{
  "best_metric": 26.328800988875155,
  "best_model_checkpoint": "results/whisper-tiny/marathi/checkpoint-12000",
  "epoch": 14.775016789791806,
  "eval_steps": 1000,
  "global_step": 22000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "grad_norm": 74.15543365478516,
      "learning_rate": 4.4e-07,
      "loss": 3.7766,
      "step": 25
    },
    {
      "epoch": 0.03,
      "grad_norm": 39.68334197998047,
      "learning_rate": 9.200000000000001e-07,
      "loss": 3.2026,
      "step": 50
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.481256484985352,
      "learning_rate": 1.42e-06,
      "loss": 2.4672,
      "step": 75
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.309144020080566,
      "learning_rate": 1.9200000000000003e-06,
      "loss": 1.9195,
      "step": 100
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.247703552246094,
      "learning_rate": 2.42e-06,
      "loss": 1.5361,
      "step": 125
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.86753511428833,
      "learning_rate": 2.92e-06,
      "loss": 1.2775,
      "step": 150
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.9364752769470215,
      "learning_rate": 3.4200000000000007e-06,
      "loss": 1.0933,
      "step": 175
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.17349910736084,
      "learning_rate": 3.920000000000001e-06,
      "loss": 0.9505,
      "step": 200
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.331369400024414,
      "learning_rate": 4.42e-06,
      "loss": 0.8507,
      "step": 225
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.466459274291992,
      "learning_rate": 4.92e-06,
      "loss": 0.7655,
      "step": 250
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.921384811401367,
      "learning_rate": 5.420000000000001e-06,
      "loss": 0.7153,
      "step": 275
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.227000713348389,
      "learning_rate": 5.92e-06,
      "loss": 0.6886,
      "step": 300
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.924015045166016,
      "learning_rate": 6.42e-06,
      "loss": 0.6324,
      "step": 325
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.0168986320495605,
      "learning_rate": 6.92e-06,
      "loss": 0.6107,
      "step": 350
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.831826686859131,
      "learning_rate": 7.420000000000001e-06,
      "loss": 0.5784,
      "step": 375
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.7476935386657715,
      "learning_rate": 7.92e-06,
      "loss": 0.5509,
      "step": 400
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.2020978927612305,
      "learning_rate": 8.42e-06,
      "loss": 0.5442,
      "step": 425
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.830783843994141,
      "learning_rate": 8.920000000000001e-06,
      "loss": 0.5297,
      "step": 450
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.747669696807861,
      "learning_rate": 9.42e-06,
      "loss": 0.5059,
      "step": 475
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.504109859466553,
      "learning_rate": 9.920000000000002e-06,
      "loss": 0.4927,
      "step": 500
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.707924842834473,
      "learning_rate": 9.997889447236182e-06,
      "loss": 0.4721,
      "step": 525
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.621720790863037,
      "learning_rate": 9.995376884422112e-06,
      "loss": 0.464,
      "step": 550
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.50490140914917,
      "learning_rate": 9.992864321608041e-06,
      "loss": 0.4518,
      "step": 575
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.592816352844238,
      "learning_rate": 9.99035175879397e-06,
      "loss": 0.4335,
      "step": 600
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.791091442108154,
      "learning_rate": 9.9878391959799e-06,
      "loss": 0.4348,
      "step": 625
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.221704959869385,
      "learning_rate": 9.98532663316583e-06,
      "loss": 0.4203,
      "step": 650
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.549515724182129,
      "learning_rate": 9.98281407035176e-06,
      "loss": 0.4086,
      "step": 675
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.485387802124023,
      "learning_rate": 9.98030150753769e-06,
      "loss": 0.405,
      "step": 700
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.758955001831055,
      "learning_rate": 9.977788944723619e-06,
      "loss": 0.4016,
      "step": 725
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.615067005157471,
      "learning_rate": 9.975276381909548e-06,
      "loss": 0.393,
      "step": 750
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.661777019500732,
      "learning_rate": 9.972763819095477e-06,
      "loss": 0.3843,
      "step": 775
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.5793609619140625,
      "learning_rate": 9.970251256281408e-06,
      "loss": 0.3832,
      "step": 800
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.030839443206787,
      "learning_rate": 9.967738693467338e-06,
      "loss": 0.3789,
      "step": 825
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.351238250732422,
      "learning_rate": 9.965226130653267e-06,
      "loss": 0.3576,
      "step": 850
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.560535907745361,
      "learning_rate": 9.962713567839198e-06,
      "loss": 0.3589,
      "step": 875
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.39430046081543,
      "learning_rate": 9.960201005025126e-06,
      "loss": 0.3554,
      "step": 900
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.813572883605957,
      "learning_rate": 9.957688442211057e-06,
      "loss": 0.3561,
      "step": 925
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.895594358444214,
      "learning_rate": 9.955175879396986e-06,
      "loss": 0.3532,
      "step": 950
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.176882266998291,
      "learning_rate": 9.952663316582915e-06,
      "loss": 0.3501,
      "step": 975
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.483668327331543,
      "learning_rate": 9.950150753768845e-06,
      "loss": 0.3485,
      "step": 1000
    },
    {
      "epoch": 0.67,
      "eval_loss": 0.23381204903125763,
      "eval_runtime": 566.2474,
      "eval_samples_per_second": 2.448,
      "eval_steps_per_second": 2.448,
      "eval_wer": 47.28059332509271,
      "step": 1000
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.013958930969238,
      "learning_rate": 9.947638190954774e-06,
      "loss": 0.3408,
      "step": 1025
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.734582424163818,
      "learning_rate": 9.945125628140703e-06,
      "loss": 0.3313,
      "step": 1050
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.5922722816467285,
      "learning_rate": 9.942613065326634e-06,
      "loss": 0.3364,
      "step": 1075
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.997859001159668,
      "learning_rate": 9.940100502512564e-06,
      "loss": 0.3283,
      "step": 1100
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.432836532592773,
      "learning_rate": 9.937587939698493e-06,
      "loss": 0.3258,
      "step": 1125
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.074716091156006,
      "learning_rate": 9.935075376884424e-06,
      "loss": 0.3338,
      "step": 1150
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.509114742279053,
      "learning_rate": 9.932562814070352e-06,
      "loss": 0.3121,
      "step": 1175
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.589898586273193,
      "learning_rate": 9.930050251256283e-06,
      "loss": 0.3161,
      "step": 1200
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.0301079750061035,
      "learning_rate": 9.927537688442212e-06,
      "loss": 0.3248,
      "step": 1225
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.21639347076416,
      "learning_rate": 9.925025125628141e-06,
      "loss": 0.3096,
      "step": 1250
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.40596866607666,
      "learning_rate": 9.922512562814072e-06,
      "loss": 0.3136,
      "step": 1275
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.144809722900391,
      "learning_rate": 9.920000000000002e-06,
      "loss": 0.3068,
      "step": 1300
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.97633695602417,
      "learning_rate": 9.917487437185931e-06,
      "loss": 0.3044,
      "step": 1325
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.247403144836426,
      "learning_rate": 9.91497487437186e-06,
      "loss": 0.307,
      "step": 1350
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9145348072052,
      "learning_rate": 9.91246231155779e-06,
      "loss": 0.3007,
      "step": 1375
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.151167869567871,
      "learning_rate": 9.909949748743719e-06,
      "loss": 0.2931,
      "step": 1400
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.783816337585449,
      "learning_rate": 9.90743718592965e-06,
      "loss": 0.2939,
      "step": 1425
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.319779872894287,
      "learning_rate": 9.904924623115578e-06,
      "loss": 0.294,
      "step": 1450
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.233304500579834,
      "learning_rate": 9.902412060301509e-06,
      "loss": 0.2918,
      "step": 1475
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.045380592346191,
      "learning_rate": 9.899899497487438e-06,
      "loss": 0.2839,
      "step": 1500
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.171890735626221,
      "learning_rate": 9.897386934673367e-06,
      "loss": 0.277,
      "step": 1525
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.11909818649292,
      "learning_rate": 9.894874371859298e-06,
      "loss": 0.2708,
      "step": 1550
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.329667568206787,
      "learning_rate": 9.892361809045228e-06,
      "loss": 0.278,
      "step": 1575
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.358795166015625,
      "learning_rate": 9.889849246231157e-06,
      "loss": 0.2702,
      "step": 1600
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.983689308166504,
      "learning_rate": 9.887336683417086e-06,
      "loss": 0.2594,
      "step": 1625
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.065433502197266,
      "learning_rate": 9.884824120603015e-06,
      "loss": 0.2628,
      "step": 1650
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.138759136199951,
      "learning_rate": 9.882311557788945e-06,
      "loss": 0.2692,
      "step": 1675
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.43567419052124,
      "learning_rate": 9.879798994974876e-06,
      "loss": 0.2688,
      "step": 1700
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.089324951171875,
      "learning_rate": 9.877286432160805e-06,
      "loss": 0.2641,
      "step": 1725
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.174434661865234,
      "learning_rate": 9.874773869346734e-06,
      "loss": 0.2638,
      "step": 1750
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.324215888977051,
      "learning_rate": 9.872261306532664e-06,
      "loss": 0.2621,
      "step": 1775
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.167600631713867,
      "learning_rate": 9.869748743718593e-06,
      "loss": 0.2568,
      "step": 1800
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.090190410614014,
      "learning_rate": 9.867236180904524e-06,
      "loss": 0.2579,
      "step": 1825
    },
    {
      "epoch": 1.24,
      "grad_norm": 3.862471580505371,
      "learning_rate": 9.864723618090453e-06,
      "loss": 0.2549,
      "step": 1850
    },
    {
      "epoch": 1.26,
      "grad_norm": 3.9046545028686523,
      "learning_rate": 9.862211055276383e-06,
      "loss": 0.2512,
      "step": 1875
    },
    {
      "epoch": 1.28,
      "grad_norm": 3.973026990890503,
      "learning_rate": 9.859698492462312e-06,
      "loss": 0.2535,
      "step": 1900
    },
    {
      "epoch": 1.29,
      "grad_norm": 3.875776529312134,
      "learning_rate": 9.857185929648241e-06,
      "loss": 0.2454,
      "step": 1925
    },
    {
      "epoch": 1.31,
      "grad_norm": 3.815830707550049,
      "learning_rate": 9.854673366834172e-06,
      "loss": 0.2509,
      "step": 1950
    },
    {
      "epoch": 1.33,
      "grad_norm": 3.9826467037200928,
      "learning_rate": 9.852160804020102e-06,
      "loss": 0.2469,
      "step": 1975
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.199316024780273,
      "learning_rate": 9.849648241206031e-06,
      "loss": 0.2543,
      "step": 2000
    },
    {
      "epoch": 1.34,
      "eval_loss": 0.1726374477148056,
      "eval_runtime": 531.5792,
      "eval_samples_per_second": 2.607,
      "eval_steps_per_second": 2.607,
      "eval_wer": 38.892812996644885,
      "step": 2000
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.188065052032471,
      "learning_rate": 9.84713567839196e-06,
      "loss": 0.253,
      "step": 2025
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.348769187927246,
      "learning_rate": 9.84462311557789e-06,
      "loss": 0.2439,
      "step": 2050
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.025571823120117,
      "learning_rate": 9.842110552763819e-06,
      "loss": 0.2499,
      "step": 2075
    },
    {
      "epoch": 1.41,
      "grad_norm": 3.441206216812134,
      "learning_rate": 9.83959798994975e-06,
      "loss": 0.2419,
      "step": 2100
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.062358856201172,
      "learning_rate": 9.83708542713568e-06,
      "loss": 0.2428,
      "step": 2125
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.701034069061279,
      "learning_rate": 9.834572864321609e-06,
      "loss": 0.2435,
      "step": 2150
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.011937618255615,
      "learning_rate": 9.832060301507538e-06,
      "loss": 0.2429,
      "step": 2175
    },
    {
      "epoch": 1.48,
      "grad_norm": 3.9073057174682617,
      "learning_rate": 9.829547738693467e-06,
      "loss": 0.2371,
      "step": 2200
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.053809642791748,
      "learning_rate": 9.827035175879398e-06,
      "loss": 0.236,
      "step": 2225
    },
    {
      "epoch": 1.51,
      "grad_norm": 3.983830690383911,
      "learning_rate": 9.824522613065328e-06,
      "loss": 0.2393,
      "step": 2250
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.095301151275635,
      "learning_rate": 9.822010050251257e-06,
      "loss": 0.2329,
      "step": 2275
    },
    {
      "epoch": 1.54,
      "grad_norm": 3.980642318725586,
      "learning_rate": 9.819497487437186e-06,
      "loss": 0.24,
      "step": 2300
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.499876976013184,
      "learning_rate": 9.816984924623116e-06,
      "loss": 0.2307,
      "step": 2325
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.50550651550293,
      "learning_rate": 9.814472361809047e-06,
      "loss": 0.2336,
      "step": 2350
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.186766147613525,
      "learning_rate": 9.811959798994976e-06,
      "loss": 0.233,
      "step": 2375
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.362492084503174,
      "learning_rate": 9.809447236180905e-06,
      "loss": 0.2281,
      "step": 2400
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.026979446411133,
      "learning_rate": 9.806934673366835e-06,
      "loss": 0.227,
      "step": 2425
    },
    {
      "epoch": 1.65,
      "grad_norm": 3.556326389312744,
      "learning_rate": 9.804422110552764e-06,
      "loss": 0.2331,
      "step": 2450
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.109285831451416,
      "learning_rate": 9.801909547738693e-06,
      "loss": 0.2369,
      "step": 2475
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.070573329925537,
      "learning_rate": 9.799396984924624e-06,
      "loss": 0.2307,
      "step": 2500
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.282459735870361,
      "learning_rate": 9.796884422110554e-06,
      "loss": 0.2276,
      "step": 2525
    },
    {
      "epoch": 1.71,
      "grad_norm": 3.989485263824463,
      "learning_rate": 9.794371859296483e-06,
      "loss": 0.2325,
      "step": 2550
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.217010021209717,
      "learning_rate": 9.791859296482414e-06,
      "loss": 0.222,
      "step": 2575
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.2022199630737305,
      "learning_rate": 9.789346733668342e-06,
      "loss": 0.2281,
      "step": 2600
    },
    {
      "epoch": 1.76,
      "grad_norm": 3.7386114597320557,
      "learning_rate": 9.786834170854273e-06,
      "loss": 0.2165,
      "step": 2625
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.048258304595947,
      "learning_rate": 9.784321608040202e-06,
      "loss": 0.2218,
      "step": 2650
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.0867133140563965,
      "learning_rate": 9.781809045226131e-06,
      "loss": 0.2197,
      "step": 2675
    },
    {
      "epoch": 1.81,
      "grad_norm": 3.8676252365112305,
      "learning_rate": 9.77929648241206e-06,
      "loss": 0.2261,
      "step": 2700
    },
    {
      "epoch": 1.83,
      "grad_norm": 3.8840291500091553,
      "learning_rate": 9.77678391959799e-06,
      "loss": 0.2188,
      "step": 2725
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.130185127258301,
      "learning_rate": 9.774271356783921e-06,
      "loss": 0.2176,
      "step": 2750
    },
    {
      "epoch": 1.86,
      "grad_norm": 3.8641357421875,
      "learning_rate": 9.77175879396985e-06,
      "loss": 0.2165,
      "step": 2775
    },
    {
      "epoch": 1.88,
      "grad_norm": 3.8261783123016357,
      "learning_rate": 9.76924623115578e-06,
      "loss": 0.2149,
      "step": 2800
    },
    {
      "epoch": 1.9,
      "grad_norm": 3.861722946166992,
      "learning_rate": 9.766733668341709e-06,
      "loss": 0.2122,
      "step": 2825
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.013296127319336,
      "learning_rate": 9.76422110552764e-06,
      "loss": 0.2183,
      "step": 2850
    },
    {
      "epoch": 1.93,
      "grad_norm": 3.78545880317688,
      "learning_rate": 9.761708542713568e-06,
      "loss": 0.2151,
      "step": 2875
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.247804641723633,
      "learning_rate": 9.759195979899499e-06,
      "loss": 0.2213,
      "step": 2900
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.548637390136719,
      "learning_rate": 9.756683417085428e-06,
      "loss": 0.2198,
      "step": 2925
    },
    {
      "epoch": 1.98,
      "grad_norm": 3.617631435394287,
      "learning_rate": 9.754170854271357e-06,
      "loss": 0.2103,
      "step": 2950
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.9520044326782227,
      "learning_rate": 9.751658291457288e-06,
      "loss": 0.2053,
      "step": 2975
    },
    {
      "epoch": 2.01,
      "grad_norm": 3.903465747833252,
      "learning_rate": 9.749145728643216e-06,
      "loss": 0.1961,
      "step": 3000
    },
    {
      "epoch": 2.01,
      "eval_loss": 0.1471080482006073,
      "eval_runtime": 533.9926,
      "eval_samples_per_second": 2.596,
      "eval_steps_per_second": 2.596,
      "eval_wer": 33.365707222320324,
      "step": 3000
    },
    {
      "epoch": 2.03,
      "grad_norm": 3.3298327922821045,
      "learning_rate": 9.746633165829147e-06,
      "loss": 0.1968,
      "step": 3025
    },
    {
      "epoch": 2.05,
      "grad_norm": 3.907670259475708,
      "learning_rate": 9.744120603015076e-06,
      "loss": 0.1938,
      "step": 3050
    },
    {
      "epoch": 2.07,
      "grad_norm": 3.819309711456299,
      "learning_rate": 9.741608040201006e-06,
      "loss": 0.1879,
      "step": 3075
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.644184112548828,
      "learning_rate": 9.739095477386935e-06,
      "loss": 0.1933,
      "step": 3100
    },
    {
      "epoch": 2.1,
      "grad_norm": 3.5478782653808594,
      "learning_rate": 9.736582914572866e-06,
      "loss": 0.19,
      "step": 3125
    },
    {
      "epoch": 2.12,
      "grad_norm": 3.4926066398620605,
      "learning_rate": 9.734070351758794e-06,
      "loss": 0.1929,
      "step": 3150
    },
    {
      "epoch": 2.13,
      "grad_norm": 3.6318588256835938,
      "learning_rate": 9.731557788944725e-06,
      "loss": 0.1921,
      "step": 3175
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.020270824432373,
      "learning_rate": 9.729045226130654e-06,
      "loss": 0.19,
      "step": 3200
    },
    {
      "epoch": 2.17,
      "grad_norm": 3.391878128051758,
      "learning_rate": 9.726532663316583e-06,
      "loss": 0.1912,
      "step": 3225
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.8649306297302246,
      "learning_rate": 9.724020100502514e-06,
      "loss": 0.1965,
      "step": 3250
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.8927695751190186,
      "learning_rate": 9.721507537688444e-06,
      "loss": 0.1901,
      "step": 3275
    },
    {
      "epoch": 2.22,
      "grad_norm": 3.7473957538604736,
      "learning_rate": 9.718994974874373e-06,
      "loss": 0.1932,
      "step": 3300
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.2613677978515625,
      "learning_rate": 9.716482412060302e-06,
      "loss": 0.1925,
      "step": 3325
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.175868988037109,
      "learning_rate": 9.713969849246232e-06,
      "loss": 0.1898,
      "step": 3350
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.236743450164795,
      "learning_rate": 9.711457286432163e-06,
      "loss": 0.1899,
      "step": 3375
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.136856555938721,
      "learning_rate": 9.708944723618092e-06,
      "loss": 0.1912,
      "step": 3400
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.826167345046997,
      "learning_rate": 9.706432160804021e-06,
      "loss": 0.1909,
      "step": 3425
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.949150323867798,
      "learning_rate": 9.70391959798995e-06,
      "loss": 0.1889,
      "step": 3450
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.023538589477539,
      "learning_rate": 9.70140703517588e-06,
      "loss": 0.1903,
      "step": 3475
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.7844576835632324,
      "learning_rate": 9.698894472361809e-06,
      "loss": 0.1928,
      "step": 3500
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.364312171936035,
      "learning_rate": 9.69638190954774e-06,
      "loss": 0.193,
      "step": 3525
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.4202849864959717,
      "learning_rate": 9.69386934673367e-06,
      "loss": 0.186,
      "step": 3550
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.6285476684570312,
      "learning_rate": 9.691356783919599e-06,
      "loss": 0.1863,
      "step": 3575
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.26074743270874,
      "learning_rate": 9.688844221105528e-06,
      "loss": 0.1831,
      "step": 3600
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.6059014797210693,
      "learning_rate": 9.686331658291457e-06,
      "loss": 0.182,
      "step": 3625
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.773573637008667,
      "learning_rate": 9.683819095477388e-06,
      "loss": 0.1824,
      "step": 3650
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.112974643707275,
      "learning_rate": 9.681306532663318e-06,
      "loss": 0.1828,
      "step": 3675
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.467885732650757,
      "learning_rate": 9.678793969849247e-06,
      "loss": 0.1819,
      "step": 3700
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.418673038482666,
      "learning_rate": 9.676281407035176e-06,
      "loss": 0.1831,
      "step": 3725
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.263250350952148,
      "learning_rate": 9.673768844221106e-06,
      "loss": 0.186,
      "step": 3750
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.5524044036865234,
      "learning_rate": 9.671256281407035e-06,
      "loss": 0.1889,
      "step": 3775
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.693559408187866,
      "learning_rate": 9.668743718592966e-06,
      "loss": 0.1842,
      "step": 3800
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.682617425918579,
      "learning_rate": 9.666231155778895e-06,
      "loss": 0.1826,
      "step": 3825
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.4766149520874023,
      "learning_rate": 9.663718592964825e-06,
      "loss": 0.1766,
      "step": 3850
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.3245768547058105,
      "learning_rate": 9.661206030150754e-06,
      "loss": 0.1816,
      "step": 3875
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.028345584869385,
      "learning_rate": 9.658693467336683e-06,
      "loss": 0.1803,
      "step": 3900
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.665334463119507,
      "learning_rate": 9.656180904522614e-06,
      "loss": 0.178,
      "step": 3925
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.212314128875732,
      "learning_rate": 9.653668341708544e-06,
      "loss": 0.1818,
      "step": 3950
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.093043804168701,
      "learning_rate": 9.651155778894473e-06,
      "loss": 0.1784,
      "step": 3975
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.8363521099090576,
      "learning_rate": 9.648643216080404e-06,
      "loss": 0.1786,
      "step": 4000
    },
    {
      "epoch": 2.69,
      "eval_loss": 0.1332446187734604,
      "eval_runtime": 541.0077,
      "eval_samples_per_second": 2.562,
      "eval_steps_per_second": 2.562,
      "eval_wer": 30.64630054741303,
      "step": 4000
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.6020116806030273,
      "learning_rate": 9.646130653266332e-06,
      "loss": 0.1755,
      "step": 4025
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.253662586212158,
      "learning_rate": 9.643618090452263e-06,
      "loss": 0.1767,
      "step": 4050
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.3012306690216064,
      "learning_rate": 9.641105527638192e-06,
      "loss": 0.1813,
      "step": 4075
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.7597391605377197,
      "learning_rate": 9.638592964824121e-06,
      "loss": 0.1806,
      "step": 4100
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.914498805999756,
      "learning_rate": 9.63608040201005e-06,
      "loss": 0.1781,
      "step": 4125
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.5466084480285645,
      "learning_rate": 9.63356783919598e-06,
      "loss": 0.1784,
      "step": 4150
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.5035176277160645,
      "learning_rate": 9.63105527638191e-06,
      "loss": 0.1693,
      "step": 4175
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.6211013793945312,
      "learning_rate": 9.62854271356784e-06,
      "loss": 0.1831,
      "step": 4200
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.355555772781372,
      "learning_rate": 9.62603015075377e-06,
      "loss": 0.1737,
      "step": 4225
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.14884614944458,
      "learning_rate": 9.623517587939699e-06,
      "loss": 0.1773,
      "step": 4250
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.896099805831909,
      "learning_rate": 9.62100502512563e-06,
      "loss": 0.1712,
      "step": 4275
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.69228196144104,
      "learning_rate": 9.618492462311558e-06,
      "loss": 0.1747,
      "step": 4300
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.4385323524475098,
      "learning_rate": 9.615979899497489e-06,
      "loss": 0.1806,
      "step": 4325
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.37261438369751,
      "learning_rate": 9.613467336683418e-06,
      "loss": 0.1705,
      "step": 4350
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.549129009246826,
      "learning_rate": 9.610954773869347e-06,
      "loss": 0.1683,
      "step": 4375
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.8860154151916504,
      "learning_rate": 9.608442211055277e-06,
      "loss": 0.1732,
      "step": 4400
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.8280348777770996,
      "learning_rate": 9.605929648241206e-06,
      "loss": 0.1739,
      "step": 4425
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.9021239280700684,
      "learning_rate": 9.603417085427137e-06,
      "loss": 0.1744,
      "step": 4450
    },
    {
      "epoch": 3.01,
      "grad_norm": 3.436377763748169,
      "learning_rate": 9.600904522613066e-06,
      "loss": 0.1674,
      "step": 4475
    },
    {
      "epoch": 3.02,
      "grad_norm": 3.5898520946502686,
      "learning_rate": 9.598391959798996e-06,
      "loss": 0.1546,
      "step": 4500
    },
    {
      "epoch": 3.04,
      "grad_norm": 3.295307159423828,
      "learning_rate": 9.595879396984925e-06,
      "loss": 0.1543,
      "step": 4525
    },
    {
      "epoch": 3.06,
      "grad_norm": 3.3402857780456543,
      "learning_rate": 9.593366834170856e-06,
      "loss": 0.1585,
      "step": 4550
    },
    {
      "epoch": 3.07,
      "grad_norm": 3.4992740154266357,
      "learning_rate": 9.590854271356784e-06,
      "loss": 0.1584,
      "step": 4575
    },
    {
      "epoch": 3.09,
      "grad_norm": 3.301234245300293,
      "learning_rate": 9.588341708542715e-06,
      "loss": 0.1556,
      "step": 4600
    },
    {
      "epoch": 3.11,
      "grad_norm": 3.2298898696899414,
      "learning_rate": 9.585829145728644e-06,
      "loss": 0.1557,
      "step": 4625
    },
    {
      "epoch": 3.12,
      "grad_norm": 3.81208872795105,
      "learning_rate": 9.583316582914573e-06,
      "loss": 0.1488,
      "step": 4650
    },
    {
      "epoch": 3.14,
      "grad_norm": 3.7610058784484863,
      "learning_rate": 9.580804020100504e-06,
      "loss": 0.1537,
      "step": 4675
    },
    {
      "epoch": 3.16,
      "grad_norm": 3.394169569015503,
      "learning_rate": 9.578291457286432e-06,
      "loss": 0.1546,
      "step": 4700
    },
    {
      "epoch": 3.17,
      "grad_norm": 3.5936498641967773,
      "learning_rate": 9.575778894472363e-06,
      "loss": 0.1544,
      "step": 4725
    },
    {
      "epoch": 3.19,
      "grad_norm": 3.714808464050293,
      "learning_rate": 9.573266331658292e-06,
      "loss": 0.1554,
      "step": 4750
    },
    {
      "epoch": 3.21,
      "grad_norm": 3.731008768081665,
      "learning_rate": 9.570753768844222e-06,
      "loss": 0.157,
      "step": 4775
    },
    {
      "epoch": 3.22,
      "grad_norm": 3.4987032413482666,
      "learning_rate": 9.568241206030151e-06,
      "loss": 0.1506,
      "step": 4800
    },
    {
      "epoch": 3.24,
      "grad_norm": 3.487567186355591,
      "learning_rate": 9.565728643216082e-06,
      "loss": 0.1553,
      "step": 4825
    },
    {
      "epoch": 3.26,
      "grad_norm": 3.537971258163452,
      "learning_rate": 9.563216080402011e-06,
      "loss": 0.1518,
      "step": 4850
    },
    {
      "epoch": 3.27,
      "grad_norm": 3.5595390796661377,
      "learning_rate": 9.56070351758794e-06,
      "loss": 0.1519,
      "step": 4875
    },
    {
      "epoch": 3.29,
      "grad_norm": 3.397580146789551,
      "learning_rate": 9.55819095477387e-06,
      "loss": 0.1516,
      "step": 4900
    },
    {
      "epoch": 3.31,
      "grad_norm": 3.758497714996338,
      "learning_rate": 9.5556783919598e-06,
      "loss": 0.1562,
      "step": 4925
    },
    {
      "epoch": 3.32,
      "grad_norm": 3.1611812114715576,
      "learning_rate": 9.55316582914573e-06,
      "loss": 0.1522,
      "step": 4950
    },
    {
      "epoch": 3.34,
      "grad_norm": 3.4654600620269775,
      "learning_rate": 9.550653266331658e-06,
      "loss": 0.1544,
      "step": 4975
    },
    {
      "epoch": 3.36,
      "grad_norm": 3.0207717418670654,
      "learning_rate": 9.548140703517589e-06,
      "loss": 0.1574,
      "step": 5000
    },
    {
      "epoch": 3.36,
      "eval_loss": 0.1250012367963791,
      "eval_runtime": 532.3886,
      "eval_samples_per_second": 2.603,
      "eval_steps_per_second": 2.603,
      "eval_wer": 28.50962387427159,
      "step": 5000
    },
    {
      "epoch": 3.37,
      "grad_norm": 3.4317171573638916,
      "learning_rate": 9.545628140703518e-06,
      "loss": 0.1563,
      "step": 5025
    },
    {
      "epoch": 3.39,
      "grad_norm": 3.7062742710113525,
      "learning_rate": 9.543115577889448e-06,
      "loss": 0.1559,
      "step": 5050
    },
    {
      "epoch": 3.41,
      "grad_norm": 3.580521821975708,
      "learning_rate": 9.540603015075379e-06,
      "loss": 0.1521,
      "step": 5075
    },
    {
      "epoch": 3.43,
      "grad_norm": 3.364760160446167,
      "learning_rate": 9.538090452261308e-06,
      "loss": 0.1637,
      "step": 5100
    },
    {
      "epoch": 3.44,
      "grad_norm": 3.812782049179077,
      "learning_rate": 9.535577889447237e-06,
      "loss": 0.1564,
      "step": 5125
    },
    {
      "epoch": 3.46,
      "grad_norm": 3.065197229385376,
      "learning_rate": 9.533065326633166e-06,
      "loss": 0.1484,
      "step": 5150
    },
    {
      "epoch": 3.48,
      "grad_norm": 3.456214427947998,
      "learning_rate": 9.530552763819096e-06,
      "loss": 0.1517,
      "step": 5175
    },
    {
      "epoch": 3.49,
      "grad_norm": 3.731849193572998,
      "learning_rate": 9.528040201005025e-06,
      "loss": 0.1499,
      "step": 5200
    },
    {
      "epoch": 3.51,
      "grad_norm": 4.133150577545166,
      "learning_rate": 9.525527638190956e-06,
      "loss": 0.1475,
      "step": 5225
    },
    {
      "epoch": 3.53,
      "grad_norm": 3.353069543838501,
      "learning_rate": 9.523015075376885e-06,
      "loss": 0.1479,
      "step": 5250
    },
    {
      "epoch": 3.54,
      "grad_norm": 3.885782480239868,
      "learning_rate": 9.520502512562815e-06,
      "loss": 0.1458,
      "step": 5275
    },
    {
      "epoch": 3.56,
      "grad_norm": 3.5367889404296875,
      "learning_rate": 9.517989949748744e-06,
      "loss": 0.1531,
      "step": 5300
    },
    {
      "epoch": 3.58,
      "grad_norm": 3.8021907806396484,
      "learning_rate": 9.515477386934673e-06,
      "loss": 0.1498,
      "step": 5325
    },
    {
      "epoch": 3.59,
      "grad_norm": 3.308176279067993,
      "learning_rate": 9.512964824120604e-06,
      "loss": 0.1519,
      "step": 5350
    },
    {
      "epoch": 3.61,
      "grad_norm": 3.668410539627075,
      "learning_rate": 9.510452261306534e-06,
      "loss": 0.1526,
      "step": 5375
    },
    {
      "epoch": 3.63,
      "grad_norm": 3.228257417678833,
      "learning_rate": 9.507939698492463e-06,
      "loss": 0.1509,
      "step": 5400
    },
    {
      "epoch": 3.64,
      "grad_norm": 3.433962345123291,
      "learning_rate": 9.505427135678392e-06,
      "loss": 0.1521,
      "step": 5425
    },
    {
      "epoch": 3.66,
      "grad_norm": 3.707969903945923,
      "learning_rate": 9.502914572864322e-06,
      "loss": 0.1439,
      "step": 5450
    },
    {
      "epoch": 3.68,
      "grad_norm": 3.447314739227295,
      "learning_rate": 9.500402010050253e-06,
      "loss": 0.1473,
      "step": 5475
    },
    {
      "epoch": 3.69,
      "grad_norm": 3.579751968383789,
      "learning_rate": 9.497889447236182e-06,
      "loss": 0.1507,
      "step": 5500
    },
    {
      "epoch": 3.71,
      "grad_norm": 3.470454454421997,
      "learning_rate": 9.495376884422111e-06,
      "loss": 0.1473,
      "step": 5525
    },
    {
      "epoch": 3.73,
      "grad_norm": 3.2754967212677,
      "learning_rate": 9.49286432160804e-06,
      "loss": 0.1458,
      "step": 5550
    },
    {
      "epoch": 3.74,
      "grad_norm": 3.724622964859009,
      "learning_rate": 9.49035175879397e-06,
      "loss": 0.1474,
      "step": 5575
    },
    {
      "epoch": 3.76,
      "grad_norm": 3.176765203475952,
      "learning_rate": 9.4878391959799e-06,
      "loss": 0.1484,
      "step": 5600
    },
    {
      "epoch": 3.78,
      "grad_norm": 3.3496909141540527,
      "learning_rate": 9.48532663316583e-06,
      "loss": 0.1485,
      "step": 5625
    },
    {
      "epoch": 3.79,
      "grad_norm": 3.3814542293548584,
      "learning_rate": 9.48281407035176e-06,
      "loss": 0.1491,
      "step": 5650
    },
    {
      "epoch": 3.81,
      "grad_norm": 3.9236228466033936,
      "learning_rate": 9.480301507537689e-06,
      "loss": 0.1522,
      "step": 5675
    },
    {
      "epoch": 3.83,
      "grad_norm": 3.3441123962402344,
      "learning_rate": 9.47778894472362e-06,
      "loss": 0.1442,
      "step": 5700
    },
    {
      "epoch": 3.84,
      "grad_norm": 3.3952231407165527,
      "learning_rate": 9.475276381909548e-06,
      "loss": 0.1487,
      "step": 5725
    },
    {
      "epoch": 3.86,
      "grad_norm": 3.1410765647888184,
      "learning_rate": 9.472763819095479e-06,
      "loss": 0.1522,
      "step": 5750
    },
    {
      "epoch": 3.88,
      "grad_norm": 3.267335891723633,
      "learning_rate": 9.470251256281408e-06,
      "loss": 0.1425,
      "step": 5775
    },
    {
      "epoch": 3.9,
      "grad_norm": 3.547773838043213,
      "learning_rate": 9.467738693467337e-06,
      "loss": 0.1416,
      "step": 5800
    },
    {
      "epoch": 3.91,
      "grad_norm": 3.4462673664093018,
      "learning_rate": 9.465226130653267e-06,
      "loss": 0.1453,
      "step": 5825
    },
    {
      "epoch": 3.93,
      "grad_norm": 3.4584672451019287,
      "learning_rate": 9.462713567839196e-06,
      "loss": 0.1445,
      "step": 5850
    },
    {
      "epoch": 3.95,
      "grad_norm": 3.501403331756592,
      "learning_rate": 9.460201005025127e-06,
      "loss": 0.1423,
      "step": 5875
    },
    {
      "epoch": 3.96,
      "grad_norm": 3.912052631378174,
      "learning_rate": 9.457688442211056e-06,
      "loss": 0.1481,
      "step": 5900
    },
    {
      "epoch": 3.98,
      "grad_norm": 3.257798433303833,
      "learning_rate": 9.455175879396986e-06,
      "loss": 0.1385,
      "step": 5925
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.7633163928985596,
      "learning_rate": 9.452663316582915e-06,
      "loss": 0.1444,
      "step": 5950
    },
    {
      "epoch": 4.01,
      "grad_norm": 3.1884329319000244,
      "learning_rate": 9.450150753768846e-06,
      "loss": 0.1314,
      "step": 5975
    },
    {
      "epoch": 4.03,
      "grad_norm": 3.5105104446411133,
      "learning_rate": 9.447638190954774e-06,
      "loss": 0.1284,
      "step": 6000
    },
    {
      "epoch": 4.03,
      "eval_loss": 0.11968862265348434,
      "eval_runtime": 533.7582,
      "eval_samples_per_second": 2.597,
      "eval_steps_per_second": 2.597,
      "eval_wer": 27.644358114073814,
      "step": 6000
    },
    {
      "epoch": 4.05,
      "grad_norm": 3.4444563388824463,
      "learning_rate": 9.445125628140705e-06,
      "loss": 0.1278,
      "step": 6025
    },
    {
      "epoch": 4.06,
      "grad_norm": 3.446941375732422,
      "learning_rate": 9.442613065326634e-06,
      "loss": 0.1316,
      "step": 6050
    },
    {
      "epoch": 4.08,
      "grad_norm": 3.251770496368408,
      "learning_rate": 9.440100502512563e-06,
      "loss": 0.1289,
      "step": 6075
    },
    {
      "epoch": 4.1,
      "grad_norm": 3.1929450035095215,
      "learning_rate": 9.437587939698494e-06,
      "loss": 0.1257,
      "step": 6100
    },
    {
      "epoch": 4.11,
      "grad_norm": 3.137993097305298,
      "learning_rate": 9.435075376884422e-06,
      "loss": 0.1259,
      "step": 6125
    },
    {
      "epoch": 4.13,
      "grad_norm": 3.5924248695373535,
      "learning_rate": 9.432562814070353e-06,
      "loss": 0.1246,
      "step": 6150
    },
    {
      "epoch": 4.15,
      "grad_norm": 3.7657840251922607,
      "learning_rate": 9.430050251256282e-06,
      "loss": 0.1263,
      "step": 6175
    },
    {
      "epoch": 4.16,
      "grad_norm": 3.8803839683532715,
      "learning_rate": 9.427537688442212e-06,
      "loss": 0.1278,
      "step": 6200
    },
    {
      "epoch": 4.18,
      "grad_norm": 3.049147844314575,
      "learning_rate": 9.425025125628141e-06,
      "loss": 0.1248,
      "step": 6225
    },
    {
      "epoch": 4.2,
      "grad_norm": 3.5847809314727783,
      "learning_rate": 9.422512562814072e-06,
      "loss": 0.1326,
      "step": 6250
    },
    {
      "epoch": 4.21,
      "grad_norm": 3.208193063735962,
      "learning_rate": 9.42e-06,
      "loss": 0.1278,
      "step": 6275
    },
    {
      "epoch": 4.23,
      "grad_norm": 3.787940740585327,
      "learning_rate": 9.41748743718593e-06,
      "loss": 0.1286,
      "step": 6300
    },
    {
      "epoch": 4.25,
      "grad_norm": 2.801053762435913,
      "learning_rate": 9.41497487437186e-06,
      "loss": 0.1309,
      "step": 6325
    },
    {
      "epoch": 4.26,
      "grad_norm": 3.1014838218688965,
      "learning_rate": 9.41246231155779e-06,
      "loss": 0.1265,
      "step": 6350
    },
    {
      "epoch": 4.28,
      "grad_norm": 3.3319690227508545,
      "learning_rate": 9.40994974874372e-06,
      "loss": 0.1273,
      "step": 6375
    },
    {
      "epoch": 4.3,
      "grad_norm": 3.366464376449585,
      "learning_rate": 9.407437185929648e-06,
      "loss": 0.1266,
      "step": 6400
    },
    {
      "epoch": 4.31,
      "grad_norm": 3.5356907844543457,
      "learning_rate": 9.404924623115579e-06,
      "loss": 0.13,
      "step": 6425
    },
    {
      "epoch": 4.33,
      "grad_norm": 3.325680732727051,
      "learning_rate": 9.402412060301508e-06,
      "loss": 0.1262,
      "step": 6450
    },
    {
      "epoch": 4.35,
      "grad_norm": 3.4266843795776367,
      "learning_rate": 9.399899497487438e-06,
      "loss": 0.1284,
      "step": 6475
    },
    {
      "epoch": 4.37,
      "grad_norm": 3.1395492553710938,
      "learning_rate": 9.397386934673369e-06,
      "loss": 0.1276,
      "step": 6500
    },
    {
      "epoch": 4.38,
      "grad_norm": 3.323065757751465,
      "learning_rate": 9.394874371859298e-06,
      "loss": 0.1246,
      "step": 6525
    },
    {
      "epoch": 4.4,
      "grad_norm": 3.3577070236206055,
      "learning_rate": 9.392361809045227e-06,
      "loss": 0.1326,
      "step": 6550
    },
    {
      "epoch": 4.42,
      "grad_norm": 3.4483211040496826,
      "learning_rate": 9.389849246231157e-06,
      "loss": 0.1287,
      "step": 6575
    },
    {
      "epoch": 4.43,
      "grad_norm": 3.939202308654785,
      "learning_rate": 9.387336683417086e-06,
      "loss": 0.1295,
      "step": 6600
    },
    {
      "epoch": 4.45,
      "grad_norm": 3.5882346630096436,
      "learning_rate": 9.384824120603015e-06,
      "loss": 0.1257,
      "step": 6625
    },
    {
      "epoch": 4.47,
      "grad_norm": 3.9268131256103516,
      "learning_rate": 9.382311557788946e-06,
      "loss": 0.1308,
      "step": 6650
    },
    {
      "epoch": 4.48,
      "grad_norm": 3.2181479930877686,
      "learning_rate": 9.379798994974874e-06,
      "loss": 0.1289,
      "step": 6675
    },
    {
      "epoch": 4.5,
      "grad_norm": 3.4467923641204834,
      "learning_rate": 9.377286432160805e-06,
      "loss": 0.1286,
      "step": 6700
    },
    {
      "epoch": 4.52,
      "grad_norm": 3.227398157119751,
      "learning_rate": 9.374773869346734e-06,
      "loss": 0.1279,
      "step": 6725
    },
    {
      "epoch": 4.53,
      "grad_norm": 3.5086820125579834,
      "learning_rate": 9.372261306532664e-06,
      "loss": 0.1295,
      "step": 6750
    },
    {
      "epoch": 4.55,
      "grad_norm": 3.0712101459503174,
      "learning_rate": 9.369748743718595e-06,
      "loss": 0.1285,
      "step": 6775
    },
    {
      "epoch": 4.57,
      "grad_norm": 3.3961784839630127,
      "learning_rate": 9.367236180904524e-06,
      "loss": 0.1311,
      "step": 6800
    },
    {
      "epoch": 4.58,
      "grad_norm": 3.6800429821014404,
      "learning_rate": 9.364723618090453e-06,
      "loss": 0.1284,
      "step": 6825
    },
    {
      "epoch": 4.6,
      "grad_norm": 3.6793227195739746,
      "learning_rate": 9.362211055276383e-06,
      "loss": 0.1311,
      "step": 6850
    },
    {
      "epoch": 4.62,
      "grad_norm": 3.1020681858062744,
      "learning_rate": 9.359698492462312e-06,
      "loss": 0.1287,
      "step": 6875
    },
    {
      "epoch": 4.63,
      "grad_norm": 3.738802909851074,
      "learning_rate": 9.357185929648241e-06,
      "loss": 0.1241,
      "step": 6900
    },
    {
      "epoch": 4.65,
      "grad_norm": 3.344667911529541,
      "learning_rate": 9.354673366834172e-06,
      "loss": 0.1247,
      "step": 6925
    },
    {
      "epoch": 4.67,
      "grad_norm": 3.011655330657959,
      "learning_rate": 9.352160804020101e-06,
      "loss": 0.1237,
      "step": 6950
    },
    {
      "epoch": 4.68,
      "grad_norm": 3.486971139907837,
      "learning_rate": 9.34964824120603e-06,
      "loss": 0.1275,
      "step": 6975
    },
    {
      "epoch": 4.7,
      "grad_norm": 3.265568971633911,
      "learning_rate": 9.34713567839196e-06,
      "loss": 0.1216,
      "step": 7000
    },
    {
      "epoch": 4.7,
      "eval_loss": 0.11660390347242355,
      "eval_runtime": 534.0823,
      "eval_samples_per_second": 2.595,
      "eval_steps_per_second": 2.595,
      "eval_wer": 26.823238566131025,
      "step": 7000
    },
    {
      "epoch": 4.72,
      "grad_norm": 3.0894901752471924,
      "learning_rate": 9.34462311557789e-06,
      "loss": 0.1258,
      "step": 7025
    },
    {
      "epoch": 4.73,
      "grad_norm": 3.5530054569244385,
      "learning_rate": 9.34211055276382e-06,
      "loss": 0.1294,
      "step": 7050
    },
    {
      "epoch": 4.75,
      "grad_norm": 3.127763271331787,
      "learning_rate": 9.33959798994975e-06,
      "loss": 0.129,
      "step": 7075
    },
    {
      "epoch": 4.77,
      "grad_norm": 3.453204393386841,
      "learning_rate": 9.337085427135679e-06,
      "loss": 0.1264,
      "step": 7100
    },
    {
      "epoch": 4.79,
      "grad_norm": 3.470991611480713,
      "learning_rate": 9.334572864321608e-06,
      "loss": 0.1272,
      "step": 7125
    },
    {
      "epoch": 4.8,
      "grad_norm": 3.498213768005371,
      "learning_rate": 9.332060301507538e-06,
      "loss": 0.1257,
      "step": 7150
    },
    {
      "epoch": 4.82,
      "grad_norm": 3.052225351333618,
      "learning_rate": 9.329547738693469e-06,
      "loss": 0.1242,
      "step": 7175
    },
    {
      "epoch": 4.84,
      "grad_norm": 2.9512875080108643,
      "learning_rate": 9.327035175879398e-06,
      "loss": 0.1226,
      "step": 7200
    },
    {
      "epoch": 4.85,
      "grad_norm": 3.124257802963257,
      "learning_rate": 9.324522613065327e-06,
      "loss": 0.1276,
      "step": 7225
    },
    {
      "epoch": 4.87,
      "grad_norm": 3.763948678970337,
      "learning_rate": 9.322010050251257e-06,
      "loss": 0.123,
      "step": 7250
    },
    {
      "epoch": 4.89,
      "grad_norm": 3.859360694885254,
      "learning_rate": 9.319497487437186e-06,
      "loss": 0.1288,
      "step": 7275
    },
    {
      "epoch": 4.9,
      "grad_norm": 3.406261682510376,
      "learning_rate": 9.316984924623115e-06,
      "loss": 0.1268,
      "step": 7300
    },
    {
      "epoch": 4.92,
      "grad_norm": 3.5981762409210205,
      "learning_rate": 9.314472361809046e-06,
      "loss": 0.1267,
      "step": 7325
    },
    {
      "epoch": 4.94,
      "grad_norm": 3.2677414417266846,
      "learning_rate": 9.311959798994976e-06,
      "loss": 0.1228,
      "step": 7350
    },
    {
      "epoch": 4.95,
      "grad_norm": 3.4176025390625,
      "learning_rate": 9.309447236180905e-06,
      "loss": 0.1292,
      "step": 7375
    },
    {
      "epoch": 4.97,
      "grad_norm": 3.702085018157959,
      "learning_rate": 9.306934673366836e-06,
      "loss": 0.1212,
      "step": 7400
    },
    {
      "epoch": 4.99,
      "grad_norm": 3.075143337249756,
      "learning_rate": 9.304422110552764e-06,
      "loss": 0.1208,
      "step": 7425
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.96437406539917,
      "learning_rate": 9.301909547738695e-06,
      "loss": 0.1223,
      "step": 7450
    },
    {
      "epoch": 5.02,
      "grad_norm": 3.359867572784424,
      "learning_rate": 9.299396984924624e-06,
      "loss": 0.1083,
      "step": 7475
    },
    {
      "epoch": 5.04,
      "grad_norm": 3.1340601444244385,
      "learning_rate": 9.296884422110553e-06,
      "loss": 0.1088,
      "step": 7500
    },
    {
      "epoch": 5.05,
      "grad_norm": 3.4933323860168457,
      "learning_rate": 9.294371859296483e-06,
      "loss": 0.1103,
      "step": 7525
    },
    {
      "epoch": 5.07,
      "grad_norm": 2.8419055938720703,
      "learning_rate": 9.291859296482412e-06,
      "loss": 0.1074,
      "step": 7550
    },
    {
      "epoch": 5.09,
      "grad_norm": 2.699908971786499,
      "learning_rate": 9.289346733668343e-06,
      "loss": 0.1074,
      "step": 7575
    },
    {
      "epoch": 5.1,
      "grad_norm": 3.4752280712127686,
      "learning_rate": 9.286834170854272e-06,
      "loss": 0.1074,
      "step": 7600
    },
    {
      "epoch": 5.12,
      "grad_norm": 3.5037472248077393,
      "learning_rate": 9.284321608040202e-06,
      "loss": 0.1114,
      "step": 7625
    },
    {
      "epoch": 5.14,
      "grad_norm": 3.3195717334747314,
      "learning_rate": 9.281809045226131e-06,
      "loss": 0.1111,
      "step": 7650
    },
    {
      "epoch": 5.15,
      "grad_norm": 3.210256338119507,
      "learning_rate": 9.279296482412062e-06,
      "loss": 0.1095,
      "step": 7675
    },
    {
      "epoch": 5.17,
      "grad_norm": 3.4619410037994385,
      "learning_rate": 9.27678391959799e-06,
      "loss": 0.1118,
      "step": 7700
    },
    {
      "epoch": 5.19,
      "grad_norm": 3.2132604122161865,
      "learning_rate": 9.27427135678392e-06,
      "loss": 0.1065,
      "step": 7725
    },
    {
      "epoch": 5.2,
      "grad_norm": 3.379657030105591,
      "learning_rate": 9.27175879396985e-06,
      "loss": 0.1121,
      "step": 7750
    },
    {
      "epoch": 5.22,
      "grad_norm": 3.6748008728027344,
      "learning_rate": 9.26924623115578e-06,
      "loss": 0.1062,
      "step": 7775
    },
    {
      "epoch": 5.24,
      "grad_norm": 3.063694715499878,
      "learning_rate": 9.26673366834171e-06,
      "loss": 0.1077,
      "step": 7800
    },
    {
      "epoch": 5.26,
      "grad_norm": 3.237032413482666,
      "learning_rate": 9.264221105527638e-06,
      "loss": 0.1079,
      "step": 7825
    },
    {
      "epoch": 5.27,
      "grad_norm": 3.2364072799682617,
      "learning_rate": 9.261708542713569e-06,
      "loss": 0.1075,
      "step": 7850
    },
    {
      "epoch": 5.29,
      "grad_norm": 3.2186496257781982,
      "learning_rate": 9.259195979899498e-06,
      "loss": 0.107,
      "step": 7875
    },
    {
      "epoch": 5.31,
      "grad_norm": 3.249338150024414,
      "learning_rate": 9.256683417085428e-06,
      "loss": 0.1101,
      "step": 7900
    },
    {
      "epoch": 5.32,
      "grad_norm": 2.9037253856658936,
      "learning_rate": 9.254170854271357e-06,
      "loss": 0.1049,
      "step": 7925
    },
    {
      "epoch": 5.34,
      "grad_norm": 3.467984914779663,
      "learning_rate": 9.251658291457288e-06,
      "loss": 0.1164,
      "step": 7950
    },
    {
      "epoch": 5.36,
      "grad_norm": 3.047340154647827,
      "learning_rate": 9.249145728643217e-06,
      "loss": 0.1079,
      "step": 7975
    },
    {
      "epoch": 5.37,
      "grad_norm": 3.2782435417175293,
      "learning_rate": 9.246633165829147e-06,
      "loss": 0.1063,
      "step": 8000
    },
    {
      "epoch": 5.37,
      "eval_loss": 0.11793605983257294,
      "eval_runtime": 535.0689,
      "eval_samples_per_second": 2.59,
      "eval_steps_per_second": 2.59,
      "eval_wer": 27.19406674907293,
      "step": 8000
    },
    {
      "epoch": 5.39,
      "grad_norm": 3.045055389404297,
      "learning_rate": 9.244120603015076e-06,
      "loss": 0.1068,
      "step": 8025
    },
    {
      "epoch": 5.41,
      "grad_norm": 3.3729374408721924,
      "learning_rate": 9.241608040201005e-06,
      "loss": 0.1097,
      "step": 8050
    },
    {
      "epoch": 5.42,
      "grad_norm": 3.581709861755371,
      "learning_rate": 9.239095477386936e-06,
      "loss": 0.109,
      "step": 8075
    },
    {
      "epoch": 5.44,
      "grad_norm": 3.690354585647583,
      "learning_rate": 9.236582914572864e-06,
      "loss": 0.1105,
      "step": 8100
    },
    {
      "epoch": 5.46,
      "grad_norm": 3.4395689964294434,
      "learning_rate": 9.234070351758795e-06,
      "loss": 0.1047,
      "step": 8125
    },
    {
      "epoch": 5.47,
      "grad_norm": 3.5904619693756104,
      "learning_rate": 9.231557788944724e-06,
      "loss": 0.1098,
      "step": 8150
    },
    {
      "epoch": 5.49,
      "grad_norm": 3.4449338912963867,
      "learning_rate": 9.229045226130654e-06,
      "loss": 0.1094,
      "step": 8175
    },
    {
      "epoch": 5.51,
      "grad_norm": 3.081770181655884,
      "learning_rate": 9.226532663316585e-06,
      "loss": 0.1046,
      "step": 8200
    },
    {
      "epoch": 5.52,
      "grad_norm": 3.2109663486480713,
      "learning_rate": 9.224020100502514e-06,
      "loss": 0.11,
      "step": 8225
    },
    {
      "epoch": 5.54,
      "grad_norm": 3.8002219200134277,
      "learning_rate": 9.221507537688443e-06,
      "loss": 0.1103,
      "step": 8250
    },
    {
      "epoch": 5.56,
      "grad_norm": 3.423508644104004,
      "learning_rate": 9.218994974874373e-06,
      "loss": 0.1046,
      "step": 8275
    },
    {
      "epoch": 5.57,
      "grad_norm": 3.408816337585449,
      "learning_rate": 9.216482412060302e-06,
      "loss": 0.1137,
      "step": 8300
    },
    {
      "epoch": 5.59,
      "grad_norm": 3.349015235900879,
      "learning_rate": 9.213969849246231e-06,
      "loss": 0.1063,
      "step": 8325
    },
    {
      "epoch": 5.61,
      "grad_norm": 3.255462169647217,
      "learning_rate": 9.211457286432162e-06,
      "loss": 0.1081,
      "step": 8350
    },
    {
      "epoch": 5.62,
      "grad_norm": 3.0760374069213867,
      "learning_rate": 9.20894472361809e-06,
      "loss": 0.1124,
      "step": 8375
    },
    {
      "epoch": 5.64,
      "grad_norm": 3.469221830368042,
      "learning_rate": 9.206432160804021e-06,
      "loss": 0.1095,
      "step": 8400
    },
    {
      "epoch": 5.66,
      "grad_norm": 3.20563006401062,
      "learning_rate": 9.20391959798995e-06,
      "loss": 0.1091,
      "step": 8425
    },
    {
      "epoch": 5.67,
      "grad_norm": 3.58202862739563,
      "learning_rate": 9.20140703517588e-06,
      "loss": 0.1084,
      "step": 8450
    },
    {
      "epoch": 5.69,
      "grad_norm": 3.2912611961364746,
      "learning_rate": 9.19889447236181e-06,
      "loss": 0.109,
      "step": 8475
    },
    {
      "epoch": 5.71,
      "grad_norm": 3.2603135108947754,
      "learning_rate": 9.19638190954774e-06,
      "loss": 0.1051,
      "step": 8500
    },
    {
      "epoch": 5.73,
      "grad_norm": 3.3398075103759766,
      "learning_rate": 9.19386934673367e-06,
      "loss": 0.1086,
      "step": 8525
    },
    {
      "epoch": 5.74,
      "grad_norm": 3.480815887451172,
      "learning_rate": 9.191356783919599e-06,
      "loss": 0.1059,
      "step": 8550
    },
    {
      "epoch": 5.76,
      "grad_norm": 3.1898598670959473,
      "learning_rate": 9.188844221105528e-06,
      "loss": 0.104,
      "step": 8575
    },
    {
      "epoch": 5.78,
      "grad_norm": 3.3440845012664795,
      "learning_rate": 9.186331658291459e-06,
      "loss": 0.1126,
      "step": 8600
    },
    {
      "epoch": 5.79,
      "grad_norm": 3.6762936115264893,
      "learning_rate": 9.183819095477388e-06,
      "loss": 0.1064,
      "step": 8625
    },
    {
      "epoch": 5.81,
      "grad_norm": 3.66489315032959,
      "learning_rate": 9.181306532663317e-06,
      "loss": 0.1096,
      "step": 8650
    },
    {
      "epoch": 5.83,
      "grad_norm": 3.3494789600372314,
      "learning_rate": 9.178793969849247e-06,
      "loss": 0.1111,
      "step": 8675
    },
    {
      "epoch": 5.84,
      "grad_norm": 3.4388012886047363,
      "learning_rate": 9.176281407035176e-06,
      "loss": 0.1107,
      "step": 8700
    },
    {
      "epoch": 5.86,
      "grad_norm": 3.2451605796813965,
      "learning_rate": 9.173768844221105e-06,
      "loss": 0.1089,
      "step": 8725
    },
    {
      "epoch": 5.88,
      "grad_norm": 3.6606147289276123,
      "learning_rate": 9.171256281407036e-06,
      "loss": 0.1084,
      "step": 8750
    },
    {
      "epoch": 5.89,
      "grad_norm": 3.2145121097564697,
      "learning_rate": 9.168743718592966e-06,
      "loss": 0.1063,
      "step": 8775
    },
    {
      "epoch": 5.91,
      "grad_norm": 3.6518869400024414,
      "learning_rate": 9.166231155778895e-06,
      "loss": 0.1043,
      "step": 8800
    },
    {
      "epoch": 5.93,
      "grad_norm": 2.9405784606933594,
      "learning_rate": 9.163718592964826e-06,
      "loss": 0.1117,
      "step": 8825
    },
    {
      "epoch": 5.94,
      "grad_norm": 3.5626678466796875,
      "learning_rate": 9.161206030150754e-06,
      "loss": 0.1048,
      "step": 8850
    },
    {
      "epoch": 5.96,
      "grad_norm": 3.2351441383361816,
      "learning_rate": 9.158693467336685e-06,
      "loss": 0.1093,
      "step": 8875
    },
    {
      "epoch": 5.98,
      "grad_norm": 3.439530372619629,
      "learning_rate": 9.156180904522614e-06,
      "loss": 0.1073,
      "step": 8900
    },
    {
      "epoch": 5.99,
      "grad_norm": 3.4655144214630127,
      "learning_rate": 9.153668341708543e-06,
      "loss": 0.1067,
      "step": 8925
    },
    {
      "epoch": 6.01,
      "grad_norm": 2.794478178024292,
      "learning_rate": 9.151155778894473e-06,
      "loss": 0.0995,
      "step": 8950
    },
    {
      "epoch": 6.03,
      "grad_norm": 3.5291810035705566,
      "learning_rate": 9.148643216080402e-06,
      "loss": 0.0925,
      "step": 8975
    },
    {
      "epoch": 6.04,
      "grad_norm": 2.9376721382141113,
      "learning_rate": 9.146130653266331e-06,
      "loss": 0.0879,
      "step": 9000
    },
    {
      "epoch": 6.04,
      "eval_loss": 0.1166120246052742,
      "eval_runtime": 531.7233,
      "eval_samples_per_second": 2.607,
      "eval_steps_per_second": 2.607,
      "eval_wer": 26.876214020837015,
      "step": 9000
    },
    {
      "epoch": 6.06,
      "grad_norm": 3.121159076690674,
      "learning_rate": 9.143618090452262e-06,
      "loss": 0.0925,
      "step": 9025
    },
    {
      "epoch": 6.08,
      "grad_norm": 3.190279722213745,
      "learning_rate": 9.141105527638192e-06,
      "loss": 0.0966,
      "step": 9050
    },
    {
      "epoch": 6.09,
      "grad_norm": 2.9551713466644287,
      "learning_rate": 9.138592964824121e-06,
      "loss": 0.0927,
      "step": 9075
    },
    {
      "epoch": 6.11,
      "grad_norm": 2.6916284561157227,
      "learning_rate": 9.136080402010052e-06,
      "loss": 0.0905,
      "step": 9100
    },
    {
      "epoch": 6.13,
      "grad_norm": 3.1297528743743896,
      "learning_rate": 9.13356783919598e-06,
      "loss": 0.0966,
      "step": 9125
    },
    {
      "epoch": 6.15,
      "grad_norm": 3.3253660202026367,
      "learning_rate": 9.13105527638191e-06,
      "loss": 0.0977,
      "step": 9150
    },
    {
      "epoch": 6.16,
      "grad_norm": 3.1732029914855957,
      "learning_rate": 9.12854271356784e-06,
      "loss": 0.0943,
      "step": 9175
    },
    {
      "epoch": 6.18,
      "grad_norm": 3.00846791267395,
      "learning_rate": 9.12603015075377e-06,
      "loss": 0.095,
      "step": 9200
    },
    {
      "epoch": 6.2,
      "grad_norm": 3.4318153858184814,
      "learning_rate": 9.1235175879397e-06,
      "loss": 0.0918,
      "step": 9225
    },
    {
      "epoch": 6.21,
      "grad_norm": 2.615586519241333,
      "learning_rate": 9.121005025125628e-06,
      "loss": 0.0918,
      "step": 9250
    },
    {
      "epoch": 6.23,
      "grad_norm": 3.2654173374176025,
      "learning_rate": 9.118492462311559e-06,
      "loss": 0.0892,
      "step": 9275
    },
    {
      "epoch": 6.25,
      "grad_norm": 3.255948066711426,
      "learning_rate": 9.115979899497488e-06,
      "loss": 0.0955,
      "step": 9300
    },
    {
      "epoch": 6.26,
      "grad_norm": 3.593632221221924,
      "learning_rate": 9.113467336683418e-06,
      "loss": 0.0951,
      "step": 9325
    },
    {
      "epoch": 6.28,
      "grad_norm": 3.3398244380950928,
      "learning_rate": 9.110954773869347e-06,
      "loss": 0.0926,
      "step": 9350
    },
    {
      "epoch": 6.3,
      "grad_norm": 3.4789888858795166,
      "learning_rate": 9.108442211055278e-06,
      "loss": 0.0946,
      "step": 9375
    },
    {
      "epoch": 6.31,
      "grad_norm": 3.1585254669189453,
      "learning_rate": 9.105929648241206e-06,
      "loss": 0.0921,
      "step": 9400
    },
    {
      "epoch": 6.33,
      "grad_norm": 3.3125743865966797,
      "learning_rate": 9.103417085427137e-06,
      "loss": 0.0912,
      "step": 9425
    },
    {
      "epoch": 6.35,
      "grad_norm": 2.899616241455078,
      "learning_rate": 9.100904522613066e-06,
      "loss": 0.0935,
      "step": 9450
    },
    {
      "epoch": 6.36,
      "grad_norm": 2.9725539684295654,
      "learning_rate": 9.098391959798995e-06,
      "loss": 0.0934,
      "step": 9475
    },
    {
      "epoch": 6.38,
      "grad_norm": 3.340712070465088,
      "learning_rate": 9.095879396984926e-06,
      "loss": 0.0968,
      "step": 9500
    },
    {
      "epoch": 6.4,
      "grad_norm": 3.4166252613067627,
      "learning_rate": 9.093366834170854e-06,
      "loss": 0.0902,
      "step": 9525
    },
    {
      "epoch": 6.41,
      "grad_norm": 3.42030930519104,
      "learning_rate": 9.090854271356785e-06,
      "loss": 0.0903,
      "step": 9550
    },
    {
      "epoch": 6.43,
      "grad_norm": 2.913060188293457,
      "learning_rate": 9.088341708542714e-06,
      "loss": 0.0967,
      "step": 9575
    },
    {
      "epoch": 6.45,
      "grad_norm": 2.9808599948883057,
      "learning_rate": 9.085829145728644e-06,
      "loss": 0.0898,
      "step": 9600
    },
    {
      "epoch": 6.46,
      "grad_norm": 3.318812847137451,
      "learning_rate": 9.083316582914573e-06,
      "loss": 0.0964,
      "step": 9625
    },
    {
      "epoch": 6.48,
      "grad_norm": 2.8281571865081787,
      "learning_rate": 9.080804020100504e-06,
      "loss": 0.0925,
      "step": 9650
    },
    {
      "epoch": 6.5,
      "grad_norm": 3.3148748874664307,
      "learning_rate": 9.078291457286433e-06,
      "loss": 0.0955,
      "step": 9675
    },
    {
      "epoch": 6.51,
      "grad_norm": 3.047445297241211,
      "learning_rate": 9.075778894472363e-06,
      "loss": 0.0907,
      "step": 9700
    },
    {
      "epoch": 6.53,
      "grad_norm": 3.201747417449951,
      "learning_rate": 9.073266331658292e-06,
      "loss": 0.0891,
      "step": 9725
    },
    {
      "epoch": 6.55,
      "grad_norm": 3.5526840686798096,
      "learning_rate": 9.070753768844221e-06,
      "loss": 0.0937,
      "step": 9750
    },
    {
      "epoch": 6.56,
      "grad_norm": 3.3490021228790283,
      "learning_rate": 9.068241206030152e-06,
      "loss": 0.0924,
      "step": 9775
    },
    {
      "epoch": 6.58,
      "grad_norm": 3.195934534072876,
      "learning_rate": 9.06572864321608e-06,
      "loss": 0.0941,
      "step": 9800
    },
    {
      "epoch": 6.6,
      "grad_norm": 3.1133546829223633,
      "learning_rate": 9.063216080402011e-06,
      "loss": 0.093,
      "step": 9825
    },
    {
      "epoch": 6.62,
      "grad_norm": 3.5979671478271484,
      "learning_rate": 9.06070351758794e-06,
      "loss": 0.0881,
      "step": 9850
    },
    {
      "epoch": 6.63,
      "grad_norm": 3.7291669845581055,
      "learning_rate": 9.05819095477387e-06,
      "loss": 0.0955,
      "step": 9875
    },
    {
      "epoch": 6.65,
      "grad_norm": 3.2835400104522705,
      "learning_rate": 9.0556783919598e-06,
      "loss": 0.0902,
      "step": 9900
    },
    {
      "epoch": 6.67,
      "grad_norm": 3.1277029514312744,
      "learning_rate": 9.05316582914573e-06,
      "loss": 0.0936,
      "step": 9925
    },
    {
      "epoch": 6.68,
      "grad_norm": 3.2376766204833984,
      "learning_rate": 9.05065326633166e-06,
      "loss": 0.0896,
      "step": 9950
    },
    {
      "epoch": 6.7,
      "grad_norm": 2.698474168777466,
      "learning_rate": 9.048140703517589e-06,
      "loss": 0.0915,
      "step": 9975
    },
    {
      "epoch": 6.72,
      "grad_norm": 3.623647451400757,
      "learning_rate": 9.045628140703518e-06,
      "loss": 0.0924,
      "step": 10000
    },
    {
      "epoch": 6.72,
      "eval_loss": 0.1171552762389183,
      "eval_runtime": 533.9147,
      "eval_samples_per_second": 2.596,
      "eval_steps_per_second": 2.596,
      "eval_wer": 26.558361292601095,
      "step": 10000
    },
    {
      "epoch": 6.73,
      "grad_norm": 3.608774423599243,
      "learning_rate": 9.043115577889447e-06,
      "loss": 0.0927,
      "step": 10025
    },
    {
      "epoch": 6.75,
      "grad_norm": Infinity,
      "learning_rate": 9.04070351758794e-06,
      "loss": 0.0952,
      "step": 10050
    },
    {
      "epoch": 6.77,
      "grad_norm": 2.832880735397339,
      "learning_rate": 9.03819095477387e-06,
      "loss": 0.0909,
      "step": 10075
    },
    {
      "epoch": 6.78,
      "grad_norm": 3.0156736373901367,
      "learning_rate": 9.0356783919598e-06,
      "loss": 0.0944,
      "step": 10100
    },
    {
      "epoch": 6.8,
      "grad_norm": 3.3390650749206543,
      "learning_rate": 9.033165829145728e-06,
      "loss": 0.0919,
      "step": 10125
    },
    {
      "epoch": 6.82,
      "grad_norm": 3.394937515258789,
      "learning_rate": 9.03065326633166e-06,
      "loss": 0.0932,
      "step": 10150
    },
    {
      "epoch": 6.83,
      "grad_norm": 3.443366765975952,
      "learning_rate": 9.028140703517589e-06,
      "loss": 0.0934,
      "step": 10175
    },
    {
      "epoch": 6.85,
      "grad_norm": 3.167790174484253,
      "learning_rate": 9.025628140703518e-06,
      "loss": 0.0934,
      "step": 10200
    },
    {
      "epoch": 6.87,
      "grad_norm": 3.151536464691162,
      "learning_rate": 9.023115577889447e-06,
      "loss": 0.0935,
      "step": 10225
    },
    {
      "epoch": 6.88,
      "grad_norm": 3.475541114807129,
      "learning_rate": 9.020603015075378e-06,
      "loss": 0.0924,
      "step": 10250
    },
    {
      "epoch": 6.9,
      "grad_norm": 3.254150629043579,
      "learning_rate": 9.018090452261308e-06,
      "loss": 0.0946,
      "step": 10275
    },
    {
      "epoch": 6.92,
      "grad_norm": 3.126755714416504,
      "learning_rate": 9.015577889447237e-06,
      "loss": 0.0921,
      "step": 10300
    },
    {
      "epoch": 6.93,
      "grad_norm": 3.1626737117767334,
      "learning_rate": 9.013065326633166e-06,
      "loss": 0.0908,
      "step": 10325
    },
    {
      "epoch": 6.95,
      "grad_norm": 3.488074779510498,
      "learning_rate": 9.010552763819096e-06,
      "loss": 0.0956,
      "step": 10350
    },
    {
      "epoch": 6.97,
      "grad_norm": 3.0085911750793457,
      "learning_rate": 9.008040201005027e-06,
      "loss": 0.0915,
      "step": 10375
    },
    {
      "epoch": 6.98,
      "grad_norm": 3.424804925918579,
      "learning_rate": 9.005527638190954e-06,
      "loss": 0.0968,
      "step": 10400
    },
    {
      "epoch": 7.0,
      "grad_norm": 3.1618521213531494,
      "learning_rate": 9.003015075376885e-06,
      "loss": 0.0911,
      "step": 10425
    },
    {
      "epoch": 7.02,
      "grad_norm": 3.355823040008545,
      "learning_rate": 9.000502512562815e-06,
      "loss": 0.0775,
      "step": 10450
    },
    {
      "epoch": 7.03,
      "grad_norm": 2.7716736793518066,
      "learning_rate": 8.997989949748744e-06,
      "loss": 0.0776,
      "step": 10475
    },
    {
      "epoch": 7.05,
      "grad_norm": 2.89070987701416,
      "learning_rate": 8.995477386934675e-06,
      "loss": 0.0803,
      "step": 10500
    },
    {
      "epoch": 7.07,
      "grad_norm": 3.0273945331573486,
      "learning_rate": 8.992964824120604e-06,
      "loss": 0.0731,
      "step": 10525
    },
    {
      "epoch": 7.09,
      "grad_norm": 2.902979612350464,
      "learning_rate": 8.990452261306534e-06,
      "loss": 0.0805,
      "step": 10550
    },
    {
      "epoch": 7.1,
      "grad_norm": 2.9858810901641846,
      "learning_rate": 8.987939698492463e-06,
      "loss": 0.0761,
      "step": 10575
    },
    {
      "epoch": 7.12,
      "grad_norm": 2.780200958251953,
      "learning_rate": 8.985427135678392e-06,
      "loss": 0.0786,
      "step": 10600
    },
    {
      "epoch": 7.14,
      "grad_norm": 3.0452048778533936,
      "learning_rate": 8.982914572864322e-06,
      "loss": 0.078,
      "step": 10625
    },
    {
      "epoch": 7.15,
      "grad_norm": 3.0429253578186035,
      "learning_rate": 8.980402010050253e-06,
      "loss": 0.078,
      "step": 10650
    },
    {
      "epoch": 7.17,
      "grad_norm": 2.758443593978882,
      "learning_rate": 8.977889447236182e-06,
      "loss": 0.0787,
      "step": 10675
    },
    {
      "epoch": 7.19,
      "grad_norm": 3.3259782791137695,
      "learning_rate": 8.975376884422111e-06,
      "loss": 0.0818,
      "step": 10700
    },
    {
      "epoch": 7.2,
      "grad_norm": 3.1599812507629395,
      "learning_rate": 8.97286432160804e-06,
      "loss": 0.0788,
      "step": 10725
    },
    {
      "epoch": 7.22,
      "grad_norm": 3.163283348083496,
      "learning_rate": 8.97035175879397e-06,
      "loss": 0.0801,
      "step": 10750
    },
    {
      "epoch": 7.24,
      "grad_norm": 3.883058547973633,
      "learning_rate": 8.967839195979901e-06,
      "loss": 0.0818,
      "step": 10775
    },
    {
      "epoch": 7.25,
      "grad_norm": 3.0166139602661133,
      "learning_rate": 8.96532663316583e-06,
      "loss": 0.079,
      "step": 10800
    },
    {
      "epoch": 7.27,
      "grad_norm": 3.532127857208252,
      "learning_rate": 8.96281407035176e-06,
      "loss": 0.0764,
      "step": 10825
    },
    {
      "epoch": 7.29,
      "grad_norm": 2.8934993743896484,
      "learning_rate": 8.960301507537689e-06,
      "loss": 0.0791,
      "step": 10850
    },
    {
      "epoch": 7.3,
      "grad_norm": 3.4274938106536865,
      "learning_rate": 8.957788944723618e-06,
      "loss": 0.0788,
      "step": 10875
    },
    {
      "epoch": 7.32,
      "grad_norm": 2.964526891708374,
      "learning_rate": 8.95527638190955e-06,
      "loss": 0.0781,
      "step": 10900
    },
    {
      "epoch": 7.34,
      "grad_norm": 3.1131231784820557,
      "learning_rate": 8.952763819095479e-06,
      "loss": 0.0776,
      "step": 10925
    },
    {
      "epoch": 7.35,
      "grad_norm": 2.757322072982788,
      "learning_rate": 8.950251256281408e-06,
      "loss": 0.0793,
      "step": 10950
    },
    {
      "epoch": 7.37,
      "grad_norm": 2.8853962421417236,
      "learning_rate": 8.947738693467337e-06,
      "loss": 0.08,
      "step": 10975
    },
    {
      "epoch": 7.39,
      "grad_norm": 3.2388052940368652,
      "learning_rate": 8.945226130653267e-06,
      "loss": 0.0837,
      "step": 11000
    },
    {
      "epoch": 7.39,
      "eval_loss": 0.11983851343393326,
      "eval_runtime": 541.1838,
      "eval_samples_per_second": 2.561,
      "eval_steps_per_second": 2.561,
      "eval_wer": 27.052798869856964,
      "step": 11000
    },
    {
      "epoch": 7.4,
      "grad_norm": 3.431065559387207,
      "learning_rate": 8.942713567839196e-06,
      "loss": 0.0797,
      "step": 11025
    },
    {
      "epoch": 7.42,
      "grad_norm": 3.1514389514923096,
      "learning_rate": 8.940201005025127e-06,
      "loss": 0.0809,
      "step": 11050
    },
    {
      "epoch": 7.44,
      "grad_norm": 3.1348989009857178,
      "learning_rate": 8.937688442211056e-06,
      "loss": 0.0796,
      "step": 11075
    },
    {
      "epoch": 7.45,
      "grad_norm": 3.4892783164978027,
      "learning_rate": 8.935175879396986e-06,
      "loss": 0.0813,
      "step": 11100
    },
    {
      "epoch": 7.47,
      "grad_norm": 3.3423171043395996,
      "learning_rate": 8.932663316582915e-06,
      "loss": 0.0768,
      "step": 11125
    },
    {
      "epoch": 7.49,
      "grad_norm": 3.119539499282837,
      "learning_rate": 8.930150753768844e-06,
      "loss": 0.0833,
      "step": 11150
    },
    {
      "epoch": 7.51,
      "grad_norm": 3.181475877761841,
      "learning_rate": 8.927638190954775e-06,
      "loss": 0.0803,
      "step": 11175
    },
    {
      "epoch": 7.52,
      "grad_norm": 3.3543057441711426,
      "learning_rate": 8.925125628140705e-06,
      "loss": 0.0806,
      "step": 11200
    },
    {
      "epoch": 7.54,
      "grad_norm": 3.1575417518615723,
      "learning_rate": 8.922613065326634e-06,
      "loss": 0.0812,
      "step": 11225
    },
    {
      "epoch": 7.56,
      "grad_norm": 3.0198452472686768,
      "learning_rate": 8.920100502512563e-06,
      "loss": 0.0805,
      "step": 11250
    },
    {
      "epoch": 7.57,
      "grad_norm": 2.9735798835754395,
      "learning_rate": 8.917587939698493e-06,
      "loss": 0.0791,
      "step": 11275
    },
    {
      "epoch": 7.59,
      "grad_norm": 3.363503932952881,
      "learning_rate": 8.915075376884424e-06,
      "loss": 0.0817,
      "step": 11300
    },
    {
      "epoch": 7.61,
      "grad_norm": 3.10579514503479,
      "learning_rate": 8.912562814070353e-06,
      "loss": 0.0833,
      "step": 11325
    },
    {
      "epoch": 7.62,
      "grad_norm": 3.5427165031433105,
      "learning_rate": 8.910050251256282e-06,
      "loss": 0.0827,
      "step": 11350
    },
    {
      "epoch": 7.64,
      "grad_norm": 2.9739034175872803,
      "learning_rate": 8.907537688442212e-06,
      "loss": 0.0795,
      "step": 11375
    },
    {
      "epoch": 7.66,
      "grad_norm": 3.0262250900268555,
      "learning_rate": 8.905025125628143e-06,
      "loss": 0.0777,
      "step": 11400
    },
    {
      "epoch": 7.67,
      "grad_norm": 2.9359376430511475,
      "learning_rate": 8.90251256281407e-06,
      "loss": 0.0807,
      "step": 11425
    },
    {
      "epoch": 7.69,
      "grad_norm": 3.158572196960449,
      "learning_rate": 8.900000000000001e-06,
      "loss": 0.0772,
      "step": 11450
    },
    {
      "epoch": 7.71,
      "grad_norm": 3.330089807510376,
      "learning_rate": 8.89748743718593e-06,
      "loss": 0.0793,
      "step": 11475
    },
    {
      "epoch": 7.72,
      "grad_norm": 3.2174530029296875,
      "learning_rate": 8.89497487437186e-06,
      "loss": 0.079,
      "step": 11500
    },
    {
      "epoch": 7.74,
      "grad_norm": 3.673243522644043,
      "learning_rate": 8.892462311557791e-06,
      "loss": 0.0775,
      "step": 11525
    },
    {
      "epoch": 7.76,
      "grad_norm": 3.3094096183776855,
      "learning_rate": 8.889949748743718e-06,
      "loss": 0.078,
      "step": 11550
    },
    {
      "epoch": 7.77,
      "grad_norm": 3.426079273223877,
      "learning_rate": 8.88743718592965e-06,
      "loss": 0.0777,
      "step": 11575
    },
    {
      "epoch": 7.79,
      "grad_norm": 3.517086982727051,
      "learning_rate": 8.884924623115579e-06,
      "loss": 0.0776,
      "step": 11600
    },
    {
      "epoch": 7.81,
      "grad_norm": 2.9824516773223877,
      "learning_rate": 8.882412060301508e-06,
      "loss": 0.0805,
      "step": 11625
    },
    {
      "epoch": 7.82,
      "grad_norm": 2.965653896331787,
      "learning_rate": 8.879899497487437e-06,
      "loss": 0.0786,
      "step": 11650
    },
    {
      "epoch": 7.84,
      "grad_norm": 2.9882099628448486,
      "learning_rate": 8.877386934673368e-06,
      "loss": 0.0822,
      "step": 11675
    },
    {
      "epoch": 7.86,
      "grad_norm": 3.118823289871216,
      "learning_rate": 8.874874371859296e-06,
      "loss": 0.0773,
      "step": 11700
    },
    {
      "epoch": 7.87,
      "grad_norm": 4.2748188972473145,
      "learning_rate": 8.872361809045227e-06,
      "loss": 0.0812,
      "step": 11725
    },
    {
      "epoch": 7.89,
      "grad_norm": 3.5226612091064453,
      "learning_rate": 8.869849246231156e-06,
      "loss": 0.0801,
      "step": 11750
    },
    {
      "epoch": 7.91,
      "grad_norm": 3.2962095737457275,
      "learning_rate": 8.867336683417086e-06,
      "loss": 0.0779,
      "step": 11775
    },
    {
      "epoch": 7.92,
      "grad_norm": 3.037177801132202,
      "learning_rate": 8.864824120603017e-06,
      "loss": 0.0811,
      "step": 11800
    },
    {
      "epoch": 7.94,
      "grad_norm": 3.207000255584717,
      "learning_rate": 8.862311557788944e-06,
      "loss": 0.0813,
      "step": 11825
    },
    {
      "epoch": 7.96,
      "grad_norm": 3.5045995712280273,
      "learning_rate": 8.859798994974875e-06,
      "loss": 0.0793,
      "step": 11850
    },
    {
      "epoch": 7.98,
      "grad_norm": 2.9062917232513428,
      "learning_rate": 8.857286432160805e-06,
      "loss": 0.0808,
      "step": 11875
    },
    {
      "epoch": 7.99,
      "grad_norm": 3.086449146270752,
      "learning_rate": 8.854773869346734e-06,
      "loss": 0.0757,
      "step": 11900
    },
    {
      "epoch": 8.01,
      "grad_norm": 3.4503021240234375,
      "learning_rate": 8.852261306532665e-06,
      "loss": 0.0698,
      "step": 11925
    },
    {
      "epoch": 8.03,
      "grad_norm": 2.755633592605591,
      "learning_rate": 8.849748743718594e-06,
      "loss": 0.0619,
      "step": 11950
    },
    {
      "epoch": 8.04,
      "grad_norm": 3.3875789642333984,
      "learning_rate": 8.847236180904524e-06,
      "loss": 0.0651,
      "step": 11975
    },
    {
      "epoch": 8.06,
      "grad_norm": 2.697042465209961,
      "learning_rate": 8.844723618090453e-06,
      "loss": 0.0654,
      "step": 12000
    },
    {
      "epoch": 8.06,
      "eval_loss": 0.12158209830522537,
      "eval_runtime": 532.8467,
      "eval_samples_per_second": 2.601,
      "eval_steps_per_second": 2.601,
      "eval_wer": 26.328800988875155,
      "step": 12000
    },
    {
      "epoch": 8.08,
      "grad_norm": 2.8202855587005615,
      "learning_rate": 8.842211055276382e-06,
      "loss": 0.0658,
      "step": 12025
    },
    {
      "epoch": 8.09,
      "grad_norm": 2.7945172786712646,
      "learning_rate": 8.839698492462312e-06,
      "loss": 0.0627,
      "step": 12050
    },
    {
      "epoch": 8.11,
      "grad_norm": 3.1584692001342773,
      "learning_rate": 8.837185929648243e-06,
      "loss": 0.0673,
      "step": 12075
    },
    {
      "epoch": 8.13,
      "grad_norm": 3.1642470359802246,
      "learning_rate": 8.83467336683417e-06,
      "loss": 0.0683,
      "step": 12100
    },
    {
      "epoch": 8.14,
      "grad_norm": 2.9188601970672607,
      "learning_rate": 8.832160804020101e-06,
      "loss": 0.0682,
      "step": 12125
    },
    {
      "epoch": 8.16,
      "grad_norm": 3.276679039001465,
      "learning_rate": 8.829748743718593e-06,
      "loss": 0.0656,
      "step": 12150
    },
    {
      "epoch": 8.18,
      "grad_norm": 2.683711051940918,
      "learning_rate": 8.827236180904524e-06,
      "loss": 0.0625,
      "step": 12175
    },
    {
      "epoch": 8.19,
      "grad_norm": 3.232003688812256,
      "learning_rate": 8.824723618090453e-06,
      "loss": 0.066,
      "step": 12200
    },
    {
      "epoch": 8.21,
      "grad_norm": 2.7374961376190186,
      "learning_rate": 8.822211055276383e-06,
      "loss": 0.0647,
      "step": 12225
    },
    {
      "epoch": 8.23,
      "grad_norm": 3.423482656478882,
      "learning_rate": 8.819698492462312e-06,
      "loss": 0.0673,
      "step": 12250
    },
    {
      "epoch": 8.24,
      "grad_norm": 2.9813687801361084,
      "learning_rate": 8.817185929648241e-06,
      "loss": 0.0685,
      "step": 12275
    },
    {
      "epoch": 8.26,
      "grad_norm": 3.047753095626831,
      "learning_rate": 8.81467336683417e-06,
      "loss": 0.0658,
      "step": 12300
    },
    {
      "epoch": 8.28,
      "grad_norm": 3.4329652786254883,
      "learning_rate": 8.812160804020102e-06,
      "loss": 0.0662,
      "step": 12325
    },
    {
      "epoch": 8.29,
      "grad_norm": 3.080573081970215,
      "learning_rate": 8.809648241206031e-06,
      "loss": 0.0674,
      "step": 12350
    },
    {
      "epoch": 8.31,
      "grad_norm": 2.828704833984375,
      "learning_rate": 8.80713567839196e-06,
      "loss": 0.0694,
      "step": 12375
    },
    {
      "epoch": 8.33,
      "grad_norm": 3.132976531982422,
      "learning_rate": 8.804623115577891e-06,
      "loss": 0.0685,
      "step": 12400
    },
    {
      "epoch": 8.34,
      "grad_norm": 3.154456615447998,
      "learning_rate": 8.802110552763819e-06,
      "loss": 0.0679,
      "step": 12425
    },
    {
      "epoch": 8.36,
      "grad_norm": 3.4193313121795654,
      "learning_rate": 8.79959798994975e-06,
      "loss": 0.0674,
      "step": 12450
    },
    {
      "epoch": 8.38,
      "grad_norm": 3.2318356037139893,
      "learning_rate": 8.79708542713568e-06,
      "loss": 0.0658,
      "step": 12475
    },
    {
      "epoch": 8.39,
      "grad_norm": 2.9559836387634277,
      "learning_rate": 8.794572864321609e-06,
      "loss": 0.0647,
      "step": 12500
    },
    {
      "epoch": 8.41,
      "grad_norm": 3.459628105163574,
      "learning_rate": 8.792060301507538e-06,
      "loss": 0.0693,
      "step": 12525
    },
    {
      "epoch": 8.43,
      "grad_norm": 3.2934398651123047,
      "learning_rate": 8.789547738693467e-06,
      "loss": 0.0696,
      "step": 12550
    },
    {
      "epoch": 8.45,
      "grad_norm": 3.2100000381469727,
      "learning_rate": 8.787035175879398e-06,
      "loss": 0.0703,
      "step": 12575
    },
    {
      "epoch": 8.46,
      "grad_norm": 3.280884265899658,
      "learning_rate": 8.784522613065328e-06,
      "loss": 0.0661,
      "step": 12600
    },
    {
      "epoch": 8.48,
      "grad_norm": 3.1474897861480713,
      "learning_rate": 8.782010050251257e-06,
      "loss": 0.0663,
      "step": 12625
    },
    {
      "epoch": 8.5,
      "grad_norm": 2.9876487255096436,
      "learning_rate": 8.779497487437186e-06,
      "loss": 0.0693,
      "step": 12650
    },
    {
      "epoch": 8.51,
      "grad_norm": 3.278313159942627,
      "learning_rate": 8.776984924623117e-06,
      "loss": 0.0699,
      "step": 12675
    },
    {
      "epoch": 8.53,
      "grad_norm": 3.023169755935669,
      "learning_rate": 8.774472361809045e-06,
      "loss": 0.0712,
      "step": 12700
    },
    {
      "epoch": 8.55,
      "grad_norm": 3.168148994445801,
      "learning_rate": 8.771959798994976e-06,
      "loss": 0.0698,
      "step": 12725
    },
    {
      "epoch": 8.56,
      "grad_norm": 3.177262544631958,
      "learning_rate": 8.769447236180905e-06,
      "loss": 0.0686,
      "step": 12750
    },
    {
      "epoch": 8.58,
      "grad_norm": 3.1487865447998047,
      "learning_rate": 8.766934673366834e-06,
      "loss": 0.0684,
      "step": 12775
    },
    {
      "epoch": 8.6,
      "grad_norm": 2.9590165615081787,
      "learning_rate": 8.764422110552765e-06,
      "loss": 0.0691,
      "step": 12800
    },
    {
      "epoch": 8.61,
      "grad_norm": 3.0423812866210938,
      "learning_rate": 8.761909547738693e-06,
      "loss": 0.0682,
      "step": 12825
    },
    {
      "epoch": 8.63,
      "grad_norm": 3.3768019676208496,
      "learning_rate": 8.759396984924624e-06,
      "loss": 0.0709,
      "step": 12850
    },
    {
      "epoch": 8.65,
      "grad_norm": 3.7296512126922607,
      "learning_rate": 8.756884422110553e-06,
      "loss": 0.0701,
      "step": 12875
    },
    {
      "epoch": 8.66,
      "grad_norm": 3.148634433746338,
      "learning_rate": 8.754371859296483e-06,
      "loss": 0.0634,
      "step": 12900
    },
    {
      "epoch": 8.68,
      "grad_norm": 2.908444881439209,
      "learning_rate": 8.751859296482412e-06,
      "loss": 0.0659,
      "step": 12925
    },
    {
      "epoch": 8.7,
      "grad_norm": 3.3164865970611572,
      "learning_rate": 8.749346733668343e-06,
      "loss": 0.0655,
      "step": 12950
    },
    {
      "epoch": 8.71,
      "grad_norm": 2.9725685119628906,
      "learning_rate": 8.746834170854272e-06,
      "loss": 0.0659,
      "step": 12975
    },
    {
      "epoch": 8.73,
      "grad_norm": 3.171374797821045,
      "learning_rate": 8.744321608040202e-06,
      "loss": 0.068,
      "step": 13000
    },
    {
      "epoch": 8.73,
      "eval_loss": 0.12423743307590485,
      "eval_runtime": 533.8353,
      "eval_samples_per_second": 2.596,
      "eval_steps_per_second": 2.596,
      "eval_wer": 26.86738477838602,
      "step": 13000
    },
    {
      "epoch": 8.75,
      "grad_norm": 3.3160324096679688,
      "learning_rate": 8.741809045226131e-06,
      "loss": 0.0692,
      "step": 13025
    },
    {
      "epoch": 8.76,
      "grad_norm": 3.2802672386169434,
      "learning_rate": 8.73929648241206e-06,
      "loss": 0.067,
      "step": 13050
    },
    {
      "epoch": 8.78,
      "grad_norm": 3.2849535942077637,
      "learning_rate": 8.736783919597991e-06,
      "loss": 0.0654,
      "step": 13075
    },
    {
      "epoch": 8.8,
      "grad_norm": 3.685974359512329,
      "learning_rate": 8.734271356783919e-06,
      "loss": 0.0687,
      "step": 13100
    },
    {
      "epoch": 8.81,
      "grad_norm": 2.9581081867218018,
      "learning_rate": 8.73175879396985e-06,
      "loss": 0.0658,
      "step": 13125
    },
    {
      "epoch": 8.83,
      "grad_norm": 3.3408470153808594,
      "learning_rate": 8.72924623115578e-06,
      "loss": 0.0726,
      "step": 13150
    },
    {
      "epoch": 8.85,
      "grad_norm": 3.5375308990478516,
      "learning_rate": 8.726733668341709e-06,
      "loss": 0.0688,
      "step": 13175
    },
    {
      "epoch": 8.87,
      "grad_norm": 2.7572827339172363,
      "learning_rate": 8.72422110552764e-06,
      "loss": 0.0688,
      "step": 13200
    },
    {
      "epoch": 8.88,
      "grad_norm": 3.0948410034179688,
      "learning_rate": 8.721708542713569e-06,
      "loss": 0.0686,
      "step": 13225
    },
    {
      "epoch": 8.9,
      "grad_norm": 3.076904773712158,
      "learning_rate": 8.719195979899498e-06,
      "loss": 0.0683,
      "step": 13250
    },
    {
      "epoch": 8.92,
      "grad_norm": 3.060412645339966,
      "learning_rate": 8.716683417085428e-06,
      "loss": 0.0692,
      "step": 13275
    },
    {
      "epoch": 8.93,
      "grad_norm": 3.1852357387542725,
      "learning_rate": 8.714170854271357e-06,
      "loss": 0.0647,
      "step": 13300
    },
    {
      "epoch": 8.95,
      "grad_norm": 3.427971601486206,
      "learning_rate": 8.711658291457286e-06,
      "loss": 0.0675,
      "step": 13325
    },
    {
      "epoch": 8.97,
      "grad_norm": 3.221360683441162,
      "learning_rate": 8.709145728643217e-06,
      "loss": 0.0702,
      "step": 13350
    },
    {
      "epoch": 8.98,
      "grad_norm": 3.490898847579956,
      "learning_rate": 8.706633165829147e-06,
      "loss": 0.0693,
      "step": 13375
    },
    {
      "epoch": 9.0,
      "grad_norm": 3.1776282787323,
      "learning_rate": 8.704120603015076e-06,
      "loss": 0.0725,
      "step": 13400
    },
    {
      "epoch": 9.02,
      "grad_norm": 2.52174973487854,
      "learning_rate": 8.701608040201005e-06,
      "loss": 0.0542,
      "step": 13425
    },
    {
      "epoch": 9.03,
      "grad_norm": 2.8436169624328613,
      "learning_rate": 8.699095477386935e-06,
      "loss": 0.0543,
      "step": 13450
    },
    {
      "epoch": 9.05,
      "grad_norm": 3.0883164405822754,
      "learning_rate": 8.696582914572866e-06,
      "loss": 0.0565,
      "step": 13475
    },
    {
      "epoch": 9.07,
      "grad_norm": 3.2945592403411865,
      "learning_rate": 8.694070351758795e-06,
      "loss": 0.0554,
      "step": 13500
    },
    {
      "epoch": 9.08,
      "grad_norm": 3.1277835369110107,
      "learning_rate": 8.691557788944724e-06,
      "loss": 0.0575,
      "step": 13525
    },
    {
      "epoch": 9.1,
      "grad_norm": 2.555258274078369,
      "learning_rate": 8.689045226130654e-06,
      "loss": 0.0557,
      "step": 13550
    },
    {
      "epoch": 9.12,
      "grad_norm": 2.6981780529022217,
      "learning_rate": 8.686532663316583e-06,
      "loss": 0.056,
      "step": 13575
    },
    {
      "epoch": 9.13,
      "grad_norm": 2.9988884925842285,
      "learning_rate": 8.684020100502514e-06,
      "loss": 0.0575,
      "step": 13600
    },
    {
      "epoch": 9.15,
      "grad_norm": 2.7814390659332275,
      "learning_rate": 8.681507537688443e-06,
      "loss": 0.0543,
      "step": 13625
    },
    {
      "epoch": 9.17,
      "grad_norm": 2.8165695667266846,
      "learning_rate": 8.678994974874373e-06,
      "loss": 0.0542,
      "step": 13650
    },
    {
      "epoch": 9.18,
      "grad_norm": 2.8924388885498047,
      "learning_rate": 8.676482412060302e-06,
      "loss": 0.0584,
      "step": 13675
    },
    {
      "epoch": 9.2,
      "grad_norm": 2.8846709728240967,
      "learning_rate": 8.673969849246231e-06,
      "loss": 0.0546,
      "step": 13700
    },
    {
      "epoch": 9.22,
      "grad_norm": 3.0931618213653564,
      "learning_rate": 8.67145728643216e-06,
      "loss": 0.0541,
      "step": 13725
    },
    {
      "epoch": 9.23,
      "grad_norm": 3.0044896602630615,
      "learning_rate": 8.668944723618092e-06,
      "loss": 0.0566,
      "step": 13750
    },
    {
      "epoch": 9.25,
      "grad_norm": 2.992866039276123,
      "learning_rate": 8.666432160804021e-06,
      "loss": 0.0568,
      "step": 13775
    },
    {
      "epoch": 9.27,
      "grad_norm": 3.3243565559387207,
      "learning_rate": 8.66391959798995e-06,
      "loss": 0.0575,
      "step": 13800
    },
    {
      "epoch": 9.28,
      "grad_norm": 3.164736747741699,
      "learning_rate": 8.661407035175881e-06,
      "loss": 0.0565,
      "step": 13825
    },
    {
      "epoch": 9.3,
      "grad_norm": 2.89432430267334,
      "learning_rate": 8.658894472361809e-06,
      "loss": 0.0571,
      "step": 13850
    },
    {
      "epoch": 9.32,
      "grad_norm": 3.053514242172241,
      "learning_rate": 8.65638190954774e-06,
      "loss": 0.0582,
      "step": 13875
    },
    {
      "epoch": 9.34,
      "grad_norm": 2.7615840435028076,
      "learning_rate": 8.65386934673367e-06,
      "loss": 0.0566,
      "step": 13900
    },
    {
      "epoch": 9.35,
      "grad_norm": 3.1976537704467773,
      "learning_rate": 8.651356783919599e-06,
      "loss": 0.0578,
      "step": 13925
    },
    {
      "epoch": 9.37,
      "grad_norm": 3.1072587966918945,
      "learning_rate": 8.648844221105528e-06,
      "loss": 0.0577,
      "step": 13950
    },
    {
      "epoch": 9.39,
      "grad_norm": 3.4911906719207764,
      "learning_rate": 8.646331658291457e-06,
      "loss": 0.0548,
      "step": 13975
    },
    {
      "epoch": 9.4,
      "grad_norm": 2.923501968383789,
      "learning_rate": 8.643819095477388e-06,
      "loss": 0.0586,
      "step": 14000
    },
    {
      "epoch": 9.4,
      "eval_loss": 0.1282009482383728,
      "eval_runtime": 533.8178,
      "eval_samples_per_second": 2.596,
      "eval_steps_per_second": 2.596,
      "eval_wer": 26.982164930248985,
      "step": 14000
    },
    {
      "epoch": 9.42,
      "grad_norm": 3.0205700397491455,
      "learning_rate": 8.641306532663318e-06,
      "loss": 0.0552,
      "step": 14025
    },
    {
      "epoch": 9.44,
      "grad_norm": 3.022747278213501,
      "learning_rate": 8.638793969849247e-06,
      "loss": 0.0574,
      "step": 14050
    },
    {
      "epoch": 9.45,
      "grad_norm": 3.2978105545043945,
      "learning_rate": 8.636281407035176e-06,
      "loss": 0.0571,
      "step": 14075
    },
    {
      "epoch": 9.47,
      "grad_norm": 3.0741355419158936,
      "learning_rate": 8.633768844221107e-06,
      "loss": 0.0556,
      "step": 14100
    },
    {
      "epoch": 9.49,
      "grad_norm": 2.8877174854278564,
      "learning_rate": 8.631256281407035e-06,
      "loss": 0.0538,
      "step": 14125
    },
    {
      "epoch": 9.5,
      "grad_norm": 3.618729591369629,
      "learning_rate": 8.628743718592966e-06,
      "loss": 0.0592,
      "step": 14150
    },
    {
      "epoch": 9.52,
      "grad_norm": 3.005646228790283,
      "learning_rate": 8.626231155778895e-06,
      "loss": 0.057,
      "step": 14175
    },
    {
      "epoch": 9.54,
      "grad_norm": 3.3048083782196045,
      "learning_rate": 8.623718592964825e-06,
      "loss": 0.0564,
      "step": 14200
    },
    {
      "epoch": 9.55,
      "grad_norm": 3.2562224864959717,
      "learning_rate": 8.621206030150756e-06,
      "loss": 0.0571,
      "step": 14225
    },
    {
      "epoch": 9.57,
      "grad_norm": 2.980013608932495,
      "learning_rate": 8.618693467336683e-06,
      "loss": 0.0564,
      "step": 14250
    },
    {
      "epoch": 9.59,
      "grad_norm": 3.220036745071411,
      "learning_rate": 8.616180904522614e-06,
      "loss": 0.0588,
      "step": 14275
    },
    {
      "epoch": 9.6,
      "grad_norm": 3.4643850326538086,
      "learning_rate": 8.613668341708544e-06,
      "loss": 0.0565,
      "step": 14300
    },
    {
      "epoch": 9.62,
      "grad_norm": 3.2021632194519043,
      "learning_rate": 8.611155778894473e-06,
      "loss": 0.0586,
      "step": 14325
    },
    {
      "epoch": 9.64,
      "grad_norm": 3.2279539108276367,
      "learning_rate": 8.608643216080402e-06,
      "loss": 0.0562,
      "step": 14350
    },
    {
      "epoch": 9.65,
      "grad_norm": 3.429431438446045,
      "learning_rate": 8.606130653266333e-06,
      "loss": 0.0585,
      "step": 14375
    },
    {
      "epoch": 9.67,
      "grad_norm": 3.278526544570923,
      "learning_rate": 8.60361809045226e-06,
      "loss": 0.0584,
      "step": 14400
    },
    {
      "epoch": 9.69,
      "grad_norm": 3.5569005012512207,
      "learning_rate": 8.601105527638192e-06,
      "loss": 0.0587,
      "step": 14425
    },
    {
      "epoch": 9.7,
      "grad_norm": 3.0540413856506348,
      "learning_rate": 8.598592964824121e-06,
      "loss": 0.0582,
      "step": 14450
    },
    {
      "epoch": 9.72,
      "grad_norm": 2.9771244525909424,
      "learning_rate": 8.59608040201005e-06,
      "loss": 0.0544,
      "step": 14475
    },
    {
      "epoch": 9.74,
      "grad_norm": 3.271925926208496,
      "learning_rate": 8.593567839195981e-06,
      "loss": 0.0556,
      "step": 14500
    },
    {
      "epoch": 9.75,
      "grad_norm": 3.2107813358306885,
      "learning_rate": 8.591055276381909e-06,
      "loss": 0.0556,
      "step": 14525
    },
    {
      "epoch": 9.77,
      "grad_norm": 2.9411368370056152,
      "learning_rate": 8.58854271356784e-06,
      "loss": 0.06,
      "step": 14550
    },
    {
      "epoch": 9.79,
      "grad_norm": 2.9419991970062256,
      "learning_rate": 8.58603015075377e-06,
      "loss": 0.055,
      "step": 14575
    },
    {
      "epoch": 9.81,
      "grad_norm": 3.3104031085968018,
      "learning_rate": 8.583517587939699e-06,
      "loss": 0.0586,
      "step": 14600
    },
    {
      "epoch": 9.82,
      "grad_norm": 3.488868236541748,
      "learning_rate": 8.58100502512563e-06,
      "loss": 0.0608,
      "step": 14625
    },
    {
      "epoch": 9.84,
      "grad_norm": 2.7537827491760254,
      "learning_rate": 8.578492462311559e-06,
      "loss": 0.061,
      "step": 14650
    },
    {
      "epoch": 9.86,
      "grad_norm": 2.967761278152466,
      "learning_rate": 8.575979899497488e-06,
      "loss": 0.0616,
      "step": 14675
    },
    {
      "epoch": 9.87,
      "grad_norm": 2.6756021976470947,
      "learning_rate": 8.573467336683418e-06,
      "loss": 0.0572,
      "step": 14700
    },
    {
      "epoch": 9.89,
      "grad_norm": 3.6669530868530273,
      "learning_rate": 8.570954773869347e-06,
      "loss": 0.0545,
      "step": 14725
    },
    {
      "epoch": 9.91,
      "grad_norm": 3.402998208999634,
      "learning_rate": 8.568442211055276e-06,
      "loss": 0.0595,
      "step": 14750
    },
    {
      "epoch": 9.92,
      "grad_norm": 3.397134304046631,
      "learning_rate": 8.565929648241207e-06,
      "loss": 0.0582,
      "step": 14775
    },
    {
      "epoch": 9.94,
      "grad_norm": 3.193824291229248,
      "learning_rate": 8.563417085427135e-06,
      "loss": 0.0558,
      "step": 14800
    },
    {
      "epoch": 9.96,
      "grad_norm": 3.0948803424835205,
      "learning_rate": 8.560904522613066e-06,
      "loss": 0.0572,
      "step": 14825
    },
    {
      "epoch": 9.97,
      "grad_norm": 3.6509146690368652,
      "learning_rate": 8.558391959798995e-06,
      "loss": 0.0595,
      "step": 14850
    },
    {
      "epoch": 9.99,
      "grad_norm": 3.0662288665771484,
      "learning_rate": 8.555879396984925e-06,
      "loss": 0.057,
      "step": 14875
    },
    {
      "epoch": 10.01,
      "grad_norm": 2.2760088443756104,
      "learning_rate": 8.553366834170856e-06,
      "loss": 0.0524,
      "step": 14900
    },
    {
      "epoch": 10.02,
      "grad_norm": 2.8303427696228027,
      "learning_rate": 8.550854271356785e-06,
      "loss": 0.0494,
      "step": 14925
    },
    {
      "epoch": 10.04,
      "grad_norm": 3.1542868614196777,
      "learning_rate": 8.548341708542714e-06,
      "loss": 0.0445,
      "step": 14950
    },
    {
      "epoch": 10.06,
      "grad_norm": 2.8265697956085205,
      "learning_rate": 8.545829145728644e-06,
      "loss": 0.0464,
      "step": 14975
    },
    {
      "epoch": 10.07,
      "grad_norm": 3.163896322250366,
      "learning_rate": 8.543316582914573e-06,
      "loss": 0.047,
      "step": 15000
    },
    {
      "epoch": 10.07,
      "eval_loss": 0.13359740376472473,
      "eval_runtime": 533.7428,
      "eval_samples_per_second": 2.597,
      "eval_steps_per_second": 2.597,
      "eval_wer": 27.405968567896878,
      "step": 15000
    },
    {
      "epoch": 10.09,
      "grad_norm": 2.813354253768921,
      "learning_rate": 8.540804020100502e-06,
      "loss": 0.0476,
      "step": 15025
    },
    {
      "epoch": 10.11,
      "grad_norm": 2.448727607727051,
      "learning_rate": 8.538291457286433e-06,
      "loss": 0.0448,
      "step": 15050
    },
    {
      "epoch": 10.12,
      "grad_norm": 2.798645257949829,
      "learning_rate": 8.535778894472363e-06,
      "loss": 0.0458,
      "step": 15075
    },
    {
      "epoch": 10.14,
      "grad_norm": 2.969273090362549,
      "learning_rate": 8.533266331658292e-06,
      "loss": 0.0442,
      "step": 15100
    },
    {
      "epoch": 10.16,
      "grad_norm": 2.901127576828003,
      "learning_rate": 8.530753768844221e-06,
      "loss": 0.0431,
      "step": 15125
    },
    {
      "epoch": 10.17,
      "grad_norm": 3.0042836666107178,
      "learning_rate": 8.52824120603015e-06,
      "loss": 0.049,
      "step": 15150
    },
    {
      "epoch": 10.19,
      "grad_norm": 2.694744825363159,
      "learning_rate": 8.525728643216082e-06,
      "loss": 0.0474,
      "step": 15175
    },
    {
      "epoch": 10.21,
      "grad_norm": 2.79301118850708,
      "learning_rate": 8.523216080402011e-06,
      "loss": 0.0459,
      "step": 15200
    },
    {
      "epoch": 10.22,
      "grad_norm": 3.328848123550415,
      "learning_rate": 8.52070351758794e-06,
      "loss": 0.0481,
      "step": 15225
    },
    {
      "epoch": 10.24,
      "grad_norm": 3.0490903854370117,
      "learning_rate": 8.518190954773871e-06,
      "loss": 0.0467,
      "step": 15250
    },
    {
      "epoch": 10.26,
      "grad_norm": 2.891860246658325,
      "learning_rate": 8.515678391959799e-06,
      "loss": 0.0482,
      "step": 15275
    },
    {
      "epoch": 10.28,
      "grad_norm": 3.29339599609375,
      "learning_rate": 8.51316582914573e-06,
      "loss": 0.0468,
      "step": 15300
    },
    {
      "epoch": 10.29,
      "grad_norm": 2.871262550354004,
      "learning_rate": 8.51065326633166e-06,
      "loss": 0.0465,
      "step": 15325
    },
    {
      "epoch": 10.31,
      "grad_norm": 2.673008680343628,
      "learning_rate": 8.508140703517589e-06,
      "loss": 0.0457,
      "step": 15350
    },
    {
      "epoch": 10.33,
      "grad_norm": 2.5940115451812744,
      "learning_rate": 8.505628140703518e-06,
      "loss": 0.049,
      "step": 15375
    },
    {
      "epoch": 10.34,
      "grad_norm": 2.8226072788238525,
      "learning_rate": 8.503115577889447e-06,
      "loss": 0.0472,
      "step": 15400
    },
    {
      "epoch": 10.36,
      "grad_norm": 2.800179958343506,
      "learning_rate": 8.500603015075377e-06,
      "loss": 0.0477,
      "step": 15425
    },
    {
      "epoch": 10.38,
      "grad_norm": 3.0697898864746094,
      "learning_rate": 8.498090452261308e-06,
      "loss": 0.0448,
      "step": 15450
    },
    {
      "epoch": 10.39,
      "grad_norm": 2.9394161701202393,
      "learning_rate": 8.495577889447237e-06,
      "loss": 0.0464,
      "step": 15475
    },
    {
      "epoch": 10.41,
      "grad_norm": 3.055058479309082,
      "learning_rate": 8.493065326633166e-06,
      "loss": 0.0486,
      "step": 15500
    },
    {
      "epoch": 10.43,
      "grad_norm": 3.4436676502227783,
      "learning_rate": 8.490552763819097e-06,
      "loss": 0.0479,
      "step": 15525
    },
    {
      "epoch": 10.44,
      "grad_norm": 3.167590379714966,
      "learning_rate": 8.488040201005025e-06,
      "loss": 0.049,
      "step": 15550
    },
    {
      "epoch": 10.46,
      "grad_norm": 2.786879539489746,
      "learning_rate": 8.485527638190956e-06,
      "loss": 0.0476,
      "step": 15575
    },
    {
      "epoch": 10.48,
      "grad_norm": 3.0949158668518066,
      "learning_rate": 8.483015075376885e-06,
      "loss": 0.0463,
      "step": 15600
    },
    {
      "epoch": 10.49,
      "grad_norm": 3.426304340362549,
      "learning_rate": 8.480502512562815e-06,
      "loss": 0.0475,
      "step": 15625
    },
    {
      "epoch": 10.51,
      "grad_norm": 3.1173408031463623,
      "learning_rate": 8.477989949748744e-06,
      "loss": 0.0476,
      "step": 15650
    },
    {
      "epoch": 10.53,
      "grad_norm": 2.856600046157837,
      "learning_rate": 8.475477386934673e-06,
      "loss": 0.0471,
      "step": 15675
    },
    {
      "epoch": 10.54,
      "grad_norm": 3.2512564659118652,
      "learning_rate": 8.472964824120604e-06,
      "loss": 0.0483,
      "step": 15700
    },
    {
      "epoch": 10.56,
      "grad_norm": 3.3549506664276123,
      "learning_rate": 8.470452261306534e-06,
      "loss": 0.0462,
      "step": 15725
    },
    {
      "epoch": 10.58,
      "grad_norm": 2.7729334831237793,
      "learning_rate": 8.467939698492463e-06,
      "loss": 0.0472,
      "step": 15750
    },
    {
      "epoch": 10.59,
      "grad_norm": 2.711257219314575,
      "learning_rate": 8.465427135678392e-06,
      "loss": 0.0472,
      "step": 15775
    },
    {
      "epoch": 10.61,
      "grad_norm": 3.229771375656128,
      "learning_rate": 8.462914572864323e-06,
      "loss": 0.0479,
      "step": 15800
    },
    {
      "epoch": 10.63,
      "grad_norm": 3.0402400493621826,
      "learning_rate": 8.460402010050251e-06,
      "loss": 0.0503,
      "step": 15825
    },
    {
      "epoch": 10.64,
      "grad_norm": 2.9210867881774902,
      "learning_rate": 8.457989949748744e-06,
      "loss": 0.0497,
      "step": 15850
    },
    {
      "epoch": 10.66,
      "grad_norm": 3.3483831882476807,
      "learning_rate": 8.455577889447237e-06,
      "loss": 0.0475,
      "step": 15875
    },
    {
      "epoch": 10.68,
      "grad_norm": 3.053593873977661,
      "learning_rate": 8.453065326633167e-06,
      "loss": 0.046,
      "step": 15900
    },
    {
      "epoch": 10.7,
      "grad_norm": 3.136958599090576,
      "learning_rate": 8.450552763819096e-06,
      "loss": 0.0509,
      "step": 15925
    },
    {
      "epoch": 10.71,
      "grad_norm": 3.1040425300598145,
      "learning_rate": 8.448040201005025e-06,
      "loss": 0.0504,
      "step": 15950
    },
    {
      "epoch": 10.73,
      "grad_norm": 2.8489692211151123,
      "learning_rate": 8.445527638190956e-06,
      "loss": 0.0484,
      "step": 15975
    },
    {
      "epoch": 10.75,
      "grad_norm": 2.8868560791015625,
      "learning_rate": 8.443015075376884e-06,
      "loss": 0.0475,
      "step": 16000
    },
    {
      "epoch": 10.75,
      "eval_loss": 0.1362370103597641,
      "eval_runtime": 536.1147,
      "eval_samples_per_second": 2.585,
      "eval_steps_per_second": 2.585,
      "eval_wer": 27.441285537700864,
      "step": 16000
    },
    {
      "epoch": 10.76,
      "grad_norm": 3.188688039779663,
      "learning_rate": 8.440502512562815e-06,
      "loss": 0.0502,
      "step": 16025
    },
    {
      "epoch": 10.78,
      "grad_norm": 2.4469282627105713,
      "learning_rate": 8.437989949748744e-06,
      "loss": 0.0459,
      "step": 16050
    },
    {
      "epoch": 10.8,
      "grad_norm": 2.948697328567505,
      "learning_rate": 8.435477386934674e-06,
      "loss": 0.0472,
      "step": 16075
    },
    {
      "epoch": 10.81,
      "grad_norm": 3.236891508102417,
      "learning_rate": 8.432964824120605e-06,
      "loss": 0.0494,
      "step": 16100
    },
    {
      "epoch": 10.83,
      "grad_norm": 3.0507919788360596,
      "learning_rate": 8.430452261306534e-06,
      "loss": 0.0494,
      "step": 16125
    },
    {
      "epoch": 10.85,
      "grad_norm": 2.8577802181243896,
      "learning_rate": 8.427939698492463e-06,
      "loss": 0.0487,
      "step": 16150
    },
    {
      "epoch": 10.86,
      "grad_norm": 3.035109758377075,
      "learning_rate": 8.425427135678393e-06,
      "loss": 0.0486,
      "step": 16175
    },
    {
      "epoch": 10.88,
      "grad_norm": 3.5497820377349854,
      "learning_rate": 8.422914572864322e-06,
      "loss": 0.0497,
      "step": 16200
    },
    {
      "epoch": 10.9,
      "grad_norm": 2.838867664337158,
      "learning_rate": 8.420402010050251e-06,
      "loss": 0.0451,
      "step": 16225
    },
    {
      "epoch": 10.91,
      "grad_norm": 3.316819190979004,
      "learning_rate": 8.417889447236182e-06,
      "loss": 0.0489,
      "step": 16250
    },
    {
      "epoch": 10.93,
      "grad_norm": 3.3198862075805664,
      "learning_rate": 8.415376884422112e-06,
      "loss": 0.0528,
      "step": 16275
    },
    {
      "epoch": 10.95,
      "grad_norm": 3.4924492835998535,
      "learning_rate": 8.412864321608041e-06,
      "loss": 0.0492,
      "step": 16300
    },
    {
      "epoch": 10.96,
      "grad_norm": 3.0983831882476807,
      "learning_rate": 8.41035175879397e-06,
      "loss": 0.0498,
      "step": 16325
    },
    {
      "epoch": 10.98,
      "grad_norm": 3.4345991611480713,
      "learning_rate": 8.4078391959799e-06,
      "loss": 0.0483,
      "step": 16350
    },
    {
      "epoch": 11.0,
      "grad_norm": 3.294377326965332,
      "learning_rate": 8.40532663316583e-06,
      "loss": 0.0485,
      "step": 16375
    },
    {
      "epoch": 11.01,
      "grad_norm": 2.1766245365142822,
      "learning_rate": 8.40281407035176e-06,
      "loss": 0.0371,
      "step": 16400
    },
    {
      "epoch": 11.03,
      "grad_norm": 2.683638334274292,
      "learning_rate": 8.40030150753769e-06,
      "loss": 0.0355,
      "step": 16425
    },
    {
      "epoch": 11.05,
      "grad_norm": 2.8458847999572754,
      "learning_rate": 8.397788944723619e-06,
      "loss": 0.038,
      "step": 16450
    },
    {
      "epoch": 11.06,
      "grad_norm": 2.7042036056518555,
      "learning_rate": 8.395276381909548e-06,
      "loss": 0.0375,
      "step": 16475
    },
    {
      "epoch": 11.08,
      "grad_norm": 2.0865659713745117,
      "learning_rate": 8.392763819095479e-06,
      "loss": 0.039,
      "step": 16500
    },
    {
      "epoch": 11.1,
      "grad_norm": 2.3241260051727295,
      "learning_rate": 8.390251256281408e-06,
      "loss": 0.0365,
      "step": 16525
    },
    {
      "epoch": 11.11,
      "grad_norm": 2.7509355545043945,
      "learning_rate": 8.387738693467338e-06,
      "loss": 0.0392,
      "step": 16550
    },
    {
      "epoch": 11.13,
      "grad_norm": 2.3158955574035645,
      "learning_rate": 8.385226130653267e-06,
      "loss": 0.0399,
      "step": 16575
    },
    {
      "epoch": 11.15,
      "grad_norm": 2.368791103363037,
      "learning_rate": 8.382713567839196e-06,
      "loss": 0.0366,
      "step": 16600
    },
    {
      "epoch": 11.17,
      "grad_norm": 3.157816171646118,
      "learning_rate": 8.380201005025126e-06,
      "loss": 0.0386,
      "step": 16625
    },
    {
      "epoch": 11.18,
      "grad_norm": 2.391731023788452,
      "learning_rate": 8.377688442211057e-06,
      "loss": 0.0409,
      "step": 16650
    },
    {
      "epoch": 11.2,
      "grad_norm": 2.881032943725586,
      "learning_rate": 8.375175879396986e-06,
      "loss": 0.0399,
      "step": 16675
    },
    {
      "epoch": 11.22,
      "grad_norm": 2.8162527084350586,
      "learning_rate": 8.372663316582915e-06,
      "loss": 0.0386,
      "step": 16700
    },
    {
      "epoch": 11.23,
      "grad_norm": 2.798832654953003,
      "learning_rate": 8.370150753768845e-06,
      "loss": 0.0389,
      "step": 16725
    },
    {
      "epoch": 11.25,
      "grad_norm": 2.4073362350463867,
      "learning_rate": 8.367638190954774e-06,
      "loss": 0.038,
      "step": 16750
    },
    {
      "epoch": 11.27,
      "grad_norm": 3.539222002029419,
      "learning_rate": 8.365125628140705e-06,
      "loss": 0.0385,
      "step": 16775
    },
    {
      "epoch": 11.28,
      "grad_norm": 3.047471761703491,
      "learning_rate": 8.362613065326634e-06,
      "loss": 0.0386,
      "step": 16800
    },
    {
      "epoch": 11.3,
      "grad_norm": 2.62675142288208,
      "learning_rate": 8.360100502512563e-06,
      "loss": 0.0388,
      "step": 16825
    },
    {
      "epoch": 11.32,
      "grad_norm": 2.6403391361236572,
      "learning_rate": 8.357587939698493e-06,
      "loss": 0.041,
      "step": 16850
    },
    {
      "epoch": 11.33,
      "grad_norm": 2.7048850059509277,
      "learning_rate": 8.355075376884422e-06,
      "loss": 0.0405,
      "step": 16875
    },
    {
      "epoch": 11.35,
      "grad_norm": 2.8291220664978027,
      "learning_rate": 8.352562814070353e-06,
      "loss": 0.0375,
      "step": 16900
    },
    {
      "epoch": 11.37,
      "grad_norm": 2.9671170711517334,
      "learning_rate": 8.350050251256282e-06,
      "loss": 0.0377,
      "step": 16925
    },
    {
      "epoch": 11.38,
      "grad_norm": 3.0989413261413574,
      "learning_rate": 8.347537688442212e-06,
      "loss": 0.039,
      "step": 16950
    },
    {
      "epoch": 11.4,
      "grad_norm": 2.738807201385498,
      "learning_rate": 8.345025125628141e-06,
      "loss": 0.0399,
      "step": 16975
    },
    {
      "epoch": 11.42,
      "grad_norm": 2.9761691093444824,
      "learning_rate": 8.34251256281407e-06,
      "loss": 0.0402,
      "step": 17000
    },
    {
      "epoch": 11.42,
      "eval_loss": 0.13800786435604095,
      "eval_runtime": 531.8418,
      "eval_samples_per_second": 2.606,
      "eval_steps_per_second": 2.606,
      "eval_wer": 27.76796750838778,
      "step": 17000
    },
    {
      "epoch": 11.43,
      "grad_norm": 3.1192235946655273,
      "learning_rate": 8.34e-06,
      "loss": 0.0389,
      "step": 17025
    },
    {
      "epoch": 11.45,
      "grad_norm": 3.019216299057007,
      "learning_rate": 8.33748743718593e-06,
      "loss": 0.0413,
      "step": 17050
    },
    {
      "epoch": 11.47,
      "grad_norm": 2.6235885620117188,
      "learning_rate": 8.33497487437186e-06,
      "loss": 0.043,
      "step": 17075
    },
    {
      "epoch": 11.48,
      "grad_norm": 3.3072292804718018,
      "learning_rate": 8.33246231155779e-06,
      "loss": 0.0384,
      "step": 17100
    },
    {
      "epoch": 11.5,
      "grad_norm": 3.032578706741333,
      "learning_rate": 8.32994974874372e-06,
      "loss": 0.0394,
      "step": 17125
    },
    {
      "epoch": 11.52,
      "grad_norm": 3.0692577362060547,
      "learning_rate": 8.327437185929648e-06,
      "loss": 0.0402,
      "step": 17150
    },
    {
      "epoch": 11.53,
      "grad_norm": 3.113739252090454,
      "learning_rate": 8.324924623115579e-06,
      "loss": 0.038,
      "step": 17175
    },
    {
      "epoch": 11.55,
      "grad_norm": 3.1510965824127197,
      "learning_rate": 8.322412060301508e-06,
      "loss": 0.0423,
      "step": 17200
    },
    {
      "epoch": 11.57,
      "grad_norm": 3.110407590866089,
      "learning_rate": 8.319899497487438e-06,
      "loss": 0.0381,
      "step": 17225
    },
    {
      "epoch": 11.58,
      "grad_norm": 2.9603676795959473,
      "learning_rate": 8.317386934673367e-06,
      "loss": 0.0421,
      "step": 17250
    },
    {
      "epoch": 11.6,
      "grad_norm": 2.7330162525177,
      "learning_rate": 8.314874371859298e-06,
      "loss": 0.04,
      "step": 17275
    },
    {
      "epoch": 11.62,
      "grad_norm": 3.783348798751831,
      "learning_rate": 8.312361809045226e-06,
      "loss": 0.0428,
      "step": 17300
    },
    {
      "epoch": 11.64,
      "grad_norm": 3.3141326904296875,
      "learning_rate": 8.309849246231157e-06,
      "loss": 0.04,
      "step": 17325
    },
    {
      "epoch": 11.65,
      "grad_norm": 3.1341404914855957,
      "learning_rate": 8.307336683417086e-06,
      "loss": 0.0389,
      "step": 17350
    },
    {
      "epoch": 11.67,
      "grad_norm": 2.5702879428863525,
      "learning_rate": 8.304824120603015e-06,
      "loss": 0.0411,
      "step": 17375
    },
    {
      "epoch": 11.69,
      "grad_norm": 2.7597875595092773,
      "learning_rate": 8.302311557788946e-06,
      "loss": 0.0387,
      "step": 17400
    },
    {
      "epoch": 11.7,
      "grad_norm": 3.1602911949157715,
      "learning_rate": 8.299798994974874e-06,
      "loss": 0.0401,
      "step": 17425
    },
    {
      "epoch": 11.72,
      "grad_norm": 2.9719858169555664,
      "learning_rate": 8.297286432160805e-06,
      "loss": 0.04,
      "step": 17450
    },
    {
      "epoch": 11.74,
      "grad_norm": 2.7361767292022705,
      "learning_rate": 8.294773869346734e-06,
      "loss": 0.041,
      "step": 17475
    },
    {
      "epoch": 11.75,
      "grad_norm": 2.7034785747528076,
      "learning_rate": 8.292261306532664e-06,
      "loss": 0.0413,
      "step": 17500
    },
    {
      "epoch": 11.77,
      "grad_norm": 3.2431066036224365,
      "learning_rate": 8.289748743718595e-06,
      "loss": 0.0396,
      "step": 17525
    },
    {
      "epoch": 11.79,
      "grad_norm": 2.7960753440856934,
      "learning_rate": 8.287236180904524e-06,
      "loss": 0.0406,
      "step": 17550
    },
    {
      "epoch": 11.8,
      "grad_norm": 3.0115575790405273,
      "learning_rate": 8.284723618090453e-06,
      "loss": 0.0395,
      "step": 17575
    },
    {
      "epoch": 11.82,
      "grad_norm": 2.4014508724212646,
      "learning_rate": 8.282211055276383e-06,
      "loss": 0.0404,
      "step": 17600
    },
    {
      "epoch": 11.84,
      "grad_norm": 3.1004748344421387,
      "learning_rate": 8.279698492462312e-06,
      "loss": 0.0385,
      "step": 17625
    },
    {
      "epoch": 11.85,
      "grad_norm": 2.5941948890686035,
      "learning_rate": 8.277185929648241e-06,
      "loss": 0.0398,
      "step": 17650
    },
    {
      "epoch": 11.87,
      "grad_norm": 2.6056137084960938,
      "learning_rate": 8.274673366834172e-06,
      "loss": 0.0381,
      "step": 17675
    },
    {
      "epoch": 11.89,
      "grad_norm": 2.8399932384490967,
      "learning_rate": 8.2721608040201e-06,
      "loss": 0.0401,
      "step": 17700
    },
    {
      "epoch": 11.9,
      "grad_norm": 2.9396562576293945,
      "learning_rate": 8.269648241206031e-06,
      "loss": 0.0409,
      "step": 17725
    },
    {
      "epoch": 11.92,
      "grad_norm": 3.1237053871154785,
      "learning_rate": 8.26713567839196e-06,
      "loss": 0.039,
      "step": 17750
    },
    {
      "epoch": 11.94,
      "grad_norm": 3.0028700828552246,
      "learning_rate": 8.26462311557789e-06,
      "loss": 0.0421,
      "step": 17775
    },
    {
      "epoch": 11.95,
      "grad_norm": 3.055807590484619,
      "learning_rate": 8.26211055276382e-06,
      "loss": 0.0405,
      "step": 17800
    },
    {
      "epoch": 11.97,
      "grad_norm": 3.251986026763916,
      "learning_rate": 8.25959798994975e-06,
      "loss": 0.0433,
      "step": 17825
    },
    {
      "epoch": 11.99,
      "grad_norm": 2.845550537109375,
      "learning_rate": 8.25708542713568e-06,
      "loss": 0.0385,
      "step": 17850
    },
    {
      "epoch": 12.0,
      "grad_norm": 2.913346290588379,
      "learning_rate": 8.254572864321609e-06,
      "loss": 0.0378,
      "step": 17875
    },
    {
      "epoch": 12.02,
      "grad_norm": 2.3991270065307617,
      "learning_rate": 8.252060301507538e-06,
      "loss": 0.0294,
      "step": 17900
    },
    {
      "epoch": 12.04,
      "grad_norm": 2.4414055347442627,
      "learning_rate": 8.249547738693467e-06,
      "loss": 0.0306,
      "step": 17925
    },
    {
      "epoch": 12.06,
      "grad_norm": 2.274725914001465,
      "learning_rate": 8.247035175879398e-06,
      "loss": 0.0295,
      "step": 17950
    },
    {
      "epoch": 12.07,
      "grad_norm": 2.767655849456787,
      "learning_rate": 8.244522613065328e-06,
      "loss": 0.0307,
      "step": 17975
    },
    {
      "epoch": 12.09,
      "grad_norm": 2.5598373413085938,
      "learning_rate": 8.242010050251257e-06,
      "loss": 0.0307,
      "step": 18000
    },
    {
      "epoch": 12.09,
      "eval_loss": 0.1446864753961563,
      "eval_runtime": 537.4834,
      "eval_samples_per_second": 2.579,
      "eval_steps_per_second": 2.579,
      "eval_wer": 27.238212961327918,
      "step": 18000
    },
    {
      "epoch": 12.11,
      "grad_norm": 2.654730796813965,
      "learning_rate": 8.239497487437186e-06,
      "loss": 0.0303,
      "step": 18025
    },
    {
      "epoch": 12.12,
      "grad_norm": 2.6578266620635986,
      "learning_rate": 8.236984924623116e-06,
      "loss": 0.0298,
      "step": 18050
    },
    {
      "epoch": 12.14,
      "grad_norm": 3.2597641944885254,
      "learning_rate": 8.234472361809047e-06,
      "loss": 0.0307,
      "step": 18075
    },
    {
      "epoch": 12.16,
      "grad_norm": 3.1756911277770996,
      "learning_rate": 8.231959798994976e-06,
      "loss": 0.0303,
      "step": 18100
    },
    {
      "epoch": 12.17,
      "grad_norm": 2.3517801761627197,
      "learning_rate": 8.229447236180905e-06,
      "loss": 0.0299,
      "step": 18125
    },
    {
      "epoch": 12.19,
      "grad_norm": 2.7081449031829834,
      "learning_rate": 8.226934673366835e-06,
      "loss": 0.0317,
      "step": 18150
    },
    {
      "epoch": 12.21,
      "grad_norm": 2.9442265033721924,
      "learning_rate": 8.224422110552764e-06,
      "loss": 0.0309,
      "step": 18175
    },
    {
      "epoch": 12.22,
      "grad_norm": 2.202742099761963,
      "learning_rate": 8.221909547738695e-06,
      "loss": 0.0299,
      "step": 18200
    },
    {
      "epoch": 12.24,
      "grad_norm": 2.683105230331421,
      "learning_rate": 8.219396984924624e-06,
      "loss": 0.0303,
      "step": 18225
    },
    {
      "epoch": 12.26,
      "grad_norm": 2.4034810066223145,
      "learning_rate": 8.216884422110554e-06,
      "loss": 0.0319,
      "step": 18250
    },
    {
      "epoch": 12.27,
      "grad_norm": 2.621290683746338,
      "learning_rate": 8.214371859296483e-06,
      "loss": 0.0318,
      "step": 18275
    },
    {
      "epoch": 12.29,
      "grad_norm": 2.842874765396118,
      "learning_rate": 8.211859296482412e-06,
      "loss": 0.0332,
      "step": 18300
    },
    {
      "epoch": 12.31,
      "grad_norm": 2.4797563552856445,
      "learning_rate": 8.209346733668342e-06,
      "loss": 0.0325,
      "step": 18325
    },
    {
      "epoch": 12.32,
      "grad_norm": 2.8069446086883545,
      "learning_rate": 8.206834170854273e-06,
      "loss": 0.033,
      "step": 18350
    },
    {
      "epoch": 12.34,
      "grad_norm": 2.9851083755493164,
      "learning_rate": 8.204321608040202e-06,
      "loss": 0.0321,
      "step": 18375
    },
    {
      "epoch": 12.36,
      "grad_norm": 2.948084592819214,
      "learning_rate": 8.201809045226131e-06,
      "loss": 0.0338,
      "step": 18400
    },
    {
      "epoch": 12.37,
      "grad_norm": 2.7898919582366943,
      "learning_rate": 8.19929648241206e-06,
      "loss": 0.0315,
      "step": 18425
    },
    {
      "epoch": 12.39,
      "grad_norm": 2.366434097290039,
      "learning_rate": 8.19678391959799e-06,
      "loss": 0.032,
      "step": 18450
    },
    {
      "epoch": 12.41,
      "grad_norm": 2.9562463760375977,
      "learning_rate": 8.194271356783921e-06,
      "loss": 0.0334,
      "step": 18475
    },
    {
      "epoch": 12.42,
      "grad_norm": 2.5975656509399414,
      "learning_rate": 8.19175879396985e-06,
      "loss": 0.0331,
      "step": 18500
    },
    {
      "epoch": 12.44,
      "grad_norm": 2.8374183177948,
      "learning_rate": 8.18924623115578e-06,
      "loss": 0.0318,
      "step": 18525
    },
    {
      "epoch": 12.46,
      "grad_norm": 2.839860439300537,
      "learning_rate": 8.186733668341709e-06,
      "loss": 0.0324,
      "step": 18550
    },
    {
      "epoch": 12.47,
      "grad_norm": 2.800180196762085,
      "learning_rate": 8.184221105527638e-06,
      "loss": 0.0309,
      "step": 18575
    },
    {
      "epoch": 12.49,
      "grad_norm": 2.644583225250244,
      "learning_rate": 8.18170854271357e-06,
      "loss": 0.0331,
      "step": 18600
    },
    {
      "epoch": 12.51,
      "grad_norm": 3.0358402729034424,
      "learning_rate": 8.179195979899498e-06,
      "loss": 0.0327,
      "step": 18625
    },
    {
      "epoch": 12.53,
      "grad_norm": 2.807608127593994,
      "learning_rate": 8.176683417085428e-06,
      "loss": 0.032,
      "step": 18650
    },
    {
      "epoch": 12.54,
      "grad_norm": 3.115736961364746,
      "learning_rate": 8.174170854271357e-06,
      "loss": 0.034,
      "step": 18675
    },
    {
      "epoch": 12.56,
      "grad_norm": 2.563960313796997,
      "learning_rate": 8.171658291457286e-06,
      "loss": 0.0325,
      "step": 18700
    },
    {
      "epoch": 12.58,
      "grad_norm": 2.6218457221984863,
      "learning_rate": 8.169145728643216e-06,
      "loss": 0.0312,
      "step": 18725
    },
    {
      "epoch": 12.59,
      "grad_norm": 2.6230452060699463,
      "learning_rate": 8.166633165829147e-06,
      "loss": 0.0318,
      "step": 18750
    },
    {
      "epoch": 12.61,
      "grad_norm": 3.0028395652770996,
      "learning_rate": 8.164120603015076e-06,
      "loss": 0.0339,
      "step": 18775
    },
    {
      "epoch": 12.63,
      "grad_norm": 2.810173273086548,
      "learning_rate": 8.161608040201005e-06,
      "loss": 0.0337,
      "step": 18800
    },
    {
      "epoch": 12.64,
      "grad_norm": 2.7154364585876465,
      "learning_rate": 8.159095477386936e-06,
      "loss": 0.0315,
      "step": 18825
    },
    {
      "epoch": 12.66,
      "grad_norm": 2.9645156860351562,
      "learning_rate": 8.156582914572864e-06,
      "loss": 0.0341,
      "step": 18850
    },
    {
      "epoch": 12.68,
      "grad_norm": 2.558562755584717,
      "learning_rate": 8.154070351758795e-06,
      "loss": 0.0321,
      "step": 18875
    },
    {
      "epoch": 12.69,
      "grad_norm": 3.045975923538208,
      "learning_rate": 8.151557788944724e-06,
      "loss": 0.0328,
      "step": 18900
    },
    {
      "epoch": 12.71,
      "grad_norm": 2.605736494064331,
      "learning_rate": 8.149045226130654e-06,
      "loss": 0.0338,
      "step": 18925
    },
    {
      "epoch": 12.73,
      "grad_norm": 2.6503992080688477,
      "learning_rate": 8.146532663316583e-06,
      "loss": 0.0349,
      "step": 18950
    },
    {
      "epoch": 12.74,
      "grad_norm": 2.7485363483428955,
      "learning_rate": 8.144020100502512e-06,
      "loss": 0.0331,
      "step": 18975
    },
    {
      "epoch": 12.76,
      "grad_norm": 3.0558133125305176,
      "learning_rate": 8.141507537688443e-06,
      "loss": 0.0331,
      "step": 19000
    },
    {
      "epoch": 12.76,
      "eval_loss": 0.15126191079616547,
      "eval_runtime": 542.0176,
      "eval_samples_per_second": 2.557,
      "eval_steps_per_second": 2.557,
      "eval_wer": 28.129966448878683,
      "step": 19000
    },
    {
      "epoch": 12.78,
      "grad_norm": 3.117704391479492,
      "learning_rate": 8.138994974874373e-06,
      "loss": 0.0336,
      "step": 19025
    },
    {
      "epoch": 12.79,
      "grad_norm": 2.7645487785339355,
      "learning_rate": 8.136482412060302e-06,
      "loss": 0.0324,
      "step": 19050
    },
    {
      "epoch": 12.81,
      "grad_norm": 2.742771625518799,
      "learning_rate": 8.133969849246231e-06,
      "loss": 0.0331,
      "step": 19075
    },
    {
      "epoch": 12.83,
      "grad_norm": 2.8407609462738037,
      "learning_rate": 8.131457286432162e-06,
      "loss": 0.0317,
      "step": 19100
    },
    {
      "epoch": 12.84,
      "grad_norm": 2.5845396518707275,
      "learning_rate": 8.12894472361809e-06,
      "loss": 0.0335,
      "step": 19125
    },
    {
      "epoch": 12.86,
      "grad_norm": 2.8739688396453857,
      "learning_rate": 8.126432160804021e-06,
      "loss": 0.0333,
      "step": 19150
    },
    {
      "epoch": 12.88,
      "grad_norm": 3.1160261631011963,
      "learning_rate": 8.12391959798995e-06,
      "loss": 0.033,
      "step": 19175
    },
    {
      "epoch": 12.89,
      "grad_norm": 2.978895902633667,
      "learning_rate": 8.12140703517588e-06,
      "loss": 0.0358,
      "step": 19200
    },
    {
      "epoch": 12.91,
      "grad_norm": 3.0800576210021973,
      "learning_rate": 8.11889447236181e-06,
      "loss": 0.0335,
      "step": 19225
    },
    {
      "epoch": 12.93,
      "grad_norm": 2.4890170097351074,
      "learning_rate": 8.11638190954774e-06,
      "loss": 0.034,
      "step": 19250
    },
    {
      "epoch": 12.94,
      "grad_norm": 2.8995964527130127,
      "learning_rate": 8.11386934673367e-06,
      "loss": 0.0342,
      "step": 19275
    },
    {
      "epoch": 12.96,
      "grad_norm": 2.8822238445281982,
      "learning_rate": 8.111356783919599e-06,
      "loss": 0.0338,
      "step": 19300
    },
    {
      "epoch": 12.98,
      "grad_norm": 2.3847439289093018,
      "learning_rate": 8.108844221105528e-06,
      "loss": 0.0345,
      "step": 19325
    },
    {
      "epoch": 13.0,
      "grad_norm": 2.5077168941497803,
      "learning_rate": 8.106331658291457e-06,
      "loss": 0.0323,
      "step": 19350
    },
    {
      "epoch": 13.01,
      "grad_norm": 2.0860869884490967,
      "learning_rate": 8.103819095477388e-06,
      "loss": 0.0256,
      "step": 19375
    },
    {
      "epoch": 13.03,
      "grad_norm": 2.4186856746673584,
      "learning_rate": 8.101306532663318e-06,
      "loss": 0.025,
      "step": 19400
    },
    {
      "epoch": 13.05,
      "grad_norm": 2.169545888900757,
      "learning_rate": 8.098793969849247e-06,
      "loss": 0.024,
      "step": 19425
    },
    {
      "epoch": 13.06,
      "grad_norm": 2.250295877456665,
      "learning_rate": 8.096281407035176e-06,
      "loss": 0.0227,
      "step": 19450
    },
    {
      "epoch": 13.08,
      "grad_norm": 2.8207223415374756,
      "learning_rate": 8.093768844221106e-06,
      "loss": 0.0254,
      "step": 19475
    },
    {
      "epoch": 13.1,
      "grad_norm": 2.4845900535583496,
      "learning_rate": 8.091256281407037e-06,
      "loss": 0.0251,
      "step": 19500
    },
    {
      "epoch": 13.11,
      "grad_norm": 2.9678895473480225,
      "learning_rate": 8.088743718592966e-06,
      "loss": 0.0255,
      "step": 19525
    },
    {
      "epoch": 13.13,
      "grad_norm": 3.0639657974243164,
      "learning_rate": 8.086231155778895e-06,
      "loss": 0.0266,
      "step": 19550
    },
    {
      "epoch": 13.15,
      "grad_norm": 2.5778753757476807,
      "learning_rate": 8.083718592964825e-06,
      "loss": 0.0258,
      "step": 19575
    },
    {
      "epoch": 13.16,
      "grad_norm": 2.3090131282806396,
      "learning_rate": 8.081206030150754e-06,
      "loss": 0.0234,
      "step": 19600
    },
    {
      "epoch": 13.18,
      "grad_norm": 2.645989418029785,
      "learning_rate": 8.078693467336685e-06,
      "loss": 0.0243,
      "step": 19625
    },
    {
      "epoch": 13.2,
      "grad_norm": 2.4817280769348145,
      "learning_rate": 8.076180904522614e-06,
      "loss": 0.0274,
      "step": 19650
    },
    {
      "epoch": 13.21,
      "grad_norm": 2.17031192779541,
      "learning_rate": 8.073668341708544e-06,
      "loss": 0.024,
      "step": 19675
    },
    {
      "epoch": 13.23,
      "grad_norm": 2.587280035018921,
      "learning_rate": 8.071155778894473e-06,
      "loss": 0.0258,
      "step": 19700
    },
    {
      "epoch": 13.25,
      "grad_norm": 2.3844306468963623,
      "learning_rate": 8.068643216080402e-06,
      "loss": 0.0264,
      "step": 19725
    },
    {
      "epoch": 13.26,
      "grad_norm": 2.440300226211548,
      "learning_rate": 8.066130653266332e-06,
      "loss": 0.0259,
      "step": 19750
    },
    {
      "epoch": 13.28,
      "grad_norm": 2.120274543762207,
      "learning_rate": 8.063618090452263e-06,
      "loss": 0.0253,
      "step": 19775
    },
    {
      "epoch": 13.3,
      "grad_norm": 2.412203073501587,
      "learning_rate": 8.061105527638192e-06,
      "loss": 0.0256,
      "step": 19800
    },
    {
      "epoch": 13.31,
      "grad_norm": 2.3215441703796387,
      "learning_rate": 8.058592964824121e-06,
      "loss": 0.0247,
      "step": 19825
    },
    {
      "epoch": 13.33,
      "grad_norm": 2.0729939937591553,
      "learning_rate": 8.05608040201005e-06,
      "loss": 0.0248,
      "step": 19850
    },
    {
      "epoch": 13.35,
      "grad_norm": 2.622880697250366,
      "learning_rate": 8.05356783919598e-06,
      "loss": 0.0271,
      "step": 19875
    },
    {
      "epoch": 13.36,
      "grad_norm": 2.5304481983184814,
      "learning_rate": 8.051055276381911e-06,
      "loss": 0.0255,
      "step": 19900
    },
    {
      "epoch": 13.38,
      "grad_norm": 2.6204922199249268,
      "learning_rate": 8.04854271356784e-06,
      "loss": 0.0261,
      "step": 19925
    },
    {
      "epoch": 13.4,
      "grad_norm": 2.284783363342285,
      "learning_rate": 8.04603015075377e-06,
      "loss": 0.0257,
      "step": 19950
    },
    {
      "epoch": 13.42,
      "grad_norm": 3.0914671421051025,
      "learning_rate": 8.043517587939699e-06,
      "loss": 0.027,
      "step": 19975
    },
    {
      "epoch": 13.43,
      "grad_norm": 2.8612654209136963,
      "learning_rate": 8.041005025125628e-06,
      "loss": 0.0258,
      "step": 20000
    },
    {
      "epoch": 13.43,
      "eval_loss": 0.15857619047164917,
      "eval_runtime": 534.77,
      "eval_samples_per_second": 2.592,
      "eval_steps_per_second": 2.592,
      "eval_wer": 28.809818117605506,
      "step": 20000
    },
    {
      "epoch": 13.45,
      "grad_norm": 3.074786424636841,
      "learning_rate": 8.03849246231156e-06,
      "loss": 0.026,
      "step": 20025
    },
    {
      "epoch": 13.47,
      "grad_norm": 2.40915584564209,
      "learning_rate": 8.035979899497489e-06,
      "loss": 0.029,
      "step": 20050
    },
    {
      "epoch": 13.48,
      "grad_norm": 2.7619211673736572,
      "learning_rate": 8.033467336683418e-06,
      "loss": 0.0261,
      "step": 20075
    },
    {
      "epoch": 13.5,
      "grad_norm": 2.8454036712646484,
      "learning_rate": 8.030954773869347e-06,
      "loss": 0.0257,
      "step": 20100
    },
    {
      "epoch": 13.52,
      "grad_norm": 2.519239664077759,
      "learning_rate": 8.028442211055277e-06,
      "loss": 0.0255,
      "step": 20125
    },
    {
      "epoch": 13.53,
      "grad_norm": 2.798295736312866,
      "learning_rate": 8.025929648241206e-06,
      "loss": 0.0256,
      "step": 20150
    },
    {
      "epoch": 13.55,
      "grad_norm": 2.658249855041504,
      "learning_rate": 8.023417085427137e-06,
      "loss": 0.0252,
      "step": 20175
    },
    {
      "epoch": 13.57,
      "grad_norm": 2.55195689201355,
      "learning_rate": 8.020904522613066e-06,
      "loss": 0.0281,
      "step": 20200
    },
    {
      "epoch": 13.58,
      "grad_norm": 2.282550096511841,
      "learning_rate": 8.018391959798996e-06,
      "loss": 0.0262,
      "step": 20225
    },
    {
      "epoch": 13.6,
      "grad_norm": 2.6260697841644287,
      "learning_rate": 8.015879396984927e-06,
      "loss": 0.0249,
      "step": 20250
    },
    {
      "epoch": 13.62,
      "grad_norm": 2.61671781539917,
      "learning_rate": 8.013366834170854e-06,
      "loss": 0.0276,
      "step": 20275
    },
    {
      "epoch": 13.63,
      "grad_norm": 2.5859358310699463,
      "learning_rate": 8.010854271356785e-06,
      "loss": 0.0265,
      "step": 20300
    },
    {
      "epoch": 13.65,
      "grad_norm": 2.6100573539733887,
      "learning_rate": 8.008341708542714e-06,
      "loss": 0.0258,
      "step": 20325
    },
    {
      "epoch": 13.67,
      "grad_norm": 2.5182266235351562,
      "learning_rate": 8.005829145728644e-06,
      "loss": 0.028,
      "step": 20350
    },
    {
      "epoch": 13.68,
      "grad_norm": 3.105220317840576,
      "learning_rate": 8.003316582914573e-06,
      "loss": 0.027,
      "step": 20375
    },
    {
      "epoch": 13.7,
      "grad_norm": 2.7697339057922363,
      "learning_rate": 8.000804020100502e-06,
      "loss": 0.0274,
      "step": 20400
    },
    {
      "epoch": 13.72,
      "grad_norm": 2.74824857711792,
      "learning_rate": 7.998291457286432e-06,
      "loss": 0.0264,
      "step": 20425
    },
    {
      "epoch": 13.73,
      "grad_norm": 2.1460442543029785,
      "learning_rate": 7.995778894472363e-06,
      "loss": 0.0266,
      "step": 20450
    },
    {
      "epoch": 13.75,
      "grad_norm": 2.700098991394043,
      "learning_rate": 7.993266331658292e-06,
      "loss": 0.0271,
      "step": 20475
    },
    {
      "epoch": 13.77,
      "grad_norm": 3.0646328926086426,
      "learning_rate": 7.990753768844221e-06,
      "loss": 0.0273,
      "step": 20500
    },
    {
      "epoch": 13.78,
      "grad_norm": 2.4817585945129395,
      "learning_rate": 7.988241206030152e-06,
      "loss": 0.0267,
      "step": 20525
    },
    {
      "epoch": 13.8,
      "grad_norm": 2.383892059326172,
      "learning_rate": 7.98572864321608e-06,
      "loss": 0.0281,
      "step": 20550
    },
    {
      "epoch": 13.82,
      "grad_norm": 2.6712028980255127,
      "learning_rate": 7.983216080402011e-06,
      "loss": 0.0262,
      "step": 20575
    },
    {
      "epoch": 13.83,
      "grad_norm": 2.8054888248443604,
      "learning_rate": 7.98070351758794e-06,
      "loss": 0.0277,
      "step": 20600
    },
    {
      "epoch": 13.85,
      "grad_norm": 2.520451545715332,
      "learning_rate": 7.97819095477387e-06,
      "loss": 0.0256,
      "step": 20625
    },
    {
      "epoch": 13.87,
      "grad_norm": 2.6715471744537354,
      "learning_rate": 7.975678391959799e-06,
      "loss": 0.0271,
      "step": 20650
    },
    {
      "epoch": 13.89,
      "grad_norm": 2.936898946762085,
      "learning_rate": 7.973165829145728e-06,
      "loss": 0.0271,
      "step": 20675
    },
    {
      "epoch": 13.9,
      "grad_norm": 2.5876598358154297,
      "learning_rate": 7.97065326633166e-06,
      "loss": 0.0254,
      "step": 20700
    },
    {
      "epoch": 13.92,
      "grad_norm": 2.576573133468628,
      "learning_rate": 7.968140703517589e-06,
      "loss": 0.0268,
      "step": 20725
    },
    {
      "epoch": 13.94,
      "grad_norm": 2.962134838104248,
      "learning_rate": 7.965628140703518e-06,
      "loss": 0.028,
      "step": 20750
    },
    {
      "epoch": 13.95,
      "grad_norm": 2.4978857040405273,
      "learning_rate": 7.963115577889447e-06,
      "loss": 0.0268,
      "step": 20775
    },
    {
      "epoch": 13.97,
      "grad_norm": 2.7507359981536865,
      "learning_rate": 7.960603015075378e-06,
      "loss": 0.0264,
      "step": 20800
    },
    {
      "epoch": 13.99,
      "grad_norm": 2.290602922439575,
      "learning_rate": 7.958090452261306e-06,
      "loss": 0.0268,
      "step": 20825
    },
    {
      "epoch": 14.0,
      "grad_norm": 1.895709753036499,
      "learning_rate": 7.955577889447237e-06,
      "loss": 0.0267,
      "step": 20850
    },
    {
      "epoch": 14.02,
      "grad_norm": 2.577284097671509,
      "learning_rate": 7.953065326633166e-06,
      "loss": 0.02,
      "step": 20875
    },
    {
      "epoch": 14.04,
      "grad_norm": 2.139061450958252,
      "learning_rate": 7.950552763819096e-06,
      "loss": 0.0182,
      "step": 20900
    },
    {
      "epoch": 14.05,
      "grad_norm": 2.31142520904541,
      "learning_rate": 7.948040201005027e-06,
      "loss": 0.0189,
      "step": 20925
    },
    {
      "epoch": 14.07,
      "grad_norm": 2.4628167152404785,
      "learning_rate": 7.945527638190954e-06,
      "loss": 0.0191,
      "step": 20950
    },
    {
      "epoch": 14.09,
      "grad_norm": 2.2550642490386963,
      "learning_rate": 7.943015075376885e-06,
      "loss": 0.0205,
      "step": 20975
    },
    {
      "epoch": 14.1,
      "grad_norm": 2.5067131519317627,
      "learning_rate": 7.940502512562815e-06,
      "loss": 0.0193,
      "step": 21000
    },
    {
      "epoch": 14.1,
      "eval_loss": 0.16441361606121063,
      "eval_runtime": 532.0683,
      "eval_samples_per_second": 2.605,
      "eval_steps_per_second": 2.605,
      "eval_wer": 28.28006357054565,
      "step": 21000
    },
    {
      "epoch": 14.12,
      "grad_norm": 2.0792436599731445,
      "learning_rate": 7.937989949748744e-06,
      "loss": 0.0202,
      "step": 21025
    },
    {
      "epoch": 14.14,
      "grad_norm": 2.0055572986602783,
      "learning_rate": 7.935477386934673e-06,
      "loss": 0.02,
      "step": 21050
    },
    {
      "epoch": 14.15,
      "grad_norm": 2.557342052459717,
      "learning_rate": 7.932964824120604e-06,
      "loss": 0.0202,
      "step": 21075
    },
    {
      "epoch": 14.17,
      "grad_norm": 2.351605176925659,
      "learning_rate": 7.930452261306534e-06,
      "loss": 0.0205,
      "step": 21100
    },
    {
      "epoch": 14.19,
      "grad_norm": 2.4522876739501953,
      "learning_rate": 7.927939698492463e-06,
      "loss": 0.0197,
      "step": 21125
    },
    {
      "epoch": 14.2,
      "grad_norm": 1.9259110689163208,
      "learning_rate": 7.925527638190955e-06,
      "loss": 0.019,
      "step": 21150
    },
    {
      "epoch": 14.22,
      "grad_norm": 2.6869237422943115,
      "learning_rate": 7.923015075376886e-06,
      "loss": 0.0191,
      "step": 21175
    },
    {
      "epoch": 14.24,
      "grad_norm": 2.1610636711120605,
      "learning_rate": 7.920502512562815e-06,
      "loss": 0.0199,
      "step": 21200
    },
    {
      "epoch": 14.25,
      "grad_norm": 2.3419833183288574,
      "learning_rate": 7.917989949748744e-06,
      "loss": 0.0205,
      "step": 21225
    },
    {
      "epoch": 14.27,
      "grad_norm": 2.655822277069092,
      "learning_rate": 7.915477386934674e-06,
      "loss": 0.0211,
      "step": 21250
    },
    {
      "epoch": 14.29,
      "grad_norm": 2.3895249366760254,
      "learning_rate": 7.912964824120603e-06,
      "loss": 0.02,
      "step": 21275
    },
    {
      "epoch": 14.3,
      "grad_norm": 2.626079559326172,
      "learning_rate": 7.910452261306534e-06,
      "loss": 0.0204,
      "step": 21300
    },
    {
      "epoch": 14.32,
      "grad_norm": 2.4946000576019287,
      "learning_rate": 7.907939698492463e-06,
      "loss": 0.0211,
      "step": 21325
    },
    {
      "epoch": 14.34,
      "grad_norm": 2.2254092693328857,
      "learning_rate": 7.905427135678393e-06,
      "loss": 0.0209,
      "step": 21350
    },
    {
      "epoch": 14.36,
      "grad_norm": 2.813023328781128,
      "learning_rate": 7.902914572864322e-06,
      "loss": 0.0205,
      "step": 21375
    },
    {
      "epoch": 14.37,
      "grad_norm": 2.3448939323425293,
      "learning_rate": 7.900402010050253e-06,
      "loss": 0.0204,
      "step": 21400
    },
    {
      "epoch": 14.39,
      "grad_norm": 2.1861133575439453,
      "learning_rate": 7.89788944723618e-06,
      "loss": 0.0211,
      "step": 21425
    },
    {
      "epoch": 14.41,
      "grad_norm": 2.1422207355499268,
      "learning_rate": 7.895376884422111e-06,
      "loss": 0.0224,
      "step": 21450
    },
    {
      "epoch": 14.42,
      "grad_norm": 2.713761329650879,
      "learning_rate": 7.89286432160804e-06,
      "loss": 0.02,
      "step": 21475
    },
    {
      "epoch": 14.44,
      "grad_norm": 2.430680274963379,
      "learning_rate": 7.89035175879397e-06,
      "loss": 0.0218,
      "step": 21500
    },
    {
      "epoch": 14.46,
      "grad_norm": 2.974393606185913,
      "learning_rate": 7.887839195979901e-06,
      "loss": 0.0197,
      "step": 21525
    },
    {
      "epoch": 14.47,
      "grad_norm": 2.530994415283203,
      "learning_rate": 7.885326633165829e-06,
      "loss": 0.0221,
      "step": 21550
    },
    {
      "epoch": 14.49,
      "grad_norm": 2.5071282386779785,
      "learning_rate": 7.88281407035176e-06,
      "loss": 0.0214,
      "step": 21575
    },
    {
      "epoch": 14.51,
      "grad_norm": 2.2111854553222656,
      "learning_rate": 7.880301507537689e-06,
      "loss": 0.0208,
      "step": 21600
    },
    {
      "epoch": 14.52,
      "grad_norm": 2.194091320037842,
      "learning_rate": 7.877788944723618e-06,
      "loss": 0.0203,
      "step": 21625
    },
    {
      "epoch": 14.54,
      "grad_norm": 2.2206263542175293,
      "learning_rate": 7.875276381909548e-06,
      "loss": 0.0221,
      "step": 21650
    },
    {
      "epoch": 14.56,
      "grad_norm": 2.425065279006958,
      "learning_rate": 7.872763819095479e-06,
      "loss": 0.0211,
      "step": 21675
    },
    {
      "epoch": 14.57,
      "grad_norm": 2.6152865886688232,
      "learning_rate": 7.870251256281408e-06,
      "loss": 0.0207,
      "step": 21700
    },
    {
      "epoch": 14.59,
      "grad_norm": 2.2612714767456055,
      "learning_rate": 7.867738693467337e-06,
      "loss": 0.0209,
      "step": 21725
    },
    {
      "epoch": 14.61,
      "grad_norm": 2.1470086574554443,
      "learning_rate": 7.865226130653267e-06,
      "loss": 0.021,
      "step": 21750
    },
    {
      "epoch": 14.62,
      "grad_norm": 2.484851598739624,
      "learning_rate": 7.862713567839196e-06,
      "loss": 0.02,
      "step": 21775
    },
    {
      "epoch": 14.64,
      "grad_norm": 2.4667041301727295,
      "learning_rate": 7.860201005025127e-06,
      "loss": 0.0206,
      "step": 21800
    },
    {
      "epoch": 14.66,
      "grad_norm": 2.9903693199157715,
      "learning_rate": 7.857688442211055e-06,
      "loss": 0.0219,
      "step": 21825
    },
    {
      "epoch": 14.67,
      "grad_norm": 2.6542530059814453,
      "learning_rate": 7.855175879396986e-06,
      "loss": 0.0213,
      "step": 21850
    },
    {
      "epoch": 14.69,
      "grad_norm": 2.333191394805908,
      "learning_rate": 7.852663316582915e-06,
      "loss": 0.0214,
      "step": 21875
    },
    {
      "epoch": 14.71,
      "grad_norm": 2.71769380569458,
      "learning_rate": 7.850150753768844e-06,
      "loss": 0.0215,
      "step": 21900
    },
    {
      "epoch": 14.72,
      "grad_norm": 2.4674861431121826,
      "learning_rate": 7.847638190954775e-06,
      "loss": 0.0211,
      "step": 21925
    },
    {
      "epoch": 14.74,
      "grad_norm": 2.931941270828247,
      "learning_rate": 7.845125628140705e-06,
      "loss": 0.0244,
      "step": 21950
    },
    {
      "epoch": 14.76,
      "grad_norm": 2.738786458969116,
      "learning_rate": 7.842613065326634e-06,
      "loss": 0.0218,
      "step": 21975
    },
    {
      "epoch": 14.78,
      "grad_norm": 2.375138521194458,
      "learning_rate": 7.840100502512563e-06,
      "loss": 0.0219,
      "step": 22000
    },
    {
      "epoch": 14.78,
      "eval_loss": 0.16828645765781403,
      "eval_runtime": 539.6518,
      "eval_samples_per_second": 2.568,
      "eval_steps_per_second": 2.568,
      "eval_wer": 28.11230796397669,
      "step": 22000
    }
  ],
  "logging_steps": 25,
  "max_steps": 100000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 68,
  "save_steps": 1000,
  "total_flos": 3.465787561869312e+19,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}