{
  "best_global_step": 5300,
  "best_metric": 18.77442273534636,
  "best_model_checkpoint": "C:\\Users\\tcham\\Wokspace\\african-speech-recognition\\models\\whisper-tiny-nufi-cpu\\checkpoint-5300",
  "epoch": 4.386952545473021,
  "eval_steps": 100,
  "global_step": 5400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0203231378924906,
      "grad_norm": 504.22711181640625,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 44.52904296875,
      "step": 25
    },
    {
      "epoch": 0.0406462757849812,
      "grad_norm": 300.1306457519531,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 28.0701708984375,
      "step": 50
    },
    {
      "epoch": 0.0609694136774718,
      "grad_norm": 273.4499206542969,
      "learning_rate": 7.4e-06,
      "loss": 14.294178466796875,
      "step": 75
    },
    {
      "epoch": 0.0812925515699624,
      "grad_norm": 239.64649963378906,
      "learning_rate": 9.9e-06,
      "loss": 10.49992919921875,
      "step": 100
    },
    {
      "epoch": 0.0812925515699624,
      "eval_cer": 173.6056838365897,
      "eval_loss": 1.2681573629379272,
      "eval_runtime": 711.555,
      "eval_samples_per_second": 0.769,
      "eval_steps_per_second": 0.769,
      "eval_wer": 119.61620469083157,
      "step": 100
    },
    {
      "epoch": 0.101615689462453,
      "grad_norm": 191.65396118164062,
      "learning_rate": 9.96036333608588e-06,
      "loss": 9.902861328125,
      "step": 125
    },
    {
      "epoch": 0.1219388273549436,
      "grad_norm": 221.65945434570312,
      "learning_rate": 9.91907514450867e-06,
      "loss": 8.375130615234376,
      "step": 150
    },
    {
      "epoch": 0.1422619652474342,
      "grad_norm": 321.12677001953125,
      "learning_rate": 9.877786952931463e-06,
      "loss": 7.383707275390625,
      "step": 175
    },
    {
      "epoch": 0.1625851031399248,
      "grad_norm": 284.2441711425781,
      "learning_rate": 9.836498761354254e-06,
      "loss": 6.827086181640625,
      "step": 200
    },
    {
      "epoch": 0.1625851031399248,
      "eval_cer": 71.04795737122558,
      "eval_loss": 0.8489342331886292,
      "eval_runtime": 456.5823,
      "eval_samples_per_second": 1.198,
      "eval_steps_per_second": 1.198,
      "eval_wer": 111.62046908315565,
      "step": 200
    },
    {
      "epoch": 0.1829082410324154,
      "grad_norm": 240.58543395996094,
      "learning_rate": 9.795210569777046e-06,
      "loss": 6.5289239501953125,
      "step": 225
    },
    {
      "epoch": 0.203231378924906,
      "grad_norm": 200.3292236328125,
      "learning_rate": 9.753922378199837e-06,
      "loss": 6.070548706054687,
      "step": 250
    },
    {
      "epoch": 0.2235545168173966,
      "grad_norm": 213.40771484375,
      "learning_rate": 9.712634186622628e-06,
      "loss": 5.768544311523438,
      "step": 275
    },
    {
      "epoch": 0.2438776547098872,
      "grad_norm": 231.1772003173828,
      "learning_rate": 9.671345995045417e-06,
      "loss": 5.214599609375,
      "step": 300
    },
    {
      "epoch": 0.2438776547098872,
      "eval_cer": 66.83836589698046,
      "eval_loss": 0.6892489194869995,
      "eval_runtime": 360.2103,
      "eval_samples_per_second": 1.519,
      "eval_steps_per_second": 1.519,
      "eval_wer": 118.86993603411513,
      "step": 300
    },
    {
      "epoch": 0.26420079260237783,
      "grad_norm": 215.71893310546875,
      "learning_rate": 9.630057803468208e-06,
      "loss": 5.269068603515625,
      "step": 325
    },
    {
      "epoch": 0.2845239304948684,
      "grad_norm": 184.10113525390625,
      "learning_rate": 9.588769611891e-06,
      "loss": 5.511924438476562,
      "step": 350
    },
    {
      "epoch": 0.30484706838735903,
      "grad_norm": 186.4422607421875,
      "learning_rate": 9.54748142031379e-06,
      "loss": 4.8811328125,
      "step": 375
    },
    {
      "epoch": 0.3251702062798496,
      "grad_norm": 128.75047302246094,
      "learning_rate": 9.506193228736582e-06,
      "loss": 4.461416320800781,
      "step": 400
    },
    {
      "epoch": 0.3251702062798496,
      "eval_cer": 60.142095914742455,
      "eval_loss": 0.5795261263847351,
      "eval_runtime": 694.1925,
      "eval_samples_per_second": 0.788,
      "eval_steps_per_second": 0.788,
      "eval_wer": 112.57995735607676,
      "step": 400
    },
    {
      "epoch": 0.34549334417234023,
      "grad_norm": 251.42494201660156,
      "learning_rate": 9.464905037159373e-06,
      "loss": 4.127410278320313,
      "step": 425
    },
    {
      "epoch": 0.3658164820648308,
      "grad_norm": 200.18704223632812,
      "learning_rate": 9.423616845582164e-06,
      "loss": 4.442520446777344,
      "step": 450
    },
    {
      "epoch": 0.38613961995732143,
      "grad_norm": 145.26776123046875,
      "learning_rate": 9.382328654004955e-06,
      "loss": 3.964246826171875,
      "step": 475
    },
    {
      "epoch": 0.406462757849812,
      "grad_norm": 115.70714569091797,
      "learning_rate": 9.341040462427746e-06,
      "loss": 3.8950460815429686,
      "step": 500
    },
    {
      "epoch": 0.406462757849812,
      "eval_cer": 58.774422735346356,
      "eval_loss": 0.5118994116783142,
      "eval_runtime": 456.082,
      "eval_samples_per_second": 1.199,
      "eval_steps_per_second": 1.199,
      "eval_wer": 108.2089552238806,
      "step": 500
    },
    {
      "epoch": 0.42678589574230263,
      "grad_norm": 205.18502807617188,
      "learning_rate": 9.299752270850537e-06,
      "loss": 4.013980712890625,
      "step": 525
    },
    {
      "epoch": 0.4471090336347932,
      "grad_norm": 147.5428009033203,
      "learning_rate": 9.258464079273328e-06,
      "loss": 3.7656964111328124,
      "step": 550
    },
    {
      "epoch": 0.46743217152728384,
      "grad_norm": 154.23712158203125,
      "learning_rate": 9.21717588769612e-06,
      "loss": 3.709325866699219,
      "step": 575
    },
    {
      "epoch": 0.4877553094197744,
      "grad_norm": 159.62420654296875,
      "learning_rate": 9.17588769611891e-06,
      "loss": 3.898363037109375,
      "step": 600
    },
    {
      "epoch": 0.4877553094197744,
      "eval_cer": 47.282415630550624,
      "eval_loss": 0.46684497594833374,
      "eval_runtime": 570.6455,
      "eval_samples_per_second": 0.959,
      "eval_steps_per_second": 0.959,
      "eval_wer": 105.97014925373134,
      "step": 600
    },
    {
      "epoch": 0.508078447312265,
      "grad_norm": 203.92694091796875,
      "learning_rate": 9.134599504541702e-06,
      "loss": 3.59960205078125,
      "step": 625
    },
    {
      "epoch": 0.5284015852047557,
      "grad_norm": 167.29525756835938,
      "learning_rate": 9.093311312964493e-06,
      "loss": 3.441510009765625,
      "step": 650
    },
    {
      "epoch": 0.5487247230972462,
      "grad_norm": 151.82179260253906,
      "learning_rate": 9.052023121387284e-06,
      "loss": 3.416817626953125,
      "step": 675
    },
    {
      "epoch": 0.5690478609897368,
      "grad_norm": 144.99624633789062,
      "learning_rate": 9.010734929810075e-06,
      "loss": 3.3079843139648437,
      "step": 700
    },
    {
      "epoch": 0.5690478609897368,
      "eval_cer": 44.1563055062167,
      "eval_loss": 0.4140050411224365,
      "eval_runtime": 552.4984,
      "eval_samples_per_second": 0.99,
      "eval_steps_per_second": 0.99,
      "eval_wer": 97.86780383795309,
      "step": 700
    },
    {
      "epoch": 0.5893709988822274,
      "grad_norm": 161.67771911621094,
      "learning_rate": 8.969446738232866e-06,
      "loss": 2.644310302734375,
      "step": 725
    },
    {
      "epoch": 0.6096941367747181,
      "grad_norm": 125.45323944091797,
      "learning_rate": 8.928158546655657e-06,
      "loss": 2.9411090087890623,
      "step": 750
    },
    {
      "epoch": 0.6300172746672086,
      "grad_norm": 188.03909301757812,
      "learning_rate": 8.886870355078449e-06,
      "loss": 2.788016662597656,
      "step": 775
    },
    {
      "epoch": 0.6503404125596992,
      "grad_norm": 144.45726013183594,
      "learning_rate": 8.84558216350124e-06,
      "loss": 2.7427078247070313,
      "step": 800
    },
    {
      "epoch": 0.6503404125596992,
      "eval_cer": 44.7246891651865,
      "eval_loss": 0.3983415365219116,
      "eval_runtime": 626.5463,
      "eval_samples_per_second": 0.873,
      "eval_steps_per_second": 0.873,
      "eval_wer": 101.1727078891258,
      "step": 800
    },
    {
      "epoch": 0.6706635504521898,
      "grad_norm": 169.8371124267578,
      "learning_rate": 8.80429397192403e-06,
      "loss": 2.5388978576660155,
      "step": 825
    },
    {
      "epoch": 0.6909866883446805,
      "grad_norm": 151.0153045654297,
      "learning_rate": 8.763005780346822e-06,
      "loss": 2.9790170288085935,
      "step": 850
    },
    {
      "epoch": 0.711309826237171,
      "grad_norm": 117.27120208740234,
      "learning_rate": 8.721717588769613e-06,
      "loss": 3.095129089355469,
      "step": 875
    },
    {
      "epoch": 0.7316329641296616,
      "grad_norm": 253.4383544921875,
      "learning_rate": 8.680429397192404e-06,
      "loss": 3.0029559326171875,
      "step": 900
    },
    {
      "epoch": 0.7316329641296616,
      "eval_cer": 33.83658969804618,
      "eval_loss": 0.36028751730918884,
      "eval_runtime": 1458.0583,
      "eval_samples_per_second": 0.375,
      "eval_steps_per_second": 0.375,
      "eval_wer": 84.00852878464818,
      "step": 900
    },
    {
      "epoch": 0.7519561020221522,
      "grad_norm": 159.44955444335938,
      "learning_rate": 8.639141205615195e-06,
      "loss": 2.815823669433594,
      "step": 925
    },
    {
      "epoch": 0.7722792399146429,
      "grad_norm": 127.19710540771484,
      "learning_rate": 8.597853014037986e-06,
      "loss": 2.8693475341796875,
      "step": 950
    },
    {
      "epoch": 0.7926023778071334,
      "grad_norm": 97.33570098876953,
      "learning_rate": 8.556564822460777e-06,
      "loss": 2.7277728271484376,
      "step": 975
    },
    {
      "epoch": 0.812925515699624,
      "grad_norm": 117.55218505859375,
      "learning_rate": 8.515276630883567e-06,
      "loss": 3.1459829711914065,
      "step": 1000
    },
    {
      "epoch": 0.812925515699624,
      "eval_cer": 34.13854351687389,
      "eval_loss": 0.34841328859329224,
      "eval_runtime": 1369.5851,
      "eval_samples_per_second": 0.399,
      "eval_steps_per_second": 0.399,
      "eval_wer": 89.87206823027718,
      "step": 1000
    },
    {
      "epoch": 0.8332486535921146,
      "grad_norm": 119.52729034423828,
      "learning_rate": 8.473988439306358e-06,
      "loss": 2.886960754394531,
      "step": 1025
    },
    {
      "epoch": 0.8535717914846053,
      "grad_norm": 106.33351135253906,
      "learning_rate": 8.43270024772915e-06,
      "loss": 2.5022161865234374,
      "step": 1050
    },
    {
      "epoch": 0.8738949293770958,
      "grad_norm": 121.87995910644531,
      "learning_rate": 8.39141205615194e-06,
      "loss": 2.57005615234375,
      "step": 1075
    },
    {
      "epoch": 0.8942180672695864,
      "grad_norm": 152.60342407226562,
      "learning_rate": 8.350123864574731e-06,
      "loss": 2.6209893798828126,
      "step": 1100
    },
    {
      "epoch": 0.8942180672695864,
      "eval_cer": 36.731793960923625,
      "eval_loss": 0.335877001285553,
      "eval_runtime": 1373.7369,
      "eval_samples_per_second": 0.398,
      "eval_steps_per_second": 0.398,
      "eval_wer": 84.43496801705757,
      "step": 1100
    },
    {
      "epoch": 0.914541205162077,
      "grad_norm": 135.85157775878906,
      "learning_rate": 8.308835672997524e-06,
      "loss": 2.756929931640625,
      "step": 1125
    },
    {
      "epoch": 0.9348643430545677,
      "grad_norm": 91.14208984375,
      "learning_rate": 8.267547481420315e-06,
      "loss": 2.87466796875,
      "step": 1150
    },
    {
      "epoch": 0.9551874809470582,
      "grad_norm": 150.2317352294922,
      "learning_rate": 8.226259289843106e-06,
      "loss": 2.5747671508789063,
      "step": 1175
    },
    {
      "epoch": 0.9755106188395488,
      "grad_norm": 159.78672790527344,
      "learning_rate": 8.184971098265898e-06,
      "loss": 2.2414396667480467,
      "step": 1200
    },
    {
      "epoch": 0.9755106188395488,
      "eval_cer": 30.959147424511546,
      "eval_loss": 0.31040313839912415,
      "eval_runtime": 1306.8216,
      "eval_samples_per_second": 0.419,
      "eval_steps_per_second": 0.419,
      "eval_wer": 83.15565031982942,
      "step": 1200
    },
    {
      "epoch": 0.9958337567320394,
      "grad_norm": 109.06317901611328,
      "learning_rate": 8.143682906688689e-06,
      "loss": 2.4120233154296873,
      "step": 1225
    },
    {
      "epoch": 1.0154455847982928,
      "grad_norm": 98.93025970458984,
      "learning_rate": 8.10239471511148e-06,
      "loss": 2.1240538024902342,
      "step": 1250
    },
    {
      "epoch": 1.0357687226907835,
      "grad_norm": 132.09133911132812,
      "learning_rate": 8.06110652353427e-06,
      "loss": 1.9155911254882811,
      "step": 1275
    },
    {
      "epoch": 1.056091860583274,
      "grad_norm": 134.10826110839844,
      "learning_rate": 8.01981833195706e-06,
      "loss": 1.9743263244628906,
      "step": 1300
    },
    {
      "epoch": 1.056091860583274,
      "eval_cer": 32.06039076376554,
      "eval_loss": 0.3002552390098572,
      "eval_runtime": 1389.5338,
      "eval_samples_per_second": 0.394,
      "eval_steps_per_second": 0.394,
      "eval_wer": 82.72921108742004,
      "step": 1300
    },
    {
      "epoch": 1.0764149984757647,
      "grad_norm": 101.12194061279297,
      "learning_rate": 7.978530140379851e-06,
      "loss": 2.1293214416503905,
      "step": 1325
    },
    {
      "epoch": 1.0967381363682553,
      "grad_norm": 135.47772216796875,
      "learning_rate": 7.937241948802643e-06,
      "loss": 1.777157745361328,
      "step": 1350
    },
    {
      "epoch": 1.1170612742607458,
      "grad_norm": 147.2037811279297,
      "learning_rate": 7.895953757225434e-06,
      "loss": 1.9273220825195312,
      "step": 1375
    },
    {
      "epoch": 1.1373844121532364,
      "grad_norm": 71.98359680175781,
      "learning_rate": 7.854665565648225e-06,
      "loss": 1.7837376403808594,
      "step": 1400
    },
    {
      "epoch": 1.1373844121532364,
      "eval_cer": 35.168738898756665,
      "eval_loss": 0.3094908595085144,
      "eval_runtime": 1411.0796,
      "eval_samples_per_second": 0.388,
      "eval_steps_per_second": 0.388,
      "eval_wer": 87.31343283582089,
      "step": 1400
    },
    {
      "epoch": 1.157707550045727,
      "grad_norm": 68.9445571899414,
      "learning_rate": 7.813377374071016e-06,
      "loss": 2.114809112548828,
      "step": 1425
    },
    {
      "epoch": 1.1780306879382176,
      "grad_norm": 103.13447570800781,
      "learning_rate": 7.772089182493807e-06,
      "loss": 1.7878659057617188,
      "step": 1450
    },
    {
      "epoch": 1.1983538258307083,
      "grad_norm": 55.20072555541992,
      "learning_rate": 7.730800990916598e-06,
      "loss": 1.8778643798828125,
      "step": 1475
    },
    {
      "epoch": 1.218676963723199,
      "grad_norm": 74.9879379272461,
      "learning_rate": 7.68951279933939e-06,
      "loss": 1.835758819580078,
      "step": 1500
    },
    {
      "epoch": 1.218676963723199,
      "eval_cer": 28.667850799289518,
      "eval_loss": 0.28896716237068176,
      "eval_runtime": 1391.0235,
      "eval_samples_per_second": 0.393,
      "eval_steps_per_second": 0.393,
      "eval_wer": 73.98720682302772,
      "step": 1500
    },
    {
      "epoch": 1.2390001016156895,
      "grad_norm": 171.2066192626953,
      "learning_rate": 7.64822460776218e-06,
      "loss": 1.8034989929199219,
      "step": 1525
    },
    {
      "epoch": 1.2593232395081801,
      "grad_norm": 157.14198303222656,
      "learning_rate": 7.606936416184972e-06,
      "loss": 1.9245022583007811,
      "step": 1550
    },
    {
      "epoch": 1.2796463774006708,
      "grad_norm": 74.64144134521484,
      "learning_rate": 7.565648224607762e-06,
      "loss": 1.8367242431640625,
      "step": 1575
    },
    {
      "epoch": 1.2999695152931612,
      "grad_norm": 169.1046600341797,
      "learning_rate": 7.524360033030555e-06,
      "loss": 1.7905462646484376,
      "step": 1600
    },
    {
      "epoch": 1.2999695152931612,
      "eval_cer": 28.82770870337478,
      "eval_loss": 0.27988311648368835,
      "eval_runtime": 1172.1568,
      "eval_samples_per_second": 0.467,
      "eval_steps_per_second": 0.467,
      "eval_wer": 78.03837953091684,
      "step": 1600
    },
    {
      "epoch": 1.3202926531856518,
      "grad_norm": 59.238563537597656,
      "learning_rate": 7.483071841453346e-06,
      "loss": 1.887559814453125,
      "step": 1625
    },
    {
      "epoch": 1.3406157910781424,
      "grad_norm": 75.89434814453125,
      "learning_rate": 7.441783649876136e-06,
      "loss": 1.8114033508300782,
      "step": 1650
    },
    {
      "epoch": 1.360938928970633,
      "grad_norm": 100.34668731689453,
      "learning_rate": 7.400495458298927e-06,
      "loss": 1.664589385986328,
      "step": 1675
    },
    {
      "epoch": 1.3812620668631237,
      "grad_norm": 66.03353118896484,
      "learning_rate": 7.359207266721718e-06,
      "loss": 1.7770977783203126,
      "step": 1700
    },
    {
      "epoch": 1.3812620668631237,
      "eval_cer": 28.24156305506217,
      "eval_loss": 0.2707199156284332,
      "eval_runtime": 1279.8534,
      "eval_samples_per_second": 0.427,
      "eval_steps_per_second": 0.427,
      "eval_wer": 74.20042643923242,
      "step": 1700
    },
    {
      "epoch": 1.4015852047556143,
      "grad_norm": 148.64271545410156,
      "learning_rate": 7.3179190751445094e-06,
      "loss": 1.7526657104492187,
      "step": 1725
    },
    {
      "epoch": 1.421908342648105,
      "grad_norm": 123.52094268798828,
      "learning_rate": 7.2766308835673006e-06,
      "loss": 1.5236294555664063,
      "step": 1750
    },
    {
      "epoch": 1.4422314805405954,
      "grad_norm": 135.69247436523438,
      "learning_rate": 7.235342691990092e-06,
      "loss": 1.786600341796875,
      "step": 1775
    },
    {
      "epoch": 1.462554618433086,
      "grad_norm": 181.1728057861328,
      "learning_rate": 7.194054500412883e-06,
      "loss": 1.72043701171875,
      "step": 1800
    },
    {
      "epoch": 1.462554618433086,
      "eval_cer": 25.044404973357015,
      "eval_loss": 0.2616724967956543,
      "eval_runtime": 428.661,
      "eval_samples_per_second": 1.276,
      "eval_steps_per_second": 1.276,
      "eval_wer": 67.37739872068231,
      "step": 1800
    },
    {
      "epoch": 1.4828777563255766,
      "grad_norm": 102.76298522949219,
      "learning_rate": 9.43797881010595e-06,
      "loss": 1.8223385620117187,
      "step": 1825
    },
    {
      "epoch": 1.5032008942180672,
      "grad_norm": 221.53797912597656,
      "learning_rate": 9.429828850855747e-06,
      "loss": 1.6174728393554687,
      "step": 1850
    },
    {
      "epoch": 1.5235240321105579,
      "grad_norm": 105.45674896240234,
      "learning_rate": 9.421678891605542e-06,
      "loss": 1.8269915771484375,
      "step": 1875
    },
    {
      "epoch": 1.5438471700030485,
      "grad_norm": 133.1260986328125,
      "learning_rate": 9.413528932355339e-06,
      "loss": 1.5793560791015624,
      "step": 1900
    },
    {
      "epoch": 1.5438471700030485,
      "eval_cer": 26.962699822380102,
      "eval_loss": 0.26534637808799744,
      "eval_runtime": 670.1701,
      "eval_samples_per_second": 0.816,
      "eval_steps_per_second": 0.816,
      "eval_wer": 68.7633262260128,
      "step": 1900
    },
    {
      "epoch": 1.5641703078955391,
      "grad_norm": 120.65760040283203,
      "learning_rate": 9.405378973105134e-06,
      "loss": 1.7896224975585937,
      "step": 1925
    },
    {
      "epoch": 1.5844934457880298,
      "grad_norm": 154.523193359375,
      "learning_rate": 9.397229013854931e-06,
      "loss": 1.3732839965820312,
      "step": 1950
    },
    {
      "epoch": 1.6048165836805204,
      "grad_norm": 85.28186798095703,
      "learning_rate": 9.389079054604727e-06,
      "loss": 1.8148405456542969,
      "step": 1975
    },
    {
      "epoch": 1.625139721573011,
      "grad_norm": 116.69713592529297,
      "learning_rate": 9.380929095354524e-06,
      "loss": 1.5959693908691406,
      "step": 2000
    },
    {
      "epoch": 1.625139721573011,
      "eval_cer": 26.554174067495563,
      "eval_loss": 0.25513890385627747,
      "eval_runtime": 658.413,
      "eval_samples_per_second": 0.831,
      "eval_steps_per_second": 0.831,
      "eval_wer": 66.09808102345416,
      "step": 2000
    },
    {
      "epoch": 1.6454628594655016,
      "grad_norm": 99.85759735107422,
      "learning_rate": 9.37277913610432e-06,
      "loss": 1.7173544311523437,
      "step": 2025
    },
    {
      "epoch": 1.665785997357992,
      "grad_norm": 95.3363265991211,
      "learning_rate": 9.364629176854116e-06,
      "loss": 1.6536795043945312,
      "step": 2050
    },
    {
      "epoch": 1.6861091352504827,
      "grad_norm": 112.90980529785156,
      "learning_rate": 9.356479217603913e-06,
      "loss": 1.5276934814453125,
      "step": 2075
    },
    {
      "epoch": 1.7064322731429733,
      "grad_norm": 131.311767578125,
      "learning_rate": 9.348329258353708e-06,
      "loss": 1.4632241821289063,
      "step": 2100
    },
    {
      "epoch": 1.7064322731429733,
      "eval_cer": 27.015985790408525,
      "eval_loss": 0.24846957623958588,
      "eval_runtime": 758.5985,
      "eval_samples_per_second": 0.721,
      "eval_steps_per_second": 0.721,
      "eval_wer": 65.67164179104478,
      "step": 2100
    },
    {
      "epoch": 1.726755411035464,
      "grad_norm": 144.18362426757812,
      "learning_rate": 9.340179299103505e-06,
      "loss": 1.486290283203125,
      "step": 2125
    },
    {
      "epoch": 1.7470785489279543,
      "grad_norm": 171.3587188720703,
      "learning_rate": 9.3320293398533e-06,
      "loss": 1.673651123046875,
      "step": 2150
    },
    {
      "epoch": 1.767401686820445,
      "grad_norm": 64.67139434814453,
      "learning_rate": 9.323879380603098e-06,
      "loss": 1.5918267822265626,
      "step": 2175
    },
    {
      "epoch": 1.7877248247129356,
      "grad_norm": 102.27267456054688,
      "learning_rate": 9.315729421352893e-06,
      "loss": 1.5113105773925781,
      "step": 2200
    },
    {
      "epoch": 1.7877248247129356,
      "eval_cer": 24.33392539964476,
      "eval_loss": 0.24699242413043976,
      "eval_runtime": 680.562,
      "eval_samples_per_second": 0.804,
      "eval_steps_per_second": 0.804,
      "eval_wer": 62.68656716417911,
      "step": 2200
    },
    {
      "epoch": 1.8080479626054262,
      "grad_norm": 142.9913787841797,
      "learning_rate": 9.30757946210269e-06,
      "loss": 1.7434786987304687,
      "step": 2225
    },
    {
      "epoch": 1.8283711004979168,
      "grad_norm": 105.40516662597656,
      "learning_rate": 9.299429502852487e-06,
      "loss": 1.4946308898925782,
      "step": 2250
    },
    {
      "epoch": 1.8486942383904075,
      "grad_norm": 111.58350372314453,
      "learning_rate": 9.291279543602282e-06,
      "loss": 1.917201690673828,
      "step": 2275
    },
    {
      "epoch": 1.869017376282898,
      "grad_norm": 168.474609375,
      "learning_rate": 9.28312958435208e-06,
      "loss": 1.7116949462890625,
      "step": 2300
    },
    {
      "epoch": 1.869017376282898,
      "eval_cer": 27.54884547069272,
      "eval_loss": 0.2382805198431015,
      "eval_runtime": 679.487,
      "eval_samples_per_second": 0.805,
      "eval_steps_per_second": 0.805,
      "eval_wer": 69.82942430703625,
      "step": 2300
    },
    {
      "epoch": 1.8893405141753887,
      "grad_norm": 157.70494079589844,
      "learning_rate": 9.274979625101875e-06,
      "loss": 1.6206892395019532,
      "step": 2325
    },
    {
      "epoch": 1.9096636520678794,
      "grad_norm": 107.10155487060547,
      "learning_rate": 9.266829665851672e-06,
      "loss": 1.4464591979980468,
      "step": 2350
    },
    {
      "epoch": 1.92998678996037,
      "grad_norm": 55.5534782409668,
      "learning_rate": 9.258679706601467e-06,
      "loss": 1.6919075012207032,
      "step": 2375
    },
    {
      "epoch": 1.9503099278528606,
      "grad_norm": 93.59611511230469,
      "learning_rate": 9.250529747351264e-06,
      "loss": 1.6986639404296875,
      "step": 2400
    },
    {
      "epoch": 1.9503099278528606,
      "eval_cer": 25.843694493783303,
      "eval_loss": 0.24499696493148804,
      "eval_runtime": 698.1936,
      "eval_samples_per_second": 0.783,
      "eval_steps_per_second": 0.783,
      "eval_wer": 67.16417910447761,
      "step": 2400
    },
    {
      "epoch": 1.9706330657453512,
      "grad_norm": 64.02003479003906,
      "learning_rate": 9.242379788101061e-06,
      "loss": 1.6129344177246094,
      "step": 2425
    },
    {
      "epoch": 1.9909562036378416,
      "grad_norm": 38.325313568115234,
      "learning_rate": 9.234229828850856e-06,
      "loss": 1.7849627685546876,
      "step": 2450
    },
    {
      "epoch": 2.010568031704095,
      "grad_norm": 144.74317932128906,
      "learning_rate": 9.226079869600653e-06,
      "loss": 1.6095491027832032,
      "step": 2475
    },
    {
      "epoch": 2.0308911695965857,
      "grad_norm": 128.37503051757812,
      "learning_rate": 9.217929910350449e-06,
      "loss": 1.2414840698242187,
      "step": 2500
    },
    {
      "epoch": 2.0308911695965857,
      "eval_cer": 22.184724689165186,
      "eval_loss": 0.23757652938365936,
      "eval_runtime": 693.1535,
      "eval_samples_per_second": 0.789,
      "eval_steps_per_second": 0.789,
      "eval_wer": 60.66098081023454,
      "step": 2500
    },
    {
      "epoch": 2.0512143074890763,
      "grad_norm": 92.03895568847656,
      "learning_rate": 9.209779951100246e-06,
      "loss": 1.0699855041503907,
      "step": 2525
    },
    {
      "epoch": 2.071537445381567,
      "grad_norm": 153.54556274414062,
      "learning_rate": 9.201629991850043e-06,
      "loss": 1.1367869567871094,
      "step": 2550
    },
    {
      "epoch": 2.0918605832740576,
      "grad_norm": 82.37834930419922,
      "learning_rate": 9.193480032599838e-06,
      "loss": 1.1217189025878906,
      "step": 2575
    },
    {
      "epoch": 2.112183721166548,
      "grad_norm": 68.00988006591797,
      "learning_rate": 9.185330073349635e-06,
      "loss": 1.0083988952636718,
      "step": 2600
    },
    {
      "epoch": 2.112183721166548,
      "eval_cer": 24.58259325044405,
      "eval_loss": 0.22681163251399994,
      "eval_runtime": 691.8909,
      "eval_samples_per_second": 0.791,
      "eval_steps_per_second": 0.791,
      "eval_wer": 60.23454157782516,
      "step": 2600
    },
    {
      "epoch": 2.132506859059039,
      "grad_norm": 63.60848617553711,
      "learning_rate": 9.17718011409943e-06,
      "loss": 1.0976016998291016,
      "step": 2625
    },
    {
      "epoch": 2.1528299969515294,
      "grad_norm": 33.86166763305664,
      "learning_rate": 9.169030154849227e-06,
      "loss": 1.1681001281738281,
      "step": 2650
    },
    {
      "epoch": 2.17315313484402,
      "grad_norm": 82.70641326904297,
      "learning_rate": 9.160880195599023e-06,
      "loss": 1.103014907836914,
      "step": 2675
    },
    {
      "epoch": 2.1934762727365107,
      "grad_norm": 111.75276947021484,
      "learning_rate": 9.15273023634882e-06,
      "loss": 1.2061747741699218,
      "step": 2700
    },
    {
      "epoch": 2.1934762727365107,
      "eval_cer": 23.907637655417407,
      "eval_loss": 0.23165647685527802,
      "eval_runtime": 701.7116,
      "eval_samples_per_second": 0.78,
      "eval_steps_per_second": 0.78,
      "eval_wer": 66.09808102345416,
      "step": 2700
    },
    {
      "epoch": 2.2137994106290013,
      "grad_norm": 95.47616577148438,
      "learning_rate": 9.144580277098615e-06,
      "loss": 1.0050942993164063,
      "step": 2725
    },
    {
      "epoch": 2.2341225485214915,
      "grad_norm": 39.8724250793457,
      "learning_rate": 9.136430317848412e-06,
      "loss": 1.1623261260986328,
      "step": 2750
    },
    {
      "epoch": 2.254445686413982,
      "grad_norm": 156.6307373046875,
      "learning_rate": 9.128280358598209e-06,
      "loss": 1.201123046875,
      "step": 2775
    },
    {
      "epoch": 2.2747688243064728,
      "grad_norm": 105.96151733398438,
      "learning_rate": 9.120130399348004e-06,
      "loss": 1.2658418273925782,
      "step": 2800
    },
    {
      "epoch": 2.2747688243064728,
      "eval_cer": 21.77619893428064,
      "eval_loss": 0.2180229276418686,
      "eval_runtime": 717.5352,
      "eval_samples_per_second": 0.762,
      "eval_steps_per_second": 0.762,
      "eval_wer": 60.12793176972282,
      "step": 2800
    },
    {
      "epoch": 2.2950919621989634,
      "grad_norm": 97.60123443603516,
      "learning_rate": 9.111980440097801e-06,
      "loss": 1.155967254638672,
      "step": 2825
    },
    {
      "epoch": 2.315415100091454,
      "grad_norm": 275.60992431640625,
      "learning_rate": 9.103830480847597e-06,
      "loss": 1.0560490417480468,
      "step": 2850
    },
    {
      "epoch": 2.3357382379839446,
      "grad_norm": 84.4440689086914,
      "learning_rate": 9.095680521597394e-06,
      "loss": 1.128302993774414,
      "step": 2875
    },
    {
      "epoch": 2.3560613758764353,
      "grad_norm": 94.70333862304688,
      "learning_rate": 9.087530562347189e-06,
      "loss": 1.286135711669922,
      "step": 2900
    },
    {
      "epoch": 2.3560613758764353,
      "eval_cer": 22.841918294849023,
      "eval_loss": 0.21774369478225708,
      "eval_runtime": 693.436,
      "eval_samples_per_second": 0.789,
      "eval_steps_per_second": 0.789,
      "eval_wer": 60.021321961620465,
      "step": 2900
    },
    {
      "epoch": 2.376384513768926,
      "grad_norm": 155.65313720703125,
      "learning_rate": 9.079380603096986e-06,
      "loss": 1.16890869140625,
      "step": 2925
    },
    {
      "epoch": 2.3967076516614165,
      "grad_norm": 98.38611602783203,
      "learning_rate": 9.071230643846781e-06,
      "loss": 1.0425555419921875,
      "step": 2950
    },
    {
      "epoch": 2.417030789553907,
      "grad_norm": 58.345516204833984,
      "learning_rate": 9.063080684596578e-06,
      "loss": 1.1348511505126953,
      "step": 2975
    },
    {
      "epoch": 2.437353927446398,
      "grad_norm": 95.1644515991211,
      "learning_rate": 9.054930725346374e-06,
      "loss": 1.0004090881347656,
      "step": 3000
    },
    {
      "epoch": 2.437353927446398,
      "eval_cer": 22.75310834813499,
      "eval_loss": 0.2199828028678894,
      "eval_runtime": 717.7838,
      "eval_samples_per_second": 0.762,
      "eval_steps_per_second": 0.762,
      "eval_wer": 61.833688699360344,
      "step": 3000
    },
    {
      "epoch": 2.4576770653388884,
      "grad_norm": 99.55670928955078,
      "learning_rate": 9.04678076609617e-06,
      "loss": 1.081207504272461,
      "step": 3025
    },
    {
      "epoch": 2.478000203231379,
      "grad_norm": 143.2071075439453,
      "learning_rate": 9.038630806845966e-06,
      "loss": 1.0549359130859375,
      "step": 3050
    },
    {
      "epoch": 2.4983233411238697,
      "grad_norm": 78.72933197021484,
      "learning_rate": 9.030480847595763e-06,
      "loss": 1.1238302612304687,
      "step": 3075
    },
    {
      "epoch": 2.5186464790163603,
      "grad_norm": 172.69789123535156,
      "learning_rate": 9.022330888345558e-06,
      "loss": 1.1517737579345704,
      "step": 3100
    },
    {
      "epoch": 2.5186464790163603,
      "eval_cer": 22.36234458259325,
      "eval_loss": 0.2184779793024063,
      "eval_runtime": 704.5453,
      "eval_samples_per_second": 0.776,
      "eval_steps_per_second": 0.776,
      "eval_wer": 58.52878464818764,
      "step": 3100
    },
    {
      "epoch": 2.538969616908851,
      "grad_norm": 109.82661437988281,
      "learning_rate": 9.014180929095355e-06,
      "loss": 0.9760607147216797,
      "step": 3125
    },
    {
      "epoch": 2.5592927548013416,
      "grad_norm": 151.6119842529297,
      "learning_rate": 9.006030969845152e-06,
      "loss": 1.0380972290039063,
      "step": 3150
    },
    {
      "epoch": 2.579615892693832,
      "grad_norm": 72.74021911621094,
      "learning_rate": 8.997881010594948e-06,
      "loss": 1.096968307495117,
      "step": 3175
    },
    {
      "epoch": 2.5999390305863224,
      "grad_norm": 74.72578430175781,
      "learning_rate": 8.989731051344745e-06,
      "loss": 1.1900202178955077,
      "step": 3200
    },
    {
      "epoch": 2.5999390305863224,
      "eval_cer": 24.7246891651865,
      "eval_loss": 0.2234128713607788,
      "eval_runtime": 714.3069,
      "eval_samples_per_second": 0.766,
      "eval_steps_per_second": 0.766,
      "eval_wer": 64.0724946695096,
      "step": 3200
    },
    {
      "epoch": 2.620262168478813,
      "grad_norm": 113.59678649902344,
      "learning_rate": 8.98158109209454e-06,
      "loss": 1.063079605102539,
      "step": 3225
    },
    {
      "epoch": 2.6405853063713036,
      "grad_norm": 71.74285888671875,
      "learning_rate": 8.973431132844337e-06,
      "loss": 0.9074827575683594,
      "step": 3250
    },
    {
      "epoch": 2.6609084442637942,
      "grad_norm": 72.7939453125,
      "learning_rate": 8.965281173594132e-06,
      "loss": 0.9882894134521485,
      "step": 3275
    },
    {
      "epoch": 2.681231582156285,
      "grad_norm": 173.79615783691406,
      "learning_rate": 8.95713121434393e-06,
      "loss": 1.0640850067138672,
      "step": 3300
    },
    {
      "epoch": 2.681231582156285,
      "eval_cer": 21.651865008880993,
      "eval_loss": 0.22209937870502472,
      "eval_runtime": 710.5422,
      "eval_samples_per_second": 0.77,
      "eval_steps_per_second": 0.77,
      "eval_wer": 57.46268656716418,
      "step": 3300
    },
    {
      "epoch": 2.7015547200487755,
      "grad_norm": 131.877685546875,
      "learning_rate": 8.948981255093725e-06,
      "loss": 1.3128117370605468,
      "step": 3325
    },
    {
      "epoch": 2.721877857941266,
      "grad_norm": 149.9304962158203,
      "learning_rate": 8.940831295843522e-06,
      "loss": 1.1322238159179687,
      "step": 3350
    },
    {
      "epoch": 2.7422009958337568,
      "grad_norm": 70.97515869140625,
      "learning_rate": 8.932681336593317e-06,
      "loss": 1.1012673950195313,
      "step": 3375
    },
    {
      "epoch": 2.7625241337262474,
      "grad_norm": 45.893707275390625,
      "learning_rate": 8.924531377343114e-06,
      "loss": 1.0022483825683595,
      "step": 3400
    },
    {
      "epoch": 2.7625241337262474,
      "eval_cer": 22.07815275310835,
      "eval_loss": 0.22389553487300873,
      "eval_runtime": 698.5818,
      "eval_samples_per_second": 0.783,
      "eval_steps_per_second": 0.783,
      "eval_wer": 56.60980810234542,
      "step": 3400
    },
    {
      "epoch": 2.782847271618738,
      "grad_norm": 131.48675537109375,
      "learning_rate": 8.91638141809291e-06,
      "loss": 1.2124028015136719,
      "step": 3425
    },
    {
      "epoch": 2.8031704095112286,
      "grad_norm": 146.05023193359375,
      "learning_rate": 8.908231458842706e-06,
      "loss": 1.1416338348388673,
      "step": 3450
    },
    {
      "epoch": 2.8234935474037193,
      "grad_norm": 53.297996520996094,
      "learning_rate": 8.900081499592502e-06,
      "loss": 1.1624464416503906,
      "step": 3475
    },
    {
      "epoch": 2.84381668529621,
      "grad_norm": 138.91616821289062,
      "learning_rate": 8.891931540342299e-06,
      "loss": 1.1016802215576171,
      "step": 3500
    },
    {
      "epoch": 2.84381668529621,
      "eval_cer": 21.70515097690941,
      "eval_loss": 0.21198414266109467,
      "eval_runtime": 705.3615,
      "eval_samples_per_second": 0.775,
      "eval_steps_per_second": 0.775,
      "eval_wer": 56.39658848614072,
      "step": 3500
    },
    {
      "epoch": 2.8641398231887,
      "grad_norm": 87.6238021850586,
      "learning_rate": 8.883781581092096e-06,
      "loss": 1.2523902130126954,
      "step": 3525
    },
    {
      "epoch": 2.8844629610811907,
      "grad_norm": 63.05250549316406,
      "learning_rate": 8.875631621841891e-06,
      "loss": 1.0616297149658203,
      "step": 3550
    },
    {
      "epoch": 2.9047860989736813,
      "grad_norm": 142.04672241210938,
      "learning_rate": 8.867481662591688e-06,
      "loss": 1.1016445922851563,
      "step": 3575
    },
    {
      "epoch": 2.925109236866172,
      "grad_norm": 80.58750915527344,
      "learning_rate": 8.859331703341483e-06,
      "loss": 1.098918685913086,
      "step": 3600
    },
    {
      "epoch": 2.925109236866172,
      "eval_cer": 22.45115452930728,
      "eval_loss": 0.21089890599250793,
      "eval_runtime": 710.7245,
      "eval_samples_per_second": 0.77,
      "eval_steps_per_second": 0.77,
      "eval_wer": 59.1684434968017,
      "step": 3600
    },
    {
      "epoch": 2.9454323747586626,
      "grad_norm": 93.55347442626953,
      "learning_rate": 8.85118174409128e-06,
      "loss": 1.09771728515625,
      "step": 3625
    },
    {
      "epoch": 2.9657555126511532,
      "grad_norm": 59.22978591918945,
      "learning_rate": 8.843031784841076e-06,
      "loss": 1.0049332427978515,
      "step": 3650
    },
    {
      "epoch": 2.986078650543644,
      "grad_norm": 73.22894287109375,
      "learning_rate": 8.834881825590873e-06,
      "loss": 1.1537026214599608,
      "step": 3675
    },
    {
      "epoch": 3.0056904786098975,
      "grad_norm": 65.94957733154297,
      "learning_rate": 8.826731866340668e-06,
      "loss": 1.0833206176757812,
      "step": 3700
    },
    {
      "epoch": 3.0056904786098975,
      "eval_cer": 21.42095914742451,
      "eval_loss": 0.2071550339460373,
      "eval_runtime": 706.0699,
      "eval_samples_per_second": 0.775,
      "eval_steps_per_second": 0.775,
      "eval_wer": 54.15778251599147,
      "step": 3700
    },
    {
      "epoch": 3.026013616502388,
      "grad_norm": 66.91118621826172,
      "learning_rate": 8.818581907090465e-06,
      "loss": 0.7976800537109375,
      "step": 3725
    },
    {
      "epoch": 3.0463367543948787,
      "grad_norm": 52.65355682373047,
      "learning_rate": 8.81043194784026e-06,
      "loss": 0.7081849670410156,
      "step": 3750
    },
    {
      "epoch": 3.0666598922873693,
      "grad_norm": 245.1796417236328,
      "learning_rate": 8.802281988590057e-06,
      "loss": 0.8120671081542968,
      "step": 3775
    },
    {
      "epoch": 3.08698303017986,
      "grad_norm": 52.94565200805664,
      "learning_rate": 8.794132029339853e-06,
      "loss": 0.65879150390625,
      "step": 3800
    },
    {
      "epoch": 3.08698303017986,
      "eval_cer": 21.154529307282417,
      "eval_loss": 0.20488880574703217,
      "eval_runtime": 705.6609,
      "eval_samples_per_second": 0.775,
      "eval_steps_per_second": 0.775,
      "eval_wer": 57.995735607675904,
      "step": 3800
    },
    {
      "epoch": 3.10730616807235,
      "grad_norm": 91.6010513305664,
      "learning_rate": 8.78598207008965e-06,
      "loss": 0.6262943649291992,
      "step": 3825
    },
    {
      "epoch": 3.127629305964841,
      "grad_norm": 45.20219421386719,
      "learning_rate": 8.777832110839447e-06,
      "loss": 0.7677568817138671,
      "step": 3850
    },
    {
      "epoch": 3.1479524438573314,
      "grad_norm": 50.439029693603516,
      "learning_rate": 8.769682151589242e-06,
      "loss": 0.7979000091552735,
      "step": 3875
    },
    {
      "epoch": 3.168275581749822,
      "grad_norm": 72.62454986572266,
      "learning_rate": 8.761532192339039e-06,
      "loss": 0.6406471252441406,
      "step": 3900
    },
    {
      "epoch": 3.168275581749822,
      "eval_cer": 20.923623445825932,
      "eval_loss": 0.199077770113945,
      "eval_runtime": 702.1971,
      "eval_samples_per_second": 0.779,
      "eval_steps_per_second": 0.779,
      "eval_wer": 55.330490405117274,
      "step": 3900
    },
    {
      "epoch": 3.1885987196423127,
      "grad_norm": 67.39865112304688,
      "learning_rate": 8.753382233088834e-06,
      "loss": 0.6185147857666016,
      "step": 3925
    },
    {
      "epoch": 3.2089218575348033,
      "grad_norm": 66.15321350097656,
      "learning_rate": 8.745232273838631e-06,
      "loss": 0.668779296875,
      "step": 3950
    },
    {
      "epoch": 3.229244995427294,
      "grad_norm": 113.56475067138672,
      "learning_rate": 8.737082314588427e-06,
      "loss": 0.6111898422241211,
      "step": 3975
    },
    {
      "epoch": 3.2495681333197846,
      "grad_norm": 83.1806640625,
      "learning_rate": 8.728932355338224e-06,
      "loss": 0.7096160125732421,
      "step": 4000
    },
    {
      "epoch": 3.2495681333197846,
      "eval_cer": 22.184724689165186,
      "eval_loss": 0.20164236426353455,
      "eval_runtime": 712.1842,
      "eval_samples_per_second": 0.768,
      "eval_steps_per_second": 0.768,
      "eval_wer": 58.52878464818764,
      "step": 4000
    },
    {
      "epoch": 3.269891271212275,
      "grad_norm": 51.69230270385742,
      "learning_rate": 8.72078239608802e-06,
      "loss": 0.7522864532470703,
      "step": 4025
    },
    {
      "epoch": 3.290214409104766,
      "grad_norm": 114.41486358642578,
      "learning_rate": 8.712632436837816e-06,
      "loss": 0.6682227325439453,
      "step": 4050
    },
    {
      "epoch": 3.3105375469972564,
      "grad_norm": 148.02732849121094,
      "learning_rate": 8.704482477587613e-06,
      "loss": 0.8854801177978515,
      "step": 4075
    },
    {
      "epoch": 3.330860684889747,
      "grad_norm": 75.74806213378906,
      "learning_rate": 8.696332518337408e-06,
      "loss": 0.743462142944336,
      "step": 4100
    },
    {
      "epoch": 3.330860684889747,
      "eval_cer": 21.119005328596803,
      "eval_loss": 0.1968332678079605,
      "eval_runtime": 682.547,
      "eval_samples_per_second": 0.801,
      "eval_steps_per_second": 0.801,
      "eval_wer": 55.330490405117274,
      "step": 4100
    },
    {
      "epoch": 3.3511838227822377,
      "grad_norm": 76.1290054321289,
      "learning_rate": 8.688182559087205e-06,
      "loss": 0.6166913223266601,
      "step": 4125
    },
    {
      "epoch": 3.3715069606747283,
      "grad_norm": 57.191162109375,
      "learning_rate": 8.680032599837002e-06,
      "loss": 0.8192992401123047,
      "step": 4150
    },
    {
      "epoch": 3.391830098567219,
      "grad_norm": 38.85801696777344,
      "learning_rate": 8.671882640586798e-06,
      "loss": 0.6981607818603516,
      "step": 4175
    },
    {
      "epoch": 3.4121532364597096,
      "grad_norm": 83.87921142578125,
      "learning_rate": 8.663732681336595e-06,
      "loss": 0.7230227661132812,
      "step": 4200
    },
    {
      "epoch": 3.4121532364597096,
      "eval_cer": 19.609236234458262,
      "eval_loss": 0.19413423538208008,
      "eval_runtime": 701.4376,
      "eval_samples_per_second": 0.78,
      "eval_steps_per_second": 0.78,
      "eval_wer": 52.87846481876333,
      "step": 4200
    },
    {
      "epoch": 3.4324763743522,
      "grad_norm": 80.26457977294922,
      "learning_rate": 8.65558272208639e-06,
      "loss": 0.6954840850830079,
      "step": 4225
    },
    {
      "epoch": 3.4527995122446904,
      "grad_norm": 145.0185546875,
      "learning_rate": 8.647432762836187e-06,
      "loss": 0.7332251739501953,
      "step": 4250
    },
    {
      "epoch": 3.473122650137181,
      "grad_norm": 92.92949676513672,
      "learning_rate": 8.639282803585982e-06,
      "loss": 0.8173981475830078,
      "step": 4275
    },
    {
      "epoch": 3.4934457880296717,
      "grad_norm": 156.13197326660156,
      "learning_rate": 8.63113284433578e-06,
      "loss": 0.6192795181274414,
      "step": 4300
    },
    {
      "epoch": 3.4934457880296717,
      "eval_cer": 21.349911190053287,
      "eval_loss": 0.20569011569023132,
      "eval_runtime": 695.3326,
      "eval_samples_per_second": 0.787,
      "eval_steps_per_second": 0.787,
      "eval_wer": 56.929637526652456,
      "step": 4300
    },
    {
      "epoch": 3.5137689259221623,
      "grad_norm": 62.01288604736328,
      "learning_rate": 8.622982885085577e-06,
      "loss": 0.7418531036376953,
      "step": 4325
    },
    {
      "epoch": 3.534092063814653,
      "grad_norm": 90.77095794677734,
      "learning_rate": 8.614832925835372e-06,
      "loss": 0.8014112091064454,
      "step": 4350
    },
    {
      "epoch": 3.5544152017071435,
      "grad_norm": 97.51771545410156,
      "learning_rate": 8.606682966585169e-06,
      "loss": 0.7216325378417969,
      "step": 4375
    },
    {
      "epoch": 3.574738339599634,
      "grad_norm": 147.5199737548828,
      "learning_rate": 8.598533007334964e-06,
      "loss": 0.6958029174804687,
      "step": 4400
    },
    {
      "epoch": 3.574738339599634,
      "eval_cer": 21.22557726465364,
      "eval_loss": 0.19071508944034576,
      "eval_runtime": 1059.0019,
      "eval_samples_per_second": 0.517,
      "eval_steps_per_second": 0.517,
      "eval_wer": 55.11727078891258,
      "step": 4400
    },
    {
      "epoch": 3.595061477492125,
      "grad_norm": 187.49392700195312,
      "learning_rate": 8.590383048084761e-06,
      "loss": 0.7554714202880859,
      "step": 4425
    },
    {
      "epoch": 3.6153846153846154,
      "grad_norm": 11.664432525634766,
      "learning_rate": 8.582233088834557e-06,
      "loss": 0.7832601165771484,
      "step": 4450
    },
    {
      "epoch": 3.635707753277106,
      "grad_norm": 133.3282470703125,
      "learning_rate": 8.574083129584354e-06,
      "loss": 0.6756803894042969,
      "step": 4475
    },
    {
      "epoch": 3.6560308911695967,
      "grad_norm": 47.35822677612305,
      "learning_rate": 8.565933170334149e-06,
      "loss": 0.542722511291504,
      "step": 4500
    },
    {
      "epoch": 3.6560308911695967,
      "eval_cer": 22.25577264653641,
      "eval_loss": 0.20622798800468445,
      "eval_runtime": 712.3794,
      "eval_samples_per_second": 0.768,
      "eval_steps_per_second": 0.768,
      "eval_wer": 56.71641791044776,
      "step": 4500
    },
    {
      "epoch": 3.6763540290620873,
      "grad_norm": 152.44204711914062,
      "learning_rate": 8.557783211083946e-06,
      "loss": 0.822818374633789,
      "step": 4525
    },
    {
      "epoch": 3.696677166954578,
      "grad_norm": 95.3579330444336,
      "learning_rate": 8.549633251833741e-06,
      "loss": 0.8210031127929688,
      "step": 4550
    },
    {
      "epoch": 3.717000304847068,
      "grad_norm": 91.85175323486328,
      "learning_rate": 8.541483292583538e-06,
      "loss": 0.6622520446777344,
      "step": 4575
    },
    {
      "epoch": 3.7373234427395587,
      "grad_norm": 27.532859802246094,
      "learning_rate": 8.533333333333335e-06,
      "loss": 0.6281048965454101,
      "step": 4600
    },
    {
      "epoch": 3.7373234427395587,
      "eval_cer": 21.08348134991119,
      "eval_loss": 0.19978412985801697,
      "eval_runtime": 461.687,
      "eval_samples_per_second": 1.185,
      "eval_steps_per_second": 1.185,
      "eval_wer": 55.75692963752665,
      "step": 4600
    },
    {
      "epoch": 3.7576465806320494,
      "grad_norm": 66.97003936767578,
      "learning_rate": 8.52518337408313e-06,
      "loss": 0.8327887725830078,
      "step": 4625
    },
    {
      "epoch": 3.77796971852454,
      "grad_norm": 53.52529525756836,
      "learning_rate": 8.517033414832928e-06,
      "loss": 0.7108338928222656,
      "step": 4650
    },
    {
      "epoch": 3.7982928564170306,
      "grad_norm": 66.4482650756836,
      "learning_rate": 8.508883455582723e-06,
      "loss": 0.7522552490234375,
      "step": 4675
    },
    {
      "epoch": 3.8186159943095213,
      "grad_norm": 91.04061126708984,
      "learning_rate": 8.50073349633252e-06,
      "loss": 0.7105814361572266,
      "step": 4700
    },
    {
      "epoch": 3.8186159943095213,
      "eval_cer": 21.03019538188277,
      "eval_loss": 0.2015153467655182,
      "eval_runtime": 447.4064,
      "eval_samples_per_second": 1.223,
      "eval_steps_per_second": 1.223,
      "eval_wer": 55.11727078891258,
      "step": 4700
    },
    {
      "epoch": 3.838939132202012,
      "grad_norm": 83.4188232421875,
      "learning_rate": 8.492583537082315e-06,
      "loss": 0.8240100860595703,
      "step": 4725
    },
    {
      "epoch": 3.8592622700945025,
      "grad_norm": 73.50299835205078,
      "learning_rate": 8.484433577832112e-06,
      "loss": 0.6377122116088867,
      "step": 4750
    },
    {
      "epoch": 3.879585407986993,
      "grad_norm": 38.1170768737793,
      "learning_rate": 8.476283618581908e-06,
      "loss": 0.7637339782714844,
      "step": 4775
    },
    {
      "epoch": 3.8999085458794838,
      "grad_norm": 97.35997772216797,
      "learning_rate": 8.468133659331705e-06,
      "loss": 0.8303545379638672,
      "step": 4800
    },
    {
      "epoch": 3.8999085458794838,
      "eval_cer": 19.928952042628776,
      "eval_loss": 0.20565949380397797,
      "eval_runtime": 413.2811,
      "eval_samples_per_second": 1.324,
      "eval_steps_per_second": 1.324,
      "eval_wer": 53.83795309168443,
      "step": 4800
    },
    {
      "epoch": 3.9202316837719744,
      "grad_norm": 127.98067474365234,
      "learning_rate": 8.4599837000815e-06,
      "loss": 0.9052116394042968,
      "step": 4825
    },
    {
      "epoch": 3.940554821664465,
      "grad_norm": 25.169031143188477,
      "learning_rate": 8.451833740831297e-06,
      "loss": 0.7790591430664062,
      "step": 4850
    },
    {
      "epoch": 3.9608779595569557,
      "grad_norm": 86.80704498291016,
      "learning_rate": 8.443683781581092e-06,
      "loss": 0.8170637512207031,
      "step": 4875
    },
    {
      "epoch": 3.9812010974494463,
      "grad_norm": 77.93894958496094,
      "learning_rate": 8.43553382233089e-06,
      "loss": 0.7851848602294922,
      "step": 4900
    },
    {
      "epoch": 3.9812010974494463,
      "eval_cer": 20.515097690941385,
      "eval_loss": 0.19961005449295044,
      "eval_runtime": 483.4097,
      "eval_samples_per_second": 1.132,
      "eval_steps_per_second": 1.132,
      "eval_wer": 53.304904051172706,
      "step": 4900
    },
    {
      "epoch": 4.0008129255156994,
      "grad_norm": 72.7056884765625,
      "learning_rate": 8.427383863080685e-06,
      "loss": 0.6390523910522461,
      "step": 4925
    },
    {
      "epoch": 4.02113606340819,
      "grad_norm": 21.632434844970703,
      "learning_rate": 8.419233903830482e-06,
      "loss": 0.4280753326416016,
      "step": 4950
    },
    {
      "epoch": 4.041459201300681,
      "grad_norm": 78.34844970703125,
      "learning_rate": 8.411083944580279e-06,
      "loss": 0.4393476104736328,
      "step": 4975
    },
    {
      "epoch": 4.061782339193171,
      "grad_norm": 54.52299499511719,
      "learning_rate": 8.402933985330074e-06,
      "loss": 0.4570903778076172,
      "step": 5000
    },
    {
      "epoch": 4.061782339193171,
      "eval_cer": 21.27886323268206,
      "eval_loss": 0.19543775916099548,
      "eval_runtime": 515.0994,
      "eval_samples_per_second": 1.062,
      "eval_steps_per_second": 1.062,
      "eval_wer": 58.2089552238806,
      "step": 5000
    },
    {
      "epoch": 4.082105477085662,
      "grad_norm": 33.551124572753906,
      "learning_rate": 8.394784026079871e-06,
      "loss": 0.5063205337524415,
      "step": 5025
    },
    {
      "epoch": 4.102428614978153,
      "grad_norm": 31.981117248535156,
      "learning_rate": 8.386634066829666e-06,
      "loss": 0.5069622039794922,
      "step": 5050
    },
    {
      "epoch": 4.122751752870643,
      "grad_norm": 64.63758850097656,
      "learning_rate": 8.378484107579463e-06,
      "loss": 0.4465951919555664,
      "step": 5075
    },
    {
      "epoch": 4.143074890763134,
      "grad_norm": 46.17578125,
      "learning_rate": 8.370334148329259e-06,
      "loss": 0.4238921737670898,
      "step": 5100
    },
    {
      "epoch": 4.143074890763134,
      "eval_cer": 18.82770870337478,
      "eval_loss": 0.18936777114868164,
      "eval_runtime": 520.2822,
      "eval_samples_per_second": 1.051,
      "eval_steps_per_second": 1.051,
      "eval_wer": 51.279317697228144,
      "step": 5100
    },
    {
      "epoch": 4.1633980286556245,
      "grad_norm": 34.64641571044922,
      "learning_rate": 8.362184189079056e-06,
      "loss": 0.30932424545288084,
      "step": 5125
    },
    {
      "epoch": 4.183721166548115,
      "grad_norm": 60.691871643066406,
      "learning_rate": 8.354034229828851e-06,
      "loss": 0.4232832717895508,
      "step": 5150
    },
    {
      "epoch": 4.204044304440606,
      "grad_norm": 47.806434631347656,
      "learning_rate": 8.345884270578648e-06,
      "loss": 0.5768677520751954,
      "step": 5175
    },
    {
      "epoch": 4.224367442333096,
      "grad_norm": 41.621978759765625,
      "learning_rate": 8.337734311328443e-06,
      "loss": 0.5497477340698242,
      "step": 5200
    },
    {
      "epoch": 4.224367442333096,
      "eval_cer": 19.69804618117229,
      "eval_loss": 0.19797834753990173,
      "eval_runtime": 545.9798,
      "eval_samples_per_second": 1.002,
      "eval_steps_per_second": 1.002,
      "eval_wer": 52.87846481876333,
      "step": 5200
    },
    {
      "epoch": 4.244690580225587,
      "grad_norm": 86.3643798828125,
      "learning_rate": 8.32958435207824e-06,
      "loss": 0.4311883544921875,
      "step": 5225
    },
    {
      "epoch": 4.265013718118078,
      "grad_norm": 30.649311065673828,
      "learning_rate": 8.321434392828036e-06,
      "loss": 0.4010509490966797,
      "step": 5250
    },
    {
      "epoch": 4.285336856010568,
      "grad_norm": 36.32900619506836,
      "learning_rate": 8.313284433577833e-06,
      "loss": 0.5138769149780273,
      "step": 5275
    },
    {
      "epoch": 4.305659993903059,
      "grad_norm": 22.783056259155273,
      "learning_rate": 8.305134474327628e-06,
      "loss": 0.4101755142211914,
      "step": 5300
    },
    {
      "epoch": 4.305659993903059,
      "eval_cer": 18.77442273534636,
      "eval_loss": 0.19671358168125153,
      "eval_runtime": 688.0551,
      "eval_samples_per_second": 0.795,
      "eval_steps_per_second": 0.795,
      "eval_wer": 51.59914712153518,
      "step": 5300
    },
    {
      "epoch": 4.3259831317955495,
      "grad_norm": 97.10392761230469,
      "learning_rate": 8.296984515077425e-06,
      "loss": 0.629931526184082,
      "step": 5325
    },
    {
      "epoch": 4.34630626968804,
      "grad_norm": 85.37991333007812,
      "learning_rate": 8.288834555827222e-06,
      "loss": 0.5344712829589844,
      "step": 5350
    },
    {
      "epoch": 4.366629407580531,
      "grad_norm": 50.0683479309082,
      "learning_rate": 8.280684596577017e-06,
      "loss": 0.4125747299194336,
      "step": 5375
    },
    {
      "epoch": 4.386952545473021,
      "grad_norm": 32.11968231201172,
      "learning_rate": 8.272534637326814e-06,
      "loss": 0.5142041778564453,
      "step": 5400
    },
    {
      "epoch": 4.386952545473021,
      "eval_cer": 21.456483126110122,
      "eval_loss": 0.19670283794403076,
      "eval_runtime": 708.757,
      "eval_samples_per_second": 0.772,
      "eval_steps_per_second": 0.772,
      "eval_wer": 59.80810234541578,
      "step": 5400
    }
  ],
  "logging_steps": 25,
  "max_steps": 30775,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 25,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.06284650840064e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}