{
  "best_global_step": 5000,
  "best_metric": 0.20783165635834974,
  "best_model_checkpoint": "/home/florent/milo/models/whisper-mg-v1/checkpoint-5000",
  "epoch": 3.946285069787114,
  "eval_steps": 1000,
  "global_step": 7000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.014098406880022557,
      "grad_norm": 84.86420440673828,
      "learning_rate": 4.800000000000001e-07,
      "loss": 17.227703857421876,
      "step": 25
    },
    {
      "epoch": 0.028196813760045115,
      "grad_norm": 66.53484344482422,
      "learning_rate": 9.800000000000001e-07,
      "loss": 13.832662353515625,
      "step": 50
    },
    {
      "epoch": 0.042295220640067674,
      "grad_norm": 43.68435287475586,
      "learning_rate": 1.48e-06,
      "loss": 10.33947021484375,
      "step": 75
    },
    {
      "epoch": 0.05639362752009023,
      "grad_norm": 34.214447021484375,
      "learning_rate": 1.98e-06,
      "loss": 7.374788818359375,
      "step": 100
    },
    {
      "epoch": 0.07049203440011279,
      "grad_norm": 27.010826110839844,
      "learning_rate": 2.4800000000000004e-06,
      "loss": 5.627406005859375,
      "step": 125
    },
    {
      "epoch": 0.08459044128013535,
      "grad_norm": 24.344482421875,
      "learning_rate": 2.9800000000000003e-06,
      "loss": 4.657888793945313,
      "step": 150
    },
    {
      "epoch": 0.0986888481601579,
      "grad_norm": 24.166223526000977,
      "learning_rate": 3.48e-06,
      "loss": 3.687208251953125,
      "step": 175
    },
    {
      "epoch": 0.11278725504018046,
      "grad_norm": 23.202571868896484,
      "learning_rate": 3.980000000000001e-06,
      "loss": 2.8217697143554688,
      "step": 200
    },
    {
      "epoch": 0.126885661920203,
      "grad_norm": 21.142698287963867,
      "learning_rate": 4.48e-06,
      "loss": 2.384236602783203,
      "step": 225
    },
    {
      "epoch": 0.14098406880022557,
      "grad_norm": 21.08614730834961,
      "learning_rate": 4.980000000000001e-06,
      "loss": 2.1562757873535157,
      "step": 250
    },
    {
      "epoch": 0.15508247568024813,
      "grad_norm": 20.73036003112793,
      "learning_rate": 5.480000000000001e-06,
      "loss": 2.0236280822753905,
      "step": 275
    },
    {
      "epoch": 0.1691808825602707,
      "grad_norm": 21.132606506347656,
      "learning_rate": 5.98e-06,
      "loss": 1.865457763671875,
      "step": 300
    },
    {
      "epoch": 0.18327928944029326,
      "grad_norm": 20.846128463745117,
      "learning_rate": 6.480000000000001e-06,
      "loss": 1.7576261901855468,
      "step": 325
    },
    {
      "epoch": 0.1973776963203158,
      "grad_norm": 18.715116500854492,
      "learning_rate": 6.98e-06,
      "loss": 1.6428683471679688,
      "step": 350
    },
    {
      "epoch": 0.21147610320033836,
      "grad_norm": 18.342042922973633,
      "learning_rate": 7.48e-06,
      "loss": 1.6423768615722656,
      "step": 375
    },
    {
      "epoch": 0.22557451008036092,
      "grad_norm": 19.184682846069336,
      "learning_rate": 7.980000000000002e-06,
      "loss": 1.6796397399902343,
      "step": 400
    },
    {
      "epoch": 0.23967291696038348,
      "grad_norm": 15.670419692993164,
      "learning_rate": 8.48e-06,
      "loss": 1.5983003234863282,
      "step": 425
    },
    {
      "epoch": 0.253771323840406,
      "grad_norm": 17.033227920532227,
      "learning_rate": 8.98e-06,
      "loss": 1.5592161560058593,
      "step": 450
    },
    {
      "epoch": 0.2678697307204286,
      "grad_norm": 17.4881649017334,
      "learning_rate": 9.48e-06,
      "loss": 1.5372105407714844,
      "step": 475
    },
    {
      "epoch": 0.28196813760045114,
      "grad_norm": 14.954177856445312,
      "learning_rate": 9.980000000000001e-06,
      "loss": 1.3610572814941406,
      "step": 500
    },
    {
      "epoch": 0.2960665444804737,
      "grad_norm": 14.768590927124023,
      "learning_rate": 9.986078886310906e-06,
      "loss": 1.4396812438964843,
      "step": 525
    },
    {
      "epoch": 0.31016495136049627,
      "grad_norm": 14.442098617553711,
      "learning_rate": 9.971577726218099e-06,
      "loss": 1.3178257751464844,
      "step": 550
    },
    {
      "epoch": 0.32426335824051883,
      "grad_norm": 15.411622047424316,
      "learning_rate": 9.957076566125291e-06,
      "loss": 1.4055836486816407,
      "step": 575
    },
    {
      "epoch": 0.3383617651205414,
      "grad_norm": 15.339015007019043,
      "learning_rate": 9.942575406032482e-06,
      "loss": 1.263025665283203,
      "step": 600
    },
    {
      "epoch": 0.35246017200056395,
      "grad_norm": 12.730027198791504,
      "learning_rate": 9.928074245939677e-06,
      "loss": 1.3695268249511718,
      "step": 625
    },
    {
      "epoch": 0.3665585788805865,
      "grad_norm": 15.123625755310059,
      "learning_rate": 9.913573085846868e-06,
      "loss": 1.2718246459960938,
      "step": 650
    },
    {
      "epoch": 0.380656985760609,
      "grad_norm": 15.623430252075195,
      "learning_rate": 9.899071925754062e-06,
      "loss": 1.2624960327148438,
      "step": 675
    },
    {
      "epoch": 0.3947553926406316,
      "grad_norm": 14.083202362060547,
      "learning_rate": 9.884570765661253e-06,
      "loss": 1.312706756591797,
      "step": 700
    },
    {
      "epoch": 0.40885379952065415,
      "grad_norm": 14.791698455810547,
      "learning_rate": 9.870069605568446e-06,
      "loss": 1.1517729187011718,
      "step": 725
    },
    {
      "epoch": 0.4229522064006767,
      "grad_norm": 16.083740234375,
      "learning_rate": 9.855568445475639e-06,
      "loss": 1.3111384582519532,
      "step": 750
    },
    {
      "epoch": 0.4370506132806993,
      "grad_norm": 12.33295726776123,
      "learning_rate": 9.841067285382831e-06,
      "loss": 1.1583899688720702,
      "step": 775
    },
    {
      "epoch": 0.45114902016072184,
      "grad_norm": 15.303627014160156,
      "learning_rate": 9.826566125290024e-06,
      "loss": 1.1577481079101561,
      "step": 800
    },
    {
      "epoch": 0.4652474270407444,
      "grad_norm": 14.345124244689941,
      "learning_rate": 9.812064965197217e-06,
      "loss": 1.1901895904541016,
      "step": 825
    },
    {
      "epoch": 0.47934583392076696,
      "grad_norm": 14.913029670715332,
      "learning_rate": 9.79756380510441e-06,
      "loss": 1.1890618896484375,
      "step": 850
    },
    {
      "epoch": 0.4934442408007895,
      "grad_norm": 13.710859298706055,
      "learning_rate": 9.783062645011602e-06,
      "loss": 1.1438979339599609,
      "step": 875
    },
    {
      "epoch": 0.507542647680812,
      "grad_norm": 13.514805793762207,
      "learning_rate": 9.768561484918795e-06,
      "loss": 1.1654217529296875,
      "step": 900
    },
    {
      "epoch": 0.5216410545608346,
      "grad_norm": 12.146404266357422,
      "learning_rate": 9.754060324825988e-06,
      "loss": 1.1279725646972656,
      "step": 925
    },
    {
      "epoch": 0.5357394614408572,
      "grad_norm": 14.747977256774902,
      "learning_rate": 9.73955916473318e-06,
      "loss": 1.0921778106689453,
      "step": 950
    },
    {
      "epoch": 0.5498378683208798,
      "grad_norm": 12.012781143188477,
      "learning_rate": 9.725058004640371e-06,
      "loss": 1.089789581298828,
      "step": 975
    },
    {
      "epoch": 0.5639362752009023,
      "grad_norm": 14.096403121948242,
      "learning_rate": 9.710556844547566e-06,
      "loss": 1.09536865234375,
      "step": 1000
    },
    {
      "epoch": 0.5639362752009023,
      "eval_loss": 0.3027215600013733,
      "eval_runtime": 1414.4639,
      "eval_samples_per_second": 2.191,
      "eval_steps_per_second": 0.274,
      "eval_wer": 0.25130644178215644,
      "step": 1000
    },
    {
      "epoch": 0.5780346820809249,
      "grad_norm": 12.812657356262207,
      "learning_rate": 9.696055684454757e-06,
      "loss": 1.073106155395508,
      "step": 1025
    },
    {
      "epoch": 0.5921330889609474,
      "grad_norm": 12.071333885192871,
      "learning_rate": 9.68155452436195e-06,
      "loss": 1.0373800659179688,
      "step": 1050
    },
    {
      "epoch": 0.60623149584097,
      "grad_norm": 14.140727043151855,
      "learning_rate": 9.667053364269142e-06,
      "loss": 1.0855345916748047,
      "step": 1075
    },
    {
      "epoch": 0.6203299027209925,
      "grad_norm": 12.884740829467773,
      "learning_rate": 9.652552204176335e-06,
      "loss": 1.0462813568115235,
      "step": 1100
    },
    {
      "epoch": 0.634428309601015,
      "grad_norm": 12.38447093963623,
      "learning_rate": 9.638051044083528e-06,
      "loss": 1.0593311309814453,
      "step": 1125
    },
    {
      "epoch": 0.6485267164810377,
      "grad_norm": 13.830496788024902,
      "learning_rate": 9.62354988399072e-06,
      "loss": 1.0979310607910155,
      "step": 1150
    },
    {
      "epoch": 0.6626251233610602,
      "grad_norm": 14.614953994750977,
      "learning_rate": 9.609048723897913e-06,
      "loss": 1.1127804565429686,
      "step": 1175
    },
    {
      "epoch": 0.6767235302410828,
      "grad_norm": 13.682169914245605,
      "learning_rate": 9.594547563805106e-06,
      "loss": 1.0035169982910157,
      "step": 1200
    },
    {
      "epoch": 0.6908219371211053,
      "grad_norm": 13.420117378234863,
      "learning_rate": 9.580046403712297e-06,
      "loss": 1.0495610046386719,
      "step": 1225
    },
    {
      "epoch": 0.7049203440011279,
      "grad_norm": 12.9420166015625,
      "learning_rate": 9.565545243619491e-06,
      "loss": 1.060257339477539,
      "step": 1250
    },
    {
      "epoch": 0.7190187508811504,
      "grad_norm": 12.879499435424805,
      "learning_rate": 9.551044083526682e-06,
      "loss": 1.0332550811767578,
      "step": 1275
    },
    {
      "epoch": 0.733117157761173,
      "grad_norm": 13.018166542053223,
      "learning_rate": 9.536542923433877e-06,
      "loss": 1.0661671447753907,
      "step": 1300
    },
    {
      "epoch": 0.7472155646411955,
      "grad_norm": 13.178157806396484,
      "learning_rate": 9.522041763341068e-06,
      "loss": 0.9954322052001953,
      "step": 1325
    },
    {
      "epoch": 0.761313971521218,
      "grad_norm": 15.271596908569336,
      "learning_rate": 9.50754060324826e-06,
      "loss": 1.0291824340820312,
      "step": 1350
    },
    {
      "epoch": 0.7754123784012407,
      "grad_norm": 17.327726364135742,
      "learning_rate": 9.493039443155453e-06,
      "loss": 1.0398453521728515,
      "step": 1375
    },
    {
      "epoch": 0.7895107852812632,
      "grad_norm": 13.380585670471191,
      "learning_rate": 9.478538283062646e-06,
      "loss": 0.9851990509033203,
      "step": 1400
    },
    {
      "epoch": 0.8036091921612858,
      "grad_norm": 11.947903633117676,
      "learning_rate": 9.464037122969838e-06,
      "loss": 0.9933275604248046,
      "step": 1425
    },
    {
      "epoch": 0.8177075990413083,
      "grad_norm": 11.313209533691406,
      "learning_rate": 9.449535962877031e-06,
      "loss": 0.95019775390625,
      "step": 1450
    },
    {
      "epoch": 0.8318060059213309,
      "grad_norm": 9.134963035583496,
      "learning_rate": 9.435034802784224e-06,
      "loss": 0.9632527160644532,
      "step": 1475
    },
    {
      "epoch": 0.8459044128013534,
      "grad_norm": 11.307052612304688,
      "learning_rate": 9.420533642691417e-06,
      "loss": 1.0088771820068358,
      "step": 1500
    },
    {
      "epoch": 0.860002819681376,
      "grad_norm": 11.490585327148438,
      "learning_rate": 9.406032482598608e-06,
      "loss": 0.975683822631836,
      "step": 1525
    },
    {
      "epoch": 0.8741012265613985,
      "grad_norm": 11.09744930267334,
      "learning_rate": 9.391531322505802e-06,
      "loss": 0.9516730499267578,
      "step": 1550
    },
    {
      "epoch": 0.8881996334414212,
      "grad_norm": 12.852828979492188,
      "learning_rate": 9.377030162412993e-06,
      "loss": 0.9338680267333984,
      "step": 1575
    },
    {
      "epoch": 0.9022980403214437,
      "grad_norm": 13.335673332214355,
      "learning_rate": 9.362529002320186e-06,
      "loss": 0.9313024139404297,
      "step": 1600
    },
    {
      "epoch": 0.9163964472014662,
      "grad_norm": 11.356801986694336,
      "learning_rate": 9.348027842227378e-06,
      "loss": 1.0073654174804687,
      "step": 1625
    },
    {
      "epoch": 0.9304948540814888,
      "grad_norm": 13.629708290100098,
      "learning_rate": 9.333526682134571e-06,
      "loss": 1.007170639038086,
      "step": 1650
    },
    {
      "epoch": 0.9445932609615113,
      "grad_norm": 9.606148719787598,
      "learning_rate": 9.319025522041764e-06,
      "loss": 0.9286507415771484,
      "step": 1675
    },
    {
      "epoch": 0.9586916678415339,
      "grad_norm": 9.82172679901123,
      "learning_rate": 9.304524361948957e-06,
      "loss": 0.9336747741699218,
      "step": 1700
    },
    {
      "epoch": 0.9727900747215564,
      "grad_norm": 11.669585227966309,
      "learning_rate": 9.29002320185615e-06,
      "loss": 0.9786383819580078,
      "step": 1725
    },
    {
      "epoch": 0.986888481601579,
      "grad_norm": 11.385014533996582,
      "learning_rate": 9.275522041763342e-06,
      "loss": 0.9959779357910157,
      "step": 1750
    },
    {
      "epoch": 1.0005639362752008,
      "grad_norm": 11.361459732055664,
      "learning_rate": 9.261020881670535e-06,
      "loss": 0.9319418334960937,
      "step": 1775
    },
    {
      "epoch": 1.0146623431552235,
      "grad_norm": 10.459379196166992,
      "learning_rate": 9.246519721577727e-06,
      "loss": 0.7039449310302734,
      "step": 1800
    },
    {
      "epoch": 1.028760750035246,
      "grad_norm": 8.165692329406738,
      "learning_rate": 9.23201856148492e-06,
      "loss": 0.7685092926025391,
      "step": 1825
    },
    {
      "epoch": 1.0428591569152685,
      "grad_norm": 9.223814010620117,
      "learning_rate": 9.217517401392111e-06,
      "loss": 0.7643940734863282,
      "step": 1850
    },
    {
      "epoch": 1.056957563795291,
      "grad_norm": 12.819178581237793,
      "learning_rate": 9.203016241299306e-06,
      "loss": 0.7673802185058594,
      "step": 1875
    },
    {
      "epoch": 1.0710559706753138,
      "grad_norm": 8.873549461364746,
      "learning_rate": 9.188515081206497e-06,
      "loss": 0.7162540435791016,
      "step": 1900
    },
    {
      "epoch": 1.0851543775553363,
      "grad_norm": 11.80384349822998,
      "learning_rate": 9.174013921113691e-06,
      "loss": 0.7582861328125,
      "step": 1925
    },
    {
      "epoch": 1.0992527844353588,
      "grad_norm": 10.77530288696289,
      "learning_rate": 9.159512761020882e-06,
      "loss": 0.7159561920166015,
      "step": 1950
    },
    {
      "epoch": 1.1133511913153813,
      "grad_norm": 11.646292686462402,
      "learning_rate": 9.145011600928075e-06,
      "loss": 0.7990260314941406,
      "step": 1975
    },
    {
      "epoch": 1.1274495981954038,
      "grad_norm": 11.254467010498047,
      "learning_rate": 9.130510440835267e-06,
      "loss": 0.7786456298828125,
      "step": 2000
    },
    {
      "epoch": 1.1274495981954038,
      "eval_loss": 0.2605888545513153,
      "eval_runtime": 1451.9513,
      "eval_samples_per_second": 2.134,
      "eval_steps_per_second": 0.267,
      "eval_wer": 0.22206013129365215,
      "step": 2000
    },
    {
      "epoch": 1.1415480050754265,
      "grad_norm": 8.868310928344727,
      "learning_rate": 9.11600928074246e-06,
      "loss": 0.7087932586669922,
      "step": 2025
    },
    {
      "epoch": 1.155646411955449,
      "grad_norm": 11.000504493713379,
      "learning_rate": 9.101508120649653e-06,
      "loss": 0.7248979949951172,
      "step": 2050
    },
    {
      "epoch": 1.1697448188354715,
      "grad_norm": 12.025858879089355,
      "learning_rate": 9.087006960556846e-06,
      "loss": 0.7824922943115235,
      "step": 2075
    },
    {
      "epoch": 1.1838432257154943,
      "grad_norm": 12.042730331420898,
      "learning_rate": 9.072505800464038e-06,
      "loss": 0.8172999572753906,
      "step": 2100
    },
    {
      "epoch": 1.1979416325955168,
      "grad_norm": 9.696090698242188,
      "learning_rate": 9.058004640371231e-06,
      "loss": 0.8125002288818359,
      "step": 2125
    },
    {
      "epoch": 1.2120400394755393,
      "grad_norm": 10.16943359375,
      "learning_rate": 9.043503480278422e-06,
      "loss": 0.76618408203125,
      "step": 2150
    },
    {
      "epoch": 1.2261384463555618,
      "grad_norm": 9.093573570251465,
      "learning_rate": 9.029002320185616e-06,
      "loss": 0.7104488372802734,
      "step": 2175
    },
    {
      "epoch": 1.2402368532355843,
      "grad_norm": 10.546662330627441,
      "learning_rate": 9.014501160092808e-06,
      "loss": 0.7818730926513672,
      "step": 2200
    },
    {
      "epoch": 1.254335260115607,
      "grad_norm": 12.367358207702637,
      "learning_rate": 9e-06,
      "loss": 0.8542975616455079,
      "step": 2225
    },
    {
      "epoch": 1.2684336669956295,
      "grad_norm": 10.952892303466797,
      "learning_rate": 8.985498839907193e-06,
      "loss": 0.7449074554443359,
      "step": 2250
    },
    {
      "epoch": 1.282532073875652,
      "grad_norm": 8.846220970153809,
      "learning_rate": 8.970997679814386e-06,
      "loss": 0.7692269897460937,
      "step": 2275
    },
    {
      "epoch": 1.2966304807556746,
      "grad_norm": 11.092299461364746,
      "learning_rate": 8.956496519721578e-06,
      "loss": 0.770924072265625,
      "step": 2300
    },
    {
      "epoch": 1.310728887635697,
      "grad_norm": 11.600993156433105,
      "learning_rate": 8.941995359628771e-06,
      "loss": 0.8001760864257812,
      "step": 2325
    },
    {
      "epoch": 1.3248272945157198,
      "grad_norm": 9.988462448120117,
      "learning_rate": 8.927494199535964e-06,
      "loss": 0.736719970703125,
      "step": 2350
    },
    {
      "epoch": 1.3389257013957423,
      "grad_norm": 11.111364364624023,
      "learning_rate": 8.912993039443157e-06,
      "loss": 0.7216539001464843,
      "step": 2375
    },
    {
      "epoch": 1.3530241082757648,
      "grad_norm": 12.4348726272583,
      "learning_rate": 8.898491879350348e-06,
      "loss": 0.8057781219482422,
      "step": 2400
    },
    {
      "epoch": 1.3671225151557875,
      "grad_norm": 11.37057876586914,
      "learning_rate": 8.883990719257542e-06,
      "loss": 0.7377596282958985,
      "step": 2425
    },
    {
      "epoch": 1.3812209220358098,
      "grad_norm": 8.443863868713379,
      "learning_rate": 8.869489559164733e-06,
      "loss": 0.764500732421875,
      "step": 2450
    },
    {
      "epoch": 1.3953193289158325,
      "grad_norm": 11.640331268310547,
      "learning_rate": 8.854988399071927e-06,
      "loss": 0.7854479217529297,
      "step": 2475
    },
    {
      "epoch": 1.409417735795855,
      "grad_norm": 9.915708541870117,
      "learning_rate": 8.840487238979118e-06,
      "loss": 0.7995922088623046,
      "step": 2500
    },
    {
      "epoch": 1.4235161426758776,
      "grad_norm": 12.385272979736328,
      "learning_rate": 8.825986078886311e-06,
      "loss": 0.7540443420410157,
      "step": 2525
    },
    {
      "epoch": 1.4376145495559003,
      "grad_norm": 10.713929176330566,
      "learning_rate": 8.811484918793504e-06,
      "loss": 0.7403028869628906,
      "step": 2550
    },
    {
      "epoch": 1.4517129564359228,
      "grad_norm": 10.878042221069336,
      "learning_rate": 8.796983758700697e-06,
      "loss": 0.7737533569335937,
      "step": 2575
    },
    {
      "epoch": 1.4658113633159453,
      "grad_norm": 9.902074813842773,
      "learning_rate": 8.78248259860789e-06,
      "loss": 0.752545166015625,
      "step": 2600
    },
    {
      "epoch": 1.4799097701959678,
      "grad_norm": 9.259173393249512,
      "learning_rate": 8.767981438515082e-06,
      "loss": 0.7878742980957031,
      "step": 2625
    },
    {
      "epoch": 1.4940081770759903,
      "grad_norm": 11.374395370483398,
      "learning_rate": 8.753480278422275e-06,
      "loss": 0.7678947448730469,
      "step": 2650
    },
    {
      "epoch": 1.508106583956013,
      "grad_norm": 10.379016876220703,
      "learning_rate": 8.738979118329467e-06,
      "loss": 0.7748676300048828,
      "step": 2675
    },
    {
      "epoch": 1.5222049908360356,
      "grad_norm": 11.049686431884766,
      "learning_rate": 8.72447795823666e-06,
      "loss": 0.7766862487792969,
      "step": 2700
    },
    {
      "epoch": 1.536303397716058,
      "grad_norm": 10.903966903686523,
      "learning_rate": 8.709976798143853e-06,
      "loss": 0.7612065124511719,
      "step": 2725
    },
    {
      "epoch": 1.5504018045960808,
      "grad_norm": 11.366527557373047,
      "learning_rate": 8.695475638051046e-06,
      "loss": 0.7193534088134765,
      "step": 2750
    },
    {
      "epoch": 1.564500211476103,
      "grad_norm": 9.587455749511719,
      "learning_rate": 8.680974477958237e-06,
      "loss": 0.6888518524169922,
      "step": 2775
    },
    {
      "epoch": 1.5785986183561258,
      "grad_norm": 10.296839714050293,
      "learning_rate": 8.666473317865431e-06,
      "loss": 0.7584939575195313,
      "step": 2800
    },
    {
      "epoch": 1.5926970252361483,
      "grad_norm": 13.031759262084961,
      "learning_rate": 8.651972157772622e-06,
      "loss": 0.8102165985107422,
      "step": 2825
    },
    {
      "epoch": 1.6067954321161708,
      "grad_norm": 10.904520988464355,
      "learning_rate": 8.637470997679815e-06,
      "loss": 0.7805465698242188,
      "step": 2850
    },
    {
      "epoch": 1.6208938389961935,
      "grad_norm": 9.31521224975586,
      "learning_rate": 8.622969837587007e-06,
      "loss": 0.7300022125244141,
      "step": 2875
    },
    {
      "epoch": 1.6349922458762158,
      "grad_norm": 11.240145683288574,
      "learning_rate": 8.6084686774942e-06,
      "loss": 0.783562240600586,
      "step": 2900
    },
    {
      "epoch": 1.6490906527562386,
      "grad_norm": 10.595625877380371,
      "learning_rate": 8.593967517401393e-06,
      "loss": 0.7286166381835938,
      "step": 2925
    },
    {
      "epoch": 1.663189059636261,
      "grad_norm": 9.539634704589844,
      "learning_rate": 8.579466357308586e-06,
      "loss": 0.7374664306640625,
      "step": 2950
    },
    {
      "epoch": 1.6772874665162836,
      "grad_norm": 11.489903450012207,
      "learning_rate": 8.564965197215778e-06,
      "loss": 0.7891261291503906,
      "step": 2975
    },
    {
      "epoch": 1.6913858733963063,
      "grad_norm": 11.152668952941895,
      "learning_rate": 8.550464037122971e-06,
      "loss": 0.750862808227539,
      "step": 3000
    },
    {
      "epoch": 1.6913858733963063,
      "eval_loss": 0.2470168024301529,
      "eval_runtime": 1517.4032,
      "eval_samples_per_second": 2.042,
      "eval_steps_per_second": 0.256,
      "eval_wer": 0.21129884793317413,
      "step": 3000
    },
    {
      "epoch": 1.7054842802763288,
      "grad_norm": 10.490551948547363,
      "learning_rate": 8.535962877030162e-06,
      "loss": 0.7084814453125,
      "step": 3025
    },
    {
      "epoch": 1.7195826871563513,
      "grad_norm": 11.868510246276855,
      "learning_rate": 8.521461716937356e-06,
      "loss": 0.7457318115234375,
      "step": 3050
    },
    {
      "epoch": 1.733681094036374,
      "grad_norm": 9.93341064453125,
      "learning_rate": 8.506960556844547e-06,
      "loss": 0.7481878662109375,
      "step": 3075
    },
    {
      "epoch": 1.7477795009163963,
      "grad_norm": 9.801637649536133,
      "learning_rate": 8.492459396751742e-06,
      "loss": 0.7577263641357422,
      "step": 3100
    },
    {
      "epoch": 1.761877907796419,
      "grad_norm": 9.962018013000488,
      "learning_rate": 8.477958236658933e-06,
      "loss": 0.7176610565185547,
      "step": 3125
    },
    {
      "epoch": 1.7759763146764416,
      "grad_norm": 10.016621589660645,
      "learning_rate": 8.463457076566126e-06,
      "loss": 0.8129973602294922,
      "step": 3150
    },
    {
      "epoch": 1.790074721556464,
      "grad_norm": 10.95247745513916,
      "learning_rate": 8.448955916473318e-06,
      "loss": 0.76224609375,
      "step": 3175
    },
    {
      "epoch": 1.8041731284364868,
      "grad_norm": 10.412908554077148,
      "learning_rate": 8.434454756380511e-06,
      "loss": 0.72124267578125,
      "step": 3200
    },
    {
      "epoch": 1.818271535316509,
      "grad_norm": 10.183904647827148,
      "learning_rate": 8.419953596287704e-06,
      "loss": 0.7625586700439453,
      "step": 3225
    },
    {
      "epoch": 1.8323699421965318,
      "grad_norm": 10.891714096069336,
      "learning_rate": 8.405452436194896e-06,
      "loss": 0.719871826171875,
      "step": 3250
    },
    {
      "epoch": 1.8464683490765543,
      "grad_norm": 10.172575950622559,
      "learning_rate": 8.390951276102089e-06,
      "loss": 0.753576889038086,
      "step": 3275
    },
    {
      "epoch": 1.8605667559565768,
      "grad_norm": 9.48585033416748,
      "learning_rate": 8.376450116009282e-06,
      "loss": 0.7430252838134765,
      "step": 3300
    },
    {
      "epoch": 1.8746651628365996,
      "grad_norm": 11.499151229858398,
      "learning_rate": 8.361948955916473e-06,
      "loss": 0.7010813903808594,
      "step": 3325
    },
    {
      "epoch": 1.888763569716622,
      "grad_norm": 9.432136535644531,
      "learning_rate": 8.347447795823667e-06,
      "loss": 0.71728759765625,
      "step": 3350
    },
    {
      "epoch": 1.9028619765966446,
      "grad_norm": 8.735350608825684,
      "learning_rate": 8.332946635730858e-06,
      "loss": 0.8144132232666016,
      "step": 3375
    },
    {
      "epoch": 1.916960383476667,
      "grad_norm": 9.190115928649902,
      "learning_rate": 8.318445475638051e-06,
      "loss": 0.7769715881347656,
      "step": 3400
    },
    {
      "epoch": 1.9310587903566896,
      "grad_norm": 8.803107261657715,
      "learning_rate": 8.303944315545245e-06,
      "loss": 0.7007711791992187,
      "step": 3425
    },
    {
      "epoch": 1.9451571972367123,
      "grad_norm": 10.293452262878418,
      "learning_rate": 8.289443155452436e-06,
      "loss": 0.7087128448486328,
      "step": 3450
    },
    {
      "epoch": 1.9592556041167348,
      "grad_norm": 11.334561347961426,
      "learning_rate": 8.27494199535963e-06,
      "loss": 0.7811639404296875,
      "step": 3475
    },
    {
      "epoch": 1.9733540109967573,
      "grad_norm": 9.957331657409668,
      "learning_rate": 8.260440835266822e-06,
      "loss": 0.7865208435058594,
      "step": 3500
    },
    {
      "epoch": 1.98745241787678,
      "grad_norm": 12.995558738708496,
      "learning_rate": 8.245939675174015e-06,
      "loss": 0.7895949554443359,
      "step": 3525
    },
    {
      "epoch": 2.0011278725504016,
      "grad_norm": 9.675638198852539,
      "learning_rate": 8.231438515081207e-06,
      "loss": 0.721746826171875,
      "step": 3550
    },
    {
      "epoch": 2.0152262794304243,
      "grad_norm": 18.80123519897461,
      "learning_rate": 8.2169373549884e-06,
      "loss": 0.557889175415039,
      "step": 3575
    },
    {
      "epoch": 2.029324686310447,
      "grad_norm": 9.540290832519531,
      "learning_rate": 8.202436194895593e-06,
      "loss": 0.5269033432006835,
      "step": 3600
    },
    {
      "epoch": 2.0434230931904693,
      "grad_norm": 8.743149757385254,
      "learning_rate": 8.187935034802785e-06,
      "loss": 0.5314161300659179,
      "step": 3625
    },
    {
      "epoch": 2.057521500070492,
      "grad_norm": 8.326909065246582,
      "learning_rate": 8.173433874709976e-06,
      "loss": 0.5158148193359375,
      "step": 3650
    },
    {
      "epoch": 2.071619906950515,
      "grad_norm": 8.55147933959961,
      "learning_rate": 8.158932714617171e-06,
      "loss": 0.543358497619629,
      "step": 3675
    },
    {
      "epoch": 2.085718313830537,
      "grad_norm": 9.927057266235352,
      "learning_rate": 8.144431554524362e-06,
      "loss": 0.5452922058105468,
      "step": 3700
    },
    {
      "epoch": 2.09981672071056,
      "grad_norm": 7.827384948730469,
      "learning_rate": 8.129930394431556e-06,
      "loss": 0.5253535079956054,
      "step": 3725
    },
    {
      "epoch": 2.113915127590582,
      "grad_norm": 9.569947242736816,
      "learning_rate": 8.115429234338747e-06,
      "loss": 0.5380427169799805,
      "step": 3750
    },
    {
      "epoch": 2.128013534470605,
      "grad_norm": 8.02529525756836,
      "learning_rate": 8.10092807424594e-06,
      "loss": 0.50870849609375,
      "step": 3775
    },
    {
      "epoch": 2.1421119413506275,
      "grad_norm": 10.682024002075195,
      "learning_rate": 8.086426914153133e-06,
      "loss": 0.5357696914672851,
      "step": 3800
    },
    {
      "epoch": 2.15621034823065,
      "grad_norm": 7.825737476348877,
      "learning_rate": 8.071925754060325e-06,
      "loss": 0.5354624176025391,
      "step": 3825
    },
    {
      "epoch": 2.1703087551106726,
      "grad_norm": 8.716205596923828,
      "learning_rate": 8.057424593967518e-06,
      "loss": 0.5709107208251953,
      "step": 3850
    },
    {
      "epoch": 2.184407161990695,
      "grad_norm": 8.497817993164062,
      "learning_rate": 8.042923433874711e-06,
      "loss": 0.559893798828125,
      "step": 3875
    },
    {
      "epoch": 2.1985055688707176,
      "grad_norm": 10.781414985656738,
      "learning_rate": 8.028422273781904e-06,
      "loss": 0.5544267272949219,
      "step": 3900
    },
    {
      "epoch": 2.2126039757507403,
      "grad_norm": 10.681407928466797,
      "learning_rate": 8.013921113689096e-06,
      "loss": 0.533585090637207,
      "step": 3925
    },
    {
      "epoch": 2.2267023826307626,
      "grad_norm": 9.05926513671875,
      "learning_rate": 7.999419953596287e-06,
      "loss": 0.5441395950317383,
      "step": 3950
    },
    {
      "epoch": 2.2408007895107853,
      "grad_norm": 10.925969123840332,
      "learning_rate": 7.984918793503482e-06,
      "loss": 0.5743826675415039,
      "step": 3975
    },
    {
      "epoch": 2.2548991963908076,
      "grad_norm": 9.930057525634766,
      "learning_rate": 7.970417633410673e-06,
      "loss": 0.5230157089233398,
      "step": 4000
    },
    {
      "epoch": 2.2548991963908076,
      "eval_loss": 0.25168365240097046,
      "eval_runtime": 1521.1061,
      "eval_samples_per_second": 2.037,
      "eval_steps_per_second": 0.255,
      "eval_wer": 0.20968015907115237,
      "step": 4000
    },
    {
      "epoch": 2.2689976032708303,
      "grad_norm": 11.271224021911621,
      "learning_rate": 7.955916473317865e-06,
      "loss": 0.5519426727294922,
      "step": 4025
    },
    {
      "epoch": 2.283096010150853,
      "grad_norm": 9.23180103302002,
      "learning_rate": 7.941415313225058e-06,
      "loss": 0.5422761917114258,
      "step": 4050
    },
    {
      "epoch": 2.2971944170308753,
      "grad_norm": 8.428093910217285,
      "learning_rate": 7.926914153132251e-06,
      "loss": 0.5573156356811524,
      "step": 4075
    },
    {
      "epoch": 2.311292823910898,
      "grad_norm": 9.771265029907227,
      "learning_rate": 7.912412993039444e-06,
      "loss": 0.5444869995117188,
      "step": 4100
    },
    {
      "epoch": 2.325391230790921,
      "grad_norm": 8.109742164611816,
      "learning_rate": 7.897911832946636e-06,
      "loss": 0.565984001159668,
      "step": 4125
    },
    {
      "epoch": 2.339489637670943,
      "grad_norm": 10.035289764404297,
      "learning_rate": 7.883410672853829e-06,
      "loss": 0.5919873809814453,
      "step": 4150
    },
    {
      "epoch": 2.353588044550966,
      "grad_norm": 7.981551647186279,
      "learning_rate": 7.868909512761022e-06,
      "loss": 0.5728730392456055,
      "step": 4175
    },
    {
      "epoch": 2.3676864514309885,
      "grad_norm": 9.20626163482666,
      "learning_rate": 7.854408352668213e-06,
      "loss": 0.5345810317993164,
      "step": 4200
    },
    {
      "epoch": 2.381784858311011,
      "grad_norm": 10.04966926574707,
      "learning_rate": 7.839907192575407e-06,
      "loss": 0.5514390563964844,
      "step": 4225
    },
    {
      "epoch": 2.3958832651910336,
      "grad_norm": 9.066740989685059,
      "learning_rate": 7.8254060324826e-06,
      "loss": 0.5442893600463867,
      "step": 4250
    },
    {
      "epoch": 2.409981672071056,
      "grad_norm": 8.016524314880371,
      "learning_rate": 7.810904872389791e-06,
      "loss": 0.5711633682250976,
      "step": 4275
    },
    {
      "epoch": 2.4240800789510786,
      "grad_norm": 9.421856880187988,
      "learning_rate": 7.796403712296985e-06,
      "loss": 0.5826901626586914,
      "step": 4300
    },
    {
      "epoch": 2.4381784858311013,
      "grad_norm": 9.801589965820312,
      "learning_rate": 7.781902552204176e-06,
      "loss": 0.5632424545288086,
      "step": 4325
    },
    {
      "epoch": 2.4522768927111236,
      "grad_norm": 7.847035884857178,
      "learning_rate": 7.76740139211137e-06,
      "loss": 0.5748075485229492,
      "step": 4350
    },
    {
      "epoch": 2.4663752995911463,
      "grad_norm": 8.85059928894043,
      "learning_rate": 7.752900232018562e-06,
      "loss": 0.5258598327636719,
      "step": 4375
    },
    {
      "epoch": 2.4804737064711686,
      "grad_norm": 9.43190860748291,
      "learning_rate": 7.738399071925755e-06,
      "loss": 0.5532307815551758,
      "step": 4400
    },
    {
      "epoch": 2.4945721133511913,
      "grad_norm": 8.644095420837402,
      "learning_rate": 7.723897911832947e-06,
      "loss": 0.5246799850463867,
      "step": 4425
    },
    {
      "epoch": 2.508670520231214,
      "grad_norm": 10.166682243347168,
      "learning_rate": 7.70939675174014e-06,
      "loss": 0.5615069198608399,
      "step": 4450
    },
    {
      "epoch": 2.5227689271112363,
      "grad_norm": 8.29073429107666,
      "learning_rate": 7.694895591647333e-06,
      "loss": 0.5665618133544922,
      "step": 4475
    },
    {
      "epoch": 2.536867333991259,
      "grad_norm": 11.046154975891113,
      "learning_rate": 7.680394431554525e-06,
      "loss": 0.5597576904296875,
      "step": 4500
    },
    {
      "epoch": 2.5509657408712814,
      "grad_norm": 8.011266708374023,
      "learning_rate": 7.665893271461718e-06,
      "loss": 0.5483290863037109,
      "step": 4525
    },
    {
      "epoch": 2.565064147751304,
      "grad_norm": 8.506767272949219,
      "learning_rate": 7.65139211136891e-06,
      "loss": 0.5561467361450195,
      "step": 4550
    },
    {
      "epoch": 2.579162554631327,
      "grad_norm": 8.871068954467773,
      "learning_rate": 7.636890951276102e-06,
      "loss": 0.581519546508789,
      "step": 4575
    },
    {
      "epoch": 2.593260961511349,
      "grad_norm": 7.879359722137451,
      "learning_rate": 7.622389791183295e-06,
      "loss": 0.5407870483398437,
      "step": 4600
    },
    {
      "epoch": 2.607359368391372,
      "grad_norm": 6.408371448516846,
      "learning_rate": 7.607888631090487e-06,
      "loss": 0.5560993576049804,
      "step": 4625
    },
    {
      "epoch": 2.621457775271394,
      "grad_norm": 8.178828239440918,
      "learning_rate": 7.593387470997681e-06,
      "loss": 0.5660905838012695,
      "step": 4650
    },
    {
      "epoch": 2.635556182151417,
      "grad_norm": 8.56795597076416,
      "learning_rate": 7.578886310904873e-06,
      "loss": 0.5393736648559571,
      "step": 4675
    },
    {
      "epoch": 2.6496545890314396,
      "grad_norm": 8.56303596496582,
      "learning_rate": 7.564385150812066e-06,
      "loss": 0.5549613952636718,
      "step": 4700
    },
    {
      "epoch": 2.663752995911462,
      "grad_norm": 8.995079040527344,
      "learning_rate": 7.549883990719258e-06,
      "loss": 0.5543619537353516,
      "step": 4725
    },
    {
      "epoch": 2.6778514027914846,
      "grad_norm": 9.455941200256348,
      "learning_rate": 7.535382830626451e-06,
      "loss": 0.5422988510131836,
      "step": 4750
    },
    {
      "epoch": 2.691949809671507,
      "grad_norm": 9.779569625854492,
      "learning_rate": 7.520881670533643e-06,
      "loss": 0.5424752044677734,
      "step": 4775
    },
    {
      "epoch": 2.7060482165515296,
      "grad_norm": 9.737711906433105,
      "learning_rate": 7.506380510440836e-06,
      "loss": 0.5491796112060547,
      "step": 4800
    },
    {
      "epoch": 2.7201466234315523,
      "grad_norm": 10.25398063659668,
      "learning_rate": 7.491879350348028e-06,
      "loss": 0.5705435180664062,
      "step": 4825
    },
    {
      "epoch": 2.734245030311575,
      "grad_norm": 9.573110580444336,
      "learning_rate": 7.477378190255221e-06,
      "loss": 0.5880419540405274,
      "step": 4850
    },
    {
      "epoch": 2.7483434371915973,
      "grad_norm": 11.062814712524414,
      "learning_rate": 7.4628770301624135e-06,
      "loss": 0.556083984375,
      "step": 4875
    },
    {
      "epoch": 2.7624418440716196,
      "grad_norm": 9.247238159179688,
      "learning_rate": 7.448375870069606e-06,
      "loss": 0.5655975341796875,
      "step": 4900
    },
    {
      "epoch": 2.7765402509516424,
      "grad_norm": 9.007148742675781,
      "learning_rate": 7.433874709976798e-06,
      "loss": 0.5616880035400391,
      "step": 4925
    },
    {
      "epoch": 2.790638657831665,
      "grad_norm": 8.514342308044434,
      "learning_rate": 7.419373549883992e-06,
      "loss": 0.5538288116455078,
      "step": 4950
    },
    {
      "epoch": 2.804737064711688,
      "grad_norm": 9.440685272216797,
      "learning_rate": 7.4048723897911835e-06,
      "loss": 0.5948199081420898,
      "step": 4975
    },
    {
      "epoch": 2.81883547159171,
      "grad_norm": 6.956933975219727,
      "learning_rate": 7.390371229698376e-06,
      "loss": 0.5486139678955078,
      "step": 5000
    },
    {
      "epoch": 2.81883547159171,
      "eval_loss": 0.24652095139026642,
      "eval_runtime": 1532.9367,
      "eval_samples_per_second": 2.022,
      "eval_steps_per_second": 0.253,
      "eval_wer": 0.20783165635834974,
      "step": 5000
    },
    {
      "epoch": 2.832933878471733,
      "grad_norm": 9.058638572692871,
      "learning_rate": 7.375870069605568e-06,
      "loss": 0.5798805999755859,
      "step": 5025
    },
    {
      "epoch": 2.847032285351755,
      "grad_norm": 7.628641605377197,
      "learning_rate": 7.361368909512762e-06,
      "loss": 0.5301958847045899,
      "step": 5050
    },
    {
      "epoch": 2.861130692231778,
      "grad_norm": 8.487007141113281,
      "learning_rate": 7.346867749419954e-06,
      "loss": 0.566702880859375,
      "step": 5075
    },
    {
      "epoch": 2.8752290991118006,
      "grad_norm": 9.008893013000488,
      "learning_rate": 7.332366589327147e-06,
      "loss": 0.564128532409668,
      "step": 5100
    },
    {
      "epoch": 2.889327505991823,
      "grad_norm": 11.27266788482666,
      "learning_rate": 7.31786542923434e-06,
      "loss": 0.5648199462890625,
      "step": 5125
    },
    {
      "epoch": 2.9034259128718456,
      "grad_norm": 9.269525527954102,
      "learning_rate": 7.303364269141532e-06,
      "loss": 0.5541753005981446,
      "step": 5150
    },
    {
      "epoch": 2.917524319751868,
      "grad_norm": 9.549739837646484,
      "learning_rate": 7.288863109048725e-06,
      "loss": 0.5463447952270508,
      "step": 5175
    },
    {
      "epoch": 2.9316227266318906,
      "grad_norm": 9.327741622924805,
      "learning_rate": 7.274361948955917e-06,
      "loss": 0.5498114395141601,
      "step": 5200
    },
    {
      "epoch": 2.9457211335119133,
      "grad_norm": 10.359440803527832,
      "learning_rate": 7.25986078886311e-06,
      "loss": 0.5578446578979492,
      "step": 5225
    },
    {
      "epoch": 2.9598195403919356,
      "grad_norm": 7.895180702209473,
      "learning_rate": 7.245359628770302e-06,
      "loss": 0.5610840606689453,
      "step": 5250
    },
    {
      "epoch": 2.9739179472719584,
      "grad_norm": 11.096906661987305,
      "learning_rate": 7.230858468677495e-06,
      "loss": 0.5936727523803711,
      "step": 5275
    },
    {
      "epoch": 2.9880163541519806,
      "grad_norm": 10.793472290039062,
      "learning_rate": 7.216357308584687e-06,
      "loss": 0.5792999649047852,
      "step": 5300
    },
    {
      "epoch": 3.0016918088256026,
      "grad_norm": 8.622193336486816,
      "learning_rate": 7.201856148491881e-06,
      "loss": 0.5246665573120117,
      "step": 5325
    },
    {
      "epoch": 3.0157902157056253,
      "grad_norm": 6.704568386077881,
      "learning_rate": 7.1873549883990726e-06,
      "loss": 0.40513866424560546,
      "step": 5350
    },
    {
      "epoch": 3.0298886225856476,
      "grad_norm": 7.683806419372559,
      "learning_rate": 7.172853828306265e-06,
      "loss": 0.3967144775390625,
      "step": 5375
    },
    {
      "epoch": 3.0439870294656703,
      "grad_norm": 7.537134647369385,
      "learning_rate": 7.158352668213457e-06,
      "loss": 0.39253467559814453,
      "step": 5400
    },
    {
      "epoch": 3.058085436345693,
      "grad_norm": 6.951216697692871,
      "learning_rate": 7.143851508120651e-06,
      "loss": 0.374131965637207,
      "step": 5425
    },
    {
      "epoch": 3.0721838432257154,
      "grad_norm": 7.260530471801758,
      "learning_rate": 7.129350348027843e-06,
      "loss": 0.35717914581298826,
      "step": 5450
    },
    {
      "epoch": 3.086282250105738,
      "grad_norm": 7.6884989738464355,
      "learning_rate": 7.114849187935035e-06,
      "loss": 0.35510223388671874,
      "step": 5475
    },
    {
      "epoch": 3.100380656985761,
      "grad_norm": 7.6629109382629395,
      "learning_rate": 7.100348027842228e-06,
      "loss": 0.35819530487060547,
      "step": 5500
    },
    {
      "epoch": 3.114479063865783,
      "grad_norm": 7.406759262084961,
      "learning_rate": 7.085846867749421e-06,
      "loss": 0.3919057846069336,
      "step": 5525
    },
    {
      "epoch": 3.128577470745806,
      "grad_norm": 8.94927978515625,
      "learning_rate": 7.071345707656613e-06,
      "loss": 0.3659718704223633,
      "step": 5550
    },
    {
      "epoch": 3.142675877625828,
      "grad_norm": 8.768269538879395,
      "learning_rate": 7.056844547563806e-06,
      "loss": 0.37319766998291015,
      "step": 5575
    },
    {
      "epoch": 3.156774284505851,
      "grad_norm": 8.374007225036621,
      "learning_rate": 7.042343387470998e-06,
      "loss": 0.3749269104003906,
      "step": 5600
    },
    {
      "epoch": 3.1708726913858736,
      "grad_norm": 6.927631855010986,
      "learning_rate": 7.027842227378191e-06,
      "loss": 0.3902094650268555,
      "step": 5625
    },
    {
      "epoch": 3.184971098265896,
      "grad_norm": 8.199101448059082,
      "learning_rate": 7.013341067285383e-06,
      "loss": 0.3912439727783203,
      "step": 5650
    },
    {
      "epoch": 3.1990695051459186,
      "grad_norm": 6.719486713409424,
      "learning_rate": 6.998839907192576e-06,
      "loss": 0.36518966674804687,
      "step": 5675
    },
    {
      "epoch": 3.213167912025941,
      "grad_norm": 8.620352745056152,
      "learning_rate": 6.984338747099768e-06,
      "loss": 0.3811537551879883,
      "step": 5700
    },
    {
      "epoch": 3.2272663189059636,
      "grad_norm": 7.2147040367126465,
      "learning_rate": 6.969837587006962e-06,
      "loss": 0.37800453186035154,
      "step": 5725
    },
    {
      "epoch": 3.2413647257859863,
      "grad_norm": 8.890436172485352,
      "learning_rate": 6.9553364269141535e-06,
      "loss": 0.3713486480712891,
      "step": 5750
    },
    {
      "epoch": 3.2554631326660086,
      "grad_norm": 8.093574523925781,
      "learning_rate": 6.940835266821346e-06,
      "loss": 0.4041537857055664,
      "step": 5775
    },
    {
      "epoch": 3.2695615395460313,
      "grad_norm": 9.059134483337402,
      "learning_rate": 6.926334106728538e-06,
      "loss": 0.397373046875,
      "step": 5800
    },
    {
      "epoch": 3.283659946426054,
      "grad_norm": 9.615391731262207,
      "learning_rate": 6.911832946635732e-06,
      "loss": 0.3877538299560547,
      "step": 5825
    },
    {
      "epoch": 3.2977583533060764,
      "grad_norm": 7.0843186378479,
      "learning_rate": 6.8973317865429235e-06,
      "loss": 0.38612773895263675,
      "step": 5850
    },
    {
      "epoch": 3.311856760186099,
      "grad_norm": 7.724761009216309,
      "learning_rate": 6.882830626450116e-06,
      "loss": 0.3913743209838867,
      "step": 5875
    },
    {
      "epoch": 3.3259551670661214,
      "grad_norm": 7.574341297149658,
      "learning_rate": 6.86832946635731e-06,
      "loss": 0.36103832244873046,
      "step": 5900
    },
    {
      "epoch": 3.340053573946144,
      "grad_norm": 9.554155349731445,
      "learning_rate": 6.853828306264502e-06,
      "loss": 0.39909862518310546,
      "step": 5925
    },
    {
      "epoch": 3.354151980826167,
      "grad_norm": 9.424493789672852,
      "learning_rate": 6.839327146171695e-06,
      "loss": 0.36129764556884764,
      "step": 5950
    },
    {
      "epoch": 3.368250387706189,
      "grad_norm": 8.635687828063965,
      "learning_rate": 6.824825986078887e-06,
      "loss": 0.4005467987060547,
      "step": 5975
    },
    {
      "epoch": 3.382348794586212,
      "grad_norm": 8.711161613464355,
      "learning_rate": 6.81032482598608e-06,
      "loss": 0.37857559204101565,
      "step": 6000
    },
    {
      "epoch": 3.382348794586212,
      "eval_loss": 0.2660383880138397,
      "eval_runtime": 1540.0218,
      "eval_samples_per_second": 2.012,
      "eval_steps_per_second": 0.252,
      "eval_wer": 0.21214815999040776,
      "step": 6000
    },
    {
      "epoch": 3.396447201466234,
      "grad_norm": 8.4149751663208,
      "learning_rate": 6.795823665893272e-06,
      "loss": 0.38818477630615233,
      "step": 6025
    },
    {
      "epoch": 3.410545608346257,
      "grad_norm": 7.9141974449157715,
      "learning_rate": 6.781322505800465e-06,
      "loss": 0.3774140930175781,
      "step": 6050
    },
    {
      "epoch": 3.4246440152262796,
      "grad_norm": 8.901697158813477,
      "learning_rate": 6.766821345707657e-06,
      "loss": 0.37076019287109374,
      "step": 6075
    },
    {
      "epoch": 3.438742422106302,
      "grad_norm": 7.5369696617126465,
      "learning_rate": 6.75232018561485e-06,
      "loss": 0.3822758102416992,
      "step": 6100
    },
    {
      "epoch": 3.4528408289863246,
      "grad_norm": 8.009127616882324,
      "learning_rate": 6.7378190255220425e-06,
      "loss": 0.38495445251464844,
      "step": 6125
    },
    {
      "epoch": 3.466939235866347,
      "grad_norm": 8.295459747314453,
      "learning_rate": 6.723317865429235e-06,
      "loss": 0.3907606887817383,
      "step": 6150
    },
    {
      "epoch": 3.4810376427463696,
      "grad_norm": 10.099355697631836,
      "learning_rate": 6.708816705336427e-06,
      "loss": 0.407639274597168,
      "step": 6175
    },
    {
      "epoch": 3.4951360496263923,
      "grad_norm": 7.597548961639404,
      "learning_rate": 6.694315545243621e-06,
      "loss": 0.37315830230712893,
      "step": 6200
    },
    {
      "epoch": 3.5092344565064146,
      "grad_norm": 8.139415740966797,
      "learning_rate": 6.6798143851508125e-06,
      "loss": 0.3879343795776367,
      "step": 6225
    },
    {
      "epoch": 3.5233328633864374,
      "grad_norm": 8.672515869140625,
      "learning_rate": 6.665313225058005e-06,
      "loss": 0.37412925720214846,
      "step": 6250
    },
    {
      "epoch": 3.5374312702664596,
      "grad_norm": 6.852619171142578,
      "learning_rate": 6.650812064965198e-06,
      "loss": 0.3825421142578125,
      "step": 6275
    },
    {
      "epoch": 3.5515296771464824,
      "grad_norm": 7.91625452041626,
      "learning_rate": 6.636310904872391e-06,
      "loss": 0.3875956726074219,
      "step": 6300
    },
    {
      "epoch": 3.565628084026505,
      "grad_norm": 8.165389060974121,
      "learning_rate": 6.6218097447795825e-06,
      "loss": 0.39700664520263673,
      "step": 6325
    },
    {
      "epoch": 3.579726490906528,
      "grad_norm": 7.832555294036865,
      "learning_rate": 6.607308584686776e-06,
      "loss": 0.37678192138671873,
      "step": 6350
    },
    {
      "epoch": 3.59382489778655,
      "grad_norm": 9.242694854736328,
      "learning_rate": 6.592807424593968e-06,
      "loss": 0.3898345184326172,
      "step": 6375
    },
    {
      "epoch": 3.6079233046665724,
      "grad_norm": 10.566645622253418,
      "learning_rate": 6.578306264501161e-06,
      "loss": 0.41113948822021484,
      "step": 6400
    },
    {
      "epoch": 3.622021711546595,
      "grad_norm": 8.03178882598877,
      "learning_rate": 6.5638051044083525e-06,
      "loss": 0.3778879165649414,
      "step": 6425
    },
    {
      "epoch": 3.636120118426618,
      "grad_norm": 9.302940368652344,
      "learning_rate": 6.549303944315546e-06,
      "loss": 0.4171958541870117,
      "step": 6450
    },
    {
      "epoch": 3.6502185253066406,
      "grad_norm": 8.460973739624023,
      "learning_rate": 6.534802784222738e-06,
      "loss": 0.39614273071289063,
      "step": 6475
    },
    {
      "epoch": 3.664316932186663,
      "grad_norm": 7.800125598907471,
      "learning_rate": 6.520301624129931e-06,
      "loss": 0.3654580307006836,
      "step": 6500
    },
    {
      "epoch": 3.6784153390666856,
      "grad_norm": 9.029678344726562,
      "learning_rate": 6.505800464037123e-06,
      "loss": 0.37184043884277346,
      "step": 6525
    },
    {
      "epoch": 3.692513745946708,
      "grad_norm": 7.120487689971924,
      "learning_rate": 6.491299303944316e-06,
      "loss": 0.3724634552001953,
      "step": 6550
    },
    {
      "epoch": 3.7066121528267306,
      "grad_norm": 7.917516708374023,
      "learning_rate": 6.476798143851508e-06,
      "loss": 0.3739112091064453,
      "step": 6575
    },
    {
      "epoch": 3.7207105597067534,
      "grad_norm": 8.271352767944336,
      "learning_rate": 6.4622969837587015e-06,
      "loss": 0.401793212890625,
      "step": 6600
    },
    {
      "epoch": 3.7348089665867756,
      "grad_norm": 7.502166748046875,
      "learning_rate": 6.447795823665893e-06,
      "loss": 0.42757793426513674,
      "step": 6625
    },
    {
      "epoch": 3.7489073734667984,
      "grad_norm": 7.828073024749756,
      "learning_rate": 6.433294663573086e-06,
      "loss": 0.4309410095214844,
      "step": 6650
    },
    {
      "epoch": 3.7630057803468207,
      "grad_norm": 8.782448768615723,
      "learning_rate": 6.418793503480279e-06,
      "loss": 0.40100780487060544,
      "step": 6675
    },
    {
      "epoch": 3.7771041872268434,
      "grad_norm": 6.84846305847168,
      "learning_rate": 6.4042923433874715e-06,
      "loss": 0.3828923797607422,
      "step": 6700
    },
    {
      "epoch": 3.791202594106866,
      "grad_norm": 8.250489234924316,
      "learning_rate": 6.389791183294664e-06,
      "loss": 0.3788992691040039,
      "step": 6725
    },
    {
      "epoch": 3.8053010009868884,
      "grad_norm": 8.07790470123291,
      "learning_rate": 6.375290023201857e-06,
      "loss": 0.3897978210449219,
      "step": 6750
    },
    {
      "epoch": 3.819399407866911,
      "grad_norm": 8.17908000946045,
      "learning_rate": 6.36078886310905e-06,
      "loss": 0.3732691955566406,
      "step": 6775
    },
    {
      "epoch": 3.8334978147469334,
      "grad_norm": 10.077781677246094,
      "learning_rate": 6.3462877030162415e-06,
      "loss": 0.38536983489990234,
      "step": 6800
    },
    {
      "epoch": 3.847596221626956,
      "grad_norm": 7.02805757522583,
      "learning_rate": 6.331786542923435e-06,
      "loss": 0.39071887969970703,
      "step": 6825
    },
    {
      "epoch": 3.861694628506979,
      "grad_norm": 7.91436243057251,
      "learning_rate": 6.317285382830627e-06,
      "loss": 0.3588584899902344,
      "step": 6850
    },
    {
      "epoch": 3.875793035387001,
      "grad_norm": 8.013671875,
      "learning_rate": 6.30278422273782e-06,
      "loss": 0.37952312469482424,
      "step": 6875
    },
    {
      "epoch": 3.889891442267024,
      "grad_norm": 8.154217720031738,
      "learning_rate": 6.288283062645012e-06,
      "loss": 0.37390522003173826,
      "step": 6900
    },
    {
      "epoch": 3.903989849147046,
      "grad_norm": 8.916945457458496,
      "learning_rate": 6.273781902552205e-06,
      "loss": 0.38926349639892577,
      "step": 6925
    },
    {
      "epoch": 3.918088256027069,
      "grad_norm": 7.6121296882629395,
      "learning_rate": 6.259280742459397e-06,
      "loss": 0.3879304504394531,
      "step": 6950
    },
    {
      "epoch": 3.9321866629070916,
      "grad_norm": 9.87363052368164,
      "learning_rate": 6.2447795823665905e-06,
      "loss": 0.3799150466918945,
      "step": 6975
    },
    {
      "epoch": 3.946285069787114,
      "grad_norm": 6.678126335144043,
      "learning_rate": 6.230278422273782e-06,
      "loss": 0.3740867614746094,
      "step": 7000
    },
    {
      "epoch": 3.946285069787114,
      "eval_loss": 0.270298033952713,
      "eval_runtime": 1521.4464,
      "eval_samples_per_second": 2.037,
      "eval_steps_per_second": 0.255,
      "eval_wer": 0.21098909882994774,
      "step": 7000
    }
  ],
  "logging_steps": 25,
  "max_steps": 17740,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1426794605084672e+20,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}