{
  "best_metric": 0.23469093535410654,
  "best_model_checkpoint": "./checkpoints/w2v-multilingual-v1.3/checkpoint-2000",
  "epoch": 0.3295254833040422,
  "eval_steps": 2000,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001647627416520211,
      "grad_norm": 0.2767059803009033,
      "learning_rate": 4.9431537320810675e-08,
      "loss": 0.1923,
      "step": 20
    },
    {
      "epoch": 0.003295254833040422,
      "grad_norm": 2.006911039352417,
      "learning_rate": 1.977261492832427e-07,
      "loss": 0.2947,
      "step": 40
    },
    {
      "epoch": 0.004942882249560633,
      "grad_norm": 0.3714193105697632,
      "learning_rate": 3.624979403526116e-07,
      "loss": 0.2955,
      "step": 60
    },
    {
      "epoch": 0.006590509666080844,
      "grad_norm": 0.3106062412261963,
      "learning_rate": 5.272697314219806e-07,
      "loss": 0.2754,
      "step": 80
    },
    {
      "epoch": 0.008238137082601054,
      "grad_norm": 3.2078254222869873,
      "learning_rate": 6.838029329378811e-07,
      "loss": 0.3177,
      "step": 100
    },
    {
      "epoch": 0.009885764499121265,
      "grad_norm": 0.38503074645996094,
      "learning_rate": 8.485747240072501e-07,
      "loss": 0.1905,
      "step": 120
    },
    {
      "epoch": 0.011533391915641476,
      "grad_norm": 0.15724419057369232,
      "learning_rate": 1.013346515076619e-06,
      "loss": 0.2942,
      "step": 140
    },
    {
      "epoch": 0.013181019332161687,
      "grad_norm": 0.5840986967086792,
      "learning_rate": 1.1781183061459877e-06,
      "loss": 0.2907,
      "step": 160
    },
    {
      "epoch": 0.014828646748681899,
      "grad_norm": 0.23122897744178772,
      "learning_rate": 1.3346515076618883e-06,
      "loss": 0.2811,
      "step": 180
    },
    {
      "epoch": 0.016476274165202108,
      "grad_norm": 2.7248637676239014,
      "learning_rate": 1.4994232987312573e-06,
      "loss": 0.3168,
      "step": 200
    },
    {
      "epoch": 0.01812390158172232,
      "grad_norm": 1.0846350193023682,
      "learning_rate": 1.6641950898006263e-06,
      "loss": 0.1979,
      "step": 220
    },
    {
      "epoch": 0.01977152899824253,
      "grad_norm": 1.7905633449554443,
      "learning_rate": 1.828966880869995e-06,
      "loss": 0.2656,
      "step": 240
    },
    {
      "epoch": 0.021419156414762743,
      "grad_norm": 0.31305554509162903,
      "learning_rate": 1.993738671939364e-06,
      "loss": 0.272,
      "step": 260
    },
    {
      "epoch": 0.023066783831282953,
      "grad_norm": 0.31258106231689453,
      "learning_rate": 2.158510463008733e-06,
      "loss": 0.2693,
      "step": 280
    },
    {
      "epoch": 0.024714411247803162,
      "grad_norm": 1.783349871635437,
      "learning_rate": 2.323282254078102e-06,
      "loss": 0.3116,
      "step": 300
    },
    {
      "epoch": 0.026362038664323375,
      "grad_norm": 0.6936488747596741,
      "learning_rate": 2.488054045147471e-06,
      "loss": 0.1931,
      "step": 320
    },
    {
      "epoch": 0.028009666080843584,
      "grad_norm": 0.6185577511787415,
      "learning_rate": 2.65282583621684e-06,
      "loss": 0.258,
      "step": 340
    },
    {
      "epoch": 0.029657293497363797,
      "grad_norm": 0.3592207729816437,
      "learning_rate": 2.8175976272862085e-06,
      "loss": 0.2401,
      "step": 360
    },
    {
      "epoch": 0.03130492091388401,
      "grad_norm": 1.2324920892715454,
      "learning_rate": 2.982369418355578e-06,
      "loss": 0.2371,
      "step": 380
    },
    {
      "epoch": 0.032952548330404216,
      "grad_norm": 1.5560214519500732,
      "learning_rate": 3.147141209424947e-06,
      "loss": 0.2414,
      "step": 400
    },
    {
      "epoch": 0.03460017574692443,
      "grad_norm": 0.8133947253227234,
      "learning_rate": 3.3119130004943154e-06,
      "loss": 0.1744,
      "step": 420
    },
    {
      "epoch": 0.03624780316344464,
      "grad_norm": 0.8599107265472412,
      "learning_rate": 3.4766847915636844e-06,
      "loss": 0.2536,
      "step": 440
    },
    {
      "epoch": 0.03789543057996485,
      "grad_norm": 0.3699595034122467,
      "learning_rate": 3.641456582633054e-06,
      "loss": 0.2552,
      "step": 460
    },
    {
      "epoch": 0.03954305799648506,
      "grad_norm": 1.2955116033554077,
      "learning_rate": 3.797989784148954e-06,
      "loss": 0.2532,
      "step": 480
    },
    {
      "epoch": 0.041190685413005274,
      "grad_norm": 11.518170356750488,
      "learning_rate": 3.962761575218322e-06,
      "loss": 0.2507,
      "step": 500
    },
    {
      "epoch": 0.042838312829525486,
      "grad_norm": 0.9779248833656311,
      "learning_rate": 4.127533366287692e-06,
      "loss": 0.1733,
      "step": 520
    },
    {
      "epoch": 0.04448594024604569,
      "grad_norm": 0.5386682152748108,
      "learning_rate": 4.29230515735706e-06,
      "loss": 0.2415,
      "step": 540
    },
    {
      "epoch": 0.046133567662565905,
      "grad_norm": 2.7366793155670166,
      "learning_rate": 4.457076948426429e-06,
      "loss": 0.2409,
      "step": 560
    },
    {
      "epoch": 0.04778119507908612,
      "grad_norm": 1.9912066459655762,
      "learning_rate": 4.621848739495799e-06,
      "loss": 0.2134,
      "step": 580
    },
    {
      "epoch": 0.049428822495606324,
      "grad_norm": 4.058573246002197,
      "learning_rate": 4.786620530565167e-06,
      "loss": 0.2456,
      "step": 600
    },
    {
      "epoch": 0.05107644991212654,
      "grad_norm": 0.37456750869750977,
      "learning_rate": 4.951392321634536e-06,
      "loss": 0.1589,
      "step": 620
    },
    {
      "epoch": 0.05272407732864675,
      "grad_norm": 0.7950440645217896,
      "learning_rate": 5.116164112703905e-06,
      "loss": 0.2485,
      "step": 640
    },
    {
      "epoch": 0.05437170474516696,
      "grad_norm": 1.884665846824646,
      "learning_rate": 5.280935903773274e-06,
      "loss": 0.2197,
      "step": 660
    },
    {
      "epoch": 0.05601933216168717,
      "grad_norm": 0.32170844078063965,
      "learning_rate": 5.445707694842643e-06,
      "loss": 0.2111,
      "step": 680
    },
    {
      "epoch": 0.05766695957820738,
      "grad_norm": 1.8230172395706177,
      "learning_rate": 5.610479485912012e-06,
      "loss": 0.2379,
      "step": 700
    },
    {
      "epoch": 0.059314586994727594,
      "grad_norm": 1.2472524642944336,
      "learning_rate": 5.775251276981381e-06,
      "loss": 0.1684,
      "step": 720
    },
    {
      "epoch": 0.0609622144112478,
      "grad_norm": 0.29022061824798584,
      "learning_rate": 5.94002306805075e-06,
      "loss": 0.2201,
      "step": 740
    },
    {
      "epoch": 0.06260984182776802,
      "grad_norm": 0.49721184372901917,
      "learning_rate": 6.104794859120118e-06,
      "loss": 0.2297,
      "step": 760
    },
    {
      "epoch": 0.06425746924428823,
      "grad_norm": 0.6133716106414795,
      "learning_rate": 6.269566650189487e-06,
      "loss": 0.3173,
      "step": 780
    },
    {
      "epoch": 0.06590509666080843,
      "grad_norm": 0.9667792320251465,
      "learning_rate": 6.434338441258857e-06,
      "loss": 0.2349,
      "step": 800
    },
    {
      "epoch": 0.06755272407732865,
      "grad_norm": 0.3177216053009033,
      "learning_rate": 6.599110232328226e-06,
      "loss": 0.1635,
      "step": 820
    },
    {
      "epoch": 0.06920035149384886,
      "grad_norm": 0.8457621335983276,
      "learning_rate": 6.763882023397594e-06,
      "loss": 0.2777,
      "step": 840
    },
    {
      "epoch": 0.07084797891036906,
      "grad_norm": 0.3330087661743164,
      "learning_rate": 6.928653814466963e-06,
      "loss": 0.2154,
      "step": 860
    },
    {
      "epoch": 0.07249560632688928,
      "grad_norm": 0.5845814943313599,
      "learning_rate": 7.093425605536333e-06,
      "loss": 0.2164,
      "step": 880
    },
    {
      "epoch": 0.07414323374340949,
      "grad_norm": 2.325303554534912,
      "learning_rate": 7.258197396605701e-06,
      "loss": 0.2068,
      "step": 900
    },
    {
      "epoch": 0.0757908611599297,
      "grad_norm": 0.21893823146820068,
      "learning_rate": 7.414730598121602e-06,
      "loss": 0.1621,
      "step": 920
    },
    {
      "epoch": 0.07743848857644992,
      "grad_norm": 0.5854327082633972,
      "learning_rate": 7.5795023891909705e-06,
      "loss": 0.2281,
      "step": 940
    },
    {
      "epoch": 0.07908611599297012,
      "grad_norm": 0.2406030148267746,
      "learning_rate": 7.74427418026034e-06,
      "loss": 0.2342,
      "step": 960
    },
    {
      "epoch": 0.08073374340949033,
      "grad_norm": 1.3764126300811768,
      "learning_rate": 7.90904597132971e-06,
      "loss": 0.2018,
      "step": 980
    },
    {
      "epoch": 0.08238137082601055,
      "grad_norm": 0.9587862491607666,
      "learning_rate": 8.073817762399077e-06,
      "loss": 0.2489,
      "step": 1000
    },
    {
      "epoch": 0.08402899824253075,
      "grad_norm": 0.9726558923721313,
      "learning_rate": 8.238589553468447e-06,
      "loss": 0.154,
      "step": 1020
    },
    {
      "epoch": 0.08567662565905097,
      "grad_norm": 1.9828400611877441,
      "learning_rate": 8.403361344537817e-06,
      "loss": 0.2336,
      "step": 1040
    },
    {
      "epoch": 0.08732425307557118,
      "grad_norm": 0.14761961996555328,
      "learning_rate": 8.568133135607183e-06,
      "loss": 0.1949,
      "step": 1060
    },
    {
      "epoch": 0.08897188049209138,
      "grad_norm": 0.39077144861221313,
      "learning_rate": 8.732904926676553e-06,
      "loss": 0.2067,
      "step": 1080
    },
    {
      "epoch": 0.0906195079086116,
      "grad_norm": 2.3257837295532227,
      "learning_rate": 8.897676717745921e-06,
      "loss": 0.2184,
      "step": 1100
    },
    {
      "epoch": 0.09226713532513181,
      "grad_norm": 1.0966060161590576,
      "learning_rate": 9.062448508815291e-06,
      "loss": 0.1655,
      "step": 1120
    },
    {
      "epoch": 0.09391476274165202,
      "grad_norm": 0.5648412704467773,
      "learning_rate": 9.227220299884661e-06,
      "loss": 0.2097,
      "step": 1140
    },
    {
      "epoch": 0.09556239015817224,
      "grad_norm": 0.49816444516181946,
      "learning_rate": 9.39199209095403e-06,
      "loss": 0.2052,
      "step": 1160
    },
    {
      "epoch": 0.09721001757469244,
      "grad_norm": 1.1164054870605469,
      "learning_rate": 9.556763882023397e-06,
      "loss": 0.2298,
      "step": 1180
    },
    {
      "epoch": 0.09885764499121265,
      "grad_norm": 0.9453270435333252,
      "learning_rate": 9.721535673092767e-06,
      "loss": 0.2157,
      "step": 1200
    },
    {
      "epoch": 0.10050527240773287,
      "grad_norm": 0.4588276743888855,
      "learning_rate": 9.886307464162135e-06,
      "loss": 0.1433,
      "step": 1220
    },
    {
      "epoch": 0.10215289982425307,
      "grad_norm": 0.47535696625709534,
      "learning_rate": 1.0051079255231505e-05,
      "loss": 0.2115,
      "step": 1240
    },
    {
      "epoch": 0.10380052724077328,
      "grad_norm": 0.585959792137146,
      "learning_rate": 1.0215851046300875e-05,
      "loss": 0.2245,
      "step": 1260
    },
    {
      "epoch": 0.1054481546572935,
      "grad_norm": 0.24861204624176025,
      "learning_rate": 1.0380622837370241e-05,
      "loss": 0.1981,
      "step": 1280
    },
    {
      "epoch": 0.1070957820738137,
      "grad_norm": 31.602128982543945,
      "learning_rate": 1.0545394628439611e-05,
      "loss": 0.2265,
      "step": 1300
    },
    {
      "epoch": 0.10874340949033393,
      "grad_norm": 0.6234269142150879,
      "learning_rate": 1.0710166419508981e-05,
      "loss": 0.154,
      "step": 1320
    },
    {
      "epoch": 0.11039103690685413,
      "grad_norm": 1.2423540353775024,
      "learning_rate": 1.087493821057835e-05,
      "loss": 0.2062,
      "step": 1340
    },
    {
      "epoch": 0.11203866432337434,
      "grad_norm": 0.2090279757976532,
      "learning_rate": 1.1039710001647719e-05,
      "loss": 0.2214,
      "step": 1360
    },
    {
      "epoch": 0.11368629173989456,
      "grad_norm": 0.6145613193511963,
      "learning_rate": 1.1204481792717087e-05,
      "loss": 0.1942,
      "step": 1380
    },
    {
      "epoch": 0.11533391915641476,
      "grad_norm": 0.9004138708114624,
      "learning_rate": 1.1369253583786455e-05,
      "loss": 0.2271,
      "step": 1400
    },
    {
      "epoch": 0.11698154657293497,
      "grad_norm": 1.609165906906128,
      "learning_rate": 1.1534025374855825e-05,
      "loss": 0.1606,
      "step": 1420
    },
    {
      "epoch": 0.11862917398945519,
      "grad_norm": 0.8725568652153015,
      "learning_rate": 1.1698797165925195e-05,
      "loss": 0.204,
      "step": 1440
    },
    {
      "epoch": 0.1202768014059754,
      "grad_norm": 1.8169455528259277,
      "learning_rate": 1.1863568956994563e-05,
      "loss": 0.207,
      "step": 1460
    },
    {
      "epoch": 0.1219244288224956,
      "grad_norm": 0.37334388494491577,
      "learning_rate": 1.2028340748063933e-05,
      "loss": 0.2033,
      "step": 1480
    },
    {
      "epoch": 0.12357205623901582,
      "grad_norm": 15.727474212646484,
      "learning_rate": 1.2193112539133301e-05,
      "loss": 0.3298,
      "step": 1500
    },
    {
      "epoch": 0.12521968365553604,
      "grad_norm": 4.70903205871582,
      "learning_rate": 1.2357884330202669e-05,
      "loss": 0.159,
      "step": 1520
    },
    {
      "epoch": 0.12686731107205623,
      "grad_norm": 0.4843326807022095,
      "learning_rate": 1.2522656121272039e-05,
      "loss": 0.2141,
      "step": 1540
    },
    {
      "epoch": 0.12851493848857645,
      "grad_norm": 0.572084367275238,
      "learning_rate": 1.2687427912341407e-05,
      "loss": 0.2182,
      "step": 1560
    },
    {
      "epoch": 0.13016256590509667,
      "grad_norm": 0.31078797578811646,
      "learning_rate": 1.2852199703410777e-05,
      "loss": 0.1994,
      "step": 1580
    },
    {
      "epoch": 0.13181019332161686,
      "grad_norm": 9.014205932617188,
      "learning_rate": 1.3016971494480145e-05,
      "loss": 0.2089,
      "step": 1600
    },
    {
      "epoch": 0.13345782073813708,
      "grad_norm": 1.0150245428085327,
      "learning_rate": 1.3181743285549513e-05,
      "loss": 0.1521,
      "step": 1620
    },
    {
      "epoch": 0.1351054481546573,
      "grad_norm": 0.25271451473236084,
      "learning_rate": 1.3346515076618885e-05,
      "loss": 0.1996,
      "step": 1640
    },
    {
      "epoch": 0.1367530755711775,
      "grad_norm": 0.47118502855300903,
      "learning_rate": 1.3511286867688253e-05,
      "loss": 0.2059,
      "step": 1660
    },
    {
      "epoch": 0.13840070298769772,
      "grad_norm": 0.5134350657463074,
      "learning_rate": 1.3676058658757621e-05,
      "loss": 0.1935,
      "step": 1680
    },
    {
      "epoch": 0.14004833040421794,
      "grad_norm": 1.0354816913604736,
      "learning_rate": 1.384083044982699e-05,
      "loss": 0.2103,
      "step": 1700
    },
    {
      "epoch": 0.14169595782073813,
      "grad_norm": 0.5588876605033875,
      "learning_rate": 1.4005602240896359e-05,
      "loss": 0.1598,
      "step": 1720
    },
    {
      "epoch": 0.14334358523725835,
      "grad_norm": 0.7309175133705139,
      "learning_rate": 1.4170374031965727e-05,
      "loss": 0.2204,
      "step": 1740
    },
    {
      "epoch": 0.14499121265377857,
      "grad_norm": 0.6155902743339539,
      "learning_rate": 1.4335145823035099e-05,
      "loss": 0.2133,
      "step": 1760
    },
    {
      "epoch": 0.14663884007029876,
      "grad_norm": 0.7660940885543823,
      "learning_rate": 1.4499917614104467e-05,
      "loss": 0.2065,
      "step": 1780
    },
    {
      "epoch": 0.14828646748681898,
      "grad_norm": 1.1954026222229004,
      "learning_rate": 1.4664689405173835e-05,
      "loss": 0.2147,
      "step": 1800
    },
    {
      "epoch": 0.1499340949033392,
      "grad_norm": 0.4249323606491089,
      "learning_rate": 1.4829461196243205e-05,
      "loss": 0.1553,
      "step": 1820
    },
    {
      "epoch": 0.1515817223198594,
      "grad_norm": 2.9014129638671875,
      "learning_rate": 1.4994232987312573e-05,
      "loss": 0.2208,
      "step": 1840
    },
    {
      "epoch": 0.1532293497363796,
      "grad_norm": 1.6474498510360718,
      "learning_rate": 1.5159004778381941e-05,
      "loss": 0.209,
      "step": 1860
    },
    {
      "epoch": 0.15487697715289983,
      "grad_norm": 0.1585623174905777,
      "learning_rate": 1.532377656945131e-05,
      "loss": 0.1873,
      "step": 1880
    },
    {
      "epoch": 0.15652460456942002,
      "grad_norm": 1.171941876411438,
      "learning_rate": 1.548854836052068e-05,
      "loss": 0.2389,
      "step": 1900
    },
    {
      "epoch": 0.15817223198594024,
      "grad_norm": 0.48890382051467896,
      "learning_rate": 1.5653320151590047e-05,
      "loss": 0.1679,
      "step": 1920
    },
    {
      "epoch": 0.15981985940246046,
      "grad_norm": 0.5568016767501831,
      "learning_rate": 1.581809194265942e-05,
      "loss": 0.1968,
      "step": 1940
    },
    {
      "epoch": 0.16146748681898065,
      "grad_norm": 0.9775394797325134,
      "learning_rate": 1.5982863733728787e-05,
      "loss": 0.2208,
      "step": 1960
    },
    {
      "epoch": 0.16311511423550087,
      "grad_norm": 0.60302734375,
      "learning_rate": 1.6147635524798155e-05,
      "loss": 0.1929,
      "step": 1980
    },
    {
      "epoch": 0.1647627416520211,
      "grad_norm": 1.7513552904129028,
      "learning_rate": 1.6312407315867526e-05,
      "loss": 0.2055,
      "step": 2000
    },
    {
      "epoch": 0.1647627416520211,
      "eval_loss": 0.5699400305747986,
      "eval_runtime": 686.8117,
      "eval_samples_per_second": 31.725,
      "eval_steps_per_second": 7.932,
      "eval_wer": 0.23469093535410654,
      "step": 2000
    },
    {
      "epoch": 0.16641036906854131,
      "grad_norm": 1.049734354019165,
      "learning_rate": 1.6477179106936894e-05,
      "loss": 0.1608,
      "step": 2020
    },
    {
      "epoch": 0.1680579964850615,
      "grad_norm": 1.7113618850708008,
      "learning_rate": 1.6641950898006263e-05,
      "loss": 0.2008,
      "step": 2040
    },
    {
      "epoch": 0.16970562390158173,
      "grad_norm": 0.3202134370803833,
      "learning_rate": 1.6806722689075634e-05,
      "loss": 0.1973,
      "step": 2060
    },
    {
      "epoch": 0.17135325131810195,
      "grad_norm": 0.3612610697746277,
      "learning_rate": 1.6971494480145e-05,
      "loss": 0.1732,
      "step": 2080
    },
    {
      "epoch": 0.17300087873462214,
      "grad_norm": 1.8115849494934082,
      "learning_rate": 1.7136266271214367e-05,
      "loss": 0.2138,
      "step": 2100
    },
    {
      "epoch": 0.17464850615114236,
      "grad_norm": 0.7046949863433838,
      "learning_rate": 1.7301038062283735e-05,
      "loss": 0.1571,
      "step": 2120
    },
    {
      "epoch": 0.17629613356766258,
      "grad_norm": 0.5983096957206726,
      "learning_rate": 1.7465809853353107e-05,
      "loss": 0.2092,
      "step": 2140
    },
    {
      "epoch": 0.17794376098418277,
      "grad_norm": 0.17064958810806274,
      "learning_rate": 1.7630581644422475e-05,
      "loss": 0.2083,
      "step": 2160
    },
    {
      "epoch": 0.179591388400703,
      "grad_norm": 1.422013759613037,
      "learning_rate": 1.7795353435491843e-05,
      "loss": 0.2087,
      "step": 2180
    },
    {
      "epoch": 0.1812390158172232,
      "grad_norm": 0.982097864151001,
      "learning_rate": 1.7960125226561214e-05,
      "loss": 0.2161,
      "step": 2200
    },
    {
      "epoch": 0.1828866432337434,
      "grad_norm": 0.2690947949886322,
      "learning_rate": 1.8124897017630583e-05,
      "loss": 0.1693,
      "step": 2220
    },
    {
      "epoch": 0.18453427065026362,
      "grad_norm": 0.24867244064807892,
      "learning_rate": 1.828966880869995e-05,
      "loss": 0.2058,
      "step": 2240
    },
    {
      "epoch": 0.18618189806678384,
      "grad_norm": 0.9435555934906006,
      "learning_rate": 1.8454440599769322e-05,
      "loss": 0.1991,
      "step": 2260
    },
    {
      "epoch": 0.18782952548330403,
      "grad_norm": 0.23964335024356842,
      "learning_rate": 1.861921239083869e-05,
      "loss": 0.1932,
      "step": 2280
    },
    {
      "epoch": 0.18947715289982425,
      "grad_norm": 2.1007418632507324,
      "learning_rate": 1.878398418190806e-05,
      "loss": 0.2075,
      "step": 2300
    },
    {
      "epoch": 0.19112478031634447,
      "grad_norm": 0.31368857622146606,
      "learning_rate": 1.8948755972977427e-05,
      "loss": 0.1557,
      "step": 2320
    },
    {
      "epoch": 0.19277240773286466,
      "grad_norm": 0.4029647409915924,
      "learning_rate": 1.9113527764046795e-05,
      "loss": 0.1952,
      "step": 2340
    },
    {
      "epoch": 0.19442003514938488,
      "grad_norm": 0.38545289635658264,
      "learning_rate": 1.9278299555116163e-05,
      "loss": 0.1998,
      "step": 2360
    },
    {
      "epoch": 0.1960676625659051,
      "grad_norm": 0.44573166966438293,
      "learning_rate": 1.9443071346185534e-05,
      "loss": 0.2078,
      "step": 2380
    },
    {
      "epoch": 0.1977152899824253,
      "grad_norm": 9.132265090942383,
      "learning_rate": 1.9607843137254903e-05,
      "loss": 0.2175,
      "step": 2400
    },
    {
      "epoch": 0.19936291739894552,
      "grad_norm": 0.42929643392562866,
      "learning_rate": 1.977261492832427e-05,
      "loss": 0.1762,
      "step": 2420
    },
    {
      "epoch": 0.20101054481546574,
      "grad_norm": 0.6267173886299133,
      "learning_rate": 1.9937386719393642e-05,
      "loss": 0.2101,
      "step": 2440
    },
    {
      "epoch": 0.20265817223198593,
      "grad_norm": 9.70997142791748,
      "learning_rate": 2.010215851046301e-05,
      "loss": 0.2101,
      "step": 2460
    },
    {
      "epoch": 0.20430579964850615,
      "grad_norm": 0.47748956084251404,
      "learning_rate": 2.026693030153238e-05,
      "loss": 0.2039,
      "step": 2480
    },
    {
      "epoch": 0.20595342706502637,
      "grad_norm": 1.3222582340240479,
      "learning_rate": 2.043170209260175e-05,
      "loss": 0.213,
      "step": 2500
    },
    {
      "epoch": 0.20760105448154656,
      "grad_norm": 0.4152863919734955,
      "learning_rate": 2.0596473883671115e-05,
      "loss": 0.148,
      "step": 2520
    },
    {
      "epoch": 0.20924868189806678,
      "grad_norm": 0.7384160757064819,
      "learning_rate": 2.0761245674740483e-05,
      "loss": 0.2138,
      "step": 2540
    },
    {
      "epoch": 0.210896309314587,
      "grad_norm": 0.27651092410087585,
      "learning_rate": 2.0926017465809854e-05,
      "loss": 0.2046,
      "step": 2560
    },
    {
      "epoch": 0.21254393673110722,
      "grad_norm": 0.226897194981575,
      "learning_rate": 2.1090789256879222e-05,
      "loss": 0.1904,
      "step": 2580
    },
    {
      "epoch": 0.2141915641476274,
      "grad_norm": 1.2391464710235596,
      "learning_rate": 2.125556104794859e-05,
      "loss": 0.204,
      "step": 2600
    },
    {
      "epoch": 0.21583919156414763,
      "grad_norm": 1.6048617362976074,
      "learning_rate": 2.1420332839017962e-05,
      "loss": 0.1548,
      "step": 2620
    },
    {
      "epoch": 0.21748681898066785,
      "grad_norm": 0.28409889340400696,
      "learning_rate": 2.158510463008733e-05,
      "loss": 0.201,
      "step": 2640
    },
    {
      "epoch": 0.21913444639718804,
      "grad_norm": 0.354885995388031,
      "learning_rate": 2.17498764211567e-05,
      "loss": 0.2083,
      "step": 2660
    },
    {
      "epoch": 0.22078207381370826,
      "grad_norm": 0.2778099775314331,
      "learning_rate": 2.191464821222607e-05,
      "loss": 0.1891,
      "step": 2680
    },
    {
      "epoch": 0.22242970123022848,
      "grad_norm": 1.007686734199524,
      "learning_rate": 2.2079420003295438e-05,
      "loss": 0.2152,
      "step": 2700
    },
    {
      "epoch": 0.22407732864674867,
      "grad_norm": 0.9725649952888489,
      "learning_rate": 2.2244191794364806e-05,
      "loss": 0.1581,
      "step": 2720
    },
    {
      "epoch": 0.2257249560632689,
      "grad_norm": 0.2451123297214508,
      "learning_rate": 2.2408963585434174e-05,
      "loss": 0.2019,
      "step": 2740
    },
    {
      "epoch": 0.22737258347978911,
      "grad_norm": 0.3667006194591522,
      "learning_rate": 2.2573735376503542e-05,
      "loss": 0.2083,
      "step": 2760
    },
    {
      "epoch": 0.2290202108963093,
      "grad_norm": 3.1283884048461914,
      "learning_rate": 2.273850716757291e-05,
      "loss": 0.194,
      "step": 2780
    },
    {
      "epoch": 0.23066783831282953,
      "grad_norm": 0.7148507237434387,
      "learning_rate": 2.2903278958642282e-05,
      "loss": 0.1931,
      "step": 2800
    },
    {
      "epoch": 0.23231546572934975,
      "grad_norm": 0.5805519223213196,
      "learning_rate": 2.306805074971165e-05,
      "loss": 0.1552,
      "step": 2820
    },
    {
      "epoch": 0.23396309314586994,
      "grad_norm": 0.8168196082115173,
      "learning_rate": 2.323282254078102e-05,
      "loss": 0.2107,
      "step": 2840
    },
    {
      "epoch": 0.23561072056239016,
      "grad_norm": 0.17171867191791534,
      "learning_rate": 2.339759433185039e-05,
      "loss": 0.2195,
      "step": 2860
    },
    {
      "epoch": 0.23725834797891038,
      "grad_norm": 0.6692082285881042,
      "learning_rate": 2.3562366122919758e-05,
      "loss": 0.1975,
      "step": 2880
    },
    {
      "epoch": 0.23890597539543057,
      "grad_norm": 1.5185160636901855,
      "learning_rate": 2.3727137913989126e-05,
      "loss": 0.2086,
      "step": 2900
    },
    {
      "epoch": 0.2405536028119508,
      "grad_norm": 0.8978987336158752,
      "learning_rate": 2.3891909705058498e-05,
      "loss": 0.1537,
      "step": 2920
    },
    {
      "epoch": 0.242201230228471,
      "grad_norm": 1.1462221145629883,
      "learning_rate": 2.4056681496127866e-05,
      "loss": 0.215,
      "step": 2940
    },
    {
      "epoch": 0.2438488576449912,
      "grad_norm": 0.2455727905035019,
      "learning_rate": 2.422145328719723e-05,
      "loss": 0.2137,
      "step": 2960
    },
    {
      "epoch": 0.24549648506151142,
      "grad_norm": 0.21464231610298157,
      "learning_rate": 2.4386225078266602e-05,
      "loss": 0.203,
      "step": 2980
    },
    {
      "epoch": 0.24714411247803164,
      "grad_norm": 0.8478316068649292,
      "learning_rate": 2.455099686933597e-05,
      "loss": 0.2085,
      "step": 3000
    },
    {
      "epoch": 0.24879173989455183,
      "grad_norm": 0.37225690484046936,
      "learning_rate": 2.4715768660405338e-05,
      "loss": 0.1645,
      "step": 3020
    },
    {
      "epoch": 0.2504393673110721,
      "grad_norm": 1.3999593257904053,
      "learning_rate": 2.488054045147471e-05,
      "loss": 0.2129,
      "step": 3040
    },
    {
      "epoch": 0.2520869947275923,
      "grad_norm": 2.0909199714660645,
      "learning_rate": 2.5045312242544078e-05,
      "loss": 0.1926,
      "step": 3060
    },
    {
      "epoch": 0.25373462214411246,
      "grad_norm": 0.19655053317546844,
      "learning_rate": 2.5210084033613446e-05,
      "loss": 0.2007,
      "step": 3080
    },
    {
      "epoch": 0.2553822495606327,
      "grad_norm": 1.2680870294570923,
      "learning_rate": 2.5374855824682814e-05,
      "loss": 0.2107,
      "step": 3100
    },
    {
      "epoch": 0.2570298769771529,
      "grad_norm": 0.19821316003799438,
      "learning_rate": 2.5539627615752182e-05,
      "loss": 0.1511,
      "step": 3120
    },
    {
      "epoch": 0.2586775043936731,
      "grad_norm": 0.22427937388420105,
      "learning_rate": 2.5704399406821554e-05,
      "loss": 0.2139,
      "step": 3140
    },
    {
      "epoch": 0.26032513181019334,
      "grad_norm": 0.4531656503677368,
      "learning_rate": 2.5869171197890922e-05,
      "loss": 0.1995,
      "step": 3160
    },
    {
      "epoch": 0.26197275922671354,
      "grad_norm": 0.3967747986316681,
      "learning_rate": 2.603394298896029e-05,
      "loss": 0.1981,
      "step": 3180
    },
    {
      "epoch": 0.26362038664323373,
      "grad_norm": 1.0957462787628174,
      "learning_rate": 2.6198714780029658e-05,
      "loss": 0.2092,
      "step": 3200
    },
    {
      "epoch": 0.265268014059754,
      "grad_norm": 0.5567193627357483,
      "learning_rate": 2.6363486571099026e-05,
      "loss": 0.1642,
      "step": 3220
    },
    {
      "epoch": 0.26691564147627417,
      "grad_norm": 0.3523741066455841,
      "learning_rate": 2.6528258362168395e-05,
      "loss": 0.2059,
      "step": 3240
    },
    {
      "epoch": 0.26856326889279436,
      "grad_norm": 0.40257710218429565,
      "learning_rate": 2.669303015323777e-05,
      "loss": 0.195,
      "step": 3260
    },
    {
      "epoch": 0.2702108963093146,
      "grad_norm": 0.3187640309333801,
      "learning_rate": 2.6849563354753665e-05,
      "loss": 0.1827,
      "step": 3280
    },
    {
      "epoch": 0.2718585237258348,
      "grad_norm": 0.7375414967536926,
      "learning_rate": 2.701433514582304e-05,
      "loss": 0.2247,
      "step": 3300
    },
    {
      "epoch": 0.273506151142355,
      "grad_norm": 0.45597076416015625,
      "learning_rate": 2.7179106936892408e-05,
      "loss": 0.1654,
      "step": 3320
    },
    {
      "epoch": 0.27515377855887524,
      "grad_norm": 0.21507132053375244,
      "learning_rate": 2.7343878727961776e-05,
      "loss": 0.1858,
      "step": 3340
    },
    {
      "epoch": 0.27680140597539543,
      "grad_norm": 0.7203060388565063,
      "learning_rate": 2.7508650519031144e-05,
      "loss": 0.1908,
      "step": 3360
    },
    {
      "epoch": 0.2784490333919156,
      "grad_norm": 0.8007901906967163,
      "learning_rate": 2.7673422310100512e-05,
      "loss": 0.1793,
      "step": 3380
    },
    {
      "epoch": 0.28009666080843587,
      "grad_norm": 3.210064649581909,
      "learning_rate": 2.783819410116988e-05,
      "loss": 0.206,
      "step": 3400
    },
    {
      "epoch": 0.28174428822495606,
      "grad_norm": 0.8950255513191223,
      "learning_rate": 2.8002965892239252e-05,
      "loss": 0.1531,
      "step": 3420
    },
    {
      "epoch": 0.28339191564147626,
      "grad_norm": 0.4942973256111145,
      "learning_rate": 2.816773768330862e-05,
      "loss": 0.1916,
      "step": 3440
    },
    {
      "epoch": 0.2850395430579965,
      "grad_norm": 0.31426137685775757,
      "learning_rate": 2.8332509474377988e-05,
      "loss": 0.2013,
      "step": 3460
    },
    {
      "epoch": 0.2866871704745167,
      "grad_norm": 0.47154414653778076,
      "learning_rate": 2.8497281265447356e-05,
      "loss": 0.1831,
      "step": 3480
    },
    {
      "epoch": 0.2883347978910369,
      "grad_norm": 0.8456603288650513,
      "learning_rate": 2.8662053056516724e-05,
      "loss": 0.2043,
      "step": 3500
    },
    {
      "epoch": 0.28998242530755713,
      "grad_norm": 0.29031482338905334,
      "learning_rate": 2.8826824847586092e-05,
      "loss": 0.1607,
      "step": 3520
    },
    {
      "epoch": 0.2916300527240773,
      "grad_norm": 0.3170378804206848,
      "learning_rate": 2.8991596638655467e-05,
      "loss": 0.1916,
      "step": 3540
    },
    {
      "epoch": 0.2932776801405975,
      "grad_norm": 0.3800877332687378,
      "learning_rate": 2.9156368429724836e-05,
      "loss": 0.2051,
      "step": 3560
    },
    {
      "epoch": 0.29492530755711777,
      "grad_norm": 0.5847609639167786,
      "learning_rate": 2.9321140220794204e-05,
      "loss": 0.196,
      "step": 3580
    },
    {
      "epoch": 0.29657293497363796,
      "grad_norm": 1.0933667421340942,
      "learning_rate": 2.9485912011863572e-05,
      "loss": 0.2154,
      "step": 3600
    },
    {
      "epoch": 0.29822056239015815,
      "grad_norm": 4.349573135375977,
      "learning_rate": 2.9650683802932937e-05,
      "loss": 0.1606,
      "step": 3620
    },
    {
      "epoch": 0.2998681898066784,
      "grad_norm": 0.4264489710330963,
      "learning_rate": 2.9815455594002305e-05,
      "loss": 0.2117,
      "step": 3640
    },
    {
      "epoch": 0.3015158172231986,
      "grad_norm": 0.47935691475868225,
      "learning_rate": 2.998022738507168e-05,
      "loss": 0.1901,
      "step": 3660
    },
    {
      "epoch": 0.3031634446397188,
      "grad_norm": 0.7258153557777405,
      "learning_rate": 3.0144999176141048e-05,
      "loss": 0.189,
      "step": 3680
    },
    {
      "epoch": 0.30481107205623903,
      "grad_norm": 2.0093941688537598,
      "learning_rate": 3.0309770967210416e-05,
      "loss": 0.2104,
      "step": 3700
    },
    {
      "epoch": 0.3064586994727592,
      "grad_norm": 1.1718699932098389,
      "learning_rate": 3.0474542758279784e-05,
      "loss": 0.1577,
      "step": 3720
    },
    {
      "epoch": 0.3081063268892794,
      "grad_norm": 0.19388867914676666,
      "learning_rate": 3.063931454934915e-05,
      "loss": 0.2011,
      "step": 3740
    },
    {
      "epoch": 0.30975395430579966,
      "grad_norm": 0.2112320065498352,
      "learning_rate": 3.080408634041852e-05,
      "loss": 0.2077,
      "step": 3760
    },
    {
      "epoch": 0.31140158172231985,
      "grad_norm": 1.9554697275161743,
      "learning_rate": 3.096885813148789e-05,
      "loss": 0.1861,
      "step": 3780
    },
    {
      "epoch": 0.31304920913884005,
      "grad_norm": 0.7065563201904297,
      "learning_rate": 3.113362992255726e-05,
      "loss": 0.2071,
      "step": 3800
    },
    {
      "epoch": 0.3146968365553603,
      "grad_norm": 0.4599238634109497,
      "learning_rate": 3.129840171362663e-05,
      "loss": 0.1655,
      "step": 3820
    },
    {
      "epoch": 0.3163444639718805,
      "grad_norm": 0.5441445708274841,
      "learning_rate": 3.1463173504695996e-05,
      "loss": 0.1971,
      "step": 3840
    },
    {
      "epoch": 0.3179920913884007,
      "grad_norm": 1.1993862390518188,
      "learning_rate": 3.1627945295765364e-05,
      "loss": 0.2048,
      "step": 3860
    },
    {
      "epoch": 0.3196397188049209,
      "grad_norm": 0.3095191717147827,
      "learning_rate": 3.179271708683473e-05,
      "loss": 0.2009,
      "step": 3880
    },
    {
      "epoch": 0.3212873462214411,
      "grad_norm": 1.0743999481201172,
      "learning_rate": 3.195748887790411e-05,
      "loss": 0.2371,
      "step": 3900
    },
    {
      "epoch": 0.3229349736379613,
      "grad_norm": 0.5000220537185669,
      "learning_rate": 3.2122260668973475e-05,
      "loss": 0.1722,
      "step": 3920
    },
    {
      "epoch": 0.32458260105448156,
      "grad_norm": 1.1417018175125122,
      "learning_rate": 3.2287032460042844e-05,
      "loss": 0.2079,
      "step": 3940
    },
    {
      "epoch": 0.32623022847100175,
      "grad_norm": 1.099433422088623,
      "learning_rate": 3.245180425111221e-05,
      "loss": 0.2009,
      "step": 3960
    },
    {
      "epoch": 0.32787785588752194,
      "grad_norm": 0.3827146589756012,
      "learning_rate": 3.261657604218158e-05,
      "loss": 0.2171,
      "step": 3980
    },
    {
      "epoch": 0.3295254833040422,
      "grad_norm": 1.1845418214797974,
      "learning_rate": 3.278134783325095e-05,
      "loss": 0.2195,
      "step": 4000
    },
    {
      "epoch": 0.3295254833040422,
      "eval_loss": 0.5868579149246216,
      "eval_runtime": 260.92,
      "eval_samples_per_second": 83.508,
      "eval_steps_per_second": 20.88,
      "eval_wer": 0.2350240606008012,
      "step": 4000
    }
  ],
  "logging_steps": 20,
  "max_steps": 60690,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.157571471334336e+20,
  "train_batch_size": 24,
  "trial_name": null,
  "trial_params": null
}