate": 0.00014411764705882352,
      "loss": 0.0233,
      "step": 3152
    },
    {
      "epoch": 28.15,
      "grad_norm": 0.055099520832300186,
      "learning_rate": 0.00014405882352941176,
      "loss": 0.0243,
      "step": 3153
    },
    {
      "epoch": 28.16,
      "grad_norm": 0.055144671350717545,
      "learning_rate": 0.00014399999999999998,
      "loss": 0.0174,
      "step": 3154
    },
    {
      "epoch": 28.17,
      "grad_norm": 0.08644551038742065,
      "learning_rate": 0.00014394117647058822,
      "loss": 0.0229,
      "step": 3155
    },
    {
      "epoch": 28.18,
      "grad_norm": 0.05233171582221985,
      "learning_rate": 0.00014388235294117646,
      "loss": 0.0221,
      "step": 3156
    },
    {
      "epoch": 28.19,
      "grad_norm": 0.05046974867582321,
      "learning_rate": 0.0001438235294117647,
      "loss": 0.021,
      "step": 3157
    },
    {
      "epoch": 28.2,
      "grad_norm": 0.051787909120321274,
      "learning_rate": 0.00014376470588235292,
      "loss": 0.0208,
      "step": 3158
    },
    {
      "epoch": 28.21,
      "grad_norm": 0.052158426493406296,
      "learning_rate": 0.00014370588235294117,
      "loss": 0.0182,
      "step": 3159
    },
    {
      "epoch": 28.21,
      "grad_norm": 0.05279925838112831,
      "learning_rate": 0.0001436470588235294,
      "loss": 0.0183,
      "step": 3160
    },
    {
      "epoch": 28.22,
      "grad_norm": 0.1067124456167221,
      "learning_rate": 0.00014358823529411763,
      "loss": 0.0253,
      "step": 3161
    },
    {
      "epoch": 28.23,
      "grad_norm": 0.05293858423829079,
      "learning_rate": 0.00014352941176470587,
      "loss": 0.023,
      "step": 3162
    },
    {
      "epoch": 28.24,
      "grad_norm": 0.0508616678416729,
      "learning_rate": 0.00014347058823529409,
      "loss": 0.0244,
      "step": 3163
    },
    {
      "epoch": 28.25,
      "grad_norm": 0.050752900540828705,
      "learning_rate": 0.00014341176470588233,
      "loss": 0.0204,
      "step": 3164
    },
    {
      "epoch": 28.26,
      "grad_norm": 0.04990452527999878,
      "learning_rate": 0.00014335294117647057,
      "loss": 0.0204,
      "step": 3165
    },
    {
      "epoch": 28.27,
      "grad_norm": 0.053398337215185165,
      "learning_rate": 0.00014329411764705882,
      "loss": 0.0208,
      "step": 3166
    },
    {
      "epoch": 28.28,
      "grad_norm": 0.07779485732316971,
      "learning_rate": 0.00014323529411764706,
      "loss": 0.0235,
      "step": 3167
    },
    {
      "epoch": 28.29,
      "grad_norm": 0.08986517041921616,
      "learning_rate": 0.00014317647058823528,
      "loss": 0.0256,
      "step": 3168
    },
    {
      "epoch": 28.29,
      "grad_norm": 0.044942453503608704,
      "learning_rate": 0.00014311764705882352,
      "loss": 0.0213,
      "step": 3169
    },
    {
      "epoch": 28.3,
      "grad_norm": 0.05150241032242775,
      "learning_rate": 0.00014305882352941176,
      "loss": 0.0204,
      "step": 3170
    },
    {
      "epoch": 28.31,
      "grad_norm": 0.0570935495197773,
      "learning_rate": 0.00014299999999999998,
      "loss": 0.022,
      "step": 3171
    },
    {
      "epoch": 28.32,
      "grad_norm": 0.05327785015106201,
      "learning_rate": 0.00014294117647058822,
      "loss": 0.0188,
      "step": 3172
    },
    {
      "epoch": 28.33,
      "grad_norm": 0.07646356523036957,
      "learning_rate": 0.00014288235294117647,
      "loss": 0.0222,
      "step": 3173
    },
    {
      "epoch": 28.34,
      "grad_norm": 0.09917483478784561,
      "learning_rate": 0.00014282352941176468,
      "loss": 0.0308,
      "step": 3174
    },
    {
      "epoch": 28.35,
      "grad_norm": 0.04881855845451355,
      "learning_rate": 0.00014276470588235293,
      "loss": 0.0191,
      "step": 3175
    },
    {
      "epoch": 28.36,
      "grad_norm": 0.0523366816341877,
      "learning_rate": 0.00014270588235294117,
      "loss": 0.0247,
      "step": 3176
    },
    {
      "epoch": 28.37,
      "grad_norm": 0.058703478425741196,
      "learning_rate": 0.00014264705882352939,
      "loss": 0.023,
      "step": 3177
    },
    {
      "epoch": 28.38,
      "grad_norm": 0.04521499201655388,
      "learning_rate": 0.00014258823529411763,
      "loss": 0.0193,
      "step": 3178
    },
    {
      "epoch": 28.38,
      "grad_norm": 0.054483313113451004,
      "learning_rate": 0.00014252941176470587,
      "loss": 0.0209,
      "step": 3179
    },
    {
      "epoch": 28.39,
      "grad_norm": 0.08454073220491409,
      "learning_rate": 0.00014247058823529412,
      "loss": 0.0274,
      "step": 3180
    },
    {
      "epoch": 28.4,
      "grad_norm": 0.04647323489189148,
      "learning_rate": 0.00014241176470588236,
      "loss": 0.0236,
      "step": 3181
    },
    {
      "epoch": 28.41,
      "grad_norm": 0.04771995544433594,
      "learning_rate": 0.00014235294117647058,
      "loss": 0.0222,
      "step": 3182
    },
    {
      "epoch": 28.42,
      "grad_norm": 0.050001874566078186,
      "learning_rate": 0.00014229411764705882,
      "loss": 0.0243,
      "step": 3183
    },
    {
      "epoch": 28.43,
      "grad_norm": 0.051152534782886505,
      "learning_rate": 0.00014223529411764704,
      "loss": 0.0229,
      "step": 3184
    },
    {
      "epoch": 28.44,
      "grad_norm": 0.06005259230732918,
      "learning_rate": 0.00014217647058823528,
      "loss": 0.0234,
      "step": 3185
    },
    {
      "epoch": 28.45,
      "grad_norm": 0.11991535872220993,
      "learning_rate": 0.00014211764705882352,
      "loss": 0.029,
      "step": 3186
    },
    {
      "epoch": 28.46,
      "grad_norm": 0.0499068982899189,
      "learning_rate": 0.00014205882352941174,
      "loss": 0.0209,
      "step": 3187
    },
    {
      "epoch": 28.46,
      "grad_norm": 0.049948178231716156,
      "learning_rate": 0.00014199999999999998,
      "loss": 0.0231,
      "step": 3188
    },
    {
      "epoch": 28.47,
      "grad_norm": 0.0567111000418663,
      "learning_rate": 0.00014194117647058823,
      "loss": 0.0238,
      "step": 3189
    },
    {
      "epoch": 28.48,
      "grad_norm": 0.04997025057673454,
      "learning_rate": 0.00014188235294117647,
      "loss": 0.0232,
      "step": 3190
    },
    {
      "epoch": 28.49,
      "grad_norm": 0.062033213675022125,
      "learning_rate": 0.00014182352941176469,
      "loss": 0.0215,
      "step": 3191
    },
    {
      "epoch": 28.5,
      "grad_norm": 0.06742071360349655,
      "learning_rate": 0.00014176470588235293,
      "loss": 0.0227,
      "step": 3192
    },
    {
      "epoch": 28.51,
      "grad_norm": 0.057490911334753036,
      "learning_rate": 0.00014170588235294117,
      "loss": 0.023,
      "step": 3193
    },
    {
      "epoch": 28.52,
      "grad_norm": 0.05032145604491234,
      "learning_rate": 0.00014164705882352942,
      "loss": 0.0238,
      "step": 3194
    },
    {
      "epoch": 28.53,
      "grad_norm": 0.045561935752630234,
      "learning_rate": 0.00014158823529411763,
      "loss": 0.0233,
      "step": 3195
    },
    {
      "epoch": 28.54,
      "grad_norm": 0.05160699039697647,
      "learning_rate": 0.00014152941176470588,
      "loss": 0.0257,
      "step": 3196
    },
    {
      "epoch": 28.54,
      "grad_norm": 0.05225726217031479,
      "learning_rate": 0.0001414705882352941,
      "loss": 0.0179,
      "step": 3197
    },
    {
      "epoch": 28.55,
      "grad_norm": 0.06850314885377884,
      "learning_rate": 0.00014141176470588234,
      "loss": 0.0291,
      "step": 3198
    },
    {
      "epoch": 28.56,
      "grad_norm": 0.0695592612028122,
      "learning_rate": 0.00014135294117647058,
      "loss": 0.0255,
      "step": 3199
    },
    {
      "epoch": 28.57,
      "grad_norm": 0.046914249658584595,
      "learning_rate": 0.00014129411764705882,
      "loss": 0.025,
      "step": 3200
    },
    {
      "epoch": 28.57,
      "eval_cer": 0.03299094609374119,
      "eval_loss": 0.20456813275814056,
      "eval_runtime": 22.4984,
      "eval_samples_per_second": 117.431,
      "eval_steps_per_second": 1.867,
      "eval_wer": 0.10767552558508528,
      "step": 3200
    },
    {
      "epoch": 28.58,
      "grad_norm": 0.05197370797395706,
      "learning_rate": 0.00014123529411764704,
      "loss": 0.0214,
      "step": 3201
    },
    {
      "epoch": 28.59,
      "grad_norm": 0.04819939285516739,
      "learning_rate": 0.00014117647058823528,
      "loss": 0.0221,
      "step": 3202
    },
    {
      "epoch": 28.6,
      "grad_norm": 0.05303650721907616,
      "learning_rate": 0.00014111764705882353,
      "loss": 0.0229,
      "step": 3203
    },
    {
      "epoch": 28.61,
      "grad_norm": 0.05289100483059883,
      "learning_rate": 0.00014105882352941177,
      "loss": 0.0197,
      "step": 3204
    },
    {
      "epoch": 28.62,
      "grad_norm": 0.10235423594713211,
      "learning_rate": 0.00014099999999999998,
      "loss": 0.0314,
      "step": 3205
    },
    {
      "epoch": 28.62,
      "grad_norm": 0.0496661514043808,
      "learning_rate": 0.00014094117647058823,
      "loss": 0.0215,
      "step": 3206
    },
    {
      "epoch": 28.63,
      "grad_norm": 0.047629259526729584,
      "learning_rate": 0.00014088235294117644,
      "loss": 0.019,
      "step": 3207
    },
    {
      "epoch": 28.64,
      "grad_norm": 0.05727589130401611,
      "learning_rate": 0.0001408235294117647,
      "loss": 0.0231,
      "step": 3208
    },
    {
      "epoch": 28.65,
      "grad_norm": 0.05282297357916832,
      "learning_rate": 0.00014076470588235293,
      "loss": 0.0237,
      "step": 3209
    },
    {
      "epoch": 28.66,
      "grad_norm": 0.04988083615899086,
      "learning_rate": 0.00014070588235294115,
      "loss": 0.0178,
      "step": 3210
    },
    {
      "epoch": 28.67,
      "grad_norm": 0.09559240192174911,
      "learning_rate": 0.0001406470588235294,
      "loss": 0.0331,
      "step": 3211
    },
    {
      "epoch": 28.68,
      "grad_norm": 0.05182569473981857,
      "learning_rate": 0.00014058823529411763,
      "loss": 0.02,
      "step": 3212
    },
    {
      "epoch": 28.69,
      "grad_norm": 0.05532494932413101,
      "learning_rate": 0.00014052941176470588,
      "loss": 0.0198,
      "step": 3213
    },
    {
      "epoch": 28.7,
      "grad_norm": 0.053896158933639526,
      "learning_rate": 0.00014047058823529412,
      "loss": 0.0231,
      "step": 3214
    },
    {
      "epoch": 28.71,
      "grad_norm": 0.04777355492115021,
      "learning_rate": 0.00014041176470588234,
      "loss": 0.0188,
      "step": 3215
    },
    {
      "epoch": 28.71,
      "grad_norm": 0.05201899632811546,
      "learning_rate": 0.00014035294117647058,
      "loss": 0.0209,
      "step": 3216
    },
    {
      "epoch": 28.72,
      "grad_norm": 0.0713396966457367,
      "learning_rate": 0.00014029411764705882,
      "loss": 0.0261,
      "step": 3217
    },
    {
      "epoch": 28.73,
      "grad_norm": 0.08003824949264526,
      "learning_rate": 0.00014023529411764704,
      "loss": 0.0246,
      "step": 3218
    },
    {
      "epoch": 28.74,
      "grad_norm": 0.04647595062851906,
      "learning_rate": 0.00014017647058823528,
      "loss": 0.0243,
      "step": 3219
    },
    {
      "epoch": 28.75,
      "grad_norm": 0.053717732429504395,
      "learning_rate": 0.0001401176470588235,
      "loss": 0.0256,
      "step": 3220
    },
    {
      "epoch": 28.76,
      "grad_norm": 0.05227770283818245,
      "learning_rate": 0.00014005882352941174,
      "loss": 0.0222,
      "step": 3221
    },
    {
      "epoch": 28.77,
      "grad_norm": 0.0462813600897789,
      "learning_rate": 0.00014,
      "loss": 0.0194,
      "step": 3222
    },
    {
      "epoch": 28.78,
      "grad_norm": 0.060716137290000916,
      "learning_rate": 0.00013994117647058823,
      "loss": 0.0261,
      "step": 3223
    },
    {
      "epoch": 28.79,
      "grad_norm": 0.07485483586788177,
      "learning_rate": 0.00013988235294117645,
      "loss": 0.0307,
      "step": 3224
    },
    {
      "epoch": 28.79,
      "grad_norm": 0.04654227942228317,
      "learning_rate": 0.0001398235294117647,
      "loss": 0.0225,
      "step": 3225
    },
    {
      "epoch": 28.8,
      "grad_norm": 0.053567755967378616,
      "learning_rate": 0.00013976470588235293,
      "loss": 0.0209,
      "step": 3226
    },
    {
      "epoch": 28.81,
      "grad_norm": 0.048128023743629456,
      "learning_rate": 0.00013970588235294118,
      "loss": 0.024,
      "step": 3227
    },
    {
      "epoch": 28.82,
      "grad_norm": 0.057889826595783234,
      "learning_rate": 0.0001396470588235294,
      "loss": 0.0229,
      "step": 3228
    },
    {
      "epoch": 28.83,
      "grad_norm": 0.052900783717632294,
      "learning_rate": 0.00013958823529411764,
      "loss": 0.0196,
      "step": 3229
    },
    {
      "epoch": 28.84,
      "grad_norm": 0.0898534432053566,
      "learning_rate": 0.00013952941176470585,
      "loss": 0.0254,
      "step": 3230
    },
    {
      "epoch": 28.85,
      "grad_norm": 0.05033555254340172,
      "learning_rate": 0.0001394705882352941,
      "loss": 0.0216,
      "step": 3231
    },
    {
      "epoch": 28.86,
      "grad_norm": 0.05251815915107727,
      "learning_rate": 0.00013941176470588234,
      "loss": 0.0218,
      "step": 3232
    },
    {
      "epoch": 28.87,
      "grad_norm": 0.0519377738237381,
      "learning_rate": 0.00013935294117647058,
      "loss": 0.0218,
      "step": 3233
    },
    {
      "epoch": 28.88,
      "grad_norm": 0.0559382326900959,
      "learning_rate": 0.0001392941176470588,
      "loss": 0.0266,
      "step": 3234
    },
    {
      "epoch": 28.88,
      "grad_norm": 0.05021519958972931,
      "learning_rate": 0.00013923529411764704,
      "loss": 0.0203,
      "step": 3235
    },
    {
      "epoch": 28.89,
      "grad_norm": 0.123111791908741,
      "learning_rate": 0.0001391764705882353,
      "loss": 0.0308,
      "step": 3236
    },
    {
      "epoch": 28.9,
      "grad_norm": 0.04679188132286072,
      "learning_rate": 0.00013911764705882353,
      "loss": 0.0211,
      "step": 3237
    },
    {
      "epoch": 28.91,
      "grad_norm": 0.0480525977909565,
      "learning_rate": 0.00013905882352941175,
      "loss": 0.0225,
      "step": 3238
    },
    {
      "epoch": 28.92,
      "grad_norm": 0.04935698211193085,
      "learning_rate": 0.000139,
      "loss": 0.0196,
      "step": 3239
    },
    {
      "epoch": 28.93,
      "grad_norm": 0.051021214574575424,
      "learning_rate": 0.00013894117647058823,
      "loss": 0.0189,
      "step": 3240
    },
    {
      "epoch": 28.94,
      "grad_norm": 0.05619300529360771,
      "learning_rate": 0.00013888235294117645,
      "loss": 0.0227,
      "step": 3241
    },
    {
      "epoch": 28.95,
      "grad_norm": 0.08613027632236481,
      "learning_rate": 0.0001388235294117647,
      "loss": 0.0266,
      "step": 3242
    },
    {
      "epoch": 28.96,
      "grad_norm": 0.12794989347457886,
      "learning_rate": 0.0001387647058823529,
      "loss": 0.0283,
      "step": 3243
    },
    {
      "epoch": 28.96,
      "grad_norm": 0.05472174286842346,
      "learning_rate": 0.00013870588235294115,
      "loss": 0.024,
      "step": 3244
    },
    {
      "epoch": 28.97,
      "grad_norm": 0.05780802294611931,
      "learning_rate": 0.0001386470588235294,
      "loss": 0.0243,
      "step": 3245
    },
    {
      "epoch": 28.98,
      "grad_norm": 0.056403495371341705,
      "learning_rate": 0.00013858823529411764,
      "loss": 0.0217,
      "step": 3246
    },
    {
      "epoch": 28.99,
      "grad_norm": 0.0666576698422432,
      "learning_rate": 0.00013852941176470588,
      "loss": 0.0237,
      "step": 3247
    },
    {
      "epoch": 29.0,
      "grad_norm": 0.07459690421819687,
      "learning_rate": 0.0001384705882352941,
      "loss": 0.0266,
      "step": 3248
    },
    {
      "epoch": 29.01,
      "grad_norm": 0.045693084597587585,
      "learning_rate": 0.00013841176470588234,
      "loss": 0.0212,
      "step": 3249
    },
    {
      "epoch": 29.02,
      "grad_norm": 0.04682189226150513,
      "learning_rate": 0.0001383529411764706,
      "loss": 0.0211,
      "step": 3250
    },
    {
      "epoch": 29.03,
      "grad_norm": 0.047736797481775284,
      "learning_rate": 0.0001382941176470588,
      "loss": 0.0214,
      "step": 3251
    },
    {
      "epoch": 29.04,
      "grad_norm": 0.0597694106400013,
      "learning_rate": 0.00013823529411764705,
      "loss": 0.0249,
      "step": 3252
    },
    {
      "epoch": 29.04,
      "grad_norm": 0.047346778213977814,
      "learning_rate": 0.0001381764705882353,
      "loss": 0.0183,
      "step": 3253
    },
    {
      "epoch": 29.05,
      "grad_norm": 0.0749417245388031,
      "learning_rate": 0.0001381176470588235,
      "loss": 0.0281,
      "step": 3254
    },
    {
      "epoch": 29.06,
      "grad_norm": 0.06579748541116714,
      "learning_rate": 0.00013805882352941175,
      "loss": 0.0239,
      "step": 3255
    },
    {
      "epoch": 29.07,
      "grad_norm": 0.043980877846479416,
      "learning_rate": 0.000138,
      "loss": 0.0192,
      "step": 3256
    },
    {
      "epoch": 29.08,
      "grad_norm": 0.045786090195178986,
      "learning_rate": 0.0001379411764705882,
      "loss": 0.0203,
      "step": 3257
    },
    {
      "epoch": 29.09,
      "grad_norm": 0.051101308315992355,
      "learning_rate": 0.00013788235294117645,
      "loss": 0.0213,
      "step": 3258
    },
    {
      "epoch": 29.1,
      "grad_norm": 0.05994032323360443,
      "learning_rate": 0.0001378235294117647,
      "loss": 0.0212,
      "step": 3259
    },
    {
      "epoch": 29.11,
      "grad_norm": 0.0744439885020256,
      "learning_rate": 0.00013776470588235294,
      "loss": 0.0262,
      "step": 3260
    },
    {
      "epoch": 29.12,
      "grad_norm": 0.09793171286582947,
      "learning_rate": 0.00013770588235294118,
      "loss": 0.0345,
      "step": 3261
    },
    {
      "epoch": 29.12,
      "grad_norm": 0.04691844806075096,
      "learning_rate": 0.0001376470588235294,
      "loss": 0.0216,
      "step": 3262
    },
    {
      "epoch": 29.13,
      "grad_norm": 0.0566234216094017,
      "learning_rate": 0.00013758823529411764,
      "loss": 0.0238,
      "step": 3263
    },
    {
      "epoch": 29.14,
      "grad_norm": 0.04758448526263237,
      "learning_rate": 0.00013752941176470586,
      "loss": 0.0194,
      "step": 3264
    },
    {
      "epoch": 29.15,
      "grad_norm": 0.04558186233043671,
      "learning_rate": 0.0001374705882352941,
      "loss": 0.0187,
      "step": 3265
    },
    {
      "epoch": 29.16,
      "grad_norm": 0.053277499973773956,
      "learning_rate": 0.00013741176470588235,
      "loss": 0.0185,
      "step": 3266
    },
    {
      "epoch": 29.17,
      "grad_norm": 0.09081804752349854,
      "learning_rate": 0.00013735294117647056,
      "loss": 0.0283,
      "step": 3267
    },
    {
      "epoch": 29.18,
      "grad_norm": 0.04838769510388374,
      "learning_rate": 0.0001372941176470588,
      "loss": 0.0222,
      "step": 3268
    },
    {
      "epoch": 29.19,
      "grad_norm": 0.056981537491083145,
      "learning_rate": 0.00013723529411764705,
      "loss": 0.0234,
      "step": 3269
    },
    {
      "epoch": 29.2,
      "grad_norm": 0.05015343055129051,
      "learning_rate": 0.0001371764705882353,
      "loss": 0.0222,
      "step": 3270
    },
    {
      "epoch": 29.21,
      "grad_norm": 0.053255997598171234,
      "learning_rate": 0.0001371176470588235,
      "loss": 0.0241,
      "step": 3271
    },
    {
      "epoch": 29.21,
      "grad_norm": 0.057104263454675674,
      "learning_rate": 0.00013705882352941175,
      "loss": 0.0201,
      "step": 3272
    },
    {
      "epoch": 29.22,
      "grad_norm": 0.07775372266769409,
      "learning_rate": 0.000137,
      "loss": 0.0229,
      "step": 3273
    },
    {
      "epoch": 29.23,
      "grad_norm": 0.05271748825907707,
      "learning_rate": 0.00013694117647058824,
      "loss": 0.0217,
      "step": 3274
    },
    {
      "epoch": 29.24,
      "grad_norm": 0.04710380360484123,
      "learning_rate": 0.00013688235294117646,
      "loss": 0.0215,
      "step": 3275
    },
    {
      "epoch": 29.25,
      "grad_norm": 0.05564838647842407,
      "learning_rate": 0.0001368235294117647,
      "loss": 0.02,
      "step": 3276
    },
    {
      "epoch": 29.26,
      "grad_norm": 0.05574845150113106,
      "learning_rate": 0.00013676470588235292,
      "loss": 0.0219,
      "step": 3277
    },
    {
      "epoch": 29.27,
      "grad_norm": 0.06458115577697754,
      "learning_rate": 0.00013670588235294116,
      "loss": 0.0229,
      "step": 3278
    },
    {
      "epoch": 29.28,
      "grad_norm": 0.08612285554409027,
      "learning_rate": 0.0001366470588235294,
      "loss": 0.0282,
      "step": 3279
    },
    {
      "epoch": 29.29,
      "grad_norm": 0.08460797369480133,
      "learning_rate": 0.00013658823529411765,
      "loss": 0.029,
      "step": 3280
    },
    {
      "epoch": 29.29,
      "grad_norm": 0.05518344044685364,
      "learning_rate": 0.00013652941176470586,
      "loss": 0.024,
      "step": 3281
    },
    {
      "epoch": 29.3,
      "grad_norm": 0.048608794808387756,
      "learning_rate": 0.0001364705882352941,
      "loss": 0.0203,
      "step": 3282
    },
    {
      "epoch": 29.31,
      "grad_norm": 0.049384672194719315,
      "learning_rate": 0.00013641176470588235,
      "loss": 0.0214,
      "step": 3283
    },
    {
      "epoch": 29.32,
      "grad_norm": 0.050028521567583084,
      "learning_rate": 0.0001363529411764706,
      "loss": 0.022,
      "step": 3284
    },
    {
      "epoch": 29.33,
      "grad_norm": 0.06495417654514313,
      "learning_rate": 0.0001362941176470588,
      "loss": 0.022,
      "step": 3285
    },
    {
      "epoch": 29.34,
      "grad_norm": 0.08429013937711716,
      "learning_rate": 0.00013623529411764705,
      "loss": 0.0259,
      "step": 3286
    },
    {
      "epoch": 29.35,
      "grad_norm": 0.046002425253391266,
      "learning_rate": 0.00013617647058823527,
      "loss": 0.0206,
      "step": 3287
    },
    {
      "epoch": 29.36,
      "grad_norm": 0.04369529336690903,
      "learning_rate": 0.0001361176470588235,
      "loss": 0.0181,
      "step": 3288
    },
    {
      "epoch": 29.37,
      "grad_norm": 0.05745384097099304,
      "learning_rate": 0.00013605882352941176,
      "loss": 0.0219,
      "step": 3289
    },
    {
      "epoch": 29.38,
      "grad_norm": 0.05233566835522652,
      "learning_rate": 0.00013599999999999997,
      "loss": 0.0225,
      "step": 3290
    },
    {
      "epoch": 29.38,
      "grad_norm": 0.0635610893368721,
      "learning_rate": 0.00013594117647058822,
      "loss": 0.0234,
      "step": 3291
    },
    {
      "epoch": 29.39,
      "grad_norm": 0.10268981754779816,
      "learning_rate": 0.00013588235294117646,
      "loss": 0.0322,
      "step": 3292
    },
    {
      "epoch": 29.4,
      "grad_norm": 0.055386073887348175,
      "learning_rate": 0.0001358235294117647,
      "loss": 0.0246,
      "step": 3293
    },
    {
      "epoch": 29.41,
      "grad_norm": 0.04810195416212082,
      "learning_rate": 0.00013576470588235295,
      "loss": 0.0222,
      "step": 3294
    },
    {
      "epoch": 29.42,
      "grad_norm": 0.04568944871425629,
      "learning_rate": 0.00013570588235294116,
      "loss": 0.0213,
      "step": 3295
    },
    {
      "epoch": 29.43,
      "grad_norm": 0.05103462189435959,
      "learning_rate": 0.0001356470588235294,
      "loss": 0.0198,
      "step": 3296
    },
    {
      "epoch": 29.44,
      "grad_norm": 0.059654537588357925,
      "learning_rate": 0.00013558823529411765,
      "loss": 0.0252,
      "step": 3297
    },
    {
      "epoch": 29.45,
      "grad_norm": 0.09684297442436218,
      "learning_rate": 0.00013552941176470587,
      "loss": 0.026,
      "step": 3298
    },
    {
      "epoch": 29.46,
      "grad_norm": 0.046199869364500046,
      "learning_rate": 0.0001354705882352941,
      "loss": 0.0195,
      "step": 3299
    },
    {
      "epoch": 29.46,
      "grad_norm": 0.05403624847531319,
      "learning_rate": 0.00013541176470588232,
      "loss": 0.0242,
      "step": 3300
    },
    {
      "epoch": 29.46,
      "eval_cer": 0.03347577544508462,
      "eval_loss": 0.20549054443836212,
      "eval_runtime": 22.6841,
      "eval_samples_per_second": 116.47,
      "eval_steps_per_second": 1.852,
      "eval_wer": 0.10715985719952399,
      "step": 3300
    },
    {
      "epoch": 29.47,
      "grad_norm": 0.04708437621593475,
      "learning_rate": 0.00013535294117647057,
      "loss": 0.0222,
      "step": 3301
    },
    {
      "epoch": 29.48,
      "grad_norm": 0.060763850808143616,
      "learning_rate": 0.0001352941176470588,
      "loss": 0.0237,
      "step": 3302
    },
    {
      "epoch": 29.49,
      "grad_norm": 0.05629892274737358,
      "learning_rate": 0.00013523529411764706,
      "loss": 0.0214,
      "step": 3303
    },
    {
      "epoch": 29.5,
      "grad_norm": 0.07087591290473938,
      "learning_rate": 0.00013517647058823527,
      "loss": 0.0256,
      "step": 3304
    },
    {
      "epoch": 29.51,
      "grad_norm": 0.06636977940797806,
      "learning_rate": 0.00013511764705882351,
      "loss": 0.0255,
      "step": 3305
    },
    {
      "epoch": 29.52,
      "grad_norm": 0.0395834743976593,
      "learning_rate": 0.00013505882352941176,
      "loss": 0.0185,
      "step": 3306
    },
    {
      "epoch": 29.53,
      "grad_norm": 0.05122927576303482,
      "learning_rate": 0.000135,
      "loss": 0.0215,
      "step": 3307
    },
    {
      "epoch": 29.54,
      "grad_norm": 0.04181566834449768,
      "learning_rate": 0.00013494117647058822,
      "loss": 0.0185,
      "step": 3308
    },
    {
      "epoch": 29.54,
      "grad_norm": 0.04806053638458252,
      "learning_rate": 0.00013488235294117646,
      "loss": 0.017,
      "step": 3309
    },
    {
      "epoch": 29.55,
      "grad_norm": 0.0685345008969307,
      "learning_rate": 0.0001348235294117647,
      "loss": 0.0258,
      "step": 3310
    },
    {
      "epoch": 29.56,
      "grad_norm": 0.08832594007253647,
      "learning_rate": 0.00013476470588235292,
      "loss": 0.0248,
      "step": 3311
    },
    {
      "epoch": 29.57,
      "grad_norm": 0.04239046201109886,
      "learning_rate": 0.00013470588235294116,
      "loss": 0.0213,
      "step": 3312
    },
    {
      "epoch": 29.58,
      "grad_norm": 0.04903341084718704,
      "learning_rate": 0.0001346470588235294,
      "loss": 0.0207,
      "step": 3313
    },
    {
      "epoch": 29.59,
      "grad_norm": 0.05598336458206177,
      "learning_rate": 0.00013458823529411762,
      "loss": 0.0221,
      "step": 3314
    },
    {
      "epoch": 29.6,
      "grad_norm": 0.055826783180236816,
      "learning_rate": 0.00013452941176470587,
      "loss": 0.0191,
      "step": 3315
    },
    {
      "epoch": 29.61,
      "grad_norm": 0.06717198342084885,
      "learning_rate": 0.0001344705882352941,
      "loss": 0.0222,
      "step": 3316
    },
    {
      "epoch": 29.62,
      "grad_norm": 0.10662803053855896,
      "learning_rate": 0.00013441176470588235,
      "loss": 0.0336,
      "step": 3317
    },
    {
      "epoch": 29.62,
      "grad_norm": 0.04396955668926239,
      "learning_rate": 0.00013435294117647057,
      "loss": 0.0221,
      "step": 3318
    },
    {
      "epoch": 29.63,
      "grad_norm": 0.0503300242125988,
      "learning_rate": 0.00013429411764705881,
      "loss": 0.023,
      "step": 3319
    },
    {
      "epoch": 29.64,
      "grad_norm": 0.05000848323106766,
      "learning_rate": 0.00013423529411764706,
      "loss": 0.0206,
      "step": 3320
    },
    {
      "epoch": 29.65,
      "grad_norm": 0.046584829688072205,
      "learning_rate": 0.00013417647058823527,
      "loss": 0.021,
      "step": 3321
    },
    {
      "epoch": 29.66,
      "grad_norm": 0.05395794287323952,
      "learning_rate": 0.00013411764705882352,
      "loss": 0.0231,
      "step": 3322
    },
    {
      "epoch": 29.67,
      "grad_norm": 0.134524866938591,
      "learning_rate": 0.00013405882352941173,
      "loss": 0.0255,
      "step": 3323
    },
    {
      "epoch": 29.68,
      "grad_norm": 0.04994124546647072,
      "learning_rate": 0.00013399999999999998,
      "loss": 0.0194,
      "step": 3324
    },
    {
      "epoch": 29.69,
      "grad_norm": 0.04826788231730461,
      "learning_rate": 0.00013394117647058822,
      "loss": 0.0222,
      "step": 3325
    },
    {
      "epoch": 29.7,
      "grad_norm": 0.04609817638993263,
      "learning_rate": 0.00013388235294117646,
      "loss": 0.021,
      "step": 3326
    },
    {
      "epoch": 29.71,
      "grad_norm": 0.05121495947241783,
      "learning_rate": 0.0001338235294117647,
      "loss": 0.0178,
      "step": 3327
    },
    {
      "epoch": 29.71,
      "grad_norm": 0.057989224791526794,
      "learning_rate": 0.00013376470588235292,
      "loss": 0.0205,
      "step": 3328
    },
    {
      "epoch": 29.72,
      "grad_norm": 0.08133722841739655,
      "learning_rate": 0.00013370588235294117,
      "loss": 0.0216,
      "step": 3329
    },
    {
      "epoch": 29.73,
      "grad_norm": 0.09465450793504715,
      "learning_rate": 0.0001336470588235294,
      "loss": 0.0279,
      "step": 3330
    },
    {
      "epoch": 29.74,
      "grad_norm": 0.05890243127942085,
      "learning_rate": 0.00013358823529411763,
      "loss": 0.0235,
      "step": 3331
    },
    {
      "epoch": 29.75,
      "grad_norm": 0.04863334447145462,
      "learning_rate": 0.00013352941176470587,
      "loss": 0.0187,
      "step": 3332
    },
    {
      "epoch": 29.76,
      "grad_norm": 0.04999236762523651,
      "learning_rate": 0.00013347058823529411,
      "loss": 0.0226,
      "step": 3333
    },
    {
      "epoch": 29.77,
      "grad_norm": 0.055948127061128616,
      "learning_rate": 0.00013341176470588233,
      "loss": 0.0221,
      "step": 3334
    },
    {
      "epoch": 29.78,
      "grad_norm": 0.06907929480075836,
      "learning_rate": 0.00013335294117647057,
      "loss": 0.0208,
      "step": 3335
    },
    {
      "epoch": 29.79,
      "grad_norm": 0.0889093205332756,
      "learning_rate": 0.00013329411764705882,
      "loss": 0.0271,
      "step": 3336
    },
    {
      "epoch": 29.79,
      "grad_norm": 0.04269901290535927,
      "learning_rate": 0.00013323529411764703,
      "loss": 0.0189,
      "step": 3337
    },
    {
      "epoch": 29.8,
      "grad_norm": 0.050411589443683624,
      "learning_rate": 0.00013317647058823528,
      "loss": 0.0205,
      "step": 3338
    },
    {
      "epoch": 29.81,
      "grad_norm": 0.05397701635956764,
      "learning_rate": 0.00013311764705882352,
      "loss": 0.0206,
      "step": 3339
    },
    {
      "epoch": 29.82,
      "grad_norm": 0.057393450289964676,
      "learning_rate": 0.00013305882352941176,
      "loss": 0.0192,
      "step": 3340
    },
    {
      "epoch": 29.83,
      "grad_norm": 0.06307059526443481,
      "learning_rate": 0.000133,
      "loss": 0.0225,
      "step": 3341
    },
    {
      "epoch": 29.84,
      "grad_norm": 0.08708269149065018,
      "learning_rate": 0.00013294117647058822,
      "loss": 0.0258,
      "step": 3342
    },
    {
      "epoch": 29.85,
      "grad_norm": 0.049405332654714584,
      "learning_rate": 0.00013288235294117647,
      "loss": 0.0235,
      "step": 3343
    },
    {
      "epoch": 29.86,
      "grad_norm": 0.05103577673435211,
      "learning_rate": 0.00013282352941176468,
      "loss": 0.0215,
      "step": 3344
    },
    {
      "epoch": 29.87,
      "grad_norm": 0.05492822080850601,
      "learning_rate": 0.00013276470588235293,
      "loss": 0.0238,
      "step": 3345
    },
    {
      "epoch": 29.88,
      "grad_norm": 0.06334534287452698,
      "learning_rate": 0.00013270588235294117,
      "loss": 0.0228,
      "step": 3346
    },
    {
      "epoch": 29.88,
      "grad_norm": 0.05494748055934906,
      "learning_rate": 0.0001326470588235294,
      "loss": 0.0226,
      "step": 3347
    },
    {
      "epoch": 29.89,
      "grad_norm": 0.10256792604923248,
      "learning_rate": 0.00013258823529411763,
      "loss": 0.0306,
      "step": 3348
    },
    {
      "epoch": 29.9,
      "grad_norm": 0.046946603804826736,
      "learning_rate": 0.00013252941176470587,
      "loss": 0.0224,
      "step": 3349
    },
    {
      "epoch": 29.91,
      "grad_norm": 0.04373834282159805,
      "learning_rate": 0.00013247058823529412,
      "loss": 0.0189,
      "step": 3350
    },
    {
      "epoch": 29.92,
      "grad_norm": 0.052029527723789215,
      "learning_rate": 0.00013241176470588233,
      "loss": 0.0203,
      "step": 3351
    },
    {
      "epoch": 29.93,
      "grad_norm": 0.04216480255126953,
      "learning_rate": 0.00013235294117647058,
      "loss": 0.016,
      "step": 3352
    },
    {
      "epoch": 29.94,
      "grad_norm": 0.06439980864524841,
      "learning_rate": 0.00013229411764705882,
      "loss": 0.0214,
      "step": 3353
    },
    {
      "epoch": 29.95,
      "grad_norm": 0.062002141028642654,
      "learning_rate": 0.00013223529411764706,
      "loss": 0.0211,
      "step": 3354
    },
    {
      "epoch": 29.96,
      "grad_norm": 0.0867651030421257,
      "learning_rate": 0.00013217647058823528,
      "loss": 0.0248,
      "step": 3355
    },
    {
      "epoch": 29.96,
      "grad_norm": 0.05184529349207878,
      "learning_rate": 0.00013211764705882352,
      "loss": 0.0219,
      "step": 3356
    },
    {
      "epoch": 29.97,
      "grad_norm": 0.051039062440395355,
      "learning_rate": 0.00013205882352941174,
      "loss": 0.0204,
      "step": 3357
    },
    {
      "epoch": 29.98,
      "grad_norm": 0.05570162087678909,
      "learning_rate": 0.00013199999999999998,
      "loss": 0.0189,
      "step": 3358
    },
    {
      "epoch": 29.99,
      "grad_norm": 0.05899661406874657,
      "learning_rate": 0.00013194117647058823,
      "loss": 0.0224,
      "step": 3359
    },
    {
      "epoch": 30.0,
      "grad_norm": 0.084539495408535,
      "learning_rate": 0.00013188235294117647,
      "loss": 0.0236,
      "step": 3360
    },
    {
      "epoch": 30.01,
      "grad_norm": 0.0488189272582531,
      "learning_rate": 0.00013182352941176469,
      "loss": 0.0214,
      "step": 3361
    },
    {
      "epoch": 30.02,
      "grad_norm": 0.05479676276445389,
      "learning_rate": 0.00013176470588235293,
      "loss": 0.0225,
      "step": 3362
    },
    {
      "epoch": 30.03,
      "grad_norm": 0.042999375611543655,
      "learning_rate": 0.00013170588235294117,
      "loss": 0.0192,
      "step": 3363
    },
    {
      "epoch": 30.04,
      "grad_norm": 0.045781463384628296,
      "learning_rate": 0.00013164705882352942,
      "loss": 0.0194,
      "step": 3364
    },
    {
      "epoch": 30.04,
      "grad_norm": 0.1375766098499298,
      "learning_rate": 0.00013158823529411763,
      "loss": 0.0191,
      "step": 3365
    },
    {
      "epoch": 30.05,
      "grad_norm": 0.06471619009971619,
      "learning_rate": 0.00013152941176470588,
      "loss": 0.0212,
      "step": 3366
    },
    {
      "epoch": 30.06,
      "grad_norm": 0.10753370076417923,
      "learning_rate": 0.0001314705882352941,
      "loss": 0.0238,
      "step": 3367
    },
    {
      "epoch": 30.07,
      "grad_norm": 0.053320981562137604,
      "learning_rate": 0.00013141176470588234,
      "loss": 0.0221,
      "step": 3368
    },
    {
      "epoch": 30.08,
      "grad_norm": 0.051087040454149246,
      "learning_rate": 0.00013135294117647058,
      "loss": 0.0255,
      "step": 3369
    },
    {
      "epoch": 30.09,
      "grad_norm": 0.056379638612270355,
      "learning_rate": 0.0001312941176470588,
      "loss": 0.0229,
      "step": 3370
    },
    {
      "epoch": 30.1,
      "grad_norm": 0.05336599051952362,
      "learning_rate": 0.00013123529411764704,
      "loss": 0.0183,
      "step": 3371
    },
    {
      "epoch": 30.11,
      "grad_norm": 0.0666266679763794,
      "learning_rate": 0.00013117647058823528,
      "loss": 0.0199,
      "step": 3372
    },
    {
      "epoch": 30.12,
      "grad_norm": 0.10086492449045181,
      "learning_rate": 0.00013111764705882353,
      "loss": 0.0284,
      "step": 3373
    },
    {
      "epoch": 30.12,
      "grad_norm": 0.05312999337911606,
      "learning_rate": 0.00013105882352941177,
      "loss": 0.0225,
      "step": 3374
    },
    {
      "epoch": 30.13,
      "grad_norm": 0.04641437530517578,
      "learning_rate": 0.00013099999999999999,
      "loss": 0.0217,
      "step": 3375
    },
    {
      "epoch": 30.14,
      "grad_norm": 0.051450230181217194,
      "learning_rate": 0.00013094117647058823,
      "loss": 0.0248,
      "step": 3376
    },
    {
      "epoch": 30.15,
      "grad_norm": 0.04565277323126793,
      "learning_rate": 0.00013088235294117647,
      "loss": 0.0158,
      "step": 3377
    },
    {
      "epoch": 30.16,
      "grad_norm": 0.05888838320970535,
      "learning_rate": 0.0001308235294117647,
      "loss": 0.023,
      "step": 3378
    },
    {
      "epoch": 30.17,
      "grad_norm": 0.08209196478128433,
      "learning_rate": 0.00013076470588235293,
      "loss": 0.0239,
      "step": 3379
    },
    {
      "epoch": 30.18,
      "grad_norm": 0.050883445888757706,
      "learning_rate": 0.00013070588235294115,
      "loss": 0.023,
      "step": 3380
    },
    {
      "epoch": 30.19,
      "grad_norm": 0.04774251580238342,
      "learning_rate": 0.0001306470588235294,
      "loss": 0.0212,
      "step": 3381
    },
    {
      "epoch": 30.2,
      "grad_norm": 0.044998809695243835,
      "learning_rate": 0.00013058823529411764,
      "loss": 0.018,
      "step": 3382
    },
    {
      "epoch": 30.21,
      "grad_norm": 0.052356570959091187,
      "learning_rate": 0.00013052941176470588,
      "loss": 0.02,
      "step": 3383
    },
    {
      "epoch": 30.21,
      "grad_norm": 0.05536366254091263,
      "learning_rate": 0.0001304705882352941,
      "loss": 0.0182,
      "step": 3384
    },
    {
      "epoch": 30.22,
      "grad_norm": 0.09938780218362808,
      "learning_rate": 0.00013041176470588234,
      "loss": 0.0274,
      "step": 3385
    },
    {
      "epoch": 30.23,
      "grad_norm": 0.05490725114941597,
      "learning_rate": 0.00013035294117647058,
      "loss": 0.0213,
      "step": 3386
    },
    {
      "epoch": 30.24,
      "grad_norm": 0.056850604712963104,
      "learning_rate": 0.00013029411764705883,
      "loss": 0.0238,
      "step": 3387
    },
    {
      "epoch": 30.25,
      "grad_norm": 0.05505095049738884,
      "learning_rate": 0.00013023529411764704,
      "loss": 0.0226,
      "step": 3388
    },
    {
      "epoch": 30.26,
      "grad_norm": 0.05054643005132675,
      "learning_rate": 0.00013017647058823529,
      "loss": 0.0202,
      "step": 3389
    },
    {
      "epoch": 30.27,
      "grad_norm": 0.05233348160982132,
      "learning_rate": 0.00013011764705882353,
      "loss": 0.0194,
      "step": 3390
    },
    {
      "epoch": 30.28,
      "grad_norm": 0.07340871542692184,
      "learning_rate": 0.00013005882352941175,
      "loss": 0.0244,
      "step": 3391
    },
    {
      "epoch": 30.29,
      "grad_norm": 0.08080914616584778,
      "learning_rate": 0.00013,
      "loss": 0.0257,
      "step": 3392
    },
    {
      "epoch": 30.29,
      "grad_norm": 0.04434896633028984,
      "learning_rate": 0.00012994117647058823,
      "loss": 0.0215,
      "step": 3393
    },
    {
      "epoch": 30.3,
      "grad_norm": 0.0490557886660099,
      "learning_rate": 0.00012988235294117645,
      "loss": 0.0214,
      "step": 3394
    },
    {
      "epoch": 30.31,
      "grad_norm": 0.05217267945408821,
      "learning_rate": 0.0001298235294117647,
      "loss": 0.0225,
      "step": 3395
    },
    {
      "epoch": 30.32,
      "grad_norm": 0.04906807467341423,
      "learning_rate": 0.00012976470588235294,
      "loss": 0.02,
      "step": 3396
    },
    {
      "epoch": 30.33,
      "grad_norm": 0.06618456542491913,
      "learning_rate": 0.00012970588235294118,
      "loss": 0.0227,
      "step": 3397
    },
    {
      "epoch": 30.34,
      "grad_norm": 0.07943788915872574,
      "learning_rate": 0.0001296470588235294,
      "loss": 0.0219,
      "step": 3398
    },
    {
      "epoch": 30.35,
      "grad_norm": 0.044673431664705276,
      "learning_rate": 0.00012958823529411764,
      "loss": 0.0216,
      "step": 3399
    },
    {
      "epoch": 30.36,
      "grad_norm": 0.046142809092998505,
      "learning_rate": 0.00012952941176470588,
      "loss": 0.0187,
      "step": 3400
    },
    {
      "epoch": 30.36,
      "eval_cer": 0.030735925854934548,
      "eval_loss": 0.19797120988368988,
      "eval_runtime": 22.4052,
      "eval_samples_per_second": 117.919,
      "eval_steps_per_second": 1.875,
      "eval_wer": 0.10212217374057914,
      "step": 3400
    },
    {
      "epoch": 30.37,
      "grad_norm": 0.047959551215171814,
      "learning_rate": 0.0001294705882352941,
      "loss": 0.0207,
      "step": 3401
    },
    {
      "epoch": 30.38,
      "grad_norm": 0.05231958255171776,
      "learning_rate": 0.00012941176470588234,
      "loss": 0.0206,
      "step": 3402
    },
    {
      "epoch": 30.38,
      "grad_norm": 0.053719572722911835,
      "learning_rate": 0.00012935294117647056,
      "loss": 0.0198,
      "step": 3403
    },
    {
      "epoch": 30.39,
      "grad_norm": 0.08503369241952896,
      "learning_rate": 0.0001292941176470588,
      "loss": 0.0281,
      "step": 3404
    },
    {
      "epoch": 30.4,
      "grad_norm": 0.04950686916708946,
      "learning_rate": 0.00012923529411764704,
      "loss": 0.0206,
      "step": 3405
    },
    {
      "epoch": 30.41,
      "grad_norm": 0.045385610312223434,
      "learning_rate": 0.0001291764705882353,
      "loss": 0.0218,
      "step": 3406
    },
    {
      "epoch": 30.42,
      "grad_norm": 0.04846157878637314,
      "learning_rate": 0.00012911764705882353,
      "loss": 0.0219,
      "step": 3407
    },
    {
      "epoch": 30.43,
      "grad_norm": 0.04513081908226013,
      "learning_rate": 0.00012905882352941175,
      "loss": 0.0192,
      "step": 3408
    },
    {
      "epoch": 30.44,
      "grad_norm": 0.056301578879356384,
      "learning_rate": 0.000129,
      "loss": 0.0208,
      "step": 3409
    },
    {
      "epoch": 30.45,
      "grad_norm": 0.11125416308641434,
      "learning_rate": 0.00012894117647058823,
      "loss": 0.0276,
      "step": 3410
    },
    {
      "epoch": 30.46,
      "grad_norm": 0.041852883994579315,
      "learning_rate": 0.00012888235294117645,
      "loss": 0.0199,
      "step": 3411
    },
    {
      "epoch": 30.46,
      "grad_norm": 0.044221457093954086,
      "learning_rate": 0.0001288235294117647,
      "loss": 0.0198,
      "step": 3412
    },
    {
      "epoch": 30.47,
      "grad_norm": 0.050722718238830566,
      "learning_rate": 0.00012876470588235294,
      "loss": 0.0213,
      "step": 3413
    },
    {
      "epoch": 30.48,
      "grad_norm": 0.054453521966934204,
      "learning_rate": 0.00012870588235294115,
      "loss": 0.0222,
      "step": 3414
    },
    {
      "epoch": 30.49,
      "grad_norm": 0.05060184746980667,
      "learning_rate": 0.0001286470588235294,
      "loss": 0.0189,
      "step": 3415
    },
    {
      "epoch": 30.5,
      "grad_norm": 0.06100284308195114,
      "learning_rate": 0.00012858823529411764,
      "loss": 0.0255,
      "step": 3416
    },
    {
      "epoch": 30.51,
      "grad_norm": 0.07308360189199448,
      "learning_rate": 0.00012852941176470586,
      "loss": 0.0293,
      "step": 3417
    },
    {
      "epoch": 30.52,
      "grad_norm": 0.04617452993988991,
      "learning_rate": 0.0001284705882352941,
      "loss": 0.0208,
      "step": 3418
    },
    {
      "epoch": 30.53,
      "grad_norm": 0.04911469668149948,
      "learning_rate": 0.00012841176470588234,
      "loss": 0.0226,
      "step": 3419
    },
    {
      "epoch": 30.54,
      "grad_norm": 0.0460105799138546,
      "learning_rate": 0.0001283529411764706,
      "loss": 0.0191,
      "step": 3420
    },
    {
      "epoch": 30.54,
      "grad_norm": 0.047294821590185165,
      "learning_rate": 0.00012829411764705883,
      "loss": 0.0182,
      "step": 3421
    },
    {
      "epoch": 30.55,
      "grad_norm": 0.07154631614685059,
      "learning_rate": 0.00012823529411764705,
      "loss": 0.0246,
      "step": 3422
    },
    {
      "epoch": 30.56,
      "grad_norm": 0.08403802663087845,
      "learning_rate": 0.0001281764705882353,
      "loss": 0.0279,
      "step": 3423
    },
    {
      "epoch": 30.57,
      "grad_norm": 0.046277470886707306,
      "learning_rate": 0.0001281176470588235,
      "loss": 0.0184,
      "step": 3424
    },
    {
      "epoch": 30.58,
      "grad_norm": 0.04642440378665924,
      "learning_rate": 0.00012805882352941175,
      "loss": 0.0199,
      "step": 3425
    },
    {
      "epoch": 30.59,
      "grad_norm": 0.055028121918439865,
      "learning_rate": 0.000128,
      "loss": 0.0212,
      "step": 3426
    },
    {
      "epoch": 30.6,
      "grad_norm": 0.05246156454086304,
      "learning_rate": 0.0001279411764705882,
      "loss": 0.0198,
      "step": 3427
    },
    {
      "epoch": 30.61,
      "grad_norm": 0.05042044073343277,
      "learning_rate": 0.00012788235294117645,
      "loss": 0.0172,
      "step": 3428
    },
    {
      "epoch": 30.62,
      "grad_norm": 0.08460158109664917,
      "learning_rate": 0.0001278235294117647,
      "loss": 0.0302,
      "step": 3429
    },
    {
      "epoch": 30.62,
      "grad_norm": 0.05066191777586937,
      "learning_rate": 0.00012776470588235294,
      "loss": 0.0195,
      "step": 3430
    },
    {
      "epoch": 30.63,
      "grad_norm": 0.054355815052986145,
      "learning_rate": 0.00012770588235294116,
      "loss": 0.0215,
      "step": 3431
    },
    {
      "epoch": 30.64,
      "grad_norm": 0.050329890102148056,
      "learning_rate": 0.0001276470588235294,
      "loss": 0.0176,
      "step": 3432
    },
    {
      "epoch": 30.65,
      "grad_norm": 0.05507971718907356,
      "learning_rate": 0.00012758823529411764,
      "loss": 0.0204,
      "step": 3433
    },
    {
      "epoch": 30.66,
      "grad_norm": 0.056048277765512466,
      "learning_rate": 0.0001275294117647059,
      "loss": 0.0214,
      "step": 3434
    },
    {
      "epoch": 30.67,
      "grad_norm": 0.09453506767749786,
      "learning_rate": 0.0001274705882352941,
      "loss": 0.0278,
      "step": 3435
    },
    {
      "epoch": 30.68,
      "grad_norm": 0.04392372444272041,
      "learning_rate": 0.00012741176470588235,
      "loss": 0.0206,
      "step": 3436
    },
    {
      "epoch": 30.69,
      "grad_norm": 0.04168267548084259,
      "learning_rate": 0.00012735294117647056,
      "loss": 0.0207,
      "step": 3437
    },
    {
      "epoch": 30.7,
      "grad_norm": 0.04299396649003029,
      "learning_rate": 0.0001272941176470588,
      "loss": 0.019,
      "step": 3438
    },
    {
      "epoch": 30.71,
      "grad_norm": 0.04762311652302742,
      "learning_rate": 0.00012723529411764705,
      "loss": 0.02,
      "step": 3439
    },
    {
      "epoch": 30.71,
      "grad_norm": 0.052612196654081345,
      "learning_rate": 0.0001271764705882353,
      "loss": 0.0212,
      "step": 3440
    },
    {
      "epoch": 30.72,
      "grad_norm": 0.06647343188524246,
      "learning_rate": 0.0001271176470588235,
      "loss": 0.0225,
      "step": 3441
    },
    {
      "epoch": 30.73,
      "grad_norm": 0.12204380333423615,
      "learning_rate": 0.00012705882352941175,
      "loss": 0.0249,
      "step": 3442
    },
    {
      "epoch": 30.74,
      "grad_norm": 0.04954591393470764,
      "learning_rate": 0.000127,
      "loss": 0.0179,
      "step": 3443
    },
    {
      "epoch": 30.75,
      "grad_norm": 0.05146835371851921,
      "learning_rate": 0.00012694117647058824,
      "loss": 0.02,
      "step": 3444
    },
    {
      "epoch": 30.76,
      "grad_norm": 0.053591907024383545,
      "learning_rate": 0.00012688235294117646,
      "loss": 0.021,
      "step": 3445
    },
    {
      "epoch": 30.77,
      "grad_norm": 0.060286153107881546,
      "learning_rate": 0.0001268235294117647,
      "loss": 0.0202,
      "step": 3446
    },
    {
      "epoch": 30.78,
      "grad_norm": 0.055687978863716125,
      "learning_rate": 0.00012676470588235292,
      "loss": 0.0178,
      "step": 3447
    },
    {
      "epoch": 30.79,
      "grad_norm": 0.09846653789281845,
      "learning_rate": 0.00012670588235294116,
      "loss": 0.0261,
      "step": 3448
    },
    {
      "epoch": 30.79,
      "grad_norm": 0.057401638478040695,
      "learning_rate": 0.0001266470588235294,
      "loss": 0.0251,
      "step": 3449
    },
    {
      "epoch": 30.8,
      "grad_norm": 0.050923582166433334,
      "learning_rate": 0.00012658823529411762,
      "loss": 0.0235,
      "step": 3450
    },
    {
      "epoch": 30.81,
      "grad_norm": 0.05702921375632286,
      "learning_rate": 0.00012652941176470586,
      "loss": 0.0206,
      "step": 3451
    },
    {
      "epoch": 30.82,
      "grad_norm": 0.05089607834815979,
      "learning_rate": 0.0001264705882352941,
      "loss": 0.021,
      "step": 3452
    },
    {
      "epoch": 30.83,
      "grad_norm": 0.058155208826065063,
      "learning_rate": 0.00012641176470588235,
      "loss": 0.0216,
      "step": 3453
    },
    {
      "epoch": 30.84,
      "grad_norm": 0.08319725096225739,
      "learning_rate": 0.0001263529411764706,
      "loss": 0.028,
      "step": 3454
    },
    {
      "epoch": 30.85,
      "grad_norm": 0.048014167696237564,
      "learning_rate": 0.0001262941176470588,
      "loss": 0.0204,
      "step": 3455
    },
    {
      "epoch": 30.86,
      "grad_norm": 0.04385518655180931,
      "learning_rate": 0.00012623529411764705,
      "loss": 0.0192,
      "step": 3456
    },
    {
      "epoch": 30.87,
      "grad_norm": 0.04214124009013176,
      "learning_rate": 0.0001261764705882353,
      "loss": 0.0205,
      "step": 3457
    },
    {
      "epoch": 30.88,
      "grad_norm": 0.05014554038643837,
      "learning_rate": 0.0001261176470588235,
      "loss": 0.0219,
      "step": 3458
    },
    {
      "epoch": 30.88,
      "grad_norm": 0.06376246362924576,
      "learning_rate": 0.00012605882352941176,
      "loss": 0.0248,
      "step": 3459
    },
    {
      "epoch": 30.89,
      "grad_norm": 0.09171823412179947,
      "learning_rate": 0.00012599999999999997,
      "loss": 0.0252,
      "step": 3460
    },
    {
      "epoch": 30.9,
      "grad_norm": 0.0505376011133194,
      "learning_rate": 0.00012594117647058822,
      "loss": 0.0229,
      "step": 3461
    },
    {
      "epoch": 30.91,
      "grad_norm": 0.05193379148840904,
      "learning_rate": 0.00012588235294117646,
      "loss": 0.0214,
      "step": 3462
    },
    {
      "epoch": 30.92,
      "grad_norm": 0.054054778069257736,
      "learning_rate": 0.0001258235294117647,
      "loss": 0.0243,
      "step": 3463
    },
    {
      "epoch": 30.93,
      "grad_norm": 0.04865433648228645,
      "learning_rate": 0.00012576470588235292,
      "loss": 0.0196,
      "step": 3464
    },
    {
      "epoch": 30.94,
      "grad_norm": 0.04885263368487358,
      "learning_rate": 0.00012570588235294116,
      "loss": 0.0196,
      "step": 3465
    },
    {
      "epoch": 30.95,
      "grad_norm": 0.07468660920858383,
      "learning_rate": 0.0001256470588235294,
      "loss": 0.0243,
      "step": 3466
    },
    {
      "epoch": 30.96,
      "grad_norm": 0.13972200453281403,
      "learning_rate": 0.00012558823529411765,
      "loss": 0.0323,
      "step": 3467
    },
    {
      "epoch": 30.96,
      "grad_norm": 0.04695039987564087,
      "learning_rate": 0.00012552941176470587,
      "loss": 0.0213,
      "step": 3468
    },
    {
      "epoch": 30.97,
      "grad_norm": 0.04661361500620842,
      "learning_rate": 0.0001254705882352941,
      "loss": 0.0214,
      "step": 3469
    },
    {
      "epoch": 30.98,
      "grad_norm": 0.04745853319764137,
      "learning_rate": 0.00012541176470588235,
      "loss": 0.0195,
      "step": 3470
    },
    {
      "epoch": 30.99,
      "grad_norm": 0.06471362709999084,
      "learning_rate": 0.00012535294117647057,
      "loss": 0.0188,
      "step": 3471
    },
    {
      "epoch": 31.0,
      "grad_norm": 0.06648192554712296,
      "learning_rate": 0.0001252941176470588,
      "loss": 0.0236,
      "step": 3472
    },
    {
      "epoch": 31.01,
      "grad_norm": 0.047431010752916336,
      "learning_rate": 0.00012523529411764706,
      "loss": 0.0181,
      "step": 3473
    },
    {
      "epoch": 31.02,
      "grad_norm": 0.052671611309051514,
      "learning_rate": 0.00012517647058823527,
      "loss": 0.017,
      "step": 3474
    },
    {
      "epoch": 31.03,
      "grad_norm": 0.04441322386264801,
      "learning_rate": 0.00012511764705882352,
      "loss": 0.019,
      "step": 3475
    },
    {
      "epoch": 31.04,
      "grad_norm": 0.05547754466533661,
      "learning_rate": 0.00012505882352941176,
      "loss": 0.0246,
      "step": 3476
    },
    {
      "epoch": 31.04,
      "grad_norm": 0.0603085495531559,
      "learning_rate": 0.000125,
      "loss": 0.0224,
      "step": 3477
    },
    {
      "epoch": 31.05,
      "grad_norm": 0.09202206134796143,
      "learning_rate": 0.00012494117647058822,
      "loss": 0.0293,
      "step": 3478
    },
    {
      "epoch": 31.06,
      "grad_norm": 0.09541846811771393,
      "learning_rate": 0.00012488235294117646,
      "loss": 0.0292,
      "step": 3479
    },
    {
      "epoch": 31.07,
      "grad_norm": 0.04697448015213013,
      "learning_rate": 0.0001248235294117647,
      "loss": 0.0213,
      "step": 3480
    },
    {
      "epoch": 31.08,
      "grad_norm": 0.04767817258834839,
      "learning_rate": 0.00012476470588235292,
      "loss": 0.0182,
      "step": 3481
    },
    {
      "epoch": 31.09,
      "grad_norm": 0.05568642169237137,
      "learning_rate": 0.00012470588235294117,
      "loss": 0.0232,
      "step": 3482
    },
    {
      "epoch": 31.1,
      "grad_norm": 0.052640173584222794,
      "learning_rate": 0.00012464705882352938,
      "loss": 0.0214,
      "step": 3483
    },
    {
      "epoch": 31.11,
      "grad_norm": 0.06075693294405937,
      "learning_rate": 0.00012458823529411763,
      "loss": 0.0234,
      "step": 3484
    },
    {
      "epoch": 31.12,
      "grad_norm": 0.1130075678229332,
      "learning_rate": 0.00012452941176470587,
      "loss": 0.0285,
      "step": 3485
    },
    {
      "epoch": 31.12,
      "grad_norm": 0.04859291762113571,
      "learning_rate": 0.0001244705882352941,
      "loss": 0.0224,
      "step": 3486
    },
    {
      "epoch": 31.13,
      "grad_norm": 0.044511616230010986,
      "learning_rate": 0.00012441176470588236,
      "loss": 0.023,
      "step": 3487
    },
    {
      "epoch": 31.14,
      "grad_norm": 0.04058460891246796,
      "learning_rate": 0.00012435294117647057,
      "loss": 0.0185,
      "step": 3488
    },
    {
      "epoch": 31.15,
      "grad_norm": 0.08004245907068253,
      "learning_rate": 0.00012429411764705882,
      "loss": 0.0174,
      "step": 3489
    },
    {
      "epoch": 31.16,
      "grad_norm": 0.05423317849636078,
      "learning_rate": 0.00012423529411764706,
      "loss": 0.0202,
      "step": 3490
    },
    {
      "epoch": 31.17,
      "grad_norm": 0.09387530386447906,
      "learning_rate": 0.00012417647058823528,
      "loss": 0.0255,
      "step": 3491
    },
    {
      "epoch": 31.18,
      "grad_norm": 0.044507239013910294,
      "learning_rate": 0.00012411764705882352,
      "loss": 0.0204,
      "step": 3492
    },
    {
      "epoch": 31.19,
      "grad_norm": 0.04497398063540459,
      "learning_rate": 0.00012405882352941176,
      "loss": 0.0199,
      "step": 3493
    },
    {
      "epoch": 31.2,
      "grad_norm": 0.05612332001328468,
      "learning_rate": 0.00012399999999999998,
      "loss": 0.0233,
      "step": 3494
    },
    {
      "epoch": 31.21,
      "grad_norm": 0.05282190069556236,
      "learning_rate": 0.00012394117647058822,
      "loss": 0.0203,
      "step": 3495
    },
    {
      "epoch": 31.21,
      "grad_norm": 0.05601921305060387,
      "learning_rate": 0.00012388235294117647,
      "loss": 0.0236,
      "step": 3496
    },
    {
      "epoch": 31.22,
      "grad_norm": 0.12544254958629608,
      "learning_rate": 0.00012382352941176468,
      "loss": 0.0308,
      "step": 3497
    },
    {
      "epoch": 31.23,
      "grad_norm": 0.04651530086994171,
      "learning_rate": 0.00012376470588235292,
      "loss": 0.0221,
      "step": 3498
    },
    {
      "epoch": 31.24,
      "grad_norm": 0.049244217574596405,
      "learning_rate": 0.00012370588235294117,
      "loss": 0.0207,
      "step": 3499
    },
    {
      "epoch": 31.25,
      "grad_norm": 0.049539964646101,
      "learning_rate": 0.0001236470588235294,
      "loss": 0.0219,
      "step": 3500
    },
    {
      "epoch": 31.25,
      "eval_cer": 0.03221672247841758,
      "eval_loss": 0.1998356431722641,
      "eval_runtime": 23.0411,
      "eval_samples_per_second": 114.665,
      "eval_steps_per_second": 1.823,
      "eval_wer": 0.1053550178500595,
      "step": 3500
    },
    {
      "epoch": 31.26,
      "grad_norm": 0.04684394225478172,
      "learning_rate": 0.00012358823529411766,
      "loss": 0.0191,
      "step": 3501
    },
    {
      "epoch": 31.27,
      "grad_norm": 0.058877211064100266,
      "learning_rate": 0.00012352941176470587,
      "loss": 0.0239,
      "step": 3502
    },
    {
      "epoch": 31.28,
      "grad_norm": 0.07442399859428406,
      "learning_rate": 0.00012347058823529411,
      "loss": 0.024,
      "step": 3503
    },
    {
      "epoch": 31.29,
      "grad_norm": 0.07524219155311584,
      "learning_rate": 0.00012341176470588233,
      "loss": 0.0264,
      "step": 3504
    },
    {
      "epoch": 31.29,
      "grad_norm": 0.04997357353568077,
      "learning_rate": 0.00012335294117647057,
      "loss": 0.0218,
      "step": 3505
    },
    {
      "epoch": 31.3,
      "grad_norm": 0.04596414044499397,
      "learning_rate": 0.00012329411764705882,
      "loss": 0.0203,
      "step": 3506
    },
    {
      "epoch": 31.31,
      "grad_norm": 0.04907511547207832,
      "learning_rate": 0.00012323529411764703,
      "loss": 0.0222,
      "step": 3507
    },
    {
      "epoch": 31.32,
      "grad_norm": 0.048574049025774,
      "learning_rate": 0.00012317647058823528,
      "loss": 0.022,
      "step": 3508
    },
    {
      "epoch": 31.33,
      "grad_norm": 0.0650213211774826,
      "learning_rate": 0.00012311764705882352,
      "loss": 0.0206,
      "step": 3509
    },
    {
      "epoch": 31.34,
      "grad_norm": 0.09144242107868195,
      "learning_rate": 0.00012305882352941176,
      "loss": 0.0297,
      "step": 3510
    },
    {
      "epoch": 31.35,
      "grad_norm": 0.04303023964166641,
      "learning_rate": 0.00012299999999999998,
      "loss": 0.02,
      "step": 3511
    },
    {
      "epoch": 31.36,
      "grad_norm": 0.05274583771824837,
      "learning_rate": 0.00012294117647058822,
      "loss": 0.0229,
      "step": 3512
    },
    {
      "epoch": 31.37,
      "grad_norm": 0.047474268823862076,
      "learning_rate": 0.00012288235294117647,
      "loss": 0.0208,
      "step": 3513
    },
    {
      "epoch": 31.38,
      "grad_norm": 0.04687033221125603,
      "learning_rate": 0.0001228235294117647,
      "loss": 0.0179,
      "step": 3514
    },
    {
      "epoch": 31.38,
      "grad_norm": 0.047554392367601395,
      "learning_rate": 0.00012276470588235293,
      "loss": 0.0174,
      "step": 3515
    },
    {
      "epoch": 31.39,
      "grad_norm": 0.09634137898683548,
      "learning_rate": 0.00012270588235294117,
      "loss": 0.0285,
      "step": 3516
    },
    {
      "epoch": 31.4,
      "grad_norm": 0.04787968844175339,
      "learning_rate": 0.0001226470588235294,
      "loss": 0.0226,
      "step": 3517
    },
    {
      "epoch": 31.41,
      "grad_norm": 0.047528766095638275,
      "learning_rate": 0.00012258823529411763,
      "loss": 0.0198,
      "step": 3518
    },
    {
      "epoch": 31.42,
      "grad_norm": 0.0573684461414814,
      "learning_rate": 0.00012252941176470587,
      "loss": 0.0225,
      "step": 3519
    },
    {
      "epoch": 31.43,
      "grad_norm": 0.04796380549669266,
      "learning_rate": 0.00012247058823529412,
      "loss": 0.0185,
      "step": 3520
    },
    {
      "epoch": 31.44,
      "grad_norm": 0.06380965560674667,
      "learning_rate": 0.00012241176470588233,
      "loss": 0.0253,
      "step": 3521
    },
    {
      "epoch": 31.45,
      "grad_norm": 0.09760231524705887,
      "learning_rate": 0.00012235294117647058,
      "loss": 0.0237,
      "step": 3522
    },
    {
      "epoch": 31.46,
      "grad_norm": 0.04646872729063034,
      "learning_rate": 0.00012229411764705882,
      "loss": 0.0227,
      "step": 3523
    },
    {
      "epoch": 31.46,
      "grad_norm": 0.041683658957481384,
      "learning_rate": 0.00012223529411764706,
      "loss": 0.0187,
      "step": 3524
    },
    {
      "epoch": 31.47,
      "grad_norm": 0.04618990048766136,
      "learning_rate": 0.00012217647058823528,
      "loss": 0.0227,
      "step": 3525
    },
    {
      "epoch": 31.48,
      "grad_norm": 0.045315660536289215,
      "learning_rate": 0.00012211764705882352,
      "loss": 0.0187,
      "step": 3526
    },
    {
      "epoch": 31.49,
      "grad_norm": 0.04802365228533745,
      "learning_rate": 0.00012205882352941175,
      "loss": 0.021,
      "step": 3527
    },
    {
      "epoch": 31.5,
      "grad_norm": 0.07292218506336212,
      "learning_rate": 0.000122,
      "loss": 0.0179,
      "step": 3528
    },
    {
      "epoch": 31.51,
      "grad_norm": 0.0829540491104126,
      "learning_rate": 0.00012194117647058823,
      "loss": 0.0248,
      "step": 3529
    },
    {
      "epoch": 31.52,
      "grad_norm": 0.05062733218073845,
      "learning_rate": 0.00012188235294117646,
      "loss": 0.0223,
      "step": 3530
    },
    {
      "epoch": 31.53,
      "grad_norm": 0.056058090180158615,
      "learning_rate": 0.00012182352941176469,
      "loss": 0.023,
      "step": 3531
    },
    {
      "epoch": 31.54,
      "grad_norm": 0.05849380046129227,
      "learning_rate": 0.00012176470588235293,
      "loss": 0.0224,
      "step": 3532
    },
    {
      "epoch": 31.54,
      "grad_norm": 0.04975719004869461,
      "learning_rate": 0.00012170588235294117,
      "loss": 0.0169,
      "step": 3533
    },
    {
      "epoch": 31.55,
      "grad_norm": 0.07418406009674072,
      "learning_rate": 0.0001216470588235294,
      "loss": 0.0262,
      "step": 3534
    },
    {
      "epoch": 31.56,
      "grad_norm": 0.07387860119342804,
      "learning_rate": 0.00012158823529411763,
      "loss": 0.0241,
      "step": 3535
    },
    {
      "epoch": 31.57,
      "grad_norm": 0.05272188410162926,
      "learning_rate": 0.00012152941176470586,
      "loss": 0.0236,
      "step": 3536
    },
    {
      "epoch": 31.58,
      "grad_norm": 0.04087989777326584,
      "learning_rate": 0.00012147058823529411,
      "loss": 0.0172,
      "step": 3537
    },
    {
      "epoch": 31.59,
      "grad_norm": 0.04647759720683098,
      "learning_rate": 0.00012141176470588235,
      "loss": 0.0199,
      "step": 3538
    },
    {
      "epoch": 31.6,
      "grad_norm": 0.045854777097702026,
      "learning_rate": 0.00012135294117647057,
      "loss": 0.0177,
      "step": 3539
    },
    {
      "epoch": 31.61,
      "grad_norm": 0.06473439931869507,
      "learning_rate": 0.00012129411764705881,
      "loss": 0.0232,
      "step": 3540
    },
    {
      "epoch": 31.62,
      "grad_norm": 0.0961967185139656,
      "learning_rate": 0.00012123529411764705,
      "loss": 0.0339,
      "step": 3541
    },
    {
      "epoch": 31.62,
      "grad_norm": 0.04991842806339264,
      "learning_rate": 0.00012117647058823528,
      "loss": 0.0232,
      "step": 3542
    },
    {
      "epoch": 31.63,
      "grad_norm": 0.038326505571603775,
      "learning_rate": 0.00012111764705882353,
      "loss": 0.0155,
      "step": 3543
    },
    {
      "epoch": 31.64,
      "grad_norm": 0.0446661151945591,
      "learning_rate": 0.00012105882352941174,
      "loss": 0.0186,
      "step": 3544
    },
    {
      "epoch": 31.65,
      "grad_norm": 0.04957854002714157,
      "learning_rate": 0.00012099999999999999,
      "loss": 0.0205,
      "step": 3545
    },
    {
      "epoch": 31.66,
      "grad_norm": 0.05040052533149719,
      "learning_rate": 0.00012094117647058823,
      "loss": 0.0198,
      "step": 3546
    },
    {
      "epoch": 31.67,
      "grad_norm": 0.0885852724313736,
      "learning_rate": 0.00012088235294117646,
      "loss": 0.0229,
      "step": 3547
    },
    {
      "epoch": 31.68,
      "grad_norm": 0.04862269014120102,
      "learning_rate": 0.0001208235294117647,
      "loss": 0.0227,
      "step": 3548
    },
    {
      "epoch": 31.69,
      "grad_norm": 0.045693445950746536,
      "learning_rate": 0.00012076470588235292,
      "loss": 0.0176,
      "step": 3549
    },
    {
      "epoch": 31.7,
      "grad_norm": 0.04793444648385048,
      "learning_rate": 0.00012070588235294116,
      "loss": 0.0207,
      "step": 3550
    },
    {
      "epoch": 31.71,
      "grad_norm": 0.054520826786756516,
      "learning_rate": 0.0001206470588235294,
      "loss": 0.0224,
      "step": 3551
    },
    {
      "epoch": 31.71,
      "grad_norm": 0.06703490763902664,
      "learning_rate": 0.00012058823529411764,
      "loss": 0.0221,
      "step": 3552
    },
    {
      "epoch": 31.72,
      "grad_norm": 0.08838488906621933,
      "learning_rate": 0.00012052941176470588,
      "loss": 0.0262,
      "step": 3553
    },
    {
      "epoch": 31.73,
      "grad_norm": 0.07164445519447327,
      "learning_rate": 0.0001204705882352941,
      "loss": 0.0194,
      "step": 3554
    },
    {
      "epoch": 31.74,
      "grad_norm": 0.05277720466256142,
      "learning_rate": 0.00012041176470588234,
      "loss": 0.0208,
      "step": 3555
    },
    {
      "epoch": 31.75,
      "grad_norm": 0.05103946849703789,
      "learning_rate": 0.00012035294117647058,
      "loss": 0.0208,
      "step": 3556
    },
    {
      "epoch": 31.76,
      "grad_norm": 0.0512225367128849,
      "learning_rate": 0.00012029411764705881,
      "loss": 0.0189,
      "step": 3557
    },
    {
      "epoch": 31.77,
      "grad_norm": 0.05225786566734314,
      "learning_rate": 0.00012023529411764704,
      "loss": 0.0194,
      "step": 3558
    },
    {
      "epoch": 31.78,
      "grad_norm": 0.0629500150680542,
      "learning_rate": 0.00012017647058823529,
      "loss": 0.0233,
      "step": 3559
    },
    {
      "epoch": 31.79,
      "grad_norm": 0.06870309263467789,
      "learning_rate": 0.00012011764705882352,
      "loss": 0.0245,
      "step": 3560
    },
    {
      "epoch": 31.79,
      "grad_norm": 0.04595329612493515,
      "learning_rate": 0.00012005882352941176,
      "loss": 0.0209,
      "step": 3561
    },
    {
      "epoch": 31.8,
      "grad_norm": 0.043259911239147186,
      "learning_rate": 0.00011999999999999999,
      "loss": 0.0175,
      "step": 3562
    },
    {
      "epoch": 31.81,
      "grad_norm": 0.04623683542013168,
      "learning_rate": 0.00011994117647058822,
      "loss": 0.0205,
      "step": 3563
    },
    {
      "epoch": 31.82,
      "grad_norm": 0.046441450715065,
      "learning_rate": 0.00011988235294117646,
      "loss": 0.0174,
      "step": 3564
    },
    {
      "epoch": 31.83,
      "grad_norm": 0.08617646992206573,
      "learning_rate": 0.00011982352941176469,
      "loss": 0.02,
      "step": 3565
    },
    {
      "epoch": 31.84,
      "grad_norm": 0.11688149720430374,
      "learning_rate": 0.00011976470588235294,
      "loss": 0.0322,
      "step": 3566
    },
    {
      "epoch": 31.85,
      "grad_norm": 0.049706801772117615,
      "learning_rate": 0.00011970588235294118,
      "loss": 0.0194,
      "step": 3567
    },
    {
      "epoch": 31.86,
      "grad_norm": 0.046902839094400406,
      "learning_rate": 0.0001196470588235294,
      "loss": 0.0201,
      "step": 3568
    },
    {
      "epoch": 31.87,
      "grad_norm": 0.0508330799639225,
      "learning_rate": 0.00011958823529411764,
      "loss": 0.0187,
      "step": 3569
    },
    {
      "epoch": 31.88,
      "grad_norm": 0.04902074113488197,
      "learning_rate": 0.00011952941176470587,
      "loss": 0.0191,
      "step": 3570
    },
    {
      "epoch": 31.88,
      "grad_norm": 0.050239987671375275,
      "learning_rate": 0.00011947058823529411,
      "loss": 0.0184,
      "step": 3571
    },
    {
      "epoch": 31.89,
      "grad_norm": 0.11308218538761139,
      "learning_rate": 0.00011941176470588233,
      "loss": 0.0264,
      "step": 3572
    },
    {
      "epoch": 31.9,
      "grad_norm": 0.039108309894800186,
      "learning_rate": 0.00011935294117647057,
      "loss": 0.0181,
      "step": 3573
    },
    {
      "epoch": 31.91,
      "grad_norm": 0.05109729990363121,
      "learning_rate": 0.00011929411764705882,
      "loss": 0.0226,
      "step": 3574
    },
    {
      "epoch": 31.92,
      "grad_norm": 0.04487459361553192,
      "learning_rate": 0.00011923529411764705,
      "loss": 0.0201,
      "step": 3575
    },
    {
      "epoch": 31.93,
      "grad_norm": 0.04437851533293724,
      "learning_rate": 0.00011917647058823529,
      "loss": 0.0212,
      "step": 3576
    },
    {
      "epoch": 31.94,
      "grad_norm": 0.052766717970371246,
      "learning_rate": 0.0001191176470588235,
      "loss": 0.0175,
      "step": 3577
    },
    {
      "epoch": 31.95,
      "grad_norm": 0.08078915625810623,
      "learning_rate": 0.00011905882352941175,
      "loss": 0.0271,
      "step": 3578
    },
    {
      "epoch": 31.96,
      "grad_norm": 0.07316518574953079,
      "learning_rate": 0.00011899999999999999,
      "loss": 0.0249,
      "step": 3579
    },
    {
      "epoch": 31.96,
      "grad_norm": 0.0467817448079586,
      "learning_rate": 0.00011894117647058822,
      "loss": 0.0198,
      "step": 3580
    },
    {
      "epoch": 31.97,
      "grad_norm": 0.052814096212387085,
      "learning_rate": 0.00011888235294117647,
      "loss": 0.0258,
      "step": 3581
    },
    {
      "epoch": 31.98,
      "grad_norm": 0.06177174299955368,
      "learning_rate": 0.0001188235294117647,
      "loss": 0.0246,
      "step": 3582
    },
    {
      "epoch": 31.99,
      "grad_norm": 0.05956928804516792,
      "learning_rate": 0.00011876470588235293,
      "loss": 0.0201,
      "step": 3583
    },
    {
      "epoch": 32.0,
      "grad_norm": 0.08306241035461426,
      "learning_rate": 0.00011870588235294117,
      "loss": 0.0259,
      "step": 3584
    },
    {
      "epoch": 32.01,
      "grad_norm": 0.04788234084844589,
      "learning_rate": 0.00011864705882352941,
      "loss": 0.0238,
      "step": 3585
    },
    {
      "epoch": 32.02,
      "grad_norm": 0.04594253748655319,
      "learning_rate": 0.00011858823529411763,
      "loss": 0.0175,
      "step": 3586
    },
    {
      "epoch": 32.03,
      "grad_norm": 0.046111781150102615,
      "learning_rate": 0.00011852941176470587,
      "loss": 0.0187,
      "step": 3587
    },
    {
      "epoch": 32.04,
      "grad_norm": 0.051978856325149536,
      "learning_rate": 0.0001184705882352941,
      "loss": 0.0181,
      "step": 3588
    },
    {
      "epoch": 32.04,
      "grad_norm": 0.05756892263889313,
      "learning_rate": 0.00011841176470588235,
      "loss": 0.0231,
      "step": 3589
    },
    {
      "epoch": 32.05,
      "grad_norm": 0.08719819784164429,
      "learning_rate": 0.00011835294117647059,
      "loss": 0.0274,
      "step": 3590
    },
    {
      "epoch": 32.06,
      "grad_norm": 0.09158667922019958,
      "learning_rate": 0.0001182941176470588,
      "loss": 0.0258,
      "step": 3591
    },
    {
      "epoch": 32.07,
      "grad_norm": 0.04844530671834946,
      "learning_rate": 0.00011823529411764705,
      "loss": 0.0208,
      "step": 3592
    },
    {
      "epoch": 32.08,
      "grad_norm": 0.053919319063425064,
      "learning_rate": 0.00011817647058823528,
      "loss": 0.0245,
      "step": 3593
    },
    {
      "epoch": 32.09,
      "grad_norm": 0.04653266444802284,
      "learning_rate": 0.00011811764705882352,
      "loss": 0.0195,
      "step": 3594
    },
    {
      "epoch": 32.1,
      "grad_norm": 0.05394500866532326,
      "learning_rate": 0.00011805882352941177,
      "loss": 0.0184,
      "step": 3595
    },
    {
      "epoch": 32.11,
      "grad_norm": 0.07050443440675735,
      "learning_rate": 0.00011799999999999998,
      "loss": 0.024,
      "step": 3596
    },
    {
      "epoch": 32.12,
      "grad_norm": 0.0826314389705658,
      "learning_rate": 0.00011794117647058822,
      "loss": 0.0269,
      "step": 3597
    },
    {
      "epoch": 32.12,
      "grad_norm": 0.04215487465262413,
      "learning_rate": 0.00011788235294117645,
      "loss": 0.0198,
      "step": 3598
    },
    {
      "epoch": 32.13,
      "grad_norm": 0.041640754789114,
      "learning_rate": 0.0001178235294117647,
      "loss": 0.0176,
      "step": 3599
    },
    {
      "epoch": 32.14,
      "grad_norm": 0.044302526861429214,
      "learning_rate": 0.00011776470588235294,
      "loss": 0.0198,
      "step": 3600
    },
    {
      "epoch": 32.14,
      "eval_cer": 0.032235514313740964,
      "eval_loss": 0.21042084693908691,
      "eval_runtime": 22.9332,
      "eval_samples_per_second": 115.204,
      "eval_steps_per_second": 1.831,
      "eval_wer": 0.10479968266560888,
      "step": 3600
    },
    {
      "epoch": 32.15,
      "grad_norm": 0.05428871512413025,
      "learning_rate": 0.00011770588235294116,
      "loss": 0.0215,
      "step": 3601
    },
    {
      "epoch": 32.16,
      "grad_norm": 0.06890869140625,
      "learning_rate": 0.0001176470588235294,
      "loss": 0.0216,
      "step": 3602
    },
    {
      "epoch": 32.17,
      "grad_norm": 0.0953051894903183,
      "learning_rate": 0.00011758823529411764,
      "loss": 0.0205,
      "step": 3603
    },
    {
      "epoch": 32.18,
      "grad_norm": 0.047257766127586365,
      "learning_rate": 0.00011752941176470587,
      "loss": 0.0216,
      "step": 3604
    },
    {
      "epoch": 32.19,
      "grad_norm": 0.04303435608744621,
      "learning_rate": 0.0001174705882352941,
      "loss": 0.0203,
      "step": 3605
    },
    {
      "epoch": 32.2,
      "grad_norm": 0.05104793608188629,
      "learning_rate": 0.00011741176470588233,
      "loss": 0.0211,
      "step": 3606
    },
    {
      "epoch": 32.21,
      "grad_norm": 0.05943785235285759,
      "learning_rate": 0.00011735294117647058,
      "loss": 0.0183,
      "step": 3607
    },
    {
      "epoch": 32.21,
      "grad_norm": 0.06392104923725128,
      "learning_rate": 0.00011729411764705882,
      "loss": 0.0198,
      "step": 3608
    },
    {
      "epoch": 32.22,
      "grad_norm": 0.11663123220205307,
      "learning_rate": 0.00011723529411764705,
      "loss": 0.0274,
      "step": 3609
    },
    {
      "epoch": 32.23,
      "grad_norm": 0.047014717012643814,
      "learning_rate": 0.00011717647058823528,
      "loss": 0.0175,
      "step": 3610
    },
    {
      "epoch": 32.24,
      "grad_norm": 0.047937631607055664,
      "learning_rate": 0.00011711764705882351,
      "loss": 0.0195,
      "step": 3611
    },
    {
      "epoch": 32.25,
      "grad_norm": 0.046994805335998535,
      "learning_rate": 0.00011705882352941175,
      "loss": 0.0183,
      "step": 3612
    },
    {
      "epoch": 32.26,
      "grad_norm": 0.05170201137661934,
      "learning_rate": 0.000117,
      "loss": 0.0189,
      "step": 3613
    },
    {
      "epoch": 32.27,
      "grad_norm": 0.05205686017870903,
      "learning_rate": 0.00011694117647058823,
      "loss": 0.021,
      "step": 3614
    },
    {
      "epoch": 32.28,
      "grad_norm": 0.057531144469976425,
      "learning_rate": 0.00011688235294117646,
      "loss": 0.0179,
      "step": 3615
    },
    {
      "epoch": 32.29,
      "grad_norm": 0.10487011820077896,
      "learning_rate": 0.00011682352941176469,
      "loss": 0.0239,
      "step": 3616
    },
    {
      "epoch": 32.29,
      "grad_norm": 0.054197829216718674,
      "learning_rate": 0.00011676470588235293,
      "loss": 0.0221,
      "step": 3617
    },
    {
      "epoch": 32.3,
      "grad_norm": 0.05336236581206322,
      "learning_rate": 0.00011670588235294117,
      "loss": 0.0248,
      "step": 3618
    },
    {
      "epoch": 32.31,
      "grad_norm": 0.054558366537094116,
      "learning_rate": 0.00011664705882352939,
      "loss": 0.0186,
      "step": 3619
    },
    {
      "epoch": 32.32,
      "grad_norm": 0.05783456563949585,
      "learning_rate": 0.00011658823529411763,
      "loss": 0.0174,
      "step": 3620
    },
    {
      "epoch": 32.33,
      "grad_norm": 0.06637098640203476,
      "learning_rate": 0.00011652941176470588,
      "loss": 0.0192,
      "step": 3621
    },
    {
      "epoch": 32.34,
      "grad_norm": 0.12480707466602325,
      "learning_rate": 0.00011647058823529411,
      "loss": 0.0258,
      "step": 3622
    },
    {
      "epoch": 32.35,
      "grad_norm": 0.04600154235959053,
      "learning_rate": 0.00011641176470588235,
      "loss": 0.018,
      "step": 3623
    },
    {
      "epoch": 32.36,
      "grad_norm": 0.05718235671520233,
      "learning_rate": 0.00011635294117647057,
      "loss": 0.022,
      "step": 3624
    },
    {
      "epoch": 32.37,
      "grad_norm": 0.056042853742837906,
      "learning_rate": 0.00011629411764705881,
      "loss": 0.0222,
      "step": 3625
    },
    {
      "epoch": 32.38,
      "grad_norm": 0.04886164516210556,
      "learning_rate": 0.00011623529411764705,
      "loss": 0.0194,
      "step": 3626
    },
    {
      "epoch": 32.38,
      "grad_norm": 0.05563600733876228,
      "learning_rate": 0.00011617647058823528,
      "loss": 0.0184,
      "step": 3627
    },
    {
      "epoch": 32.39,
      "grad_norm": 0.09011834114789963,
      "learning_rate": 0.00011611764705882353,
      "loss": 0.0307,
      "step": 3628
    },
    {
      "epoch": 32.4,
      "grad_norm": 0.04582998901605606,
      "learning_rate": 0.00011605882352941174,
      "loss": 0.0195,
      "step": 3629
    },
    {
      "epoch": 32.41,
      "grad_norm": 0.05276224762201309,
      "learning_rate": 0.00011599999999999999,
      "loss": 0.0231,
      "step": 3630
    },
    {
      "epoch": 32.42,
      "grad_norm": 0.04544449225068092,
      "learning_rate": 0.00011594117647058823,
      "loss": 0.0172,
      "step": 3631
    },
    {
      "epoch": 32.43,
      "grad_norm": 0.04735076054930687,
      "learning_rate": 0.00011588235294117646,
      "loss": 0.02,
      "step": 3632
    },
    {
      "epoch": 32.44,
      "grad_norm": 0.05830598250031471,
      "learning_rate": 0.0001158235294117647,
      "loss": 0.0202,
      "step": 3633
    },
    {
      "epoch": 32.45,
      "grad_norm": 0.12295148521661758,
      "learning_rate": 0.00011576470588235292,
      "loss": 0.0314,
      "step": 3634
    },
    {
      "epoch": 32.46,
      "grad_norm": 0.047054532915353775,
      "learning_rate": 0.00011570588235294116,
      "loss": 0.0237,
      "step": 3635
    },
    {
      "epoch": 32.46,
      "grad_norm": 0.050387535244226456,
      "learning_rate": 0.00011564705882352941,
      "loss": 0.0203,
      "step": 3636
    },
    {
      "epoch": 32.47,
      "grad_norm": 0.04612543806433678,
      "learning_rate": 0.00011558823529411764,
      "loss": 0.0199,
      "step": 3637
    },
    {
      "epoch": 32.48,
      "grad_norm": 0.04046722501516342,
      "learning_rate": 0.00011552941176470587,
      "loss": 0.015,
      "step": 3638
    },
    {
      "epoch": 32.49,
      "grad_norm": 0.06123023107647896,
      "learning_rate": 0.00011547058823529411,
      "loss": 0.0256,
      "step": 3639
    },
    {
      "epoch": 32.5,
      "grad_norm": 0.06880581378936768,
      "learning_rate": 0.00011541176470588234,
      "loss": 0.0196,
      "step": 3640
    },
    {
      "epoch": 32.51,
      "grad_norm": 0.08232057094573975,
      "learning_rate": 0.00011535294117647058,
      "loss": 0.0253,
      "step": 3641
    },
    {
      "epoch": 32.52,
      "grad_norm": 0.045623622834682465,
      "learning_rate": 0.00011529411764705881,
      "loss": 0.0201,
      "step": 3642
    },
    {
      "epoch": 32.53,
      "grad_norm": 0.05098617449402809,
      "learning_rate": 0.00011523529411764704,
      "loss": 0.0225,
      "step": 3643
    },
    {
      "epoch": 32.54,
      "grad_norm": 0.04750572890043259,
      "learning_rate": 0.00011517647058823529,
      "loss": 0.0186,
      "step": 3644
    },
    {
      "epoch": 32.54,
      "grad_norm": 0.047422174364328384,
      "learning_rate": 0.00011511764705882352,
      "loss": 0.0178,
      "step": 3645
    },
    {
      "epoch": 32.55,
      "grad_norm": 0.06636884808540344,
      "learning_rate": 0.00011505882352941176,
      "loss": 0.0215,
      "step": 3646
    },
    {
      "epoch": 32.56,
      "grad_norm": 0.0959538146853447,
      "learning_rate": 0.000115,
      "loss": 0.0243,
      "step": 3647
    },
    {
      "epoch": 32.57,
      "grad_norm": 0.044255390763282776,
      "learning_rate": 0.00011494117647058822,
      "loss": 0.0195,
      "step": 3648
    },
    {
      "epoch": 32.58,
      "grad_norm": 0.04697493091225624,
      "learning_rate": 0.00011488235294117646,
      "loss": 0.0229,
      "step": 3649
    },
    {
      "epoch": 32.59,
      "grad_norm": 0.055186036974191666,
      "learning_rate": 0.00011482352941176469,
      "loss": 0.0214,
      "step": 3650
    },
    {
      "epoch": 32.6,
      "grad_norm": 0.048233091831207275,
      "learning_rate": 0.00011476470588235294,
      "loss": 0.0182,
      "step": 3651
    },
    {
      "epoch": 32.61,
      "grad_norm": 0.058868657797575,
      "learning_rate": 0.00011470588235294115,
      "loss": 0.0211,
      "step": 3652
    },
    {
      "epoch": 32.62,
      "grad_norm": 0.07956228405237198,
      "learning_rate": 0.0001146470588235294,
      "loss": 0.0212,
      "step": 3653
    },
    {
      "epoch": 32.62,
      "grad_norm": 0.04887668043375015,
      "learning_rate": 0.00011458823529411764,
      "loss": 0.0198,
      "step": 3654
    },
    {
      "epoch": 32.63,
      "grad_norm": 0.048192281275987625,
      "learning_rate": 0.00011452941176470587,
      "loss": 0.0227,
      "step": 3655
    },
    {
      "epoch": 32.64,
      "grad_norm": 0.04844574257731438,
      "learning_rate": 0.00011447058823529411,
      "loss": 0.0201,
      "step": 3656
    },
    {
      "epoch": 32.65,
      "grad_norm": 0.04375741258263588,
      "learning_rate": 0.00011441176470588233,
      "loss": 0.0205,
      "step": 3657
    },
    {
      "epoch": 32.66,
      "grad_norm": 0.051589082926511765,
      "learning_rate": 0.00011435294117647057,
      "loss": 0.0212,
      "step": 3658
    },
    {
      "epoch": 32.67,
      "grad_norm": 0.09967084974050522,
      "learning_rate": 0.00011429411764705882,
      "loss": 0.0285,
      "step": 3659
    },
    {
      "epoch": 32.68,
      "grad_norm": 0.05101574957370758,
      "learning_rate": 0.00011423529411764705,
      "loss": 0.0214,
      "step": 3660
    },
    {
      "epoch": 32.69,
      "grad_norm": 0.04472147300839424,
      "learning_rate": 0.00011417647058823529,
      "loss": 0.0194,
      "step": 3661
    },
    {
      "epoch": 32.7,
      "grad_norm": 0.04980763792991638,
      "learning_rate": 0.00011411764705882352,
      "loss": 0.0203,
      "step": 3662
    },
    {
      "epoch": 32.71,
      "grad_norm": 0.04720604419708252,
      "learning_rate": 0.00011405882352941175,
      "loss": 0.0166,
      "step": 3663
    },
    {
      "epoch": 32.71,
      "grad_norm": 0.061146367341279984,
      "learning_rate": 0.00011399999999999999,
      "loss": 0.019,
      "step": 3664
    },
    {
      "epoch": 32.72,
      "grad_norm": 0.0784432664513588,
      "learning_rate": 0.00011394117647058824,
      "loss": 0.0185,
      "step": 3665
    },
    {
      "epoch": 32.73,
      "grad_norm": 0.07377848029136658,
      "learning_rate": 0.00011388235294117645,
      "loss": 0.0245,
      "step": 3666
    },
    {
      "epoch": 32.74,
      "grad_norm": 0.054255399852991104,
      "learning_rate": 0.0001138235294117647,
      "loss": 0.0206,
      "step": 3667
    },
    {
      "epoch": 32.75,
      "grad_norm": 0.05981878191232681,
      "learning_rate": 0.00011376470588235293,
      "loss": 0.0236,
      "step": 3668
    },
    {
      "epoch": 32.76,
      "grad_norm": 0.04551875963807106,
      "learning_rate": 0.00011370588235294117,
      "loss": 0.0215,
      "step": 3669
    },
    {
      "epoch": 32.77,
      "grad_norm": 0.055024564266204834,
      "learning_rate": 0.00011364705882352941,
      "loss": 0.0208,
      "step": 3670
    },
    {
      "epoch": 32.78,
      "grad_norm": 0.06826770305633545,
      "learning_rate": 0.00011358823529411763,
      "loss": 0.0225,
      "step": 3671
    },
    {
      "epoch": 32.79,
      "grad_norm": 0.09451206028461456,
      "learning_rate": 0.00011352941176470587,
      "loss": 0.021,
      "step": 3672
    },
    {
      "epoch": 32.79,
      "grad_norm": 0.043758291751146317,
      "learning_rate": 0.0001134705882352941,
      "loss": 0.0175,
      "step": 3673
    },
    {
      "epoch": 32.8,
      "grad_norm": 0.04687466844916344,
      "learning_rate": 0.00011341176470588235,
      "loss": 0.0198,
      "step": 3674
    },
    {
      "epoch": 32.81,
      "grad_norm": 0.046850625425577164,
      "learning_rate": 0.00011335294117647059,
      "loss": 0.0226,
      "step": 3675
    },
    {
      "epoch": 32.82,
      "grad_norm": 0.045804087072610855,
      "learning_rate": 0.0001132941176470588,
      "loss": 0.0196,
      "step": 3676
    },
    {
      "epoch": 32.83,
      "grad_norm": 0.057633958756923676,
      "learning_rate": 0.00011323529411764705,
      "loss": 0.0183,
      "step": 3677
    },
    {
      "epoch": 32.84,
      "grad_norm": 0.08373904228210449,
      "learning_rate": 0.00011317647058823528,
      "loss": 0.0265,
      "step": 3678
    },
    {
      "epoch": 32.85,
      "grad_norm": 0.049049582332372665,
      "learning_rate": 0.00011311764705882352,
      "loss": 0.0224,
      "step": 3679
    },
    {
      "epoch": 32.86,
      "grad_norm": 0.043412089347839355,
      "learning_rate": 0.00011305882352941177,
      "loss": 0.019,
      "step": 3680
    },
    {
      "epoch": 32.87,
      "grad_norm": 0.041254326701164246,
      "learning_rate": 0.00011299999999999998,
      "loss": 0.0185,
      "step": 3681
    },
    {
      "epoch": 32.88,
      "grad_norm": 0.05116904899477959,
      "learning_rate": 0.00011294117647058823,
      "loss": 0.0217,
      "step": 3682
    },
    {
      "epoch": 32.88,
      "grad_norm": 0.05248168110847473,
      "learning_rate": 0.00011288235294117647,
      "loss": 0.02,
      "step": 3683
    },
    {
      "epoch": 32.89,
      "grad_norm": 0.09650642424821854,
      "learning_rate": 0.0001128235294117647,
      "loss": 0.0219,
      "step": 3684
    },
    {
      "epoch": 32.9,
      "grad_norm": 0.05646537244319916,
      "learning_rate": 0.00011276470588235293,
      "loss": 0.0218,
      "step": 3685
    },
    {
      "epoch": 32.91,
      "grad_norm": 0.04979201406240463,
      "learning_rate": 0.00011270588235294116,
      "loss": 0.021,
      "step": 3686
    },
    {
      "epoch": 32.92,
      "grad_norm": 0.04833272099494934,
      "learning_rate": 0.0001126470588235294,
      "loss": 0.0226,
      "step": 3687
    },
    {
      "epoch": 32.93,
      "grad_norm": 0.04990481212735176,
      "learning_rate": 0.00011258823529411765,
      "loss": 0.0193,
      "step": 3688
    },
    {
      "epoch": 32.94,
      "grad_norm": 0.057368963956832886,
      "learning_rate": 0.00011252941176470588,
      "loss": 0.0204,
      "step": 3689
    },
    {
      "epoch": 32.95,
      "grad_norm": 0.06582289934158325,
      "learning_rate": 0.0001124705882352941,
      "loss": 0.0221,
      "step": 3690
    },
    {
      "epoch": 32.96,
      "grad_norm": 0.06527213007211685,
      "learning_rate": 0.00011241176470588233,
      "loss": 0.0229,
      "step": 3691
    },
    {
      "epoch": 32.96,
      "grad_norm": 0.04593219608068466,
      "learning_rate": 0.00011235294117647058,
      "loss": 0.0216,
      "step": 3692
    },
    {
      "epoch": 32.97,
      "grad_norm": 0.048174042254686356,
      "learning_rate": 0.00011229411764705882,
      "loss": 0.0196,
      "step": 3693
    },
    {
      "epoch": 32.98,
      "grad_norm": 0.04866315796971321,
      "learning_rate": 0.00011223529411764705,
      "loss": 0.0179,
      "step": 3694
    },
    {
      "epoch": 32.99,
      "grad_norm": 0.05889221280813217,
      "learning_rate": 0.00011217647058823528,
      "loss": 0.0179,
      "step": 3695
    },
    {
      "epoch": 33.0,
      "grad_norm": 0.07265275716781616,
      "learning_rate": 0.00011211764705882351,
      "loss": 0.021,
      "step": 3696
    },
    {
      "epoch": 33.01,
      "grad_norm": 0.043408121913671494,
      "learning_rate": 0.00011205882352941175,
      "loss": 0.018,
      "step": 3697
    },
    {
      "epoch": 33.02,
      "grad_norm": 0.041275300085544586,
      "learning_rate": 0.000112,
      "loss": 0.0167,
      "step": 3698
    },
    {
      "epoch": 33.03,
      "grad_norm": 0.04630756005644798,
      "learning_rate": 0.00011194117647058821,
      "loss": 0.0185,
      "step": 3699
    },
    {
      "epoch": 33.04,
      "grad_norm": 0.04922295734286308,
      "learning_rate": 0.00011188235294117646,
      "loss": 0.0181,
      "step": 3700
    },
    {
      "epoch": 33.04,
      "eval_cer": 0.032483566540009695,
      "eval_loss": 0.20927734673023224,
      "eval_runtime": 22.7958,
      "eval_samples_per_second": 115.898,
      "eval_steps_per_second": 1.842,
      "eval_wer": 0.10497818326061087,
      "step": 3700
    },
    {
      "epoch": 33.04,
      "grad_norm": 0.059649765491485596,
      "learning_rate": 0.0001118235294117647,
      "loss": 0.0217,
      "step": 3701
    },
    {
      "epoch": 33.05,
      "grad_norm": 0.07166648656129837,
      "learning_rate": 0.00011176470588235293,
      "loss": 0.0202,
      "step": 3702
    },
    {
      "epoch": 33.06,
      "grad_norm": 0.056485336273908615,
      "learning_rate": 0.00011170588235294117,
      "loss": 0.0214,
      "step": 3703
    },
    {
      "epoch": 33.07,
      "grad_norm": 0.04659261554479599,
      "learning_rate": 0.00011164705882352939,
      "loss": 0.0187,
      "step": 3704
    },
    {
      "epoch": 33.08,
      "grad_norm": 0.054145973175764084,
      "learning_rate": 0.00011158823529411763,
      "loss": 0.0195,
      "step": 3705
    },
    {
      "epoch": 33.09,
      "grad_norm": 0.05096295475959778,
      "learning_rate": 0.00011152941176470588,
      "loss": 0.0189,
      "step": 3706
    },
    {
      "epoch": 33.1,
      "grad_norm": 0.04888540133833885,
      "learning_rate": 0.00011147058823529411,
      "loss": 0.0213,
      "step": 3707
    },
    {
      "epoch": 33.11,
      "grad_norm": 0.06510414928197861,
      "learning_rate": 0.00011141176470588235,
      "loss": 0.024,
      "step": 3708
    },
    {
      "epoch": 33.12,
      "grad_norm": 0.2117670178413391,
      "learning_rate": 0.00011135294117647057,
      "loss": 0.0211,
      "step": 3709
    },
    {
      "epoch": 33.12,
      "grad_norm": 0.04437718167901039,
      "learning_rate": 0.00011129411764705881,
      "loss": 0.0211,
      "step": 3710
    },
    {
      "epoch": 33.13,
      "grad_norm": 0.0447576679289341,
      "learning_rate": 0.00011123529411764705,
      "loss": 0.0191,
      "step": 3711
    },
    {
      "epoch": 33.14,
      "grad_norm": 0.05013945698738098,
      "learning_rate": 0.00011117647058823528,
      "loss": 0.0211,
      "step": 3712
    },
    {
      "epoch": 33.15,
      "grad_norm": 0.04286734759807587,
      "learning_rate": 0.00011111764705882353,
      "loss": 0.0164,
      "step": 3713
    },
    {
      "epoch": 33.16,
      "grad_norm": 0.06261122226715088,
      "learning_rate": 0.00011105882352941174,
      "loss": 0.0235,
      "step": 3714
    },
    {
      "epoch": 33.17,
      "grad_norm": 0.08340524137020111,
      "learning_rate": 0.00011099999999999999,
      "loss": 0.0245,
      "step": 3715
    },
    {
      "epoch": 33.18,
      "grad_norm": 0.04745287075638771,
      "learning_rate": 0.00011094117647058823,
      "loss": 0.0204,
      "step": 3716
    },
    {
      "epoch": 33.19,
      "grad_norm": 0.04093869775533676,
      "learning_rate": 0.00011088235294117646,
      "loss": 0.0189,
      "step": 3717
    },
    {
      "epoch": 33.2,
      "grad_norm": 0.046972230076789856,
      "learning_rate": 0.00011082352941176469,
      "loss": 0.0189,
      "step": 3718
    },
    {
      "epoch": 33.21,
      "grad_norm": 0.053276680409908295,
      "learning_rate": 0.00011076470588235293,
      "loss": 0.0177,
      "step": 3719
    },
    {
      "epoch": 33.21,
      "grad_norm": 0.05650439113378525,
      "learning_rate": 0.00011070588235294116,
      "loss": 0.0213,
      "step": 3720
    },
    {
      "epoch": 33.22,
      "grad_norm": 0.13323943316936493,
      "learning_rate": 0.00011064705882352941,
      "loss": 0.0245,
      "step": 3721
    },
    {
      "epoch": 33.23,
      "grad_norm": 0.04782477766275406,
      "learning_rate": 0.00011058823529411765,
      "loss": 0.0189,
      "step": 3722
    },
    {
      "epoch": 33.24,
      "grad_norm": 0.05005009099841118,
      "learning_rate": 0.00011052941176470587,
      "loss": 0.021,
      "step": 3723
    },
    {
      "epoch": 33.25,
      "grad_norm": 0.06174638122320175,
      "learning_rate": 0.00011047058823529411,
      "loss": 0.0226,
      "step": 3724
    },
    {
      "epoch": 33.26,
      "grad_norm": 0.054218094795942307,
      "learning_rate": 0.00011041176470588234,
      "loss": 0.0191,
      "step": 3725
    },
    {
      "epoch": 33.27,
      "grad_norm": 0.047325361520051956,
      "learning_rate": 0.00011035294117647058,
      "loss": 0.0144,
      "step": 3726
    },
    {
      "epoch": 33.28,
      "grad_norm": 0.07584651559591293,
      "learning_rate": 0.00011029411764705883,
      "loss": 0.0182,
      "step": 3727
    },
    {
      "epoch": 33.29,
      "grad_norm": 0.06795705109834671,
      "learning_rate": 0.00011023529411764704,
      "loss": 0.0236,
      "step": 3728
    },
    {
      "epoch": 33.29,
      "grad_norm": 0.0456150621175766,
      "learning_rate": 0.00011017647058823529,
      "loss": 0.0222,
      "step": 3729
    },
    {
      "epoch": 33.3,
      "grad_norm": 0.04702172428369522,
      "learning_rate": 0.00011011764705882352,
      "loss": 0.0219,
      "step": 3730
    },
    {
      "epoch": 33.31,
      "grad_norm": 0.05131182819604874,
      "learning_rate": 0.00011005882352941176,
      "loss": 0.0214,
      "step": 3731
    },
    {
      "epoch": 33.32,
      "grad_norm": 0.06191159412264824,
      "learning_rate": 0.00010999999999999998,
      "loss": 0.0216,
      "step": 3732
    },
    {
      "epoch": 33.33,
      "grad_norm": 0.053905926644802094,
      "learning_rate": 0.00010994117647058822,
      "loss": 0.0185,
      "step": 3733
    },
    {
      "epoch": 33.34,
      "grad_norm": 0.08019814640283585,
      "learning_rate": 0.00010988235294117646,
      "loss": 0.0232,
      "step": 3734
    },
    {
      "epoch": 33.35,
      "grad_norm": 0.04236734285950661,
      "learning_rate": 0.0001098235294117647,
      "loss": 0.0223,
      "step": 3735
    },
    {
      "epoch": 33.36,
      "grad_norm": 0.0380890928208828,
      "learning_rate": 0.00010976470588235294,
      "loss": 0.0195,
      "step": 3736
    },
    {
      "epoch": 33.37,
      "grad_norm": 0.05268734320998192,
      "learning_rate": 0.00010970588235294117,
      "loss": 0.0206,
      "step": 3737
    },
    {
      "epoch": 33.38,
      "grad_norm": 0.049979280680418015,
      "learning_rate": 0.0001096470588235294,
      "loss": 0.018,
      "step": 3738
    },
    {
      "epoch": 33.38,
      "grad_norm": 0.05195252224802971,
      "learning_rate": 0.00010958823529411764,
      "loss": 0.0166,
      "step": 3739
    },
    {
      "epoch": 33.39,
      "grad_norm": 0.12251890450716019,
      "learning_rate": 0.00010952941176470587,
      "loss": 0.0271,
      "step": 3740
    },
    {
      "epoch": 33.4,
      "grad_norm": 0.04267783463001251,
      "learning_rate": 0.00010947058823529411,
      "loss": 0.0182,
      "step": 3741
    },
    {
      "epoch": 33.41,
      "grad_norm": 0.0516081228852272,
      "learning_rate": 0.00010941176470588234,
      "loss": 0.0212,
      "step": 3742
    },
    {
      "epoch": 33.42,
      "grad_norm": 0.05559803172945976,
      "learning_rate": 0.00010935294117647057,
      "loss": 0.0176,
      "step": 3743
    },
    {
      "epoch": 33.43,
      "grad_norm": 0.059197813272476196,
      "learning_rate": 0.00010929411764705882,
      "loss": 0.0228,
      "step": 3744
    },
    {
      "epoch": 33.44,
      "grad_norm": 0.05572134256362915,
      "learning_rate": 0.00010923529411764706,
      "loss": 0.0187,
      "step": 3745
    },
    {
      "epoch": 33.45,
      "grad_norm": 0.10062196105718613,
      "learning_rate": 0.00010917647058823528,
      "loss": 0.0305,
      "step": 3746
    },
    {
      "epoch": 33.46,
      "grad_norm": 0.04885195568203926,
      "learning_rate": 0.00010911764705882352,
      "loss": 0.0215,
      "step": 3747
    },
    {
      "epoch": 33.46,
      "grad_norm": 0.04990047216415405,
      "learning_rate": 0.00010905882352941175,
      "loss": 0.0212,
      "step": 3748
    },
    {
      "epoch": 33.47,
      "grad_norm": 0.055074553936719894,
      "learning_rate": 0.00010899999999999999,
      "loss": 0.0226,
      "step": 3749
    },
    {
      "epoch": 33.48,
      "grad_norm": 0.05209925398230553,
      "learning_rate": 0.00010894117647058824,
      "loss": 0.0224,
      "step": 3750
    },
    {
      "epoch": 33.49,
      "grad_norm": 0.04867013543844223,
      "learning_rate": 0.00010888235294117645,
      "loss": 0.0178,
      "step": 3751
    },
    {
      "epoch": 33.5,
      "grad_norm": 0.07351130992174149,
      "learning_rate": 0.0001088235294117647,
      "loss": 0.0243,
      "step": 3752
    },
    {
      "epoch": 33.51,
      "grad_norm": 0.07198871672153473,
      "learning_rate": 0.00010876470588235293,
      "loss": 0.0216,
      "step": 3753
    },
    {
      "epoch": 33.52,
      "grad_norm": 0.04132643714547157,
      "learning_rate": 0.00010870588235294117,
      "loss": 0.0178,
      "step": 3754
    },
    {
      "epoch": 33.53,
      "grad_norm": 0.04665254428982735,
      "learning_rate": 0.00010864705882352941,
      "loss": 0.0202,
      "step": 3755
    },
    {
      "epoch": 33.54,
      "grad_norm": 0.05033242329955101,
      "learning_rate": 0.00010858823529411763,
      "loss": 0.0192,
      "step": 3756
    },
    {
      "epoch": 33.54,
      "grad_norm": 0.04282531514763832,
      "learning_rate": 0.00010852941176470587,
      "loss": 0.0175,
      "step": 3757
    },
    {
      "epoch": 33.55,
      "grad_norm": 0.08175268769264221,
      "learning_rate": 0.0001084705882352941,
      "loss": 0.0231,
      "step": 3758
    },
    {
      "epoch": 33.56,
      "grad_norm": 0.10714887082576752,
      "learning_rate": 0.00010841176470588235,
      "loss": 0.0389,
      "step": 3759
    },
    {
      "epoch": 33.57,
      "grad_norm": 0.04657446965575218,
      "learning_rate": 0.00010835294117647059,
      "loss": 0.022,
      "step": 3760
    },
    {
      "epoch": 33.58,
      "grad_norm": 0.04799662530422211,
      "learning_rate": 0.0001082941176470588,
      "loss": 0.0202,
      "step": 3761
    },
    {
      "epoch": 33.59,
      "grad_norm": 0.05010412260890007,
      "learning_rate": 0.00010823529411764705,
      "loss": 0.0202,
      "step": 3762
    },
    {
      "epoch": 33.6,
      "grad_norm": 0.04735333099961281,
      "learning_rate": 0.00010817647058823529,
      "loss": 0.0205,
      "step": 3763
    },
    {
      "epoch": 33.61,
      "grad_norm": 0.04919218271970749,
      "learning_rate": 0.00010811764705882352,
      "loss": 0.0154,
      "step": 3764
    },
    {
      "epoch": 33.62,
      "grad_norm": 0.09689608961343765,
      "learning_rate": 0.00010805882352941175,
      "loss": 0.0239,
      "step": 3765
    },
    {
      "epoch": 33.62,
      "grad_norm": 0.04646177217364311,
      "learning_rate": 0.00010799999999999998,
      "loss": 0.0204,
      "step": 3766
    },
    {
      "epoch": 33.63,
      "grad_norm": 0.04333686828613281,
      "learning_rate": 0.00010794117647058823,
      "loss": 0.0203,
      "step": 3767
    },
    {
      "epoch": 33.64,
      "grad_norm": 0.047065041959285736,
      "learning_rate": 0.00010788235294117647,
      "loss": 0.0178,
      "step": 3768
    },
    {
      "epoch": 33.65,
      "grad_norm": 0.04462426155805588,
      "learning_rate": 0.0001078235294117647,
      "loss": 0.017,
      "step": 3769
    },
    {
      "epoch": 33.66,
      "grad_norm": 0.052010223269462585,
      "learning_rate": 0.00010776470588235293,
      "loss": 0.0178,
      "step": 3770
    },
    {
      "epoch": 33.67,
      "grad_norm": 0.10874468088150024,
      "learning_rate": 0.00010770588235294116,
      "loss": 0.0263,
      "step": 3771
    },
    {
      "epoch": 33.68,
      "grad_norm": 0.040056925266981125,
      "learning_rate": 0.0001076470588235294,
      "loss": 0.0153,
      "step": 3772
    },
    {
      "epoch": 33.69,
      "grad_norm": 0.044297847896814346,
      "learning_rate": 0.00010758823529411765,
      "loss": 0.0181,
      "step": 3773
    },
    {
      "epoch": 33.7,
      "grad_norm": 0.04572479799389839,
      "learning_rate": 0.00010752941176470588,
      "loss": 0.0183,
      "step": 3774
    },
    {
      "epoch": 33.71,
      "grad_norm": 0.04739643633365631,
      "learning_rate": 0.0001074705882352941,
      "loss": 0.021,
      "step": 3775
    },
    {
      "epoch": 33.71,
      "grad_norm": 0.0541643463075161,
      "learning_rate": 0.00010741176470588234,
      "loss": 0.0211,
      "step": 3776
    },
    {
      "epoch": 33.72,
      "grad_norm": 0.08154944330453873,
      "learning_rate": 0.00010735294117647058,
      "loss": 0.0209,
      "step": 3777
    },
    {
      "epoch": 33.73,
      "grad_norm": 0.12991441786289215,
      "learning_rate": 0.00010729411764705882,
      "loss": 0.0265,
      "step": 3778
    },
    {
      "epoch": 33.74,
      "grad_norm": 0.046380940824747086,
      "learning_rate": 0.00010723529411764704,
      "loss": 0.0186,
      "step": 3779
    },
    {
      "epoch": 33.75,
      "grad_norm": 0.049973901361227036,
      "learning_rate": 0.00010717647058823528,
      "loss": 0.0189,
      "step": 3780
    },
    {
      "epoch": 33.76,
      "grad_norm": 0.04776526615023613,
      "learning_rate": 0.00010711764705882353,
      "loss": 0.019,
      "step": 3781
    },
    {
      "epoch": 33.77,
      "grad_norm": 0.0652221217751503,
      "learning_rate": 0.00010705882352941176,
      "loss": 0.0206,
      "step": 3782
    },
    {
      "epoch": 33.78,
      "grad_norm": 0.04913077875971794,
      "learning_rate": 0.000107,
      "loss": 0.018,
      "step": 3783
    },
    {
      "epoch": 33.79,
      "grad_norm": 0.07298099994659424,
      "learning_rate": 0.00010694117647058822,
      "loss": 0.0242,
      "step": 3784
    },
    {
      "epoch": 33.79,
      "grad_norm": 0.044130854308605194,
      "learning_rate": 0.00010688235294117646,
      "loss": 0.0181,
      "step": 3785
    },
    {
      "epoch": 33.8,
      "grad_norm": 0.04705604165792465,
      "learning_rate": 0.0001068235294117647,
      "loss": 0.0195,
      "step": 3786
    },
    {
      "epoch": 33.81,
      "grad_norm": 0.055822499096393585,
      "learning_rate": 0.00010676470588235293,
      "loss": 0.0199,
      "step": 3787
    },
    {
      "epoch": 33.82,
      "grad_norm": 0.06081099435687065,
      "learning_rate": 0.00010670588235294118,
      "loss": 0.0249,
      "step": 3788
    },
    {
      "epoch": 33.83,
      "grad_norm": 0.0651383101940155,
      "learning_rate": 0.00010664705882352939,
      "loss": 0.0208,
      "step": 3789
    },
    {
      "epoch": 33.84,
      "grad_norm": 0.09618881344795227,
      "learning_rate": 0.00010658823529411764,
      "loss": 0.0292,
      "step": 3790
    },
    {
      "epoch": 33.85,
      "grad_norm": 0.04474983736872673,
      "learning_rate": 0.00010652941176470588,
      "loss": 0.0187,
      "step": 3791
    },
    {
      "epoch": 33.86,
      "grad_norm": 0.04621360823512077,
      "learning_rate": 0.00010647058823529411,
      "loss": 0.0181,
      "step": 3792
    },
    {
      "epoch": 33.87,
      "grad_norm": 0.05100925266742706,
      "learning_rate": 0.00010641176470588235,
      "loss": 0.0178,
      "step": 3793
    },
    {
      "epoch": 33.88,
      "grad_norm": 0.04568636044859886,
      "learning_rate": 0.00010635294117647057,
      "loss": 0.0188,
      "step": 3794
    },
    {
      "epoch": 33.88,
      "grad_norm": 0.05396410822868347,
      "learning_rate": 0.00010629411764705881,
      "loss": 0.0188,
      "step": 3795
    },
    {
      "epoch": 33.89,
      "grad_norm": 0.07931025326251984,
      "learning_rate": 0.00010623529411764705,
      "loss": 0.0261,
      "step": 3796
    },
    {
      "epoch": 33.9,
      "grad_norm": 0.04671921953558922,
      "learning_rate": 0.00010617647058823528,
      "loss": 0.019,
      "step": 3797
    },
    {
      "epoch": 33.91,
      "grad_norm": 0.049630068242549896,
      "learning_rate": 0.00010611764705882351,
      "loss": 0.0197,
      "step": 3798
    },
    {
      "epoch": 33.92,
      "grad_norm": 0.05084071308374405,
      "learning_rate": 0.00010605882352941176,
      "loss": 0.0215,
      "step": 3799
    },
    {
      "epoch": 33.93,
      "grad_norm": 0.050024282187223434,
      "learning_rate": 0.00010599999999999999,
      "loss": 0.0166,
      "step": 3800
    },
    {
      "epoch": 33.93,
      "eval_cer": 0.03149887436906413,
      "eval_loss": 0.211999773979187,
      "eval_runtime": 22.198,
      "eval_samples_per_second": 119.02,
      "eval_steps_per_second": 1.892,
      "eval_wer": 0.1032130107100357,
      "step": 3800
    },
    {
      "epoch": 33.94,
      "grad_norm": 0.054959606379270554,
      "learning_rate": 0.00010594117647058823,
      "loss": 0.0214,
      "step": 3801
    },
    {
      "epoch": 33.95,
      "grad_norm": 0.08315333724021912,
      "learning_rate": 0.00010588235294117647,
      "loss": 0.0215,
      "step": 3802
    },
    {
      "epoch": 33.96,
      "grad_norm": 0.06588681787252426,
      "learning_rate": 0.00010582352941176469,
      "loss": 0.0231,
      "step": 3803
    },
    {
      "epoch": 33.96,
      "grad_norm": 0.055088456720113754,
      "learning_rate": 0.00010576470588235293,
      "loss": 0.0239,
      "step": 3804
    },
    {
      "epoch": 33.97,
      "grad_norm": 0.04421170800924301,
      "learning_rate": 0.00010570588235294116,
      "loss": 0.0185,
      "step": 3805
    },
    {
      "epoch": 33.98,
      "grad_norm": 0.051479555666446686,
      "learning_rate": 0.00010564705882352941,
      "loss": 0.0186,
      "step": 3806
    },
    {
      "epoch": 33.99,
      "grad_norm": 0.0651811733841896,
      "learning_rate": 0.00010558823529411765,
      "loss": 0.0169,
      "step": 3807
    },
    {
      "epoch": 34.0,
      "grad_norm": 0.08760152012109756,
      "learning_rate": 0.00010552941176470587,
      "loss": 0.0221,
      "step": 3808
    },
    {
      "epoch": 34.01,
      "grad_norm": 0.04792863503098488,
      "learning_rate": 0.00010547058823529411,
      "loss": 0.0193,
      "step": 3809
    },
    {
      "epoch": 34.02,
      "grad_norm": 0.04409262537956238,
      "learning_rate": 0.00010541176470588234,
      "loss": 0.0213,
      "step": 3810
    },
    {
      "epoch": 34.03,
      "grad_norm": 0.04361652955412865,
      "learning_rate": 0.00010535294117647058,
      "loss": 0.0193,
      "step": 3811
    },
    {
      "epoch": 34.04,
      "grad_norm": 0.05131509155035019,
      "learning_rate": 0.0001052941176470588,
      "loss": 0.0193,
      "step": 3812
    },
    {
      "epoch": 34.04,
      "grad_norm": 0.04921355098485947,
      "learning_rate": 0.00010523529411764704,
      "loss": 0.0188,
      "step": 3813
    },
    {
      "epoch": 34.05,
      "grad_norm": 0.060709960758686066,
      "learning_rate": 0.00010517647058823529,
      "loss": 0.0201,
      "step": 3814
    },
    {
      "epoch": 34.06,
      "grad_norm": 0.07292447984218597,
      "learning_rate": 0.00010511764705882352,
      "loss": 0.0225,
      "step": 3815
    },
    {
      "epoch": 34.07,
      "grad_norm": 0.05245755612850189,
      "learning_rate": 0.00010505882352941176,
      "loss": 0.0214,
      "step": 3816
    },
    {
      "epoch": 34.08,
      "grad_norm": 0.05187506601214409,
      "learning_rate": 0.00010499999999999999,
      "loss": 0.0223,
      "step": 3817
    },
    {
      "epoch": 34.09,
      "grad_norm": 0.04697016626596451,
      "learning_rate": 0.00010494117647058822,
      "loss": 0.0153,
      "step": 3818
    },
    {
      "epoch": 34.1,
      "grad_norm": 0.05046670138835907,
      "learning_rate": 0.00010488235294117646,
      "loss": 0.0169,
      "step": 3819
    },
    {
      "epoch": 34.11,
      "grad_norm": 0.06887195259332657,
      "learning_rate": 0.0001048235294117647,
      "loss": 0.0198,
      "step": 3820
    },
    {
      "epoch": 34.12,
      "grad_norm": 0.10050436854362488,
      "learning_rate": 0.00010476470588235294,
      "loss": 0.03,
      "step": 3821
    },
    {
      "epoch": 34.12,
      "grad_norm": 0.04660368338227272,
      "learning_rate": 0.00010470588235294117,
      "loss": 0.016,
      "step": 3822
    },
    {
      "epoch": 34.13,
      "grad_norm": 0.050970591604709625,
      "learning_rate": 0.0001046470588235294,
      "loss": 0.0194,
      "step": 3823
    },
    {
      "epoch": 34.14,
      "grad_norm": 0.0501236766576767,
      "learning_rate": 0.00010458823529411764,
      "loss": 0.0207,
      "step": 3824
    },
    {
      "epoch": 34.15,
      "grad_norm": 0.0496857650578022,
      "learning_rate": 0.00010452941176470588,
      "loss": 0.0169,
      "step": 3825
    },
    {
      "epoch": 34.16,
      "grad_norm": 0.04786838963627815,
      "learning_rate": 0.0001044705882352941,
      "loss": 0.0179,
      "step": 3826
    },
    {
      "epoch": 34.17,
      "grad_norm": 0.07660852372646332,
      "learning_rate": 0.00010441176470588234,
      "loss": 0.0208,
      "step": 3827
    },
    {
      "epoch": 34.18,
      "grad_norm": 0.04219606891274452,
      "learning_rate": 0.00010435294117647057,
      "loss": 0.0193,
      "step": 3828
    },
    {
      "epoch": 34.19,
      "grad_norm": 0.0388142429292202,
      "learning_rate": 0.00010429411764705882,
      "loss": 0.0164,
      "step": 3829
    },
    {
      "epoch": 34.2,
      "grad_norm": 0.04581207409501076,
      "learning_rate": 0.00010423529411764706,
      "loss": 0.0192,
      "step": 3830
    },
    {
      "epoch": 34.21,
      "grad_norm": 0.045865558087825775,
      "learning_rate": 0.00010417647058823528,
      "loss": 0.0194,
      "step": 3831
    },
    {
      "epoch": 34.21,
      "grad_norm": 0.052915655076503754,
      "learning_rate": 0.00010411764705882352,
      "loss": 0.0181,
      "step": 3832
    },
    {
      "epoch": 34.22,
      "grad_norm": 0.09456345438957214,
      "learning_rate": 0.00010405882352941175,
      "loss": 0.0266,
      "step": 3833
    },
    {
      "epoch": 34.23,
      "grad_norm": 0.04623633995652199,
      "learning_rate": 0.000104,
      "loss": 0.0183,
      "step": 3834
    },
    {
      "epoch": 34.24,
      "grad_norm": 0.05617140978574753,
      "learning_rate": 0.00010394117647058824,
      "loss": 0.0208,
      "step": 3835
    },
    {
      "epoch": 34.25,
      "grad_norm": 0.044178467243909836,
      "learning_rate": 0.00010388235294117645,
      "loss": 0.0167,
      "step": 3836
    },
    {
      "epoch": 34.26,
      "grad_norm": 0.055569615215063095,
      "learning_rate": 0.0001038235294117647,
      "loss": 0.0201,
      "step": 3837
    },
    {
      "epoch": 34.27,
      "grad_norm": 0.04524802044034004,
      "learning_rate": 0.00010376470588235293,
      "loss": 0.0171,
      "step": 3838
    },
    {
      "epoch": 34.28,
      "grad_norm": 0.06489437073469162,
      "learning_rate": 0.00010370588235294117,
      "loss": 0.0204,
      "step": 3839
    },
    {
      "epoch": 34.29,
      "grad_norm": 0.0554768443107605,
      "learning_rate": 0.00010364705882352941,
      "loss": 0.0189,
      "step": 3840
    },
    {
      "epoch": 34.29,
      "grad_norm": 0.046934958547353745,
      "learning_rate": 0.00010358823529411763,
      "loss": 0.0205,
      "step": 3841
    },
    {
      "epoch": 34.3,
      "grad_norm": 0.057572390884160995,
      "learning_rate": 0.00010352941176470587,
      "loss": 0.0181,
      "step": 3842
    },
    {
      "epoch": 34.31,
      "grad_norm": 0.05205719172954559,
      "learning_rate": 0.00010347058823529412,
      "loss": 0.0206,
      "step": 3843
    },
    {
      "epoch": 34.32,
      "grad_norm": 0.04701286554336548,
      "learning_rate": 0.00010341176470588235,
      "loss": 0.016,
      "step": 3844
    },
    {
      "epoch": 34.33,
      "grad_norm": 0.06379012763500214,
      "learning_rate": 0.00010335294117647058,
      "loss": 0.0227,
      "step": 3845
    },
    {
      "epoch": 34.34,
      "grad_norm": 0.07218367606401443,
      "learning_rate": 0.0001032941176470588,
      "loss": 0.0257,
      "step": 3846
    },
    {
      "epoch": 34.35,
      "grad_norm": 0.04608406499028206,
      "learning_rate": 0.00010323529411764705,
      "loss": 0.0177,
      "step": 3847
    },
    {
      "epoch": 34.36,
      "grad_norm": 0.04197123646736145,
      "learning_rate": 0.00010317647058823529,
      "loss": 0.0174,
      "step": 3848
    },
    {
      "epoch": 34.37,
      "grad_norm": 0.04382241144776344,
      "learning_rate": 0.00010311764705882352,
      "loss": 0.0203,
      "step": 3849
    },
    {
      "epoch": 34.38,
      "grad_norm": 0.03994964063167572,
      "learning_rate": 0.00010305882352941175,
      "loss": 0.0171,
      "step": 3850
    },
    {
      "epoch": 34.38,
      "grad_norm": 0.0624520443379879,
      "learning_rate": 0.00010299999999999998,
      "loss": 0.0244,
      "step": 3851
    },
    {
      "epoch": 34.39,
      "grad_norm": 0.07916943728923798,
      "learning_rate": 0.00010294117647058823,
      "loss": 0.0209,
      "step": 3852
    },
    {
      "epoch": 34.4,
      "grad_norm": 0.04561072960495949,
      "learning_rate": 0.00010288235294117647,
      "loss": 0.0223,
      "step": 3853
    },
    {
      "epoch": 34.41,
      "grad_norm": 0.04165709391236305,
      "learning_rate": 0.0001028235294117647,
      "loss": 0.0186,
      "step": 3854
    },
    {
      "epoch": 34.42,
      "grad_norm": 0.04192725569009781,
      "learning_rate": 0.00010276470588235293,
      "loss": 0.0205,
      "step": 3855
    },
    {
      "epoch": 34.43,
      "grad_norm": 0.06156204640865326,
      "learning_rate": 0.00010270588235294116,
      "loss": 0.0216,
      "step": 3856
    },
    {
      "epoch": 34.44,
      "grad_norm": 0.043788302689790726,
      "learning_rate": 0.0001026470588235294,
      "loss": 0.0158,
      "step": 3857
    },
    {
      "epoch": 34.45,
      "grad_norm": 0.08256421983242035,
      "learning_rate": 0.00010258823529411765,
      "loss": 0.0236,
      "step": 3858
    },
    {
      "epoch": 34.46,
      "grad_norm": 0.04196475073695183,
      "learning_rate": 0.00010252941176470586,
      "loss": 0.0194,
      "step": 3859
    },
    {
      "epoch": 34.46,
      "grad_norm": 0.050266288220882416,
      "learning_rate": 0.0001024705882352941,
      "loss": 0.0224,
      "step": 3860
    },
    {
      "epoch": 34.47,
      "grad_norm": 0.04009345546364784,
      "learning_rate": 0.00010241176470588235,
      "loss": 0.0143,
      "step": 3861
    },
    {
      "epoch": 34.48,
      "grad_norm": 0.05010367929935455,
      "learning_rate": 0.00010235294117647058,
      "loss": 0.0196,
      "step": 3862
    },
    {
      "epoch": 34.49,
      "grad_norm": 0.04665965214371681,
      "learning_rate": 0.00010229411764705882,
      "loss": 0.016,
      "step": 3863
    },
    {
      "epoch": 34.5,
      "grad_norm": 0.06397382915019989,
      "learning_rate": 0.00010223529411764704,
      "loss": 0.0222,
      "step": 3864
    },
    {
      "epoch": 34.51,
      "grad_norm": 0.07012230902910233,
      "learning_rate": 0.00010217647058823528,
      "loss": 0.0215,
      "step": 3865
    },
    {
      "epoch": 34.52,
      "grad_norm": 0.05256851390004158,
      "learning_rate": 0.00010211764705882353,
      "loss": 0.0208,
      "step": 3866
    },
    {
      "epoch": 34.53,
      "grad_norm": 0.051373377442359924,
      "learning_rate": 0.00010205882352941176,
      "loss": 0.0203,
      "step": 3867
    },
    {
      "epoch": 34.54,
      "grad_norm": 0.049756649881601334,
      "learning_rate": 0.000102,
      "loss": 0.0197,
      "step": 3868
    },
    {
      "epoch": 34.54,
      "grad_norm": 0.04557337984442711,
      "learning_rate": 0.00010194117647058822,
      "loss": 0.0194,
      "step": 3869
    },
    {
      "epoch": 34.55,
      "grad_norm": 0.05897529423236847,
      "learning_rate": 0.00010188235294117646,
      "loss": 0.0236,
      "step": 3870
    },
    {
      "epoch": 34.56,
      "grad_norm": 0.12045527994632721,
      "learning_rate": 0.0001018235294117647,
      "loss": 0.0208,
      "step": 3871
    },
    {
      "epoch": 34.57,
      "grad_norm": 0.041995007544755936,
      "learning_rate": 0.00010176470588235293,
      "loss": 0.0209,
      "step": 3872
    },
    {
      "epoch": 34.58,
      "grad_norm": 0.03895152360200882,
      "learning_rate": 0.00010170588235294116,
      "loss": 0.0174,
      "step": 3873
    },
    {
      "epoch": 34.59,
      "grad_norm": 0.048664640635252,
      "learning_rate": 0.00010164705882352939,
      "loss": 0.0196,
      "step": 3874
    },
    {
      "epoch": 34.6,
      "grad_norm": 0.04239026457071304,
      "learning_rate": 0.00010158823529411764,
      "loss": 0.016,
      "step": 3875
    },
    {
      "epoch": 34.61,
      "grad_norm": 0.054239142686128616,
      "learning_rate": 0.00010152941176470588,
      "loss": 0.0188,
      "step": 3876
    },
    {
      "epoch": 34.62,
      "grad_norm": 0.09040377289056778,
      "learning_rate": 0.00010147058823529411,
      "loss": 0.0234,
      "step": 3877
    },
    {
      "epoch": 34.62,
      "grad_norm": 0.0430770069360733,
      "learning_rate": 0.00010141176470588234,
      "loss": 0.0187,
      "step": 3878
    },
    {
      "epoch": 34.63,
      "grad_norm": 0.04614727944135666,
      "learning_rate": 0.00010135294117647058,
      "loss": 0.019,
      "step": 3879
    },
    {
      "epoch": 34.64,
      "grad_norm": 0.05060048773884773,
      "learning_rate": 0.00010129411764705881,
      "loss": 0.0212,
      "step": 3880
    },
    {
      "epoch": 34.65,
      "grad_norm": 0.047570060938596725,
      "learning_rate": 0.00010123529411764706,
      "loss": 0.0186,
      "step": 3881
    },
    {
      "epoch": 34.66,
      "grad_norm": 0.05347687751054764,
      "learning_rate": 0.0001011764705882353,
      "loss": 0.0183,
      "step": 3882
    },
    {
      "epoch": 34.67,
      "grad_norm": 0.11090794950723648,
      "learning_rate": 0.00010111764705882352,
      "loss": 0.0206,
      "step": 3883
    },
    {
      "epoch": 34.68,
      "grad_norm": 0.03908548131585121,
      "learning_rate": 0.00010105882352941176,
      "loss": 0.019,
      "step": 3884
    },
    {
      "epoch": 34.69,
      "grad_norm": 0.05485350266098976,
      "learning_rate": 0.00010099999999999999,
      "loss": 0.0204,
      "step": 3885
    },
    {
      "epoch": 34.7,
      "grad_norm": 0.04888375476002693,
      "learning_rate": 0.00010094117647058823,
      "loss": 0.0162,
      "step": 3886
    },
    {
      "epoch": 34.71,
      "grad_norm": 0.054330479353666306,
      "learning_rate": 0.00010088235294117648,
      "loss": 0.0217,
      "step": 3887
    },
    {
      "epoch": 34.71,
      "grad_norm": 0.04673108458518982,
      "learning_rate": 0.00010082352941176469,
      "loss": 0.0184,
      "step": 3888
    },
    {
      "epoch": 34.72,
      "grad_norm": 0.0918075367808342,
      "learning_rate": 0.00010076470588235294,
      "loss": 0.0294,
      "step": 3889
    },
    {
      "epoch": 34.73,
      "grad_norm": 0.10332155972719193,
      "learning_rate": 0.00010070588235294116,
      "loss": 0.0216,
      "step": 3890
    },
    {
      "epoch": 34.74,
      "grad_norm": 0.043055880814790726,
      "learning_rate": 0.00010064705882352941,
      "loss": 0.0203,
      "step": 3891
    },
    {
      "epoch": 34.75,
      "grad_norm": 0.04952540248632431,
      "learning_rate": 0.00010058823529411762,
      "loss": 0.0176,
      "step": 3892
    },
    {
      "epoch": 34.76,
      "grad_norm": 0.040237873792648315,
      "learning_rate": 0.00010052941176470587,
      "loss": 0.0179,
      "step": 3893
    },
    {
      "epoch": 34.77,
      "grad_norm": 0.052950188517570496,
      "learning_rate": 0.00010047058823529411,
      "loss": 0.0197,
      "step": 3894
    },
    {
      "epoch": 34.78,
      "grad_norm": 0.05970821902155876,
      "learning_rate": 0.00010041176470588234,
      "loss": 0.0201,
      "step": 3895
    },
    {
      "epoch": 34.79,
      "grad_norm": 0.06343962252140045,
      "learning_rate": 0.00010035294117647058,
      "loss": 0.0189,
      "step": 3896
    },
    {
      "epoch": 34.79,
      "grad_norm": 0.04404140263795853,
      "learning_rate": 0.00010029411764705881,
      "loss": 0.0193,
      "step": 3897
    },
    {
      "epoch": 34.8,
      "grad_norm": 0.04577597603201866,
      "learning_rate": 0.00010023529411764704,
      "loss": 0.0213,
      "step": 3898
    },
    {
      "epoch": 34.81,
      "grad_norm": 0.04664938896894455,
      "learning_rate": 0.00010017647058823529,
      "loss": 0.0199,
      "step": 3899
    },
    {
      "epoch": 34.82,
      "grad_norm": 0.05409785360097885,
      "learning_rate": 0.00010011764705882352,
      "loss": 0.0212,
      "step": 3900
    },
    {
      "epoch": 34.82,
      "eval_cer": 0.030025594479710455,
      "eval_loss": 0.2020832896232605,
      "eval_runtime": 22.9436,
      "eval_samples_per_second": 115.152,
      "eval_steps_per_second": 1.831,
      "eval_wer": 0.10029750099166997,
      "step": 3900
    },
    {
      "epoch": 34.83,
      "grad_norm": 0.0607311986386776,
      "learning_rate": 0.00010005882352941176,
      "loss": 0.019,
      "step": 3901
    },
    {
      "epoch": 34.84,
      "grad_norm": 0.1514507532119751,
      "learning_rate": 9.999999999999999e-05,
      "loss": 0.0278,
      "step": 3902
    },
    {
      "epoch": 34.85,
      "grad_norm": 0.047490138560533524,
      "learning_rate": 9.994117647058822e-05,
      "loss": 0.0189,
      "step": 3903
    },
    {
      "epoch": 34.86,
      "grad_norm": 0.04280892014503479,
      "learning_rate": 9.988235294117646e-05,
      "loss": 0.0183,
      "step": 3904
    },
    {
      "epoch": 34.87,
      "grad_norm": 0.04812157154083252,
      "learning_rate": 9.982352941176471e-05,
      "loss": 0.0237,
      "step": 3905
    },
    {
      "epoch": 34.88,
      "grad_norm": 0.052132874727249146,
      "learning_rate": 9.976470588235292e-05,
      "loss": 0.02,
      "step": 3906
    },
    {
      "epoch": 34.88,
      "grad_norm": 0.05362263694405556,
      "learning_rate": 9.970588235294117e-05,
      "loss": 0.0208,
      "step": 3907
    },
    {
      "epoch": 34.89,
      "grad_norm": 0.1373939961194992,
      "learning_rate": 9.96470588235294e-05,
      "loss": 0.0339,
      "step": 3908
    },
    {
      "epoch": 34.9,
      "grad_norm": 0.04172702878713608,
      "learning_rate": 9.958823529411764e-05,
      "loss": 0.0224,
      "step": 3909
    },
    {
      "epoch": 34.91,
      "grad_norm": 0.04621425271034241,
      "learning_rate": 9.952941176470588e-05,
      "loss": 0.0185,
      "step": 3910
    },
    {
      "epoch": 34.92,
      "grad_norm": 0.04608256742358208,
      "learning_rate": 9.94705882352941e-05,
      "loss": 0.0202,
      "step": 3911
    },
    {
      "epoch": 34.93,
      "grad_norm": 0.05011044070124626,
      "learning_rate": 9.941176470588234e-05,
      "loss": 0.0204,
      "step": 3912
    },
    {
      "epoch": 34.94,
      "grad_norm": 0.04857317730784416,
      "learning_rate": 9.935294117647057e-05,
      "loss": 0.0207,
      "step": 3913
    },
    {
      "epoch": 34.95,
      "grad_norm": 0.07684504985809326,
      "learning_rate": 9.929411764705882e-05,
      "loss": 0.0235,
      "step": 3914
    },
    {
      "epoch": 34.96,
      "grad_norm": 0.04883652925491333,
      "learning_rate": 9.923529411764706e-05,
      "loss": 0.0185,
      "step": 3915
    },
    {
      "epoch": 34.96,
      "grad_norm": 0.04158344864845276,
      "learning_rate": 9.917647058823528e-05,
      "loss": 0.0207,
      "step": 3916
    },
    {
      "epoch": 34.97,
      "grad_norm": 0.04643884673714638,
      "learning_rate": 9.911764705882352e-05,
      "loss": 0.0191,
      "step": 3917
    },
    {
      "epoch": 34.98,
      "grad_norm": 0.046996667981147766,
      "learning_rate": 9.905882352941175e-05,
      "loss": 0.0158,
      "step": 3918
    },
    {
      "epoch": 34.99,
      "grad_norm": 0.058684635907411575,
      "learning_rate": 9.9e-05,
      "loss": 0.0222,
      "step": 3919
    },
    {
      "epoch": 35.0,
      "grad_norm": 0.21057476103305817,
      "learning_rate": 9.894117647058824e-05,
      "loss": 0.0291,
      "step": 3920
    },
    {
      "epoch": 35.01,
      "grad_norm": 0.04154033213853836,
      "learning_rate": 9.888235294117645e-05,
      "loss": 0.0188,
      "step": 3921
    },
    {
      "epoch": 35.02,
      "grad_norm": 0.04205253720283508,
      "learning_rate": 9.88235294117647e-05,
      "loss": 0.016,
      "step": 3922
    },
    {
      "epoch": 35.03,
      "grad_norm": 0.05432457849383354,
      "learning_rate": 9.876470588235294e-05,
      "loss": 0.0207,
      "step": 3923
    },
    {
      "epoch": 35.04,
      "grad_norm": 0.0533113107085228,
      "learning_rate": 9.870588235294117e-05,
      "loss": 0.0192,
      "step": 3924
    },
    {
      "epoch": 35.04,
      "grad_norm": 0.051685433834791183,
      "learning_rate": 9.86470588235294e-05,
      "loss": 0.0204,
      "step": 3925
    },
    {
      "epoch": 35.05,
      "grad_norm": 0.06034713238477707,
      "learning_rate": 9.858823529411763e-05,
      "loss": 0.0193,
      "step": 3926
    },
    {
      "epoch": 35.06,
      "grad_norm": 0.0914221778512001,
      "learning_rate": 9.852941176470587e-05,
      "loss": 0.027,
      "step": 3927
    },
    {
      "epoch": 35.07,
      "grad_norm": 0.04747364670038223,
      "learning_rate": 9.847058823529412e-05,
      "loss": 0.0191,
      "step": 3928
    },
    {
      "epoch": 35.08,
      "grad_norm": 0.0516696535050869,
      "learning_rate": 9.841176470588235e-05,
      "loss": 0.0226,
      "step": 3929
    },
    {
      "epoch": 35.09,
      "grad_norm": 0.051890451461076736,
      "learning_rate": 9.835294117647058e-05,
      "loss": 0.0227,
      "step": 3930
    },
    {
      "epoch": 35.1,
      "grad_norm": 0.06238153949379921,
      "learning_rate": 9.829411764705881e-05,
      "loss": 0.024,
      "step": 3931
    },
    {
      "epoch": 35.11,
      "grad_norm": 0.06373263150453568,
      "learning_rate": 9.823529411764705e-05,
      "loss": 0.0175,
      "step": 3932
    },
    {
      "epoch": 35.12,
      "grad_norm": 0.08236797899007797,
      "learning_rate": 9.81764705882353e-05,
      "loss": 0.0239,
      "step": 3933
    },
    {
      "epoch": 35.12,
      "grad_norm": 0.04389422759413719,
      "learning_rate": 9.811764705882352e-05,
      "loss": 0.0193,
      "step": 3934
    },
    {
      "epoch": 35.13,
      "grad_norm": 0.04222368448972702,
      "learning_rate": 9.805882352941175e-05,
      "loss": 0.0163,
      "step": 3935
    },
    {
      "epoch": 35.14,
      "grad_norm": 0.04606059566140175,
      "learning_rate": 9.799999999999998e-05,
      "loss": 0.0179,
      "step": 3936
    },
    {
      "epoch": 35.15,
      "grad_norm": 0.04678035527467728,
      "learning_rate": 9.794117647058823e-05,
      "loss": 0.0196,
      "step": 3937
    },
    {
      "epoch": 35.16,
      "grad_norm": 0.05267263203859329,
      "learning_rate": 9.788235294117647e-05,
      "loss": 0.0206,
      "step": 3938
    },
    {
      "epoch": 35.17,
      "grad_norm": 0.08604593575000763,
      "learning_rate": 9.782352941176469e-05,
      "loss": 0.0193,
      "step": 3939
    },
    {
      "epoch": 35.18,
      "grad_norm": 0.040144361555576324,
      "learning_rate": 9.776470588235293e-05,
      "loss": 0.0189,
      "step": 3940
    },
    {
      "epoch": 35.19,
      "grad_norm": 0.050802409648895264,
      "learning_rate": 9.770588235294117e-05,
      "loss": 0.0204,
      "step": 3941
    },
    {
      "epoch": 35.2,
      "grad_norm": 0.04380539059638977,
      "learning_rate": 9.76470588235294e-05,
      "loss": 0.0205,
      "step": 3942
    },
    {
      "epoch": 35.21,
      "grad_norm": 0.049333445727825165,
      "learning_rate": 9.758823529411765e-05,
      "loss": 0.0171,
      "step": 3943
    },
    {
      "epoch": 35.21,
      "grad_norm": 0.0534956157207489,
      "learning_rate": 9.752941176470586e-05,
      "loss": 0.0193,
      "step": 3944
    },
    {
      "epoch": 35.22,
      "grad_norm": 0.0936095118522644,
      "learning_rate": 9.74705882352941e-05,
      "loss": 0.0238,
      "step": 3945
    },
    {
      "epoch": 35.23,
      "grad_norm": 0.043501611799001694,
      "learning_rate": 9.741176470588235e-05,
      "loss": 0.0171,
      "step": 3946
    },
    {
      "epoch": 35.24,
      "grad_norm": 0.049899764358997345,
      "learning_rate": 9.735294117647058e-05,
      "loss": 0.0176,
      "step": 3947
    },
    {
      "epoch": 35.25,
      "grad_norm": 0.047519441694021225,
      "learning_rate": 9.729411764705882e-05,
      "loss": 0.0189,
      "step": 3948
    },
    {
      "epoch": 35.26,
      "grad_norm": 0.05049138143658638,
      "learning_rate": 9.723529411764704e-05,
      "loss": 0.0159,
      "step": 3949
    },
    {
      "epoch": 35.27,
      "grad_norm": 0.04400688037276268,
      "learning_rate": 9.717647058823528e-05,
      "loss": 0.0168,
      "step": 3950
    },
    {
      "epoch": 35.28,
      "grad_norm": 0.06818078458309174,
      "learning_rate": 9.711764705882353e-05,
      "loss": 0.0175,
      "step": 3951
    },
    {
      "epoch": 35.29,
      "grad_norm": 0.0692988783121109,
      "learning_rate": 9.705882352941176e-05,
      "loss": 0.0189,
      "step": 3952
    },
    {
      "epoch": 35.29,
      "grad_norm": 0.04943229258060455,
      "learning_rate": 9.699999999999999e-05,
      "loss": 0.0195,
      "step": 3953
    },
    {
      "epoch": 35.3,
      "grad_norm": 0.042735274881124496,
      "learning_rate": 9.694117647058822e-05,
      "loss": 0.0172,
      "step": 3954
    },
    {
      "epoch": 35.31,
      "grad_norm": 0.051739778369665146,
      "learning_rate": 9.688235294117646e-05,
      "loss": 0.0185,
      "step": 3955
    },
    {
      "epoch": 35.32,
      "grad_norm": 0.04766315221786499,
      "learning_rate": 9.68235294117647e-05,
      "loss": 0.0196,
      "step": 3956
    },
    {
      "epoch": 35.33,
      "grad_norm": 0.0675685852766037,
      "learning_rate": 9.676470588235293e-05,
      "loss": 0.0204,
      "step": 3957
    },
    {
      "epoch": 35.34,
      "grad_norm": 0.07023131102323532,
      "learning_rate": 9.670588235294116e-05,
      "loss": 0.0256,
      "step": 3958
    },
    {
      "epoch": 35.35,
      "grad_norm": 0.06399403512477875,
      "learning_rate": 9.66470588235294e-05,
      "loss": 0.0192,
      "step": 3959
    },
    {
      "epoch": 35.36,
      "grad_norm": 0.04383014887571335,
      "learning_rate": 9.658823529411764e-05,
      "loss": 0.0191,
      "step": 3960
    },
    {
      "epoch": 35.37,
      "grad_norm": 0.04324600473046303,
      "learning_rate": 9.652941176470588e-05,
      "loss": 0.0193,
      "step": 3961
    },
    {
      "epoch": 35.38,
      "grad_norm": 0.04773641377687454,
      "learning_rate": 9.647058823529412e-05,
      "loss": 0.0186,
      "step": 3962
    },
    {
      "epoch": 35.38,
      "grad_norm": 0.05815904587507248,
      "learning_rate": 9.641176470588234e-05,
      "loss": 0.0205,
      "step": 3963
    },
    {
      "epoch": 35.39,
      "grad_norm": 0.10423627495765686,
      "learning_rate": 9.635294117647058e-05,
      "loss": 0.0216,
      "step": 3964
    },
    {
      "epoch": 35.4,
      "grad_norm": 0.04527971148490906,
      "learning_rate": 9.629411764705881e-05,
      "loss": 0.0215,
      "step": 3965
    },
    {
      "epoch": 35.41,
      "grad_norm": 0.04167898744344711,
      "learning_rate": 9.623529411764706e-05,
      "loss": 0.0152,
      "step": 3966
    },
    {
      "epoch": 35.42,
      "grad_norm": 0.041512541472911835,
      "learning_rate": 9.61764705882353e-05,
      "loss": 0.0191,
      "step": 3967
    },
    {
      "epoch": 35.43,
      "grad_norm": 0.04319898039102554,
      "learning_rate": 9.611764705882352e-05,
      "loss": 0.016,
      "step": 3968
    },
    {
      "epoch": 35.44,
      "grad_norm": 0.052551429718732834,
      "learning_rate": 9.605882352941176e-05,
      "loss": 0.0218,
      "step": 3969
    },
    {
      "epoch": 35.45,
      "grad_norm": 0.11623974144458771,
      "learning_rate": 9.599999999999999e-05,
      "loss": 0.0239,
      "step": 3970
    },
    {
      "epoch": 35.46,
      "grad_norm": 0.049464885145425797,
      "learning_rate": 9.594117647058823e-05,
      "loss": 0.0214,
      "step": 3971
    },
    {
      "epoch": 35.46,
      "grad_norm": 0.04722049832344055,
      "learning_rate": 9.588235294117645e-05,
      "loss": 0.0225,
      "step": 3972
    },
    {
      "epoch": 35.47,
      "grad_norm": 0.044510941952466965,
      "learning_rate": 9.582352941176469e-05,
      "loss": 0.0177,
      "step": 3973
    },
    {
      "epoch": 35.48,
      "grad_norm": 0.054078925400972366,
      "learning_rate": 9.576470588235294e-05,
      "loss": 0.0177,
      "step": 3974
    },
    {
      "epoch": 35.49,
      "grad_norm": 0.06436042487621307,
      "learning_rate": 9.570588235294117e-05,
      "loss": 0.0201,
      "step": 3975
    },
    {
      "epoch": 35.5,
      "grad_norm": 0.07849033921957016,
      "learning_rate": 9.564705882352941e-05,
      "loss": 0.0247,
      "step": 3976
    },
    {
      "epoch": 35.51,
      "grad_norm": 0.07635417580604553,
      "learning_rate": 9.558823529411764e-05,
      "loss": 0.0217,
      "step": 3977
    },
    {
      "epoch": 35.52,
      "grad_norm": 0.04666660726070404,
      "learning_rate": 9.552941176470587e-05,
      "loss": 0.0213,
      "step": 3978
    },
    {
      "epoch": 35.53,
      "grad_norm": 0.04402756690979004,
      "learning_rate": 9.547058823529411e-05,
      "loss": 0.0153,
      "step": 3979
    },
    {
      "epoch": 35.54,
      "grad_norm": 0.05166502296924591,
      "learning_rate": 9.541176470588236e-05,
      "loss": 0.0213,
      "step": 3980
    },
    {
      "epoch": 35.54,
      "grad_norm": 0.046839211136102676,
      "learning_rate": 9.535294117647059e-05,
      "loss": 0.02,
      "step": 3981
    },
    {
      "epoch": 35.55,
      "grad_norm": 0.07163360714912415,
      "learning_rate": 9.529411764705882e-05,
      "loss": 0.0209,
      "step": 3982
    },
    {
      "epoch": 35.56,
      "grad_norm": 0.1001317948102951,
      "learning_rate": 9.523529411764705e-05,
      "loss": 0.0246,
      "step": 3983
    },
    {
      "epoch": 35.57,
      "grad_norm": 0.04498804733157158,
      "learning_rate": 9.517647058823529e-05,
      "loss": 0.0207,
      "step": 3984
    },
    {
      "epoch": 35.58,
      "grad_norm": 0.04487980902194977,
      "learning_rate": 9.511764705882353e-05,
      "loss": 0.0208,
      "step": 3985
    },
    {
      "epoch": 35.59,
      "grad_norm": 0.04715997725725174,
      "learning_rate": 9.505882352941175e-05,
      "loss": 0.0182,
      "step": 3986
    },
    {
      "epoch": 35.6,
      "grad_norm": 0.04480253905057907,
      "learning_rate": 9.499999999999999e-05,
      "loss": 0.0154,
      "step": 3987
    },
    {
      "epoch": 35.61,
      "grad_norm": 0.058956392109394073,
      "learning_rate": 9.494117647058822e-05,
      "loss": 0.0189,
      "step": 3988
    },
    {
      "epoch": 35.62,
      "grad_norm": 0.09902913868427277,
      "learning_rate": 9.488235294117646e-05,
      "loss": 0.0249,
      "step": 3989
    },
    {
      "epoch": 35.62,
      "grad_norm": 0.04270798712968826,
      "learning_rate": 9.482352941176471e-05,
      "loss": 0.0196,
      "step": 3990
    },
    {
      "epoch": 35.63,
      "grad_norm": 0.05272712558507919,
      "learning_rate": 9.476470588235292e-05,
      "loss": 0.0192,
      "step": 3991
    },
    {
      "epoch": 35.64,
      "grad_norm": 0.04505373165011406,
      "learning_rate": 9.470588235294117e-05,
      "loss": 0.0176,
      "step": 3992
    },
    {
      "epoch": 35.65,
      "grad_norm": 0.04484843462705612,
      "learning_rate": 9.46470588235294e-05,
      "loss": 0.0152,
      "step": 3993
    },
    {
      "epoch": 35.66,
      "grad_norm": 0.054936349391937256,
      "learning_rate": 9.458823529411764e-05,
      "loss": 0.0197,
      "step": 3994
    },
    {
      "epoch": 35.67,
      "grad_norm": 0.10348665714263916,
      "learning_rate": 9.452941176470588e-05,
      "loss": 0.0269,
      "step": 3995
    },
    {
      "epoch": 35.68,
      "grad_norm": 0.05079462379217148,
      "learning_rate": 9.44705882352941e-05,
      "loss": 0.0222,
      "step": 3996
    },
    {
      "epoch": 35.69,
      "grad_norm": 0.04353007301688194,
      "learning_rate": 9.441176470588234e-05,
      "loss": 0.0185,
      "step": 3997
    },
    {
      "epoch": 35.7,
      "grad_norm": 0.04913564771413803,
      "learning_rate": 9.435294117647057e-05,
      "loss": 0.0201,
      "step": 3998
    },
    {
      "epoch": 35.71,
      "grad_norm": 0.0476599857211113,
      "learning_rate": 9.429411764705882e-05,
      "loss": 0.0162,
      "step": 3999
    },
    {
      "epoch": 35.71,
      "grad_norm": 0.05431988462805748,
      "learning_rate": 9.423529411764706e-05,
      "loss": 0.0214,
      "step": 4000
    },
    {
      "epoch": 35.71,
      "eval_cer": 0.031649209051651236,
      "eval_loss": 0.20453354716300964,
      "eval_runtime": 23.7481,
      "eval_samples_per_second": 111.251,
      "eval_steps_per_second": 1.769,
      "eval_wer": 0.10329234430781437,
      "step": 4000
    },
    {
      "epoch": 35.72,
      "grad_norm": 0.06919653713703156,
      "learning_rate": 9.417647058823528e-05,
      "loss": 0.0186,
      "step": 4001
    },
    {
      "epoch": 35.73,
      "grad_norm": 0.10540597140789032,
      "learning_rate": 9.411764705882352e-05,
      "loss": 0.0253,
      "step": 4002
    },
    {
      "epoch": 35.74,
      "grad_norm": 0.04335903748869896,
      "learning_rate": 9.405882352941176e-05,
      "loss": 0.0153,
      "step": 4003
    },
    {
      "epoch": 35.75,
      "grad_norm": 0.0410657674074173,
      "learning_rate": 9.4e-05,
      "loss": 0.0193,
      "step": 4004
    },
    {
      "epoch": 35.76,
      "grad_norm": 0.045593492686748505,
      "learning_rate": 9.394117647058822e-05,
      "loss": 0.0172,
      "step": 4005
    },
    {
      "epoch": 35.77,
      "grad_norm": 0.04321907460689545,
      "learning_rate": 9.388235294117645e-05,
      "loss": 0.0174,
      "step": 4006
    },
    {
      "epoch": 35.78,
      "grad_norm": 0.06603457778692245,
      "learning_rate": 9.38235294117647e-05,
      "loss": 0.0215,
      "step": 4007
    },
    {
      "epoch": 35.79,
      "grad_norm": 0.06926041096448898,
      "learning_rate": 9.376470588235294e-05,
      "loss": 0.0234,
      "step": 4008
    },
    {
      "epoch": 35.79,
      "grad_norm": 0.045095477253198624,
      "learning_rate": 9.370588235294117e-05,
      "loss": 0.0222,
      "step": 4009
    },
    {
      "epoch": 35.8,
      "grad_norm": 0.035017479211091995,
      "learning_rate": 9.36470588235294e-05,
      "loss": 0.0143,
      "step": 4010
    },
    {
      "epoch": 35.81,
      "grad_norm": 0.04700927063822746,
      "learning_rate": 9.358823529411763e-05,
      "loss": 0.0204,
      "step": 4011
    },
    {
      "epoch": 35.82,
      "grad_norm": 0.05358951911330223,
      "learning_rate": 9.352941176470587e-05,
      "loss": 0.0207,
      "step": 4012
    },
    {
      "epoch": 35.83,
      "grad_norm": 0.058761801570653915,
      "learning_rate": 9.347058823529412e-05,
      "loss": 0.0175,
      "step": 4013
    },
    {
      "epoch": 35.84,
      "grad_norm": 0.08514871448278427,
      "learning_rate": 9.341176470588235e-05,
      "loss": 0.0244,
      "step": 4014
    },
    {
      "epoch": 35.85,
      "grad_norm": 0.042977988719940186,
      "learning_rate": 9.335294117647058e-05,
      "loss": 0.0203,
      "step": 4015
    },
    {
      "epoch": 35.86,
      "grad_norm": 0.03963282331824303,
      "learning_rate": 9.329411764705881e-05,
      "loss": 0.0152,
      "step": 4016
    },
    {
      "epoch": 35.87,
      "grad_norm": 0.04526777192950249,
      "learning_rate": 9.323529411764705e-05,
      "loss": 0.0185,
      "step": 4017
    },
    {
      "epoch": 35.88,
      "grad_norm": 0.04637821763753891,
      "learning_rate": 9.31764705882353e-05,
      "loss": 0.0175,
      "step": 4018
    },
    {
      "epoch": 35.88,
      "grad_norm": 0.052102718502283096,
      "learning_rate": 9.311764705882351e-05,
      "loss": 0.0221,
      "step": 4019
    },
    {
      "epoch": 35.89,
      "grad_norm": 0.0893474668264389,
      "learning_rate": 9.305882352941175e-05,
      "loss": 0.0294,
      "step": 4020
    },
    {
      "epoch": 35.9,
      "grad_norm": 0.04137646406888962,
      "learning_rate": 9.3e-05,
      "loss": 0.0186,
      "step": 4021
    },
    {
      "epoch": 35.91,
      "grad_norm": 0.049010612070560455,
      "learning_rate": 9.294117647058823e-05,
      "loss": 0.0213,
      "step": 4022
    },
    {
      "epoch": 35.92,
      "grad_norm": 0.039913684129714966,
      "learning_rate": 9.288235294117647e-05,
      "loss": 0.0164,
      "step": 4023
    },
    {
      "epoch": 35.93,
      "grad_norm": 0.04661644250154495,
      "learning_rate": 9.282352941176469e-05,
      "loss": 0.017,
      "step": 4024
    },
    {
      "epoch": 35.94,
      "grad_norm": 0.05610841512680054,
      "learning_rate": 9.276470588235293e-05,
      "loss": 0.0206,
      "step": 4025
    },
    {
      "epoch": 35.95,
      "grad_norm": 0.07188771665096283,
      "learning_rate": 9.270588235294117e-05,
      "loss": 0.0212,
      "step": 4026
    },
    {
      "epoch": 35.96,
      "grad_norm": 0.07807855308055878,
      "learning_rate": 9.26470588235294e-05,
      "loss": 0.0211,
      "step": 4027
    },
    {
      "epoch": 35.96,
      "grad_norm": 0.042308419942855835,
      "learning_rate": 9.258823529411765e-05,
      "loss": 0.0179,
      "step": 4028
    },
    {
      "epoch": 35.97,
      "grad_norm": 0.052682895213365555,
      "learning_rate": 9.252941176470586e-05,
      "loss": 0.0162,
      "step": 4029
    },
    {
      "epoch": 35.98,
      "grad_norm": 0.0560329295694828,
      "learning_rate": 9.247058823529411e-05,
      "loss": 0.0214,
      "step": 4030
    },
    {
      "epoch": 35.99,
      "grad_norm": 0.06603354215621948,
      "learning_rate": 9.241176470588235e-05,
      "loss": 0.0203,
      "step": 4031
    },
    {
      "epoch": 36.0,
      "grad_norm": 0.0614035464823246,
      "learning_rate": 9.235294117647058e-05,
      "loss": 0.0215,
      "step": 4032
    },
    {
      "epoch": 36.01,
      "grad_norm": 0.0384756401181221,
      "learning_rate": 9.229411764705881e-05,
      "loss": 0.0172,
      "step": 4033
    },
    {
      "epoch": 36.02,
      "grad_norm": 0.045768752694129944,
      "learning_rate": 9.223529411764704e-05,
      "loss": 0.0173,
      "step": 4034
    },
    {
      "epoch": 36.03,
      "grad_norm": 0.04661966860294342,
      "learning_rate": 9.217647058823528e-05,
      "loss": 0.0182,
      "step": 4035
    },
    {
      "epoch": 36.04,
      "grad_norm": 0.04212106764316559,
      "learning_rate": 9.211764705882353e-05,
      "loss": 0.0162,
      "step": 4036
    },
    {
      "epoch": 36.04,
      "grad_norm": 0.050992123782634735,
      "learning_rate": 9.205882352941176e-05,
      "loss": 0.0176,
      "step": 4037
    },
    {
      "epoch": 36.05,
      "grad_norm": 0.07657718658447266,
      "learning_rate": 9.199999999999999e-05,
      "loss": 0.0215,
      "step": 4038
    },
    {
      "epoch": 36.06,
      "grad_norm": 0.06231314316391945,
      "learning_rate": 9.194117647058823e-05,
      "loss": 0.0195,
      "step": 4039
    },
    {
      "epoch": 36.07,
      "grad_norm": 0.044922493398189545,
      "learning_rate": 9.188235294117646e-05,
      "loss": 0.0204,
      "step": 4040
    },
    {
      "epoch": 36.08,
      "grad_norm": 0.046494387090206146,
      "learning_rate": 9.18235294117647e-05,
      "loss": 0.0175,
      "step": 4041
    },
    {
      "epoch": 36.09,
      "grad_norm": 0.051921382546424866,
      "learning_rate": 9.176470588235295e-05,
      "loss": 0.024,
      "step": 4042
    },
    {
      "epoch": 36.1,
      "grad_norm": 0.04883875325322151,
      "learning_rate": 9.170588235294116e-05,
      "loss": 0.0187,
      "step": 4043
    },
    {
      "epoch": 36.11,
      "grad_norm": 0.05999523401260376,
      "learning_rate": 9.16470588235294e-05,
      "loss": 0.0209,
      "step": 4044
    },
    {
      "epoch": 36.12,
      "grad_norm": 0.08448347449302673,
      "learning_rate": 9.158823529411764e-05,
      "loss": 0.022,
      "step": 4045
    },
    {
      "epoch": 36.12,
      "grad_norm": 0.04675189405679703,
      "learning_rate": 9.152941176470588e-05,
      "loss": 0.0223,
      "step": 4046
    },
    {
      "epoch": 36.13,
      "grad_norm": 0.04032502323389053,
      "learning_rate": 9.147058823529412e-05,
      "loss": 0.0186,
      "step": 4047
    },
    {
      "epoch": 36.14,
      "grad_norm": 0.048110414296388626,
      "learning_rate": 9.141176470588234e-05,
      "loss": 0.0186,
      "step": 4048
    },
    {
      "epoch": 36.15,
      "grad_norm": 0.04452439397573471,
      "learning_rate": 9.135294117647058e-05,
      "loss": 0.0164,
      "step": 4049
    },
    {
      "epoch": 36.16,
      "grad_norm": 0.057770442217588425,
      "learning_rate": 9.129411764705881e-05,
      "loss": 0.0219,
      "step": 4050
    },
    {
      "epoch": 36.17,
      "grad_norm": 0.0856478363275528,
      "learning_rate": 9.123529411764706e-05,
      "loss": 0.0207,
      "step": 4051
    },
    {
      "epoch": 36.18,
      "grad_norm": 0.04736426845192909,
      "learning_rate": 9.117647058823527e-05,
      "loss": 0.0182,
      "step": 4052
    },
    {
      "epoch": 36.19,
      "grad_norm": 0.03857048973441124,
      "learning_rate": 9.111764705882352e-05,
      "loss": 0.0165,
      "step": 4053
    },
    {
      "epoch": 36.2,
      "grad_norm": 0.04223356395959854,
      "learning_rate": 9.105882352941176e-05,
      "loss": 0.0176,
      "step": 4054
    },
    {
      "epoch": 36.21,
      "grad_norm": 0.046761445701122284,
      "learning_rate": 9.099999999999999e-05,
      "loss": 0.0172,
      "step": 4055
    },
    {
      "epoch": 36.21,
      "grad_norm": 0.045401718467473984,
      "learning_rate": 9.094117647058823e-05,
      "loss": 0.0161,
      "step": 4056
    },
    {
      "epoch": 36.22,
      "grad_norm": 0.0933755412697792,
      "learning_rate": 9.088235294117646e-05,
      "loss": 0.0208,
      "step": 4057
    },
    {
      "epoch": 36.23,
      "grad_norm": 0.042552392929792404,
      "learning_rate": 9.082352941176469e-05,
      "loss": 0.0172,
      "step": 4058
    },
    {
      "epoch": 36.24,
      "grad_norm": 0.04614667221903801,
      "learning_rate": 9.076470588235294e-05,
      "loss": 0.0172,
      "step": 4059
    },
    {
      "epoch": 36.25,
      "grad_norm": 0.044079042971134186,
      "learning_rate": 9.070588235294118e-05,
      "loss": 0.0191,
      "step": 4060
    },
    {
      "epoch": 36.26,
      "grad_norm": 0.0437043160200119,
      "learning_rate": 9.064705882352941e-05,
      "loss": 0.0185,
      "step": 4061
    },
    {
      "epoch": 36.27,
      "grad_norm": 0.046947527676820755,
      "learning_rate": 9.058823529411764e-05,
      "loss": 0.0161,
      "step": 4062
    },
    {
      "epoch": 36.28,
      "grad_norm": 0.06822053343057632,
      "learning_rate": 9.052941176470587e-05,
      "loss": 0.0201,
      "step": 4063
    },
    {
      "epoch": 36.29,
      "grad_norm": 0.06094614788889885,
      "learning_rate": 9.047058823529411e-05,
      "loss": 0.0169,
      "step": 4064
    },
    {
      "epoch": 36.29,
      "grad_norm": 0.04314233362674713,
      "learning_rate": 9.041176470588236e-05,
      "loss": 0.0161,
      "step": 4065
    },
    {
      "epoch": 36.3,
      "grad_norm": 0.03811066225171089,
      "learning_rate": 9.035294117647057e-05,
      "loss": 0.0159,
      "step": 4066
    },
    {
      "epoch": 36.31,
      "grad_norm": 0.05600721389055252,
      "learning_rate": 9.029411764705882e-05,
      "loss": 0.0204,
      "step": 4067
    },
    {
      "epoch": 36.32,
      "grad_norm": 0.05661261826753616,
      "learning_rate": 9.023529411764705e-05,
      "loss": 0.0191,
      "step": 4068
    },
    {
      "epoch": 36.33,
      "grad_norm": 0.06936700642108917,
      "learning_rate": 9.017647058823529e-05,
      "loss": 0.0241,
      "step": 4069
    },
    {
      "epoch": 36.34,
      "grad_norm": 0.09189575910568237,
      "learning_rate": 9.011764705882353e-05,
      "loss": 0.0262,
      "step": 4070
    },
    {
      "epoch": 36.35,
      "grad_norm": 0.046048182994127274,
      "learning_rate": 9.005882352941175e-05,
      "loss": 0.0212,
      "step": 4071
    },
    {
      "epoch": 36.36,
      "grad_norm": 0.04294374957680702,
      "learning_rate": 8.999999999999999e-05,
      "loss": 0.0173,
      "step": 4072
    },
    {
      "epoch": 36.37,
      "grad_norm": 0.04520189017057419,
      "learning_rate": 8.994117647058822e-05,
      "loss": 0.0199,
      "step": 4073
    },
    {
      "epoch": 36.38,
      "grad_norm": 0.043684907257556915,
      "learning_rate": 8.988235294117647e-05,
      "loss": 0.0172,
      "step": 4074
    },
    {
      "epoch": 36.38,
      "grad_norm": 0.05723635107278824,
      "learning_rate": 8.982352941176471e-05,
      "loss": 0.0235,
      "step": 4075
    },
    {
      "epoch": 36.39,
      "grad_norm": 0.0909346342086792,
      "learning_rate": 8.976470588235293e-05,
      "loss": 0.0252,
      "step": 4076
    },
    {
      "epoch": 36.4,
      "grad_norm": 0.042179208248853683,
      "learning_rate": 8.970588235294117e-05,
      "loss": 0.0169,
      "step": 4077
    },
    {
      "epoch": 36.41,
      "grad_norm": 0.04011724889278412,
      "learning_rate": 8.96470588235294e-05,
      "loss": 0.0195,
      "step": 4078
    },
    {
      "epoch": 36.42,
      "grad_norm": 0.042223189026117325,
      "learning_rate": 8.958823529411764e-05,
      "loss": 0.0153,
      "step": 4079
    },
    {
      "epoch": 36.43,
      "grad_norm": 0.0425608791410923,
      "learning_rate": 8.952941176470587e-05,
      "loss": 0.0162,
      "step": 4080
    },
    {
      "epoch": 36.44,
      "grad_norm": 0.06180000677704811,
      "learning_rate": 8.94705882352941e-05,
      "loss": 0.018,
      "step": 4081
    },
    {
      "epoch": 36.45,
      "grad_norm": 0.08952069282531738,
      "learning_rate": 8.941176470588235e-05,
      "loss": 0.0241,
      "step": 4082
    },
    {
      "epoch": 36.46,
      "grad_norm": 0.05010952427983284,
      "learning_rate": 8.935294117647059e-05,
      "loss": 0.0201,
      "step": 4083
    },
    {
      "epoch": 36.46,
      "grad_norm": 0.04697617143392563,
      "learning_rate": 8.929411764705882e-05,
      "loss": 0.0205,
      "step": 4084
    },
    {
      "epoch": 36.47,
      "grad_norm": 0.04716704785823822,
      "learning_rate": 8.923529411764705e-05,
      "loss": 0.0161,
      "step": 4085
    },
    {
      "epoch": 36.48,
      "grad_norm": 0.04796275869011879,
      "learning_rate": 8.917647058823528e-05,
      "loss": 0.0177,
      "step": 4086
    },
    {
      "epoch": 36.49,
      "grad_norm": 0.05288788303732872,
      "learning_rate": 8.911764705882352e-05,
      "loss": 0.0215,
      "step": 4087
    },
    {
      "epoch": 36.5,
      "grad_norm": 0.07674020528793335,
      "learning_rate": 8.905882352941177e-05,
      "loss": 0.0205,
      "step": 4088
    },
    {
      "epoch": 36.51,
      "grad_norm": 0.06745994836091995,
      "learning_rate": 8.9e-05,
      "loss": 0.0198,
      "step": 4089
    },
    {
      "epoch": 36.52,
      "grad_norm": 0.04017637297511101,
      "learning_rate": 8.894117647058822e-05,
      "loss": 0.0172,
      "step": 4090
    },
    {
      "epoch": 36.53,
      "grad_norm": 0.0458281934261322,
      "learning_rate": 8.888235294117645e-05,
      "loss": 0.0196,
      "step": 4091
    },
    {
      "epoch": 36.54,
      "grad_norm": 0.04622480645775795,
      "learning_rate": 8.88235294117647e-05,
      "loss": 0.0187,
      "step": 4092
    },
    {
      "epoch": 36.54,
      "grad_norm": 0.04795752465724945,
      "learning_rate": 8.876470588235294e-05,
      "loss": 0.0191,
      "step": 4093
    },
    {
      "epoch": 36.55,
      "grad_norm": 0.05634784698486328,
      "learning_rate": 8.870588235294117e-05,
      "loss": 0.0209,
      "step": 4094
    },
    {
      "epoch": 36.56,
      "grad_norm": 0.09274792671203613,
      "learning_rate": 8.86470588235294e-05,
      "loss": 0.0268,
      "step": 4095
    },
    {
      "epoch": 36.57,
      "grad_norm": 0.04380248114466667,
      "learning_rate": 8.858823529411763e-05,
      "loss": 0.0178,
      "step": 4096
    },
    {
      "epoch": 36.58,
      "grad_norm": 0.04211769253015518,
      "learning_rate": 8.852941176470587e-05,
      "loss": 0.0188,
      "step": 4097
    },
    {
      "epoch": 36.59,
      "grad_norm": 0.04355277493596077,
      "learning_rate": 8.847058823529412e-05,
      "loss": 0.0181,
      "step": 4098
    },
    {
      "epoch": 36.6,
      "grad_norm": 0.05291120707988739,
      "learning_rate": 8.841176470588233e-05,
      "loss": 0.0204,
      "step": 4099
    },
    {
      "epoch": 36.61,
      "grad_norm": 0.04758208990097046,
      "learning_rate": 8.835294117647058e-05,
      "loss": 0.016,
      "step": 4100
    },
    {
      "epoch": 36.61,
      "eval_cer": 0.030153378959909497,
      "eval_loss": 0.20217667520046234,
      "eval_runtime": 22.4099,
      "eval_samples_per_second": 117.894,
      "eval_steps_per_second": 1.874,
      "eval_wer": 0.09998016660055534,
      "step": 4100
    },
    {
      "epoch": 36.62,
      "grad_norm": 0.0827101469039917,
      "learning_rate": 8.829411764705882e-05,
      "loss": 0.0198,
      "step": 4101
    },
    {
      "epoch": 36.62,
      "grad_norm": 0.04562339186668396,
      "learning_rate": 8.823529411764705e-05,
      "loss": 0.0183,
      "step": 4102
    },
    {
      "epoch": 36.63,
      "grad_norm": 0.04250386729836464,
      "learning_rate": 8.81764705882353e-05,
      "loss": 0.0182,
      "step": 4103
    },
    {
      "epoch": 36.64,
      "grad_norm": 0.04504351317882538,
      "learning_rate": 8.811764705882351e-05,
      "loss": 0.019,
      "step": 4104
    },
    {
      "epoch": 36.65,
      "grad_norm": 0.062047552317380905,
      "learning_rate": 8.805882352941175e-05,
      "loss": 0.0209,
      "step": 4105
    },
    {
      "epoch": 36.66,
      "grad_norm": 0.06217120215296745,
      "learning_rate": 8.8e-05,
      "loss": 0.0194,
      "step": 4106
    },
    {
      "epoch": 36.67,
      "grad_norm": 0.09967740625143051,
      "learning_rate": 8.794117647058823e-05,
      "loss": 0.0274,
      "step": 4107
    },
    {
      "epoch": 36.68,
      "grad_norm": 0.043474357575178146,
      "learning_rate": 8.788235294117647e-05,
      "loss": 0.0198,
      "step": 4108
    },
    {
      "epoch": 36.69,
      "grad_norm": 0.04572237655520439,
      "learning_rate": 8.782352941176469e-05,
      "loss": 0.0202,
      "step": 4109
    },
    {
      "epoch": 36.7,
      "grad_norm": 0.04573224112391472,
      "learning_rate": 8.776470588235293e-05,
      "loss": 0.0173,
      "step": 4110
    },
    {
      "epoch": 36.71,
      "grad_norm": 0.05247281864285469,
      "learning_rate": 8.770588235294117e-05,
      "loss": 0.0183,
      "step": 4111
    },
    {
      "epoch": 36.71,
      "grad_norm": 0.047982342541217804,
      "learning_rate": 8.76470588235294e-05,
      "loss": 0.018,
      "step": 4112
    },
    {
      "epoch": 36.72,
      "grad_norm": 0.06717761605978012,
      "learning_rate": 8.758823529411763e-05,
      "loss": 0.0192,
      "step": 4113
    },
    {
      "epoch": 36.73,
      "grad_norm": 0.06690295040607452,
      "learning_rate": 8.752941176470586e-05,
      "loss": 0.0215,
      "step": 4114
    },
    {
      "epoch": 36.74,
      "grad_norm": 0.045360762625932693,
      "learning_rate": 8.747058823529411e-05,
      "loss": 0.0167,
      "step": 4115
    },
    {
      "epoch": 36.75,
      "grad_norm": 0.04655643552541733,
      "learning_rate": 8.741176470588235e-05,
      "loss": 0.0211,
      "step": 4116
    },
    {
      "epoch": 36.76,
      "grad_norm": 0.04601249471306801,
      "learning_rate": 8.735294117647058e-05,
      "loss": 0.0187,
      "step": 4117
    },
    {
      "epoch": 36.77,
      "grad_norm": 0.04942779242992401,
      "learning_rate": 8.729411764705881e-05,
      "loss": 0.0191,
      "step": 4118
    },
    {
      "epoch": 36.78,
      "grad_norm": 0.05830894038081169,
      "learning_rate": 8.723529411764705e-05,
      "loss": 0.0152,
      "step": 4119
    },
    {
      "epoch": 36.79,
      "grad_norm": 0.09998707473278046,
      "learning_rate": 8.717647058823528e-05,
      "loss": 0.0249,
      "step": 4120
    },
    {
      "epoch": 36.79,
      "grad_norm": 0.049037862569093704,
      "learning_rate": 8.711764705882353e-05,
      "loss": 0.0196,
      "step": 4121
    },
    {
      "epoch": 36.8,
      "grad_norm": 0.05310535803437233,
      "learning_rate": 8.705882352941177e-05,
      "loss": 0.0181,
      "step": 4122
    },
    {
      "epoch": 36.81,
      "grad_norm": 0.051805902272462845,
      "learning_rate": 8.699999999999999e-05,
      "loss": 0.021,
      "step": 4123
    },
    {
      "epoch": 36.82,
      "grad_norm": 0.04846580699086189,
      "learning_rate": 8.694117647058823e-05,
      "loss": 0.0151,
      "step": 4124
    },
    {
      "epoch": 36.83,
      "grad_norm": 0.05232789367437363,
      "learning_rate": 8.688235294117646e-05,
      "loss": 0.0183,
      "step": 4125
    },
    {
      "epoch": 36.84,
      "grad_norm": 0.07918441295623779,
      "learning_rate": 8.68235294117647e-05,
      "loss": 0.0209,
      "step": 4126
    },
    {
      "epoch": 36.85,
      "grad_norm": 0.04547763615846634,
      "learning_rate": 8.676470588235295e-05,
      "loss": 0.0195,
      "step": 4127
    },
    {
      "epoch": 36.86,
      "grad_norm": 0.04671810194849968,
      "learning_rate": 8.670588235294116e-05,
      "loss": 0.0198,
      "step": 4128
    },
    {
      "epoch": 36.87,
      "grad_norm": 0.045081958174705505,
      "learning_rate": 8.664705882352941e-05,
      "loss": 0.0202,
      "step": 4129
    },
    {
      "epoch": 36.88,
      "grad_norm": 0.05202661454677582,
      "learning_rate": 8.658823529411764e-05,
      "loss": 0.0196,
      "step": 4130
    },
    {
      "epoch": 36.88,
      "grad_norm": 0.04858111962676048,
      "learning_rate": 8.652941176470588e-05,
      "loss": 0.017,
      "step": 4131
    },
    {
      "epoch": 36.89,
      "grad_norm": 0.09790980815887451,
      "learning_rate": 8.64705882352941e-05,
      "loss": 0.022,
      "step": 4132
    },
    {
      "epoch": 36.9,
      "grad_norm": 0.045605581253767014,
      "learning_rate": 8.641176470588234e-05,
      "loss": 0.0202,
      "step": 4133
    },
    {
      "epoch": 36.91,
      "grad_norm": 0.041663676500320435,
      "learning_rate": 8.635294117647058e-05,
      "loss": 0.017,
      "step": 4134
    },
    {
      "epoch": 36.92,
      "grad_norm": 0.04285924509167671,
      "learning_rate": 8.629411764705881e-05,
      "loss": 0.0177,
      "step": 4135
    },
    {
      "epoch": 36.93,
      "grad_norm": 0.05176946520805359,
      "learning_rate": 8.623529411764706e-05,
      "loss": 0.0205,
      "step": 4136
    },
    {
      "epoch": 36.94,
      "grad_norm": 0.04874508082866669,
      "learning_rate": 8.617647058823529e-05,
      "loss": 0.0184,
      "step": 4137
    },
    {
      "epoch": 36.95,
      "grad_norm": 0.06821689754724503,
      "learning_rate": 8.611764705882352e-05,
      "loss": 0.0228,
      "step": 4138
    },
    {
      "epoch": 36.96,
      "grad_norm": 0.11184251308441162,
      "learning_rate": 8.605882352941176e-05,
      "loss": 0.0263,
      "step": 4139
    },
    {
      "epoch": 36.96,
      "grad_norm": 0.04449713975191116,
      "learning_rate": 8.6e-05,
      "loss": 0.0196,
      "step": 4140
    },
    {
      "epoch": 36.97,
      "grad_norm": 0.04029373079538345,
      "learning_rate": 8.594117647058823e-05,
      "loss": 0.0177,
      "step": 4141
    },
    {
      "epoch": 36.98,
      "grad_norm": 0.04268214851617813,
      "learning_rate": 8.588235294117646e-05,
      "loss": 0.018,
      "step": 4142
    },
    {
      "epoch": 36.99,
      "grad_norm": 0.04401444271206856,
      "learning_rate": 8.582352941176469e-05,
      "loss": 0.0158,
      "step": 4143
    },
    {
      "epoch": 37.0,
      "grad_norm": 0.06233417987823486,
      "learning_rate": 8.576470588235294e-05,
      "loss": 0.0222,
      "step": 4144
    },
    {
      "epoch": 37.01,
      "grad_norm": 0.03744405508041382,
      "learning_rate": 8.570588235294118e-05,
      "loss": 0.0207,
      "step": 4145
    },
    {
      "epoch": 37.02,
      "grad_norm": 0.04306573420763016,
      "learning_rate": 8.56470588235294e-05,
      "loss": 0.0197,
      "step": 4146
    },
    {
      "epoch": 37.03,
      "grad_norm": 0.044871244579553604,
      "learning_rate": 8.558823529411764e-05,
      "loss": 0.0178,
      "step": 4147
    },
    {
      "epoch": 37.04,
      "grad_norm": 0.042695172131061554,
      "learning_rate": 8.552941176470587e-05,
      "loss": 0.016,
      "step": 4148
    },
    {
      "epoch": 37.04,
      "grad_norm": 0.051238398998975754,
      "learning_rate": 8.547058823529411e-05,
      "loss": 0.0196,
      "step": 4149
    },
    {
      "epoch": 37.05,
      "grad_norm": 0.09372657537460327,
      "learning_rate": 8.541176470588236e-05,
      "loss": 0.0251,
      "step": 4150
    },
    {
      "epoch": 37.06,
      "grad_norm": 0.08845796436071396,
      "learning_rate": 8.535294117647057e-05,
      "loss": 0.0253,
      "step": 4151
    },
    {
      "epoch": 37.07,
      "grad_norm": 0.041753947734832764,
      "learning_rate": 8.529411764705882e-05,
      "loss": 0.0161,
      "step": 4152
    },
    {
      "epoch": 37.08,
      "grad_norm": 0.04683902859687805,
      "learning_rate": 8.523529411764705e-05,
      "loss": 0.0199,
      "step": 4153
    },
    {
      "epoch": 37.09,
      "grad_norm": 0.04193998500704765,
      "learning_rate": 8.517647058823529e-05,
      "loss": 0.0163,
      "step": 4154
    },
    {
      "epoch": 37.1,
      "grad_norm": 0.04733918607234955,
      "learning_rate": 8.511764705882353e-05,
      "loss": 0.0159,
      "step": 4155
    },
    {
      "epoch": 37.11,
      "grad_norm": 0.05876201391220093,
      "learning_rate": 8.505882352941175e-05,
      "loss": 0.0163,
      "step": 4156
    },
    {
      "epoch": 37.12,
      "grad_norm": 0.09874025732278824,
      "learning_rate": 8.499999999999999e-05,
      "loss": 0.02,
      "step": 4157
    },
    {
      "epoch": 37.12,
      "grad_norm": 0.04608433321118355,
      "learning_rate": 8.494117647058822e-05,
      "loss": 0.0182,
      "step": 4158
    },
    {
      "epoch": 37.13,
      "grad_norm": 0.04740406572818756,
      "learning_rate": 8.488235294117647e-05,
      "loss": 0.0185,
      "step": 4159
    },
    {
      "epoch": 37.14,
      "grad_norm": 0.04387908801436424,
      "learning_rate": 8.48235294117647e-05,
      "loss": 0.016,
      "step": 4160
    },
    {
      "epoch": 37.15,
      "grad_norm": 0.05817878246307373,
      "learning_rate": 8.476470588235293e-05,
      "loss": 0.0212,
      "step": 4161
    },
    {
      "epoch": 37.16,
      "grad_norm": 0.055451344698667526,
      "learning_rate": 8.470588235294117e-05,
      "loss": 0.0166,
      "step": 4162
    },
    {
      "epoch": 37.17,
      "grad_norm": 0.0807330384850502,
      "learning_rate": 8.464705882352941e-05,
      "loss": 0.0222,
      "step": 4163
    },
    {
      "epoch": 37.18,
      "grad_norm": 0.04307100921869278,
      "learning_rate": 8.458823529411764e-05,
      "loss": 0.018,
      "step": 4164
    },
    {
      "epoch": 37.19,
      "grad_norm": 0.04464540630578995,
      "learning_rate": 8.452941176470587e-05,
      "loss": 0.0148,
      "step": 4165
    },
    {
      "epoch": 37.2,
      "grad_norm": 0.04523914307355881,
      "learning_rate": 8.44705882352941e-05,
      "loss": 0.017,
      "step": 4166
    },
    {
      "epoch": 37.21,
      "grad_norm": 0.04177824407815933,
      "learning_rate": 8.441176470588235e-05,
      "loss": 0.0193,
      "step": 4167
    },
    {
      "epoch": 37.21,
      "grad_norm": 0.057433731853961945,
      "learning_rate": 8.435294117647059e-05,
      "loss": 0.0192,
      "step": 4168
    },
    {
      "epoch": 37.22,
      "grad_norm": 0.13495777547359467,
      "learning_rate": 8.429411764705882e-05,
      "loss": 0.0221,
      "step": 4169
    },
    {
      "epoch": 37.23,
      "grad_norm": 0.04041381552815437,
      "learning_rate": 8.423529411764705e-05,
      "loss": 0.0198,
      "step": 4170
    },
    {
      "epoch": 37.24,
      "grad_norm": 0.04346789792180061,
      "learning_rate": 8.417647058823528e-05,
      "loss": 0.0191,
      "step": 4171
    },
    {
      "epoch": 37.25,
      "grad_norm": 0.035024456679821014,
      "learning_rate": 8.411764705882352e-05,
      "loss": 0.0143,
      "step": 4172
    },
    {
      "epoch": 37.26,
      "grad_norm": 0.044780805706977844,
      "learning_rate": 8.405882352941177e-05,
      "loss": 0.0169,
      "step": 4173
    },
    {
      "epoch": 37.27,
      "grad_norm": 0.049140334129333496,
      "learning_rate": 8.4e-05,
      "loss": 0.0215,
      "step": 4174
    },
    {
      "epoch": 37.28,
      "grad_norm": 0.07246115058660507,
      "learning_rate": 8.394117647058823e-05,
      "loss": 0.0228,
      "step": 4175
    },
    {
      "epoch": 37.29,
      "grad_norm": 0.06579442322254181,
      "learning_rate": 8.388235294117646e-05,
      "loss": 0.022,
      "step": 4176
    },
    {
      "epoch": 37.29,
      "grad_norm": 0.043875981122255325,
      "learning_rate": 8.38235294117647e-05,
      "loss": 0.0197,
      "step": 4177
    },
    {
      "epoch": 37.3,
      "grad_norm": 0.04688160493969917,
      "learning_rate": 8.376470588235294e-05,
      "loss": 0.019,
      "step": 4178
    },
    {
      "epoch": 37.31,
      "grad_norm": 0.04777594655752182,
      "learning_rate": 8.370588235294116e-05,
      "loss": 0.0202,
      "step": 4179
    },
    {
      "epoch": 37.32,
      "grad_norm": 0.044735386967659,
      "learning_rate": 8.36470588235294e-05,
      "loss": 0.0152,
      "step": 4180
    },
    {
      "epoch": 37.33,
      "grad_norm": 0.06980415433645248,
      "learning_rate": 8.358823529411765e-05,
      "loss": 0.019,
      "step": 4181
    },
    {
      "epoch": 37.34,
      "grad_norm": 0.07926139235496521,
      "learning_rate": 8.352941176470588e-05,
      "loss": 0.0247,
      "step": 4182
    },
    {
      "epoch": 37.35,
      "grad_norm": 0.04676423966884613,
      "learning_rate": 8.347058823529412e-05,
      "loss": 0.0181,
      "step": 4183
    },
    {
      "epoch": 37.36,
      "grad_norm": 0.04603610187768936,
      "learning_rate": 8.341176470588233e-05,
      "loss": 0.017,
      "step": 4184
    },
    {
      "epoch": 37.37,
      "grad_norm": 0.04675346612930298,
      "learning_rate": 8.335294117647058e-05,
      "loss": 0.0179,
      "step": 4185
    },
    {
      "epoch": 37.38,
      "grad_norm": 0.053286049515008926,
      "learning_rate": 8.329411764705882e-05,
      "loss": 0.0162,
      "step": 4186
    },
    {
      "epoch": 37.38,
      "grad_norm": 0.06341850012540817,
      "learning_rate": 8.323529411764705e-05,
      "loss": 0.0162,
      "step": 4187
    },
    {
      "epoch": 37.39,
      "grad_norm": 0.09439020603895187,
      "learning_rate": 8.31764705882353e-05,
      "loss": 0.0233,
      "step": 4188
    },
    {
      "epoch": 37.4,
      "grad_norm": 0.045090291649103165,
      "learning_rate": 8.311764705882351e-05,
      "loss": 0.017,
      "step": 4189
    },
    {
      "epoch": 37.41,
      "grad_norm": 0.04638542979955673,
      "learning_rate": 8.305882352941175e-05,
      "loss": 0.0214,
      "step": 4190
    },
    {
      "epoch": 37.42,
      "grad_norm": 0.051234204322099686,
      "learning_rate": 8.3e-05,
      "loss": 0.0184,
      "step": 4191
    },
    {
      "epoch": 37.43,
      "grad_norm": 0.04307663068175316,
      "learning_rate": 8.294117647058823e-05,
      "loss": 0.0177,
      "step": 4192
    },
    {
      "epoch": 37.44,
      "grad_norm": 0.045090969651937485,
      "learning_rate": 8.288235294117646e-05,
      "loss": 0.0123,
      "step": 4193
    },
    {
      "epoch": 37.45,
      "grad_norm": 0.07967652380466461,
      "learning_rate": 8.282352941176469e-05,
      "loss": 0.0204,
      "step": 4194
    },
    {
      "epoch": 37.46,
      "grad_norm": 0.04527030885219574,
      "learning_rate": 8.276470588235293e-05,
      "loss": 0.0201,
      "step": 4195
    },
    {
      "epoch": 37.46,
      "grad_norm": 0.04721197485923767,
      "learning_rate": 8.270588235294117e-05,
      "loss": 0.0227,
      "step": 4196
    },
    {
      "epoch": 37.47,
      "grad_norm": 0.05305152386426926,
      "learning_rate": 8.26470588235294e-05,
      "loss": 0.0184,
      "step": 4197
    },
    {
      "epoch": 37.48,
      "grad_norm": 0.04736615717411041,
      "learning_rate": 8.258823529411763e-05,
      "loss": 0.0155,
      "step": 4198
    },
    {
      "epoch": 37.49,
      "grad_norm": 0.046044182032346725,
      "learning_rate": 8.252941176470588e-05,
      "loss": 0.0201,
      "step": 4199
    },
    {
      "epoch": 37.5,
      "grad_norm": 0.06805248558521271,
      "learning_rate": 8.247058823529411e-05,
      "loss": 0.0169,
      "step": 4200
    },
    {
      "epoch": 37.5,
      "eval_cer": 0.02986774306299399,
      "eval_loss": 0.20600314438343048,
      "eval_runtime": 22.4441,
      "eval_samples_per_second": 117.715,
      "eval_steps_per_second": 1.871,
      "eval_wer": 0.09956366521221738,
      "step": 4200
    },
    {
      "epoch": 37.51,
      "grad_norm": 0.060498278588056564,
      "learning_rate": 8.241176470588235e-05,
      "loss": 0.0182,
      "step": 4201
    },
    {
      "epoch": 37.52,
      "grad_norm": 0.043822020292282104,
      "learning_rate": 8.23529411764706e-05,
      "loss": 0.0185,
      "step": 4202
    },
    {
      "epoch": 37.53,
      "grad_norm": 0.045863326638936996,
      "learning_rate": 8.229411764705881e-05,
      "loss": 0.0185,
      "step": 4203
    },
    {
      "epoch": 37.54,
      "grad_norm": 0.04996486380696297,
      "learning_rate": 8.223529411764705e-05,
      "loss": 0.0194,
      "step": 4204
    },
    {
      "epoch": 37.54,
      "grad_norm": 0.04862746596336365,
      "learning_rate": 8.217647058823528e-05,
      "loss": 0.0155,
      "step": 4205
    },
    {
      "epoch": 37.55,
      "grad_norm": 0.060789793729782104,
      "learning_rate": 8.211764705882353e-05,
      "loss": 0.0184,
      "step": 4206
    },
    {
      "epoch": 37.56,
      "grad_norm": 0.10485976934432983,
      "learning_rate": 8.205882352941177e-05,
      "loss": 0.0249,
      "step": 4207
    },
    {
      "epoch": 37.57,
      "grad_norm": 0.04610714688897133,
      "learning_rate": 8.199999999999999e-05,
      "loss": 0.0178,
      "step": 4208
    },
    {
      "epoch": 37.58,
      "grad_norm": 0.04796239733695984,
      "learning_rate": 8.194117647058823e-05,
      "loss": 0.0184,
      "step": 4209
    },
    {
      "epoch": 37.59,
      "grad_norm": 0.04679669439792633,
      "learning_rate": 8.188235294117646e-05,
      "loss": 0.0207,
      "step": 4210
    },
    {
      "epoch": 37.6,
      "grad_norm": 0.05065898597240448,
      "learning_rate": 8.18235294117647e-05,
      "loss": 0.0157,
      "step": 4211
    },
    {
      "epoch": 37.61,
      "grad_norm": 0.051082100719213486,
      "learning_rate": 8.176470588235292e-05,
      "loss": 0.0177,
      "step": 4212
    },
    {
      "epoch": 37.62,
      "grad_norm": 0.06115908920764923,
      "learning_rate": 8.170588235294116e-05,
      "loss": 0.0177,
      "step": 4213
    },
    {
      "epoch": 37.62,
      "grad_norm": 0.04532017186284065,
      "learning_rate": 8.164705882352941e-05,
      "loss": 0.0177,
      "step": 4214
    },
    {
      "epoch": 37.63,
      "grad_norm": 0.04716581851243973,
      "learning_rate": 8.158823529411764e-05,
      "loss": 0.0189,
      "step": 4215
    },
    {
      "epoch": 37.64,
      "grad_norm": 0.05453303083777428,
      "learning_rate": 8.152941176470588e-05,
      "loss": 0.0187,
      "step": 4216
    },
    {
      "epoch": 37.65,
      "grad_norm": 0.04072986915707588,
      "learning_rate": 8.147058823529411e-05,
      "loss": 0.0147,
      "step": 4217
    },
    {
      "epoch": 37.66,
      "grad_norm": 0.0515437088906765,
      "learning_rate": 8.141176470588234e-05,
      "loss": 0.0187,
      "step": 4218
    },
    {
      "epoch": 37.67,
      "grad_norm": 0.0794481411576271,
      "learning_rate": 8.135294117647058e-05,
      "loss": 0.0175,
      "step": 4219
    },
    {
      "epoch": 37.68,
      "grad_norm": 0.038771871477365494,
      "learning_rate": 8.129411764705883e-05,
      "loss": 0.0176,
      "step": 4220
    },
    {
      "epoch": 37.69,
      "grad_norm": 0.04678460955619812,
      "learning_rate": 8.123529411764706e-05,
      "loss": 0.0206,
      "step": 4221
    },
    {
      "epoch": 37.7,
      "grad_norm": 0.04655611515045166,
      "learning_rate": 8.117647058823529e-05,
      "loss": 0.0186,
      "step": 4222
    },
    {
      "epoch": 37.71,
      "grad_norm": 0.04548626393079758,
      "learning_rate": 8.111764705882352e-05,
      "loss": 0.0167,
      "step": 4223
    },
    {
      "epoch": 37.71,
      "grad_norm": 0.05780279263854027,
      "learning_rate": 8.105882352941176e-05,
      "loss": 0.0199,
      "step": 4224
    },
    {
      "epoch": 37.72,
      "grad_norm": 0.06784381717443466,
      "learning_rate": 8.1e-05,
      "loss": 0.0198,
      "step": 4225
    },
    {
      "epoch": 37.73,
      "grad_norm": 0.07435184717178345,
      "learning_rate": 8.094117647058822e-05,
      "loss": 0.0204,
      "step": 4226
    },
    {
      "epoch": 37.74,
      "grad_norm": 0.039560895413160324,
      "learning_rate": 8.088235294117646e-05,
      "loss": 0.0158,
      "step": 4227
    },
    {
      "epoch": 37.75,
      "grad_norm": 0.042115263640880585,
      "learning_rate": 8.08235294117647e-05,
      "loss": 0.0168,
      "step": 4228
    },
    {
      "epoch": 37.76,
      "grad_norm": 0.041196148842573166,
      "learning_rate": 8.076470588235294e-05,
      "loss": 0.0169,
      "step": 4229
    },
    {
      "epoch": 37.77,
      "grad_norm": 0.04904539883136749,
      "learning_rate": 8.070588235294118e-05,
      "loss": 0.0166,
      "step": 4230
    },
    {
      "epoch": 37.78,
      "grad_norm": 0.0626995861530304,
      "learning_rate": 8.06470588235294e-05,
      "loss": 0.0196,
      "step": 4231
    },
    {
      "epoch": 37.79,
      "grad_norm": 0.07392898947000504,
      "learning_rate": 8.058823529411764e-05,
      "loss": 0.0184,
      "step": 4232
    },
    {
      "epoch": 37.79,
      "grad_norm": 0.04533496871590614,
      "learning_rate": 8.052941176470587e-05,
      "loss": 0.0181,
      "step": 4233
    },
    {
      "epoch": 37.8,
      "grad_norm": 0.04209979623556137,
      "learning_rate": 8.047058823529411e-05,
      "loss": 0.0177,
      "step": 4234
    },
    {
      "epoch": 37.81,
      "grad_norm": 0.037162382155656815,
      "learning_rate": 8.041176470588236e-05,
      "loss": 0.0146,
      "step": 4235
    },
    {
      "epoch": 37.82,
      "grad_norm": 0.04646997153759003,
      "learning_rate": 8.035294117647057e-05,
      "loss": 0.0168,
      "step": 4236
    },
    {
      "epoch": 37.83,
      "grad_norm": 0.05494370311498642,
      "learning_rate": 8.029411764705882e-05,
      "loss": 0.0172,
      "step": 4237
    },
    {
      "epoch": 37.84,
      "grad_norm": 0.08447180688381195,
      "learning_rate": 8.023529411764706e-05,
      "loss": 0.0212,
      "step": 4238
    },
    {
      "epoch": 37.85,
      "grad_norm": 0.04605470597743988,
      "learning_rate": 8.017647058823529e-05,
      "loss": 0.0175,
      "step": 4239
    },
    {
      "epoch": 37.86,
      "grad_norm": 0.041734080761671066,
      "learning_rate": 8.011764705882352e-05,
      "loss": 0.0133,
      "step": 4240
    },
    {
      "epoch": 37.87,
      "grad_norm": 0.04305846244096756,
      "learning_rate": 8.005882352941175e-05,
      "loss": 0.0156,
      "step": 4241
    },
    {
      "epoch": 37.88,
      "grad_norm": 0.049476027488708496,
      "learning_rate": 7.999999999999999e-05,
      "loss": 0.0164,
      "step": 4242
    },
    {
      "epoch": 37.88,
      "grad_norm": 0.047128137201070786,
      "learning_rate": 7.994117647058824e-05,
      "loss": 0.0155,
      "step": 4243
    },
    {
      "epoch": 37.89,
      "grad_norm": 0.10725882649421692,
      "learning_rate": 7.988235294117647e-05,
      "loss": 0.0232,
      "step": 4244
    },
    {
      "epoch": 37.9,
      "grad_norm": 0.04166736453771591,
      "learning_rate": 7.98235294117647e-05,
      "loss": 0.0171,
      "step": 4245
    },
    {
      "epoch": 37.91,
      "grad_norm": 0.04831862077116966,
      "learning_rate": 7.976470588235293e-05,
      "loss": 0.0203,
      "step": 4246
    },
    {
      "epoch": 37.92,
      "grad_norm": 0.051550157368183136,
      "learning_rate": 7.970588235294117e-05,
      "loss": 0.0189,
      "step": 4247
    },
    {
      "epoch": 37.93,
      "grad_norm": 0.045276615768671036,
      "learning_rate": 7.964705882352941e-05,
      "loss": 0.0187,
      "step": 4248
    },
    {
      "epoch": 37.94,
      "grad_norm": 0.05321819707751274,
      "learning_rate": 7.958823529411764e-05,
      "loss": 0.0161,
      "step": 4249
    },
    {
      "epoch": 37.95,
      "grad_norm": 0.06772742420434952,
      "learning_rate": 7.952941176470587e-05,
      "loss": 0.02,
      "step": 4250
    },
    {
      "epoch": 37.96,
      "grad_norm": 0.057556118816137314,
      "learning_rate": 7.94705882352941e-05,
      "loss": 0.02,
      "step": 4251
    },
    {
      "epoch": 37.96,
      "grad_norm": 0.039384014904499054,
      "learning_rate": 7.941176470588235e-05,
      "loss": 0.0175,
      "step": 4252
    },
    {
      "epoch": 37.97,
      "grad_norm": 0.04523039981722832,
      "learning_rate": 7.935294117647059e-05,
      "loss": 0.019,
      "step": 4253
    },
    {
      "epoch": 37.98,
      "grad_norm": 0.040083229541778564,
      "learning_rate": 7.929411764705882e-05,
      "loss": 0.0147,
      "step": 4254
    },
    {
      "epoch": 37.99,
      "grad_norm": 0.04562624543905258,
      "learning_rate": 7.923529411764705e-05,
      "loss": 0.0167,
      "step": 4255
    },
    {
      "epoch": 38.0,
      "grad_norm": 0.06696755439043045,
      "learning_rate": 7.917647058823528e-05,
      "loss": 0.0192,
      "step": 4256
    },
    {
      "epoch": 38.01,
      "grad_norm": 0.04006008058786392,
      "learning_rate": 7.911764705882352e-05,
      "loss": 0.0157,
      "step": 4257
    },
    {
      "epoch": 38.02,
      "grad_norm": 0.04034001752734184,
      "learning_rate": 7.905882352941177e-05,
      "loss": 0.0196,
      "step": 4258
    },
    {
      "epoch": 38.03,
      "grad_norm": 0.04585159569978714,
      "learning_rate": 7.899999999999998e-05,
      "loss": 0.0194,
      "step": 4259
    },
    {
      "epoch": 38.04,
      "grad_norm": 0.04633148014545441,
      "learning_rate": 7.894117647058823e-05,
      "loss": 0.0168,
      "step": 4260
    },
    {
      "epoch": 38.04,
      "grad_norm": 0.0424240380525589,
      "learning_rate": 7.888235294117647e-05,
      "loss": 0.0163,
      "step": 4261
    },
    {
      "epoch": 38.05,
      "grad_norm": 0.07280178368091583,
      "learning_rate": 7.88235294117647e-05,
      "loss": 0.0228,
      "step": 4262
    },
    {
      "epoch": 38.06,
      "grad_norm": 0.09230028092861176,
      "learning_rate": 7.876470588235294e-05,
      "loss": 0.0218,
      "step": 4263
    },
    {
      "epoch": 38.07,
      "grad_norm": 0.047616370022296906,
      "learning_rate": 7.870588235294116e-05,
      "loss": 0.0151,
      "step": 4264
    },
    {
      "epoch": 38.08,
      "grad_norm": 0.04466402158141136,
      "learning_rate": 7.86470588235294e-05,
      "loss": 0.017,
      "step": 4265
    },
    {
      "epoch": 38.09,
      "grad_norm": 0.04777903854846954,
      "learning_rate": 7.858823529411765e-05,
      "loss": 0.0198,
      "step": 4266
    },
    {
      "epoch": 38.1,
      "grad_norm": 0.04484889283776283,
      "learning_rate": 7.852941176470588e-05,
      "loss": 0.0179,
      "step": 4267
    },
    {
      "epoch": 38.11,
      "grad_norm": 0.06533332914113998,
      "learning_rate": 7.847058823529412e-05,
      "loss": 0.0221,
      "step": 4268
    },
    {
      "epoch": 38.12,
      "grad_norm": 0.10680500417947769,
      "learning_rate": 7.841176470588234e-05,
      "loss": 0.0221,
      "step": 4269
    },
    {
      "epoch": 38.12,
      "grad_norm": 0.03996514156460762,
      "learning_rate": 7.835294117647058e-05,
      "loss": 0.02,
      "step": 4270
    },
    {
      "epoch": 38.13,
      "grad_norm": 0.039795588701963425,
      "learning_rate": 7.829411764705882e-05,
      "loss": 0.0156,
      "step": 4271
    },
    {
      "epoch": 38.14,
      "grad_norm": 0.037465307861566544,
      "learning_rate": 7.823529411764705e-05,
      "loss": 0.0166,
      "step": 4272
    },
    {
      "epoch": 38.15,
      "grad_norm": 0.048377782106399536,
      "learning_rate": 7.817647058823528e-05,
      "loss": 0.0178,
      "step": 4273
    },
    {
      "epoch": 38.16,
      "grad_norm": 0.04678390920162201,
      "learning_rate": 7.811764705882351e-05,
      "loss": 0.0201,
      "step": 4274
    },
    {
      "epoch": 38.17,
      "grad_norm": 0.08644243329763412,
      "learning_rate": 7.805882352941176e-05,
      "loss": 0.0216,
      "step": 4275
    },
    {
      "epoch": 38.18,
      "grad_norm": 0.043905939906835556,
      "learning_rate": 7.8e-05,
      "loss": 0.0187,
      "step": 4276
    },
    {
      "epoch": 38.19,
      "grad_norm": 0.04375680908560753,
      "learning_rate": 7.794117647058823e-05,
      "loss": 0.021,
      "step": 4277
    },
    {
      "epoch": 38.2,
      "grad_norm": 0.04755211994051933,
      "learning_rate": 7.788235294117646e-05,
      "loss": 0.0173,
      "step": 4278
    },
    {
      "epoch": 38.21,
      "grad_norm": 0.04453584924340248,
      "learning_rate": 7.78235294117647e-05,
      "loss": 0.0154,
      "step": 4279
    },
    {
      "epoch": 38.21,
      "grad_norm": 0.04129549860954285,
      "learning_rate": 7.776470588235293e-05,
      "loss": 0.0167,
      "step": 4280
    },
    {
      "epoch": 38.22,
      "grad_norm": 0.0804203599691391,
      "learning_rate": 7.770588235294118e-05,
      "loss": 0.0187,
      "step": 4281
    },
    {
      "epoch": 38.23,
      "grad_norm": 0.04315444827079773,
      "learning_rate": 7.764705882352942e-05,
      "loss": 0.0168,
      "step": 4282
    },
    {
      "epoch": 38.24,
      "grad_norm": 0.0411263182759285,
      "learning_rate": 7.758823529411763e-05,
      "loss": 0.0186,
      "step": 4283
    },
    {
      "epoch": 38.25,
      "grad_norm": 0.05210395157337189,
      "learning_rate": 7.752941176470588e-05,
      "loss": 0.0223,
      "step": 4284
    },
    {
      "epoch": 38.26,
      "grad_norm": 0.05146675929427147,
      "learning_rate": 7.747058823529411e-05,
      "loss": 0.0195,
      "step": 4285
    },
    {
      "epoch": 38.27,
      "grad_norm": 0.05376441404223442,
      "learning_rate": 7.741176470588235e-05,
      "loss": 0.0158,
      "step": 4286
    },
    {
      "epoch": 38.28,
      "grad_norm": 0.059104304760694504,
      "learning_rate": 7.73529411764706e-05,
      "loss": 0.0161,
      "step": 4287
    },
    {
      "epoch": 38.29,
      "grad_norm": 0.05555211752653122,
      "learning_rate": 7.729411764705881e-05,
      "loss": 0.0175,
      "step": 4288
    },
    {
      "epoch": 38.29,
      "grad_norm": 0.03919730708003044,
      "learning_rate": 7.723529411764705e-05,
      "loss": 0.0167,
      "step": 4289
    },
    {
      "epoch": 38.3,
      "grad_norm": 0.05061023309826851,
      "learning_rate": 7.717647058823528e-05,
      "loss": 0.0172,
      "step": 4290
    },
    {
      "epoch": 38.31,
      "grad_norm": 0.04149852320551872,
      "learning_rate": 7.711764705882353e-05,
      "loss": 0.0154,
      "step": 4291
    },
    {
      "epoch": 38.32,
      "grad_norm": 0.04099132493138313,
      "learning_rate": 7.705882352941174e-05,
      "loss": 0.0136,
      "step": 4292
    },
    {
      "epoch": 38.33,
      "grad_norm": 0.0593915656208992,
      "learning_rate": 7.699999999999999e-05,
      "loss": 0.02,
      "step": 4293
    },
    {
      "epoch": 38.34,
      "grad_norm": 0.0756407380104065,
      "learning_rate": 7.694117647058823e-05,
      "loss": 0.0211,
      "step": 4294
    },
    {
      "epoch": 38.35,
      "grad_norm": 0.04416878521442413,
      "learning_rate": 7.688235294117646e-05,
      "loss": 0.0173,
      "step": 4295
    },
    {
      "epoch": 38.36,
      "grad_norm": 0.04282741993665695,
      "learning_rate": 7.68235294117647e-05,
      "loss": 0.0175,
      "step": 4296
    },
    {
      "epoch": 38.37,
      "grad_norm": 0.036725807934999466,
      "learning_rate": 7.676470588235293e-05,
      "loss": 0.0154,
      "step": 4297
    },
    {
      "epoch": 38.38,
      "grad_norm": 0.05139268562197685,
      "learning_rate": 7.670588235294116e-05,
      "loss": 0.0212,
      "step": 4298
    },
    {
      "epoch": 38.38,
      "grad_norm": 0.04303603619337082,
      "learning_rate": 7.664705882352941e-05,
      "loss": 0.0141,
      "step": 4299
    },
    {
      "epoch": 38.39,
      "grad_norm": 0.10238958895206451,
      "learning_rate": 7.658823529411765e-05,
      "loss": 0.0191,
      "step": 4300
    },
    {
      "epoch": 38.39,
      "eval_cer": 0.030653241779511637,
      "eval_loss": 0.2114274501800537,
      "eval_runtime": 22.5509,
      "eval_samples_per_second": 117.157,
      "eval_steps_per_second": 1.862,
      "eval_wer": 0.10063466878222928,
      "step": 4300
    },
    {
      "epoch": 38.4,
      "grad_norm": 0.04634826257824898,
      "learning_rate": 7.652941176470588e-05,
      "loss": 0.0206,
      "step": 4301
    },
    {
      "epoch": 38.41,
      "grad_norm": 0.04878462851047516,
      "learning_rate": 7.647058823529411e-05,
      "loss": 0.0211,
      "step": 4302
    },
    {
      "epoch": 38.42,
      "grad_norm": 0.041815824806690216,
      "learning_rate": 7.641176470588234e-05,
      "loss": 0.0179,
      "step": 4303
    },
    {
      "epoch": 38.43,
      "grad_norm": 0.044155798852443695,
      "learning_rate": 7.635294117647058e-05,
      "loss": 0.0155,
      "step": 4304
    },
    {
      "epoch": 38.44,
      "grad_norm": 0.04693682864308357,
      "learning_rate": 7.629411764705883e-05,
      "loss": 0.0159,
      "step": 4305
    },
    {
      "epoch": 38.45,
      "grad_norm": 0.07583024352788925,
      "learning_rate": 7.623529411764704e-05,
      "loss": 0.0189,
      "step": 4306
    },
    {
      "epoch": 38.46,
      "grad_norm": 0.04122854396700859,
      "learning_rate": 7.617647058823529e-05,
      "loss": 0.0151,
      "step": 4307
    },
    {
      "epoch": 38.46,
      "grad_norm": 0.0455055758357048,
      "learning_rate": 7.611764705882352e-05,
      "loss": 0.0193,
      "step": 4308
    },
    {
      "epoch": 38.47,
      "grad_norm": 0.04155740141868591,
      "learning_rate": 7.605882352941176e-05,
      "loss": 0.0153,
      "step": 4309
    },
    {
      "epoch": 38.48,
      "grad_norm": 0.04967249929904938,
      "learning_rate": 7.6e-05,
      "loss": 0.0151,
      "step": 4310
    },
    {
      "epoch": 38.49,
      "grad_norm": 0.04226543381810188,
      "learning_rate": 7.594117647058822e-05,
      "loss": 0.0155,
      "step": 4311
    },
    {
      "epoch": 38.5,
      "grad_norm": 0.06497447937726974,
      "learning_rate": 7.588235294117646e-05,
      "loss": 0.0171,
      "step": 4312
    },
    {
      "epoch": 38.51,
      "grad_norm": 0.05942431464791298,
      "learning_rate": 7.58235294117647e-05,
      "loss": 0.0176,
      "step": 4313
    },
    {
      "epoch": 38.52,
      "grad_norm": 0.05040823668241501,
      "learning_rate": 7.576470588235294e-05,
      "loss": 0.0171,
      "step": 4314
    },
    {
      "epoch": 38.53,
      "grad_norm": 0.051547449082136154,
      "learning_rate": 7.570588235294118e-05,
      "loss": 0.0191,
      "step": 4315
    },
    {
      "epoch": 38.54,
      "grad_norm": 0.04978009685873985,
      "learning_rate": 7.56470588235294e-05,
      "loss": 0.0178,
      "step": 4316
    },
    {
      "epoch": 38.54,
      "grad_norm": 0.05885303393006325,
      "learning_rate": 7.558823529411764e-05,
      "loss": 0.0193,
      "step": 4317
    },
    {
      "epoch": 38.55,
      "grad_norm": 0.0688789114356041,
      "learning_rate": 7.552941176470588e-05,
      "loss": 0.0202,
      "step": 4318
    },
    {
      "epoch": 38.56,
      "grad_norm": 0.08288246393203735,
      "learning_rate": 7.547058823529411e-05,
      "loss": 0.0181,
      "step": 4319
    },
    {
      "epoch": 38.57,
      "grad_norm": 0.04574671387672424,
      "learning_rate": 7.541176470588234e-05,
      "loss": 0.0188,
      "step": 4320
    },
    {
      "epoch": 38.58,
      "grad_norm": 0.040807656943798065,
      "learning_rate": 7.535294117647057e-05,
      "loss": 0.0183,
      "step": 4321
    },
    {
      "epoch": 38.59,
      "grad_norm": 0.041974350810050964,
      "learning_rate": 7.529411764705882e-05,
      "loss": 0.0168,
      "step": 4322
    },
    {
      "epoch": 38.6,
      "grad_norm": 0.047496818006038666,
      "learning_rate": 7.523529411764706e-05,
      "loss": 0.0168,
      "step": 4323
    },
    {
      "epoch": 38.61,
      "grad_norm": 0.05184216424822807,
      "learning_rate": 7.517647058823529e-05,
      "loss": 0.0192,
      "step": 4324
    },
    {
      "epoch": 38.62,
      "grad_norm": 0.07647722959518433,
      "learning_rate": 7.511764705882352e-05,
      "loss": 0.0216,
      "step": 4325
    },
    {
      "epoch": 38.62,
      "grad_norm": 0.04705112800002098,
      "learning_rate": 7.505882352941175e-05,
      "loss": 0.0194,
      "step": 4326
    },
    {
      "epoch": 38.63,
      "grad_norm": 0.05239846184849739,
      "learning_rate": 7.5e-05,
      "loss": 0.0177,
      "step": 4327
    },
    {
      "epoch": 38.64,
      "grad_norm": 0.03698711097240448,
      "learning_rate": 7.494117647058824e-05,
      "loss": 0.0128,
      "step": 4328
    },
    {
      "epoch": 38.65,
      "grad_norm": 0.04837555065751076,
      "learning_rate": 7.488235294117647e-05,
      "loss": 0.0162,
      "step": 4329
    },
    {
      "epoch": 38.66,
      "grad_norm": 0.04549936205148697,
      "learning_rate": 7.48235294117647e-05,
      "loss": 0.0151,
      "step": 4330
    },
    {
      "epoch": 38.67,
      "grad_norm": 0.08335675299167633,
      "learning_rate": 7.476470588235293e-05,
      "loss": 0.0224,
      "step": 4331
    },
    {
      "epoch": 38.68,
      "grad_norm": 0.040344540029764175,
      "learning_rate": 7.470588235294117e-05,
      "loss": 0.0165,
      "step": 4332
    },
    {
      "epoch": 38.69,
      "grad_norm": 0.04174313321709633,
      "learning_rate": 7.46470588235294e-05,
      "loss": 0.0146,
      "step": 4333
    },
    {
      "epoch": 38.7,
      "grad_norm": 0.04302332177758217,
      "learning_rate": 7.458823529411764e-05,
      "loss": 0.0148,
      "step": 4334
    },
    {
      "epoch": 38.71,
      "grad_norm": 0.049152255058288574,
      "learning_rate": 7.452941176470587e-05,
      "loss": 0.0216,
      "step": 4335
    },
    {
      "epoch": 38.71,
      "grad_norm": 0.04938128590583801,
      "learning_rate": 7.44705882352941e-05,
      "loss": 0.017,
      "step": 4336
    },
    {
      "epoch": 38.72,
      "grad_norm": 0.07563067972660065,
      "learning_rate": 7.441176470588235e-05,
      "loss": 0.0213,
      "step": 4337
    },
    {
      "epoch": 38.73,
      "grad_norm": 0.056373558938503265,
      "learning_rate": 7.435294117647058e-05,
      "loss": 0.018,
      "step": 4338
    },
    {
      "epoch": 38.74,
      "grad_norm": 0.04104134440422058,
      "learning_rate": 7.429411764705882e-05,
      "loss": 0.018,
      "step": 4339
    },
    {
      "epoch": 38.75,
      "grad_norm": 0.03918365389108658,
      "learning_rate": 7.423529411764705e-05,
      "loss": 0.0149,
      "step": 4340
    },
    {
      "epoch": 38.76,
      "grad_norm": 0.052823957055807114,
      "learning_rate": 7.417647058823529e-05,
      "loss": 0.02,
      "step": 4341
    },
    {
      "epoch": 38.77,
      "grad_norm": 0.053039487451314926,
      "learning_rate": 7.411764705882352e-05,
      "loss": 0.0162,
      "step": 4342
    },
    {
      "epoch": 38.78,
      "grad_norm": 0.06652284413576126,
      "learning_rate": 7.405882352941175e-05,
      "loss": 0.0183,
      "step": 4343
    },
    {
      "epoch": 38.79,
      "grad_norm": 0.08099169284105301,
      "learning_rate": 7.4e-05,
      "loss": 0.022,
      "step": 4344
    },
    {
      "epoch": 38.79,
      "grad_norm": 0.038255538791418076,
      "learning_rate": 7.394117647058823e-05,
      "loss": 0.0173,
      "step": 4345
    },
    {
      "epoch": 38.8,
      "grad_norm": 0.04626018926501274,
      "learning_rate": 7.388235294117647e-05,
      "loss": 0.021,
      "step": 4346
    },
    {
      "epoch": 38.81,
      "grad_norm": 0.04810623824596405,
      "learning_rate": 7.38235294117647e-05,
      "loss": 0.0159,
      "step": 4347
    },
    {
      "epoch": 38.82,
      "grad_norm": 0.043795496225357056,
      "learning_rate": 7.376470588235293e-05,
      "loss": 0.0184,
      "step": 4348
    },
    {
      "epoch": 38.83,
      "grad_norm": 0.04710904508829117,
      "learning_rate": 7.370588235294117e-05,
      "loss": 0.0184,
      "step": 4349
    },
    {
      "epoch": 38.84,
      "grad_norm": 0.08242934197187424,
      "learning_rate": 7.36470588235294e-05,
      "loss": 0.0223,
      "step": 4350
    },
    {
      "epoch": 38.85,
      "grad_norm": 0.039028432220220566,
      "learning_rate": 7.358823529411765e-05,
      "loss": 0.0149,
      "step": 4351
    },
    {
      "epoch": 38.86,
      "grad_norm": 0.03827488794922829,
      "learning_rate": 7.352941176470588e-05,
      "loss": 0.014,
      "step": 4352
    },
    {
      "epoch": 38.87,
      "grad_norm": 0.04675621539354324,
      "learning_rate": 7.34705882352941e-05,
      "loss": 0.0199,
      "step": 4353
    },
    {
      "epoch": 38.88,
      "grad_norm": 0.04818340763449669,
      "learning_rate": 7.341176470588235e-05,
      "loss": 0.0174,
      "step": 4354
    },
    {
      "epoch": 38.88,
      "grad_norm": 0.04653319716453552,
      "learning_rate": 7.335294117647058e-05,
      "loss": 0.0168,
      "step": 4355
    },
    {
      "epoch": 38.89,
      "grad_norm": 0.11294801533222198,
      "learning_rate": 7.329411764705882e-05,
      "loss": 0.025,
      "step": 4356
    },
    {
      "epoch": 38.9,
      "grad_norm": 0.04957776889204979,
      "learning_rate": 7.323529411764705e-05,
      "loss": 0.0183,
      "step": 4357
    },
    {
      "epoch": 38.91,
      "grad_norm": 0.04400325566530228,
      "learning_rate": 7.31764705882353e-05,
      "loss": 0.0179,
      "step": 4358
    },
    {
      "epoch": 38.92,
      "grad_norm": 0.04249448701739311,
      "learning_rate": 7.311764705882353e-05,
      "loss": 0.016,
      "step": 4359
    },
    {
      "epoch": 38.93,
      "grad_norm": 0.041930440813302994,
      "learning_rate": 7.305882352941176e-05,
      "loss": 0.0176,
      "step": 4360
    },
    {
      "epoch": 38.94,
      "grad_norm": 0.04815790429711342,
      "learning_rate": 7.3e-05,
      "loss": 0.018,
      "step": 4361
    },
    {
      "epoch": 38.95,
      "grad_norm": 0.06433267146348953,
      "learning_rate": 7.294117647058823e-05,
      "loss": 0.0176,
      "step": 4362
    },
    {
      "epoch": 38.96,
      "grad_norm": 0.09689892828464508,
      "learning_rate": 7.288235294117647e-05,
      "loss": 0.0342,
      "step": 4363
    },
    {
      "epoch": 38.96,
      "grad_norm": 0.04072459414601326,
      "learning_rate": 7.28235294117647e-05,
      "loss": 0.017,
      "step": 4364
    },
    {
      "epoch": 38.97,
      "grad_norm": 0.04257611185312271,
      "learning_rate": 7.276470588235293e-05,
      "loss": 0.0188,
      "step": 4365
    },
    {
      "epoch": 38.98,
      "grad_norm": 0.04054246470332146,
      "learning_rate": 7.270588235294116e-05,
      "loss": 0.0182,
      "step": 4366
    },
    {
      "epoch": 38.99,
      "grad_norm": 0.05006944015622139,
      "learning_rate": 7.26470588235294e-05,
      "loss": 0.0189,
      "step": 4367
    },
    {
      "epoch": 39.0,
      "grad_norm": 0.06838885694742203,
      "learning_rate": 7.258823529411765e-05,
      "loss": 0.0183,
      "step": 4368
    },
    {
      "epoch": 39.01,
      "grad_norm": 0.03908037766814232,
      "learning_rate": 7.252941176470588e-05,
      "loss": 0.0162,
      "step": 4369
    },
    {
      "epoch": 39.02,
      "grad_norm": 0.03869599848985672,
      "learning_rate": 7.247058823529411e-05,
      "loss": 0.0168,
      "step": 4370
    },
    {
      "epoch": 39.03,
      "grad_norm": 0.040932200849056244,
      "learning_rate": 7.241176470588234e-05,
      "loss": 0.0166,
      "step": 4371
    },
    {
      "epoch": 39.04,
      "grad_norm": 0.04948969930410385,
      "learning_rate": 7.235294117647058e-05,
      "loss": 0.0189,
      "step": 4372
    },
    {
      "epoch": 39.04,
      "grad_norm": 0.04653513804078102,
      "learning_rate": 7.229411764705881e-05,
      "loss": 0.017,
      "step": 4373
    },
    {
      "epoch": 39.05,
      "grad_norm": 0.1132410317659378,
      "learning_rate": 7.223529411764706e-05,
      "loss": 0.024,
      "step": 4374
    },
    {
      "epoch": 39.06,
      "grad_norm": 0.07041914016008377,
      "learning_rate": 7.217647058823529e-05,
      "loss": 0.0245,
      "step": 4375
    },
    {
      "epoch": 39.07,
      "grad_norm": 0.044276244938373566,
      "learning_rate": 7.211764705882351e-05,
      "loss": 0.0171,
      "step": 4376
    },
    {
      "epoch": 39.08,
      "grad_norm": 0.04436330497264862,
      "learning_rate": 7.205882352941176e-05,
      "loss": 0.0176,
      "step": 4377
    },
    {
      "epoch": 39.09,
      "grad_norm": 0.04595024138689041,
      "learning_rate": 7.199999999999999e-05,
      "loss": 0.018,
      "step": 4378
    },
    {
      "epoch": 39.1,
      "grad_norm": 0.04338918626308441,
      "learning_rate": 7.194117647058823e-05,
      "loss": 0.0146,
      "step": 4379
    },
    {
      "epoch": 39.11,
      "grad_norm": 0.05065865069627762,
      "learning_rate": 7.188235294117646e-05,
      "loss": 0.0159,
      "step": 4380
    },
    {
      "epoch": 39.12,
      "grad_norm": 0.10124563425779343,
      "learning_rate": 7.18235294117647e-05,
      "loss": 0.0232,
      "step": 4381
    },
    {
      "epoch": 39.12,
      "grad_norm": 0.04452907294034958,
      "learning_rate": 7.176470588235293e-05,
      "loss": 0.0187,
      "step": 4382
    },
    {
      "epoch": 39.13,
      "grad_norm": 0.04646245390176773,
      "learning_rate": 7.170588235294116e-05,
      "loss": 0.0181,
      "step": 4383
    },
    {
      "epoch": 39.14,
      "grad_norm": 0.045267120003700256,
      "learning_rate": 7.164705882352941e-05,
      "loss": 0.0167,
      "step": 4384
    },
    {
      "epoch": 39.15,
      "grad_norm": 0.04318848252296448,
      "learning_rate": 7.158823529411764e-05,
      "loss": 0.0161,
      "step": 4385
    },
    {
      "epoch": 39.16,
      "grad_norm": 0.05045012757182121,
      "learning_rate": 7.152941176470588e-05,
      "loss": 0.0166,
      "step": 4386
    },
    {
      "epoch": 39.17,
      "grad_norm": 0.08273526281118393,
      "learning_rate": 7.147058823529411e-05,
      "loss": 0.0214,
      "step": 4387
    },
    {
      "epoch": 39.18,
      "grad_norm": 0.03756927698850632,
      "learning_rate": 7.141176470588234e-05,
      "loss": 0.0162,
      "step": 4388
    },
    {
      "epoch": 39.19,
      "grad_norm": 0.042943406850099564,
      "learning_rate": 7.135294117647058e-05,
      "loss": 0.018,
      "step": 4389
    },
    {
      "epoch": 39.2,
      "grad_norm": 0.03948609158396721,
      "learning_rate": 7.129411764705881e-05,
      "loss": 0.0166,
      "step": 4390
    },
    {
      "epoch": 39.21,
      "grad_norm": 0.03829723596572876,
      "learning_rate": 7.123529411764706e-05,
      "loss": 0.0154,
      "step": 4391
    },
    {
      "epoch": 39.21,
      "grad_norm": 0.052331846207380295,
      "learning_rate": 7.117647058823529e-05,
      "loss": 0.0165,
      "step": 4392
    },
    {
      "epoch": 39.22,
      "grad_norm": 0.06658195704221725,
      "learning_rate": 7.111764705882352e-05,
      "loss": 0.0143,
      "step": 4393
    },
    {
      "epoch": 39.23,
      "grad_norm": 0.04346032440662384,
      "learning_rate": 7.105882352941176e-05,
      "loss": 0.0195,
      "step": 4394
    },
    {
      "epoch": 39.24,
      "grad_norm": 0.04886326938867569,
      "learning_rate": 7.099999999999999e-05,
      "loss": 0.0224,
      "step": 4395
    },
    {
      "epoch": 39.25,
      "grad_norm": 0.04518451169133186,
      "learning_rate": 7.094117647058823e-05,
      "loss": 0.0157,
      "step": 4396
    },
    {
      "epoch": 39.26,
      "grad_norm": 0.04408060014247894,
      "learning_rate": 7.088235294117646e-05,
      "loss": 0.0166,
      "step": 4397
    },
    {
      "epoch": 39.27,
      "grad_norm": 0.06991055607795715,
      "learning_rate": 7.082352941176471e-05,
      "loss": 0.0185,
      "step": 4398
    },
    {
      "epoch": 39.28,
      "grad_norm": 0.0835832804441452,
      "learning_rate": 7.076470588235294e-05,
      "loss": 0.021,
      "step": 4399
    },
    {
      "epoch": 39.29,
      "grad_norm": 0.20428909361362457,
      "learning_rate": 7.070588235294117e-05,
      "loss": 0.0218,
      "step": 4400
    },
    {
      "epoch": 39.29,
      "eval_cer": 0.031431223761899926,
      "eval_loss": 0.20662860572338104,
      "eval_runtime": 22.7039,
      "eval_samples_per_second": 116.368,
      "eval_steps_per_second": 1.85,
      "eval_wer": 0.1014676715589052,
      "step": 4400
    },
    {
      "epoch": 39.29,
      "grad_norm": 0.044423457235097885,
      "learning_rate": 7.064705882352941e-05,
      "loss": 0.0197,
      "step": 4401
    },
    {
      "epoch": 39.3,
      "grad_norm": 0.04140207916498184,
      "learning_rate": 7.058823529411764e-05,
      "loss": 0.0176,
      "step": 4402
    },
    {
      "epoch": 39.31,
      "grad_norm": 0.04382094740867615,
      "learning_rate": 7.052941176470588e-05,
      "loss": 0.0157,
      "step": 4403
    },
    {
      "epoch": 39.32,
      "grad_norm": 0.05342730134725571,
      "learning_rate": 7.047058823529411e-05,
      "loss": 0.016,
      "step": 4404
    },
    {
      "epoch": 39.33,
      "grad_norm": 0.05287891626358032,
      "learning_rate": 7.041176470588234e-05,
      "loss": 0.0135,
      "step": 4405
    },
    {
      "epoch": 39.34,
      "grad_norm": 0.08517567068338394,
      "learning_rate": 7.035294117647057e-05,
      "loss": 0.0209,
      "step": 4406
    },
    {
      "epoch": 39.35,
      "grad_norm": 0.04383644461631775,
      "learning_rate": 7.029411764705882e-05,
      "loss": 0.0189,
      "step": 4407
    },
    {
      "epoch": 39.36,
      "grad_norm": 0.04496624693274498,
      "learning_rate": 7.023529411764706e-05,
      "loss": 0.0167,
      "step": 4408
    },
    {
      "epoch": 39.37,
      "grad_norm": 0.0436580665409565,
      "learning_rate": 7.017647058823529e-05,
      "loss": 0.017,
      "step": 4409
    },
    {
      "epoch": 39.38,
      "grad_norm": 0.04130517691373825,
      "learning_rate": 7.011764705882352e-05,
      "loss": 0.0184,
      "step": 4410
    },
    {
      "epoch": 39.38,
      "grad_norm": 0.05226563289761543,
      "learning_rate": 7.005882352941175e-05,
      "loss": 0.0139,
      "step": 4411
    },
    {
      "epoch": 39.39,
      "grad_norm": 0.10196669399738312,
      "learning_rate": 7e-05,
      "loss": 0.0243,
      "step": 4412
    },
    {
      "epoch": 39.4,
      "grad_norm": 0.04151444882154465,
      "learning_rate": 6.994117647058822e-05,
      "loss": 0.0179,
      "step": 4413
    },
    {
      "epoch": 39.41,
      "grad_norm": 0.035112831741571426,
      "learning_rate": 6.988235294117647e-05,
      "loss": 0.016,
      "step": 4414
    },
    {
      "epoch": 39.42,
      "grad_norm": 0.043129149824380875,
      "learning_rate": 6.98235294117647e-05,
      "loss": 0.0169,
      "step": 4415
    },
    {
      "epoch": 39.43,
      "grad_norm": 0.0493992418050766,
      "learning_rate": 6.976470588235293e-05,
      "loss": 0.0201,
      "step": 4416
    },
    {
      "epoch": 39.44,
      "grad_norm": 0.04564020782709122,
      "learning_rate": 6.970588235294117e-05,
      "loss": 0.0146,
      "step": 4417
    },
    {
      "epoch": 39.45,
      "grad_norm": 0.08126378059387207,
      "learning_rate": 6.96470588235294e-05,
      "loss": 0.0221,
      "step": 4418
    },
    {
      "epoch": 39.46,
      "grad_norm": 0.04035772755742073,
      "learning_rate": 6.958823529411764e-05,
      "loss": 0.0179,
      "step": 4419
    },
    {
      "epoch": 39.46,
      "grad_norm": 0.03929206356406212,
      "learning_rate": 6.952941176470587e-05,
      "loss": 0.0151,
      "step": 4420
    },
    {
      "epoch": 39.47,
      "grad_norm": 0.04197614639997482,
      "learning_rate": 6.947058823529412e-05,
      "loss": 0.0192,
      "step": 4421
    },
    {
      "epoch": 39.48,
      "grad_norm": 0.04631732031702995,
      "learning_rate": 6.941176470588235e-05,
      "loss": 0.0178,
      "step": 4422
    },
    {
      "epoch": 39.49,
      "grad_norm": 0.04176364466547966,
      "learning_rate": 6.935294117647058e-05,
      "loss": 0.0158,
      "step": 4423
    },
    {
      "epoch": 39.5,
      "grad_norm": 0.054150938987731934,
      "learning_rate": 6.929411764705882e-05,
      "loss": 0.0187,
      "step": 4424
    },
    {
      "epoch": 39.51,
      "grad_norm": 0.07182719558477402,
      "learning_rate": 6.923529411764705e-05,
      "loss": 0.0203,
      "step": 4425
    },
    {
      "epoch": 39.52,
      "grad_norm": 0.043137047439813614,
      "learning_rate": 6.91764705882353e-05,
      "loss": 0.0177,
      "step": 4426
    },
    {
      "epoch": 39.53,
      "grad_norm": 0.040281008929014206,
      "learning_rate": 6.911764705882352e-05,
      "loss": 0.0163,
      "step": 4427
    },
    {
      "epoch": 39.54,
      "grad_norm": 0.0436340756714344,
      "learning_rate": 6.905882352941175e-05,
      "loss": 0.0146,
      "step": 4428
    },
    {
      "epoch": 39.54,
      "grad_norm": 0.05586494505405426,
      "learning_rate": 6.9e-05,
      "loss": 0.0181,
      "step": 4429
    },
    {
      "epoch": 39.55,
      "grad_norm": 0.053915951400995255,
      "learning_rate": 6.894117647058823e-05,
      "loss": 0.0156,
      "step": 4430
    },
    {
      "epoch": 39.56,
      "grad_norm": 0.09184318780899048,
      "learning_rate": 6.888235294117647e-05,
      "loss": 0.0257,
      "step": 4431
    },
    {
      "epoch": 39.57,
      "grad_norm": 0.04375550150871277,
      "learning_rate": 6.88235294117647e-05,
      "loss": 0.015,
      "step": 4432
    },
    {
      "epoch": 39.58,
      "grad_norm": 0.04235227778553963,
      "learning_rate": 6.876470588235293e-05,
      "loss": 0.0168,
      "step": 4433
    },
    {
      "epoch": 39.59,
      "grad_norm": 0.04491400346159935,
      "learning_rate": 6.870588235294117e-05,
      "loss": 0.0184,
      "step": 4434
    },
    {
      "epoch": 39.6,
      "grad_norm": 0.0541360005736351,
      "learning_rate": 6.86470588235294e-05,
      "loss": 0.0157,
      "step": 4435
    },
    {
      "epoch": 39.61,
      "grad_norm": 0.048810943961143494,
      "learning_rate": 6.858823529411765e-05,
      "loss": 0.0155,
      "step": 4436
    },
    {
      "epoch": 39.62,
      "grad_norm": 0.07412078976631165,
      "learning_rate": 6.852941176470588e-05,
      "loss": 0.0175,
      "step": 4437
    },
    {
      "epoch": 39.62,
      "grad_norm": 0.04780080169439316,
      "learning_rate": 6.847058823529412e-05,
      "loss": 0.0191,
      "step": 4438
    },
    {
      "epoch": 39.63,
      "grad_norm": 0.043971408158540726,
      "learning_rate": 6.841176470588235e-05,
      "loss": 0.0177,
      "step": 4439
    },
    {
      "epoch": 39.64,
      "grad_norm": 0.03924727067351341,
      "learning_rate": 6.835294117647058e-05,
      "loss": 0.0156,
      "step": 4440
    },
    {
      "epoch": 39.65,
      "grad_norm": 0.05375504493713379,
      "learning_rate": 6.829411764705882e-05,
      "loss": 0.0203,
      "step": 4441
    },
    {
      "epoch": 39.66,
      "grad_norm": 0.05465371906757355,
      "learning_rate": 6.823529411764705e-05,
      "loss": 0.0202,
      "step": 4442
    },
    {
      "epoch": 39.67,
      "grad_norm": 0.10145536810159683,
      "learning_rate": 6.81764705882353e-05,
      "loss": 0.022,
      "step": 4443
    },
    {
      "epoch": 39.68,
      "grad_norm": 0.039379965513944626,
      "learning_rate": 6.811764705882353e-05,
      "loss": 0.0159,
      "step": 4444
    },
    {
      "epoch": 39.69,
      "grad_norm": 0.04439634829759598,
      "learning_rate": 6.805882352941176e-05,
      "loss": 0.0162,
      "step": 4445
    },
    {
      "epoch": 39.7,
      "grad_norm": 0.04451480880379677,
      "learning_rate": 6.799999999999999e-05,
      "loss": 0.0201,
      "step": 4446
    },
    {
      "epoch": 39.71,
      "grad_norm": 0.04450339451432228,
      "learning_rate": 6.794117647058823e-05,
      "loss": 0.0163,
      "step": 4447
    },
    {
      "epoch": 39.71,
      "grad_norm": 0.058022525161504745,
      "learning_rate": 6.788235294117647e-05,
      "loss": 0.0199,
      "step": 4448
    },
    {
      "epoch": 39.72,
      "grad_norm": 0.0665673092007637,
      "learning_rate": 6.78235294117647e-05,
      "loss": 0.0201,
      "step": 4449
    },
    {
      "epoch": 39.73,
      "grad_norm": 0.053832486271858215,
      "learning_rate": 6.776470588235293e-05,
      "loss": 0.018,
      "step": 4450
    },
    {
      "epoch": 39.74,
      "grad_norm": 0.03415067121386528,
      "learning_rate": 6.770588235294116e-05,
      "loss": 0.0145,
      "step": 4451
    },
    {
      "epoch": 39.75,
      "grad_norm": 0.04492098465561867,
      "learning_rate": 6.76470588235294e-05,
      "loss": 0.0179,
      "step": 4452
    },
    {
      "epoch": 39.76,
      "grad_norm": 0.03915470838546753,
      "learning_rate": 6.758823529411764e-05,
      "loss": 0.0178,
      "step": 4453
    },
    {
      "epoch": 39.77,
      "grad_norm": 0.04149465635418892,
      "learning_rate": 6.752941176470588e-05,
      "loss": 0.018,
      "step": 4454
    },
    {
      "epoch": 39.78,
      "grad_norm": 0.06206176429986954,
      "learning_rate": 6.747058823529411e-05,
      "loss": 0.0223,
      "step": 4455
    },
    {
      "epoch": 39.79,
      "grad_norm": 0.07275174558162689,
      "learning_rate": 6.741176470588235e-05,
      "loss": 0.02,
      "step": 4456
    },
    {
      "epoch": 39.79,
      "grad_norm": 0.04092457890510559,
      "learning_rate": 6.735294117647058e-05,
      "loss": 0.0172,
      "step": 4457
    },
    {
      "epoch": 39.8,
      "grad_norm": 0.046152371913194656,
      "learning_rate": 6.729411764705881e-05,
      "loss": 0.0161,
      "step": 4458
    },
    {
      "epoch": 39.81,
      "grad_norm": 0.04276079684495926,
      "learning_rate": 6.723529411764706e-05,
      "loss": 0.0133,
      "step": 4459
    },
    {
      "epoch": 39.82,
      "grad_norm": 0.0418611615896225,
      "learning_rate": 6.717647058823529e-05,
      "loss": 0.015,
      "step": 4460
    },
    {
      "epoch": 39.83,
      "grad_norm": 0.06505148112773895,
      "learning_rate": 6.711764705882353e-05,
      "loss": 0.0163,
      "step": 4461
    },
    {
      "epoch": 39.84,
      "grad_norm": 0.12159974128007889,
      "learning_rate": 6.705882352941176e-05,
      "loss": 0.0279,
      "step": 4462
    },
    {
      "epoch": 39.85,
      "grad_norm": 0.041626088321208954,
      "learning_rate": 6.699999999999999e-05,
      "loss": 0.0167,
      "step": 4463
    },
    {
      "epoch": 39.86,
      "grad_norm": 0.04645446315407753,
      "learning_rate": 6.694117647058823e-05,
      "loss": 0.0165,
      "step": 4464
    },
    {
      "epoch": 39.87,
      "grad_norm": 0.04155229032039642,
      "learning_rate": 6.688235294117646e-05,
      "loss": 0.015,
      "step": 4465
    },
    {
      "epoch": 39.88,
      "grad_norm": 0.05169856548309326,
      "learning_rate": 6.68235294117647e-05,
      "loss": 0.0178,
      "step": 4466
    },
    {
      "epoch": 39.88,
      "grad_norm": 0.04497070983052254,
      "learning_rate": 6.676470588235294e-05,
      "loss": 0.0154,
      "step": 4467
    },
    {
      "epoch": 39.89,
      "grad_norm": 0.11043273657560349,
      "learning_rate": 6.670588235294117e-05,
      "loss": 0.0293,
      "step": 4468
    },
    {
      "epoch": 39.9,
      "grad_norm": 0.041215650737285614,
      "learning_rate": 6.664705882352941e-05,
      "loss": 0.0193,
      "step": 4469
    },
    {
      "epoch": 39.91,
      "grad_norm": 0.04200184717774391,
      "learning_rate": 6.658823529411764e-05,
      "loss": 0.0177,
      "step": 4470
    },
    {
      "epoch": 39.92,
      "grad_norm": 0.035983189940452576,
      "learning_rate": 6.652941176470588e-05,
      "loss": 0.0131,
      "step": 4471
    },
    {
      "epoch": 39.93,
      "grad_norm": 0.042236682027578354,
      "learning_rate": 6.647058823529411e-05,
      "loss": 0.0138,
      "step": 4472
    },
    {
      "epoch": 39.94,
      "grad_norm": 0.05005483701825142,
      "learning_rate": 6.641176470588234e-05,
      "loss": 0.0147,
      "step": 4473
    },
    {
      "epoch": 39.95,
      "grad_norm": 0.06001787260174751,
      "learning_rate": 6.635294117647059e-05,
      "loss": 0.0146,
      "step": 4474
    },
    {
      "epoch": 39.96,
      "grad_norm": 0.07263220101594925,
      "learning_rate": 6.629411764705882e-05,
      "loss": 0.0212,
      "step": 4475
    },
    {
      "epoch": 39.96,
      "grad_norm": 0.04336221143603325,
      "learning_rate": 6.623529411764706e-05,
      "loss": 0.0196,
      "step": 4476
    },
    {
      "epoch": 39.97,
      "grad_norm": 0.05087766796350479,
      "learning_rate": 6.617647058823529e-05,
      "loss": 0.0202,
      "step": 4477
    },
    {
      "epoch": 39.98,
      "grad_norm": 0.04861607402563095,
      "learning_rate": 6.611764705882353e-05,
      "loss": 0.0185,
      "step": 4478
    },
    {
      "epoch": 39.99,
      "grad_norm": 0.043793126940727234,
      "learning_rate": 6.605882352941176e-05,
      "loss": 0.0137,
      "step": 4479
    },
    {
      "epoch": 40.0,
      "grad_norm": 0.05250184237957001,
      "learning_rate": 6.599999999999999e-05,
      "loss": 0.0179,
      "step": 4480
    },
    {
      "epoch": 40.01,
      "grad_norm": 0.04231712967157364,
      "learning_rate": 6.594117647058823e-05,
      "loss": 0.0181,
      "step": 4481
    },
    {
      "epoch": 40.02,
      "grad_norm": 0.04119037091732025,
      "learning_rate": 6.588235294117646e-05,
      "loss": 0.0162,
      "step": 4482
    },
    {
      "epoch": 40.03,
      "grad_norm": 0.04248598963022232,
      "learning_rate": 6.582352941176471e-05,
      "loss": 0.0175,
      "step": 4483
    },
    {
      "epoch": 40.04,
      "grad_norm": 0.04289815574884415,
      "learning_rate": 6.576470588235294e-05,
      "loss": 0.0173,
      "step": 4484
    },
    {
      "epoch": 40.04,
      "grad_norm": 0.04525744542479515,
      "learning_rate": 6.570588235294117e-05,
      "loss": 0.0172,
      "step": 4485
    },
    {
      "epoch": 40.05,
      "grad_norm": 0.07593417912721634,
      "learning_rate": 6.56470588235294e-05,
      "loss": 0.0168,
      "step": 4486
    },
    {
      "epoch": 40.06,
      "grad_norm": 0.07838865369558334,
      "learning_rate": 6.558823529411764e-05,
      "loss": 0.0159,
      "step": 4487
    },
    {
      "epoch": 40.07,
      "grad_norm": 0.0439491868019104,
      "learning_rate": 6.552941176470588e-05,
      "loss": 0.0165,
      "step": 4488
    },
    {
      "epoch": 40.08,
      "grad_norm": 0.0446203351020813,
      "learning_rate": 6.547058823529411e-05,
      "loss": 0.0171,
      "step": 4489
    },
    {
      "epoch": 40.09,
      "grad_norm": 0.04487207904458046,
      "learning_rate": 6.541176470588234e-05,
      "loss": 0.0168,
      "step": 4490
    },
    {
      "epoch": 40.1,
      "grad_norm": 0.046270936727523804,
      "learning_rate": 6.535294117647057e-05,
      "loss": 0.0171,
      "step": 4491
    },
    {
      "epoch": 40.11,
      "grad_norm": 0.06900790333747864,
      "learning_rate": 6.529411764705882e-05,
      "loss": 0.0198,
      "step": 4492
    },
    {
      "epoch": 40.12,
      "grad_norm": 0.0703432485461235,
      "learning_rate": 6.523529411764705e-05,
      "loss": 0.0219,
      "step": 4493
    },
    {
      "epoch": 40.12,
      "grad_norm": 0.04324033483862877,
      "learning_rate": 6.517647058823529e-05,
      "loss": 0.0188,
      "step": 4494
    },
    {
      "epoch": 40.13,
      "grad_norm": 0.044322215020656586,
      "learning_rate": 6.511764705882352e-05,
      "loss": 0.0189,
      "step": 4495
    },
    {
      "epoch": 40.14,
      "grad_norm": 0.04601391777396202,
      "learning_rate": 6.505882352941176e-05,
      "loss": 0.0176,
      "step": 4496
    },
    {
      "epoch": 40.15,
      "grad_norm": 0.04588237777352333,
      "learning_rate": 6.5e-05,
      "loss": 0.0147,
      "step": 4497
    },
    {
      "epoch": 40.16,
      "grad_norm": 0.04775720089673996,
      "learning_rate": 6.494117647058822e-05,
      "loss": 0.0145,
      "step": 4498
    },
    {
      "epoch": 40.17,
      "grad_norm": 0.08080917596817017,
      "learning_rate": 6.488235294117647e-05,
      "loss": 0.0193,
      "step": 4499
    },
    {
      "epoch": 40.18,
      "grad_norm": 0.04132635518908501,
      "learning_rate": 6.48235294117647e-05,
      "loss": 0.0182,
      "step": 4500
    },
    {
      "epoch": 40.18,
      "eval_cer": 0.030010561011451744,
      "eval_loss": 0.20539185404777527,
      "eval_runtime": 22.3198,
      "eval_samples_per_second": 118.37,
      "eval_steps_per_second": 1.882,
      "eval_wer": 0.09884966283220945,
      "step": 4500
    },
    {
      "epoch": 40.19,
      "grad_norm": 0.04232960566878319,
      "learning_rate": 6.476470588235294e-05,
      "loss": 0.0183,
      "step": 4501
    },
    {
      "epoch": 40.2,
      "grad_norm": 0.04780032858252525,
      "learning_rate": 6.470588235294117e-05,
      "loss": 0.021,
      "step": 4502
    },
    {
      "epoch": 40.21,
      "grad_norm": 0.05101853981614113,
      "learning_rate": 6.46470588235294e-05,
      "loss": 0.0179,
      "step": 4503
    },
    {
      "epoch": 40.21,
      "grad_norm": 0.05485577508807182,
      "learning_rate": 6.458823529411764e-05,
      "loss": 0.0171,
      "step": 4504
    },
    {
      "epoch": 40.22,
      "grad_norm": 0.09795645624399185,
      "learning_rate": 6.452941176470587e-05,
      "loss": 0.0246,
      "step": 4505
    },
    {
      "epoch": 40.23,
      "grad_norm": 0.04066602513194084,
      "learning_rate": 6.447058823529412e-05,
      "loss": 0.0166,
      "step": 4506
    },
    {
      "epoch": 40.24,
      "grad_norm": 0.040202949196100235,
      "learning_rate": 6.441176470588235e-05,
      "loss": 0.018,
      "step": 4507
    },
    {
      "epoch": 40.25,
      "grad_norm": 0.043347131460905075,
      "learning_rate": 6.435294117647058e-05,
      "loss": 0.0188,
      "step": 4508
    },
    {
      "epoch": 40.26,
      "grad_norm": 0.04132955148816109,
      "learning_rate": 6.429411764705882e-05,
      "loss": 0.0144,
      "step": 4509
    },
    {
      "epoch": 40.27,
      "grad_norm": 0.04157760739326477,
      "learning_rate": 6.423529411764705e-05,
      "loss": 0.0145,
      "step": 4510
    },
    {
      "epoch": 40.28,
      "grad_norm": 0.060481470078229904,
      "learning_rate": 6.41764705882353e-05,
      "loss": 0.016,
      "step": 4511
    },
    {
      "epoch": 40.29,
      "grad_norm": 0.05811312794685364,
      "learning_rate": 6.411764705882352e-05,
      "loss": 0.0212,
      "step": 4512
    },
    {
      "epoch": 40.29,
      "grad_norm": 0.04830561950802803,
      "learning_rate": 6.405882352941175e-05,
      "loss": 0.0222,
      "step": 4513
    },
    {
      "epoch": 40.3,
      "grad_norm": 0.04640211910009384,
      "learning_rate": 6.4e-05,
      "loss": 0.0158,
      "step": 4514
    },
    {
      "epoch": 40.31,
      "grad_norm": 0.04630252346396446,
      "learning_rate": 6.394117647058823e-05,
      "loss": 0.0182,
      "step": 4515
    },
    {
      "epoch": 40.32,
      "grad_norm": 0.044155918061733246,
      "learning_rate": 6.388235294117647e-05,
      "loss": 0.0167,
      "step": 4516
    },
    {
      "epoch": 40.33,
      "grad_norm": 0.07158509641885757,
      "learning_rate": 6.38235294117647e-05,
      "loss": 0.023,
      "step": 4517
    },
    {
      "epoch": 40.34,
      "grad_norm": 0.0738370344042778,
      "learning_rate": 6.376470588235294e-05,
      "loss": 0.0217,
      "step": 4518
    },
    {
      "epoch": 40.35,
      "grad_norm": 0.04584616795182228,
      "learning_rate": 6.370588235294117e-05,
      "loss": 0.0181,
      "step": 4519
    },
    {
      "epoch": 40.36,
      "grad_norm": 0.048624083399772644,
      "learning_rate": 6.36470588235294e-05,
      "loss": 0.0194,
      "step": 4520
    },
    {
      "epoch": 40.37,
      "grad_norm": 0.04459695890545845,
      "learning_rate": 6.358823529411765e-05,
      "loss": 0.0185,
      "step": 4521
    },
    {
      "epoch": 40.38,
      "grad_norm": 0.04149356484413147,
      "learning_rate": 6.352941176470588e-05,
      "loss": 0.015,
      "step": 4522
    },
    {
      "epoch": 40.38,
      "grad_norm": 0.06206259876489639,
      "learning_rate": 6.347058823529412e-05,
      "loss": 0.0196,
      "step": 4523
    },
    {
      "epoch": 40.39,
      "grad_norm": 0.08526722341775894,
      "learning_rate": 6.341176470588235e-05,
      "loss": 0.021,
      "step": 4524
    },
    {
      "epoch": 40.4,
      "grad_norm": 0.04059320315718651,
      "learning_rate": 6.335294117647058e-05,
      "loss": 0.0152,
      "step": 4525
    },
    {
      "epoch": 40.41,
      "grad_norm": 0.03885188326239586,
      "learning_rate": 6.329411764705881e-05,
      "loss": 0.0158,
      "step": 4526
    },
    {
      "epoch": 40.42,
      "grad_norm": 0.0395585261285305,
      "learning_rate": 6.323529411764705e-05,
      "loss": 0.0139,
      "step": 4527
    },
    {
      "epoch": 40.43,
      "grad_norm": 0.04279087111353874,
      "learning_rate": 6.31764705882353e-05,
      "loss": 0.0165,
      "step": 4528
    },
    {
      "epoch": 40.44,
      "grad_norm": 0.03796738386154175,
      "learning_rate": 6.311764705882353e-05,
      "loss": 0.0114,
      "step": 4529
    },
    {
      "epoch": 40.45,
      "grad_norm": 0.0947808027267456,
      "learning_rate": 6.305882352941176e-05,
      "loss": 0.0257,
      "step": 4530
    },
    {
      "epoch": 40.46,
      "grad_norm": 0.04211699590086937,
      "learning_rate": 6.299999999999999e-05,
      "loss": 0.0173,
      "step": 4531
    },
    {
      "epoch": 40.46,
      "grad_norm": 0.04453981667757034,
      "learning_rate": 6.294117647058823e-05,
      "loss": 0.0191,
      "step": 4532
    },
    {
      "epoch": 40.47,
      "grad_norm": 0.047891560941934586,
      "learning_rate": 6.288235294117646e-05,
      "loss": 0.0171,
      "step": 4533
    },
    {
      "epoch": 40.48,
      "grad_norm": 0.055286191403865814,
      "learning_rate": 6.28235294117647e-05,
      "loss": 0.0177,
      "step": 4534
    },
    {
      "epoch": 40.49,
      "grad_norm": 0.05157831311225891,
      "learning_rate": 6.276470588235293e-05,
      "loss": 0.019,
      "step": 4535
    },
    {
      "epoch": 40.5,
      "grad_norm": 0.05864199250936508,
      "learning_rate": 6.270588235294118e-05,
      "loss": 0.0171,
      "step": 4536
    },
    {
      "epoch": 40.51,
      "grad_norm": 0.0728534385561943,
      "learning_rate": 6.26470588235294e-05,
      "loss": 0.0193,
      "step": 4537
    },
    {
      "epoch": 40.52,
      "grad_norm": 0.040822833776474,
      "learning_rate": 6.258823529411764e-05,
      "loss": 0.0199,
      "step": 4538
    },
    {
      "epoch": 40.53,
      "grad_norm": 0.04746582359075546,
      "learning_rate": 6.252941176470588e-05,
      "loss": 0.0182,
      "step": 4539
    },
    {
      "epoch": 40.54,
      "grad_norm": 0.037793178111314774,
      "learning_rate": 6.247058823529411e-05,
      "loss": 0.0151,
      "step": 4540
    },
    {
      "epoch": 40.54,
      "grad_norm": 0.044214483350515366,
      "learning_rate": 6.241176470588235e-05,
      "loss": 0.0188,
      "step": 4541
    },
    {
      "epoch": 40.55,
      "grad_norm": 0.05251981317996979,
      "learning_rate": 6.235294117647058e-05,
      "loss": 0.0166,
      "step": 4542
    },
    {
      "epoch": 40.56,
      "grad_norm": 0.0751509889960289,
      "learning_rate": 6.229411764705881e-05,
      "loss": 0.0212,
      "step": 4543
    },
    {
      "epoch": 40.57,
      "grad_norm": 0.038863301277160645,
      "learning_rate": 6.223529411764706e-05,
      "loss": 0.0176,
      "step": 4544
    },
    {
      "epoch": 40.58,
      "grad_norm": 0.041233357042074203,
      "learning_rate": 6.217647058823529e-05,
      "loss": 0.017,
      "step": 4545
    },
    {
      "epoch": 40.59,
      "grad_norm": 0.049184199422597885,
      "learning_rate": 6.211764705882353e-05,
      "loss": 0.0178,
      "step": 4546
    },
    {
      "epoch": 40.6,
      "grad_norm": 0.04078080132603645,
      "learning_rate": 6.205882352941176e-05,
      "loss": 0.0172,
      "step": 4547
    },
    {
      "epoch": 40.61,
      "grad_norm": 0.05991187319159508,
      "learning_rate": 6.199999999999999e-05,
      "loss": 0.0175,
      "step": 4548
    },
    {
      "epoch": 40.62,
      "grad_norm": 0.07770812511444092,
      "learning_rate": 6.194117647058823e-05,
      "loss": 0.0192,
      "step": 4549
    },
    {
      "epoch": 40.62,
      "grad_norm": 0.0437345877289772,
      "learning_rate": 6.188235294117646e-05,
      "loss": 0.017,
      "step": 4550
    },
    {
      "epoch": 40.63,
      "grad_norm": 0.04400186240673065,
      "learning_rate": 6.18235294117647e-05,
      "loss": 0.0189,
      "step": 4551
    },
    {
      "epoch": 40.64,
      "grad_norm": 0.04059171676635742,
      "learning_rate": 6.176470588235294e-05,
      "loss": 0.0177,
      "step": 4552
    },
    {
      "epoch": 40.65,
      "grad_norm": 0.04959942772984505,
      "learning_rate": 6.170588235294117e-05,
      "loss": 0.0204,
      "step": 4553
    },
    {
      "epoch": 40.66,
      "grad_norm": 0.04438275843858719,
      "learning_rate": 6.164705882352941e-05,
      "loss": 0.0168,
      "step": 4554
    },
    {
      "epoch": 40.67,
      "grad_norm": 0.09111711382865906,
      "learning_rate": 6.158823529411764e-05,
      "loss": 0.0272,
      "step": 4555
    },
    {
      "epoch": 40.68,
      "grad_norm": 0.03727611526846886,
      "learning_rate": 6.152941176470588e-05,
      "loss": 0.0143,
      "step": 4556
    },
    {
      "epoch": 40.69,
      "grad_norm": 0.04391516000032425,
      "learning_rate": 6.147058823529411e-05,
      "loss": 0.0166,
      "step": 4557
    },
    {
      "epoch": 40.7,
      "grad_norm": 0.040411531925201416,
      "learning_rate": 6.141176470588236e-05,
      "loss": 0.0186,
      "step": 4558
    },
    {
      "epoch": 40.71,
      "grad_norm": 0.042106371372938156,
      "learning_rate": 6.135294117647059e-05,
      "loss": 0.0184,
      "step": 4559
    },
    {
      "epoch": 40.71,
      "grad_norm": 0.051107071340084076,
      "learning_rate": 6.129411764705882e-05,
      "loss": 0.0194,
      "step": 4560
    },
    {
      "epoch": 40.72,
      "grad_norm": 0.07036192715167999,
      "learning_rate": 6.123529411764706e-05,
      "loss": 0.0188,
      "step": 4561
    },
    {
      "epoch": 40.73,
      "grad_norm": 0.07041957974433899,
      "learning_rate": 6.117647058823529e-05,
      "loss": 0.0173,
      "step": 4562
    },
    {
      "epoch": 40.74,
      "grad_norm": 0.04342490807175636,
      "learning_rate": 6.111764705882353e-05,
      "loss": 0.0176,
      "step": 4563
    },
    {
      "epoch": 40.75,
      "grad_norm": 0.04865897819399834,
      "learning_rate": 6.105882352941176e-05,
      "loss": 0.0189,
      "step": 4564
    },
    {
      "epoch": 40.76,
      "grad_norm": 0.04451196268200874,
      "learning_rate": 6.1e-05,
      "loss": 0.0192,
      "step": 4565
    },
    {
      "epoch": 40.77,
      "grad_norm": 0.045886747539043427,
      "learning_rate": 6.094117647058823e-05,
      "loss": 0.0163,
      "step": 4566
    },
    {
      "epoch": 40.78,
      "grad_norm": 0.060928717255592346,
      "learning_rate": 6.0882352941176465e-05,
      "loss": 0.0176,
      "step": 4567
    },
    {
      "epoch": 40.79,
      "grad_norm": 0.07810866087675095,
      "learning_rate": 6.08235294117647e-05,
      "loss": 0.0216,
      "step": 4568
    },
    {
      "epoch": 40.79,
      "grad_norm": 0.03726855292916298,
      "learning_rate": 6.076470588235293e-05,
      "loss": 0.0154,
      "step": 4569
    },
    {
      "epoch": 40.8,
      "grad_norm": 0.038348231464624405,
      "learning_rate": 6.0705882352941175e-05,
      "loss": 0.0153,
      "step": 4570
    },
    {
      "epoch": 40.81,
      "grad_norm": 0.044328585267066956,
      "learning_rate": 6.0647058823529405e-05,
      "loss": 0.0191,
      "step": 4571
    },
    {
      "epoch": 40.82,
      "grad_norm": 0.043415606021881104,
      "learning_rate": 6.058823529411764e-05,
      "loss": 0.0159,
      "step": 4572
    },
    {
      "epoch": 40.83,
      "grad_norm": 0.047394152730703354,
      "learning_rate": 6.052941176470587e-05,
      "loss": 0.0154,
      "step": 4573
    },
    {
      "epoch": 40.84,
      "grad_norm": 0.06495626270771027,
      "learning_rate": 6.0470588235294115e-05,
      "loss": 0.0192,
      "step": 4574
    },
    {
      "epoch": 40.85,
      "grad_norm": 0.03826696053147316,
      "learning_rate": 6.041176470588235e-05,
      "loss": 0.0171,
      "step": 4575
    },
    {
      "epoch": 40.86,
      "grad_norm": 0.04659932851791382,
      "learning_rate": 6.035294117647058e-05,
      "loss": 0.0205,
      "step": 4576
    },
    {
      "epoch": 40.87,
      "grad_norm": 0.04046553745865822,
      "learning_rate": 6.029411764705882e-05,
      "loss": 0.0156,
      "step": 4577
    },
    {
      "epoch": 40.88,
      "grad_norm": 0.05137075111269951,
      "learning_rate": 6.023529411764705e-05,
      "loss": 0.0165,
      "step": 4578
    },
    {
      "epoch": 40.88,
      "grad_norm": 0.04387735575437546,
      "learning_rate": 6.017647058823529e-05,
      "loss": 0.0146,
      "step": 4579
    },
    {
      "epoch": 40.89,
      "grad_norm": 0.06885524094104767,
      "learning_rate": 6.011764705882352e-05,
      "loss": 0.0142,
      "step": 4580
    },
    {
      "epoch": 40.9,
      "grad_norm": 0.034240756183862686,
      "learning_rate": 6.005882352941176e-05,
      "loss": 0.0133,
      "step": 4581
    },
    {
      "epoch": 40.91,
      "grad_norm": 0.04060880467295647,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 0.0158,
      "step": 4582
    },
    {
      "epoch": 40.92,
      "grad_norm": 0.044692594558000565,
      "learning_rate": 5.994117647058823e-05,
      "loss": 0.0177,
      "step": 4583
    },
    {
      "epoch": 40.93,
      "grad_norm": 0.04203140363097191,
      "learning_rate": 5.988235294117647e-05,
      "loss": 0.0162,
      "step": 4584
    },
    {
      "epoch": 40.94,
      "grad_norm": 0.05605940520763397,
      "learning_rate": 5.98235294117647e-05,
      "loss": 0.0184,
      "step": 4585
    },
    {
      "epoch": 40.95,
      "grad_norm": 0.07269809395074844,
      "learning_rate": 5.9764705882352935e-05,
      "loss": 0.0237,
      "step": 4586
    },
    {
      "epoch": 40.96,
      "grad_norm": 0.05527862161397934,
      "learning_rate": 5.9705882352941164e-05,
      "loss": 0.0173,
      "step": 4587
    },
    {
      "epoch": 40.96,
      "grad_norm": 0.046791303902864456,
      "learning_rate": 5.964705882352941e-05,
      "loss": 0.0165,
      "step": 4588
    },
    {
      "epoch": 40.97,
      "grad_norm": 0.040417153388261795,
      "learning_rate": 5.9588235294117645e-05,
      "loss": 0.0167,
      "step": 4589
    },
    {
      "epoch": 40.98,
      "grad_norm": 0.04512058570981026,
      "learning_rate": 5.9529411764705874e-05,
      "loss": 0.0165,
      "step": 4590
    },
    {
      "epoch": 40.99,
      "grad_norm": 0.05198723450303078,
      "learning_rate": 5.947058823529411e-05,
      "loss": 0.0179,
      "step": 4591
    },
    {
      "epoch": 41.0,
      "grad_norm": 0.060485243797302246,
      "learning_rate": 5.941176470588235e-05,
      "loss": 0.0163,
      "step": 4592
    },
    {
      "epoch": 41.01,
      "grad_norm": 0.04254607483744621,
      "learning_rate": 5.9352941176470584e-05,
      "loss": 0.0164,
      "step": 4593
    },
    {
      "epoch": 41.02,
      "grad_norm": 0.04230436310172081,
      "learning_rate": 5.9294117647058814e-05,
      "loss": 0.0186,
      "step": 4594
    },
    {
      "epoch": 41.03,
      "grad_norm": 0.04634520038962364,
      "learning_rate": 5.923529411764705e-05,
      "loss": 0.0172,
      "step": 4595
    },
    {
      "epoch": 41.04,
      "grad_norm": 0.04482303559780121,
      "learning_rate": 5.9176470588235294e-05,
      "loss": 0.0193,
      "step": 4596
    },
    {
      "epoch": 41.04,
      "grad_norm": 0.04185933992266655,
      "learning_rate": 5.9117647058823524e-05,
      "loss": 0.0156,
      "step": 4597
    },
    {
      "epoch": 41.05,
      "grad_norm": 0.058788761496543884,
      "learning_rate": 5.905882352941176e-05,
      "loss": 0.0199,
      "step": 4598
    },
    {
      "epoch": 41.06,
      "grad_norm": 0.08305046707391739,
      "learning_rate": 5.899999999999999e-05,
      "loss": 0.0215,
      "step": 4599
    },
    {
      "epoch": 41.07,
      "grad_norm": 0.045421600341796875,
      "learning_rate": 5.894117647058823e-05,
      "loss": 0.0185,
      "step": 4600
    },
    {
      "epoch": 41.07,
      "eval_cer": 0.030345055680208063,
      "eval_loss": 0.20500260591506958,
      "eval_runtime": 22.2766,
      "eval_samples_per_second": 118.6,
      "eval_steps_per_second": 1.885,
      "eval_wer": 0.09938516461721539,
      "step": 4600
    },
    {
      "epoch": 41.08,
      "grad_norm": 0.038410548120737076,
      "learning_rate": 5.888235294117647e-05,
      "loss": 0.0172,
      "step": 4601
    },
    {
      "epoch": 41.09,
      "grad_norm": 0.038646068423986435,
      "learning_rate": 5.88235294117647e-05,
      "loss": 0.016,
      "step": 4602
    },
    {
      "epoch": 41.1,
      "grad_norm": 0.04110243543982506,
      "learning_rate": 5.876470588235294e-05,
      "loss": 0.0152,
      "step": 4603
    },
    {
      "epoch": 41.11,
      "grad_norm": 0.052151646465063095,
      "learning_rate": 5.870588235294117e-05,
      "loss": 0.0159,
      "step": 4604
    },
    {
      "epoch": 41.12,
      "grad_norm": 0.06517789512872696,
      "learning_rate": 5.864705882352941e-05,
      "loss": 0.0204,
      "step": 4605
    },
    {
      "epoch": 41.12,
      "grad_norm": 0.0377584770321846,
      "learning_rate": 5.858823529411764e-05,
      "loss": 0.0155,
      "step": 4606
    },
    {
      "epoch": 41.13,
      "grad_norm": 0.03711734712123871,
      "learning_rate": 5.852941176470588e-05,
      "loss": 0.0177,
      "step": 4607
    },
    {
      "epoch": 41.14,
      "grad_norm": 0.04377947002649307,
      "learning_rate": 5.8470588235294114e-05,
      "loss": 0.0149,
      "step": 4608
    },
    {
      "epoch": 41.15,
      "grad_norm": 0.04752575606107712,
      "learning_rate": 5.8411764705882344e-05,
      "loss": 0.0178,
      "step": 4609
    },
    {
      "epoch": 41.16,
      "grad_norm": 0.04799149930477142,
      "learning_rate": 5.835294117647059e-05,
      "loss": 0.0173,
      "step": 4610
    },
    {
      "epoch": 41.17,
      "grad_norm": 0.1113613173365593,
      "learning_rate": 5.829411764705882e-05,
      "loss": 0.0243,
      "step": 4611
    },
    {
      "epoch": 41.18,
      "grad_norm": 0.04745407775044441,
      "learning_rate": 5.8235294117647054e-05,
      "loss": 0.0199,
      "step": 4612
    },
    {
      "epoch": 41.19,
      "grad_norm": 0.03987278416752815,
      "learning_rate": 5.8176470588235284e-05,
      "loss": 0.0171,
      "step": 4613
    },
    {
      "epoch": 41.2,
      "grad_norm": 0.04175101965665817,
      "learning_rate": 5.811764705882353e-05,
      "loss": 0.0176,
      "step": 4614
    },
    {
      "epoch": 41.21,
      "grad_norm": 0.04692227765917778,
      "learning_rate": 5.8058823529411764e-05,
      "loss": 0.0168,
      "step": 4615
    },
    {
      "epoch": 41.21,
      "grad_norm": 0.04507969692349434,
      "learning_rate": 5.7999999999999994e-05,
      "loss": 0.0167,
      "step": 4616
    },
    {
      "epoch": 41.22,
      "grad_norm": 0.08686289191246033,
      "learning_rate": 5.794117647058823e-05,
      "loss": 0.0244,
      "step": 4617
    },
    {
      "epoch": 41.23,
      "grad_norm": 0.03982887789607048,
      "learning_rate": 5.788235294117646e-05,
      "loss": 0.0179,
      "step": 4618
    },
    {
      "epoch": 41.24,
      "grad_norm": 0.04830826818943024,
      "learning_rate": 5.7823529411764704e-05,
      "loss": 0.0195,
      "step": 4619
    },
    {
      "epoch": 41.25,
      "grad_norm": 0.0378747284412384,
      "learning_rate": 5.7764705882352933e-05,
      "loss": 0.0152,
      "step": 4620
    },
    {
      "epoch": 41.26,
      "grad_norm": 0.03851475194096565,
      "learning_rate": 5.770588235294117e-05,
      "loss": 0.0168,
      "step": 4621
    },
    {
      "epoch": 41.27,
      "grad_norm": 0.04425816982984543,
      "learning_rate": 5.764705882352941e-05,
      "loss": 0.0167,
      "step": 4622
    },
    {
      "epoch": 41.28,
      "grad_norm": 0.06909024715423584,
      "learning_rate": 5.758823529411764e-05,
      "loss": 0.0153,
      "step": 4623
    },
    {
      "epoch": 41.29,
      "grad_norm": 0.0706043615937233,
      "learning_rate": 5.752941176470588e-05,
      "loss": 0.0238,
      "step": 4624
    },
    {
      "epoch": 41.29,
      "grad_norm": 0.0388813391327858,
      "learning_rate": 5.747058823529411e-05,
      "loss": 0.0165,
      "step": 4625
    },
    {
      "epoch": 41.3,
      "grad_norm": 0.03613726422190666,
      "learning_rate": 5.7411764705882347e-05,
      "loss": 0.0146,
      "step": 4626
    },
    {
      "epoch": 41.31,
      "grad_norm": 0.044658176600933075,
      "learning_rate": 5.7352941176470576e-05,
      "loss": 0.0158,
      "step": 4627
    },
    {
      "epoch": 41.32,
      "grad_norm": 0.04512956365942955,
      "learning_rate": 5.729411764705882e-05,
      "loss": 0.0182,
      "step": 4628
    },
    {
      "epoch": 41.33,
      "grad_norm": 0.06834746897220612,
      "learning_rate": 5.7235294117647057e-05,
      "loss": 0.0176,
      "step": 4629
    },
    {
      "epoch": 41.34,
      "grad_norm": 0.09034556895494461,
      "learning_rate": 5.7176470588235286e-05,
      "loss": 0.0212,
      "step": 4630
    },
    {
      "epoch": 41.35,
      "grad_norm": 0.0401027612388134,
      "learning_rate": 5.711764705882352e-05,
      "loss": 0.0152,
      "step": 4631
    },
    {
      "epoch": 41.36,
      "grad_norm": 0.04389679804444313,
      "learning_rate": 5.705882352941176e-05,
      "loss": 0.0165,
      "step": 4632
    },
    {
      "epoch": 41.37,
      "grad_norm": 0.04307018220424652,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 0.0157,
      "step": 4633
    },
    {
      "epoch": 41.38,
      "grad_norm": 0.041806433349847794,
      "learning_rate": 5.6941176470588226e-05,
      "loss": 0.0142,
      "step": 4634
    },
    {
      "epoch": 41.38,
      "grad_norm": 0.05418125540018082,
      "learning_rate": 5.688235294117646e-05,
      "loss": 0.0184,
      "step": 4635
    },
    {
      "epoch": 41.39,
      "grad_norm": 0.07835173606872559,
      "learning_rate": 5.6823529411764706e-05,
      "loss": 0.0188,
      "step": 4636
    },
    {
      "epoch": 41.4,
      "grad_norm": 0.034479010850191116,
      "learning_rate": 5.6764705882352936e-05,
      "loss": 0.014,
      "step": 4637
    },
    {
      "epoch": 41.41,
      "grad_norm": 0.04069066047668457,
      "learning_rate": 5.670588235294117e-05,
      "loss": 0.0154,
      "step": 4638
    },
    {
      "epoch": 41.42,
      "grad_norm": 0.041832029819488525,
      "learning_rate": 5.66470588235294e-05,
      "loss": 0.0197,
      "step": 4639
    },
    {
      "epoch": 41.43,
      "grad_norm": 0.04355741664767265,
      "learning_rate": 5.658823529411764e-05,
      "loss": 0.0151,
      "step": 4640
    },
    {
      "epoch": 41.44,
      "grad_norm": 0.049824878573417664,
      "learning_rate": 5.652941176470588e-05,
      "loss": 0.0162,
      "step": 4641
    },
    {
      "epoch": 41.45,
      "grad_norm": 0.09460069239139557,
      "learning_rate": 5.647058823529411e-05,
      "loss": 0.0189,
      "step": 4642
    },
    {
      "epoch": 41.46,
      "grad_norm": 0.037275031208992004,
      "learning_rate": 5.641176470588235e-05,
      "loss": 0.0147,
      "step": 4643
    },
    {
      "epoch": 41.46,
      "grad_norm": 0.04388081654906273,
      "learning_rate": 5.635294117647058e-05,
      "loss": 0.0167,
      "step": 4644
    },
    {
      "epoch": 41.47,
      "grad_norm": 0.03630803897976875,
      "learning_rate": 5.629411764705882e-05,
      "loss": 0.0149,
      "step": 4645
    },
    {
      "epoch": 41.48,
      "grad_norm": 0.034449201077222824,
      "learning_rate": 5.623529411764705e-05,
      "loss": 0.0132,
      "step": 4646
    },
    {
      "epoch": 41.49,
      "grad_norm": 0.046683307737112045,
      "learning_rate": 5.617647058823529e-05,
      "loss": 0.0167,
      "step": 4647
    },
    {
      "epoch": 41.5,
      "grad_norm": 0.050131574273109436,
      "learning_rate": 5.6117647058823526e-05,
      "loss": 0.0146,
      "step": 4648
    },
    {
      "epoch": 41.51,
      "grad_norm": 0.0743521898984909,
      "learning_rate": 5.6058823529411756e-05,
      "loss": 0.0172,
      "step": 4649
    },
    {
      "epoch": 41.52,
      "grad_norm": 0.04171568527817726,
      "learning_rate": 5.6e-05,
      "loss": 0.0157,
      "step": 4650
    },
    {
      "epoch": 41.53,
      "grad_norm": 0.043973859399557114,
      "learning_rate": 5.594117647058823e-05,
      "loss": 0.0181,
      "step": 4651
    },
    {
      "epoch": 41.54,
      "grad_norm": 0.039228856563568115,
      "learning_rate": 5.5882352941176466e-05,
      "loss": 0.0133,
      "step": 4652
    },
    {
      "epoch": 41.54,
      "grad_norm": 0.045872762799263,
      "learning_rate": 5.5823529411764696e-05,
      "loss": 0.0153,
      "step": 4653
    },
    {
      "epoch": 41.55,
      "grad_norm": 0.06259597837924957,
      "learning_rate": 5.576470588235294e-05,
      "loss": 0.0204,
      "step": 4654
    },
    {
      "epoch": 41.56,
      "grad_norm": 0.0728481337428093,
      "learning_rate": 5.5705882352941176e-05,
      "loss": 0.022,
      "step": 4655
    },
    {
      "epoch": 41.57,
      "grad_norm": 0.039760757237672806,
      "learning_rate": 5.5647058823529406e-05,
      "loss": 0.0164,
      "step": 4656
    },
    {
      "epoch": 41.58,
      "grad_norm": 0.03884727507829666,
      "learning_rate": 5.558823529411764e-05,
      "loss": 0.0154,
      "step": 4657
    },
    {
      "epoch": 41.59,
      "grad_norm": 0.03832199051976204,
      "learning_rate": 5.552941176470587e-05,
      "loss": 0.0162,
      "step": 4658
    },
    {
      "epoch": 41.6,
      "grad_norm": 0.04435676708817482,
      "learning_rate": 5.5470588235294115e-05,
      "loss": 0.0166,
      "step": 4659
    },
    {
      "epoch": 41.61,
      "grad_norm": 0.05417013168334961,
      "learning_rate": 5.5411764705882345e-05,
      "loss": 0.0163,
      "step": 4660
    },
    {
      "epoch": 41.62,
      "grad_norm": 0.07437200099229813,
      "learning_rate": 5.535294117647058e-05,
      "loss": 0.0224,
      "step": 4661
    },
    {
      "epoch": 41.62,
      "grad_norm": 0.0526440367102623,
      "learning_rate": 5.5294117647058825e-05,
      "loss": 0.0207,
      "step": 4662
    },
    {
      "epoch": 41.63,
      "grad_norm": 0.040949463844299316,
      "learning_rate": 5.5235294117647055e-05,
      "loss": 0.0166,
      "step": 4663
    },
    {
      "epoch": 41.64,
      "grad_norm": 0.04179446026682854,
      "learning_rate": 5.517647058823529e-05,
      "loss": 0.0157,
      "step": 4664
    },
    {
      "epoch": 41.65,
      "grad_norm": 0.04304821044206619,
      "learning_rate": 5.511764705882352e-05,
      "loss": 0.0174,
      "step": 4665
    },
    {
      "epoch": 41.66,
      "grad_norm": 0.05936549976468086,
      "learning_rate": 5.505882352941176e-05,
      "loss": 0.0179,
      "step": 4666
    },
    {
      "epoch": 41.67,
      "grad_norm": 0.08342947065830231,
      "learning_rate": 5.499999999999999e-05,
      "loss": 0.0192,
      "step": 4667
    },
    {
      "epoch": 41.68,
      "grad_norm": 0.04317031800746918,
      "learning_rate": 5.494117647058823e-05,
      "loss": 0.0168,
      "step": 4668
    },
    {
      "epoch": 41.69,
      "grad_norm": 0.045395709574222565,
      "learning_rate": 5.488235294117647e-05,
      "loss": 0.0166,
      "step": 4669
    },
    {
      "epoch": 41.7,
      "grad_norm": 0.0382905974984169,
      "learning_rate": 5.48235294117647e-05,
      "loss": 0.0167,
      "step": 4670
    },
    {
      "epoch": 41.71,
      "grad_norm": 0.04265590384602547,
      "learning_rate": 5.4764705882352935e-05,
      "loss": 0.0167,
      "step": 4671
    },
    {
      "epoch": 41.71,
      "grad_norm": 0.04800283536314964,
      "learning_rate": 5.470588235294117e-05,
      "loss": 0.0176,
      "step": 4672
    },
    {
      "epoch": 41.72,
      "grad_norm": 0.08824150264263153,
      "learning_rate": 5.464705882352941e-05,
      "loss": 0.0185,
      "step": 4673
    },
    {
      "epoch": 41.73,
      "grad_norm": 0.04878758266568184,
      "learning_rate": 5.458823529411764e-05,
      "loss": 0.0187,
      "step": 4674
    },
    {
      "epoch": 41.74,
      "grad_norm": 0.0359526127576828,
      "learning_rate": 5.4529411764705875e-05,
      "loss": 0.0134,
      "step": 4675
    },
    {
      "epoch": 41.75,
      "grad_norm": 0.039270978420972824,
      "learning_rate": 5.447058823529412e-05,
      "loss": 0.0139,
      "step": 4676
    },
    {
      "epoch": 41.76,
      "grad_norm": 0.043132711201906204,
      "learning_rate": 5.441176470588235e-05,
      "loss": 0.0179,
      "step": 4677
    },
    {
      "epoch": 41.77,
      "grad_norm": 0.04192988947033882,
      "learning_rate": 5.4352941176470585e-05,
      "loss": 0.017,
      "step": 4678
    },
    {
      "epoch": 41.78,
      "grad_norm": 0.05601268634200096,
      "learning_rate": 5.4294117647058815e-05,
      "loss": 0.0139,
      "step": 4679
    },
    {
      "epoch": 41.79,
      "grad_norm": 0.11920702457427979,
      "learning_rate": 5.423529411764705e-05,
      "loss": 0.0288,
      "step": 4680
    },
    {
      "epoch": 41.79,
      "grad_norm": 0.0415286086499691,
      "learning_rate": 5.4176470588235295e-05,
      "loss": 0.0161,
      "step": 4681
    },
    {
      "epoch": 41.8,
      "grad_norm": 0.043536920100450516,
      "learning_rate": 5.4117647058823525e-05,
      "loss": 0.0141,
      "step": 4682
    },
    {
      "epoch": 41.81,
      "grad_norm": 0.04488612338900566,
      "learning_rate": 5.405882352941176e-05,
      "loss": 0.0142,
      "step": 4683
    },
    {
      "epoch": 41.82,
      "grad_norm": 0.05310561880469322,
      "learning_rate": 5.399999999999999e-05,
      "loss": 0.0129,
      "step": 4684
    },
    {
      "epoch": 41.83,
      "grad_norm": 0.05201854184269905,
      "learning_rate": 5.3941176470588235e-05,
      "loss": 0.018,
      "step": 4685
    },
    {
      "epoch": 41.84,
      "grad_norm": 0.09678861498832703,
      "learning_rate": 5.3882352941176465e-05,
      "loss": 0.0227,
      "step": 4686
    },
    {
      "epoch": 41.85,
      "grad_norm": 0.04128541052341461,
      "learning_rate": 5.38235294117647e-05,
      "loss": 0.0182,
      "step": 4687
    },
    {
      "epoch": 41.86,
      "grad_norm": 0.039448902010917664,
      "learning_rate": 5.376470588235294e-05,
      "loss": 0.0148,
      "step": 4688
    },
    {
      "epoch": 41.87,
      "grad_norm": 0.04196583852171898,
      "learning_rate": 5.370588235294117e-05,
      "loss": 0.0159,
      "step": 4689
    },
    {
      "epoch": 41.88,
      "grad_norm": 0.04726090282201767,
      "learning_rate": 5.364705882352941e-05,
      "loss": 0.0152,
      "step": 4690
    },
    {
      "epoch": 41.88,
      "grad_norm": 0.04414398968219757,
      "learning_rate": 5.358823529411764e-05,
      "loss": 0.0143,
      "step": 4691
    },
    {
      "epoch": 41.89,
      "grad_norm": 0.09072451293468475,
      "learning_rate": 5.352941176470588e-05,
      "loss": 0.0204,
      "step": 4692
    },
    {
      "epoch": 41.9,
      "grad_norm": 0.0364651121199131,
      "learning_rate": 5.347058823529411e-05,
      "loss": 0.0171,
      "step": 4693
    },
    {
      "epoch": 41.91,
      "grad_norm": 0.03598048537969589,
      "learning_rate": 5.341176470588235e-05,
      "loss": 0.016,
      "step": 4694
    },
    {
      "epoch": 41.92,
      "grad_norm": 0.039510637521743774,
      "learning_rate": 5.335294117647059e-05,
      "loss": 0.0166,
      "step": 4695
    },
    {
      "epoch": 41.93,
      "grad_norm": 0.04128700867295265,
      "learning_rate": 5.329411764705882e-05,
      "loss": 0.0155,
      "step": 4696
    },
    {
      "epoch": 41.94,
      "grad_norm": 0.04875123128294945,
      "learning_rate": 5.3235294117647054e-05,
      "loss": 0.016,
      "step": 4697
    },
    {
      "epoch": 41.95,
      "grad_norm": 0.058702390640974045,
      "learning_rate": 5.3176470588235284e-05,
      "loss": 0.0152,
      "step": 4698
    },
    {
      "epoch": 41.96,
      "grad_norm": 0.07510247826576233,
      "learning_rate": 5.311764705882353e-05,
      "loss": 0.0222,
      "step": 4699
    },
    {
      "epoch": 41.96,
      "grad_norm": 0.04359116405248642,
      "learning_rate": 5.305882352941176e-05,
      "loss": 0.0171,
      "step": 4700
    },
    {
      "epoch": 41.96,
      "eval_cer": 0.030570557704088726,
      "eval_loss": 0.21361805498600006,
      "eval_runtime": 22.1312,
      "eval_samples_per_second": 119.379,
      "eval_steps_per_second": 1.898,
      "eval_wer": 0.09936533121777072,
      "step": 4700
    },
    {
      "epoch": 41.97,
      "grad_norm": 0.04909026250243187,
      "learning_rate": 5.2999999999999994e-05,
      "loss": 0.0162,
      "step": 4701
    },
    {
      "epoch": 41.98,
      "grad_norm": 0.05668247863650322,
      "learning_rate": 5.294117647058824e-05,
      "loss": 0.0208,
      "step": 4702
    },
    {
      "epoch": 41.99,
      "grad_norm": 0.06715834885835648,
      "learning_rate": 5.288235294117647e-05,
      "loss": 0.0181,
      "step": 4703
    },
    {
      "epoch": 42.0,
      "grad_norm": 0.06368158757686615,
      "learning_rate": 5.2823529411764704e-05,
      "loss": 0.0174,
      "step": 4704
    },
    {
      "epoch": 42.01,
      "grad_norm": 0.04115816578269005,
      "learning_rate": 5.2764705882352934e-05,
      "loss": 0.0174,
      "step": 4705
    },
    {
      "epoch": 42.02,
      "grad_norm": 0.04423500970005989,
      "learning_rate": 5.270588235294117e-05,
      "loss": 0.0163,
      "step": 4706
    },
    {
      "epoch": 42.03,
      "grad_norm": 0.038058824837207794,
      "learning_rate": 5.26470588235294e-05,
      "loss": 0.0152,
      "step": 4707
    },
    {
      "epoch": 42.04,
      "grad_norm": 0.04965026304125786,
      "learning_rate": 5.2588235294117644e-05,
      "loss": 0.0188,
      "step": 4708
    },
    {
      "epoch": 42.04,
      "grad_norm": 0.05564122274518013,
      "learning_rate": 5.252941176470588e-05,
      "loss": 0.0169,
      "step": 4709
    },
    {
      "epoch": 42.05,
      "grad_norm": 0.0768323689699173,
      "learning_rate": 5.247058823529411e-05,
      "loss": 0.0192,
      "step": 4710
    },
    {
      "epoch": 42.06,
      "grad_norm": 0.048551928251981735,
      "learning_rate": 5.241176470588235e-05,
      "loss": 0.0168,
      "step": 4711
    },
    {
      "epoch": 42.07,
      "grad_norm": 0.032121993601322174,
      "learning_rate": 5.2352941176470584e-05,
      "loss": 0.0131,
      "step": 4712
    },
    {
      "epoch": 42.08,
      "grad_norm": 0.03721311315894127,
      "learning_rate": 5.229411764705882e-05,
      "loss": 0.0173,
      "step": 4713
    },
    {
      "epoch": 42.09,
      "grad_norm": 0.041603635996580124,
      "learning_rate": 5.223529411764705e-05,
      "loss": 0.0176,
      "step": 4714
    },
    {
      "epoch": 42.1,
      "grad_norm": 0.04141776263713837,
      "learning_rate": 5.217647058823529e-05,
      "loss": 0.0174,
      "step": 4715
    },
    {
      "epoch": 42.11,
      "grad_norm": 0.04728950187563896,
      "learning_rate": 5.211764705882353e-05,
      "loss": 0.0156,
      "step": 4716
    },
    {
      "epoch": 42.12,
      "grad_norm": 0.09710820019245148,
      "learning_rate": 5.205882352941176e-05,
      "loss": 0.0243,
      "step": 4717
    },
    {
      "epoch": 42.12,
      "grad_norm": 0.04254890978336334,
      "learning_rate": 5.2e-05,
      "loss": 0.0172,
      "step": 4718
    },
    {
      "epoch": 42.13,
      "grad_norm": 0.04165995493531227,
      "learning_rate": 5.194117647058823e-05,
      "loss": 0.0182,
      "step": 4719
    },
    {
      "epoch": 42.14,
      "grad_norm": 0.03956056386232376,
      "learning_rate": 5.188235294117646e-05,
      "loss": 0.0165,
      "step": 4720
    },
    {
      "epoch": 42.15,
      "grad_norm": 0.041155051440000534,
      "learning_rate": 5.182352941176471e-05,
      "loss": 0.0187,
      "step": 4721
    },
    {
      "epoch": 42.16,
      "grad_norm": 0.06585805863142014,
      "learning_rate": 5.176470588235294e-05,
      "loss": 0.0187,
      "step": 4722
    },
    {
      "epoch": 42.17,
      "grad_norm": 0.07872984558343887,
      "learning_rate": 5.170588235294117e-05,
      "loss": 0.0183,
      "step": 4723
    },
    {
      "epoch": 42.18,
      "grad_norm": 0.03979578614234924,
      "learning_rate": 5.16470588235294e-05,
      "loss": 0.0163,
      "step": 4724
    },
    {
      "epoch": 42.19,
      "grad_norm": 0.04740254208445549,
      "learning_rate": 5.1588235294117647e-05,
      "loss": 0.0172,
      "step": 4725
    },
    {
      "epoch": 42.2,
      "grad_norm": 0.03644009679555893,
      "learning_rate": 5.1529411764705876e-05,
      "loss": 0.0157,
      "step": 4726
    },
    {
      "epoch": 42.21,
      "grad_norm": 0.044104836881160736,
      "learning_rate": 5.147058823529411e-05,
      "loss": 0.0174,
      "step": 4727
    },
    {
      "epoch": 42.21,
      "grad_norm": 0.05358245223760605,
      "learning_rate": 5.141176470588235e-05,
      "loss": 0.0183,
      "step": 4728
    },
    {
      "epoch": 42.22,
      "grad_norm": 0.08560299128293991,
      "learning_rate": 5.135294117647058e-05,
      "loss": 0.0216,
      "step": 4729
    },
    {
      "epoch": 42.23,
      "grad_norm": 0.04160348325967789,
      "learning_rate": 5.129411764705882e-05,
      "loss": 0.016,
      "step": 4730
    },
    {
      "epoch": 42.24,
      "grad_norm": 0.04579866677522659,
      "learning_rate": 5.123529411764705e-05,
      "loss": 0.0182,
      "step": 4731
    },
    {
      "epoch": 42.25,
      "grad_norm": 0.05497223883867264,
      "learning_rate": 5.117647058823529e-05,
      "loss": 0.02,
      "step": 4732
    },
    {
      "epoch": 42.26,
      "grad_norm": 0.04410240799188614,
      "learning_rate": 5.111764705882352e-05,
      "loss": 0.0156,
      "step": 4733
    },
    {
      "epoch": 42.27,
      "grad_norm": 0.05195372551679611,
      "learning_rate": 5.105882352941176e-05,
      "loss": 0.0162,
      "step": 4734
    },
    {
      "epoch": 42.28,
      "grad_norm": 0.07460621744394302,
      "learning_rate": 5.1e-05,
      "loss": 0.0166,
      "step": 4735
    },
    {
      "epoch": 42.29,
      "grad_norm": 0.08454734832048416,
      "learning_rate": 5.094117647058823e-05,
      "loss": 0.0207,
      "step": 4736
    },
    {
      "epoch": 42.29,
      "grad_norm": 0.039156123995780945,
      "learning_rate": 5.0882352941176466e-05,
      "loss": 0.0172,
      "step": 4737
    },
    {
      "epoch": 42.3,
      "grad_norm": 0.049701571464538574,
      "learning_rate": 5.0823529411764696e-05,
      "loss": 0.019,
      "step": 4738
    },
    {
      "epoch": 42.31,
      "grad_norm": 0.04055621474981308,
      "learning_rate": 5.076470588235294e-05,
      "loss": 0.0166,
      "step": 4739
    },
    {
      "epoch": 42.32,
      "grad_norm": 0.04590931907296181,
      "learning_rate": 5.070588235294117e-05,
      "loss": 0.0172,
      "step": 4740
    },
    {
      "epoch": 42.33,
      "grad_norm": 0.06520258635282516,
      "learning_rate": 5.0647058823529406e-05,
      "loss": 0.0182,
      "step": 4741
    },
    {
      "epoch": 42.34,
      "grad_norm": 0.15398338437080383,
      "learning_rate": 5.058823529411765e-05,
      "loss": 0.0212,
      "step": 4742
    },
    {
      "epoch": 42.35,
      "grad_norm": 0.03619791939854622,
      "learning_rate": 5.052941176470588e-05,
      "loss": 0.0162,
      "step": 4743
    },
    {
      "epoch": 42.36,
      "grad_norm": 0.04270081967115402,
      "learning_rate": 5.0470588235294116e-05,
      "loss": 0.0184,
      "step": 4744
    },
    {
      "epoch": 42.37,
      "grad_norm": 0.03663381189107895,
      "learning_rate": 5.0411764705882346e-05,
      "loss": 0.0147,
      "step": 4745
    },
    {
      "epoch": 42.38,
      "grad_norm": 0.04784776642918587,
      "learning_rate": 5.035294117647058e-05,
      "loss": 0.0182,
      "step": 4746
    },
    {
      "epoch": 42.38,
      "grad_norm": 0.0638587549328804,
      "learning_rate": 5.029411764705881e-05,
      "loss": 0.0204,
      "step": 4747
    },
    {
      "epoch": 42.39,
      "grad_norm": 0.10195545107126236,
      "learning_rate": 5.0235294117647056e-05,
      "loss": 0.0248,
      "step": 4748
    },
    {
      "epoch": 42.4,
      "grad_norm": 0.035146091133356094,
      "learning_rate": 5.017647058823529e-05,
      "loss": 0.0155,
      "step": 4749
    },
    {
      "epoch": 42.41,
      "grad_norm": 0.04046854376792908,
      "learning_rate": 5.011764705882352e-05,
      "loss": 0.0182,
      "step": 4750
    },
    {
      "epoch": 42.42,
      "grad_norm": 0.03912757709622383,
      "learning_rate": 5.005882352941176e-05,
      "loss": 0.0129,
      "step": 4751
    },
    {
      "epoch": 42.43,
      "grad_norm": 0.04403378441929817,
      "learning_rate": 4.9999999999999996e-05,
      "loss": 0.0156,
      "step": 4752
    },
    {
      "epoch": 42.44,
      "grad_norm": 0.04925810173153877,
      "learning_rate": 4.994117647058823e-05,
      "loss": 0.0173,
      "step": 4753
    },
    {
      "epoch": 42.45,
      "grad_norm": 0.10861742496490479,
      "learning_rate": 4.988235294117646e-05,
      "loss": 0.0251,
      "step": 4754
    },
    {
      "epoch": 42.46,
      "grad_norm": 0.04123973101377487,
      "learning_rate": 4.98235294117647e-05,
      "loss": 0.0147,
      "step": 4755
    },
    {
      "epoch": 42.46,
      "grad_norm": 0.03818226233124733,
      "learning_rate": 4.976470588235294e-05,
      "loss": 0.0146,
      "step": 4756
    },
    {
      "epoch": 42.47,
      "grad_norm": 0.03866466134786606,
      "learning_rate": 4.970588235294117e-05,
      "loss": 0.0151,
      "step": 4757
    },
    {
      "epoch": 42.48,
      "grad_norm": 0.042279865592718124,
      "learning_rate": 4.964705882352941e-05,
      "loss": 0.0154,
      "step": 4758
    },
    {
      "epoch": 42.49,
      "grad_norm": 0.04880718141794205,
      "learning_rate": 4.958823529411764e-05,
      "loss": 0.0151,
      "step": 4759
    },
    {
      "epoch": 42.5,
      "grad_norm": 0.05486907437443733,
      "learning_rate": 4.9529411764705875e-05,
      "loss": 0.0171,
      "step": 4760
    },
    {
      "epoch": 42.51,
      "grad_norm": 0.06372758746147156,
      "learning_rate": 4.947058823529412e-05,
      "loss": 0.0196,
      "step": 4761
    },
    {
      "epoch": 42.52,
      "grad_norm": 0.04401450976729393,
      "learning_rate": 4.941176470588235e-05,
      "loss": 0.0183,
      "step": 4762
    },
    {
      "epoch": 42.53,
      "grad_norm": 0.042200129479169846,
      "learning_rate": 4.9352941176470585e-05,
      "loss": 0.0159,
      "step": 4763
    },
    {
      "epoch": 42.54,
      "grad_norm": 0.04643147811293602,
      "learning_rate": 4.9294117647058815e-05,
      "loss": 0.015,
      "step": 4764
    },
    {
      "epoch": 42.54,
      "grad_norm": 0.042301055043935776,
      "learning_rate": 4.923529411764706e-05,
      "loss": 0.0163,
      "step": 4765
    },
    {
      "epoch": 42.55,
      "grad_norm": 0.05764590576291084,
      "learning_rate": 4.917647058823529e-05,
      "loss": 0.0191,
      "step": 4766
    },
    {
      "epoch": 42.56,
      "grad_norm": 0.07639754563570023,
      "learning_rate": 4.9117647058823525e-05,
      "loss": 0.0232,
      "step": 4767
    },
    {
      "epoch": 42.57,
      "grad_norm": 0.050897639244794846,
      "learning_rate": 4.905882352941176e-05,
      "loss": 0.0154,
      "step": 4768
    },
    {
      "epoch": 42.58,
      "grad_norm": 0.04394242540001869,
      "learning_rate": 4.899999999999999e-05,
      "loss": 0.0177,
      "step": 4769
    },
    {
      "epoch": 42.59,
      "grad_norm": 0.03962036967277527,
      "learning_rate": 4.8941176470588235e-05,
      "loss": 0.0166,
      "step": 4770
    },
    {
      "epoch": 42.6,
      "grad_norm": 0.04944916069507599,
      "learning_rate": 4.8882352941176465e-05,
      "loss": 0.0206,
      "step": 4771
    },
    {
      "epoch": 42.61,
      "grad_norm": 0.04640146717429161,
      "learning_rate": 4.88235294117647e-05,
      "loss": 0.0132,
      "step": 4772
    },
    {
      "epoch": 42.62,
      "grad_norm": 0.08091548085212708,
      "learning_rate": 4.876470588235293e-05,
      "loss": 0.0202,
      "step": 4773
    },
    {
      "epoch": 42.62,
      "grad_norm": 0.04228280484676361,
      "learning_rate": 4.8705882352941175e-05,
      "loss": 0.0171,
      "step": 4774
    },
    {
      "epoch": 42.63,
      "grad_norm": 0.04076916724443436,
      "learning_rate": 4.864705882352941e-05,
      "loss": 0.016,
      "step": 4775
    },
    {
      "epoch": 42.64,
      "grad_norm": 0.047374263405799866,
      "learning_rate": 4.858823529411764e-05,
      "loss": 0.0169,
      "step": 4776
    },
    {
      "epoch": 42.65,
      "grad_norm": 0.047457072883844376,
      "learning_rate": 4.852941176470588e-05,
      "loss": 0.0149,
      "step": 4777
    },
    {
      "epoch": 42.66,
      "grad_norm": 0.04585172235965729,
      "learning_rate": 4.847058823529411e-05,
      "loss": 0.019,
      "step": 4778
    },
    {
      "epoch": 42.67,
      "grad_norm": 0.09318045526742935,
      "learning_rate": 4.841176470588235e-05,
      "loss": 0.0262,
      "step": 4779
    },
    {
      "epoch": 42.68,
      "grad_norm": 0.03627597540616989,
      "learning_rate": 4.835294117647058e-05,
      "loss": 0.0147,
      "step": 4780
    },
    {
      "epoch": 42.69,
      "grad_norm": 0.04001631587743759,
      "learning_rate": 4.829411764705882e-05,
      "loss": 0.0181,
      "step": 4781
    },
    {
      "epoch": 42.7,
      "grad_norm": 0.04767382889986038,
      "learning_rate": 4.823529411764706e-05,
      "loss": 0.0166,
      "step": 4782
    },
    {
      "epoch": 42.71,
      "grad_norm": 0.04053715243935585,
      "learning_rate": 4.817647058823529e-05,
      "loss": 0.0164,
      "step": 4783
    },
    {
      "epoch": 42.71,
      "grad_norm": 0.04655947536230087,
      "learning_rate": 4.811764705882353e-05,
      "loss": 0.0163,
      "step": 4784
    },
    {
      "epoch": 42.72,
      "grad_norm": 0.07540638744831085,
      "learning_rate": 4.805882352941176e-05,
      "loss": 0.0162,
      "step": 4785
    },
    {
      "epoch": 42.73,
      "grad_norm": 0.04849407076835632,
      "learning_rate": 4.7999999999999994e-05,
      "loss": 0.016,
      "step": 4786
    },
    {
      "epoch": 42.74,
      "grad_norm": 0.04162999987602234,
      "learning_rate": 4.7941176470588224e-05,
      "loss": 0.0163,
      "step": 4787
    },
    {
      "epoch": 42.75,
      "grad_norm": 0.04699215292930603,
      "learning_rate": 4.788235294117647e-05,
      "loss": 0.018,
      "step": 4788
    },
    {
      "epoch": 42.76,
      "grad_norm": 0.03915581852197647,
      "learning_rate": 4.7823529411764704e-05,
      "loss": 0.0142,
      "step": 4789
    },
    {
      "epoch": 42.77,
      "grad_norm": 0.051829997450113297,
      "learning_rate": 4.7764705882352934e-05,
      "loss": 0.0181,
      "step": 4790
    },
    {
      "epoch": 42.78,
      "grad_norm": 0.047189947217702866,
      "learning_rate": 4.770588235294118e-05,
      "loss": 0.0171,
      "step": 4791
    },
    {
      "epoch": 42.79,
      "grad_norm": 0.05920446664094925,
      "learning_rate": 4.764705882352941e-05,
      "loss": 0.0159,
      "step": 4792
    },
    {
      "epoch": 42.79,
      "grad_norm": 0.03865901008248329,
      "learning_rate": 4.7588235294117644e-05,
      "loss": 0.0179,
      "step": 4793
    },
    {
      "epoch": 42.8,
      "grad_norm": 0.043928008526563644,
      "learning_rate": 4.7529411764705874e-05,
      "loss": 0.0186,
      "step": 4794
    },
    {
      "epoch": 42.81,
      "grad_norm": 0.04029054567217827,
      "learning_rate": 4.747058823529411e-05,
      "loss": 0.0181,
      "step": 4795
    },
    {
      "epoch": 42.82,
      "grad_norm": 0.04039912298321724,
      "learning_rate": 4.7411764705882354e-05,
      "loss": 0.0148,
      "step": 4796
    },
    {
      "epoch": 42.83,
      "grad_norm": 0.05792859196662903,
      "learning_rate": 4.7352941176470584e-05,
      "loss": 0.0168,
      "step": 4797
    },
    {
      "epoch": 42.84,
      "grad_norm": 0.08704151958227158,
      "learning_rate": 4.729411764705882e-05,
      "loss": 0.0213,
      "step": 4798
    },
    {
      "epoch": 42.85,
      "grad_norm": 0.03789398819208145,
      "learning_rate": 4.723529411764705e-05,
      "loss": 0.0172,
      "step": 4799
    },
    {
      "epoch": 42.86,
      "grad_norm": 0.04100920259952545,
      "learning_rate": 4.717647058823529e-05,
      "loss": 0.0171,
      "step": 4800
    },
    {
      "epoch": 42.86,
      "eval_cer": 0.031784510265979636,
      "eval_loss": 0.20623566210269928,
      "eval_runtime": 22.326,
      "eval_samples_per_second": 118.337,
      "eval_steps_per_second": 1.881,
      "eval_wer": 0.1006743355811186,
      "step": 4800
    },
    {
      "epoch": 42.87,
      "grad_norm": 0.043528687208890915,
      "learning_rate": 4.711764705882353e-05,
      "loss": 0.0186,
      "step": 4801
    },
    {
      "epoch": 42.88,
      "grad_norm": 0.03754991665482521,
      "learning_rate": 4.705882352941176e-05,
      "loss": 0.0147,
      "step": 4802
    },
    {
      "epoch": 42.88,
      "grad_norm": 0.038673486560583115,
      "learning_rate": 4.7e-05,
      "loss": 0.0156,
      "step": 4803
    },
    {
      "epoch": 42.89,
      "grad_norm": 0.11040922999382019,
      "learning_rate": 4.694117647058823e-05,
      "loss": 0.027,
      "step": 4804
    },
    {
      "epoch": 42.9,
      "grad_norm": 0.03630939498543739,
      "learning_rate": 4.688235294117647e-05,
      "loss": 0.0167,
      "step": 4805
    },
    {
      "epoch": 42.91,
      "grad_norm": 0.042319733649492264,
      "learning_rate": 4.68235294117647e-05,
      "loss": 0.0184,
      "step": 4806
    },
    {
      "epoch": 42.92,
      "grad_norm": 0.03956311196088791,
      "learning_rate": 4.676470588235294e-05,
      "loss": 0.017,
      "step": 4807
    },
    {
      "epoch": 42.93,
      "grad_norm": 0.03828151151537895,
      "learning_rate": 4.6705882352941174e-05,
      "loss": 0.0143,
      "step": 4808
    },
    {
      "epoch": 42.94,
      "grad_norm": 0.048234179615974426,
      "learning_rate": 4.6647058823529404e-05,
      "loss": 0.0169,
      "step": 4809
    },
    {
      "epoch": 42.95,
      "grad_norm": 0.06010056287050247,
      "learning_rate": 4.658823529411765e-05,
      "loss": 0.0213,
      "step": 4810
    },
    {
      "epoch": 42.96,
      "grad_norm": 0.06197648495435715,
      "learning_rate": 4.652941176470588e-05,
      "loss": 0.018,
      "step": 4811
    },
    {
      "epoch": 42.96,
      "grad_norm": 0.04620685800909996,
      "learning_rate": 4.6470588235294114e-05,
      "loss": 0.0183,
      "step": 4812
    },
    {
      "epoch": 42.97,
      "grad_norm": 0.035234078764915466,
      "learning_rate": 4.6411764705882343e-05,
      "loss": 0.0158,
      "step": 4813
    },
    {
      "epoch": 42.98,
      "grad_norm": 0.04321025311946869,
      "learning_rate": 4.635294117647059e-05,
      "loss": 0.0147,
      "step": 4814
    },
    {
      "epoch": 42.99,
      "grad_norm": 0.04962904006242752,
      "learning_rate": 4.6294117647058824e-05,
      "loss": 0.0139,
      "step": 4815
    },
    {
      "epoch": 43.0,
      "grad_norm": 0.08016182482242584,
      "learning_rate": 4.6235294117647053e-05,
      "loss": 0.0199,
      "step": 4816
    },
    {
      "epoch": 43.01,
      "grad_norm": 0.03688027709722519,
      "learning_rate": 4.617647058823529e-05,
      "loss": 0.015,
      "step": 4817
    },
    {
      "epoch": 43.02,
      "grad_norm": 0.04421921446919441,
      "learning_rate": 4.611764705882352e-05,
      "loss": 0.0165,
      "step": 4818
    },
    {
      "epoch": 43.03,
      "grad_norm": 0.03546473756432533,
      "learning_rate": 4.605882352941176e-05,
      "loss": 0.0121,
      "step": 4819
    },
    {
      "epoch": 43.04,
      "grad_norm": 0.03673641011118889,
      "learning_rate": 4.599999999999999e-05,
      "loss": 0.0129,
      "step": 4820
    },
    {
      "epoch": 43.04,
      "grad_norm": 0.04233020171523094,
      "learning_rate": 4.594117647058823e-05,
      "loss": 0.0147,
      "step": 4821
    },
    {
      "epoch": 43.05,
      "grad_norm": 0.08312328159809113,
      "learning_rate": 4.588235294117647e-05,
      "loss": 0.0176,
      "step": 4822
    },
    {
      "epoch": 43.06,
      "grad_norm": 0.06797804683446884,
      "learning_rate": 4.58235294117647e-05,
      "loss": 0.0193,
      "step": 4823
    },
    {
      "epoch": 43.07,
      "grad_norm": 0.03830239549279213,
      "learning_rate": 4.576470588235294e-05,
      "loss": 0.0154,
      "step": 4824
    },
    {
      "epoch": 43.08,
      "grad_norm": 0.04687207564711571,
      "learning_rate": 4.570588235294117e-05,
      "loss": 0.0163,
      "step": 4825
    },
    {
      "epoch": 43.09,
      "grad_norm": 0.041979048401117325,
      "learning_rate": 4.5647058823529406e-05,
      "loss": 0.0149,
      "step": 4826
    },
    {
      "epoch": 43.1,
      "grad_norm": 0.04596380516886711,
      "learning_rate": 4.5588235294117636e-05,
      "loss": 0.017,
      "step": 4827
    },
    {
      "epoch": 43.11,
      "grad_norm": 0.06350573152303696,
      "learning_rate": 4.552941176470588e-05,
      "loss": 0.0195,
      "step": 4828
    },
    {
      "epoch": 43.12,
      "grad_norm": 0.06950381398200989,
      "learning_rate": 4.5470588235294116e-05,
      "loss": 0.0202,
      "step": 4829
    },
    {
      "epoch": 43.12,
      "grad_norm": 0.04189309477806091,
      "learning_rate": 4.5411764705882346e-05,
      "loss": 0.0155,
      "step": 4830
    },
    {
      "epoch": 43.13,
      "grad_norm": 0.039666492491960526,
      "learning_rate": 4.535294117647059e-05,
      "loss": 0.0162,
      "step": 4831
    },
    {
      "epoch": 43.14,
      "grad_norm": 0.03722996264696121,
      "learning_rate": 4.529411764705882e-05,
      "loss": 0.0145,
      "step": 4832
    },
    {
      "epoch": 43.15,
      "grad_norm": 0.04492158815264702,
      "learning_rate": 4.5235294117647056e-05,
      "loss": 0.0173,
      "step": 4833
    },
    {
      "epoch": 43.16,
      "grad_norm": 0.05497761815786362,
      "learning_rate": 4.5176470588235286e-05,
      "loss": 0.0156,
      "step": 4834
    },
    {
      "epoch": 43.17,
      "grad_norm": 0.10796255618333817,
      "learning_rate": 4.511764705882352e-05,
      "loss": 0.0221,
      "step": 4835
    },
    {
      "epoch": 43.18,
      "grad_norm": 0.046445827931165695,
      "learning_rate": 4.5058823529411766e-05,
      "loss": 0.0186,
      "step": 4836
    },
    {
      "epoch": 43.19,
      "grad_norm": 0.03872071951627731,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 0.0165,
      "step": 4837
    },
    {
      "epoch": 43.2,
      "grad_norm": 0.04532414674758911,
      "learning_rate": 4.494117647058823e-05,
      "loss": 0.016,
      "step": 4838
    },
    {
      "epoch": 43.21,
      "grad_norm": 0.04898528382182121,
      "learning_rate": 4.488235294117646e-05,
      "loss": 0.0188,
      "step": 4839
    },
    {
      "epoch": 43.21,
      "grad_norm": 0.0423404835164547,
      "learning_rate": 4.48235294117647e-05,
      "loss": 0.0133,
      "step": 4840
    },
    {
      "epoch": 43.22,
      "grad_norm": 0.09090810269117355,
      "learning_rate": 4.4764705882352936e-05,
      "loss": 0.0183,
      "step": 4841
    },
    {
      "epoch": 43.23,
      "grad_norm": 0.03748342767357826,
      "learning_rate": 4.470588235294117e-05,
      "loss": 0.0158,
      "step": 4842
    },
    {
      "epoch": 43.24,
      "grad_norm": 0.037210963666439056,
      "learning_rate": 4.464705882352941e-05,
      "loss": 0.0177,
      "step": 4843
    },
    {
      "epoch": 43.25,
      "grad_norm": 0.039508573710918427,
      "learning_rate": 4.458823529411764e-05,
      "loss": 0.0154,
      "step": 4844
    },
    {
      "epoch": 43.26,
      "grad_norm": 0.042392343282699585,
      "learning_rate": 4.452941176470588e-05,
      "loss": 0.0134,
      "step": 4845
    },
    {
      "epoch": 43.27,
      "grad_norm": 0.046576011925935745,
      "learning_rate": 4.447058823529411e-05,
      "loss": 0.0158,
      "step": 4846
    },
    {
      "epoch": 43.28,
      "grad_norm": 0.07849511504173279,
      "learning_rate": 4.441176470588235e-05,
      "loss": 0.0201,
      "step": 4847
    },
    {
      "epoch": 43.29,
      "grad_norm": 0.06206928938627243,
      "learning_rate": 4.4352941176470586e-05,
      "loss": 0.0216,
      "step": 4848
    },
    {
      "epoch": 43.29,
      "grad_norm": 0.03816213831305504,
      "learning_rate": 4.4294117647058816e-05,
      "loss": 0.0157,
      "step": 4849
    },
    {
      "epoch": 43.3,
      "grad_norm": 0.036603085696697235,
      "learning_rate": 4.423529411764706e-05,
      "loss": 0.0136,
      "step": 4850
    },
    {
      "epoch": 43.31,
      "grad_norm": 0.039472538977861404,
      "learning_rate": 4.417647058823529e-05,
      "loss": 0.0156,
      "step": 4851
    },
    {
      "epoch": 43.32,
      "grad_norm": 0.04416007548570633,
      "learning_rate": 4.4117647058823526e-05,
      "loss": 0.0174,
      "step": 4852
    },
    {
      "epoch": 43.33,
      "grad_norm": 0.045166872441768646,
      "learning_rate": 4.4058823529411755e-05,
      "loss": 0.0154,
      "step": 4853
    },
    {
      "epoch": 43.34,
      "grad_norm": 0.06845077872276306,
      "learning_rate": 4.4e-05,
      "loss": 0.0179,
      "step": 4854
    },
    {
      "epoch": 43.35,
      "grad_norm": 0.04094138368964195,
      "learning_rate": 4.3941176470588236e-05,
      "loss": 0.0167,
      "step": 4855
    },
    {
      "epoch": 43.36,
      "grad_norm": 0.041409075260162354,
      "learning_rate": 4.3882352941176465e-05,
      "loss": 0.0166,
      "step": 4856
    },
    {
      "epoch": 43.37,
      "grad_norm": 0.038814477622509,
      "learning_rate": 4.38235294117647e-05,
      "loss": 0.0158,
      "step": 4857
    },
    {
      "epoch": 43.38,
      "grad_norm": 0.04638247936964035,
      "learning_rate": 4.376470588235293e-05,
      "loss": 0.0167,
      "step": 4858
    },
    {
      "epoch": 43.38,
      "grad_norm": 0.05674579739570618,
      "learning_rate": 4.3705882352941175e-05,
      "loss": 0.0154,
      "step": 4859
    },
    {
      "epoch": 43.39,
      "grad_norm": 0.09118795394897461,
      "learning_rate": 4.3647058823529405e-05,
      "loss": 0.0194,
      "step": 4860
    },
    {
      "epoch": 43.4,
      "grad_norm": 0.04603675752878189,
      "learning_rate": 4.358823529411764e-05,
      "loss": 0.0185,
      "step": 4861
    },
    {
      "epoch": 43.41,
      "grad_norm": 0.043492477387189865,
      "learning_rate": 4.3529411764705885e-05,
      "loss": 0.0182,
      "step": 4862
    },
    {
      "epoch": 43.42,
      "grad_norm": 0.03807063400745392,
      "learning_rate": 4.3470588235294115e-05,
      "loss": 0.0146,
      "step": 4863
    },
    {
      "epoch": 43.43,
      "grad_norm": 0.04093649610877037,
      "learning_rate": 4.341176470588235e-05,
      "loss": 0.0172,
      "step": 4864
    },
    {
      "epoch": 43.44,
      "grad_norm": 0.04611308127641678,
      "learning_rate": 4.335294117647058e-05,
      "loss": 0.0156,
      "step": 4865
    },
    {
      "epoch": 43.45,
      "grad_norm": 0.09458320587873459,
      "learning_rate": 4.329411764705882e-05,
      "loss": 0.0227,
      "step": 4866
    },
    {
      "epoch": 43.46,
      "grad_norm": 0.043307170271873474,
      "learning_rate": 4.323529411764705e-05,
      "loss": 0.0149,
      "step": 4867
    },
    {
      "epoch": 43.46,
      "grad_norm": 0.04051913321018219,
      "learning_rate": 4.317647058823529e-05,
      "loss": 0.0143,
      "step": 4868
    },
    {
      "epoch": 43.47,
      "grad_norm": 0.04021545872092247,
      "learning_rate": 4.311764705882353e-05,
      "loss": 0.0156,
      "step": 4869
    },
    {
      "epoch": 43.48,
      "grad_norm": 0.046592261642217636,
      "learning_rate": 4.305882352941176e-05,
      "loss": 0.0179,
      "step": 4870
    },
    {
      "epoch": 43.49,
      "grad_norm": 0.049437109380960464,
      "learning_rate": 4.3e-05,
      "loss": 0.0164,
      "step": 4871
    },
    {
      "epoch": 43.5,
      "grad_norm": 0.05881152302026749,
      "learning_rate": 4.294117647058823e-05,
      "loss": 0.0174,
      "step": 4872
    },
    {
      "epoch": 43.51,
      "grad_norm": 0.09323722869157791,
      "learning_rate": 4.288235294117647e-05,
      "loss": 0.019,
      "step": 4873
    },
    {
      "epoch": 43.52,
      "grad_norm": 0.03924582153558731,
      "learning_rate": 4.28235294117647e-05,
      "loss": 0.0141,
      "step": 4874
    },
    {
      "epoch": 43.53,
      "grad_norm": 0.03648946061730385,
      "learning_rate": 4.2764705882352935e-05,
      "loss": 0.0137,
      "step": 4875
    },
    {
      "epoch": 43.54,
      "grad_norm": 0.04003776237368584,
      "learning_rate": 4.270588235294118e-05,
      "loss": 0.0168,
      "step": 4876
    },
    {
      "epoch": 43.54,
      "grad_norm": 0.05596655234694481,
      "learning_rate": 4.264705882352941e-05,
      "loss": 0.014,
      "step": 4877
    },
    {
      "epoch": 43.55,
      "grad_norm": 0.047849416732788086,
      "learning_rate": 4.2588235294117645e-05,
      "loss": 0.0113,
      "step": 4878
    },
    {
      "epoch": 43.56,
      "grad_norm": 0.08403313905000687,
      "learning_rate": 4.2529411764705875e-05,
      "loss": 0.0186,
      "step": 4879
    },
    {
      "epoch": 43.57,
      "grad_norm": 0.042822834104299545,
      "learning_rate": 4.247058823529411e-05,
      "loss": 0.0182,
      "step": 4880
    },
    {
      "epoch": 43.58,
      "grad_norm": 0.04393649473786354,
      "learning_rate": 4.241176470588235e-05,
      "loss": 0.0157,
      "step": 4881
    },
    {
      "epoch": 43.59,
      "grad_norm": 0.03740609437227249,
      "learning_rate": 4.2352941176470585e-05,
      "loss": 0.0148,
      "step": 4882
    },
    {
      "epoch": 43.6,
      "grad_norm": 0.049269191920757294,
      "learning_rate": 4.229411764705882e-05,
      "loss": 0.0185,
      "step": 4883
    },
    {
      "epoch": 43.61,
      "grad_norm": 0.054355815052986145,
      "learning_rate": 4.223529411764705e-05,
      "loss": 0.0173,
      "step": 4884
    },
    {
      "epoch": 43.62,
      "grad_norm": 0.09597691148519516,
      "learning_rate": 4.2176470588235294e-05,
      "loss": 0.0192,
      "step": 4885
    },
    {
      "epoch": 43.62,
      "grad_norm": 0.04877206310629845,
      "learning_rate": 4.2117647058823524e-05,
      "loss": 0.02,
      "step": 4886
    },
    {
      "epoch": 43.63,
      "grad_norm": 0.03655581921339035,
      "learning_rate": 4.205882352941176e-05,
      "loss": 0.0148,
      "step": 4887
    },
    {
      "epoch": 43.64,
      "grad_norm": 0.043534889817237854,
      "learning_rate": 4.2e-05,
      "loss": 0.0144,
      "step": 4888
    },
    {
      "epoch": 43.65,
      "grad_norm": 0.044560547918081284,
      "learning_rate": 4.194117647058823e-05,
      "loss": 0.0157,
      "step": 4889
    },
    {
      "epoch": 43.66,
      "grad_norm": 0.04925515875220299,
      "learning_rate": 4.188235294117647e-05,
      "loss": 0.0144,
      "step": 4890
    },
    {
      "epoch": 43.67,
      "grad_norm": 0.08285937458276749,
      "learning_rate": 4.18235294117647e-05,
      "loss": 0.0222,
      "step": 4891
    },
    {
      "epoch": 43.68,
      "grad_norm": 0.0362589992582798,
      "learning_rate": 4.176470588235294e-05,
      "loss": 0.0131,
      "step": 4892
    },
    {
      "epoch": 43.69,
      "grad_norm": 0.032939568161964417,
      "learning_rate": 4.170588235294117e-05,
      "loss": 0.0131,
      "step": 4893
    },
    {
      "epoch": 43.7,
      "grad_norm": 0.043853238224983215,
      "learning_rate": 4.164705882352941e-05,
      "loss": 0.0155,
      "step": 4894
    },
    {
      "epoch": 43.71,
      "grad_norm": 0.04488749802112579,
      "learning_rate": 4.158823529411765e-05,
      "loss": 0.0143,
      "step": 4895
    },
    {
      "epoch": 43.71,
      "grad_norm": 0.04842665046453476,
      "learning_rate": 4.152941176470588e-05,
      "loss": 0.0161,
      "step": 4896
    },
    {
      "epoch": 43.72,
      "grad_norm": 0.05583030730485916,
      "learning_rate": 4.1470588235294114e-05,
      "loss": 0.0146,
      "step": 4897
    },
    {
      "epoch": 43.73,
      "grad_norm": 0.07999975234270096,
      "learning_rate": 4.1411764705882344e-05,
      "loss": 0.0202,
      "step": 4898
    },
    {
      "epoch": 43.74,
      "grad_norm": 0.044483449310064316,
      "learning_rate": 4.135294117647059e-05,
      "loss": 0.0181,
      "step": 4899
    },
    {
      "epoch": 43.75,
      "grad_norm": 0.04633469134569168,
      "learning_rate": 4.129411764705882e-05,
      "loss": 0.0161,
      "step": 4900
    },
    {
      "epoch": 43.75,
      "eval_cer": 0.03188222780966126,
      "eval_loss": 0.21013225615024567,
      "eval_runtime": 22.254,
      "eval_samples_per_second": 118.72,
      "eval_steps_per_second": 1.887,
      "eval_wer": 0.10134867116223721,
      "step": 4900
    },
    {
      "epoch": 43.76,
      "grad_norm": 0.05003685876727104,
      "learning_rate": 4.1235294117647054e-05,
      "loss": 0.0185,
      "step": 4901
    },
    {
      "epoch": 43.77,
      "grad_norm": 0.04347279667854309,
      "learning_rate": 4.11764705882353e-05,
      "loss": 0.0161,
      "step": 4902
    },
    {
      "epoch": 43.78,
      "grad_norm": 0.054564401507377625,
      "learning_rate": 4.111764705882353e-05,
      "loss": 0.0145,
      "step": 4903
    },
    {
      "epoch": 43.79,
      "grad_norm": 0.09958654642105103,
      "learning_rate": 4.1058823529411764e-05,
      "loss": 0.0222,
      "step": 4904
    },
    {
      "epoch": 43.79,
      "grad_norm": 0.03584211319684982,
      "learning_rate": 4.0999999999999994e-05,
      "loss": 0.0154,
      "step": 4905
    },
    {
      "epoch": 43.8,
      "grad_norm": 0.043052252382040024,
      "learning_rate": 4.094117647058823e-05,
      "loss": 0.0149,
      "step": 4906
    },
    {
      "epoch": 43.81,
      "grad_norm": 0.045404739677906036,
      "learning_rate": 4.088235294117646e-05,
      "loss": 0.0194,
      "step": 4907
    },
    {
      "epoch": 43.82,
      "grad_norm": 0.042859796434640884,
      "learning_rate": 4.0823529411764704e-05,
      "loss": 0.0147,
      "step": 4908
    },
    {
      "epoch": 43.83,
      "grad_norm": 0.05667326971888542,
      "learning_rate": 4.076470588235294e-05,
      "loss": 0.0144,
      "step": 4909
    },
    {
      "epoch": 43.84,
      "grad_norm": 0.0632072240114212,
      "learning_rate": 4.070588235294117e-05,
      "loss": 0.0172,
      "step": 4910
    },
    {
      "epoch": 43.85,
      "grad_norm": 0.041090261191129684,
      "learning_rate": 4.0647058823529414e-05,
      "loss": 0.0151,
      "step": 4911
    },
    {
      "epoch": 43.86,
      "grad_norm": 0.04376598820090294,
      "learning_rate": 4.0588235294117644e-05,
      "loss": 0.0171,
      "step": 4912
    },
    {
      "epoch": 43.87,
      "grad_norm": 0.039289943873882294,
      "learning_rate": 4.052941176470588e-05,
      "loss": 0.016,
      "step": 4913
    },
    {
      "epoch": 43.88,
      "grad_norm": 0.047562599182128906,
      "learning_rate": 4.047058823529411e-05,
      "loss": 0.0181,
      "step": 4914
    },
    {
      "epoch": 43.88,
      "grad_norm": 0.04658946394920349,
      "learning_rate": 4.041176470588235e-05,
      "loss": 0.0156,
      "step": 4915
    },
    {
      "epoch": 43.89,
      "grad_norm": 0.09022916853427887,
      "learning_rate": 4.035294117647059e-05,
      "loss": 0.0194,
      "step": 4916
    },
    {
      "epoch": 43.9,
      "grad_norm": 0.03939886763691902,
      "learning_rate": 4.029411764705882e-05,
      "loss": 0.0153,
      "step": 4917
    },
    {
      "epoch": 43.91,
      "grad_norm": 0.04202469810843468,
      "learning_rate": 4.023529411764706e-05,
      "loss": 0.0193,
      "step": 4918
    },
    {
      "epoch": 43.92,
      "grad_norm": 0.04549553617835045,
      "learning_rate": 4.0176470588235287e-05,
      "loss": 0.0179,
      "step": 4919
    },
    {
      "epoch": 43.93,
      "grad_norm": 0.04405608028173447,
      "learning_rate": 4.011764705882353e-05,
      "loss": 0.0174,
      "step": 4920
    },
    {
      "epoch": 43.94,
      "grad_norm": 0.03873012959957123,
      "learning_rate": 4.005882352941176e-05,
      "loss": 0.0119,
      "step": 4921
    },
    {
      "epoch": 43.95,
      "grad_norm": 0.06394640356302261,
      "learning_rate": 3.9999999999999996e-05,
      "loss": 0.0202,
      "step": 4922
    },
    {
      "epoch": 43.96,
      "grad_norm": 0.07013455033302307,
      "learning_rate": 3.994117647058823e-05,
      "loss": 0.0199,
      "step": 4923
    },
    {
      "epoch": 43.96,
      "grad_norm": 0.04394387826323509,
      "learning_rate": 3.988235294117646e-05,
      "loss": 0.0177,
      "step": 4924
    },
    {
      "epoch": 43.97,
      "grad_norm": 0.04156498610973358,
      "learning_rate": 3.9823529411764706e-05,
      "loss": 0.016,
      "step": 4925
    },
    {
      "epoch": 43.98,
      "grad_norm": 0.042225249111652374,
      "learning_rate": 3.9764705882352936e-05,
      "loss": 0.0147,
      "step": 4926
    },
    {
      "epoch": 43.99,
      "grad_norm": 0.049209680408239365,
      "learning_rate": 3.970588235294117e-05,
      "loss": 0.0159,
      "step": 4927
    },
    {
      "epoch": 44.0,
      "grad_norm": 0.0688699260354042,
      "learning_rate": 3.964705882352941e-05,
      "loss": 0.0183,
      "step": 4928
    },
    {
      "epoch": 44.01,
      "grad_norm": 0.04197148606181145,
      "learning_rate": 3.958823529411764e-05,
      "loss": 0.0165,
      "step": 4929
    },
    {
      "epoch": 44.02,
      "grad_norm": 0.03934997692704201,
      "learning_rate": 3.952941176470588e-05,
      "loss": 0.015,
      "step": 4930
    },
    {
      "epoch": 44.03,
      "grad_norm": 0.04739455133676529,
      "learning_rate": 3.947058823529411e-05,
      "loss": 0.0151,
      "step": 4931
    },
    {
      "epoch": 44.04,
      "grad_norm": 0.04214252158999443,
      "learning_rate": 3.941176470588235e-05,
      "loss": 0.0153,
      "step": 4932
    },
    {
      "epoch": 44.04,
      "grad_norm": 0.046974487602710724,
      "learning_rate": 3.935294117647058e-05,
      "loss": 0.0163,
      "step": 4933
    },
    {
      "epoch": 44.05,
      "grad_norm": 0.0665048211812973,
      "learning_rate": 3.929411764705882e-05,
      "loss": 0.0185,
      "step": 4934
    },
    {
      "epoch": 44.06,
      "grad_norm": 0.06398850679397583,
      "learning_rate": 3.923529411764706e-05,
      "loss": 0.019,
      "step": 4935
    },
    {
      "epoch": 44.07,
      "grad_norm": 0.040308427065610886,
      "learning_rate": 3.917647058823529e-05,
      "loss": 0.0188,
      "step": 4936
    },
    {
      "epoch": 44.08,
      "grad_norm": 0.0501236766576767,
      "learning_rate": 3.9117647058823526e-05,
      "loss": 0.0199,
      "step": 4937
    },
    {
      "epoch": 44.09,
      "grad_norm": 0.03701817989349365,
      "learning_rate": 3.9058823529411756e-05,
      "loss": 0.0127,
      "step": 4938
    },
    {
      "epoch": 44.1,
      "grad_norm": 0.04599235579371452,
      "learning_rate": 3.9e-05,
      "loss": 0.0167,
      "step": 4939
    },
    {
      "epoch": 44.11,
      "grad_norm": 0.0533258430659771,
      "learning_rate": 3.894117647058823e-05,
      "loss": 0.0146,
      "step": 4940
    },
    {
      "epoch": 44.12,
      "grad_norm": 0.06163892522454262,
      "learning_rate": 3.8882352941176466e-05,
      "loss": 0.0141,
      "step": 4941
    },
    {
      "epoch": 44.12,
      "grad_norm": 0.042363330721855164,
      "learning_rate": 3.882352941176471e-05,
      "loss": 0.0184,
      "step": 4942
    },
    {
      "epoch": 44.13,
      "grad_norm": 0.04036133736371994,
      "learning_rate": 3.876470588235294e-05,
      "loss": 0.0148,
      "step": 4943
    },
    {
      "epoch": 44.14,
      "grad_norm": 0.04974912479519844,
      "learning_rate": 3.8705882352941176e-05,
      "loss": 0.0186,
      "step": 4944
    },
    {
      "epoch": 44.15,
      "grad_norm": 0.04702618345618248,
      "learning_rate": 3.8647058823529406e-05,
      "loss": 0.0175,
      "step": 4945
    },
    {
      "epoch": 44.16,
      "grad_norm": 0.04483968764543533,
      "learning_rate": 3.858823529411764e-05,
      "loss": 0.0134,
      "step": 4946
    },
    {
      "epoch": 44.17,
      "grad_norm": 0.06495136022567749,
      "learning_rate": 3.852941176470587e-05,
      "loss": 0.0179,
      "step": 4947
    },
    {
      "epoch": 44.18,
      "grad_norm": 0.042255327105522156,
      "learning_rate": 3.8470588235294116e-05,
      "loss": 0.0188,
      "step": 4948
    },
    {
      "epoch": 44.19,
      "grad_norm": 0.04956101253628731,
      "learning_rate": 3.841176470588235e-05,
      "loss": 0.0183,
      "step": 4949
    },
    {
      "epoch": 44.2,
      "grad_norm": 0.04366575926542282,
      "learning_rate": 3.835294117647058e-05,
      "loss": 0.0163,
      "step": 4950
    },
    {
      "epoch": 44.21,
      "grad_norm": 0.047807127237319946,
      "learning_rate": 3.8294117647058826e-05,
      "loss": 0.0176,
      "step": 4951
    },
    {
      "epoch": 44.21,
      "grad_norm": 0.04627014324069023,
      "learning_rate": 3.8235294117647055e-05,
      "loss": 0.0122,
      "step": 4952
    },
    {
      "epoch": 44.22,
      "grad_norm": 0.07453977316617966,
      "learning_rate": 3.817647058823529e-05,
      "loss": 0.0172,
      "step": 4953
    },
    {
      "epoch": 44.23,
      "grad_norm": 0.03574393689632416,
      "learning_rate": 3.811764705882352e-05,
      "loss": 0.0135,
      "step": 4954
    },
    {
      "epoch": 44.24,
      "grad_norm": 0.04239301383495331,
      "learning_rate": 3.805882352941176e-05,
      "loss": 0.0141,
      "step": 4955
    },
    {
      "epoch": 44.25,
      "grad_norm": 0.04692602902650833,
      "learning_rate": 3.8e-05,
      "loss": 0.0164,
      "step": 4956
    },
    {
      "epoch": 44.26,
      "grad_norm": 0.04097558930516243,
      "learning_rate": 3.794117647058823e-05,
      "loss": 0.0155,
      "step": 4957
    },
    {
      "epoch": 44.27,
      "grad_norm": 0.04169277474284172,
      "learning_rate": 3.788235294117647e-05,
      "loss": 0.0166,
      "step": 4958
    },
    {
      "epoch": 44.28,
      "grad_norm": 0.07089413702487946,
      "learning_rate": 3.78235294117647e-05,
      "loss": 0.0184,
      "step": 4959
    },
    {
      "epoch": 44.29,
      "grad_norm": 0.05112564191222191,
      "learning_rate": 3.776470588235294e-05,
      "loss": 0.0184,
      "step": 4960
    },
    {
      "epoch": 44.29,
      "grad_norm": 0.037595827132463455,
      "learning_rate": 3.770588235294117e-05,
      "loss": 0.0152,
      "step": 4961
    },
    {
      "epoch": 44.3,
      "grad_norm": 0.03754653409123421,
      "learning_rate": 3.764705882352941e-05,
      "loss": 0.0143,
      "step": 4962
    },
    {
      "epoch": 44.31,
      "grad_norm": 0.03680499270558357,
      "learning_rate": 3.7588235294117645e-05,
      "loss": 0.0146,
      "step": 4963
    },
    {
      "epoch": 44.32,
      "grad_norm": 0.046565115451812744,
      "learning_rate": 3.7529411764705875e-05,
      "loss": 0.0142,
      "step": 4964
    },
    {
      "epoch": 44.33,
      "grad_norm": 0.06004823371767998,
      "learning_rate": 3.747058823529412e-05,
      "loss": 0.0163,
      "step": 4965
    },
    {
      "epoch": 44.34,
      "grad_norm": 0.08733204752206802,
      "learning_rate": 3.741176470588235e-05,
      "loss": 0.022,
      "step": 4966
    },
    {
      "epoch": 44.35,
      "grad_norm": 0.041879232972860336,
      "learning_rate": 3.7352941176470585e-05,
      "loss": 0.0191,
      "step": 4967
    },
    {
      "epoch": 44.36,
      "grad_norm": 0.037341900169849396,
      "learning_rate": 3.729411764705882e-05,
      "loss": 0.0152,
      "step": 4968
    },
    {
      "epoch": 44.37,
      "grad_norm": 0.04649944603443146,
      "learning_rate": 3.723529411764705e-05,
      "loss": 0.0177,
      "step": 4969
    },
    {
      "epoch": 44.38,
      "grad_norm": 0.04623737558722496,
      "learning_rate": 3.717647058823529e-05,
      "loss": 0.0156,
      "step": 4970
    },
    {
      "epoch": 44.38,
      "grad_norm": 0.046323150396347046,
      "learning_rate": 3.7117647058823525e-05,
      "loss": 0.0126,
      "step": 4971
    },
    {
      "epoch": 44.39,
      "grad_norm": 0.06873354315757751,
      "learning_rate": 3.705882352941176e-05,
      "loss": 0.0219,
      "step": 4972
    },
    {
      "epoch": 44.4,
      "grad_norm": 0.03552534431219101,
      "learning_rate": 3.7e-05,
      "loss": 0.0153,
      "step": 4973
    },
    {
      "epoch": 44.41,
      "grad_norm": 0.04047302156686783,
      "learning_rate": 3.6941176470588235e-05,
      "loss": 0.0158,
      "step": 4974
    },
    {
      "epoch": 44.42,
      "grad_norm": 0.03792682662606239,
      "learning_rate": 3.6882352941176465e-05,
      "loss": 0.0132,
      "step": 4975
    },
    {
      "epoch": 44.43,
      "grad_norm": 0.03600406274199486,
      "learning_rate": 3.68235294117647e-05,
      "loss": 0.0142,
      "step": 4976
    },
    {
      "epoch": 44.44,
      "grad_norm": 0.0528290793299675,
      "learning_rate": 3.676470588235294e-05,
      "loss": 0.0205,
      "step": 4977
    },
    {
      "epoch": 44.45,
      "grad_norm": 0.07228340953588486,
      "learning_rate": 3.6705882352941175e-05,
      "loss": 0.0177,
      "step": 4978
    },
    {
      "epoch": 44.46,
      "grad_norm": 0.04562455415725708,
      "learning_rate": 3.664705882352941e-05,
      "loss": 0.0182,
      "step": 4979
    },
    {
      "epoch": 44.46,
      "grad_norm": 0.04094189032912254,
      "learning_rate": 3.658823529411765e-05,
      "loss": 0.0172,
      "step": 4980
    },
    {
      "epoch": 44.47,
      "grad_norm": 0.04234864190220833,
      "learning_rate": 3.652941176470588e-05,
      "loss": 0.0167,
      "step": 4981
    },
    {
      "epoch": 44.48,
      "grad_norm": 0.04225002974271774,
      "learning_rate": 3.6470588235294114e-05,
      "loss": 0.0163,
      "step": 4982
    },
    {
      "epoch": 44.49,
      "grad_norm": 0.04998965933918953,
      "learning_rate": 3.641176470588235e-05,
      "loss": 0.0146,
      "step": 4983
    },
    {
      "epoch": 44.5,
      "grad_norm": 0.09114740043878555,
      "learning_rate": 3.635294117647058e-05,
      "loss": 0.0218,
      "step": 4984
    },
    {
      "epoch": 44.51,
      "grad_norm": 0.054906509816646576,
      "learning_rate": 3.6294117647058824e-05,
      "loss": 0.0141,
      "step": 4985
    },
    {
      "epoch": 44.52,
      "grad_norm": 0.04177793860435486,
      "learning_rate": 3.6235294117647054e-05,
      "loss": 0.0172,
      "step": 4986
    },
    {
      "epoch": 44.53,
      "grad_norm": 0.05029189586639404,
      "learning_rate": 3.617647058823529e-05,
      "loss": 0.0178,
      "step": 4987
    },
    {
      "epoch": 44.54,
      "grad_norm": 0.04333079233765602,
      "learning_rate": 3.611764705882353e-05,
      "loss": 0.0151,
      "step": 4988
    },
    {
      "epoch": 44.54,
      "grad_norm": 0.042823683470487595,
      "learning_rate": 3.605882352941176e-05,
      "loss": 0.015,
      "step": 4989
    },
    {
      "epoch": 44.55,
      "grad_norm": 0.03932007774710655,
      "learning_rate": 3.5999999999999994e-05,
      "loss": 0.0102,
      "step": 4990
    },
    {
      "epoch": 44.56,
      "grad_norm": 0.07560725510120392,
      "learning_rate": 3.594117647058823e-05,
      "loss": 0.0208,
      "step": 4991
    },
    {
      "epoch": 44.57,
      "grad_norm": 0.03645060583949089,
      "learning_rate": 3.588235294117647e-05,
      "loss": 0.015,
      "step": 4992
    },
    {
      "epoch": 44.58,
      "grad_norm": 0.048673465847969055,
      "learning_rate": 3.5823529411764704e-05,
      "loss": 0.0171,
      "step": 4993
    },
    {
      "epoch": 44.59,
      "grad_norm": 0.04412667080760002,
      "learning_rate": 3.576470588235294e-05,
      "loss": 0.0187,
      "step": 4994
    },
    {
      "epoch": 44.6,
      "grad_norm": 0.04287470504641533,
      "learning_rate": 3.570588235294117e-05,
      "loss": 0.0162,
      "step": 4995
    },
    {
      "epoch": 44.61,
      "grad_norm": 0.03945418447256088,
      "learning_rate": 3.564705882352941e-05,
      "loss": 0.0128,
      "step": 4996
    },
    {
      "epoch": 44.62,
      "grad_norm": 0.13640552759170532,
      "learning_rate": 3.5588235294117644e-05,
      "loss": 0.0217,
      "step": 4997
    },
    {
      "epoch": 44.62,
      "grad_norm": 0.04495558887720108,
      "learning_rate": 3.552941176470588e-05,
      "loss": 0.0153,
      "step": 4998
    },
    {
      "epoch": 44.63,
      "grad_norm": 0.037331100553274155,
      "learning_rate": 3.547058823529412e-05,
      "loss": 0.0161,
      "step": 4999
    },
    {
      "epoch": 44.64,
      "grad_norm": 0.03802134841680527,
      "learning_rate": 3.5411764705882354e-05,
      "loss": 0.0168,
      "step": 5000
    },
    {
      "epoch": 44.64,
      "eval_cer": 0.03064948341244696,
      "eval_loss": 0.2111145257949829,
      "eval_runtime": 22.1584,
      "eval_samples_per_second": 119.232,
      "eval_steps_per_second": 1.895,
      "eval_wer": 0.09847282824276081,
      "step": 5000
    },
    {
      "epoch": 44.65,
      "grad_norm": 0.03596673160791397,
      "learning_rate": 3.5352941176470584e-05,
      "loss": 0.0124,
      "step": 5001
    },
    {
      "epoch": 44.66,
      "grad_norm": 0.039762191474437714,
      "learning_rate": 3.529411764705882e-05,
      "loss": 0.013,
      "step": 5002
    },
    {
      "epoch": 44.67,
      "grad_norm": 0.08294077217578888,
      "learning_rate": 3.523529411764706e-05,
      "loss": 0.0176,
      "step": 5003
    },
    {
      "epoch": 44.68,
      "grad_norm": 0.04089253768324852,
      "learning_rate": 3.517647058823529e-05,
      "loss": 0.016,
      "step": 5004
    },
    {
      "epoch": 44.69,
      "grad_norm": 0.04697546362876892,
      "learning_rate": 3.511764705882353e-05,
      "loss": 0.0185,
      "step": 5005
    },
    {
      "epoch": 44.7,
      "grad_norm": 0.039550378918647766,
      "learning_rate": 3.505882352941176e-05,
      "loss": 0.0156,
      "step": 5006
    },
    {
      "epoch": 44.71,
      "grad_norm": 0.056655146181583405,
      "learning_rate": 3.5e-05,
      "loss": 0.0157,
      "step": 5007
    },
    {
      "epoch": 44.71,
      "grad_norm": 0.05132512003183365,
      "learning_rate": 3.4941176470588234e-05,
      "loss": 0.0177,
      "step": 5008
    },
    {
      "epoch": 44.72,
      "grad_norm": 0.0632912665605545,
      "learning_rate": 3.4882352941176463e-05,
      "loss": 0.0153,
      "step": 5009
    },
    {
      "epoch": 44.73,
      "grad_norm": 0.11135727167129517,
      "learning_rate": 3.48235294117647e-05,
      "loss": 0.0232,
      "step": 5010
    },
    {
      "epoch": 44.74,
      "grad_norm": 0.03697282448410988,
      "learning_rate": 3.476470588235294e-05,
      "loss": 0.0155,
      "step": 5011
    },
    {
      "epoch": 44.75,
      "grad_norm": 0.03992462903261185,
      "learning_rate": 3.4705882352941173e-05,
      "loss": 0.0153,
      "step": 5012
    },
    {
      "epoch": 44.76,
      "grad_norm": 0.04411754757165909,
      "learning_rate": 3.464705882352941e-05,
      "loss": 0.018,
      "step": 5013
    },
    {
      "epoch": 44.77,
      "grad_norm": 0.04339717701077461,
      "learning_rate": 3.458823529411765e-05,
      "loss": 0.0156,
      "step": 5014
    },
    {
      "epoch": 44.78,
      "grad_norm": 0.04332556948065758,
      "learning_rate": 3.452941176470588e-05,
      "loss": 0.0116,
      "step": 5015
    },
    {
      "epoch": 44.79,
      "grad_norm": 0.061066947877407074,
      "learning_rate": 3.447058823529411e-05,
      "loss": 0.0168,
      "step": 5016
    },
    {
      "epoch": 44.79,
      "grad_norm": 0.037729501724243164,
      "learning_rate": 3.441176470588235e-05,
      "loss": 0.0144,
      "step": 5017
    },
    {
      "epoch": 44.8,
      "grad_norm": 0.0411374568939209,
      "learning_rate": 3.4352941176470587e-05,
      "loss": 0.0174,
      "step": 5018
    },
    {
      "epoch": 44.81,
      "grad_norm": 0.04277394711971283,
      "learning_rate": 3.429411764705882e-05,
      "loss": 0.0155,
      "step": 5019
    },
    {
      "epoch": 44.82,
      "grad_norm": 0.03835034370422363,
      "learning_rate": 3.423529411764706e-05,
      "loss": 0.0134,
      "step": 5020
    },
    {
      "epoch": 44.83,
      "grad_norm": 0.0481446236371994,
      "learning_rate": 3.417647058823529e-05,
      "loss": 0.016,
      "step": 5021
    },
    {
      "epoch": 44.84,
      "grad_norm": 0.07615407556295395,
      "learning_rate": 3.4117647058823526e-05,
      "loss": 0.0232,
      "step": 5022
    },
    {
      "epoch": 44.85,
      "grad_norm": 0.040802374482154846,
      "learning_rate": 3.405882352941176e-05,
      "loss": 0.0155,
      "step": 5023
    },
    {
      "epoch": 44.86,
      "grad_norm": 0.04113549739122391,
      "learning_rate": 3.399999999999999e-05,
      "loss": 0.0163,
      "step": 5024
    },
    {
      "epoch": 44.87,
      "grad_norm": 0.039817169308662415,
      "learning_rate": 3.3941176470588236e-05,
      "loss": 0.0143,
      "step": 5025
    },
    {
      "epoch": 44.88,
      "grad_norm": 0.04265448823571205,
      "learning_rate": 3.3882352941176466e-05,
      "loss": 0.0161,
      "step": 5026
    },
    {
      "epoch": 44.88,
      "grad_norm": 0.05202582851052284,
      "learning_rate": 3.38235294117647e-05,
      "loss": 0.017,
      "step": 5027
    },
    {
      "epoch": 44.89,
      "grad_norm": 0.09392998367547989,
      "learning_rate": 3.376470588235294e-05,
      "loss": 0.0193,
      "step": 5028
    },
    {
      "epoch": 44.9,
      "grad_norm": 0.037095922976732254,
      "learning_rate": 3.3705882352941176e-05,
      "loss": 0.0149,
      "step": 5029
    },
    {
      "epoch": 44.91,
      "grad_norm": 0.04058606177568436,
      "learning_rate": 3.3647058823529406e-05,
      "loss": 0.0171,
      "step": 5030
    },
    {
      "epoch": 44.92,
      "grad_norm": 0.045299794524908066,
      "learning_rate": 3.358823529411764e-05,
      "loss": 0.0166,
      "step": 5031
    },
    {
      "epoch": 44.93,
      "grad_norm": 0.04051448032259941,
      "learning_rate": 3.352941176470588e-05,
      "loss": 0.0155,
      "step": 5032
    },
    {
      "epoch": 44.94,
      "grad_norm": 0.048243798315525055,
      "learning_rate": 3.3470588235294116e-05,
      "loss": 0.0169,
      "step": 5033
    },
    {
      "epoch": 44.95,
      "grad_norm": 0.07924217730760574,
      "learning_rate": 3.341176470588235e-05,
      "loss": 0.0157,
      "step": 5034
    },
    {
      "epoch": 44.96,
      "grad_norm": 0.06391122937202454,
      "learning_rate": 3.335294117647058e-05,
      "loss": 0.0173,
      "step": 5035
    },
    {
      "epoch": 44.96,
      "grad_norm": 0.039239589124917984,
      "learning_rate": 3.329411764705882e-05,
      "loss": 0.0164,
      "step": 5036
    },
    {
      "epoch": 44.97,
      "grad_norm": 0.04166821390390396,
      "learning_rate": 3.3235294117647056e-05,
      "loss": 0.0129,
      "step": 5037
    },
    {
      "epoch": 44.98,
      "grad_norm": 0.04653739556670189,
      "learning_rate": 3.317647058823529e-05,
      "loss": 0.0143,
      "step": 5038
    },
    {
      "epoch": 44.99,
      "grad_norm": 0.04925594851374626,
      "learning_rate": 3.311764705882353e-05,
      "loss": 0.0145,
      "step": 5039
    },
    {
      "epoch": 45.0,
      "grad_norm": 0.05927234888076782,
      "learning_rate": 3.3058823529411766e-05,
      "loss": 0.0199,
      "step": 5040
    },
    {
      "epoch": 45.01,
      "grad_norm": 0.037993013858795166,
      "learning_rate": 3.2999999999999996e-05,
      "loss": 0.0177,
      "step": 5041
    },
    {
      "epoch": 45.02,
      "grad_norm": 0.03294490650296211,
      "learning_rate": 3.294117647058823e-05,
      "loss": 0.0122,
      "step": 5042
    },
    {
      "epoch": 45.03,
      "grad_norm": 0.035598304122686386,
      "learning_rate": 3.288235294117647e-05,
      "loss": 0.0148,
      "step": 5043
    },
    {
      "epoch": 45.04,
      "grad_norm": 0.04196874424815178,
      "learning_rate": 3.28235294117647e-05,
      "loss": 0.0148,
      "step": 5044
    },
    {
      "epoch": 45.04,
      "grad_norm": 0.04705050215125084,
      "learning_rate": 3.276470588235294e-05,
      "loss": 0.0162,
      "step": 5045
    },
    {
      "epoch": 45.05,
      "grad_norm": 0.06958391517400742,
      "learning_rate": 3.270588235294117e-05,
      "loss": 0.0206,
      "step": 5046
    },
    {
      "epoch": 45.06,
      "grad_norm": 0.054922085255384445,
      "learning_rate": 3.264705882352941e-05,
      "loss": 0.0195,
      "step": 5047
    },
    {
      "epoch": 45.07,
      "grad_norm": 0.03778674080967903,
      "learning_rate": 3.2588235294117646e-05,
      "loss": 0.0147,
      "step": 5048
    },
    {
      "epoch": 45.08,
      "grad_norm": 0.042429905384778976,
      "learning_rate": 3.252941176470588e-05,
      "loss": 0.0152,
      "step": 5049
    },
    {
      "epoch": 45.09,
      "grad_norm": 0.03815051540732384,
      "learning_rate": 3.247058823529411e-05,
      "loss": 0.0131,
      "step": 5050
    },
    {
      "epoch": 45.1,
      "grad_norm": 0.0463109090924263,
      "learning_rate": 3.241176470588235e-05,
      "loss": 0.0158,
      "step": 5051
    },
    {
      "epoch": 45.11,
      "grad_norm": 0.062121860682964325,
      "learning_rate": 3.2352941176470585e-05,
      "loss": 0.0204,
      "step": 5052
    },
    {
      "epoch": 45.12,
      "grad_norm": 0.07599413394927979,
      "learning_rate": 3.229411764705882e-05,
      "loss": 0.0181,
      "step": 5053
    },
    {
      "epoch": 45.12,
      "grad_norm": 0.0417756550014019,
      "learning_rate": 3.223529411764706e-05,
      "loss": 0.0146,
      "step": 5054
    },
    {
      "epoch": 45.13,
      "grad_norm": 0.039690159261226654,
      "learning_rate": 3.217647058823529e-05,
      "loss": 0.016,
      "step": 5055
    },
    {
      "epoch": 45.14,
      "grad_norm": 0.03887281194329262,
      "learning_rate": 3.2117647058823525e-05,
      "loss": 0.0154,
      "step": 5056
    },
    {
      "epoch": 45.15,
      "grad_norm": 0.050522856414318085,
      "learning_rate": 3.205882352941176e-05,
      "loss": 0.0173,
      "step": 5057
    },
    {
      "epoch": 45.16,
      "grad_norm": 0.04674147441983223,
      "learning_rate": 3.2e-05,
      "loss": 0.0145,
      "step": 5058
    },
    {
      "epoch": 45.17,
      "grad_norm": 0.0707877054810524,
      "learning_rate": 3.1941176470588235e-05,
      "loss": 0.0157,
      "step": 5059
    },
    {
      "epoch": 45.18,
      "grad_norm": 0.03815138339996338,
      "learning_rate": 3.188235294117647e-05,
      "loss": 0.0171,
      "step": 5060
    },
    {
      "epoch": 45.19,
      "grad_norm": 0.038219183683395386,
      "learning_rate": 3.18235294117647e-05,
      "loss": 0.016,
      "step": 5061
    },
    {
      "epoch": 45.2,
      "grad_norm": 0.045984767377376556,
      "learning_rate": 3.176470588235294e-05,
      "loss": 0.0177,
      "step": 5062
    },
    {
      "epoch": 45.21,
      "grad_norm": 0.043422240763902664,
      "learning_rate": 3.1705882352941175e-05,
      "loss": 0.0139,
      "step": 5063
    },
    {
      "epoch": 45.21,
      "grad_norm": 0.04400533810257912,
      "learning_rate": 3.1647058823529405e-05,
      "loss": 0.014,
      "step": 5064
    },
    {
      "epoch": 45.22,
      "grad_norm": 0.07776004076004028,
      "learning_rate": 3.158823529411765e-05,
      "loss": 0.0199,
      "step": 5065
    },
    {
      "epoch": 45.23,
      "grad_norm": 0.04204792156815529,
      "learning_rate": 3.152941176470588e-05,
      "loss": 0.0164,
      "step": 5066
    },
    {
      "epoch": 45.24,
      "grad_norm": 0.0436769463121891,
      "learning_rate": 3.1470588235294115e-05,
      "loss": 0.0174,
      "step": 5067
    },
    {
      "epoch": 45.25,
      "grad_norm": 0.04520263895392418,
      "learning_rate": 3.141176470588235e-05,
      "loss": 0.0156,
      "step": 5068
    },
    {
      "epoch": 45.26,
      "grad_norm": 0.037524010986089706,
      "learning_rate": 3.135294117647059e-05,
      "loss": 0.0144,
      "step": 5069
    },
    {
      "epoch": 45.27,
      "grad_norm": 0.04929890111088753,
      "learning_rate": 3.129411764705882e-05,
      "loss": 0.0156,
      "step": 5070
    },
    {
      "epoch": 45.28,
      "grad_norm": 0.06272881478071213,
      "learning_rate": 3.1235294117647055e-05,
      "loss": 0.0171,
      "step": 5071
    },
    {
      "epoch": 45.29,
      "grad_norm": 0.04566546902060509,
      "learning_rate": 3.117647058823529e-05,
      "loss": 0.0154,
      "step": 5072
    },
    {
      "epoch": 45.29,
      "grad_norm": 0.03950928524136543,
      "learning_rate": 3.111764705882353e-05,
      "loss": 0.0153,
      "step": 5073
    },
    {
      "epoch": 45.3,
      "grad_norm": 0.03667450323700905,
      "learning_rate": 3.1058823529411765e-05,
      "loss": 0.0128,
      "step": 5074
    },
    {
      "epoch": 45.31,
      "grad_norm": 0.046285588294267654,
      "learning_rate": 3.0999999999999995e-05,
      "loss": 0.0217,
      "step": 5075
    },
    {
      "epoch": 45.32,
      "grad_norm": 0.04641619697213173,
      "learning_rate": 3.094117647058823e-05,
      "loss": 0.0164,
      "step": 5076
    },
    {
      "epoch": 45.33,
      "grad_norm": 0.0505821518599987,
      "learning_rate": 3.088235294117647e-05,
      "loss": 0.0163,
      "step": 5077
    },
    {
      "epoch": 45.34,
      "grad_norm": 0.06749798357486725,
      "learning_rate": 3.0823529411764705e-05,
      "loss": 0.016,
      "step": 5078
    },
    {
      "epoch": 45.35,
      "grad_norm": 0.04133579507470131,
      "learning_rate": 3.076470588235294e-05,
      "loss": 0.0134,
      "step": 5079
    },
    {
      "epoch": 45.36,
      "grad_norm": 0.04400995746254921,
      "learning_rate": 3.070588235294118e-05,
      "loss": 0.0161,
      "step": 5080
    },
    {
      "epoch": 45.37,
      "grad_norm": 0.04433593526482582,
      "learning_rate": 3.064705882352941e-05,
      "loss": 0.0146,
      "step": 5081
    },
    {
      "epoch": 45.38,
      "grad_norm": 0.04578491300344467,
      "learning_rate": 3.0588235294117644e-05,
      "loss": 0.0167,
      "step": 5082
    },
    {
      "epoch": 45.38,
      "grad_norm": 0.04872238263487816,
      "learning_rate": 3.052941176470588e-05,
      "loss": 0.017,
      "step": 5083
    },
    {
      "epoch": 45.39,
      "grad_norm": 0.08894068002700806,
      "learning_rate": 3.0470588235294114e-05,
      "loss": 0.021,
      "step": 5084
    },
    {
      "epoch": 45.4,
      "grad_norm": 0.037841398268938065,
      "learning_rate": 3.041176470588235e-05,
      "loss": 0.0151,
      "step": 5085
    },
    {
      "epoch": 45.41,
      "grad_norm": 0.041993726044893265,
      "learning_rate": 3.0352941176470588e-05,
      "loss": 0.0161,
      "step": 5086
    },
    {
      "epoch": 45.42,
      "grad_norm": 0.04302545636892319,
      "learning_rate": 3.029411764705882e-05,
      "loss": 0.0161,
      "step": 5087
    },
    {
      "epoch": 45.43,
      "grad_norm": 0.045309703797101974,
      "learning_rate": 3.0235294117647058e-05,
      "loss": 0.0129,
      "step": 5088
    },
    {
      "epoch": 45.44,
      "grad_norm": 0.060696911066770554,
      "learning_rate": 3.017647058823529e-05,
      "loss": 0.0213,
      "step": 5089
    },
    {
      "epoch": 45.45,
      "grad_norm": 0.07425782829523087,
      "learning_rate": 3.0117647058823524e-05,
      "loss": 0.0193,
      "step": 5090
    },
    {
      "epoch": 45.46,
      "grad_norm": 0.03925832360982895,
      "learning_rate": 3.005882352941176e-05,
      "loss": 0.0155,
      "step": 5091
    },
    {
      "epoch": 45.46,
      "grad_norm": 0.039438169449567795,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 0.0146,
      "step": 5092
    },
    {
      "epoch": 45.47,
      "grad_norm": 0.03728676959872246,
      "learning_rate": 2.9941176470588234e-05,
      "loss": 0.0121,
      "step": 5093
    },
    {
      "epoch": 45.48,
      "grad_norm": 0.043770913034677505,
      "learning_rate": 2.9882352941176467e-05,
      "loss": 0.0164,
      "step": 5094
    },
    {
      "epoch": 45.49,
      "grad_norm": 0.06378931552171707,
      "learning_rate": 2.9823529411764704e-05,
      "loss": 0.017,
      "step": 5095
    },
    {
      "epoch": 45.5,
      "grad_norm": 0.06921689212322235,
      "learning_rate": 2.9764705882352937e-05,
      "loss": 0.0192,
      "step": 5096
    },
    {
      "epoch": 45.51,
      "grad_norm": 0.06097017973661423,
      "learning_rate": 2.9705882352941174e-05,
      "loss": 0.0173,
      "step": 5097
    },
    {
      "epoch": 45.52,
      "grad_norm": 0.036359891295433044,
      "learning_rate": 2.9647058823529407e-05,
      "loss": 0.0157,
      "step": 5098
    },
    {
      "epoch": 45.53,
      "grad_norm": 0.04372749477624893,
      "learning_rate": 2.9588235294117647e-05,
      "loss": 0.0145,
      "step": 5099
    },
    {
      "epoch": 45.54,
      "grad_norm": 0.039443690329790115,
      "learning_rate": 2.952941176470588e-05,
      "loss": 0.015,
      "step": 5100
    },
    {
      "epoch": 45.54,
      "eval_cer": 0.03175068496239754,
      "eval_loss": 0.21100349724292755,
      "eval_runtime": 22.3023,
      "eval_samples_per_second": 118.463,
      "eval_steps_per_second": 1.883,
      "eval_wer": 0.10027766759222531,
      "step": 5100
    },
    {
      "epoch": 45.54,
      "grad_norm": 0.04158908501267433,
      "learning_rate": 2.9470588235294114e-05,
      "loss": 0.0124,
      "step": 5101
    },
    {
      "epoch": 45.55,
      "grad_norm": 0.049603141844272614,
      "learning_rate": 2.941176470588235e-05,
      "loss": 0.0155,
      "step": 5102
    },
    {
      "epoch": 45.56,
      "grad_norm": 0.06320662796497345,
      "learning_rate": 2.9352941176470584e-05,
      "loss": 0.0202,
      "step": 5103
    },
    {
      "epoch": 45.57,
      "grad_norm": 0.04063093662261963,
      "learning_rate": 2.929411764705882e-05,
      "loss": 0.0139,
      "step": 5104
    },
    {
      "epoch": 45.58,
      "grad_norm": 0.038668982684612274,
      "learning_rate": 2.9235294117647057e-05,
      "loss": 0.0148,
      "step": 5105
    },
    {
      "epoch": 45.59,
      "grad_norm": 0.03724049776792526,
      "learning_rate": 2.9176470588235294e-05,
      "loss": 0.0127,
      "step": 5106
    },
    {
      "epoch": 45.6,
      "grad_norm": 0.04132268950343132,
      "learning_rate": 2.9117647058823527e-05,
      "loss": 0.0144,
      "step": 5107
    },
    {
      "epoch": 45.61,
      "grad_norm": 0.05094775930047035,
      "learning_rate": 2.9058823529411764e-05,
      "loss": 0.0143,
      "step": 5108
    },
    {
      "epoch": 45.62,
      "grad_norm": 0.09159460663795471,
      "learning_rate": 2.8999999999999997e-05,
      "loss": 0.0227,
      "step": 5109
    },
    {
      "epoch": 45.62,
      "grad_norm": 0.034520480781793594,
      "learning_rate": 2.894117647058823e-05,
      "loss": 0.0129,
      "step": 5110
    },
    {
      "epoch": 45.63,
      "grad_norm": 0.0377240888774395,
      "learning_rate": 2.8882352941176467e-05,
      "loss": 0.0165,
      "step": 5111
    },
    {
      "epoch": 45.64,
      "grad_norm": 0.040630314499139786,
      "learning_rate": 2.8823529411764703e-05,
      "loss": 0.0158,
      "step": 5112
    },
    {
      "epoch": 45.65,
      "grad_norm": 0.04070843383669853,
      "learning_rate": 2.876470588235294e-05,
      "loss": 0.0131,
      "step": 5113
    },
    {
      "epoch": 45.66,
      "grad_norm": 0.048122283071279526,
      "learning_rate": 2.8705882352941173e-05,
      "loss": 0.0172,
      "step": 5114
    },
    {
      "epoch": 45.67,
      "grad_norm": 0.08413251489400864,
      "learning_rate": 2.864705882352941e-05,
      "loss": 0.022,
      "step": 5115
    },
    {
      "epoch": 45.68,
      "grad_norm": 0.03990272432565689,
      "learning_rate": 2.8588235294117643e-05,
      "loss": 0.0168,
      "step": 5116
    },
    {
      "epoch": 45.69,
      "grad_norm": 0.04196707159280777,
      "learning_rate": 2.852941176470588e-05,
      "loss": 0.0154,
      "step": 5117
    },
    {
      "epoch": 45.7,
      "grad_norm": 0.03930666670203209,
      "learning_rate": 2.8470588235294113e-05,
      "loss": 0.0135,
      "step": 5118
    },
    {
      "epoch": 45.71,
      "grad_norm": 0.040444232523441315,
      "learning_rate": 2.8411764705882353e-05,
      "loss": 0.0146,
      "step": 5119
    },
    {
      "epoch": 45.71,
      "grad_norm": 0.04675314202904701,
      "learning_rate": 2.8352941176470586e-05,
      "loss": 0.0173,
      "step": 5120
    },
    {
      "epoch": 45.72,
      "grad_norm": 0.06005815044045448,
      "learning_rate": 2.829411764705882e-05,
      "loss": 0.0172,
      "step": 5121
    },
    {
      "epoch": 45.73,
      "grad_norm": 0.06058090925216675,
      "learning_rate": 2.8235294117647056e-05,
      "loss": 0.0159,
      "step": 5122
    },
    {
      "epoch": 45.74,
      "grad_norm": 0.04015067592263222,
      "learning_rate": 2.817647058823529e-05,
      "loss": 0.0156,
      "step": 5123
    },
    {
      "epoch": 45.75,
      "grad_norm": 0.04325965791940689,
      "learning_rate": 2.8117647058823526e-05,
      "loss": 0.0181,
      "step": 5124
    },
    {
      "epoch": 45.76,
      "grad_norm": 0.04603307694196701,
      "learning_rate": 2.8058823529411763e-05,
      "loss": 0.0144,
      "step": 5125
    },
    {
      "epoch": 45.77,
      "grad_norm": 0.04270898178219795,
      "learning_rate": 2.8e-05,
      "loss": 0.0151,
      "step": 5126
    },
    {
      "epoch": 45.78,
      "grad_norm": 0.045705217868089676,
      "learning_rate": 2.7941176470588233e-05,
      "loss": 0.0131,
      "step": 5127
    },
    {
      "epoch": 45.79,
      "grad_norm": 0.08076942712068558,
      "learning_rate": 2.788235294117647e-05,
      "loss": 0.0212,
      "step": 5128
    },
    {
      "epoch": 45.79,
      "grad_norm": 0.03954634815454483,
      "learning_rate": 2.7823529411764703e-05,
      "loss": 0.0172,
      "step": 5129
    },
    {
      "epoch": 45.8,
      "grad_norm": 0.0347314290702343,
      "learning_rate": 2.7764705882352936e-05,
      "loss": 0.0141,
      "step": 5130
    },
    {
      "epoch": 45.81,
      "grad_norm": 0.04443887993693352,
      "learning_rate": 2.7705882352941173e-05,
      "loss": 0.0152,
      "step": 5131
    },
    {
      "epoch": 45.82,
      "grad_norm": 0.04813896119594574,
      "learning_rate": 2.7647058823529413e-05,
      "loss": 0.0147,
      "step": 5132
    },
    {
      "epoch": 45.83,
      "grad_norm": 0.04440528526902199,
      "learning_rate": 2.7588235294117646e-05,
      "loss": 0.0146,
      "step": 5133
    },
    {
      "epoch": 45.84,
      "grad_norm": 0.06592968106269836,
      "learning_rate": 2.752941176470588e-05,
      "loss": 0.0177,
      "step": 5134
    },
    {
      "epoch": 45.85,
      "grad_norm": 0.04435674101114273,
      "learning_rate": 2.7470588235294116e-05,
      "loss": 0.0153,
      "step": 5135
    },
    {
      "epoch": 45.86,
      "grad_norm": 0.039653025567531586,
      "learning_rate": 2.741176470588235e-05,
      "loss": 0.0139,
      "step": 5136
    },
    {
      "epoch": 45.87,
      "grad_norm": 0.04258428514003754,
      "learning_rate": 2.7352941176470586e-05,
      "loss": 0.0142,
      "step": 5137
    },
    {
      "epoch": 45.88,
      "grad_norm": 0.04643780365586281,
      "learning_rate": 2.729411764705882e-05,
      "loss": 0.0152,
      "step": 5138
    },
    {
      "epoch": 45.88,
      "grad_norm": 0.04483505338430405,
      "learning_rate": 2.723529411764706e-05,
      "loss": 0.0164,
      "step": 5139
    },
    {
      "epoch": 45.89,
      "grad_norm": 0.06826629489660263,
      "learning_rate": 2.7176470588235292e-05,
      "loss": 0.0158,
      "step": 5140
    },
    {
      "epoch": 45.9,
      "grad_norm": 0.033806659281253815,
      "learning_rate": 2.7117647058823526e-05,
      "loss": 0.0147,
      "step": 5141
    },
    {
      "epoch": 45.91,
      "grad_norm": 0.04065383970737457,
      "learning_rate": 2.7058823529411762e-05,
      "loss": 0.0169,
      "step": 5142
    },
    {
      "epoch": 45.92,
      "grad_norm": 0.04558153077960014,
      "learning_rate": 2.6999999999999996e-05,
      "loss": 0.0182,
      "step": 5143
    },
    {
      "epoch": 45.93,
      "grad_norm": 0.042024046182632446,
      "learning_rate": 2.6941176470588232e-05,
      "loss": 0.0152,
      "step": 5144
    },
    {
      "epoch": 45.94,
      "grad_norm": 0.04814515262842178,
      "learning_rate": 2.688235294117647e-05,
      "loss": 0.0163,
      "step": 5145
    },
    {
      "epoch": 45.95,
      "grad_norm": 0.06419163197278976,
      "learning_rate": 2.6823529411764706e-05,
      "loss": 0.0151,
      "step": 5146
    },
    {
      "epoch": 45.96,
      "grad_norm": 0.08004485070705414,
      "learning_rate": 2.676470588235294e-05,
      "loss": 0.0211,
      "step": 5147
    },
    {
      "epoch": 45.96,
      "grad_norm": 0.0423152782022953,
      "learning_rate": 2.6705882352941175e-05,
      "loss": 0.0177,
      "step": 5148
    },
    {
      "epoch": 45.97,
      "grad_norm": 0.03929867595434189,
      "learning_rate": 2.664705882352941e-05,
      "loss": 0.014,
      "step": 5149
    },
    {
      "epoch": 45.98,
      "grad_norm": 0.04568082094192505,
      "learning_rate": 2.6588235294117642e-05,
      "loss": 0.0157,
      "step": 5150
    },
    {
      "epoch": 45.99,
      "grad_norm": 0.05054107680916786,
      "learning_rate": 2.652941176470588e-05,
      "loss": 0.0148,
      "step": 5151
    },
    {
      "epoch": 46.0,
      "grad_norm": 0.06225099042057991,
      "learning_rate": 2.647058823529412e-05,
      "loss": 0.0181,
      "step": 5152
    },
    {
      "epoch": 46.01,
      "grad_norm": 0.037778183817863464,
      "learning_rate": 2.6411764705882352e-05,
      "loss": 0.0152,
      "step": 5153
    },
    {
      "epoch": 46.02,
      "grad_norm": 0.035917848348617554,
      "learning_rate": 2.6352941176470585e-05,
      "loss": 0.0148,
      "step": 5154
    },
    {
      "epoch": 46.03,
      "grad_norm": 0.03493110090494156,
      "learning_rate": 2.6294117647058822e-05,
      "loss": 0.0131,
      "step": 5155
    },
    {
      "epoch": 46.04,
      "grad_norm": 0.04298359900712967,
      "learning_rate": 2.6235294117647055e-05,
      "loss": 0.0157,
      "step": 5156
    },
    {
      "epoch": 46.04,
      "grad_norm": 0.0488450862467289,
      "learning_rate": 2.6176470588235292e-05,
      "loss": 0.0162,
      "step": 5157
    },
    {
      "epoch": 46.05,
      "grad_norm": 0.0724613144993782,
      "learning_rate": 2.6117647058823525e-05,
      "loss": 0.0169,
      "step": 5158
    },
    {
      "epoch": 46.06,
      "grad_norm": 0.05420735850930214,
      "learning_rate": 2.6058823529411765e-05,
      "loss": 0.0166,
      "step": 5159
    },
    {
      "epoch": 46.07,
      "grad_norm": 0.039159055799245834,
      "learning_rate": 2.6e-05,
      "loss": 0.0176,
      "step": 5160
    },
    {
      "epoch": 46.08,
      "grad_norm": 0.04124021530151367,
      "learning_rate": 2.594117647058823e-05,
      "loss": 0.0169,
      "step": 5161
    },
    {
      "epoch": 46.09,
      "grad_norm": 0.041537050157785416,
      "learning_rate": 2.588235294117647e-05,
      "loss": 0.015,
      "step": 5162
    },
    {
      "epoch": 46.1,
      "grad_norm": 0.042550474405288696,
      "learning_rate": 2.58235294117647e-05,
      "loss": 0.0154,
      "step": 5163
    },
    {
      "epoch": 46.11,
      "grad_norm": 0.05221645161509514,
      "learning_rate": 2.5764705882352938e-05,
      "loss": 0.0127,
      "step": 5164
    },
    {
      "epoch": 46.12,
      "grad_norm": 0.08014420419931412,
      "learning_rate": 2.5705882352941175e-05,
      "loss": 0.0252,
      "step": 5165
    },
    {
      "epoch": 46.12,
      "grad_norm": 0.03902066871523857,
      "learning_rate": 2.564705882352941e-05,
      "loss": 0.0163,
      "step": 5166
    },
    {
      "epoch": 46.13,
      "grad_norm": 0.03881099820137024,
      "learning_rate": 2.5588235294117645e-05,
      "loss": 0.0153,
      "step": 5167
    },
    {
      "epoch": 46.14,
      "grad_norm": 0.03553776443004608,
      "learning_rate": 2.552941176470588e-05,
      "loss": 0.0132,
      "step": 5168
    },
    {
      "epoch": 46.15,
      "grad_norm": 0.042607516050338745,
      "learning_rate": 2.5470588235294115e-05,
      "loss": 0.0135,
      "step": 5169
    },
    {
      "epoch": 46.16,
      "grad_norm": 0.05033285915851593,
      "learning_rate": 2.5411764705882348e-05,
      "loss": 0.013,
      "step": 5170
    },
    {
      "epoch": 46.17,
      "grad_norm": 0.06352284550666809,
      "learning_rate": 2.5352941176470585e-05,
      "loss": 0.0146,
      "step": 5171
    },
    {
      "epoch": 46.18,
      "grad_norm": 0.035775329917669296,
      "learning_rate": 2.5294117647058825e-05,
      "loss": 0.0147,
      "step": 5172
    },
    {
      "epoch": 46.19,
      "grad_norm": 0.04284421727061272,
      "learning_rate": 2.5235294117647058e-05,
      "loss": 0.0138,
      "step": 5173
    },
    {
      "epoch": 46.2,
      "grad_norm": 0.04920913651585579,
      "learning_rate": 2.517647058823529e-05,
      "loss": 0.016,
      "step": 5174
    },
    {
      "epoch": 46.21,
      "grad_norm": 0.046024713665246964,
      "learning_rate": 2.5117647058823528e-05,
      "loss": 0.0166,
      "step": 5175
    },
    {
      "epoch": 46.21,
      "grad_norm": 0.047369956970214844,
      "learning_rate": 2.505882352941176e-05,
      "loss": 0.0148,
      "step": 5176
    },
    {
      "epoch": 46.22,
      "grad_norm": 0.09221918135881424,
      "learning_rate": 2.4999999999999998e-05,
      "loss": 0.0161,
      "step": 5177
    },
    {
      "epoch": 46.23,
      "grad_norm": 0.048006605356931686,
      "learning_rate": 2.494117647058823e-05,
      "loss": 0.0161,
      "step": 5178
    },
    {
      "epoch": 46.24,
      "grad_norm": 0.038232386112213135,
      "learning_rate": 2.488235294117647e-05,
      "loss": 0.0141,
      "step": 5179
    },
    {
      "epoch": 46.25,
      "grad_norm": 0.046028390526771545,
      "learning_rate": 2.4823529411764704e-05,
      "loss": 0.0182,
      "step": 5180
    },
    {
      "epoch": 46.26,
      "grad_norm": 0.039989035576581955,
      "learning_rate": 2.4764705882352938e-05,
      "loss": 0.0137,
      "step": 5181
    },
    {
      "epoch": 46.27,
      "grad_norm": 0.05071651563048363,
      "learning_rate": 2.4705882352941174e-05,
      "loss": 0.0146,
      "step": 5182
    },
    {
      "epoch": 46.28,
      "grad_norm": 0.07601336389780045,
      "learning_rate": 2.4647058823529408e-05,
      "loss": 0.0183,
      "step": 5183
    },
    {
      "epoch": 46.29,
      "grad_norm": 0.10517021268606186,
      "learning_rate": 2.4588235294117644e-05,
      "loss": 0.0192,
      "step": 5184
    },
    {
      "epoch": 46.29,
      "grad_norm": 0.03758545219898224,
      "learning_rate": 2.452941176470588e-05,
      "loss": 0.0151,
      "step": 5185
    },
    {
      "epoch": 46.3,
      "grad_norm": 0.04599851742386818,
      "learning_rate": 2.4470588235294118e-05,
      "loss": 0.0161,
      "step": 5186
    },
    {
      "epoch": 46.31,
      "grad_norm": 0.048986662179231644,
      "learning_rate": 2.441176470588235e-05,
      "loss": 0.0181,
      "step": 5187
    },
    {
      "epoch": 46.32,
      "grad_norm": 0.044982846826314926,
      "learning_rate": 2.4352941176470587e-05,
      "loss": 0.0162,
      "step": 5188
    },
    {
      "epoch": 46.33,
      "grad_norm": 0.05033083260059357,
      "learning_rate": 2.429411764705882e-05,
      "loss": 0.0168,
      "step": 5189
    },
    {
      "epoch": 46.34,
      "grad_norm": 0.07366334646940231,
      "learning_rate": 2.4235294117647054e-05,
      "loss": 0.0224,
      "step": 5190
    },
    {
      "epoch": 46.35,
      "grad_norm": 0.038727469742298126,
      "learning_rate": 2.417647058823529e-05,
      "loss": 0.0176,
      "step": 5191
    },
    {
      "epoch": 46.36,
      "grad_norm": 0.04443450644612312,
      "learning_rate": 2.411764705882353e-05,
      "loss": 0.018,
      "step": 5192
    },
    {
      "epoch": 46.37,
      "grad_norm": 0.03861584886908531,
      "learning_rate": 2.4058823529411764e-05,
      "loss": 0.0179,
      "step": 5193
    },
    {
      "epoch": 46.38,
      "grad_norm": 0.04566491022706032,
      "learning_rate": 2.3999999999999997e-05,
      "loss": 0.0154,
      "step": 5194
    },
    {
      "epoch": 46.38,
      "grad_norm": 0.052542828023433685,
      "learning_rate": 2.3941176470588234e-05,
      "loss": 0.0164,
      "step": 5195
    },
    {
      "epoch": 46.39,
      "grad_norm": 0.07896377891302109,
      "learning_rate": 2.3882352941176467e-05,
      "loss": 0.0181,
      "step": 5196
    },
    {
      "epoch": 46.4,
      "grad_norm": 0.040415626019239426,
      "learning_rate": 2.3823529411764704e-05,
      "loss": 0.0175,
      "step": 5197
    },
    {
      "epoch": 46.41,
      "grad_norm": 0.040500037372112274,
      "learning_rate": 2.3764705882352937e-05,
      "loss": 0.0162,
      "step": 5198
    },
    {
      "epoch": 46.42,
      "grad_norm": 0.040322475135326385,
      "learning_rate": 2.3705882352941177e-05,
      "loss": 0.0212,
      "step": 5199
    },
    {
      "epoch": 46.43,
      "grad_norm": 0.04169744998216629,
      "learning_rate": 2.364705882352941e-05,
      "loss": 0.0126,
      "step": 5200
    },
    {
      "epoch": 46.43,
      "eval_cer": 0.031855919240208516,
      "eval_loss": 0.20860451459884644,
      "eval_runtime": 22.948,
      "eval_samples_per_second": 115.13,
      "eval_steps_per_second": 1.83,
      "eval_wer": 0.09986116620388735,
      "step": 5200
    },
    {
      "epoch": 46.44,
      "grad_norm": 0.04425807297229767,
      "learning_rate": 2.3588235294117644e-05,
      "loss": 0.0138,
      "step": 5201
    },
    {
      "epoch": 46.45,
      "grad_norm": 0.07776098698377609,
      "learning_rate": 2.352941176470588e-05,
      "loss": 0.0187,
      "step": 5202
    },
    {
      "epoch": 46.46,
      "grad_norm": 0.045044541358947754,
      "learning_rate": 2.3470588235294114e-05,
      "loss": 0.0165,
      "step": 5203
    },
    {
      "epoch": 46.46,
      "grad_norm": 0.03875579312443733,
      "learning_rate": 2.341176470588235e-05,
      "loss": 0.016,
      "step": 5204
    },
    {
      "epoch": 46.47,
      "grad_norm": 0.04249643534421921,
      "learning_rate": 2.3352941176470587e-05,
      "loss": 0.0161,
      "step": 5205
    },
    {
      "epoch": 46.48,
      "grad_norm": 0.044398561120033264,
      "learning_rate": 2.3294117647058824e-05,
      "loss": 0.0141,
      "step": 5206
    },
    {
      "epoch": 46.49,
      "grad_norm": 0.04499765485525131,
      "learning_rate": 2.3235294117647057e-05,
      "loss": 0.0168,
      "step": 5207
    },
    {
      "epoch": 46.5,
      "grad_norm": 0.054214634001255035,
      "learning_rate": 2.3176470588235293e-05,
      "loss": 0.0179,
      "step": 5208
    },
    {
      "epoch": 46.51,
      "grad_norm": 0.06382989138364792,
      "learning_rate": 2.3117647058823527e-05,
      "loss": 0.0171,
      "step": 5209
    },
    {
      "epoch": 46.52,
      "grad_norm": 0.041569650173187256,
      "learning_rate": 2.305882352941176e-05,
      "loss": 0.0162,
      "step": 5210
    },
    {
      "epoch": 46.53,
      "grad_norm": 0.04001566022634506,
      "learning_rate": 2.2999999999999997e-05,
      "loss": 0.0159,
      "step": 5211
    },
    {
      "epoch": 46.54,
      "grad_norm": 0.03527286276221275,
      "learning_rate": 2.2941176470588237e-05,
      "loss": 0.0127,
      "step": 5212
    },
    {
      "epoch": 46.54,
      "grad_norm": 0.04458502307534218,
      "learning_rate": 2.288235294117647e-05,
      "loss": 0.0135,
      "step": 5213
    },
    {
      "epoch": 46.55,
      "grad_norm": 0.04917190596461296,
      "learning_rate": 2.2823529411764703e-05,
      "loss": 0.0151,
      "step": 5214
    },
    {
      "epoch": 46.56,
      "grad_norm": 0.0713915154337883,
      "learning_rate": 2.276470588235294e-05,
      "loss": 0.0164,
      "step": 5215
    },
    {
      "epoch": 46.57,
      "grad_norm": 0.038980573415756226,
      "learning_rate": 2.2705882352941173e-05,
      "loss": 0.016,
      "step": 5216
    },
    {
      "epoch": 46.58,
      "grad_norm": 0.03902507573366165,
      "learning_rate": 2.264705882352941e-05,
      "loss": 0.0142,
      "step": 5217
    },
    {
      "epoch": 46.59,
      "grad_norm": 0.04390474408864975,
      "learning_rate": 2.2588235294117643e-05,
      "loss": 0.016,
      "step": 5218
    },
    {
      "epoch": 46.6,
      "grad_norm": 0.04564650356769562,
      "learning_rate": 2.2529411764705883e-05,
      "loss": 0.0168,
      "step": 5219
    },
    {
      "epoch": 46.61,
      "grad_norm": 0.06496992707252502,
      "learning_rate": 2.2470588235294116e-05,
      "loss": 0.0173,
      "step": 5220
    },
    {
      "epoch": 46.62,
      "grad_norm": 0.13928504288196564,
      "learning_rate": 2.241176470588235e-05,
      "loss": 0.0202,
      "step": 5221
    },
    {
      "epoch": 46.62,
      "grad_norm": 0.03780483081936836,
      "learning_rate": 2.2352941176470586e-05,
      "loss": 0.0163,
      "step": 5222
    },
    {
      "epoch": 46.63,
      "grad_norm": 0.03962361067533493,
      "learning_rate": 2.229411764705882e-05,
      "loss": 0.0153,
      "step": 5223
    },
    {
      "epoch": 46.64,
      "grad_norm": 0.04358210042119026,
      "learning_rate": 2.2235294117647056e-05,
      "loss": 0.0145,
      "step": 5224
    },
    {
      "epoch": 46.65,
      "grad_norm": 0.04540077596902847,
      "learning_rate": 2.2176470588235293e-05,
      "loss": 0.016,
      "step": 5225
    },
    {
      "epoch": 46.66,
      "grad_norm": 0.04471196234226227,
      "learning_rate": 2.211764705882353e-05,
      "loss": 0.0133,
      "step": 5226
    },
    {
      "epoch": 46.67,
      "grad_norm": 0.10236772894859314,
      "learning_rate": 2.2058823529411763e-05,
      "loss": 0.0218,
      "step": 5227
    },
    {
      "epoch": 46.68,
      "grad_norm": 0.039916884154081345,
      "learning_rate": 2.2e-05,
      "loss": 0.0169,
      "step": 5228
    },
    {
      "epoch": 46.69,
      "grad_norm": 0.034644100815057755,
      "learning_rate": 2.1941176470588233e-05,
      "loss": 0.0129,
      "step": 5229
    },
    {
      "epoch": 46.7,
      "grad_norm": 0.044551361352205276,
      "learning_rate": 2.1882352941176466e-05,
      "loss": 0.0164,
      "step": 5230
    },
    {
      "epoch": 46.71,
      "grad_norm": 0.040567588061094284,
      "learning_rate": 2.1823529411764703e-05,
      "loss": 0.0146,
      "step": 5231
    },
    {
      "epoch": 46.71,
      "grad_norm": 0.046880677342414856,
      "learning_rate": 2.1764705882352943e-05,
      "loss": 0.017,
      "step": 5232
    },
    {
      "epoch": 46.72,
      "grad_norm": 0.08010052144527435,
      "learning_rate": 2.1705882352941176e-05,
      "loss": 0.0192,
      "step": 5233
    },
    {
      "epoch": 46.73,
      "grad_norm": 0.13500761985778809,
      "learning_rate": 2.164705882352941e-05,
      "loss": 0.0183,
      "step": 5234
    },
    {
      "epoch": 46.74,
      "grad_norm": 0.03708220645785332,
      "learning_rate": 2.1588235294117646e-05,
      "loss": 0.0165,
      "step": 5235
    },
    {
      "epoch": 46.75,
      "grad_norm": 0.03563365340232849,
      "learning_rate": 2.152941176470588e-05,
      "loss": 0.0154,
      "step": 5236
    },
    {
      "epoch": 46.76,
      "grad_norm": 0.04525330662727356,
      "learning_rate": 2.1470588235294116e-05,
      "loss": 0.0159,
      "step": 5237
    },
    {
      "epoch": 46.77,
      "grad_norm": 0.0355866402387619,
      "learning_rate": 2.141176470588235e-05,
      "loss": 0.0126,
      "step": 5238
    },
    {
      "epoch": 46.78,
      "grad_norm": 0.046234723180532455,
      "learning_rate": 2.135294117647059e-05,
      "loss": 0.0143,
      "step": 5239
    },
    {
      "epoch": 46.79,
      "grad_norm": 0.06242381036281586,
      "learning_rate": 2.1294117647058822e-05,
      "loss": 0.0146,
      "step": 5240
    },
    {
      "epoch": 46.79,
      "grad_norm": 0.037977006286382675,
      "learning_rate": 2.1235294117647056e-05,
      "loss": 0.016,
      "step": 5241
    },
    {
      "epoch": 46.8,
      "grad_norm": 0.03535516560077667,
      "learning_rate": 2.1176470588235292e-05,
      "loss": 0.0137,
      "step": 5242
    },
    {
      "epoch": 46.81,
      "grad_norm": 0.039125002920627594,
      "learning_rate": 2.1117647058823526e-05,
      "loss": 0.0144,
      "step": 5243
    },
    {
      "epoch": 46.82,
      "grad_norm": 0.04807785898447037,
      "learning_rate": 2.1058823529411762e-05,
      "loss": 0.0156,
      "step": 5244
    },
    {
      "epoch": 46.83,
      "grad_norm": 0.06447841972112656,
      "learning_rate": 2.1e-05,
      "loss": 0.0191,
      "step": 5245
    },
    {
      "epoch": 46.84,
      "grad_norm": 0.0857747420668602,
      "learning_rate": 2.0941176470588235e-05,
      "loss": 0.0244,
      "step": 5246
    },
    {
      "epoch": 46.85,
      "grad_norm": 0.038232676684856415,
      "learning_rate": 2.088235294117647e-05,
      "loss": 0.0158,
      "step": 5247
    },
    {
      "epoch": 46.86,
      "grad_norm": 0.040532663464546204,
      "learning_rate": 2.0823529411764705e-05,
      "loss": 0.0155,
      "step": 5248
    },
    {
      "epoch": 46.87,
      "grad_norm": 0.04061966761946678,
      "learning_rate": 2.076470588235294e-05,
      "loss": 0.0139,
      "step": 5249
    },
    {
      "epoch": 46.88,
      "grad_norm": 0.037793710827827454,
      "learning_rate": 2.0705882352941172e-05,
      "loss": 0.0134,
      "step": 5250
    },
    {
      "epoch": 46.88,
      "grad_norm": 0.037497226148843765,
      "learning_rate": 2.064705882352941e-05,
      "loss": 0.0113,
      "step": 5251
    },
    {
      "epoch": 46.89,
      "grad_norm": 0.09749874472618103,
      "learning_rate": 2.058823529411765e-05,
      "loss": 0.0211,
      "step": 5252
    },
    {
      "epoch": 46.9,
      "grad_norm": 0.03333595395088196,
      "learning_rate": 2.0529411764705882e-05,
      "loss": 0.0137,
      "step": 5253
    },
    {
      "epoch": 46.91,
      "grad_norm": 0.0389777310192585,
      "learning_rate": 2.0470588235294115e-05,
      "loss": 0.0166,
      "step": 5254
    },
    {
      "epoch": 46.92,
      "grad_norm": 0.038943417370319366,
      "learning_rate": 2.0411764705882352e-05,
      "loss": 0.0163,
      "step": 5255
    },
    {
      "epoch": 46.93,
      "grad_norm": 0.04082312434911728,
      "learning_rate": 2.0352941176470585e-05,
      "loss": 0.0156,
      "step": 5256
    },
    {
      "epoch": 46.94,
      "grad_norm": 0.03889601305127144,
      "learning_rate": 2.0294117647058822e-05,
      "loss": 0.0164,
      "step": 5257
    },
    {
      "epoch": 46.95,
      "grad_norm": 0.048415396362543106,
      "learning_rate": 2.0235294117647055e-05,
      "loss": 0.0168,
      "step": 5258
    },
    {
      "epoch": 46.96,
      "grad_norm": 0.05437806248664856,
      "learning_rate": 2.0176470588235295e-05,
      "loss": 0.0176,
      "step": 5259
    },
    {
      "epoch": 46.96,
      "grad_norm": 0.03630838543176651,
      "learning_rate": 2.011764705882353e-05,
      "loss": 0.0155,
      "step": 5260
    },
    {
      "epoch": 46.97,
      "grad_norm": 0.04185402765870094,
      "learning_rate": 2.0058823529411765e-05,
      "loss": 0.015,
      "step": 5261
    },
    {
      "epoch": 46.98,
      "grad_norm": 0.03746276721358299,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 0.0154,
      "step": 5262
    },
    {
      "epoch": 46.99,
      "grad_norm": 0.04353839159011841,
      "learning_rate": 1.994117647058823e-05,
      "loss": 0.013,
      "step": 5263
    },
    {
      "epoch": 47.0,
      "grad_norm": 0.07891372591257095,
      "learning_rate": 1.9882352941176468e-05,
      "loss": 0.0154,
      "step": 5264
    },
    {
      "epoch": 47.01,
      "grad_norm": 0.0363650768995285,
      "learning_rate": 1.9823529411764705e-05,
      "loss": 0.0166,
      "step": 5265
    },
    {
      "epoch": 47.02,
      "grad_norm": 0.03396986052393913,
      "learning_rate": 1.976470588235294e-05,
      "loss": 0.0143,
      "step": 5266
    },
    {
      "epoch": 47.03,
      "grad_norm": 0.038241930305957794,
      "learning_rate": 1.9705882352941175e-05,
      "loss": 0.0159,
      "step": 5267
    },
    {
      "epoch": 47.04,
      "grad_norm": 0.03874950110912323,
      "learning_rate": 1.964705882352941e-05,
      "loss": 0.0147,
      "step": 5268
    },
    {
      "epoch": 47.04,
      "grad_norm": 0.04823707416653633,
      "learning_rate": 1.9588235294117645e-05,
      "loss": 0.0169,
      "step": 5269
    },
    {
      "epoch": 47.05,
      "grad_norm": 0.05711193010210991,
      "learning_rate": 1.9529411764705878e-05,
      "loss": 0.0155,
      "step": 5270
    },
    {
      "epoch": 47.06,
      "grad_norm": 0.04007866233587265,
      "learning_rate": 1.9470588235294115e-05,
      "loss": 0.0119,
      "step": 5271
    },
    {
      "epoch": 47.07,
      "grad_norm": 0.03977508470416069,
      "learning_rate": 1.9411764705882355e-05,
      "loss": 0.0154,
      "step": 5272
    },
    {
      "epoch": 47.08,
      "grad_norm": 0.038793738931417465,
      "learning_rate": 1.9352941176470588e-05,
      "loss": 0.0157,
      "step": 5273
    },
    {
      "epoch": 47.09,
      "grad_norm": 0.04206732288002968,
      "learning_rate": 1.929411764705882e-05,
      "loss": 0.0124,
      "step": 5274
    },
    {
      "epoch": 47.1,
      "grad_norm": 0.041772741824388504,
      "learning_rate": 1.9235294117647058e-05,
      "loss": 0.0145,
      "step": 5275
    },
    {
      "epoch": 47.11,
      "grad_norm": 0.05594871938228607,
      "learning_rate": 1.917647058823529e-05,
      "loss": 0.0165,
      "step": 5276
    },
    {
      "epoch": 47.12,
      "grad_norm": 0.08875638246536255,
      "learning_rate": 1.9117647058823528e-05,
      "loss": 0.0207,
      "step": 5277
    },
    {
      "epoch": 47.12,
      "grad_norm": 0.03953978419303894,
      "learning_rate": 1.905882352941176e-05,
      "loss": 0.0151,
      "step": 5278
    },
    {
      "epoch": 47.13,
      "grad_norm": 0.049001071602106094,
      "learning_rate": 1.9e-05,
      "loss": 0.0145,
      "step": 5279
    },
    {
      "epoch": 47.14,
      "grad_norm": 0.04553956538438797,
      "learning_rate": 1.8941176470588234e-05,
      "loss": 0.016,
      "step": 5280
    },
    {
      "epoch": 47.15,
      "grad_norm": 0.04373745247721672,
      "learning_rate": 1.888235294117647e-05,
      "loss": 0.0159,
      "step": 5281
    },
    {
      "epoch": 47.16,
      "grad_norm": 0.051922187209129333,
      "learning_rate": 1.8823529411764704e-05,
      "loss": 0.0155,
      "step": 5282
    },
    {
      "epoch": 47.17,
      "grad_norm": 0.0890134647488594,
      "learning_rate": 1.8764705882352937e-05,
      "loss": 0.0161,
      "step": 5283
    },
    {
      "epoch": 47.18,
      "grad_norm": 0.040788739919662476,
      "learning_rate": 1.8705882352941174e-05,
      "loss": 0.0136,
      "step": 5284
    },
    {
      "epoch": 47.19,
      "grad_norm": 0.037018533796072006,
      "learning_rate": 1.864705882352941e-05,
      "loss": 0.0144,
      "step": 5285
    },
    {
      "epoch": 47.2,
      "grad_norm": 0.04045609012246132,
      "learning_rate": 1.8588235294117644e-05,
      "loss": 0.0127,
      "step": 5286
    },
    {
      "epoch": 47.21,
      "grad_norm": 0.04248913750052452,
      "learning_rate": 1.852941176470588e-05,
      "loss": 0.014,
      "step": 5287
    },
    {
      "epoch": 47.21,
      "grad_norm": 0.04623138904571533,
      "learning_rate": 1.8470588235294117e-05,
      "loss": 0.0155,
      "step": 5288
    },
    {
      "epoch": 47.22,
      "grad_norm": 0.08569624274969101,
      "learning_rate": 1.841176470588235e-05,
      "loss": 0.0138,
      "step": 5289
    },
    {
      "epoch": 47.23,
      "grad_norm": 0.04352473095059395,
      "learning_rate": 1.8352941176470587e-05,
      "loss": 0.0181,
      "step": 5290
    },
    {
      "epoch": 47.24,
      "grad_norm": 0.03902510926127434,
      "learning_rate": 1.8294117647058824e-05,
      "loss": 0.0145,
      "step": 5291
    },
    {
      "epoch": 47.25,
      "grad_norm": 0.04916934296488762,
      "learning_rate": 1.8235294117647057e-05,
      "loss": 0.0175,
      "step": 5292
    },
    {
      "epoch": 47.26,
      "grad_norm": 0.04030926525592804,
      "learning_rate": 1.817647058823529e-05,
      "loss": 0.0138,
      "step": 5293
    },
    {
      "epoch": 47.27,
      "grad_norm": 0.03972399979829788,
      "learning_rate": 1.8117647058823527e-05,
      "loss": 0.0166,
      "step": 5294
    },
    {
      "epoch": 47.28,
      "grad_norm": 0.06011297181248665,
      "learning_rate": 1.8058823529411764e-05,
      "loss": 0.0174,
      "step": 5295
    },
    {
      "epoch": 47.29,
      "grad_norm": 0.1028251126408577,
      "learning_rate": 1.7999999999999997e-05,
      "loss": 0.0202,
      "step": 5296
    },
    {
      "epoch": 47.29,
      "grad_norm": 0.04311250522732735,
      "learning_rate": 1.7941176470588234e-05,
      "loss": 0.0158,
      "step": 5297
    },
    {
      "epoch": 47.3,
      "grad_norm": 0.03926079720258713,
      "learning_rate": 1.788235294117647e-05,
      "loss": 0.0144,
      "step": 5298
    },
    {
      "epoch": 47.31,
      "grad_norm": 0.03284962475299835,
      "learning_rate": 1.7823529411764704e-05,
      "loss": 0.0129,
      "step": 5299
    },
    {
      "epoch": 47.32,
      "grad_norm": 0.04275604709982872,
      "learning_rate": 1.776470588235294e-05,
      "loss": 0.0153,
      "step": 5300
    },
    {
      "epoch": 47.32,
      "eval_cer": 0.03095015277762118,
      "eval_loss": 0.2094593048095703,
      "eval_runtime": 22.6317,
      "eval_samples_per_second": 116.739,
      "eval_steps_per_second": 1.856,
      "eval_wer": 0.09813566045220151,
      "step": 5300
    },
    {
      "epoch": 47.33,
      "grad_norm": 0.06282328814268112,
      "learning_rate": 1.7705882352941177e-05,
      "loss": 0.0175,
      "step": 5301
    },
    {
      "epoch": 47.34,
      "grad_norm": 0.055917032063007355,
      "learning_rate": 1.764705882352941e-05,
      "loss": 0.0158,
      "step": 5302
    },
    {
      "epoch": 47.35,
      "grad_norm": 0.037404872477054596,
      "learning_rate": 1.7588235294117643e-05,
      "loss": 0.0141,
      "step": 5303
    },
    {
      "epoch": 47.36,
      "grad_norm": 0.03317873552441597,
      "learning_rate": 1.752941176470588e-05,
      "loss": 0.0151,
      "step": 5304
    },
    {
      "epoch": 47.37,
      "grad_norm": 0.036500830203294754,
      "learning_rate": 1.7470588235294117e-05,
      "loss": 0.0139,
      "step": 5305
    },
    {
      "epoch": 47.38,
      "grad_norm": 0.041090648621320724,
      "learning_rate": 1.741176470588235e-05,
      "loss": 0.0165,
      "step": 5306
    },
    {
      "epoch": 47.38,
      "grad_norm": 0.06290547549724579,
      "learning_rate": 1.7352941176470587e-05,
      "loss": 0.0151,
      "step": 5307
    },
    {
      "epoch": 47.39,
      "grad_norm": 0.07896995544433594,
      "learning_rate": 1.7294117647058823e-05,
      "loss": 0.0195,
      "step": 5308
    },
    {
      "epoch": 47.4,
      "grad_norm": 0.03444727882742882,
      "learning_rate": 1.7235294117647057e-05,
      "loss": 0.0147,
      "step": 5309
    },
    {
      "epoch": 47.41,
      "grad_norm": 0.03905628249049187,
      "learning_rate": 1.7176470588235293e-05,
      "loss": 0.0149,
      "step": 5310
    },
    {
      "epoch": 47.42,
      "grad_norm": 0.035952065140008926,
      "learning_rate": 1.711764705882353e-05,
      "loss": 0.0123,
      "step": 5311
    },
    {
      "epoch": 47.43,
      "grad_norm": 0.041040096431970596,
      "learning_rate": 1.7058823529411763e-05,
      "loss": 0.0141,
      "step": 5312
    },
    {
      "epoch": 47.44,
      "grad_norm": 0.04559514299035072,
      "learning_rate": 1.6999999999999996e-05,
      "loss": 0.0163,
      "step": 5313
    },
    {
      "epoch": 47.45,
      "grad_norm": 0.07138743251562119,
      "learning_rate": 1.6941176470588233e-05,
      "loss": 0.0159,
      "step": 5314
    },
    {
      "epoch": 47.46,
      "grad_norm": 0.03392981365323067,
      "learning_rate": 1.688235294117647e-05,
      "loss": 0.0153,
      "step": 5315
    },
    {
      "epoch": 47.46,
      "grad_norm": 0.03276912122964859,
      "learning_rate": 1.6823529411764703e-05,
      "loss": 0.013,
      "step": 5316
    },
    {
      "epoch": 47.47,
      "grad_norm": 0.041713979095220566,
      "learning_rate": 1.676470588235294e-05,
      "loss": 0.0157,
      "step": 5317
    },
    {
      "epoch": 47.48,
      "grad_norm": 0.04545548930764198,
      "learning_rate": 1.6705882352941176e-05,
      "loss": 0.0174,
      "step": 5318
    },
    {
      "epoch": 47.49,
      "grad_norm": 0.044357605278491974,
      "learning_rate": 1.664705882352941e-05,
      "loss": 0.0149,
      "step": 5319
    },
    {
      "epoch": 47.5,
      "grad_norm": 0.06216815114021301,
      "learning_rate": 1.6588235294117646e-05,
      "loss": 0.0126,
      "step": 5320
    },
    {
      "epoch": 47.51,
      "grad_norm": 0.0639604777097702,
      "learning_rate": 1.6529411764705883e-05,
      "loss": 0.0159,
      "step": 5321
    },
    {
      "epoch": 47.52,
      "grad_norm": 0.04166419431567192,
      "learning_rate": 1.6470588235294116e-05,
      "loss": 0.0158,
      "step": 5322
    },
    {
      "epoch": 47.53,
      "grad_norm": 0.040787551552057266,
      "learning_rate": 1.641176470588235e-05,
      "loss": 0.016,
      "step": 5323
    },
    {
      "epoch": 47.54,
      "grad_norm": 0.037571802735328674,
      "learning_rate": 1.6352941176470586e-05,
      "loss": 0.0143,
      "step": 5324
    },
    {
      "epoch": 47.54,
      "grad_norm": 0.03916793689131737,
      "learning_rate": 1.6294117647058823e-05,
      "loss": 0.0159,
      "step": 5325
    },
    {
      "epoch": 47.55,
      "grad_norm": 0.05931222438812256,
      "learning_rate": 1.6235294117647056e-05,
      "loss": 0.0173,
      "step": 5326
    },
    {
      "epoch": 47.56,
      "grad_norm": 0.05944420397281647,
      "learning_rate": 1.6176470588235293e-05,
      "loss": 0.0205,
      "step": 5327
    },
    {
      "epoch": 47.57,
      "grad_norm": 0.043424081057310104,
      "learning_rate": 1.611764705882353e-05,
      "loss": 0.0143,
      "step": 5328
    },
    {
      "epoch": 47.58,
      "grad_norm": 0.040597956627607346,
      "learning_rate": 1.6058823529411763e-05,
      "loss": 0.0174,
      "step": 5329
    },
    {
      "epoch": 47.59,
      "grad_norm": 0.0380982905626297,
      "learning_rate": 1.6e-05,
      "loss": 0.0129,
      "step": 5330
    },
    {
      "epoch": 47.6,
      "grad_norm": 0.04688919708132744,
      "learning_rate": 1.5941176470588236e-05,
      "loss": 0.0174,
      "step": 5331
    },
    {
      "epoch": 47.61,
      "grad_norm": 0.04591430351138115,
      "learning_rate": 1.588235294117647e-05,
      "loss": 0.0142,
      "step": 5332
    },
    {
      "epoch": 47.62,
      "grad_norm": 0.07394150644540787,
      "learning_rate": 1.5823529411764702e-05,
      "loss": 0.0167,
      "step": 5333
    },
    {
      "epoch": 47.62,
      "grad_norm": 0.0348970890045166,
      "learning_rate": 1.576470588235294e-05,
      "loss": 0.0155,
      "step": 5334
    },
    {
      "epoch": 47.63,
      "grad_norm": 0.03675265982747078,
      "learning_rate": 1.5705882352941176e-05,
      "loss": 0.0138,
      "step": 5335
    },
    {
      "epoch": 47.64,
      "grad_norm": 0.04313450679183006,
      "learning_rate": 1.564705882352941e-05,
      "loss": 0.016,
      "step": 5336
    },
    {
      "epoch": 47.65,
      "grad_norm": 0.035124484449625015,
      "learning_rate": 1.5588235294117646e-05,
      "loss": 0.0127,
      "step": 5337
    },
    {
      "epoch": 47.66,
      "grad_norm": 0.05071822553873062,
      "learning_rate": 1.5529411764705882e-05,
      "loss": 0.0154,
      "step": 5338
    },
    {
      "epoch": 47.67,
      "grad_norm": 0.06335127353668213,
      "learning_rate": 1.5470588235294116e-05,
      "loss": 0.0128,
      "step": 5339
    },
    {
      "epoch": 47.68,
      "grad_norm": 0.04092599079012871,
      "learning_rate": 1.5411764705882352e-05,
      "loss": 0.0171,
      "step": 5340
    },
    {
      "epoch": 47.69,
      "grad_norm": 0.049333348870277405,
      "learning_rate": 1.535294117647059e-05,
      "loss": 0.0187,
      "step": 5341
    },
    {
      "epoch": 47.7,
      "grad_norm": 0.0430619940161705,
      "learning_rate": 1.5294117647058822e-05,
      "loss": 0.0155,
      "step": 5342
    },
    {
      "epoch": 47.71,
      "grad_norm": 0.03829503059387207,
      "learning_rate": 1.5235294117647057e-05,
      "loss": 0.014,
      "step": 5343
    },
    {
      "epoch": 47.71,
      "grad_norm": 0.05448585003614426,
      "learning_rate": 1.5176470588235294e-05,
      "loss": 0.0169,
      "step": 5344
    },
    {
      "epoch": 47.72,
      "grad_norm": 0.07088486850261688,
      "learning_rate": 1.5117647058823529e-05,
      "loss": 0.0209,
      "step": 5345
    },
    {
      "epoch": 47.73,
      "grad_norm": 0.06297363340854645,
      "learning_rate": 1.5058823529411762e-05,
      "loss": 0.0175,
      "step": 5346
    },
    {
      "epoch": 47.74,
      "grad_norm": 0.04363665357232094,
      "learning_rate": 1.4999999999999999e-05,
      "loss": 0.0179,
      "step": 5347
    },
    {
      "epoch": 47.75,
      "grad_norm": 0.04440702870488167,
      "learning_rate": 1.4941176470588234e-05,
      "loss": 0.0176,
      "step": 5348
    },
    {
      "epoch": 47.76,
      "grad_norm": 0.03349072486162186,
      "learning_rate": 1.4882352941176469e-05,
      "loss": 0.0134,
      "step": 5349
    },
    {
      "epoch": 47.77,
      "grad_norm": 0.04793049022555351,
      "learning_rate": 1.4823529411764704e-05,
      "loss": 0.0153,
      "step": 5350
    },
    {
      "epoch": 47.78,
      "grad_norm": 0.05858717858791351,
      "learning_rate": 1.476470588235294e-05,
      "loss": 0.0132,
      "step": 5351
    },
    {
      "epoch": 47.79,
      "grad_norm": 0.05767468735575676,
      "learning_rate": 1.4705882352941175e-05,
      "loss": 0.014,
      "step": 5352
    },
    {
      "epoch": 47.79,
      "grad_norm": 0.03719104826450348,
      "learning_rate": 1.464705882352941e-05,
      "loss": 0.0159,
      "step": 5353
    },
    {
      "epoch": 47.8,
      "grad_norm": 0.0334928072988987,
      "learning_rate": 1.4588235294117647e-05,
      "loss": 0.0145,
      "step": 5354
    },
    {
      "epoch": 47.81,
      "grad_norm": 0.0341482050716877,
      "learning_rate": 1.4529411764705882e-05,
      "loss": 0.0143,
      "step": 5355
    },
    {
      "epoch": 47.82,
      "grad_norm": 0.037394795566797256,
      "learning_rate": 1.4470588235294115e-05,
      "loss": 0.0141,
      "step": 5356
    },
    {
      "epoch": 47.83,
      "grad_norm": 0.0508609376847744,
      "learning_rate": 1.4411764705882352e-05,
      "loss": 0.0129,
      "step": 5357
    },
    {
      "epoch": 47.84,
      "grad_norm": 0.12816482782363892,
      "learning_rate": 1.4352941176470587e-05,
      "loss": 0.0277,
      "step": 5358
    },
    {
      "epoch": 47.85,
      "grad_norm": 0.045090626925230026,
      "learning_rate": 1.4294117647058822e-05,
      "loss": 0.0171,
      "step": 5359
    },
    {
      "epoch": 47.86,
      "grad_norm": 0.03483239561319351,
      "learning_rate": 1.4235294117647057e-05,
      "loss": 0.0129,
      "step": 5360
    },
    {
      "epoch": 47.87,
      "grad_norm": 0.039795368909835815,
      "learning_rate": 1.4176470588235293e-05,
      "loss": 0.0159,
      "step": 5361
    },
    {
      "epoch": 47.88,
      "grad_norm": 0.035538945347070694,
      "learning_rate": 1.4117647058823528e-05,
      "loss": 0.0132,
      "step": 5362
    },
    {
      "epoch": 47.88,
      "grad_norm": 0.047424495220184326,
      "learning_rate": 1.4058823529411763e-05,
      "loss": 0.0154,
      "step": 5363
    },
    {
      "epoch": 47.89,
      "grad_norm": 0.0885155200958252,
      "learning_rate": 1.4e-05,
      "loss": 0.0154,
      "step": 5364
    },
    {
      "epoch": 47.9,
      "grad_norm": 0.03710082918405533,
      "learning_rate": 1.3941176470588235e-05,
      "loss": 0.0155,
      "step": 5365
    },
    {
      "epoch": 47.91,
      "grad_norm": 0.03797599673271179,
      "learning_rate": 1.3882352941176468e-05,
      "loss": 0.0148,
      "step": 5366
    },
    {
      "epoch": 47.92,
      "grad_norm": 0.03618619963526726,
      "learning_rate": 1.3823529411764706e-05,
      "loss": 0.0116,
      "step": 5367
    },
    {
      "epoch": 47.93,
      "grad_norm": 0.04147461801767349,
      "learning_rate": 1.376470588235294e-05,
      "loss": 0.0126,
      "step": 5368
    },
    {
      "epoch": 47.94,
      "grad_norm": 0.053551096469163895,
      "learning_rate": 1.3705882352941175e-05,
      "loss": 0.0166,
      "step": 5369
    },
    {
      "epoch": 47.95,
      "grad_norm": 0.059746768325567245,
      "learning_rate": 1.364705882352941e-05,
      "loss": 0.0179,
      "step": 5370
    },
    {
      "epoch": 47.96,
      "grad_norm": 0.13540853559970856,
      "learning_rate": 1.3588235294117646e-05,
      "loss": 0.0184,
      "step": 5371
    },
    {
      "epoch": 47.96,
      "grad_norm": 0.04125310480594635,
      "learning_rate": 1.3529411764705881e-05,
      "loss": 0.0155,
      "step": 5372
    },
    {
      "epoch": 47.97,
      "grad_norm": 0.035127729177474976,
      "learning_rate": 1.3470588235294116e-05,
      "loss": 0.0135,
      "step": 5373
    },
    {
      "epoch": 47.98,
      "grad_norm": 0.043600305914878845,
      "learning_rate": 1.3411764705882353e-05,
      "loss": 0.0155,
      "step": 5374
    },
    {
      "epoch": 47.99,
      "grad_norm": 0.05032142624258995,
      "learning_rate": 1.3352941176470588e-05,
      "loss": 0.0171,
      "step": 5375
    },
    {
      "epoch": 48.0,
      "grad_norm": 0.06508412957191467,
      "learning_rate": 1.3294117647058821e-05,
      "loss": 0.0156,
      "step": 5376
    },
    {
      "epoch": 48.01,
      "grad_norm": 0.036650948226451874,
      "learning_rate": 1.323529411764706e-05,
      "loss": 0.0155,
      "step": 5377
    },
    {
      "epoch": 48.02,
      "grad_norm": 0.04516515135765076,
      "learning_rate": 1.3176470588235293e-05,
      "loss": 0.0149,
      "step": 5378
    },
    {
      "epoch": 48.03,
      "grad_norm": 0.035986579954624176,
      "learning_rate": 1.3117647058823528e-05,
      "loss": 0.0134,
      "step": 5379
    },
    {
      "epoch": 48.04,
      "grad_norm": 0.038538608700037,
      "learning_rate": 1.3058823529411763e-05,
      "loss": 0.0133,
      "step": 5380
    },
    {
      "epoch": 48.04,
      "grad_norm": 0.04843611642718315,
      "learning_rate": 1.3e-05,
      "loss": 0.0194,
      "step": 5381
    },
    {
      "epoch": 48.05,
      "grad_norm": 0.05646837130188942,
      "learning_rate": 1.2941176470588234e-05,
      "loss": 0.0204,
      "step": 5382
    },
    {
      "epoch": 48.06,
      "grad_norm": 0.07528936862945557,
      "learning_rate": 1.2882352941176469e-05,
      "loss": 0.0188,
      "step": 5383
    },
    {
      "epoch": 48.07,
      "grad_norm": 0.04549802467226982,
      "learning_rate": 1.2823529411764706e-05,
      "loss": 0.0173,
      "step": 5384
    },
    {
      "epoch": 48.08,
      "grad_norm": 0.03472273051738739,
      "learning_rate": 1.276470588235294e-05,
      "loss": 0.011,
      "step": 5385
    },
    {
      "epoch": 48.09,
      "grad_norm": 0.039434779435396194,
      "learning_rate": 1.2705882352941174e-05,
      "loss": 0.0144,
      "step": 5386
    },
    {
      "epoch": 48.1,
      "grad_norm": 0.054497748613357544,
      "learning_rate": 1.2647058823529412e-05,
      "loss": 0.0176,
      "step": 5387
    },
    {
      "epoch": 48.11,
      "grad_norm": 0.04958281293511391,
      "learning_rate": 1.2588235294117646e-05,
      "loss": 0.012,
      "step": 5388
    },
    {
      "epoch": 48.12,
      "grad_norm": 0.06760144978761673,
      "learning_rate": 1.252941176470588e-05,
      "loss": 0.0181,
      "step": 5389
    },
    {
      "epoch": 48.12,
      "grad_norm": 0.03802026808261871,
      "learning_rate": 1.2470588235294116e-05,
      "loss": 0.0144,
      "step": 5390
    },
    {
      "epoch": 48.13,
      "grad_norm": 0.04341566562652588,
      "learning_rate": 1.2411764705882352e-05,
      "loss": 0.0143,
      "step": 5391
    },
    {
      "epoch": 48.14,
      "grad_norm": 0.03979827091097832,
      "learning_rate": 1.2352941176470587e-05,
      "loss": 0.0138,
      "step": 5392
    },
    {
      "epoch": 48.15,
      "grad_norm": 0.045508190989494324,
      "learning_rate": 1.2294117647058822e-05,
      "loss": 0.0161,
      "step": 5393
    },
    {
      "epoch": 48.16,
      "grad_norm": 0.05848633125424385,
      "learning_rate": 1.2235294117647059e-05,
      "loss": 0.0162,
      "step": 5394
    },
    {
      "epoch": 48.17,
      "grad_norm": 0.07386845350265503,
      "learning_rate": 1.2176470588235294e-05,
      "loss": 0.0189,
      "step": 5395
    },
    {
      "epoch": 48.18,
      "grad_norm": 0.034790199249982834,
      "learning_rate": 1.2117647058823527e-05,
      "loss": 0.0146,
      "step": 5396
    },
    {
      "epoch": 48.19,
      "grad_norm": 0.041191063821315765,
      "learning_rate": 1.2058823529411765e-05,
      "loss": 0.013,
      "step": 5397
    },
    {
      "epoch": 48.2,
      "grad_norm": 0.03899366036057472,
      "learning_rate": 1.1999999999999999e-05,
      "loss": 0.0143,
      "step": 5398
    },
    {
      "epoch": 48.21,
      "grad_norm": 0.043636348098516464,
      "learning_rate": 1.1941176470588234e-05,
      "loss": 0.0156,
      "step": 5399
    },
    {
      "epoch": 48.21,
      "grad_norm": 0.0491810105741024,
      "learning_rate": 1.1882352941176469e-05,
      "loss": 0.0172,
      "step": 5400
    },
    {
      "epoch": 48.21,
      "eval_cer": 0.031036595220108768,
      "eval_loss": 0.21303808689117432,
      "eval_runtime": 22.4132,
      "eval_samples_per_second": 117.877,
      "eval_steps_per_second": 1.874,
      "eval_wer": 0.09851249504165013,
      "step": 5400
    },
    {
      "epoch": 48.22,
      "grad_norm": 0.09034627676010132,
      "learning_rate": 1.1823529411764705e-05,
      "loss": 0.0156,
      "step": 5401
    },
    {
      "epoch": 48.23,
      "grad_norm": 0.04086814075708389,
      "learning_rate": 1.176470588235294e-05,
      "loss": 0.0154,
      "step": 5402
    },
    {
      "epoch": 48.24,
      "grad_norm": 0.03886063024401665,
      "learning_rate": 1.1705882352941175e-05,
      "loss": 0.0164,
      "step": 5403
    },
    {
      "epoch": 48.25,
      "grad_norm": 0.038670048117637634,
      "learning_rate": 1.1647058823529412e-05,
      "loss": 0.015,
      "step": 5404
    },
    {
      "epoch": 48.26,
      "grad_norm": 0.045868970453739166,
      "learning_rate": 1.1588235294117647e-05,
      "loss": 0.0155,
      "step": 5405
    },
    {
      "epoch": 48.27,
      "grad_norm": 0.04485901817679405,
      "learning_rate": 1.152941176470588e-05,
      "loss": 0.0164,
      "step": 5406
    },
    {
      "epoch": 48.28,
      "grad_norm": 0.0518074557185173,
      "learning_rate": 1.1470588235294118e-05,
      "loss": 0.0124,
      "step": 5407
    },
    {
      "epoch": 48.29,
      "grad_norm": 0.06739792227745056,
      "learning_rate": 1.1411764705882352e-05,
      "loss": 0.018,
      "step": 5408
    },
    {
      "epoch": 48.29,
      "grad_norm": 0.035138729959726334,
      "learning_rate": 1.1352941176470587e-05,
      "loss": 0.0161,
      "step": 5409
    },
    {
      "epoch": 48.3,
      "grad_norm": 0.03395545855164528,
      "learning_rate": 1.1294117647058822e-05,
      "loss": 0.0137,
      "step": 5410
    },
    {
      "epoch": 48.31,
      "grad_norm": 0.036876097321510315,
      "learning_rate": 1.1235294117647058e-05,
      "loss": 0.0132,
      "step": 5411
    },
    {
      "epoch": 48.32,
      "grad_norm": 0.049984682351350784,
      "learning_rate": 1.1176470588235293e-05,
      "loss": 0.0188,
      "step": 5412
    },
    {
      "epoch": 48.33,
      "grad_norm": 0.04078980162739754,
      "learning_rate": 1.1117647058823528e-05,
      "loss": 0.0126,
      "step": 5413
    },
    {
      "epoch": 48.34,
      "grad_norm": 0.07755071669816971,
      "learning_rate": 1.1058823529411765e-05,
      "loss": 0.0151,
      "step": 5414
    },
    {
      "epoch": 48.35,
      "grad_norm": 0.03706435486674309,
      "learning_rate": 1.1e-05,
      "loss": 0.015,
      "step": 5415
    },
    {
      "epoch": 48.36,
      "grad_norm": 0.0402042493224144,
      "learning_rate": 1.0941176470588233e-05,
      "loss": 0.0165,
      "step": 5416
    },
    {
      "epoch": 48.37,
      "grad_norm": 0.038473259657621384,
      "learning_rate": 1.0882352941176471e-05,
      "loss": 0.0143,
      "step": 5417
    },
    {
      "epoch": 48.38,
      "grad_norm": 0.04128929600119591,
      "learning_rate": 1.0823529411764705e-05,
      "loss": 0.0173,
      "step": 5418
    },
    {
      "epoch": 48.38,
      "grad_norm": 0.04608842357993126,
      "learning_rate": 1.076470588235294e-05,
      "loss": 0.0159,
      "step": 5419
    },
    {
      "epoch": 48.39,
      "grad_norm": 0.07743281871080399,
      "learning_rate": 1.0705882352941175e-05,
      "loss": 0.0158,
      "step": 5420
    },
    {
      "epoch": 48.4,
      "grad_norm": 0.038645412772893906,
      "learning_rate": 1.0647058823529411e-05,
      "loss": 0.0138,
      "step": 5421
    },
    {
      "epoch": 48.41,
      "grad_norm": 0.040869440883398056,
      "learning_rate": 1.0588235294117646e-05,
      "loss": 0.0124,
      "step": 5422
    },
    {
      "epoch": 48.42,
      "grad_norm": 0.04238196089863777,
      "learning_rate": 1.0529411764705881e-05,
      "loss": 0.0158,
      "step": 5423
    },
    {
      "epoch": 48.43,
      "grad_norm": 0.04477444291114807,
      "learning_rate": 1.0470588235294118e-05,
      "loss": 0.0167,
      "step": 5424
    },
    {
      "epoch": 48.44,
      "grad_norm": 0.04536725580692291,
      "learning_rate": 1.0411764705882353e-05,
      "loss": 0.0138,
      "step": 5425
    },
    {
      "epoch": 48.45,
      "grad_norm": 0.08704833686351776,
      "learning_rate": 1.0352941176470586e-05,
      "loss": 0.0165,
      "step": 5426
    },
    {
      "epoch": 48.46,
      "grad_norm": 0.04076733440160751,
      "learning_rate": 1.0294117647058824e-05,
      "loss": 0.0157,
      "step": 5427
    },
    {
      "epoch": 48.46,
      "grad_norm": 0.04137210175395012,
      "learning_rate": 1.0235294117647058e-05,
      "loss": 0.0163,
      "step": 5428
    },
    {
      "epoch": 48.47,
      "grad_norm": 0.04278163984417915,
      "learning_rate": 1.0176470588235293e-05,
      "loss": 0.0157,
      "step": 5429
    },
    {
      "epoch": 48.48,
      "grad_norm": 0.04365207254886627,
      "learning_rate": 1.0117647058823528e-05,
      "loss": 0.0145,
      "step": 5430
    },
    {
      "epoch": 48.49,
      "grad_norm": 0.04416772723197937,
      "learning_rate": 1.0058823529411764e-05,
      "loss": 0.0126,
      "step": 5431
    },
    {
      "epoch": 48.5,
      "grad_norm": 0.050360389053821564,
      "learning_rate": 9.999999999999999e-06,
      "loss": 0.0146,
      "step": 5432
    },
    {
      "epoch": 48.51,
      "grad_norm": 0.04284907504916191,
      "learning_rate": 9.941176470588234e-06,
      "loss": 0.0158,
      "step": 5433
    },
    {
      "epoch": 48.52,
      "grad_norm": 0.03979944810271263,
      "learning_rate": 9.88235294117647e-06,
      "loss": 0.0151,
      "step": 5434
    },
    {
      "epoch": 48.53,
      "grad_norm": 0.03629522770643234,
      "learning_rate": 9.823529411764706e-06,
      "loss": 0.0138,
      "step": 5435
    },
    {
      "epoch": 48.54,
      "grad_norm": 0.03917134553194046,
      "learning_rate": 9.764705882352939e-06,
      "loss": 0.0155,
      "step": 5436
    },
    {
      "epoch": 48.54,
      "grad_norm": 0.04647763445973396,
      "learning_rate": 9.705882352941177e-06,
      "loss": 0.0175,
      "step": 5437
    },
    {
      "epoch": 48.55,
      "grad_norm": 0.052529849112033844,
      "learning_rate": 9.64705882352941e-06,
      "loss": 0.0131,
      "step": 5438
    },
    {
      "epoch": 48.56,
      "grad_norm": 0.07595250755548477,
      "learning_rate": 9.588235294117646e-06,
      "loss": 0.0198,
      "step": 5439
    },
    {
      "epoch": 48.57,
      "grad_norm": 0.03902169317007065,
      "learning_rate": 9.52941176470588e-06,
      "loss": 0.0163,
      "step": 5440
    },
    {
      "epoch": 48.58,
      "grad_norm": 0.03682319447398186,
      "learning_rate": 9.470588235294117e-06,
      "loss": 0.0148,
      "step": 5441
    },
    {
      "epoch": 48.59,
      "grad_norm": 0.042619552463293076,
      "learning_rate": 9.411764705882352e-06,
      "loss": 0.0148,
      "step": 5442
    },
    {
      "epoch": 48.6,
      "grad_norm": 0.04521515592932701,
      "learning_rate": 9.352941176470587e-06,
      "loss": 0.0148,
      "step": 5443
    },
    {
      "epoch": 48.61,
      "grad_norm": 0.05362657457590103,
      "learning_rate": 9.294117647058822e-06,
      "loss": 0.0177,
      "step": 5444
    },
    {
      "epoch": 48.62,
      "grad_norm": 0.07619816064834595,
      "learning_rate": 9.235294117647059e-06,
      "loss": 0.0204,
      "step": 5445
    },
    {
      "epoch": 48.62,
      "grad_norm": 0.05332028865814209,
      "learning_rate": 9.176470588235294e-06,
      "loss": 0.0163,
      "step": 5446
    },
    {
      "epoch": 48.63,
      "grad_norm": 0.03711302950978279,
      "learning_rate": 9.117647058823529e-06,
      "loss": 0.0136,
      "step": 5447
    },
    {
      "epoch": 48.64,
      "grad_norm": 0.039778921753168106,
      "learning_rate": 9.058823529411764e-06,
      "loss": 0.0161,
      "step": 5448
    },
    {
      "epoch": 48.65,
      "grad_norm": 0.042309898883104324,
      "learning_rate": 8.999999999999999e-06,
      "loss": 0.0144,
      "step": 5449
    },
    {
      "epoch": 48.66,
      "grad_norm": 0.04288448393344879,
      "learning_rate": 8.941176470588235e-06,
      "loss": 0.0142,
      "step": 5450
    },
    {
      "epoch": 48.67,
      "grad_norm": 0.09097269177436829,
      "learning_rate": 8.88235294117647e-06,
      "loss": 0.023,
      "step": 5451
    },
    {
      "epoch": 48.68,
      "grad_norm": 0.03806547820568085,
      "learning_rate": 8.823529411764705e-06,
      "loss": 0.0159,
      "step": 5452
    },
    {
      "epoch": 48.69,
      "grad_norm": 0.03374017775058746,
      "learning_rate": 8.76470588235294e-06,
      "loss": 0.0128,
      "step": 5453
    },
    {
      "epoch": 48.7,
      "grad_norm": 0.03565361723303795,
      "learning_rate": 8.705882352941175e-06,
      "loss": 0.0136,
      "step": 5454
    },
    {
      "epoch": 48.71,
      "grad_norm": 0.03919528052210808,
      "learning_rate": 8.647058823529412e-06,
      "loss": 0.0152,
      "step": 5455
    },
    {
      "epoch": 48.71,
      "grad_norm": 0.03529151901602745,
      "learning_rate": 8.588235294117647e-06,
      "loss": 0.0118,
      "step": 5456
    },
    {
      "epoch": 48.72,
      "grad_norm": 0.06929656863212585,
      "learning_rate": 8.529411764705882e-06,
      "loss": 0.0177,
      "step": 5457
    },
    {
      "epoch": 48.73,
      "grad_norm": 0.05540144443511963,
      "learning_rate": 8.470588235294117e-06,
      "loss": 0.0151,
      "step": 5458
    },
    {
      "epoch": 48.74,
      "grad_norm": 0.04252156987786293,
      "learning_rate": 8.411764705882352e-06,
      "loss": 0.0148,
      "step": 5459
    },
    {
      "epoch": 48.75,
      "grad_norm": 0.04111355170607567,
      "learning_rate": 8.352941176470588e-06,
      "loss": 0.0189,
      "step": 5460
    },
    {
      "epoch": 48.76,
      "grad_norm": 0.039277084171772,
      "learning_rate": 8.294117647058823e-06,
      "loss": 0.0132,
      "step": 5461
    },
    {
      "epoch": 48.77,
      "grad_norm": 0.03730860352516174,
      "learning_rate": 8.235294117647058e-06,
      "loss": 0.0134,
      "step": 5462
    },
    {
      "epoch": 48.78,
      "grad_norm": 0.06344401091337204,
      "learning_rate": 8.176470588235293e-06,
      "loss": 0.0195,
      "step": 5463
    },
    {
      "epoch": 48.79,
      "grad_norm": 0.060826625674963,
      "learning_rate": 8.117647058823528e-06,
      "loss": 0.0148,
      "step": 5464
    },
    {
      "epoch": 48.79,
      "grad_norm": 0.03751077502965927,
      "learning_rate": 8.058823529411765e-06,
      "loss": 0.0167,
      "step": 5465
    },
    {
      "epoch": 48.8,
      "grad_norm": 0.05045950412750244,
      "learning_rate": 8e-06,
      "loss": 0.0178,
      "step": 5466
    },
    {
      "epoch": 48.81,
      "grad_norm": 0.036700934171676636,
      "learning_rate": 7.941176470588235e-06,
      "loss": 0.0138,
      "step": 5467
    },
    {
      "epoch": 48.82,
      "grad_norm": 0.043851833790540695,
      "learning_rate": 7.88235294117647e-06,
      "loss": 0.0152,
      "step": 5468
    },
    {
      "epoch": 48.83,
      "grad_norm": 0.050441958010196686,
      "learning_rate": 7.823529411764705e-06,
      "loss": 0.0177,
      "step": 5469
    },
    {
      "epoch": 48.84,
      "grad_norm": 0.05202293023467064,
      "learning_rate": 7.764705882352941e-06,
      "loss": 0.0134,
      "step": 5470
    },
    {
      "epoch": 48.85,
      "grad_norm": 0.03672241047024727,
      "learning_rate": 7.705882352941176e-06,
      "loss": 0.0142,
      "step": 5471
    },
    {
      "epoch": 48.86,
      "grad_norm": 0.04451717808842659,
      "learning_rate": 7.647058823529411e-06,
      "loss": 0.0167,
      "step": 5472
    },
    {
      "epoch": 48.87,
      "grad_norm": 0.03132355585694313,
      "learning_rate": 7.588235294117647e-06,
      "loss": 0.0122,
      "step": 5473
    },
    {
      "epoch": 48.88,
      "grad_norm": 0.05414542928338051,
      "learning_rate": 7.529411764705881e-06,
      "loss": 0.0133,
      "step": 5474
    },
    {
      "epoch": 48.88,
      "grad_norm": 0.051125116646289825,
      "learning_rate": 7.470588235294117e-06,
      "loss": 0.0145,
      "step": 5475
    },
    {
      "epoch": 48.89,
      "grad_norm": 0.09471362084150314,
      "learning_rate": 7.411764705882352e-06,
      "loss": 0.0168,
      "step": 5476
    },
    {
      "epoch": 48.9,
      "grad_norm": 0.03679506480693817,
      "learning_rate": 7.352941176470588e-06,
      "loss": 0.0155,
      "step": 5477
    },
    {
      "epoch": 48.91,
      "grad_norm": 0.03783881291747093,
      "learning_rate": 7.294117647058823e-06,
      "loss": 0.0129,
      "step": 5478
    },
    {
      "epoch": 48.92,
      "grad_norm": 0.047866713255643845,
      "learning_rate": 7.2352941176470575e-06,
      "loss": 0.0152,
      "step": 5479
    },
    {
      "epoch": 48.93,
      "grad_norm": 0.03788332641124725,
      "learning_rate": 7.176470588235293e-06,
      "loss": 0.0132,
      "step": 5480
    },
    {
      "epoch": 48.94,
      "grad_norm": 0.04627546668052673,
      "learning_rate": 7.117647058823528e-06,
      "loss": 0.0143,
      "step": 5481
    },
    {
      "epoch": 48.95,
      "grad_norm": 0.0775122195482254,
      "learning_rate": 7.058823529411764e-06,
      "loss": 0.0208,
      "step": 5482
    },
    {
      "epoch": 48.96,
      "grad_norm": 0.051570240408182144,
      "learning_rate": 7e-06,
      "loss": 0.0156,
      "step": 5483
    },
    {
      "epoch": 48.96,
      "grad_norm": 0.04197980836033821,
      "learning_rate": 6.941176470588234e-06,
      "loss": 0.0158,
      "step": 5484
    },
    {
      "epoch": 48.97,
      "grad_norm": 0.03953948989510536,
      "learning_rate": 6.88235294117647e-06,
      "loss": 0.0136,
      "step": 5485
    },
    {
      "epoch": 48.98,
      "grad_norm": 0.05131828784942627,
      "learning_rate": 6.823529411764705e-06,
      "loss": 0.0169,
      "step": 5486
    },
    {
      "epoch": 48.99,
      "grad_norm": 0.04814542829990387,
      "learning_rate": 6.764705882352941e-06,
      "loss": 0.0148,
      "step": 5487
    },
    {
      "epoch": 49.0,
      "grad_norm": 0.04880789667367935,
      "learning_rate": 6.705882352941176e-06,
      "loss": 0.0136,
      "step": 5488
    },
    {
      "epoch": 49.01,
      "grad_norm": 0.037186071276664734,
      "learning_rate": 6.6470588235294105e-06,
      "loss": 0.0139,
      "step": 5489
    },
    {
      "epoch": 49.02,
      "grad_norm": 0.04042769595980644,
      "learning_rate": 6.588235294117646e-06,
      "loss": 0.0157,
      "step": 5490
    },
    {
      "epoch": 49.03,
      "grad_norm": 0.036447346210479736,
      "learning_rate": 6.529411764705881e-06,
      "loss": 0.0163,
      "step": 5491
    },
    {
      "epoch": 49.04,
      "grad_norm": 0.045571520924568176,
      "learning_rate": 6.470588235294117e-06,
      "loss": 0.0172,
      "step": 5492
    },
    {
      "epoch": 49.04,
      "grad_norm": 0.045244552195072174,
      "learning_rate": 6.411764705882353e-06,
      "loss": 0.0148,
      "step": 5493
    },
    {
      "epoch": 49.05,
      "grad_norm": 0.04821966215968132,
      "learning_rate": 6.352941176470587e-06,
      "loss": 0.014,
      "step": 5494
    },
    {
      "epoch": 49.06,
      "grad_norm": 0.05282486602663994,
      "learning_rate": 6.294117647058823e-06,
      "loss": 0.0156,
      "step": 5495
    },
    {
      "epoch": 49.07,
      "grad_norm": 0.03506775572896004,
      "learning_rate": 6.235294117647058e-06,
      "loss": 0.0163,
      "step": 5496
    },
    {
      "epoch": 49.08,
      "grad_norm": 0.04147464781999588,
      "learning_rate": 6.176470588235294e-06,
      "loss": 0.0138,
      "step": 5497
    },
    {
      "epoch": 49.09,
      "grad_norm": 0.049196507781744,
      "learning_rate": 6.117647058823529e-06,
      "loss": 0.0141,
      "step": 5498
    },
    {
      "epoch": 49.1,
      "grad_norm": 0.044748276472091675,
      "learning_rate": 6.0588235294117635e-06,
      "loss": 0.0172,
      "step": 5499
    },
    {
      "epoch": 49.11,
      "grad_norm": 0.0551871620118618,
      "learning_rate": 5.999999999999999e-06,
      "loss": 0.017,
      "step": 5500
    },
    {
      "epoch": 49.11,
      "eval_cer": 0.03161538374806914,
      "eval_loss": 0.2137099951505661,
      "eval_runtime": 22.2933,
      "eval_samples_per_second": 118.511,
      "eval_steps_per_second": 1.884,
      "eval_wer": 0.09944466481554938,
      "step": 5500
    },
    {
      "epoch": 49.12,
      "grad_norm": 0.06691740453243256,
      "learning_rate": 5.941176470588234e-06,
      "loss": 0.0172,
      "step": 5501
    },
    {
      "epoch": 49.12,
      "grad_norm": 0.03986073285341263,
      "learning_rate": 5.88235294117647e-06,
      "loss": 0.0155,
      "step": 5502
    },
    {
      "epoch": 49.13,
      "grad_norm": 0.04453933984041214,
      "learning_rate": 5.823529411764706e-06,
      "loss": 0.0185,
      "step": 5503
    },
    {
      "epoch": 49.14,
      "grad_norm": 0.0417616032063961,
      "learning_rate": 5.76470588235294e-06,
      "loss": 0.0156,
      "step": 5504
    },
    {
      "epoch": 49.15,
      "grad_norm": 0.03996429964900017,
      "learning_rate": 5.705882352941176e-06,
      "loss": 0.015,
      "step": 5505
    },
    {
      "epoch": 49.16,
      "grad_norm": 0.051541831344366074,
      "learning_rate": 5.647058823529411e-06,
      "loss": 0.0167,
      "step": 5506
    },
    {
      "epoch": 49.17,
      "grad_norm": 0.09497403353452682,
      "learning_rate": 5.5882352941176466e-06,
      "loss": 0.0185,
      "step": 5507
    },
    {
      "epoch": 49.18,
      "grad_norm": 0.04161723703145981,
      "learning_rate": 5.529411764705882e-06,
      "loss": 0.0147,
      "step": 5508
    },
    {
      "epoch": 49.19,
      "grad_norm": 0.04817429929971695,
      "learning_rate": 5.4705882352941165e-06,
      "loss": 0.0166,
      "step": 5509
    },
    {
      "epoch": 49.2,
      "grad_norm": 0.036694012582302094,
      "learning_rate": 5.411764705882352e-06,
      "loss": 0.013,
      "step": 5510
    },
    {
      "epoch": 49.21,
      "grad_norm": 0.04692278429865837,
      "learning_rate": 5.352941176470587e-06,
      "loss": 0.0166,
      "step": 5511
    },
    {
      "epoch": 49.21,
      "grad_norm": 0.03901823237538338,
      "learning_rate": 5.294117647058823e-06,
      "loss": 0.0117,
      "step": 5512
    },
    {
      "epoch": 49.22,
      "grad_norm": 0.08639217168092728,
      "learning_rate": 5.235294117647059e-06,
      "loss": 0.0211,
      "step": 5513
    },
    {
      "epoch": 49.23,
      "grad_norm": 0.047227051109075546,
      "learning_rate": 5.176470588235293e-06,
      "loss": 0.0171,
      "step": 5514
    },
    {
      "epoch": 49.24,
      "grad_norm": 0.03695077449083328,
      "learning_rate": 5.117647058823529e-06,
      "loss": 0.0155,
      "step": 5515
    },
    {
      "epoch": 49.25,
      "grad_norm": 0.04250597581267357,
      "learning_rate": 5.058823529411764e-06,
      "loss": 0.017,
      "step": 5516
    },
    {
      "epoch": 49.26,
      "grad_norm": 0.036251239478588104,
      "learning_rate": 4.9999999999999996e-06,
      "loss": 0.0118,
      "step": 5517
    },
    {
      "epoch": 49.27,
      "grad_norm": 0.04985490441322327,
      "learning_rate": 4.941176470588235e-06,
      "loss": 0.0163,
      "step": 5518
    },
    {
      "epoch": 49.28,
      "grad_norm": 0.08453112095594406,
      "learning_rate": 4.8823529411764695e-06,
      "loss": 0.0162,
      "step": 5519
    },
    {
      "epoch": 49.29,
      "grad_norm": 0.05833496153354645,
      "learning_rate": 4.823529411764705e-06,
      "loss": 0.018,
      "step": 5520
    },
    {
      "epoch": 49.29,
      "grad_norm": 0.03763258084654808,
      "learning_rate": 4.76470588235294e-06,
      "loss": 0.0139,
      "step": 5521
    },
    {
      "epoch": 49.3,
      "grad_norm": 0.04027899354696274,
      "learning_rate": 4.705882352941176e-06,
      "loss": 0.011,
      "step": 5522
    },
    {
      "epoch": 49.31,
      "grad_norm": 0.0386640764772892,
      "learning_rate": 4.647058823529411e-06,
      "loss": 0.0143,
      "step": 5523
    },
    {
      "epoch": 49.32,
      "grad_norm": 0.039674367755651474,
      "learning_rate": 4.588235294117647e-06,
      "loss": 0.0109,
      "step": 5524
    },
    {
      "epoch": 49.33,
      "grad_norm": 0.07444358617067337,
      "learning_rate": 4.529411764705882e-06,
      "loss": 0.0123,
      "step": 5525
    },
    {
      "epoch": 49.34,
      "grad_norm": 0.07372348755598068,
      "learning_rate": 4.470588235294118e-06,
      "loss": 0.0173,
      "step": 5526
    },
    {
      "epoch": 49.35,
      "grad_norm": 0.033435601741075516,
      "learning_rate": 4.4117647058823526e-06,
      "loss": 0.0132,
      "step": 5527
    },
    {
      "epoch": 49.36,
      "grad_norm": 0.0519268698990345,
      "learning_rate": 4.3529411764705875e-06,
      "loss": 0.0178,
      "step": 5528
    },
    {
      "epoch": 49.37,
      "grad_norm": 0.03948422893881798,
      "learning_rate": 4.294117647058823e-06,
      "loss": 0.0136,
      "step": 5529
    },
    {
      "epoch": 49.38,
      "grad_norm": 0.04186204820871353,
      "learning_rate": 4.235294117647058e-06,
      "loss": 0.0162,
      "step": 5530
    },
    {
      "epoch": 49.38,
      "grad_norm": 0.052189480513334274,
      "learning_rate": 4.176470588235294e-06,
      "loss": 0.0148,
      "step": 5531
    },
    {
      "epoch": 49.39,
      "grad_norm": 0.06453348696231842,
      "learning_rate": 4.117647058823529e-06,
      "loss": 0.0138,
      "step": 5532
    },
    {
      "epoch": 49.4,
      "grad_norm": 0.03898428753018379,
      "learning_rate": 4.058823529411764e-06,
      "loss": 0.0154,
      "step": 5533
    },
    {
      "epoch": 49.41,
      "grad_norm": 0.03803344815969467,
      "learning_rate": 4e-06,
      "loss": 0.0149,
      "step": 5534
    },
    {
      "epoch": 49.42,
      "grad_norm": 0.04040852189064026,
      "learning_rate": 3.941176470588235e-06,
      "loss": 0.0151,
      "step": 5535
    },
    {
      "epoch": 49.43,
      "grad_norm": 0.05394357070326805,
      "learning_rate": 3.882352941176471e-06,
      "loss": 0.0164,
      "step": 5536
    },
    {
      "epoch": 49.44,
      "grad_norm": 0.040444180369377136,
      "learning_rate": 3.8235294117647055e-06,
      "loss": 0.0141,
      "step": 5537
    },
    {
      "epoch": 49.45,
      "grad_norm": 0.08052466809749603,
      "learning_rate": 3.7647058823529405e-06,
      "loss": 0.0185,
      "step": 5538
    },
    {
      "epoch": 49.46,
      "grad_norm": 0.035009950399398804,
      "learning_rate": 3.705882352941176e-06,
      "loss": 0.0148,
      "step": 5539
    },
    {
      "epoch": 49.46,
      "grad_norm": 0.040605854243040085,
      "learning_rate": 3.6470588235294117e-06,
      "loss": 0.0155,
      "step": 5540
    },
    {
      "epoch": 49.47,
      "grad_norm": 0.03829166293144226,
      "learning_rate": 3.5882352941176467e-06,
      "loss": 0.013,
      "step": 5541
    },
    {
      "epoch": 49.48,
      "grad_norm": 0.04586075246334076,
      "learning_rate": 3.529411764705882e-06,
      "loss": 0.0157,
      "step": 5542
    },
    {
      "epoch": 49.49,
      "grad_norm": 0.04065732657909393,
      "learning_rate": 3.470588235294117e-06,
      "loss": 0.0151,
      "step": 5543
    },
    {
      "epoch": 49.5,
      "grad_norm": 0.0751880407333374,
      "learning_rate": 3.4117647058823524e-06,
      "loss": 0.0153,
      "step": 5544
    },
    {
      "epoch": 49.51,
      "grad_norm": 0.06613259017467499,
      "learning_rate": 3.352941176470588e-06,
      "loss": 0.0157,
      "step": 5545
    },
    {
      "epoch": 49.52,
      "grad_norm": 0.04640515521168709,
      "learning_rate": 3.294117647058823e-06,
      "loss": 0.0181,
      "step": 5546
    },
    {
      "epoch": 49.53,
      "grad_norm": 0.03859524428844452,
      "learning_rate": 3.2352941176470585e-06,
      "loss": 0.017,
      "step": 5547
    },
    {
      "epoch": 49.54,
      "grad_norm": 0.04536804184317589,
      "learning_rate": 3.1764705882352935e-06,
      "loss": 0.0149,
      "step": 5548
    },
    {
      "epoch": 49.54,
      "grad_norm": 0.04511251300573349,
      "learning_rate": 3.117647058823529e-06,
      "loss": 0.0152,
      "step": 5549
    },
    {
      "epoch": 49.55,
      "grad_norm": 0.06233420968055725,
      "learning_rate": 3.0588235294117647e-06,
      "loss": 0.0181,
      "step": 5550
    },
    {
      "epoch": 49.56,
      "grad_norm": 0.09452083706855774,
      "learning_rate": 2.9999999999999997e-06,
      "loss": 0.017,
      "step": 5551
    },
    {
      "epoch": 49.57,
      "grad_norm": 0.036748792976140976,
      "learning_rate": 2.941176470588235e-06,
      "loss": 0.0136,
      "step": 5552
    },
    {
      "epoch": 49.58,
      "grad_norm": 0.037301577627658844,
      "learning_rate": 2.88235294117647e-06,
      "loss": 0.0155,
      "step": 5553
    },
    {
      "epoch": 49.59,
      "grad_norm": 0.03743209317326546,
      "learning_rate": 2.8235294117647054e-06,
      "loss": 0.0131,
      "step": 5554
    },
    {
      "epoch": 49.6,
      "grad_norm": 0.0437840111553669,
      "learning_rate": 2.764705882352941e-06,
      "loss": 0.0169,
      "step": 5555
    },
    {
      "epoch": 49.61,
      "grad_norm": 0.05424346774816513,
      "learning_rate": 2.705882352941176e-06,
      "loss": 0.0155,
      "step": 5556
    },
    {
      "epoch": 49.62,
      "grad_norm": 0.06325390934944153,
      "learning_rate": 2.6470588235294115e-06,
      "loss": 0.0136,
      "step": 5557
    },
    {
      "epoch": 49.62,
      "grad_norm": 0.03527723625302315,
      "learning_rate": 2.5882352941176465e-06,
      "loss": 0.015,
      "step": 5558
    },
    {
      "epoch": 49.63,
      "grad_norm": 0.03985361009836197,
      "learning_rate": 2.529411764705882e-06,
      "loss": 0.0164,
      "step": 5559
    },
    {
      "epoch": 49.64,
      "grad_norm": 0.038023244589567184,
      "learning_rate": 2.4705882352941177e-06,
      "loss": 0.014,
      "step": 5560
    },
    {
      "epoch": 49.65,
      "grad_norm": 0.046528175473213196,
      "learning_rate": 2.4117647058823526e-06,
      "loss": 0.0168,
      "step": 5561
    },
    {
      "epoch": 49.66,
      "grad_norm": 0.048520345240831375,
      "learning_rate": 2.352941176470588e-06,
      "loss": 0.0132,
      "step": 5562
    },
    {
      "epoch": 49.67,
      "grad_norm": 0.06942006200551987,
      "learning_rate": 2.2941176470588234e-06,
      "loss": 0.0169,
      "step": 5563
    },
    {
      "epoch": 49.68,
      "grad_norm": 0.03572802245616913,
      "learning_rate": 2.235294117647059e-06,
      "loss": 0.0133,
      "step": 5564
    },
    {
      "epoch": 49.69,
      "grad_norm": 0.03914961963891983,
      "learning_rate": 2.1764705882352938e-06,
      "loss": 0.0148,
      "step": 5565
    },
    {
      "epoch": 49.7,
      "grad_norm": 0.04333449900150299,
      "learning_rate": 2.117647058823529e-06,
      "loss": 0.018,
      "step": 5566
    },
    {
      "epoch": 49.71,
      "grad_norm": 0.044780388474464417,
      "learning_rate": 2.0588235294117645e-06,
      "loss": 0.0123,
      "step": 5567
    },
    {
      "epoch": 49.71,
      "grad_norm": 0.03866629675030708,
      "learning_rate": 2e-06,
      "loss": 0.0125,
      "step": 5568
    },
    {
      "epoch": 49.72,
      "grad_norm": 0.07407289743423462,
      "learning_rate": 1.9411764705882353e-06,
      "loss": 0.0159,
      "step": 5569
    },
    {
      "epoch": 49.73,
      "grad_norm": 0.11904079467058182,
      "learning_rate": 1.8823529411764703e-06,
      "loss": 0.0155,
      "step": 5570
    },
    {
      "epoch": 49.74,
      "grad_norm": 0.034318406134843826,
      "learning_rate": 1.8235294117647058e-06,
      "loss": 0.0148,
      "step": 5571
    },
    {
      "epoch": 49.75,
      "grad_norm": 0.0394885390996933,
      "learning_rate": 1.764705882352941e-06,
      "loss": 0.016,
      "step": 5572
    },
    {
      "epoch": 49.76,
      "grad_norm": 0.04434511810541153,
      "learning_rate": 1.7058823529411762e-06,
      "loss": 0.0148,
      "step": 5573
    },
    {
      "epoch": 49.77,
      "grad_norm": 0.050256311893463135,
      "learning_rate": 1.6470588235294116e-06,
      "loss": 0.0153,
      "step": 5574
    },
    {
      "epoch": 49.78,
      "grad_norm": 0.04231452941894531,
      "learning_rate": 1.5882352941176468e-06,
      "loss": 0.009,
      "step": 5575
    },
    {
      "epoch": 49.79,
      "grad_norm": 0.0795440748333931,
      "learning_rate": 1.5294117647058823e-06,
      "loss": 0.0191,
      "step": 5576
    },
    {
      "epoch": 49.79,
      "grad_norm": 0.04174872487783432,
      "learning_rate": 1.4705882352941175e-06,
      "loss": 0.0166,
      "step": 5577
    },
    {
      "epoch": 49.8,
      "grad_norm": 0.03474007174372673,
      "learning_rate": 1.4117647058823527e-06,
      "loss": 0.0134,
      "step": 5578
    },
    {
      "epoch": 49.81,
      "grad_norm": 0.046486325562000275,
      "learning_rate": 1.352941176470588e-06,
      "loss": 0.0167,
      "step": 5579
    },
    {
      "epoch": 49.82,
      "grad_norm": 0.04521731659770012,
      "learning_rate": 1.2941176470588232e-06,
      "loss": 0.0144,
      "step": 5580
    },
    {
      "epoch": 49.83,
      "grad_norm": 0.052535295486450195,
      "learning_rate": 1.2352941176470588e-06,
      "loss": 0.0163,
      "step": 5581
    },
    {
      "epoch": 49.84,
      "grad_norm": 0.06909803301095963,
      "learning_rate": 1.176470588235294e-06,
      "loss": 0.017,
      "step": 5582
    },
    {
      "epoch": 49.85,
      "grad_norm": 0.04130563884973526,
      "learning_rate": 1.1176470588235294e-06,
      "loss": 0.017,
      "step": 5583
    },
    {
      "epoch": 49.86,
      "grad_norm": 0.04639488086104393,
      "learning_rate": 1.0588235294117646e-06,
      "loss": 0.0143,
      "step": 5584
    },
    {
      "epoch": 49.87,
      "grad_norm": 0.03253469243645668,
      "learning_rate": 1e-06,
      "loss": 0.0119,
      "step": 5585
    },
    {
      "epoch": 49.88,
      "grad_norm": 0.041558682918548584,
      "learning_rate": 9.411764705882351e-07,
      "loss": 0.0136,
      "step": 5586
    },
    {
      "epoch": 49.88,
      "grad_norm": 0.04843180626630783,
      "learning_rate": 8.823529411764705e-07,
      "loss": 0.0151,
      "step": 5587
    },
    {
      "epoch": 49.89,
      "grad_norm": 0.08238411694765091,
      "learning_rate": 8.235294117647058e-07,
      "loss": 0.0178,
      "step": 5588
    },
    {
      "epoch": 49.9,
      "grad_norm": 0.05005532503128052,
      "learning_rate": 7.647058823529412e-07,
      "loss": 0.0148,
      "step": 5589
    },
    {
      "epoch": 49.91,
      "grad_norm": 0.043235499411821365,
      "learning_rate": 7.058823529411763e-07,
      "loss": 0.0148,
      "step": 5590
    },
    {
      "epoch": 49.92,
      "grad_norm": 0.04127110540866852,
      "learning_rate": 6.470588235294116e-07,
      "loss": 0.0182,
      "step": 5591
    },
    {
      "epoch": 49.93,
      "grad_norm": 0.04067129269242287,
      "learning_rate": 5.88235294117647e-07,
      "loss": 0.015,
      "step": 5592
    },
    {
      "epoch": 49.94,
      "grad_norm": 0.04593358933925629,
      "learning_rate": 5.294117647058823e-07,
      "loss": 0.0161,
      "step": 5593
    },
    {
      "epoch": 49.95,
      "grad_norm": 0.07899615168571472,
      "learning_rate": 4.7058823529411756e-07,
      "loss": 0.0185,
      "step": 5594
    },
    {
      "epoch": 49.96,
      "grad_norm": 0.052048083394765854,
      "learning_rate": 4.117647058823529e-07,
      "loss": 0.0154,
      "step": 5595
    },
    {
      "epoch": 49.96,
      "grad_norm": 0.03803560510277748,
      "learning_rate": 3.5294117647058817e-07,
      "loss": 0.0144,
      "step": 5596
    },
    {
      "epoch": 49.97,
      "grad_norm": 0.032355159521102905,
      "learning_rate": 2.941176470588235e-07,
      "loss": 0.0124,
      "step": 5597
    },
    {
      "epoch": 49.98,
      "grad_norm": 0.03637293353676796,
      "learning_rate": 2.3529411764705878e-07,
      "loss": 0.0133,
      "step": 5598
    },
    {
      "epoch": 49.99,
      "grad_norm": 0.05197904258966446,
      "learning_rate": 1.7647058823529409e-07,
      "loss": 0.0155,
      "step": 5599
    },
    {
      "epoch": 50.0,
      "grad_norm": 0.060445237904787064,
      "learning_rate": 1.1764705882352939e-07,
      "loss": 0.0152,
      "step": 5600
    },
    {
      "epoch": 50.0,
      "eval_cer": 0.031607867013939785,
      "eval_loss": 0.21432273089885712,
      "eval_runtime": 22.3097,
      "eval_samples_per_second": 118.424,
      "eval_steps_per_second": 1.883,
      "eval_wer": 0.09948433161443872,
      "step": 5600
    },
    {
      "epoch": 50.0,
      "step": 5600,
      "total_flos": 5.590980477274324e+20,
      "train_loss": 0.20686535484689686,
      "train_runtime": 19804.4548,
      "train_samples_per_second": 72.049,
      "train_steps_per_second": 0.283
    },
    {
      "epoch": 50.0,
      "step": 5600,
      "total_flos": 5.590980477274324e+20,
      "train_loss": 0.0,
      "train_runtime": 249.9893,
      "train_samples_per_second": 5707.844,
      "train_steps_per_second": 22.401
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 5600,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 50,
  "save_steps": 400,
  "total_flos": 5.590980477274324e+20,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}