diff --git "a/checkpoint-24000/trainer_state.json" "b/checkpoint-24000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-24000/trainer_state.json"
@@ -0,0 +1,8541 @@
+{
+  "best_metric": 0.22245208175285466,
+  "best_model_checkpoint": "./checkpoints/w2v-multilingual-v1.3/checkpoint-24000",
+  "epoch": 1.977207820738137,
+  "eval_steps": 2000,
+  "global_step": 24000,
+  "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
+    },
+    {
+      "epoch": 0.3311731107205624,
+      "grad_norm": 1.6073412895202637,
+      "learning_rate": 3.2946119624320316e-05,
+      "loss": 0.1695,
+      "step": 4020
+    },
+    {
+      "epoch": 0.33282073813708263,
+      "grad_norm": 2.9597036838531494,
+      "learning_rate": 3.311089141538969e-05,
+      "loss": 0.2484,
+      "step": 4040
+    },
+    {
+      "epoch": 0.3344683655536028,
+      "grad_norm": 1.4150543212890625,
+      "learning_rate": 3.327566320645906e-05,
+      "loss": 0.2087,
+      "step": 4060
+    },
+    {
+      "epoch": 0.336115992970123,
+      "grad_norm": 0.255397230386734,
+      "learning_rate": 3.344043499752843e-05,
+      "loss": 0.2048,
+      "step": 4080
+    },
+    {
+      "epoch": 0.33776362038664326,
+      "grad_norm": 1.9232691526412964,
+      "learning_rate": 3.3605206788597795e-05,
+      "loss": 0.2274,
+      "step": 4100
+    },
+    {
+      "epoch": 0.33941124780316345,
+      "grad_norm": 1.0172194242477417,
+      "learning_rate": 3.3769978579667164e-05,
+      "loss": 0.1779,
+      "step": 4120
+    },
+    {
+      "epoch": 0.34105887521968364,
+      "grad_norm": 1.3862395286560059,
+      "learning_rate": 3.393475037073653e-05,
+      "loss": 0.2095,
+      "step": 4140
+    },
+    {
+      "epoch": 0.3427065026362039,
+      "grad_norm": 0.3353387117385864,
+      "learning_rate": 3.40995221618059e-05,
+      "loss": 0.2021,
+      "step": 4160
+    },
+    {
+      "epoch": 0.3443541300527241,
+      "grad_norm": 0.9549083709716797,
+      "learning_rate": 3.426429395287527e-05,
+      "loss": 0.1997,
+      "step": 4180
+    },
+    {
+      "epoch": 0.3460017574692443,
+      "grad_norm": 1.6077580451965332,
+      "learning_rate": 3.4429065743944636e-05,
+      "loss": 0.2562,
+      "step": 4200
+    },
+    {
+      "epoch": 0.3476493848857645,
+      "grad_norm": 5.387716770172119,
+      "learning_rate": 3.4593837535014004e-05,
+      "loss": 0.1726,
+      "step": 4220
+    },
+    {
+      "epoch": 0.3492970123022847,
+      "grad_norm": 0.5455642342567444,
+      "learning_rate": 3.475860932608337e-05,
+      "loss": 0.2342,
+      "step": 4240
+    },
+    {
+      "epoch": 0.3509446397188049,
+      "grad_norm": 0.18990729749202728,
+      "learning_rate": 3.492338111715274e-05,
+      "loss": 0.2334,
+      "step": 4260
+    },
+    {
+      "epoch": 0.35259226713532515,
+      "grad_norm": 0.4878564476966858,
+      "learning_rate": 3.5088152908222115e-05,
+      "loss": 0.2065,
+      "step": 4280
+    },
+    {
+      "epoch": 0.35423989455184535,
+      "grad_norm": 1.3400063514709473,
+      "learning_rate": 3.5252924699291483e-05,
+      "loss": 0.237,
+      "step": 4300
+    },
+    {
+      "epoch": 0.35588752196836554,
+      "grad_norm": 0.5822551250457764,
+      "learning_rate": 3.541769649036085e-05,
+      "loss": 0.1787,
+      "step": 4320
+    },
+    {
+      "epoch": 0.3575351493848858,
+      "grad_norm": 0.4629223644733429,
+      "learning_rate": 3.558246828143022e-05,
+      "loss": 0.2129,
+      "step": 4340
+    },
+    {
+      "epoch": 0.359182776801406,
+      "grad_norm": 1.4195072650909424,
+      "learning_rate": 3.574724007249959e-05,
+      "loss": 0.2215,
+      "step": 4360
+    },
+    {
+      "epoch": 0.36083040421792617,
+      "grad_norm": 0.3443647623062134,
+      "learning_rate": 3.5912011863568956e-05,
+      "loss": 0.2002,
+      "step": 4380
+    },
+    {
+      "epoch": 0.3624780316344464,
+      "grad_norm": 3.2932443618774414,
+      "learning_rate": 3.607678365463833e-05,
+      "loss": 0.2172,
+      "step": 4400
+    },
+    {
+      "epoch": 0.3641256590509666,
+      "grad_norm": 0.4463866055011749,
+      "learning_rate": 3.62415554457077e-05,
+      "loss": 0.1889,
+      "step": 4420
+    },
+    {
+      "epoch": 0.3657732864674868,
+      "grad_norm": 0.4873151183128357,
+      "learning_rate": 3.640632723677707e-05,
+      "loss": 0.2139,
+      "step": 4440
+    },
+    {
+      "epoch": 0.36742091388400705,
+      "grad_norm": 1.6354761123657227,
+      "learning_rate": 3.6571099027846435e-05,
+      "loss": 0.1986,
+      "step": 4460
+    },
+    {
+      "epoch": 0.36906854130052724,
+      "grad_norm": 0.5571808815002441,
+      "learning_rate": 3.6735870818915803e-05,
+      "loss": 0.2047,
+      "step": 4480
+    },
+    {
+      "epoch": 0.37071616871704743,
+      "grad_norm": 0.7461993098258972,
+      "learning_rate": 3.690064260998517e-05,
+      "loss": 0.2196,
+      "step": 4500
+    },
+    {
+      "epoch": 0.3723637961335677,
+      "grad_norm": 0.2061534970998764,
+      "learning_rate": 3.7065414401054546e-05,
+      "loss": 0.1657,
+      "step": 4520
+    },
+    {
+      "epoch": 0.3740114235500879,
+      "grad_norm": 1.0977954864501953,
+      "learning_rate": 3.7230186192123915e-05,
+      "loss": 0.2062,
+      "step": 4540
+    },
+    {
+      "epoch": 0.37565905096660807,
+      "grad_norm": 2.38232684135437,
+      "learning_rate": 3.739495798319328e-05,
+      "loss": 0.2319,
+      "step": 4560
+    },
+    {
+      "epoch": 0.3773066783831283,
+      "grad_norm": 0.6182531118392944,
+      "learning_rate": 3.755972977426265e-05,
+      "loss": 0.202,
+      "step": 4580
+    },
+    {
+      "epoch": 0.3789543057996485,
+      "grad_norm": 28.744009017944336,
+      "learning_rate": 3.772450156533201e-05,
+      "loss": 0.2235,
+      "step": 4600
+    },
+    {
+      "epoch": 0.3806019332161687,
+      "grad_norm": 0.5899057984352112,
+      "learning_rate": 3.788927335640138e-05,
+      "loss": 0.1577,
+      "step": 4620
+    },
+    {
+      "epoch": 0.38224956063268895,
+      "grad_norm": 0.5423290133476257,
+      "learning_rate": 3.8054045147470755e-05,
+      "loss": 0.2098,
+      "step": 4640
+    },
+    {
+      "epoch": 0.38389718804920914,
+      "grad_norm": 0.23139849305152893,
+      "learning_rate": 3.8218816938540123e-05,
+      "loss": 0.1989,
+      "step": 4660
+    },
+    {
+      "epoch": 0.38554481546572933,
+      "grad_norm": 0.3539600670337677,
+      "learning_rate": 3.838358872960949e-05,
+      "loss": 0.1968,
+      "step": 4680
+    },
+    {
+      "epoch": 0.3871924428822496,
+      "grad_norm": 0.7127693295478821,
+      "learning_rate": 3.854836052067886e-05,
+      "loss": 0.2225,
+      "step": 4700
+    },
+    {
+      "epoch": 0.38884007029876977,
+      "grad_norm": 0.3489459753036499,
+      "learning_rate": 3.871313231174823e-05,
+      "loss": 0.1606,
+      "step": 4720
+    },
+    {
+      "epoch": 0.39048769771528996,
+      "grad_norm": 0.27798184752464294,
+      "learning_rate": 3.886966551326413e-05,
+      "loss": 0.2096,
+      "step": 4740
+    },
+    {
+      "epoch": 0.3921353251318102,
+      "grad_norm": 0.4229481816291809,
+      "learning_rate": 3.90344373043335e-05,
+      "loss": 0.2195,
+      "step": 4760
+    },
+    {
+      "epoch": 0.3937829525483304,
+      "grad_norm": 0.25523656606674194,
+      "learning_rate": 3.9199209095402866e-05,
+      "loss": 0.2106,
+      "step": 4780
+    },
+    {
+      "epoch": 0.3954305799648506,
+      "grad_norm": 0.9676795601844788,
+      "learning_rate": 3.936398088647224e-05,
+      "loss": 0.2267,
+      "step": 4800
+    },
+    {
+      "epoch": 0.39707820738137084,
+      "grad_norm": 0.40875479578971863,
+      "learning_rate": 3.952875267754161e-05,
+      "loss": 0.155,
+      "step": 4820
+    },
+    {
+      "epoch": 0.39872583479789103,
+      "grad_norm": 0.335033655166626,
+      "learning_rate": 3.969352446861098e-05,
+      "loss": 0.2081,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4003734622144112,
+      "grad_norm": 0.3303823173046112,
+      "learning_rate": 3.9858296259680345e-05,
+      "loss": 0.2282,
+      "step": 4860
+    },
+    {
+      "epoch": 0.40202108963093147,
+      "grad_norm": 0.20934663712978363,
+      "learning_rate": 4.0023068050749714e-05,
+      "loss": 0.1965,
+      "step": 4880
+    },
+    {
+      "epoch": 0.40366871704745166,
+      "grad_norm": 1.4093564748764038,
+      "learning_rate": 4.018783984181908e-05,
+      "loss": 0.22,
+      "step": 4900
+    },
+    {
+      "epoch": 0.40531634446397186,
+      "grad_norm": 0.4276560842990875,
+      "learning_rate": 4.035261163288846e-05,
+      "loss": 0.1705,
+      "step": 4920
+    },
+    {
+      "epoch": 0.4069639718804921,
+      "grad_norm": 0.4243983030319214,
+      "learning_rate": 4.0517383423957825e-05,
+      "loss": 0.2285,
+      "step": 4940
+    },
+    {
+      "epoch": 0.4086115992970123,
+      "grad_norm": 0.16951896250247955,
+      "learning_rate": 4.068215521502719e-05,
+      "loss": 0.2234,
+      "step": 4960
+    },
+    {
+      "epoch": 0.4102592267135325,
+      "grad_norm": 0.30336254835128784,
+      "learning_rate": 4.084692700609656e-05,
+      "loss": 0.2152,
+      "step": 4980
+    },
+    {
+      "epoch": 0.41190685413005274,
+      "grad_norm": 1.0761586427688599,
+      "learning_rate": 4.101169879716593e-05,
+      "loss": 0.2093,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4135544815465729,
+      "grad_norm": 0.7221740484237671,
+      "learning_rate": 4.11764705882353e-05,
+      "loss": 0.1538,
+      "step": 5020
+    },
+    {
+      "epoch": 0.4152021089630931,
+      "grad_norm": 0.4801746904850006,
+      "learning_rate": 4.1341242379304665e-05,
+      "loss": 0.1911,
+      "step": 5040
+    },
+    {
+      "epoch": 0.41684973637961337,
+      "grad_norm": 0.311234712600708,
+      "learning_rate": 4.1506014170374034e-05,
+      "loss": 0.2015,
+      "step": 5060
+    },
+    {
+      "epoch": 0.41849736379613356,
+      "grad_norm": 0.6403760313987732,
+      "learning_rate": 4.16707859614434e-05,
+      "loss": 0.2014,
+      "step": 5080
+    },
+    {
+      "epoch": 0.42014499121265375,
+      "grad_norm": 1.2653217315673828,
+      "learning_rate": 4.183555775251277e-05,
+      "loss": 0.2388,
+      "step": 5100
+    },
+    {
+      "epoch": 0.421792618629174,
+      "grad_norm": 0.5536401867866516,
+      "learning_rate": 4.200032954358214e-05,
+      "loss": 0.1553,
+      "step": 5120
+    },
+    {
+      "epoch": 0.4234402460456942,
+      "grad_norm": 0.4605076014995575,
+      "learning_rate": 4.2165101334651506e-05,
+      "loss": 0.2105,
+      "step": 5140
+    },
+    {
+      "epoch": 0.42508787346221444,
+      "grad_norm": 0.28758004307746887,
+      "learning_rate": 4.232987312572088e-05,
+      "loss": 0.2067,
+      "step": 5160
+    },
+    {
+      "epoch": 0.42673550087873463,
+      "grad_norm": 0.36655622720718384,
+      "learning_rate": 4.249464491679025e-05,
+      "loss": 0.1976,
+      "step": 5180
+    },
+    {
+      "epoch": 0.4283831282952548,
+      "grad_norm": 0.9053062796592712,
+      "learning_rate": 4.265941670785962e-05,
+      "loss": 0.2102,
+      "step": 5200
+    },
+    {
+      "epoch": 0.43003075571177507,
+      "grad_norm": 0.5088081359863281,
+      "learning_rate": 4.2824188498928985e-05,
+      "loss": 0.1639,
+      "step": 5220
+    },
+    {
+      "epoch": 0.43167838312829526,
+      "grad_norm": 0.7981218695640564,
+      "learning_rate": 4.2988960289998354e-05,
+      "loss": 0.2024,
+      "step": 5240
+    },
+    {
+      "epoch": 0.43332601054481545,
+      "grad_norm": 0.7993011474609375,
+      "learning_rate": 4.315373208106772e-05,
+      "loss": 0.204,
+      "step": 5260
+    },
+    {
+      "epoch": 0.4349736379613357,
+      "grad_norm": 0.5183406472206116,
+      "learning_rate": 4.331850387213709e-05,
+      "loss": 0.2041,
+      "step": 5280
+    },
+    {
+      "epoch": 0.4366212653778559,
+      "grad_norm": 1.2263312339782715,
+      "learning_rate": 4.3483275663206465e-05,
+      "loss": 0.2182,
+      "step": 5300
+    },
+    {
+      "epoch": 0.4382688927943761,
+      "grad_norm": 0.7018775343894958,
+      "learning_rate": 4.364804745427583e-05,
+      "loss": 0.1685,
+      "step": 5320
+    },
+    {
+      "epoch": 0.43991652021089633,
+      "grad_norm": 0.7381752133369446,
+      "learning_rate": 4.38128192453452e-05,
+      "loss": 0.2073,
+      "step": 5340
+    },
+    {
+      "epoch": 0.4415641476274165,
+      "grad_norm": 0.4658122956752777,
+      "learning_rate": 4.397759103641457e-05,
+      "loss": 0.2031,
+      "step": 5360
+    },
+    {
+      "epoch": 0.4432117750439367,
+      "grad_norm": 0.34789761900901794,
+      "learning_rate": 4.414236282748394e-05,
+      "loss": 0.2023,
+      "step": 5380
+    },
+    {
+      "epoch": 0.44485940246045697,
+      "grad_norm": 0.9787063598632812,
+      "learning_rate": 4.4307134618553305e-05,
+      "loss": 0.2264,
+      "step": 5400
+    },
+    {
+      "epoch": 0.44650702987697716,
+      "grad_norm": 0.3786025047302246,
+      "learning_rate": 4.4471906409622673e-05,
+      "loss": 0.1656,
+      "step": 5420
+    },
+    {
+      "epoch": 0.44815465729349735,
+      "grad_norm": 0.4397692084312439,
+      "learning_rate": 4.463667820069204e-05,
+      "loss": 0.2023,
+      "step": 5440
+    },
+    {
+      "epoch": 0.4498022847100176,
+      "grad_norm": 0.20323756337165833,
+      "learning_rate": 4.480144999176141e-05,
+      "loss": 0.2146,
+      "step": 5460
+    },
+    {
+      "epoch": 0.4514499121265378,
+      "grad_norm": 0.2108180820941925,
+      "learning_rate": 4.496622178283078e-05,
+      "loss": 0.2012,
+      "step": 5480
+    },
+    {
+      "epoch": 0.453097539543058,
+      "grad_norm": 1.6603738069534302,
+      "learning_rate": 4.5130993573900146e-05,
+      "loss": 0.2257,
+      "step": 5500
+    },
+    {
+      "epoch": 0.45474516695957823,
+      "grad_norm": 1.6646907329559326,
+      "learning_rate": 4.5295765364969514e-05,
+      "loss": 0.1636,
+      "step": 5520
+    },
+    {
+      "epoch": 0.4563927943760984,
+      "grad_norm": 0.98222416639328,
+      "learning_rate": 4.546053715603889e-05,
+      "loss": 0.2018,
+      "step": 5540
+    },
+    {
+      "epoch": 0.4580404217926186,
+      "grad_norm": 0.6065666079521179,
+      "learning_rate": 4.562530894710826e-05,
+      "loss": 0.2351,
+      "step": 5560
+    },
+    {
+      "epoch": 0.45968804920913886,
+      "grad_norm": 0.4321737587451935,
+      "learning_rate": 4.5790080738177625e-05,
+      "loss": 0.2139,
+      "step": 5580
+    },
+    {
+      "epoch": 0.46133567662565905,
+      "grad_norm": 2.530203342437744,
+      "learning_rate": 4.5954852529246993e-05,
+      "loss": 0.232,
+      "step": 5600
+    },
+    {
+      "epoch": 0.46298330404217924,
+      "grad_norm": 0.8252795934677124,
+      "learning_rate": 4.611962432031636e-05,
+      "loss": 0.1792,
+      "step": 5620
+    },
+    {
+      "epoch": 0.4646309314586995,
+      "grad_norm": 4.3282880783081055,
+      "learning_rate": 4.628439611138573e-05,
+      "loss": 0.2079,
+      "step": 5640
+    },
+    {
+      "epoch": 0.4662785588752197,
+      "grad_norm": 0.2798108756542206,
+      "learning_rate": 4.6449167902455105e-05,
+      "loss": 0.2114,
+      "step": 5660
+    },
+    {
+      "epoch": 0.4679261862917399,
+      "grad_norm": 0.1499057412147522,
+      "learning_rate": 4.661393969352447e-05,
+      "loss": 0.2026,
+      "step": 5680
+    },
+    {
+      "epoch": 0.4695738137082601,
+      "grad_norm": 0.8664823770523071,
+      "learning_rate": 4.677871148459384e-05,
+      "loss": 0.2208,
+      "step": 5700
+    },
+    {
+      "epoch": 0.4712214411247803,
+      "grad_norm": 0.32247450947761536,
+      "learning_rate": 4.694348327566321e-05,
+      "loss": 0.175,
+      "step": 5720
+    },
+    {
+      "epoch": 0.4728690685413005,
+      "grad_norm": 0.4217327833175659,
+      "learning_rate": 4.710825506673258e-05,
+      "loss": 0.2207,
+      "step": 5740
+    },
+    {
+      "epoch": 0.47451669595782076,
+      "grad_norm": 0.9544996023178101,
+      "learning_rate": 4.7273026857801945e-05,
+      "loss": 0.2269,
+      "step": 5760
+    },
+    {
+      "epoch": 0.47616432337434095,
+      "grad_norm": 0.3182899057865143,
+      "learning_rate": 4.743779864887132e-05,
+      "loss": 0.224,
+      "step": 5780
+    },
+    {
+      "epoch": 0.47781195079086114,
+      "grad_norm": 0.669552743434906,
+      "learning_rate": 4.760257043994069e-05,
+      "loss": 0.2332,
+      "step": 5800
+    },
+    {
+      "epoch": 0.4794595782073814,
+      "grad_norm": 0.5297145247459412,
+      "learning_rate": 4.7767342231010056e-05,
+      "loss": 0.2233,
+      "step": 5820
+    },
+    {
+      "epoch": 0.4811072056239016,
+      "grad_norm": 0.5998116731643677,
+      "learning_rate": 4.7932114022079425e-05,
+      "loss": 0.2282,
+      "step": 5840
+    },
+    {
+      "epoch": 0.48275483304042177,
+      "grad_norm": 0.4391906261444092,
+      "learning_rate": 4.809688581314879e-05,
+      "loss": 0.2094,
+      "step": 5860
+    },
+    {
+      "epoch": 0.484402460456942,
+      "grad_norm": 0.559304416179657,
+      "learning_rate": 4.826165760421816e-05,
+      "loss": 0.2106,
+      "step": 5880
+    },
+    {
+      "epoch": 0.4860500878734622,
+      "grad_norm": 2.914066791534424,
+      "learning_rate": 4.842642939528753e-05,
+      "loss": 0.2484,
+      "step": 5900
+    },
+    {
+      "epoch": 0.4876977152899824,
+      "grad_norm": 0.29989850521087646,
+      "learning_rate": 4.85912011863569e-05,
+      "loss": 0.1631,
+      "step": 5920
+    },
+    {
+      "epoch": 0.48934534270650265,
+      "grad_norm": 2.1414361000061035,
+      "learning_rate": 4.8755972977426265e-05,
+      "loss": 0.2315,
+      "step": 5940
+    },
+    {
+      "epoch": 0.49099297012302284,
+      "grad_norm": 0.5668668746948242,
+      "learning_rate": 4.892074476849563e-05,
+      "loss": 0.218,
+      "step": 5960
+    },
+    {
+      "epoch": 0.49264059753954303,
+      "grad_norm": 0.34356266260147095,
+      "learning_rate": 4.9077277970011535e-05,
+      "loss": 0.2121,
+      "step": 5980
+    },
+    {
+      "epoch": 0.4942882249560633,
+      "grad_norm": 0.7654836773872375,
+      "learning_rate": 4.9242049761080904e-05,
+      "loss": 0.212,
+      "step": 6000
+    },
+    {
+      "epoch": 0.4942882249560633,
+      "eval_loss": 0.5638302564620972,
+      "eval_runtime": 264.9088,
+      "eval_samples_per_second": 82.251,
+      "eval_steps_per_second": 20.566,
+      "eval_wer": 0.24026168527782366,
+      "step": 6000
+    },
+    {
+      "epoch": 0.4959358523725835,
+      "grad_norm": 1.2249395847320557,
+      "learning_rate": 4.940682155215027e-05,
+      "loss": 0.1786,
+      "step": 6020
+    },
+    {
+      "epoch": 0.49758347978910367,
+      "grad_norm": 0.834241509437561,
+      "learning_rate": 4.957159334321964e-05,
+      "loss": 0.2193,
+      "step": 6040
+    },
+    {
+      "epoch": 0.4992311072056239,
+      "grad_norm": 1.9017058610916138,
+      "learning_rate": 4.9736365134289015e-05,
+      "loss": 0.2338,
+      "step": 6060
+    },
+    {
+      "epoch": 0.5008787346221442,
+      "grad_norm": 3.1669952869415283,
+      "learning_rate": 4.990113692535838e-05,
+      "loss": 0.2245,
+      "step": 6080
+    },
+    {
+      "epoch": 0.5025263620386643,
+      "grad_norm": 2.9794270992279053,
+      "learning_rate": 4.9999997353508944e-05,
+      "loss": 0.239,
+      "step": 6100
+    },
+    {
+      "epoch": 0.5041739894551845,
+      "grad_norm": 1.2019259929656982,
+      "learning_rate": 4.999996758049098e-05,
+      "loss": 0.1871,
+      "step": 6120
+    },
+    {
+      "epoch": 0.5058216168717048,
+      "grad_norm": 2.175334930419922,
+      "learning_rate": 4.999990472638076e-05,
+      "loss": 0.2265,
+      "step": 6140
+    },
+    {
+      "epoch": 0.5074692442882249,
+      "grad_norm": 1.4895635843276978,
+      "learning_rate": 4.999980879126146e-05,
+      "loss": 0.2259,
+      "step": 6160
+    },
+    {
+      "epoch": 0.5091168717047452,
+      "grad_norm": 3.355076789855957,
+      "learning_rate": 4.9999679775260015e-05,
+      "loss": 0.2316,
+      "step": 6180
+    },
+    {
+      "epoch": 0.5107644991212654,
+      "grad_norm": 3.545771598815918,
+      "learning_rate": 4.999951767854715e-05,
+      "loss": 0.2426,
+      "step": 6200
+    },
+    {
+      "epoch": 0.5124121265377856,
+      "grad_norm": 2.085190773010254,
+      "learning_rate": 4.999932250133736e-05,
+      "loss": 0.1884,
+      "step": 6220
+    },
+    {
+      "epoch": 0.5140597539543058,
+      "grad_norm": 1.6277663707733154,
+      "learning_rate": 4.999909424388892e-05,
+      "loss": 0.2332,
+      "step": 6240
+    },
+    {
+      "epoch": 0.5157073813708261,
+      "grad_norm": 3.9234516620635986,
+      "learning_rate": 4.9998832906503856e-05,
+      "loss": 0.2346,
+      "step": 6260
+    },
+    {
+      "epoch": 0.5173550087873462,
+      "grad_norm": 1.7027534246444702,
+      "learning_rate": 4.9998538489527984e-05,
+      "loss": 0.2339,
+      "step": 6280
+    },
+    {
+      "epoch": 0.5190026362038664,
+      "grad_norm": 2.7195184230804443,
+      "learning_rate": 4.9998210993350895e-05,
+      "loss": 0.2511,
+      "step": 6300
+    },
+    {
+      "epoch": 0.5206502636203867,
+      "grad_norm": 0.7267903089523315,
+      "learning_rate": 4.9997850418405945e-05,
+      "loss": 0.1765,
+      "step": 6320
+    },
+    {
+      "epoch": 0.5222978910369068,
+      "grad_norm": 0.751308798789978,
+      "learning_rate": 4.999745676517027e-05,
+      "loss": 0.2393,
+      "step": 6340
+    },
+    {
+      "epoch": 0.5239455184534271,
+      "grad_norm": 1.0236356258392334,
+      "learning_rate": 4.999703003416476e-05,
+      "loss": 0.2292,
+      "step": 6360
+    },
+    {
+      "epoch": 0.5255931458699473,
+      "grad_norm": 0.7693842649459839,
+      "learning_rate": 4.999657022595409e-05,
+      "loss": 0.2285,
+      "step": 6380
+    },
+    {
+      "epoch": 0.5272407732864675,
+      "grad_norm": 3.4103684425354004,
+      "learning_rate": 4.999607734114669e-05,
+      "loss": 0.2635,
+      "step": 6400
+    },
+    {
+      "epoch": 0.5288884007029877,
+      "grad_norm": 0.5887142419815063,
+      "learning_rate": 4.999555138039478e-05,
+      "loss": 0.1938,
+      "step": 6420
+    },
+    {
+      "epoch": 0.530536028119508,
+      "grad_norm": 1.6326848268508911,
+      "learning_rate": 4.999499234439433e-05,
+      "loss": 0.2112,
+      "step": 6440
+    },
+    {
+      "epoch": 0.5321836555360281,
+      "grad_norm": 0.4003719985485077,
+      "learning_rate": 4.9994400233885086e-05,
+      "loss": 0.2011,
+      "step": 6460
+    },
+    {
+      "epoch": 0.5338312829525483,
+      "grad_norm": 1.0633249282836914,
+      "learning_rate": 4.999377504965055e-05,
+      "loss": 0.2388,
+      "step": 6480
+    },
+    {
+      "epoch": 0.5354789103690686,
+      "grad_norm": 3.87857985496521,
+      "learning_rate": 4.999311679251799e-05,
+      "loss": 0.2222,
+      "step": 6500
+    },
+    {
+      "epoch": 0.5371265377855887,
+      "grad_norm": 1.1982243061065674,
+      "learning_rate": 4.999242546335845e-05,
+      "loss": 0.1736,
+      "step": 6520
+    },
+    {
+      "epoch": 0.538774165202109,
+      "grad_norm": 1.4149129390716553,
+      "learning_rate": 4.999170106308673e-05,
+      "loss": 0.2044,
+      "step": 6540
+    },
+    {
+      "epoch": 0.5404217926186292,
+      "grad_norm": 0.9540938138961792,
+      "learning_rate": 4.999094359266138e-05,
+      "loss": 0.2343,
+      "step": 6560
+    },
+    {
+      "epoch": 0.5420694200351494,
+      "grad_norm": 0.8150781989097595,
+      "learning_rate": 4.999015305308472e-05,
+      "loss": 0.2027,
+      "step": 6580
+    },
+    {
+      "epoch": 0.5437170474516696,
+      "grad_norm": 2.2410573959350586,
+      "learning_rate": 4.998932944540284e-05,
+      "loss": 0.2191,
+      "step": 6600
+    },
+    {
+      "epoch": 0.5453646748681898,
+      "grad_norm": 2.924471616744995,
+      "learning_rate": 4.998847277070556e-05,
+      "loss": 0.1776,
+      "step": 6620
+    },
+    {
+      "epoch": 0.54701230228471,
+      "grad_norm": 2.5102221965789795,
+      "learning_rate": 4.9987583030126484e-05,
+      "loss": 0.2172,
+      "step": 6640
+    },
+    {
+      "epoch": 0.5486599297012302,
+      "grad_norm": 1.2676066160202026,
+      "learning_rate": 4.998666022484295e-05,
+      "loss": 0.2235,
+      "step": 6660
+    },
+    {
+      "epoch": 0.5503075571177505,
+      "grad_norm": 1.7459567785263062,
+      "learning_rate": 4.998570435607605e-05,
+      "loss": 0.2114,
+      "step": 6680
+    },
+    {
+      "epoch": 0.5519551845342706,
+      "grad_norm": 2.4994170665740967,
+      "learning_rate": 4.998476565684759e-05,
+      "loss": 0.2321,
+      "step": 6700
+    },
+    {
+      "epoch": 0.5536028119507909,
+      "grad_norm": 1.436635136604309,
+      "learning_rate": 4.998374531796601e-05,
+      "loss": 0.1715,
+      "step": 6720
+    },
+    {
+      "epoch": 0.5552504393673111,
+      "grad_norm": 5.170281410217285,
+      "learning_rate": 4.9982691919458215e-05,
+      "loss": 0.2227,
+      "step": 6740
+    },
+    {
+      "epoch": 0.5568980667838312,
+      "grad_norm": 2.329806327819824,
+      "learning_rate": 4.9981605462718097e-05,
+      "loss": 0.2103,
+      "step": 6760
+    },
+    {
+      "epoch": 0.5585456942003515,
+      "grad_norm": 0.5105818510055542,
+      "learning_rate": 4.998048594918331e-05,
+      "loss": 0.2297,
+      "step": 6780
+    },
+    {
+      "epoch": 0.5601933216168717,
+      "grad_norm": 2.74815034866333,
+      "learning_rate": 4.997933338033525e-05,
+      "loss": 0.2454,
+      "step": 6800
+    },
+    {
+      "epoch": 0.5618409490333919,
+      "grad_norm": 1.0433367490768433,
+      "learning_rate": 4.997814775769904e-05,
+      "loss": 0.175,
+      "step": 6820
+    },
+    {
+      "epoch": 0.5634885764499121,
+      "grad_norm": 0.5902445316314697,
+      "learning_rate": 4.997692908284356e-05,
+      "loss": 0.2153,
+      "step": 6840
+    },
+    {
+      "epoch": 0.5651362038664324,
+      "grad_norm": 0.9332594275474548,
+      "learning_rate": 4.997567735738141e-05,
+      "loss": 0.2246,
+      "step": 6860
+    },
+    {
+      "epoch": 0.5667838312829525,
+      "grad_norm": 1.2199161052703857,
+      "learning_rate": 4.9974392582968934e-05,
+      "loss": 0.2281,
+      "step": 6880
+    },
+    {
+      "epoch": 0.5684314586994728,
+      "grad_norm": 2.2492709159851074,
+      "learning_rate": 4.9973074761306186e-05,
+      "loss": 0.238,
+      "step": 6900
+    },
+    {
+      "epoch": 0.570079086115993,
+      "grad_norm": 9.928802490234375,
+      "learning_rate": 4.997172389413699e-05,
+      "loss": 0.1731,
+      "step": 6920
+    },
+    {
+      "epoch": 0.5717267135325131,
+      "grad_norm": 1.4187493324279785,
+      "learning_rate": 4.997033998324886e-05,
+      "loss": 0.2267,
+      "step": 6940
+    },
+    {
+      "epoch": 0.5733743409490334,
+      "grad_norm": 1.0871655941009521,
+      "learning_rate": 4.996892303047306e-05,
+      "loss": 0.2478,
+      "step": 6960
+    },
+    {
+      "epoch": 0.5750219683655536,
+      "grad_norm": 0.9939442873001099,
+      "learning_rate": 4.996747303768456e-05,
+      "loss": 0.1992,
+      "step": 6980
+    },
+    {
+      "epoch": 0.5766695957820738,
+      "grad_norm": 48.66152572631836,
+      "learning_rate": 4.996599000680206e-05,
+      "loss": 0.2341,
+      "step": 7000
+    },
+    {
+      "epoch": 0.578317223198594,
+      "grad_norm": 0.8890938758850098,
+      "learning_rate": 4.996447393978797e-05,
+      "loss": 0.1687,
+      "step": 7020
+    },
+    {
+      "epoch": 0.5799648506151143,
+      "grad_norm": 0.8341367840766907,
+      "learning_rate": 4.996292483864843e-05,
+      "loss": 0.2175,
+      "step": 7040
+    },
+    {
+      "epoch": 0.5816124780316344,
+      "grad_norm": 1.6562838554382324,
+      "learning_rate": 4.996134270543326e-05,
+      "loss": 0.2136,
+      "step": 7060
+    },
+    {
+      "epoch": 0.5832601054481547,
+      "grad_norm": 0.6537133455276489,
+      "learning_rate": 4.9959727542236025e-05,
+      "loss": 0.2139,
+      "step": 7080
+    },
+    {
+      "epoch": 0.5849077328646749,
+      "grad_norm": 2.101271390914917,
+      "learning_rate": 4.9958079351193976e-05,
+      "loss": 0.2359,
+      "step": 7100
+    },
+    {
+      "epoch": 0.586555360281195,
+      "grad_norm": 0.9616307616233826,
+      "learning_rate": 4.995639813448808e-05,
+      "loss": 0.1774,
+      "step": 7120
+    },
+    {
+      "epoch": 0.5882029876977153,
+      "grad_norm": 1.0992920398712158,
+      "learning_rate": 4.9954683894343e-05,
+      "loss": 0.1962,
+      "step": 7140
+    },
+    {
+      "epoch": 0.5898506151142355,
+      "grad_norm": 0.6329948902130127,
+      "learning_rate": 4.995293663302709e-05,
+      "loss": 0.2145,
+      "step": 7160
+    },
+    {
+      "epoch": 0.5914982425307557,
+      "grad_norm": 1.0156137943267822,
+      "learning_rate": 4.9951156352852415e-05,
+      "loss": 0.2161,
+      "step": 7180
+    },
+    {
+      "epoch": 0.5931458699472759,
+      "grad_norm": 2.6131973266601562,
+      "learning_rate": 4.994943450511368e-05,
+      "loss": 0.2234,
+      "step": 7200
+    },
+    {
+      "epoch": 0.5947934973637962,
+      "grad_norm": 3.4186134338378906,
+      "learning_rate": 4.99475898449799e-05,
+      "loss": 0.171,
+      "step": 7220
+    },
+    {
+      "epoch": 0.5964411247803163,
+      "grad_norm": 4.896924018859863,
+      "learning_rate": 4.9945712173062477e-05,
+      "loss": 0.2073,
+      "step": 7240
+    },
+    {
+      "epoch": 0.5980887521968365,
+      "grad_norm": 1.2636560201644897,
+      "learning_rate": 4.994380149184601e-05,
+      "loss": 0.2144,
+      "step": 7260
+    },
+    {
+      "epoch": 0.5997363796133568,
+      "grad_norm": 1.4861242771148682,
+      "learning_rate": 4.99418578038588e-05,
+      "loss": 0.2024,
+      "step": 7280
+    },
+    {
+      "epoch": 0.6013840070298769,
+      "grad_norm": 6.527903079986572,
+      "learning_rate": 4.993988111167284e-05,
+      "loss": 0.2172,
+      "step": 7300
+    },
+    {
+      "epoch": 0.6030316344463972,
+      "grad_norm": 1.9219621419906616,
+      "learning_rate": 4.993787141790375e-05,
+      "loss": 0.1865,
+      "step": 7320
+    },
+    {
+      "epoch": 0.6046792618629174,
+      "grad_norm": 0.7625167369842529,
+      "learning_rate": 4.9935828725210874e-05,
+      "loss": 0.2238,
+      "step": 7340
+    },
+    {
+      "epoch": 0.6063268892794376,
+      "grad_norm": 1.3375577926635742,
+      "learning_rate": 4.9933753036297196e-05,
+      "loss": 0.1878,
+      "step": 7360
+    },
+    {
+      "epoch": 0.6079745166959578,
+      "grad_norm": 1.1067700386047363,
+      "learning_rate": 4.993164435390935e-05,
+      "loss": 0.2111,
+      "step": 7380
+    },
+    {
+      "epoch": 0.6096221441124781,
+      "grad_norm": 4.094808578491211,
+      "learning_rate": 4.992950268083764e-05,
+      "loss": 0.2073,
+      "step": 7400
+    },
+    {
+      "epoch": 0.6112697715289982,
+      "grad_norm": 2.683678150177002,
+      "learning_rate": 4.992732801991602e-05,
+      "loss": 0.172,
+      "step": 7420
+    },
+    {
+      "epoch": 0.6129173989455184,
+      "grad_norm": 0.9752517342567444,
+      "learning_rate": 4.992512037402212e-05,
+      "loss": 0.1924,
+      "step": 7440
+    },
+    {
+      "epoch": 0.6145650263620387,
+      "grad_norm": 0.8463476896286011,
+      "learning_rate": 4.9922879746077176e-05,
+      "loss": 0.2021,
+      "step": 7460
+    },
+    {
+      "epoch": 0.6162126537785588,
+      "grad_norm": 3.053812026977539,
+      "learning_rate": 4.992060613904611e-05,
+      "loss": 0.2086,
+      "step": 7480
+    },
+    {
+      "epoch": 0.6178602811950791,
+      "grad_norm": 2.9483275413513184,
+      "learning_rate": 4.991829955593744e-05,
+      "loss": 0.2154,
+      "step": 7500
+    },
+    {
+      "epoch": 0.6195079086115993,
+      "grad_norm": 1.3794467449188232,
+      "learning_rate": 4.9915959999803365e-05,
+      "loss": 0.1599,
+      "step": 7520
+    },
+    {
+      "epoch": 0.6211555360281195,
+      "grad_norm": 1.0529364347457886,
+      "learning_rate": 4.9913587473739666e-05,
+      "loss": 0.1891,
+      "step": 7540
+    },
+    {
+      "epoch": 0.6228031634446397,
+      "grad_norm": 0.9026776552200317,
+      "learning_rate": 4.991118198088579e-05,
+      "loss": 0.2074,
+      "step": 7560
+    },
+    {
+      "epoch": 0.62445079086116,
+      "grad_norm": 0.4504956603050232,
+      "learning_rate": 4.9908743524424806e-05,
+      "loss": 0.1931,
+      "step": 7580
+    },
+    {
+      "epoch": 0.6260984182776801,
+      "grad_norm": 4.0763726234436035,
+      "learning_rate": 4.9906272107583366e-05,
+      "loss": 0.2192,
+      "step": 7600
+    },
+    {
+      "epoch": 0.6277460456942003,
+      "grad_norm": 1.012601375579834,
+      "learning_rate": 4.990376773363178e-05,
+      "loss": 0.1604,
+      "step": 7620
+    },
+    {
+      "epoch": 0.6293936731107206,
+      "grad_norm": 0.6183291673660278,
+      "learning_rate": 4.990123040588395e-05,
+      "loss": 0.2015,
+      "step": 7640
+    },
+    {
+      "epoch": 0.6310413005272407,
+      "grad_norm": 0.718015730381012,
+      "learning_rate": 4.989866012769736e-05,
+      "loss": 0.2155,
+      "step": 7660
+    },
+    {
+      "epoch": 0.632688927943761,
+      "grad_norm": 1.5611259937286377,
+      "learning_rate": 4.989605690247315e-05,
+      "loss": 0.1975,
+      "step": 7680
+    },
+    {
+      "epoch": 0.6343365553602812,
+      "grad_norm": 3.2936527729034424,
+      "learning_rate": 4.9893420733656e-05,
+      "loss": 0.2305,
+      "step": 7700
+    },
+    {
+      "epoch": 0.6359841827768014,
+      "grad_norm": 1.3771491050720215,
+      "learning_rate": 4.9890751624734225e-05,
+      "loss": 0.1575,
+      "step": 7720
+    },
+    {
+      "epoch": 0.6376318101933216,
+      "grad_norm": 1.0577577352523804,
+      "learning_rate": 4.98880495792397e-05,
+      "loss": 0.2045,
+      "step": 7740
+    },
+    {
+      "epoch": 0.6392794376098418,
+      "grad_norm": 3.6020660400390625,
+      "learning_rate": 4.988531460074791e-05,
+      "loss": 0.2034,
+      "step": 7760
+    },
+    {
+      "epoch": 0.640927065026362,
+      "grad_norm": 0.7804440855979919,
+      "learning_rate": 4.9882546692877885e-05,
+      "loss": 0.1953,
+      "step": 7780
+    },
+    {
+      "epoch": 0.6425746924428822,
+      "grad_norm": 57.011390686035156,
+      "learning_rate": 4.987974585929226e-05,
+      "loss": 0.2207,
+      "step": 7800
+    },
+    {
+      "epoch": 0.6442223198594025,
+      "grad_norm": 1.5725603103637695,
+      "learning_rate": 4.987691210369721e-05,
+      "loss": 0.1664,
+      "step": 7820
+    },
+    {
+      "epoch": 0.6458699472759226,
+      "grad_norm": 0.6347635984420776,
+      "learning_rate": 4.98740454298425e-05,
+      "loss": 0.2134,
+      "step": 7840
+    },
+    {
+      "epoch": 0.6475175746924429,
+      "grad_norm": 1.3581891059875488,
+      "learning_rate": 4.987114584152145e-05,
+      "loss": 0.217,
+      "step": 7860
+    },
+    {
+      "epoch": 0.6491652021089631,
+      "grad_norm": 5.659873008728027,
+      "learning_rate": 4.986821334257091e-05,
+      "loss": 0.1977,
+      "step": 7880
+    },
+    {
+      "epoch": 0.6508128295254832,
+      "grad_norm": 11.7850980758667,
+      "learning_rate": 4.986524793687131e-05,
+      "loss": 0.218,
+      "step": 7900
+    },
+    {
+      "epoch": 0.6524604569420035,
+      "grad_norm": 1.2856826782226562,
+      "learning_rate": 4.986224962834659e-05,
+      "loss": 0.1675,
+      "step": 7920
+    },
+    {
+      "epoch": 0.6541080843585237,
+      "grad_norm": 1.055503249168396,
+      "learning_rate": 4.985921842096427e-05,
+      "loss": 0.2113,
+      "step": 7940
+    },
+    {
+      "epoch": 0.6557557117750439,
+      "grad_norm": 1.5587449073791504,
+      "learning_rate": 4.9856154318735374e-05,
+      "loss": 0.1939,
+      "step": 7960
+    },
+    {
+      "epoch": 0.6574033391915641,
+      "grad_norm": 2.40641450881958,
+      "learning_rate": 4.985305732571446e-05,
+      "loss": 0.206,
+      "step": 7980
+    },
+    {
+      "epoch": 0.6590509666080844,
+      "grad_norm": 4.06686544418335,
+      "learning_rate": 4.98499274459996e-05,
+      "loss": 0.2107,
+      "step": 8000
+    },
+    {
+      "epoch": 0.6590509666080844,
+      "eval_loss": 0.5655311346054077,
+      "eval_runtime": 835.4807,
+      "eval_samples_per_second": 26.08,
+      "eval_steps_per_second": 6.521,
+      "eval_wer": 0.2457764477651215,
+      "step": 8000
+    },
+    {
+      "epoch": 0.6606985940246046,
+      "grad_norm": 0.8585990071296692,
+      "learning_rate": 4.984676468373241e-05,
+      "loss": 0.1667,
+      "step": 8020
+    },
+    {
+      "epoch": 0.6623462214411248,
+      "grad_norm": 1.2851459980010986,
+      "learning_rate": 4.984356904309799e-05,
+      "loss": 0.2184,
+      "step": 8040
+    },
+    {
+      "epoch": 0.663993848857645,
+      "grad_norm": 0.8445897698402405,
+      "learning_rate": 4.984034052832496e-05,
+      "loss": 0.2169,
+      "step": 8060
+    },
+    {
+      "epoch": 0.6656414762741653,
+      "grad_norm": 1.4265161752700806,
+      "learning_rate": 4.983707914368544e-05,
+      "loss": 0.2004,
+      "step": 8080
+    },
+    {
+      "epoch": 0.6672891036906854,
+      "grad_norm": 3.0620169639587402,
+      "learning_rate": 4.983378489349504e-05,
+      "loss": 0.2467,
+      "step": 8100
+    },
+    {
+      "epoch": 0.6689367311072056,
+      "grad_norm": 1.1079747676849365,
+      "learning_rate": 4.983045778211286e-05,
+      "loss": 0.1587,
+      "step": 8120
+    },
+    {
+      "epoch": 0.6705843585237259,
+      "grad_norm": 0.8565286993980408,
+      "learning_rate": 4.982709781394148e-05,
+      "loss": 0.2101,
+      "step": 8140
+    },
+    {
+      "epoch": 0.672231985940246,
+      "grad_norm": 2.957345962524414,
+      "learning_rate": 4.982370499342698e-05,
+      "loss": 0.212,
+      "step": 8160
+    },
+    {
+      "epoch": 0.6738796133567663,
+      "grad_norm": 1.744805932044983,
+      "learning_rate": 4.982027932505887e-05,
+      "loss": 0.209,
+      "step": 8180
+    },
+    {
+      "epoch": 0.6755272407732865,
+      "grad_norm": 3.963977336883545,
+      "learning_rate": 4.9816820813370166e-05,
+      "loss": 0.214,
+      "step": 8200
+    },
+    {
+      "epoch": 0.6771748681898067,
+      "grad_norm": 1.3982582092285156,
+      "learning_rate": 4.981332946293733e-05,
+      "loss": 0.1671,
+      "step": 8220
+    },
+    {
+      "epoch": 0.6788224956063269,
+      "grad_norm": 0.5959907174110413,
+      "learning_rate": 4.9809805278380264e-05,
+      "loss": 0.2061,
+      "step": 8240
+    },
+    {
+      "epoch": 0.6804701230228472,
+      "grad_norm": 1.2237777709960938,
+      "learning_rate": 4.980624826436233e-05,
+      "loss": 0.2188,
+      "step": 8260
+    },
+    {
+      "epoch": 0.6821177504393673,
+      "grad_norm": 2.6652753353118896,
+      "learning_rate": 4.9802658425590344e-05,
+      "loss": 0.1964,
+      "step": 8280
+    },
+    {
+      "epoch": 0.6837653778558875,
+      "grad_norm": 31.890893936157227,
+      "learning_rate": 4.979903576681453e-05,
+      "loss": 0.2133,
+      "step": 8300
+    },
+    {
+      "epoch": 0.6854130052724078,
+      "grad_norm": 1.1191130876541138,
+      "learning_rate": 4.979538029282855e-05,
+      "loss": 0.1623,
+      "step": 8320
+    },
+    {
+      "epoch": 0.6870606326889279,
+      "grad_norm": 1.4404329061508179,
+      "learning_rate": 4.9791692008469514e-05,
+      "loss": 0.2087,
+      "step": 8340
+    },
+    {
+      "epoch": 0.6887082601054482,
+      "grad_norm": 1.4656809568405151,
+      "learning_rate": 4.9787970918617914e-05,
+      "loss": 0.2134,
+      "step": 8360
+    },
+    {
+      "epoch": 0.6903558875219684,
+      "grad_norm": 2.310316801071167,
+      "learning_rate": 4.978421702819767e-05,
+      "loss": 0.215,
+      "step": 8380
+    },
+    {
+      "epoch": 0.6920035149384886,
+      "grad_norm": 21.141889572143555,
+      "learning_rate": 4.978043034217609e-05,
+      "loss": 0.2206,
+      "step": 8400
+    },
+    {
+      "epoch": 0.6936511423550088,
+      "grad_norm": 1.2799972295761108,
+      "learning_rate": 4.977680261809319e-05,
+      "loss": 0.1553,
+      "step": 8420
+    },
+    {
+      "epoch": 0.695298769771529,
+      "grad_norm": 0.927029550075531,
+      "learning_rate": 4.97729519951006e-05,
+      "loss": 0.2205,
+      "step": 8440
+    },
+    {
+      "epoch": 0.6969463971880492,
+      "grad_norm": 1.8685028553009033,
+      "learning_rate": 4.976906859141309e-05,
+      "loss": 0.1938,
+      "step": 8460
+    },
+    {
+      "epoch": 0.6985940246045694,
+      "grad_norm": 0.38425010442733765,
+      "learning_rate": 4.976515241216936e-05,
+      "loss": 0.1964,
+      "step": 8480
+    },
+    {
+      "epoch": 0.7002416520210897,
+      "grad_norm": 2.6303980350494385,
+      "learning_rate": 4.976120346255146e-05,
+      "loss": 0.1984,
+      "step": 8500
+    },
+    {
+      "epoch": 0.7018892794376098,
+      "grad_norm": 7.791947841644287,
+      "learning_rate": 4.975722174778482e-05,
+      "loss": 0.1678,
+      "step": 8520
+    },
+    {
+      "epoch": 0.7035369068541301,
+      "grad_norm": 1.193328857421875,
+      "learning_rate": 4.9753207273138245e-05,
+      "loss": 0.2182,
+      "step": 8540
+    },
+    {
+      "epoch": 0.7051845342706503,
+      "grad_norm": 1.1064016819000244,
+      "learning_rate": 4.974916004392385e-05,
+      "loss": 0.2065,
+      "step": 8560
+    },
+    {
+      "epoch": 0.7068321616871704,
+      "grad_norm": 3.304832696914673,
+      "learning_rate": 4.974508006549711e-05,
+      "loss": 0.1872,
+      "step": 8580
+    },
+    {
+      "epoch": 0.7084797891036907,
+      "grad_norm": 2.5478882789611816,
+      "learning_rate": 4.974096734325686e-05,
+      "loss": 0.2295,
+      "step": 8600
+    },
+    {
+      "epoch": 0.7101274165202109,
+      "grad_norm": 14.596063613891602,
+      "learning_rate": 4.9736821882645226e-05,
+      "loss": 0.1628,
+      "step": 8620
+    },
+    {
+      "epoch": 0.7117750439367311,
+      "grad_norm": 0.9999619126319885,
+      "learning_rate": 4.973264368914766e-05,
+      "loss": 0.2107,
+      "step": 8640
+    },
+    {
+      "epoch": 0.7134226713532513,
+      "grad_norm": 1.053412675857544,
+      "learning_rate": 4.972843276829296e-05,
+      "loss": 0.2085,
+      "step": 8660
+    },
+    {
+      "epoch": 0.7150702987697716,
+      "grad_norm": 2.1676464080810547,
+      "learning_rate": 4.9724189125653195e-05,
+      "loss": 0.2048,
+      "step": 8680
+    },
+    {
+      "epoch": 0.7167179261862917,
+      "grad_norm": 10.32970142364502,
+      "learning_rate": 4.9719912766843746e-05,
+      "loss": 0.2224,
+      "step": 8700
+    },
+    {
+      "epoch": 0.718365553602812,
+      "grad_norm": 1.9829623699188232,
+      "learning_rate": 4.971560369752328e-05,
+      "loss": 0.1686,
+      "step": 8720
+    },
+    {
+      "epoch": 0.7200131810193322,
+      "grad_norm": 2.7383005619049072,
+      "learning_rate": 4.971126192339377e-05,
+      "loss": 0.2088,
+      "step": 8740
+    },
+    {
+      "epoch": 0.7216608084358523,
+      "grad_norm": 0.6413734555244446,
+      "learning_rate": 4.970688745020043e-05,
+      "loss": 0.2111,
+      "step": 8760
+    },
+    {
+      "epoch": 0.7233084358523726,
+      "grad_norm": 0.8469798564910889,
+      "learning_rate": 4.970248028373178e-05,
+      "loss": 0.2057,
+      "step": 8780
+    },
+    {
+      "epoch": 0.7249560632688928,
+      "grad_norm": 8.44117259979248,
+      "learning_rate": 4.969804042981956e-05,
+      "loss": 0.2179,
+      "step": 8800
+    },
+    {
+      "epoch": 0.726603690685413,
+      "grad_norm": 1.40958571434021,
+      "learning_rate": 4.969356789433881e-05,
+      "loss": 0.1604,
+      "step": 8820
+    },
+    {
+      "epoch": 0.7282513181019332,
+      "grad_norm": 5.328885078430176,
+      "learning_rate": 4.968906268320777e-05,
+      "loss": 0.2075,
+      "step": 8840
+    },
+    {
+      "epoch": 0.7298989455184535,
+      "grad_norm": 0.6396345496177673,
+      "learning_rate": 4.9684524802387956e-05,
+      "loss": 0.2057,
+      "step": 8860
+    },
+    {
+      "epoch": 0.7315465729349736,
+      "grad_norm": 0.7661327123641968,
+      "learning_rate": 4.967995425788409e-05,
+      "loss": 0.2064,
+      "step": 8880
+    },
+    {
+      "epoch": 0.7331942003514939,
+      "grad_norm": 2.8300130367279053,
+      "learning_rate": 4.9675351055744134e-05,
+      "loss": 0.2109,
+      "step": 8900
+    },
+    {
+      "epoch": 0.7348418277680141,
+      "grad_norm": 1.3380045890808105,
+      "learning_rate": 4.9670715202059235e-05,
+      "loss": 0.1492,
+      "step": 8920
+    },
+    {
+      "epoch": 0.7364894551845342,
+      "grad_norm": 1.478825569152832,
+      "learning_rate": 4.9666046702963784e-05,
+      "loss": 0.2133,
+      "step": 8940
+    },
+    {
+      "epoch": 0.7381370826010545,
+      "grad_norm": 1.3596333265304565,
+      "learning_rate": 4.9661345564635356e-05,
+      "loss": 0.1999,
+      "step": 8960
+    },
+    {
+      "epoch": 0.7397847100175747,
+      "grad_norm": 0.3950871229171753,
+      "learning_rate": 4.965661179329468e-05,
+      "loss": 0.1925,
+      "step": 8980
+    },
+    {
+      "epoch": 0.7414323374340949,
+      "grad_norm": 3.47658634185791,
+      "learning_rate": 4.965184539520572e-05,
+      "loss": 0.2016,
+      "step": 9000
+    },
+    {
+      "epoch": 0.7430799648506151,
+      "grad_norm": 1.6241281032562256,
+      "learning_rate": 4.9647046376675586e-05,
+      "loss": 0.1522,
+      "step": 9020
+    },
+    {
+      "epoch": 0.7447275922671354,
+      "grad_norm": 0.582032322883606,
+      "learning_rate": 4.964221474405456e-05,
+      "loss": 0.2161,
+      "step": 9040
+    },
+    {
+      "epoch": 0.7463752196836555,
+      "grad_norm": 1.1760663986206055,
+      "learning_rate": 4.963735050373608e-05,
+      "loss": 0.1986,
+      "step": 9060
+    },
+    {
+      "epoch": 0.7480228471001757,
+      "grad_norm": 1.9206633567810059,
+      "learning_rate": 4.963245366215672e-05,
+      "loss": 0.195,
+      "step": 9080
+    },
+    {
+      "epoch": 0.749670474516696,
+      "grad_norm": 2.0300755500793457,
+      "learning_rate": 4.9627524225796206e-05,
+      "loss": 0.2044,
+      "step": 9100
+    },
+    {
+      "epoch": 0.7513181019332161,
+      "grad_norm": 0.91159987449646,
+      "learning_rate": 4.962256220117739e-05,
+      "loss": 0.156,
+      "step": 9120
+    },
+    {
+      "epoch": 0.7529657293497364,
+      "grad_norm": 0.6553214192390442,
+      "learning_rate": 4.961756759486625e-05,
+      "loss": 0.1918,
+      "step": 9140
+    },
+    {
+      "epoch": 0.7546133567662566,
+      "grad_norm": 2.1197195053100586,
+      "learning_rate": 4.961254041347189e-05,
+      "loss": 0.1942,
+      "step": 9160
+    },
+    {
+      "epoch": 0.7562609841827768,
+      "grad_norm": 0.5275347232818604,
+      "learning_rate": 4.9607480663646487e-05,
+      "loss": 0.1975,
+      "step": 9180
+    },
+    {
+      "epoch": 0.757908611599297,
+      "grad_norm": 2.276308536529541,
+      "learning_rate": 4.9602388352085337e-05,
+      "loss": 0.2035,
+      "step": 9200
+    },
+    {
+      "epoch": 0.7595562390158173,
+      "grad_norm": 1.4071913957595825,
+      "learning_rate": 4.9597263485526826e-05,
+      "loss": 0.1636,
+      "step": 9220
+    },
+    {
+      "epoch": 0.7612038664323374,
+      "grad_norm": 5.9329023361206055,
+      "learning_rate": 4.959210607075239e-05,
+      "loss": 0.1954,
+      "step": 9240
+    },
+    {
+      "epoch": 0.7628514938488576,
+      "grad_norm": 1.0558881759643555,
+      "learning_rate": 4.958691611458657e-05,
+      "loss": 0.2089,
+      "step": 9260
+    },
+    {
+      "epoch": 0.7644991212653779,
+      "grad_norm": 1.1962648630142212,
+      "learning_rate": 4.958169362389695e-05,
+      "loss": 0.1966,
+      "step": 9280
+    },
+    {
+      "epoch": 0.766146748681898,
+      "grad_norm": 27.050457000732422,
+      "learning_rate": 4.957643860559417e-05,
+      "loss": 0.2254,
+      "step": 9300
+    },
+    {
+      "epoch": 0.7677943760984183,
+      "grad_norm": 1.5262819528579712,
+      "learning_rate": 4.95711510666319e-05,
+      "loss": 0.1672,
+      "step": 9320
+    },
+    {
+      "epoch": 0.7694420035149385,
+      "grad_norm": 1.5432629585266113,
+      "learning_rate": 4.956583101400685e-05,
+      "loss": 0.2201,
+      "step": 9340
+    },
+    {
+      "epoch": 0.7710896309314587,
+      "grad_norm": 1.4080497026443481,
+      "learning_rate": 4.956047845475877e-05,
+      "loss": 0.2034,
+      "step": 9360
+    },
+    {
+      "epoch": 0.7727372583479789,
+      "grad_norm": 1.6339974403381348,
+      "learning_rate": 4.9555093395970396e-05,
+      "loss": 0.1918,
+      "step": 9380
+    },
+    {
+      "epoch": 0.7743848857644992,
+      "grad_norm": 3.812486410140991,
+      "learning_rate": 4.954967584476748e-05,
+      "loss": 0.2221,
+      "step": 9400
+    },
+    {
+      "epoch": 0.7760325131810193,
+      "grad_norm": 2.0563764572143555,
+      "learning_rate": 4.954422580831879e-05,
+      "loss": 0.1611,
+      "step": 9420
+    },
+    {
+      "epoch": 0.7776801405975395,
+      "grad_norm": 0.7051644325256348,
+      "learning_rate": 4.9538743293836046e-05,
+      "loss": 0.2053,
+      "step": 9440
+    },
+    {
+      "epoch": 0.7793277680140598,
+      "grad_norm": 0.837563157081604,
+      "learning_rate": 4.9533228308573966e-05,
+      "loss": 0.2078,
+      "step": 9460
+    },
+    {
+      "epoch": 0.7809753954305799,
+      "grad_norm": 1.1410398483276367,
+      "learning_rate": 4.952768085983023e-05,
+      "loss": 0.2071,
+      "step": 9480
+    },
+    {
+      "epoch": 0.7826230228471002,
+      "grad_norm": 2.372843027114868,
+      "learning_rate": 4.952210095494546e-05,
+      "loss": 0.2154,
+      "step": 9500
+    },
+    {
+      "epoch": 0.7842706502636204,
+      "grad_norm": 25.939308166503906,
+      "learning_rate": 4.9516488601303255e-05,
+      "loss": 0.1618,
+      "step": 9520
+    },
+    {
+      "epoch": 0.7859182776801406,
+      "grad_norm": 1.780004858970642,
+      "learning_rate": 4.951084380633013e-05,
+      "loss": 0.1993,
+      "step": 9540
+    },
+    {
+      "epoch": 0.7875659050966608,
+      "grad_norm": 1.1589267253875732,
+      "learning_rate": 4.9505166577495546e-05,
+      "loss": 0.1939,
+      "step": 9560
+    },
+    {
+      "epoch": 0.789213532513181,
+      "grad_norm": 1.0380936861038208,
+      "learning_rate": 4.949945692231185e-05,
+      "loss": 0.1863,
+      "step": 9580
+    },
+    {
+      "epoch": 0.7908611599297012,
+      "grad_norm": 3.860309600830078,
+      "learning_rate": 4.949371484833433e-05,
+      "loss": 0.2199,
+      "step": 9600
+    },
+    {
+      "epoch": 0.7925087873462214,
+      "grad_norm": 2.6430704593658447,
+      "learning_rate": 4.9487940363161155e-05,
+      "loss": 0.161,
+      "step": 9620
+    },
+    {
+      "epoch": 0.7941564147627417,
+      "grad_norm": 2.1502444744110107,
+      "learning_rate": 4.948213347443339e-05,
+      "loss": 0.2101,
+      "step": 9640
+    },
+    {
+      "epoch": 0.7958040421792618,
+      "grad_norm": 0.9621193408966064,
+      "learning_rate": 4.9476294189834974e-05,
+      "loss": 0.2105,
+      "step": 9660
+    },
+    {
+      "epoch": 0.7974516695957821,
+      "grad_norm": 0.8426349759101868,
+      "learning_rate": 4.9470422517092696e-05,
+      "loss": 0.1821,
+      "step": 9680
+    },
+    {
+      "epoch": 0.7990992970123023,
+      "grad_norm": 2.1861345767974854,
+      "learning_rate": 4.9464518463976246e-05,
+      "loss": 0.2157,
+      "step": 9700
+    },
+    {
+      "epoch": 0.8007469244288224,
+      "grad_norm": 1.074504017829895,
+      "learning_rate": 4.945858203829812e-05,
+      "loss": 0.1593,
+      "step": 9720
+    },
+    {
+      "epoch": 0.8023945518453427,
+      "grad_norm": 1.4774161577224731,
+      "learning_rate": 4.945261324791367e-05,
+      "loss": 0.1981,
+      "step": 9740
+    },
+    {
+      "epoch": 0.8040421792618629,
+      "grad_norm": 0.9464648962020874,
+      "learning_rate": 4.944661210072107e-05,
+      "loss": 0.2107,
+      "step": 9760
+    },
+    {
+      "epoch": 0.8056898066783831,
+      "grad_norm": 0.34157031774520874,
+      "learning_rate": 4.94405786046613e-05,
+      "loss": 0.209,
+      "step": 9780
+    },
+    {
+      "epoch": 0.8073374340949033,
+      "grad_norm": 3.6989963054656982,
+      "learning_rate": 4.943451276771818e-05,
+      "loss": 0.221,
+      "step": 9800
+    },
+    {
+      "epoch": 0.8089850615114236,
+      "grad_norm": 0.6388671398162842,
+      "learning_rate": 4.942841459791828e-05,
+      "loss": 0.1661,
+      "step": 9820
+    },
+    {
+      "epoch": 0.8106326889279437,
+      "grad_norm": 0.6647291779518127,
+      "learning_rate": 4.9422284103330985e-05,
+      "loss": 0.192,
+      "step": 9840
+    },
+    {
+      "epoch": 0.812280316344464,
+      "grad_norm": 0.540134847164154,
+      "learning_rate": 4.941612129206844e-05,
+      "loss": 0.2126,
+      "step": 9860
+    },
+    {
+      "epoch": 0.8139279437609842,
+      "grad_norm": 0.5511148571968079,
+      "learning_rate": 4.940992617228556e-05,
+      "loss": 0.2018,
+      "step": 9880
+    },
+    {
+      "epoch": 0.8155755711775043,
+      "grad_norm": 2.1634738445281982,
+      "learning_rate": 4.9403698752180006e-05,
+      "loss": 0.2087,
+      "step": 9900
+    },
+    {
+      "epoch": 0.8172231985940246,
+      "grad_norm": 1.3764668703079224,
+      "learning_rate": 4.939743903999218e-05,
+      "loss": 0.1561,
+      "step": 9920
+    },
+    {
+      "epoch": 0.8188708260105448,
+      "grad_norm": 0.8314900398254395,
+      "learning_rate": 4.939114704400523e-05,
+      "loss": 0.2226,
+      "step": 9940
+    },
+    {
+      "epoch": 0.820518453427065,
+      "grad_norm": 1.0714060068130493,
+      "learning_rate": 4.9384822772544994e-05,
+      "loss": 0.191,
+      "step": 9960
+    },
+    {
+      "epoch": 0.8221660808435852,
+      "grad_norm": 0.5301602482795715,
+      "learning_rate": 4.937846623398003e-05,
+      "loss": 0.2069,
+      "step": 9980
+    },
+    {
+      "epoch": 0.8238137082601055,
+      "grad_norm": 2.4229841232299805,
+      "learning_rate": 4.9372077436721634e-05,
+      "loss": 0.2132,
+      "step": 10000
+    },
+    {
+      "epoch": 0.8238137082601055,
+      "eval_loss": 0.5383469462394714,
+      "eval_runtime": 248.9799,
+      "eval_samples_per_second": 87.513,
+      "eval_steps_per_second": 21.881,
+      "eval_wer": 0.23184117484036582,
+      "step": 10000
+    },
+    {
+      "epoch": 0.8254613356766256,
+      "grad_norm": 1.1694583892822266,
+      "learning_rate": 4.936565638922372e-05,
+      "loss": 0.1624,
+      "step": 10020
+    },
+    {
+      "epoch": 0.8271089630931459,
+      "grad_norm": 2.124119281768799,
+      "learning_rate": 4.9359203099982924e-05,
+      "loss": 0.2062,
+      "step": 10040
+    },
+    {
+      "epoch": 0.8287565905096661,
+      "grad_norm": 0.7251669764518738,
+      "learning_rate": 4.935271757753852e-05,
+      "loss": 0.1981,
+      "step": 10060
+    },
+    {
+      "epoch": 0.8304042179261862,
+      "grad_norm": 1.2785292863845825,
+      "learning_rate": 4.934619983047246e-05,
+      "loss": 0.2094,
+      "step": 10080
+    },
+    {
+      "epoch": 0.8320518453427065,
+      "grad_norm": 3.0615155696868896,
+      "learning_rate": 4.933964986740931e-05,
+      "loss": 0.2159,
+      "step": 10100
+    },
+    {
+      "epoch": 0.8336994727592267,
+      "grad_norm": 2.058302402496338,
+      "learning_rate": 4.933306769701629e-05,
+      "loss": 0.1585,
+      "step": 10120
+    },
+    {
+      "epoch": 0.8353471001757469,
+      "grad_norm": 1.0357364416122437,
+      "learning_rate": 4.9326453328003217e-05,
+      "loss": 0.187,
+      "step": 10140
+    },
+    {
+      "epoch": 0.8369947275922671,
+      "grad_norm": 0.6068626642227173,
+      "learning_rate": 4.931980676912252e-05,
+      "loss": 0.1971,
+      "step": 10160
+    },
+    {
+      "epoch": 0.8386423550087874,
+      "grad_norm": 3.377023220062256,
+      "learning_rate": 4.931312802916925e-05,
+      "loss": 0.1999,
+      "step": 10180
+    },
+    {
+      "epoch": 0.8402899824253075,
+      "grad_norm": 3.0354530811309814,
+      "learning_rate": 4.9306417116980996e-05,
+      "loss": 0.2008,
+      "step": 10200
+    },
+    {
+      "epoch": 0.8419376098418277,
+      "grad_norm": 2.531010150909424,
+      "learning_rate": 4.929967404143796e-05,
+      "loss": 0.1688,
+      "step": 10220
+    },
+    {
+      "epoch": 0.843585237258348,
+      "grad_norm": 0.526393711566925,
+      "learning_rate": 4.929289881146286e-05,
+      "loss": 0.1844,
+      "step": 10240
+    },
+    {
+      "epoch": 0.8452328646748682,
+      "grad_norm": 0.9323037266731262,
+      "learning_rate": 4.9286091436021015e-05,
+      "loss": 0.1968,
+      "step": 10260
+    },
+    {
+      "epoch": 0.8468804920913884,
+      "grad_norm": 1.0275589227676392,
+      "learning_rate": 4.927925192412024e-05,
+      "loss": 0.182,
+      "step": 10280
+    },
+    {
+      "epoch": 0.8485281195079086,
+      "grad_norm": 4.738681793212891,
+      "learning_rate": 4.927238028481089e-05,
+      "loss": 0.214,
+      "step": 10300
+    },
+    {
+      "epoch": 0.8501757469244289,
+      "grad_norm": 1.584230899810791,
+      "learning_rate": 4.926547652718583e-05,
+      "loss": 0.1582,
+      "step": 10320
+    },
+    {
+      "epoch": 0.851823374340949,
+      "grad_norm": 0.9227916598320007,
+      "learning_rate": 4.9258540660380434e-05,
+      "loss": 0.1921,
+      "step": 10340
+    },
+    {
+      "epoch": 0.8534710017574693,
+      "grad_norm": 7.2498250007629395,
+      "learning_rate": 4.925157269357254e-05,
+      "loss": 0.2197,
+      "step": 10360
+    },
+    {
+      "epoch": 0.8551186291739895,
+      "grad_norm": 0.8496150970458984,
+      "learning_rate": 4.924457263598248e-05,
+      "loss": 0.1798,
+      "step": 10380
+    },
+    {
+      "epoch": 0.8567662565905096,
+      "grad_norm": 9.715910911560059,
+      "learning_rate": 4.9237540496873064e-05,
+      "loss": 0.2241,
+      "step": 10400
+    },
+    {
+      "epoch": 0.8584138840070299,
+      "grad_norm": 2.707458257675171,
+      "learning_rate": 4.923047628554952e-05,
+      "loss": 0.1581,
+      "step": 10420
+    },
+    {
+      "epoch": 0.8600615114235501,
+      "grad_norm": 0.9239126443862915,
+      "learning_rate": 4.9223380011359544e-05,
+      "loss": 0.1964,
+      "step": 10440
+    },
+    {
+      "epoch": 0.8617091388400703,
+      "grad_norm": 1.0885734558105469,
+      "learning_rate": 4.9216251683693246e-05,
+      "loss": 0.2139,
+      "step": 10460
+    },
+    {
+      "epoch": 0.8633567662565905,
+      "grad_norm": 0.8583139777183533,
+      "learning_rate": 4.920909131198315e-05,
+      "loss": 0.2007,
+      "step": 10480
+    },
+    {
+      "epoch": 0.8650043936731108,
+      "grad_norm": 2.6485893726348877,
+      "learning_rate": 4.920189890570419e-05,
+      "loss": 0.2065,
+      "step": 10500
+    },
+    {
+      "epoch": 0.8666520210896309,
+      "grad_norm": 1.3636205196380615,
+      "learning_rate": 4.919467447437368e-05,
+      "loss": 0.1583,
+      "step": 10520
+    },
+    {
+      "epoch": 0.8682996485061512,
+      "grad_norm": 1.2751201391220093,
+      "learning_rate": 4.918741802755132e-05,
+      "loss": 0.2044,
+      "step": 10540
+    },
+    {
+      "epoch": 0.8699472759226714,
+      "grad_norm": 0.8484062552452087,
+      "learning_rate": 4.918012957483916e-05,
+      "loss": 0.192,
+      "step": 10560
+    },
+    {
+      "epoch": 0.8715949033391915,
+      "grad_norm": 0.5631062984466553,
+      "learning_rate": 4.917280912588163e-05,
+      "loss": 0.1997,
+      "step": 10580
+    },
+    {
+      "epoch": 0.8732425307557118,
+      "grad_norm": 2.789642572402954,
+      "learning_rate": 4.916545669036545e-05,
+      "loss": 0.2309,
+      "step": 10600
+    },
+    {
+      "epoch": 0.874890158172232,
+      "grad_norm": 7.149659156799316,
+      "learning_rate": 4.915807227801973e-05,
+      "loss": 0.1559,
+      "step": 10620
+    },
+    {
+      "epoch": 0.8765377855887522,
+      "grad_norm": 1.080868124961853,
+      "learning_rate": 4.915065589861584e-05,
+      "loss": 0.2,
+      "step": 10640
+    },
+    {
+      "epoch": 0.8781854130052724,
+      "grad_norm": 0.49652764201164246,
+      "learning_rate": 4.914320756196748e-05,
+      "loss": 0.2034,
+      "step": 10660
+    },
+    {
+      "epoch": 0.8798330404217927,
+      "grad_norm": 0.9888033270835876,
+      "learning_rate": 4.913572727793062e-05,
+      "loss": 0.1952,
+      "step": 10680
+    },
+    {
+      "epoch": 0.8814806678383128,
+      "grad_norm": 13.734339714050293,
+      "learning_rate": 4.9128215056403507e-05,
+      "loss": 0.2042,
+      "step": 10700
+    },
+    {
+      "epoch": 0.883128295254833,
+      "grad_norm": 0.7586619257926941,
+      "learning_rate": 4.912067090732667e-05,
+      "loss": 0.1449,
+      "step": 10720
+    },
+    {
+      "epoch": 0.8847759226713533,
+      "grad_norm": 0.673252284526825,
+      "learning_rate": 4.911309484068285e-05,
+      "loss": 0.2138,
+      "step": 10740
+    },
+    {
+      "epoch": 0.8864235500878734,
+      "grad_norm": 1.2692677974700928,
+      "learning_rate": 4.910548686649706e-05,
+      "loss": 0.1994,
+      "step": 10760
+    },
+    {
+      "epoch": 0.8880711775043937,
+      "grad_norm": 0.7501096725463867,
+      "learning_rate": 4.9097846994836505e-05,
+      "loss": 0.1949,
+      "step": 10780
+    },
+    {
+      "epoch": 0.8897188049209139,
+      "grad_norm": 3.0218920707702637,
+      "learning_rate": 4.909017523581062e-05,
+      "loss": 0.2137,
+      "step": 10800
+    },
+    {
+      "epoch": 0.8913664323374341,
+      "grad_norm": 1.1619638204574585,
+      "learning_rate": 4.9082471599571015e-05,
+      "loss": 0.1509,
+      "step": 10820
+    },
+    {
+      "epoch": 0.8930140597539543,
+      "grad_norm": 1.3160550594329834,
+      "learning_rate": 4.907512362815835e-05,
+      "loss": 0.1879,
+      "step": 10840
+    },
+    {
+      "epoch": 0.8946616871704746,
+      "grad_norm": 2.0613820552825928,
+      "learning_rate": 4.9067357860710327e-05,
+      "loss": 0.2094,
+      "step": 10860
+    },
+    {
+      "epoch": 0.8963093145869947,
+      "grad_norm": 0.6547063589096069,
+      "learning_rate": 4.905956024624158e-05,
+      "loss": 0.1879,
+      "step": 10880
+    },
+    {
+      "epoch": 0.897956942003515,
+      "grad_norm": 4.540472507476807,
+      "learning_rate": 4.905173079507026e-05,
+      "loss": 0.2022,
+      "step": 10900
+    },
+    {
+      "epoch": 0.8996045694200352,
+      "grad_norm": 2.7642438411712646,
+      "learning_rate": 4.904386951755665e-05,
+      "loss": 0.1566,
+      "step": 10920
+    },
+    {
+      "epoch": 0.9012521968365553,
+      "grad_norm": 1.2250425815582275,
+      "learning_rate": 4.903597642410316e-05,
+      "loss": 0.2049,
+      "step": 10940
+    },
+    {
+      "epoch": 0.9028998242530756,
+      "grad_norm": 0.6964439749717712,
+      "learning_rate": 4.902805152515427e-05,
+      "loss": 0.2003,
+      "step": 10960
+    },
+    {
+      "epoch": 0.9045474516695958,
+      "grad_norm": 0.8881003260612488,
+      "learning_rate": 4.902049342086357e-05,
+      "loss": 0.1969,
+      "step": 10980
+    },
+    {
+      "epoch": 0.906195079086116,
+      "grad_norm": 3.828152656555176,
+      "learning_rate": 4.901250653139905e-05,
+      "loss": 0.2199,
+      "step": 11000
+    },
+    {
+      "epoch": 0.9078427065026362,
+      "grad_norm": 1.2089974880218506,
+      "learning_rate": 4.900448786749557e-05,
+      "loss": 0.157,
+      "step": 11020
+    },
+    {
+      "epoch": 0.9094903339191565,
+      "grad_norm": 1.148004412651062,
+      "learning_rate": 4.8996437439763784e-05,
+      "loss": 0.2013,
+      "step": 11040
+    },
+    {
+      "epoch": 0.9111379613356766,
+      "grad_norm": 1.6459652185440063,
+      "learning_rate": 4.89883552588564e-05,
+      "loss": 0.1928,
+      "step": 11060
+    },
+    {
+      "epoch": 0.9127855887521968,
+      "grad_norm": 0.7444145083427429,
+      "learning_rate": 4.898024133546811e-05,
+      "loss": 0.1965,
+      "step": 11080
+    },
+    {
+      "epoch": 0.9144332161687171,
+      "grad_norm": 7.0522260665893555,
+      "learning_rate": 4.897209568033564e-05,
+      "loss": 0.2181,
+      "step": 11100
+    },
+    {
+      "epoch": 0.9160808435852372,
+      "grad_norm": 1.5349088907241821,
+      "learning_rate": 4.896391830423768e-05,
+      "loss": 0.1575,
+      "step": 11120
+    },
+    {
+      "epoch": 0.9177284710017575,
+      "grad_norm": 0.5103209018707275,
+      "learning_rate": 4.895570921799491e-05,
+      "loss": 0.2057,
+      "step": 11140
+    },
+    {
+      "epoch": 0.9193760984182777,
+      "grad_norm": 0.4984442889690399,
+      "learning_rate": 4.8947468432469955e-05,
+      "loss": 0.1897,
+      "step": 11160
+    },
+    {
+      "epoch": 0.9210237258347979,
+      "grad_norm": 0.5621454119682312,
+      "learning_rate": 4.893919595856742e-05,
+      "loss": 0.1912,
+      "step": 11180
+    },
+    {
+      "epoch": 0.9226713532513181,
+      "grad_norm": 5.476040363311768,
+      "learning_rate": 4.8930891807233794e-05,
+      "loss": 0.2166,
+      "step": 11200
+    },
+    {
+      "epoch": 0.9243189806678384,
+      "grad_norm": 1.3259927034378052,
+      "learning_rate": 4.892255598945753e-05,
+      "loss": 0.1561,
+      "step": 11220
+    },
+    {
+      "epoch": 0.9259666080843585,
+      "grad_norm": 0.8870605230331421,
+      "learning_rate": 4.891418851626893e-05,
+      "loss": 0.2003,
+      "step": 11240
+    },
+    {
+      "epoch": 0.9276142355008787,
+      "grad_norm": 16.381240844726562,
+      "learning_rate": 4.890578939874025e-05,
+      "loss": 0.1916,
+      "step": 11260
+    },
+    {
+      "epoch": 0.929261862917399,
+      "grad_norm": 1.6785776615142822,
+      "learning_rate": 4.889735864798556e-05,
+      "loss": 0.1825,
+      "step": 11280
+    },
+    {
+      "epoch": 0.9309094903339191,
+      "grad_norm": 5.951013565063477,
+      "learning_rate": 4.8888896275160816e-05,
+      "loss": 0.2045,
+      "step": 11300
+    },
+    {
+      "epoch": 0.9325571177504394,
+      "grad_norm": 3.8519906997680664,
+      "learning_rate": 4.888040229146382e-05,
+      "loss": 0.1573,
+      "step": 11320
+    },
+    {
+      "epoch": 0.9342047451669596,
+      "grad_norm": 1.0654627084732056,
+      "learning_rate": 4.887187670813419e-05,
+      "loss": 0.223,
+      "step": 11340
+    },
+    {
+      "epoch": 0.9358523725834798,
+      "grad_norm": 1.6851961612701416,
+      "learning_rate": 4.8863319536453364e-05,
+      "loss": 0.2058,
+      "step": 11360
+    },
+    {
+      "epoch": 0.9375,
+      "grad_norm": 0.6880635023117065,
+      "learning_rate": 4.88547307877446e-05,
+      "loss": 0.1974,
+      "step": 11380
+    },
+    {
+      "epoch": 0.9391476274165202,
+      "grad_norm": 5.7095866203308105,
+      "learning_rate": 4.884611047337289e-05,
+      "loss": 0.2169,
+      "step": 11400
+    },
+    {
+      "epoch": 0.9407952548330404,
+      "grad_norm": 8.853272438049316,
+      "learning_rate": 4.883745860474505e-05,
+      "loss": 0.1584,
+      "step": 11420
+    },
+    {
+      "epoch": 0.9424428822495606,
+      "grad_norm": 1.7952899932861328,
+      "learning_rate": 4.882877519330961e-05,
+      "loss": 0.1937,
+      "step": 11440
+    },
+    {
+      "epoch": 0.9440905096660809,
+      "grad_norm": 5.514512538909912,
+      "learning_rate": 4.882006025055685e-05,
+      "loss": 0.1989,
+      "step": 11460
+    },
+    {
+      "epoch": 0.945738137082601,
+      "grad_norm": 0.776798665523529,
+      "learning_rate": 4.88113137880188e-05,
+      "loss": 0.1862,
+      "step": 11480
+    },
+    {
+      "epoch": 0.9473857644991213,
+      "grad_norm": 2.824843645095825,
+      "learning_rate": 4.880253581726916e-05,
+      "loss": 0.2168,
+      "step": 11500
+    },
+    {
+      "epoch": 0.9490333919156415,
+      "grad_norm": 0.9011817574501038,
+      "learning_rate": 4.879372634992335e-05,
+      "loss": 0.1601,
+      "step": 11520
+    },
+    {
+      "epoch": 0.9506810193321616,
+      "grad_norm": 5.290555477142334,
+      "learning_rate": 4.8784885397638445e-05,
+      "loss": 0.1892,
+      "step": 11540
+    },
+    {
+      "epoch": 0.9523286467486819,
+      "grad_norm": 1.164330005645752,
+      "learning_rate": 4.87760129721132e-05,
+      "loss": 0.2112,
+      "step": 11560
+    },
+    {
+      "epoch": 0.9539762741652021,
+      "grad_norm": 0.8354160189628601,
+      "learning_rate": 4.876710908508801e-05,
+      "loss": 0.2176,
+      "step": 11580
+    },
+    {
+      "epoch": 0.9556239015817223,
+      "grad_norm": 3.980334758758545,
+      "learning_rate": 4.8758173748344904e-05,
+      "loss": 0.2484,
+      "step": 11600
+    },
+    {
+      "epoch": 0.9572715289982425,
+      "grad_norm": 1.7610282897949219,
+      "learning_rate": 4.874920697370753e-05,
+      "loss": 0.1618,
+      "step": 11620
+    },
+    {
+      "epoch": 0.9589191564147628,
+      "grad_norm": 1.3336617946624756,
+      "learning_rate": 4.874020877304113e-05,
+      "loss": 0.1895,
+      "step": 11640
+    },
+    {
+      "epoch": 0.9605667838312829,
+      "grad_norm": 0.9175180792808533,
+      "learning_rate": 4.873117915825252e-05,
+      "loss": 0.1945,
+      "step": 11660
+    },
+    {
+      "epoch": 0.9622144112478032,
+      "grad_norm": 0.49205484986305237,
+      "learning_rate": 4.8722118141290105e-05,
+      "loss": 0.1935,
+      "step": 11680
+    },
+    {
+      "epoch": 0.9638620386643234,
+      "grad_norm": 3.554715394973755,
+      "learning_rate": 4.871302573414384e-05,
+      "loss": 0.2135,
+      "step": 11700
+    },
+    {
+      "epoch": 0.9655096660808435,
+      "grad_norm": 1.0651150941848755,
+      "learning_rate": 4.8703901948845205e-05,
+      "loss": 0.1564,
+      "step": 11720
+    },
+    {
+      "epoch": 0.9671572934973638,
+      "grad_norm": 1.88383150100708,
+      "learning_rate": 4.869474679746721e-05,
+      "loss": 0.1994,
+      "step": 11740
+    },
+    {
+      "epoch": 0.968804920913884,
+      "grad_norm": 0.6271897554397583,
+      "learning_rate": 4.868556029212435e-05,
+      "loss": 0.2043,
+      "step": 11760
+    },
+    {
+      "epoch": 0.9704525483304042,
+      "grad_norm": 0.616521954536438,
+      "learning_rate": 4.867634244497265e-05,
+      "loss": 0.2059,
+      "step": 11780
+    },
+    {
+      "epoch": 0.9721001757469244,
+      "grad_norm": 3.2947463989257812,
+      "learning_rate": 4.8667093268209575e-05,
+      "loss": 0.205,
+      "step": 11800
+    },
+    {
+      "epoch": 0.9737478031634447,
+      "grad_norm": 0.7675888538360596,
+      "learning_rate": 4.865781277407405e-05,
+      "loss": 0.1436,
+      "step": 11820
+    },
+    {
+      "epoch": 0.9753954305799648,
+      "grad_norm": 1.3581780195236206,
+      "learning_rate": 4.8648500974846445e-05,
+      "loss": 0.1964,
+      "step": 11840
+    },
+    {
+      "epoch": 0.977043057996485,
+      "grad_norm": 0.7175058722496033,
+      "learning_rate": 4.8639157882848566e-05,
+      "loss": 0.197,
+      "step": 11860
+    },
+    {
+      "epoch": 0.9786906854130053,
+      "grad_norm": 0.7383044362068176,
+      "learning_rate": 4.8629783510443616e-05,
+      "loss": 0.1904,
+      "step": 11880
+    },
+    {
+      "epoch": 0.9803383128295254,
+      "grad_norm": 5.0527777671813965,
+      "learning_rate": 4.8620377870036184e-05,
+      "loss": 0.2117,
+      "step": 11900
+    },
+    {
+      "epoch": 0.9819859402460457,
+      "grad_norm": 1.5865782499313354,
+      "learning_rate": 4.861094097407224e-05,
+      "loss": 0.1515,
+      "step": 11920
+    },
+    {
+      "epoch": 0.9836335676625659,
+      "grad_norm": 0.5650818943977356,
+      "learning_rate": 4.860147283503912e-05,
+      "loss": 0.1926,
+      "step": 11940
+    },
+    {
+      "epoch": 0.9852811950790861,
+      "grad_norm": 1.3663684129714966,
+      "learning_rate": 4.8591973465465493e-05,
+      "loss": 0.2066,
+      "step": 11960
+    },
+    {
+      "epoch": 0.9869288224956063,
+      "grad_norm": 1.2778149843215942,
+      "learning_rate": 4.8582442877921344e-05,
+      "loss": 0.2022,
+      "step": 11980
+    },
+    {
+      "epoch": 0.9885764499121266,
+      "grad_norm": 4.921967029571533,
+      "learning_rate": 4.8572881085017996e-05,
+      "loss": 0.2158,
+      "step": 12000
+    },
+    {
+      "epoch": 0.9885764499121266,
+      "eval_loss": 0.5375325679779053,
+      "eval_runtime": 252.7679,
+      "eval_samples_per_second": 86.202,
+      "eval_steps_per_second": 21.553,
+      "eval_wer": 0.2286638878235724,
+      "step": 12000
+    },
+    {
+      "epoch": 0.9902240773286467,
+      "grad_norm": 1.044043779373169,
+      "learning_rate": 4.856328809940804e-05,
+      "loss": 0.1584,
+      "step": 12020
+    },
+    {
+      "epoch": 0.991871704745167,
+      "grad_norm": 2.4731016159057617,
+      "learning_rate": 4.855366393378535e-05,
+      "loss": 0.2071,
+      "step": 12040
+    },
+    {
+      "epoch": 0.9935193321616872,
+      "grad_norm": 0.6640995144844055,
+      "learning_rate": 4.854400860088506e-05,
+      "loss": 0.2119,
+      "step": 12060
+    },
+    {
+      "epoch": 0.9951669595782073,
+      "grad_norm": 0.43121984601020813,
+      "learning_rate": 4.853432211348356e-05,
+      "loss": 0.1918,
+      "step": 12080
+    },
+    {
+      "epoch": 0.9968145869947276,
+      "grad_norm": 5.402212619781494,
+      "learning_rate": 4.8524604484398414e-05,
+      "loss": 0.2181,
+      "step": 12100
+    },
+    {
+      "epoch": 0.9984622144112478,
+      "grad_norm": 1.417004108428955,
+      "learning_rate": 4.851485572648846e-05,
+      "loss": 0.1493,
+      "step": 12120
+    },
+    {
+      "epoch": 1.000109841827768,
+      "grad_norm": 0.9582122564315796,
+      "learning_rate": 4.8505075852653684e-05,
+      "loss": 0.207,
+      "step": 12140
+    },
+    {
+      "epoch": 1.0017574692442883,
+      "grad_norm": 1.74813973903656,
+      "learning_rate": 4.8495264875835264e-05,
+      "loss": 0.1468,
+      "step": 12160
+    },
+    {
+      "epoch": 1.0034050966608083,
+      "grad_norm": 3.2926461696624756,
+      "learning_rate": 4.8485422809015535e-05,
+      "loss": 0.2009,
+      "step": 12180
+    },
+    {
+      "epoch": 1.0050527240773286,
+      "grad_norm": 2.141998291015625,
+      "learning_rate": 4.8475549665217956e-05,
+      "loss": 0.1878,
+      "step": 12200
+    },
+    {
+      "epoch": 1.0067003514938488,
+      "grad_norm": 1.7473055124282837,
+      "learning_rate": 4.846564545750712e-05,
+      "loss": 0.1916,
+      "step": 12220
+    },
+    {
+      "epoch": 1.008347978910369,
+      "grad_norm": 0.20854823291301727,
+      "learning_rate": 4.845571019898874e-05,
+      "loss": 0.19,
+      "step": 12240
+    },
+    {
+      "epoch": 1.0099956063268893,
+      "grad_norm": 1.6081265211105347,
+      "learning_rate": 4.844574390280958e-05,
+      "loss": 0.1623,
+      "step": 12260
+    },
+    {
+      "epoch": 1.0116432337434096,
+      "grad_norm": 2.013803482055664,
+      "learning_rate": 4.843574658215751e-05,
+      "loss": 0.1916,
+      "step": 12280
+    },
+    {
+      "epoch": 1.0132908611599296,
+      "grad_norm": 1.2559956312179565,
+      "learning_rate": 4.842571825026143e-05,
+      "loss": 0.1759,
+      "step": 12300
+    },
+    {
+      "epoch": 1.0149384885764499,
+      "grad_norm": 1.103947401046753,
+      "learning_rate": 4.8415658920391296e-05,
+      "loss": 0.202,
+      "step": 12320
+    },
+    {
+      "epoch": 1.01658611599297,
+      "grad_norm": 0.5012603998184204,
+      "learning_rate": 4.840556860585805e-05,
+      "loss": 0.198,
+      "step": 12340
+    },
+    {
+      "epoch": 1.0182337434094904,
+      "grad_norm": 1.4309896230697632,
+      "learning_rate": 4.839544732001368e-05,
+      "loss": 0.163,
+      "step": 12360
+    },
+    {
+      "epoch": 1.0198813708260106,
+      "grad_norm": 0.6756483912467957,
+      "learning_rate": 4.838529507625111e-05,
+      "loss": 0.1831,
+      "step": 12380
+    },
+    {
+      "epoch": 1.0215289982425309,
+      "grad_norm": 1.141714096069336,
+      "learning_rate": 4.8375111888004263e-05,
+      "loss": 0.1888,
+      "step": 12400
+    },
+    {
+      "epoch": 1.0231766256590509,
+      "grad_norm": 0.471711665391922,
+      "learning_rate": 4.836489776874799e-05,
+      "loss": 0.1892,
+      "step": 12420
+    },
+    {
+      "epoch": 1.0248242530755711,
+      "grad_norm": 0.28789934515953064,
+      "learning_rate": 4.835465273199807e-05,
+      "loss": 0.1795,
+      "step": 12440
+    },
+    {
+      "epoch": 1.0264718804920914,
+      "grad_norm": 1.1327400207519531,
+      "learning_rate": 4.834437679131121e-05,
+      "loss": 0.169,
+      "step": 12460
+    },
+    {
+      "epoch": 1.0281195079086116,
+      "grad_norm": 1.0975223779678345,
+      "learning_rate": 4.8334069960285e-05,
+      "loss": 0.1875,
+      "step": 12480
+    },
+    {
+      "epoch": 1.0297671353251319,
+      "grad_norm": 0.5748992562294006,
+      "learning_rate": 4.832373225255791e-05,
+      "loss": 0.1888,
+      "step": 12500
+    },
+    {
+      "epoch": 1.0314147627416521,
+      "grad_norm": 0.5393682718276978,
+      "learning_rate": 4.831336368180927e-05,
+      "loss": 0.1808,
+      "step": 12520
+    },
+    {
+      "epoch": 1.0330623901581721,
+      "grad_norm": 0.22509616613388062,
+      "learning_rate": 4.830296426175922e-05,
+      "loss": 0.1803,
+      "step": 12540
+    },
+    {
+      "epoch": 1.0347100175746924,
+      "grad_norm": 1.0403659343719482,
+      "learning_rate": 4.829253400616876e-05,
+      "loss": 0.1594,
+      "step": 12560
+    },
+    {
+      "epoch": 1.0363576449912126,
+      "grad_norm": 1.5973280668258667,
+      "learning_rate": 4.828207292883968e-05,
+      "loss": 0.1907,
+      "step": 12580
+    },
+    {
+      "epoch": 1.0380052724077329,
+      "grad_norm": 0.9520076513290405,
+      "learning_rate": 4.8271581043614555e-05,
+      "loss": 0.1852,
+      "step": 12600
+    },
+    {
+      "epoch": 1.0396528998242531,
+      "grad_norm": 3.2169032096862793,
+      "learning_rate": 4.826105836437672e-05,
+      "loss": 0.1796,
+      "step": 12620
+    },
+    {
+      "epoch": 1.0413005272407734,
+      "grad_norm": 0.3701666295528412,
+      "learning_rate": 4.825050490505025e-05,
+      "loss": 0.1894,
+      "step": 12640
+    },
+    {
+      "epoch": 1.0429481546572934,
+      "grad_norm": 1.573367953300476,
+      "learning_rate": 4.823992067959998e-05,
+      "loss": 0.1565,
+      "step": 12660
+    },
+    {
+      "epoch": 1.0445957820738137,
+      "grad_norm": 0.6586406230926514,
+      "learning_rate": 4.8229305702031426e-05,
+      "loss": 0.183,
+      "step": 12680
+    },
+    {
+      "epoch": 1.046243409490334,
+      "grad_norm": 1.1278094053268433,
+      "learning_rate": 4.821865998639081e-05,
+      "loss": 0.175,
+      "step": 12700
+    },
+    {
+      "epoch": 1.0478910369068541,
+      "grad_norm": 0.9823224544525146,
+      "learning_rate": 4.820798354676504e-05,
+      "loss": 0.1939,
+      "step": 12720
+    },
+    {
+      "epoch": 1.0495386643233744,
+      "grad_norm": 0.6462484002113342,
+      "learning_rate": 4.819727639728164e-05,
+      "loss": 0.1652,
+      "step": 12740
+    },
+    {
+      "epoch": 1.0511862917398946,
+      "grad_norm": 1.3370243310928345,
+      "learning_rate": 4.818653855210882e-05,
+      "loss": 0.1605,
+      "step": 12760
+    },
+    {
+      "epoch": 1.0528339191564147,
+      "grad_norm": 0.7861006259918213,
+      "learning_rate": 4.817577002545538e-05,
+      "loss": 0.1858,
+      "step": 12780
+    },
+    {
+      "epoch": 1.054481546572935,
+      "grad_norm": 2.107189655303955,
+      "learning_rate": 4.816497083157071e-05,
+      "loss": 0.1933,
+      "step": 12800
+    },
+    {
+      "epoch": 1.0561291739894552,
+      "grad_norm": 0.9943327307701111,
+      "learning_rate": 4.81541409847448e-05,
+      "loss": 0.1949,
+      "step": 12820
+    },
+    {
+      "epoch": 1.0577768014059754,
+      "grad_norm": 0.5097872018814087,
+      "learning_rate": 4.8143280499308205e-05,
+      "loss": 0.1879,
+      "step": 12840
+    },
+    {
+      "epoch": 1.0594244288224957,
+      "grad_norm": 2.121750593185425,
+      "learning_rate": 4.8132389389631995e-05,
+      "loss": 0.1616,
+      "step": 12860
+    },
+    {
+      "epoch": 1.061072056239016,
+      "grad_norm": 0.86652010679245,
+      "learning_rate": 4.81214676701278e-05,
+      "loss": 0.2113,
+      "step": 12880
+    },
+    {
+      "epoch": 1.062719683655536,
+      "grad_norm": 2.3862969875335693,
+      "learning_rate": 4.811051535524772e-05,
+      "loss": 0.1904,
+      "step": 12900
+    },
+    {
+      "epoch": 1.0643673110720562,
+      "grad_norm": 0.5471189618110657,
+      "learning_rate": 4.8099532459484356e-05,
+      "loss": 0.182,
+      "step": 12920
+    },
+    {
+      "epoch": 1.0660149384885764,
+      "grad_norm": 1.5992333889007568,
+      "learning_rate": 4.8088518997370794e-05,
+      "loss": 0.1821,
+      "step": 12940
+    },
+    {
+      "epoch": 1.0676625659050967,
+      "grad_norm": 0.6893438696861267,
+      "learning_rate": 4.807747498348052e-05,
+      "loss": 0.1648,
+      "step": 12960
+    },
+    {
+      "epoch": 1.069310193321617,
+      "grad_norm": 4.395170211791992,
+      "learning_rate": 4.806640043242748e-05,
+      "loss": 0.1759,
+      "step": 12980
+    },
+    {
+      "epoch": 1.0709578207381372,
+      "grad_norm": 0.5166418552398682,
+      "learning_rate": 4.805529535886605e-05,
+      "loss": 0.1956,
+      "step": 13000
+    },
+    {
+      "epoch": 1.0726054481546572,
+      "grad_norm": 0.8624104261398315,
+      "learning_rate": 4.804415977749094e-05,
+      "loss": 0.1839,
+      "step": 13020
+    },
+    {
+      "epoch": 1.0742530755711774,
+      "grad_norm": 0.6681638360023499,
+      "learning_rate": 4.803299370303725e-05,
+      "loss": 0.2012,
+      "step": 13040
+    },
+    {
+      "epoch": 1.0759007029876977,
+      "grad_norm": 1.1625574827194214,
+      "learning_rate": 4.802179715028047e-05,
+      "loss": 0.165,
+      "step": 13060
+    },
+    {
+      "epoch": 1.077548330404218,
+      "grad_norm": 3.3364834785461426,
+      "learning_rate": 4.801057013403636e-05,
+      "loss": 0.1905,
+      "step": 13080
+    },
+    {
+      "epoch": 1.0791959578207382,
+      "grad_norm": 0.9237430691719055,
+      "learning_rate": 4.799931266916103e-05,
+      "loss": 0.1923,
+      "step": 13100
+    },
+    {
+      "epoch": 1.0808435852372584,
+      "grad_norm": 2.341661214828491,
+      "learning_rate": 4.798858988805215e-05,
+      "loss": 0.1918,
+      "step": 13120
+    },
+    {
+      "epoch": 1.0824912126537785,
+      "grad_norm": 0.2139461189508438,
+      "learning_rate": 4.797727309122835e-05,
+      "loss": 0.2039,
+      "step": 13140
+    },
+    {
+      "epoch": 1.0841388400702987,
+      "grad_norm": 1.3084605932235718,
+      "learning_rate": 4.796592588983348e-05,
+      "loss": 0.1808,
+      "step": 13160
+    },
+    {
+      "epoch": 1.085786467486819,
+      "grad_norm": 0.5879707336425781,
+      "learning_rate": 4.7954548298882685e-05,
+      "loss": 0.1879,
+      "step": 13180
+    },
+    {
+      "epoch": 1.0874340949033392,
+      "grad_norm": 9.557912826538086,
+      "learning_rate": 4.794314033343129e-05,
+      "loss": 0.1667,
+      "step": 13200
+    },
+    {
+      "epoch": 1.0890817223198594,
+      "grad_norm": 2.7290220260620117,
+      "learning_rate": 4.793170200857485e-05,
+      "loss": 0.1848,
+      "step": 13220
+    },
+    {
+      "epoch": 1.0907293497363797,
+      "grad_norm": 0.4870263338088989,
+      "learning_rate": 4.792023333944907e-05,
+      "loss": 0.183,
+      "step": 13240
+    },
+    {
+      "epoch": 1.0923769771528997,
+      "grad_norm": 1.1112992763519287,
+      "learning_rate": 4.790873434122982e-05,
+      "loss": 0.1628,
+      "step": 13260
+    },
+    {
+      "epoch": 1.09402460456942,
+      "grad_norm": 1.2795594930648804,
+      "learning_rate": 4.789720502913309e-05,
+      "loss": 0.1687,
+      "step": 13280
+    },
+    {
+      "epoch": 1.0956722319859402,
+      "grad_norm": 1.0361709594726562,
+      "learning_rate": 4.788564541841498e-05,
+      "loss": 0.196,
+      "step": 13300
+    },
+    {
+      "epoch": 1.0973198594024605,
+      "grad_norm": 1.4173661470413208,
+      "learning_rate": 4.787405552437171e-05,
+      "loss": 0.191,
+      "step": 13320
+    },
+    {
+      "epoch": 1.0989674868189807,
+      "grad_norm": 3.3120641708374023,
+      "learning_rate": 4.786243536233954e-05,
+      "loss": 0.1881,
+      "step": 13340
+    },
+    {
+      "epoch": 1.100615114235501,
+      "grad_norm": 1.3123334646224976,
+      "learning_rate": 4.785078494769481e-05,
+      "loss": 0.1701,
+      "step": 13360
+    },
+    {
+      "epoch": 1.1022627416520212,
+      "grad_norm": 1.0791329145431519,
+      "learning_rate": 4.783910429585386e-05,
+      "loss": 0.1883,
+      "step": 13380
+    },
+    {
+      "epoch": 1.1039103690685412,
+      "grad_norm": 4.972422122955322,
+      "learning_rate": 4.782739342227308e-05,
+      "loss": 0.1977,
+      "step": 13400
+    },
+    {
+      "epoch": 1.1055579964850615,
+      "grad_norm": 1.7628355026245117,
+      "learning_rate": 4.781565234244881e-05,
+      "loss": 0.1828,
+      "step": 13420
+    },
+    {
+      "epoch": 1.1072056239015817,
+      "grad_norm": 0.7487472295761108,
+      "learning_rate": 4.7803881071917386e-05,
+      "loss": 0.1903,
+      "step": 13440
+    },
+    {
+      "epoch": 1.108853251318102,
+      "grad_norm": 1.0923112630844116,
+      "learning_rate": 4.77920796262551e-05,
+      "loss": 0.1597,
+      "step": 13460
+    },
+    {
+      "epoch": 1.1105008787346222,
+      "grad_norm": 2.634671211242676,
+      "learning_rate": 4.778024802107814e-05,
+      "loss": 0.201,
+      "step": 13480
+    },
+    {
+      "epoch": 1.1121485061511422,
+      "grad_norm": 0.6952950358390808,
+      "learning_rate": 4.7768386272042646e-05,
+      "loss": 0.1887,
+      "step": 13500
+    },
+    {
+      "epoch": 1.1137961335676625,
+      "grad_norm": 1.2554984092712402,
+      "learning_rate": 4.77564943948446e-05,
+      "loss": 0.1862,
+      "step": 13520
+    },
+    {
+      "epoch": 1.1154437609841827,
+      "grad_norm": 0.42830732464790344,
+      "learning_rate": 4.774457240521989e-05,
+      "loss": 0.1892,
+      "step": 13540
+    },
+    {
+      "epoch": 1.117091388400703,
+      "grad_norm": 1.1879973411560059,
+      "learning_rate": 4.7732620318944224e-05,
+      "loss": 0.1576,
+      "step": 13560
+    },
+    {
+      "epoch": 1.1187390158172232,
+      "grad_norm": 0.9063071012496948,
+      "learning_rate": 4.772063815183315e-05,
+      "loss": 0.1941,
+      "step": 13580
+    },
+    {
+      "epoch": 1.1203866432337435,
+      "grad_norm": 1.071677803993225,
+      "learning_rate": 4.770862591974202e-05,
+      "loss": 0.1779,
+      "step": 13600
+    },
+    {
+      "epoch": 1.1220342706502637,
+      "grad_norm": 1.2179434299468994,
+      "learning_rate": 4.769658363856595e-05,
+      "loss": 0.1838,
+      "step": 13620
+    },
+    {
+      "epoch": 1.1236818980667838,
+      "grad_norm": 0.8977575898170471,
+      "learning_rate": 4.768451132423985e-05,
+      "loss": 0.2017,
+      "step": 13640
+    },
+    {
+      "epoch": 1.125329525483304,
+      "grad_norm": 1.6982157230377197,
+      "learning_rate": 4.767240899273835e-05,
+      "loss": 0.1608,
+      "step": 13660
+    },
+    {
+      "epoch": 1.1269771528998243,
+      "grad_norm": 0.7634503245353699,
+      "learning_rate": 4.7660276660075804e-05,
+      "loss": 0.1718,
+      "step": 13680
+    },
+    {
+      "epoch": 1.1286247803163445,
+      "grad_norm": 0.8203310966491699,
+      "learning_rate": 4.764811434230626e-05,
+      "loss": 0.1831,
+      "step": 13700
+    },
+    {
+      "epoch": 1.1302724077328647,
+      "grad_norm": 0.9817870259284973,
+      "learning_rate": 4.763592205552345e-05,
+      "loss": 0.1858,
+      "step": 13720
+    },
+    {
+      "epoch": 1.1319200351493848,
+      "grad_norm": 0.333933562040329,
+      "learning_rate": 4.762369981586077e-05,
+      "loss": 0.1785,
+      "step": 13740
+    },
+    {
+      "epoch": 1.133567662565905,
+      "grad_norm": 1.1885998249053955,
+      "learning_rate": 4.761144763949124e-05,
+      "loss": 0.1508,
+      "step": 13760
+    },
+    {
+      "epoch": 1.1352152899824253,
+      "grad_norm": 4.4533820152282715,
+      "learning_rate": 4.759916554262749e-05,
+      "loss": 0.1855,
+      "step": 13780
+    },
+    {
+      "epoch": 1.1368629173989455,
+      "grad_norm": 0.6248905062675476,
+      "learning_rate": 4.7586853541521744e-05,
+      "loss": 0.1938,
+      "step": 13800
+    },
+    {
+      "epoch": 1.1385105448154658,
+      "grad_norm": 0.7727264761924744,
+      "learning_rate": 4.757451165246583e-05,
+      "loss": 0.1809,
+      "step": 13820
+    },
+    {
+      "epoch": 1.140158172231986,
+      "grad_norm": 0.3115682005882263,
+      "learning_rate": 4.7562139891791046e-05,
+      "loss": 0.188,
+      "step": 13840
+    },
+    {
+      "epoch": 1.1418057996485063,
+      "grad_norm": 3.9221365451812744,
+      "learning_rate": 4.754973827586832e-05,
+      "loss": 0.1556,
+      "step": 13860
+    },
+    {
+      "epoch": 1.1434534270650263,
+      "grad_norm": 0.8228521347045898,
+      "learning_rate": 4.7537306821108e-05,
+      "loss": 0.185,
+      "step": 13880
+    },
+    {
+      "epoch": 1.1451010544815465,
+      "grad_norm": 3.115772247314453,
+      "learning_rate": 4.752484554395995e-05,
+      "loss": 0.1786,
+      "step": 13900
+    },
+    {
+      "epoch": 1.1467486818980668,
+      "grad_norm": 0.7074133157730103,
+      "learning_rate": 4.7512354460913524e-05,
+      "loss": 0.175,
+      "step": 13920
+    },
+    {
+      "epoch": 1.148396309314587,
+      "grad_norm": 0.4195045828819275,
+      "learning_rate": 4.7499833588497464e-05,
+      "loss": 0.1868,
+      "step": 13940
+    },
+    {
+      "epoch": 1.1500439367311073,
+      "grad_norm": 1.2703948020935059,
+      "learning_rate": 4.7487282943279965e-05,
+      "loss": 0.1517,
+      "step": 13960
+    },
+    {
+      "epoch": 1.1516915641476273,
+      "grad_norm": 2.4614219665527344,
+      "learning_rate": 4.747470254186862e-05,
+      "loss": 0.1926,
+      "step": 13980
+    },
+    {
+      "epoch": 1.1533391915641475,
+      "grad_norm": 2.0017571449279785,
+      "learning_rate": 4.746209240091038e-05,
+      "loss": 0.1852,
+      "step": 14000
+    },
+    {
+      "epoch": 1.1533391915641475,
+      "eval_loss": 0.5445581078529358,
+      "eval_runtime": 260.7987,
+      "eval_samples_per_second": 83.547,
+      "eval_steps_per_second": 20.89,
+      "eval_wer": 0.22884024824829308,
+      "step": 14000
+    },
+    {
+      "epoch": 1.1550417398945518,
+      "grad_norm": 0.7891082763671875,
+      "learning_rate": 4.744945253709156e-05,
+      "loss": 0.1683,
+      "step": 14020
+    },
+    {
+      "epoch": 1.156689367311072,
+      "grad_norm": 1.2621268033981323,
+      "learning_rate": 4.743678296713779e-05,
+      "loss": 0.2037,
+      "step": 14040
+    },
+    {
+      "epoch": 1.1583369947275923,
+      "grad_norm": 1.7061976194381714,
+      "learning_rate": 4.7424083707814035e-05,
+      "loss": 0.18,
+      "step": 14060
+    },
+    {
+      "epoch": 1.1599846221441124,
+      "grad_norm": 4.809337615966797,
+      "learning_rate": 4.7411354775924523e-05,
+      "loss": 0.1772,
+      "step": 14080
+    },
+    {
+      "epoch": 1.1616322495606326,
+      "grad_norm": 4.00537109375,
+      "learning_rate": 4.739859618831276e-05,
+      "loss": 0.2029,
+      "step": 14100
+    },
+    {
+      "epoch": 1.1632798769771528,
+      "grad_norm": 3.184988498687744,
+      "learning_rate": 4.7385807961861486e-05,
+      "loss": 0.1857,
+      "step": 14120
+    },
+    {
+      "epoch": 1.164927504393673,
+      "grad_norm": 0.24155808985233307,
+      "learning_rate": 4.737299011349265e-05,
+      "loss": 0.1808,
+      "step": 14140
+    },
+    {
+      "epoch": 1.1665751318101933,
+      "grad_norm": 1.2904281616210938,
+      "learning_rate": 4.736014266016743e-05,
+      "loss": 0.1728,
+      "step": 14160
+    },
+    {
+      "epoch": 1.1682227592267136,
+      "grad_norm": 0.8972460031509399,
+      "learning_rate": 4.734726561888616e-05,
+      "loss": 0.1712,
+      "step": 14180
+    },
+    {
+      "epoch": 1.1698703866432338,
+      "grad_norm": 0.48299163579940796,
+      "learning_rate": 4.7334359006688314e-05,
+      "loss": 0.1872,
+      "step": 14200
+    },
+    {
+      "epoch": 1.1715180140597539,
+      "grad_norm": 1.942762017250061,
+      "learning_rate": 4.732142284065251e-05,
+      "loss": 0.1742,
+      "step": 14220
+    },
+    {
+      "epoch": 1.1731656414762741,
+      "grad_norm": 0.8175634741783142,
+      "learning_rate": 4.730845713789647e-05,
+      "loss": 0.175,
+      "step": 14240
+    },
+    {
+      "epoch": 1.1748132688927944,
+      "grad_norm": 3.179103136062622,
+      "learning_rate": 4.7295461915577e-05,
+      "loss": 0.1544,
+      "step": 14260
+    },
+    {
+      "epoch": 1.1764608963093146,
+      "grad_norm": 0.7264633178710938,
+      "learning_rate": 4.7282437190889975e-05,
+      "loss": 0.1854,
+      "step": 14280
+    },
+    {
+      "epoch": 1.1781085237258349,
+      "grad_norm": 1.372662901878357,
+      "learning_rate": 4.7269382981070296e-05,
+      "loss": 0.193,
+      "step": 14300
+    },
+    {
+      "epoch": 1.179756151142355,
+      "grad_norm": 1.8890334367752075,
+      "learning_rate": 4.7256299303391895e-05,
+      "loss": 0.1908,
+      "step": 14320
+    },
+    {
+      "epoch": 1.1814037785588751,
+      "grad_norm": 0.4149373769760132,
+      "learning_rate": 4.724318617516768e-05,
+      "loss": 0.1838,
+      "step": 14340
+    },
+    {
+      "epoch": 1.1830514059753954,
+      "grad_norm": 2.029543876647949,
+      "learning_rate": 4.723004361374953e-05,
+      "loss": 0.1656,
+      "step": 14360
+    },
+    {
+      "epoch": 1.1846990333919156,
+      "grad_norm": 10.552154541015625,
+      "learning_rate": 4.721687163652829e-05,
+      "loss": 0.1828,
+      "step": 14380
+    },
+    {
+      "epoch": 1.1863466608084359,
+      "grad_norm": 0.5331339240074158,
+      "learning_rate": 4.7203670260933725e-05,
+      "loss": 0.1807,
+      "step": 14400
+    },
+    {
+      "epoch": 1.1879942882249561,
+      "grad_norm": 2.227116584777832,
+      "learning_rate": 4.719043950443448e-05,
+      "loss": 0.1903,
+      "step": 14420
+    },
+    {
+      "epoch": 1.1896419156414764,
+      "grad_norm": 0.36228641867637634,
+      "learning_rate": 4.717717938453811e-05,
+      "loss": 0.1775,
+      "step": 14440
+    },
+    {
+      "epoch": 1.1912895430579964,
+      "grad_norm": 2.8174803256988525,
+      "learning_rate": 4.7163889918790984e-05,
+      "loss": 0.1672,
+      "step": 14460
+    },
+    {
+      "epoch": 1.1929371704745166,
+      "grad_norm": 1.1632130146026611,
+      "learning_rate": 4.7150571124778344e-05,
+      "loss": 0.1798,
+      "step": 14480
+    },
+    {
+      "epoch": 1.1945847978910369,
+      "grad_norm": 2.7487854957580566,
+      "learning_rate": 4.713722302012421e-05,
+      "loss": 0.1833,
+      "step": 14500
+    },
+    {
+      "epoch": 1.1962324253075571,
+      "grad_norm": 0.7984181642532349,
+      "learning_rate": 4.712384562249141e-05,
+      "loss": 0.1934,
+      "step": 14520
+    },
+    {
+      "epoch": 1.1978800527240774,
+      "grad_norm": 0.22223146259784698,
+      "learning_rate": 4.711043894958153e-05,
+      "loss": 0.1883,
+      "step": 14540
+    },
+    {
+      "epoch": 1.1995276801405976,
+      "grad_norm": 5.944990158081055,
+      "learning_rate": 4.7097003019134876e-05,
+      "loss": 0.1584,
+      "step": 14560
+    },
+    {
+      "epoch": 1.2011753075571177,
+      "grad_norm": 0.5789378881454468,
+      "learning_rate": 4.708353784893049e-05,
+      "loss": 0.1837,
+      "step": 14580
+    },
+    {
+      "epoch": 1.202822934973638,
+      "grad_norm": 0.4476211369037628,
+      "learning_rate": 4.707004345678609e-05,
+      "loss": 0.1878,
+      "step": 14600
+    },
+    {
+      "epoch": 1.2044705623901582,
+      "grad_norm": 1.0182552337646484,
+      "learning_rate": 4.705651986055807e-05,
+      "loss": 0.1968,
+      "step": 14620
+    },
+    {
+      "epoch": 1.2061181898066784,
+      "grad_norm": 0.5336174368858337,
+      "learning_rate": 4.7042967078141466e-05,
+      "loss": 0.1849,
+      "step": 14640
+    },
+    {
+      "epoch": 1.2077658172231986,
+      "grad_norm": 0.6163532733917236,
+      "learning_rate": 4.702938512746994e-05,
+      "loss": 0.1571,
+      "step": 14660
+    },
+    {
+      "epoch": 1.209413444639719,
+      "grad_norm": 1.7392462491989136,
+      "learning_rate": 4.701577402651574e-05,
+      "loss": 0.205,
+      "step": 14680
+    },
+    {
+      "epoch": 1.2110610720562391,
+      "grad_norm": 0.9206298589706421,
+      "learning_rate": 4.70021337932897e-05,
+      "loss": 0.1872,
+      "step": 14700
+    },
+    {
+      "epoch": 1.2127086994727592,
+      "grad_norm": 1.0826961994171143,
+      "learning_rate": 4.6988464445841186e-05,
+      "loss": 0.185,
+      "step": 14720
+    },
+    {
+      "epoch": 1.2143563268892794,
+      "grad_norm": 0.3322293758392334,
+      "learning_rate": 4.6974766002258105e-05,
+      "loss": 0.1746,
+      "step": 14740
+    },
+    {
+      "epoch": 1.2160039543057997,
+      "grad_norm": 1.4564270973205566,
+      "learning_rate": 4.696103848066686e-05,
+      "loss": 0.1581,
+      "step": 14760
+    },
+    {
+      "epoch": 1.21765158172232,
+      "grad_norm": 0.7961132526397705,
+      "learning_rate": 4.6947281899232333e-05,
+      "loss": 0.1774,
+      "step": 14780
+    },
+    {
+      "epoch": 1.2192992091388402,
+      "grad_norm": 2.5025506019592285,
+      "learning_rate": 4.693349627615784e-05,
+      "loss": 0.2024,
+      "step": 14800
+    },
+    {
+      "epoch": 1.2209468365553602,
+      "grad_norm": 0.9771924614906311,
+      "learning_rate": 4.692037305103247e-05,
+      "loss": 0.1857,
+      "step": 14820
+    },
+    {
+      "epoch": 1.2225944639718804,
+      "grad_norm": 0.3456748127937317,
+      "learning_rate": 4.690653084926293e-05,
+      "loss": 0.1828,
+      "step": 14840
+    },
+    {
+      "epoch": 1.2242420913884007,
+      "grad_norm": 0.9823417067527771,
+      "learning_rate": 4.689265965977708e-05,
+      "loss": 0.1599,
+      "step": 14860
+    },
+    {
+      "epoch": 1.225889718804921,
+      "grad_norm": 1.0269907712936401,
+      "learning_rate": 4.68787595009299e-05,
+      "loss": 0.1797,
+      "step": 14880
+    },
+    {
+      "epoch": 1.2275373462214412,
+      "grad_norm": 0.7121568918228149,
+      "learning_rate": 4.686483039111472e-05,
+      "loss": 0.1851,
+      "step": 14900
+    },
+    {
+      "epoch": 1.2291849736379614,
+      "grad_norm": 0.5829301476478577,
+      "learning_rate": 4.6850872348763166e-05,
+      "loss": 0.1816,
+      "step": 14920
+    },
+    {
+      "epoch": 1.2308326010544817,
+      "grad_norm": 0.23825562000274658,
+      "learning_rate": 4.683688539234515e-05,
+      "loss": 0.1687,
+      "step": 14940
+    },
+    {
+      "epoch": 1.2324802284710017,
+      "grad_norm": 1.7142807245254517,
+      "learning_rate": 4.682286954036886e-05,
+      "loss": 0.1618,
+      "step": 14960
+    },
+    {
+      "epoch": 1.234127855887522,
+      "grad_norm": 2.196317195892334,
+      "learning_rate": 4.6808824811380704e-05,
+      "loss": 0.1868,
+      "step": 14980
+    },
+    {
+      "epoch": 1.2357754833040422,
+      "grad_norm": 0.6108691096305847,
+      "learning_rate": 4.67947512239653e-05,
+      "loss": 0.195,
+      "step": 15000
+    },
+    {
+      "epoch": 1.2374231107205624,
+      "grad_norm": 0.5913376808166504,
+      "learning_rate": 4.678135460276387e-05,
+      "loss": 0.1855,
+      "step": 15020
+    },
+    {
+      "epoch": 1.2390707381370827,
+      "grad_norm": 0.6738501191139221,
+      "learning_rate": 4.676722479501395e-05,
+      "loss": 0.1843,
+      "step": 15040
+    },
+    {
+      "epoch": 1.2407183655536027,
+      "grad_norm": 1.7320432662963867,
+      "learning_rate": 4.675306618388381e-05,
+      "loss": 0.1567,
+      "step": 15060
+    },
+    {
+      "epoch": 1.242365992970123,
+      "grad_norm": 1.1772425174713135,
+      "learning_rate": 4.6738878788108776e-05,
+      "loss": 0.1927,
+      "step": 15080
+    },
+    {
+      "epoch": 1.2440136203866432,
+      "grad_norm": 0.8685306906700134,
+      "learning_rate": 4.672466262646225e-05,
+      "loss": 0.2057,
+      "step": 15100
+    },
+    {
+      "epoch": 1.2456612478031635,
+      "grad_norm": 0.5060691833496094,
+      "learning_rate": 4.6710417717755695e-05,
+      "loss": 0.1846,
+      "step": 15120
+    },
+    {
+      "epoch": 1.2473088752196837,
+      "grad_norm": 0.2376640886068344,
+      "learning_rate": 4.6696144080838624e-05,
+      "loss": 0.1835,
+      "step": 15140
+    },
+    {
+      "epoch": 1.248956502636204,
+      "grad_norm": 1.7704603672027588,
+      "learning_rate": 4.6681841734598576e-05,
+      "loss": 0.1661,
+      "step": 15160
+    },
+    {
+      "epoch": 1.2506041300527242,
+      "grad_norm": 0.9057344198226929,
+      "learning_rate": 4.6667510697961045e-05,
+      "loss": 0.1942,
+      "step": 15180
+    },
+    {
+      "epoch": 1.2522517574692442,
+      "grad_norm": 0.7479068636894226,
+      "learning_rate": 4.6653150989889525e-05,
+      "loss": 0.1807,
+      "step": 15200
+    },
+    {
+      "epoch": 1.2538993848857645,
+      "grad_norm": 0.8971445560455322,
+      "learning_rate": 4.663876262938543e-05,
+      "loss": 0.192,
+      "step": 15220
+    },
+    {
+      "epoch": 1.2555470123022847,
+      "grad_norm": 0.6187984347343445,
+      "learning_rate": 4.6624345635488085e-05,
+      "loss": 0.1781,
+      "step": 15240
+    },
+    {
+      "epoch": 1.257194639718805,
+      "grad_norm": 1.5628702640533447,
+      "learning_rate": 4.6609900027274714e-05,
+      "loss": 0.1511,
+      "step": 15260
+    },
+    {
+      "epoch": 1.2588422671353252,
+      "grad_norm": 1.0656757354736328,
+      "learning_rate": 4.659542582386041e-05,
+      "loss": 0.1888,
+      "step": 15280
+    },
+    {
+      "epoch": 1.2604898945518452,
+      "grad_norm": 0.9140082597732544,
+      "learning_rate": 4.658092304439808e-05,
+      "loss": 0.1791,
+      "step": 15300
+    },
+    {
+      "epoch": 1.2621375219683655,
+      "grad_norm": 0.7536066770553589,
+      "learning_rate": 4.6566391708078484e-05,
+      "loss": 0.1945,
+      "step": 15320
+    },
+    {
+      "epoch": 1.2637851493848857,
+      "grad_norm": 0.6287294030189514,
+      "learning_rate": 4.655183183413012e-05,
+      "loss": 0.1844,
+      "step": 15340
+    },
+    {
+      "epoch": 1.265432776801406,
+      "grad_norm": 1.3471739292144775,
+      "learning_rate": 4.653724344181929e-05,
+      "loss": 0.1718,
+      "step": 15360
+    },
+    {
+      "epoch": 1.2670804042179262,
+      "grad_norm": 0.8668192625045776,
+      "learning_rate": 4.652262655045001e-05,
+      "loss": 0.1816,
+      "step": 15380
+    },
+    {
+      "epoch": 1.2687280316344465,
+      "grad_norm": 0.7043120861053467,
+      "learning_rate": 4.6507981179364015e-05,
+      "loss": 0.1835,
+      "step": 15400
+    },
+    {
+      "epoch": 1.2703756590509667,
+      "grad_norm": 2.341987133026123,
+      "learning_rate": 4.6493307347940735e-05,
+      "loss": 0.1779,
+      "step": 15420
+    },
+    {
+      "epoch": 1.2720232864674867,
+      "grad_norm": 0.7353787422180176,
+      "learning_rate": 4.6478605075597236e-05,
+      "loss": 0.1995,
+      "step": 15440
+    },
+    {
+      "epoch": 1.273670913884007,
+      "grad_norm": 1.131020188331604,
+      "learning_rate": 4.646387438178824e-05,
+      "loss": 0.1626,
+      "step": 15460
+    },
+    {
+      "epoch": 1.2753185413005272,
+      "grad_norm": 5.1669206619262695,
+      "learning_rate": 4.644911528600607e-05,
+      "loss": 0.1739,
+      "step": 15480
+    },
+    {
+      "epoch": 1.2769661687170475,
+      "grad_norm": 1.9862631559371948,
+      "learning_rate": 4.643432780778064e-05,
+      "loss": 0.1811,
+      "step": 15500
+    },
+    {
+      "epoch": 1.2786137961335677,
+      "grad_norm": 4.627335548400879,
+      "learning_rate": 4.641951196667939e-05,
+      "loss": 0.1755,
+      "step": 15520
+    },
+    {
+      "epoch": 1.2802614235500878,
+      "grad_norm": 1.6128586530685425,
+      "learning_rate": 4.6404667782307334e-05,
+      "loss": 0.1758,
+      "step": 15540
+    },
+    {
+      "epoch": 1.281909050966608,
+      "grad_norm": 0.8337876200675964,
+      "learning_rate": 4.638979527430696e-05,
+      "loss": 0.1543,
+      "step": 15560
+    },
+    {
+      "epoch": 1.2835566783831283,
+      "grad_norm": 1.7509500980377197,
+      "learning_rate": 4.6374894462358246e-05,
+      "loss": 0.1866,
+      "step": 15580
+    },
+    {
+      "epoch": 1.2852043057996485,
+      "grad_norm": 1.9350312948226929,
+      "learning_rate": 4.635996536617863e-05,
+      "loss": 0.1898,
+      "step": 15600
+    },
+    {
+      "epoch": 1.2868519332161688,
+      "grad_norm": 0.9639732241630554,
+      "learning_rate": 4.6345008005522966e-05,
+      "loss": 0.1757,
+      "step": 15620
+    },
+    {
+      "epoch": 1.288499560632689,
+      "grad_norm": 0.22245891392230988,
+      "learning_rate": 4.633002240018351e-05,
+      "loss": 0.1725,
+      "step": 15640
+    },
+    {
+      "epoch": 1.2901471880492092,
+      "grad_norm": 1.5688083171844482,
+      "learning_rate": 4.6315008569989905e-05,
+      "loss": 0.1682,
+      "step": 15660
+    },
+    {
+      "epoch": 1.2917948154657293,
+      "grad_norm": 0.9785603880882263,
+      "learning_rate": 4.629996653480912e-05,
+      "loss": 0.184,
+      "step": 15680
+    },
+    {
+      "epoch": 1.2934424428822495,
+      "grad_norm": 1.3254144191741943,
+      "learning_rate": 4.628489631454548e-05,
+      "loss": 0.1862,
+      "step": 15700
+    },
+    {
+      "epoch": 1.2950900702987698,
+      "grad_norm": 2.9795165061950684,
+      "learning_rate": 4.6269797929140554e-05,
+      "loss": 0.1836,
+      "step": 15720
+    },
+    {
+      "epoch": 1.29673769771529,
+      "grad_norm": 0.510050356388092,
+      "learning_rate": 4.625467139857323e-05,
+      "loss": 0.194,
+      "step": 15740
+    },
+    {
+      "epoch": 1.2983853251318103,
+      "grad_norm": 1.456040859222412,
+      "learning_rate": 4.6239516742859625e-05,
+      "loss": 0.1679,
+      "step": 15760
+    },
+    {
+      "epoch": 1.3000329525483303,
+      "grad_norm": 0.7667500376701355,
+      "learning_rate": 4.622433398205305e-05,
+      "loss": 0.1955,
+      "step": 15780
+    },
+    {
+      "epoch": 1.3016805799648505,
+      "grad_norm": 5.522034168243408,
+      "learning_rate": 4.620912313624404e-05,
+      "loss": 0.1766,
+      "step": 15800
+    },
+    {
+      "epoch": 1.3033282073813708,
+      "grad_norm": 1.5902382135391235,
+      "learning_rate": 4.619388422556026e-05,
+      "loss": 0.1843,
+      "step": 15820
+    },
+    {
+      "epoch": 1.304975834797891,
+      "grad_norm": 0.3163457214832306,
+      "learning_rate": 4.617861727016655e-05,
+      "loss": 0.1876,
+      "step": 15840
+    },
+    {
+      "epoch": 1.3066234622144113,
+      "grad_norm": 1.0746263265609741,
+      "learning_rate": 4.616332229026482e-05,
+      "loss": 0.1727,
+      "step": 15860
+    },
+    {
+      "epoch": 1.3082710896309315,
+      "grad_norm": 7.499575614929199,
+      "learning_rate": 4.6147999306094095e-05,
+      "loss": 0.191,
+      "step": 15880
+    },
+    {
+      "epoch": 1.3099187170474518,
+      "grad_norm": 1.3614699840545654,
+      "learning_rate": 4.613264833793044e-05,
+      "loss": 0.1979,
+      "step": 15900
+    },
+    {
+      "epoch": 1.3115663444639718,
+      "grad_norm": 2.371039390563965,
+      "learning_rate": 4.611726940608694e-05,
+      "loss": 0.1868,
+      "step": 15920
+    },
+    {
+      "epoch": 1.313213971880492,
+      "grad_norm": 0.4890645444393158,
+      "learning_rate": 4.6101862530913715e-05,
+      "loss": 0.181,
+      "step": 15940
+    },
+    {
+      "epoch": 1.3148615992970123,
+      "grad_norm": 1.2924984693527222,
+      "learning_rate": 4.6086427732797836e-05,
+      "loss": 0.1826,
+      "step": 15960
+    },
+    {
+      "epoch": 1.3165092267135325,
+      "grad_norm": 0.8311454653739929,
+      "learning_rate": 4.6070965032163316e-05,
+      "loss": 0.1757,
+      "step": 15980
+    },
+    {
+      "epoch": 1.3181568541300528,
+      "grad_norm": 4.735937595367432,
+      "learning_rate": 4.605547444947112e-05,
+      "loss": 0.1934,
+      "step": 16000
+    },
+    {
+      "epoch": 1.3181568541300528,
+      "eval_loss": 0.5345502495765686,
+      "eval_runtime": 711.0231,
+      "eval_samples_per_second": 30.645,
+      "eval_steps_per_second": 7.662,
+      "eval_wer": 0.23376714265318863,
+      "step": 16000
+    },
+    {
+      "epoch": 1.3198044815465728,
+      "grad_norm": 0.9712895750999451,
+      "learning_rate": 4.603995600521908e-05,
+      "loss": 0.1792,
+      "step": 16020
+    },
+    {
+      "epoch": 1.321452108963093,
+      "grad_norm": 0.3473972976207733,
+      "learning_rate": 4.6024409719941905e-05,
+      "loss": 0.1926,
+      "step": 16040
+    },
+    {
+      "epoch": 1.3230997363796133,
+      "grad_norm": 1.2832224369049072,
+      "learning_rate": 4.600883561421116e-05,
+      "loss": 0.1696,
+      "step": 16060
+    },
+    {
+      "epoch": 1.3247473637961336,
+      "grad_norm": 1.0073579549789429,
+      "learning_rate": 4.599401446384175e-05,
+      "loss": 0.1816,
+      "step": 16080
+    },
+    {
+      "epoch": 1.3263949912126538,
+      "grad_norm": 1.0329769849777222,
+      "learning_rate": 4.597838616753484e-05,
+      "loss": 0.1834,
+      "step": 16100
+    },
+    {
+      "epoch": 1.328042618629174,
+      "grad_norm": 2.175204038619995,
+      "learning_rate": 4.596273011167479e-05,
+      "loss": 0.1936,
+      "step": 16120
+    },
+    {
+      "epoch": 1.3296902460456943,
+      "grad_norm": 0.29955798387527466,
+      "learning_rate": 4.594704631697842e-05,
+      "loss": 0.1735,
+      "step": 16140
+    },
+    {
+      "epoch": 1.3313378734622145,
+      "grad_norm": 5.876577377319336,
+      "learning_rate": 4.593133480419924e-05,
+      "loss": 0.1642,
+      "step": 16160
+    },
+    {
+      "epoch": 1.3329855008787346,
+      "grad_norm": 1.0501422882080078,
+      "learning_rate": 4.591559559412743e-05,
+      "loss": 0.2176,
+      "step": 16180
+    },
+    {
+      "epoch": 1.3346331282952548,
+      "grad_norm": 2.9819207191467285,
+      "learning_rate": 4.5899828707589844e-05,
+      "loss": 0.1841,
+      "step": 16200
+    },
+    {
+      "epoch": 1.336280755711775,
+      "grad_norm": 0.8568416237831116,
+      "learning_rate": 4.588403416544992e-05,
+      "loss": 0.1916,
+      "step": 16220
+    },
+    {
+      "epoch": 1.3379283831282953,
+      "grad_norm": 1.058608889579773,
+      "learning_rate": 4.586821198860772e-05,
+      "loss": 0.1881,
+      "step": 16240
+    },
+    {
+      "epoch": 1.3395760105448153,
+      "grad_norm": 1.0041903257369995,
+      "learning_rate": 4.585236219799988e-05,
+      "loss": 0.1696,
+      "step": 16260
+    },
+    {
+      "epoch": 1.3412236379613356,
+      "grad_norm": 1.351030945777893,
+      "learning_rate": 4.5836484814599544e-05,
+      "loss": 0.183,
+      "step": 16280
+    },
+    {
+      "epoch": 1.3428712653778558,
+      "grad_norm": 0.7650487422943115,
+      "learning_rate": 4.58205798594164e-05,
+      "loss": 0.1775,
+      "step": 16300
+    },
+    {
+      "epoch": 1.344518892794376,
+      "grad_norm": 1.0618997812271118,
+      "learning_rate": 4.5804647353496604e-05,
+      "loss": 0.1846,
+      "step": 16320
+    },
+    {
+      "epoch": 1.3461665202108963,
+      "grad_norm": 1.2706764936447144,
+      "learning_rate": 4.578868731792277e-05,
+      "loss": 0.1994,
+      "step": 16340
+    },
+    {
+      "epoch": 1.3478141476274166,
+      "grad_norm": 0.9554028511047363,
+      "learning_rate": 4.5772699773813936e-05,
+      "loss": 0.1564,
+      "step": 16360
+    },
+    {
+      "epoch": 1.3494617750439368,
+      "grad_norm": 0.8473567366600037,
+      "learning_rate": 4.575668474232556e-05,
+      "loss": 0.1924,
+      "step": 16380
+    },
+    {
+      "epoch": 1.351109402460457,
+      "grad_norm": 1.2312932014465332,
+      "learning_rate": 4.574064224464944e-05,
+      "loss": 0.194,
+      "step": 16400
+    },
+    {
+      "epoch": 1.352757029876977,
+      "grad_norm": 0.5687373280525208,
+      "learning_rate": 4.5724572302013754e-05,
+      "loss": 0.1828,
+      "step": 16420
+    },
+    {
+      "epoch": 1.3544046572934973,
+      "grad_norm": 0.2973005473613739,
+      "learning_rate": 4.570847493568297e-05,
+      "loss": 0.1774,
+      "step": 16440
+    },
+    {
+      "epoch": 1.3560522847100176,
+      "grad_norm": 22.61744499206543,
+      "learning_rate": 4.5692350166957864e-05,
+      "loss": 0.16,
+      "step": 16460
+    },
+    {
+      "epoch": 1.3576999121265378,
+      "grad_norm": 0.5010253190994263,
+      "learning_rate": 4.567619801717545e-05,
+      "loss": 0.1842,
+      "step": 16480
+    },
+    {
+      "epoch": 1.3593475395430579,
+      "grad_norm": 1.405232548713684,
+      "learning_rate": 4.566001850770899e-05,
+      "loss": 0.194,
+      "step": 16500
+    },
+    {
+      "epoch": 1.3609951669595781,
+      "grad_norm": 0.8372935652732849,
+      "learning_rate": 4.564381165996796e-05,
+      "loss": 0.1866,
+      "step": 16520
+    },
+    {
+      "epoch": 1.3626427943760984,
+      "grad_norm": 0.2158607393503189,
+      "learning_rate": 4.5627577495397974e-05,
+      "loss": 0.1802,
+      "step": 16540
+    },
+    {
+      "epoch": 1.3642904217926186,
+      "grad_norm": 7.98114013671875,
+      "learning_rate": 4.561131603548082e-05,
+      "loss": 0.1832,
+      "step": 16560
+    },
+    {
+      "epoch": 1.3659380492091389,
+      "grad_norm": 0.6528833508491516,
+      "learning_rate": 4.5595027301734413e-05,
+      "loss": 0.1919,
+      "step": 16580
+    },
+    {
+      "epoch": 1.367585676625659,
+      "grad_norm": 0.5290044546127319,
+      "learning_rate": 4.557871131571274e-05,
+      "loss": 0.1693,
+      "step": 16600
+    },
+    {
+      "epoch": 1.3692333040421794,
+      "grad_norm": 1.5808736085891724,
+      "learning_rate": 4.5562368099005854e-05,
+      "loss": 0.1813,
+      "step": 16620
+    },
+    {
+      "epoch": 1.3708809314586996,
+      "grad_norm": 0.17612561583518982,
+      "learning_rate": 4.554599767323984e-05,
+      "loss": 0.1879,
+      "step": 16640
+    },
+    {
+      "epoch": 1.3725285588752196,
+      "grad_norm": 1.369644284248352,
+      "learning_rate": 4.552960006007678e-05,
+      "loss": 0.1602,
+      "step": 16660
+    },
+    {
+      "epoch": 1.3741761862917399,
+      "grad_norm": 1.1678745746612549,
+      "learning_rate": 4.551317528121476e-05,
+      "loss": 0.1897,
+      "step": 16680
+    },
+    {
+      "epoch": 1.3758238137082601,
+      "grad_norm": 0.5510971546173096,
+      "learning_rate": 4.549672335838777e-05,
+      "loss": 0.196,
+      "step": 16700
+    },
+    {
+      "epoch": 1.3774714411247804,
+      "grad_norm": 0.9784049391746521,
+      "learning_rate": 4.548024431336576e-05,
+      "loss": 0.1865,
+      "step": 16720
+    },
+    {
+      "epoch": 1.3791190685413004,
+      "grad_norm": 0.2733114957809448,
+      "learning_rate": 4.546373816795455e-05,
+      "loss": 0.1811,
+      "step": 16740
+    },
+    {
+      "epoch": 1.3807666959578206,
+      "grad_norm": 1.3034899234771729,
+      "learning_rate": 4.5447204943995816e-05,
+      "loss": 0.1528,
+      "step": 16760
+    },
+    {
+      "epoch": 1.382414323374341,
+      "grad_norm": 1.3286523818969727,
+      "learning_rate": 4.543064466336706e-05,
+      "loss": 0.1832,
+      "step": 16780
+    },
+    {
+      "epoch": 1.3840619507908611,
+      "grad_norm": 0.747806191444397,
+      "learning_rate": 4.541405734798161e-05,
+      "loss": 0.1672,
+      "step": 16800
+    },
+    {
+      "epoch": 1.3857095782073814,
+      "grad_norm": 0.5867149233818054,
+      "learning_rate": 4.539744301978856e-05,
+      "loss": 0.1814,
+      "step": 16820
+    },
+    {
+      "epoch": 1.3873572056239016,
+      "grad_norm": 0.43317413330078125,
+      "learning_rate": 4.5380801700772726e-05,
+      "loss": 0.1825,
+      "step": 16840
+    },
+    {
+      "epoch": 1.3890048330404219,
+      "grad_norm": 1.011514663696289,
+      "learning_rate": 4.536413341295467e-05,
+      "loss": 0.1553,
+      "step": 16860
+    },
+    {
+      "epoch": 1.3906524604569421,
+      "grad_norm": 0.7447202205657959,
+      "learning_rate": 4.5347438178390624e-05,
+      "loss": 0.1832,
+      "step": 16880
+    },
+    {
+      "epoch": 1.3923000878734622,
+      "grad_norm": 4.756796360015869,
+      "learning_rate": 4.533071601917248e-05,
+      "loss": 0.1923,
+      "step": 16900
+    },
+    {
+      "epoch": 1.3939477152899824,
+      "grad_norm": 1.5507986545562744,
+      "learning_rate": 4.5313966957427755e-05,
+      "loss": 0.1754,
+      "step": 16920
+    },
+    {
+      "epoch": 1.3955953427065027,
+      "grad_norm": 0.557498037815094,
+      "learning_rate": 4.529719101531957e-05,
+      "loss": 0.178,
+      "step": 16940
+    },
+    {
+      "epoch": 1.397242970123023,
+      "grad_norm": 1.611181378364563,
+      "learning_rate": 4.5280388215046624e-05,
+      "loss": 0.1557,
+      "step": 16960
+    },
+    {
+      "epoch": 1.398890597539543,
+      "grad_norm": 0.6798775792121887,
+      "learning_rate": 4.5263558578843136e-05,
+      "loss": 0.1741,
+      "step": 16980
+    },
+    {
+      "epoch": 1.4005382249560632,
+      "grad_norm": 10.896245956420898,
+      "learning_rate": 4.524670212897884e-05,
+      "loss": 0.2016,
+      "step": 17000
+    },
+    {
+      "epoch": 1.4021858523725834,
+      "grad_norm": 1.8802106380462646,
+      "learning_rate": 4.5229818887758957e-05,
+      "loss": 0.177,
+      "step": 17020
+    },
+    {
+      "epoch": 1.4038334797891037,
+      "grad_norm": 0.2604859471321106,
+      "learning_rate": 4.5212908877524165e-05,
+      "loss": 0.1791,
+      "step": 17040
+    },
+    {
+      "epoch": 1.405481107205624,
+      "grad_norm": 1.834112524986267,
+      "learning_rate": 4.519597212065056e-05,
+      "loss": 0.1664,
+      "step": 17060
+    },
+    {
+      "epoch": 1.4071287346221442,
+      "grad_norm": 0.9913097620010376,
+      "learning_rate": 4.517900863954962e-05,
+      "loss": 0.1955,
+      "step": 17080
+    },
+    {
+      "epoch": 1.4087763620386644,
+      "grad_norm": 2.1117942333221436,
+      "learning_rate": 4.51620184566682e-05,
+      "loss": 0.1798,
+      "step": 17100
+    },
+    {
+      "epoch": 1.4104239894551847,
+      "grad_norm": 5.098875999450684,
+      "learning_rate": 4.5145001594488466e-05,
+      "loss": 0.1908,
+      "step": 17120
+    },
+    {
+      "epoch": 1.4120716168717047,
+      "grad_norm": 0.7090994715690613,
+      "learning_rate": 4.512795807552792e-05,
+      "loss": 0.1727,
+      "step": 17140
+    },
+    {
+      "epoch": 1.413719244288225,
+      "grad_norm": 3.752267599105835,
+      "learning_rate": 4.511088792233932e-05,
+      "loss": 0.1542,
+      "step": 17160
+    },
+    {
+      "epoch": 1.4153668717047452,
+      "grad_norm": 1.5449800491333008,
+      "learning_rate": 4.5093791157510656e-05,
+      "loss": 0.1651,
+      "step": 17180
+    },
+    {
+      "epoch": 1.4170144991212654,
+      "grad_norm": 0.9400858283042908,
+      "learning_rate": 4.507666780366514e-05,
+      "loss": 0.1844,
+      "step": 17200
+    },
+    {
+      "epoch": 1.4186621265377855,
+      "grad_norm": 0.37838345766067505,
+      "learning_rate": 4.5059517883461196e-05,
+      "loss": 0.1733,
+      "step": 17220
+    },
+    {
+      "epoch": 1.4203097539543057,
+      "grad_norm": 0.239340141415596,
+      "learning_rate": 4.50432008728471e-05,
+      "loss": 0.1855,
+      "step": 17240
+    },
+    {
+      "epoch": 1.421957381370826,
+      "grad_norm": 1.4485529661178589,
+      "learning_rate": 4.502599921354852e-05,
+      "loss": 0.1515,
+      "step": 17260
+    },
+    {
+      "epoch": 1.4236050087873462,
+      "grad_norm": 0.6260993480682373,
+      "learning_rate": 4.500877105493846e-05,
+      "loss": 0.1781,
+      "step": 17280
+    },
+    {
+      "epoch": 1.4252526362038664,
+      "grad_norm": 2.114168882369995,
+      "learning_rate": 4.499151641981402e-05,
+      "loss": 0.1732,
+      "step": 17300
+    },
+    {
+      "epoch": 1.4269002636203867,
+      "grad_norm": 0.7402538061141968,
+      "learning_rate": 4.497423533100732e-05,
+      "loss": 0.1769,
+      "step": 17320
+    },
+    {
+      "epoch": 1.428547891036907,
+      "grad_norm": 0.7155308723449707,
+      "learning_rate": 4.4956927811385475e-05,
+      "loss": 0.1649,
+      "step": 17340
+    },
+    {
+      "epoch": 1.4301955184534272,
+      "grad_norm": 0.8413863182067871,
+      "learning_rate": 4.4939593883850576e-05,
+      "loss": 0.1563,
+      "step": 17360
+    },
+    {
+      "epoch": 1.4318431458699472,
+      "grad_norm": 2.0796799659729004,
+      "learning_rate": 4.492223357133967e-05,
+      "loss": 0.1797,
+      "step": 17380
+    },
+    {
+      "epoch": 1.4334907732864675,
+      "grad_norm": 1.1681219339370728,
+      "learning_rate": 4.490484689682471e-05,
+      "loss": 0.1819,
+      "step": 17400
+    },
+    {
+      "epoch": 1.4351384007029877,
+      "grad_norm": 1.0417784452438354,
+      "learning_rate": 4.488743388331254e-05,
+      "loss": 0.1861,
+      "step": 17420
+    },
+    {
+      "epoch": 1.436786028119508,
+      "grad_norm": 0.17414021492004395,
+      "learning_rate": 4.486999455384483e-05,
+      "loss": 0.1734,
+      "step": 17440
+    },
+    {
+      "epoch": 1.4384336555360282,
+      "grad_norm": 1.723633885383606,
+      "learning_rate": 4.485252893149812e-05,
+      "loss": 0.1629,
+      "step": 17460
+    },
+    {
+      "epoch": 1.4400812829525482,
+      "grad_norm": 0.5598846673965454,
+      "learning_rate": 4.4835037039383686e-05,
+      "loss": 0.1749,
+      "step": 17480
+    },
+    {
+      "epoch": 1.4417289103690685,
+      "grad_norm": 0.617661714553833,
+      "learning_rate": 4.481751890064761e-05,
+      "loss": 0.1734,
+      "step": 17500
+    },
+    {
+      "epoch": 1.4433765377855887,
+      "grad_norm": 0.4917828440666199,
+      "learning_rate": 4.479997453847069e-05,
+      "loss": 0.1724,
+      "step": 17520
+    },
+    {
+      "epoch": 1.445024165202109,
+      "grad_norm": 0.27072271704673767,
+      "learning_rate": 4.4782403976068417e-05,
+      "loss": 0.1664,
+      "step": 17540
+    },
+    {
+      "epoch": 1.4466717926186292,
+      "grad_norm": 1.5086781978607178,
+      "learning_rate": 4.476480723669095e-05,
+      "loss": 0.1614,
+      "step": 17560
+    },
+    {
+      "epoch": 1.4483194200351495,
+      "grad_norm": 1.092875599861145,
+      "learning_rate": 4.4747184343623113e-05,
+      "loss": 0.1827,
+      "step": 17580
+    },
+    {
+      "epoch": 1.4499670474516697,
+      "grad_norm": 0.9140943288803101,
+      "learning_rate": 4.4729535320184304e-05,
+      "loss": 0.1865,
+      "step": 17600
+    },
+    {
+      "epoch": 1.4516146748681897,
+      "grad_norm": 0.9031366109848022,
+      "learning_rate": 4.471186018972852e-05,
+      "loss": 0.1767,
+      "step": 17620
+    },
+    {
+      "epoch": 1.45326230228471,
+      "grad_norm": 0.4150620102882385,
+      "learning_rate": 4.46941589756443e-05,
+      "loss": 0.1814,
+      "step": 17640
+    },
+    {
+      "epoch": 1.4549099297012302,
+      "grad_norm": 2.178504228591919,
+      "learning_rate": 4.467643170135467e-05,
+      "loss": 0.1608,
+      "step": 17660
+    },
+    {
+      "epoch": 1.4565575571177505,
+      "grad_norm": 1.05979323387146,
+      "learning_rate": 4.4658678390317186e-05,
+      "loss": 0.175,
+      "step": 17680
+    },
+    {
+      "epoch": 1.4582051845342707,
+      "grad_norm": 1.3193062543869019,
+      "learning_rate": 4.464089906602383e-05,
+      "loss": 0.2092,
+      "step": 17700
+    },
+    {
+      "epoch": 1.4598528119507908,
+      "grad_norm": 0.6934455037117004,
+      "learning_rate": 4.462309375200101e-05,
+      "loss": 0.1695,
+      "step": 17720
+    },
+    {
+      "epoch": 1.461500439367311,
+      "grad_norm": 0.34769323468208313,
+      "learning_rate": 4.460526247180953e-05,
+      "loss": 0.1844,
+      "step": 17740
+    },
+    {
+      "epoch": 1.4631480667838312,
+      "grad_norm": 2.567417621612549,
+      "learning_rate": 4.4587405249044546e-05,
+      "loss": 0.1657,
+      "step": 17760
+    },
+    {
+      "epoch": 1.4647956942003515,
+      "grad_norm": 0.7568791508674622,
+      "learning_rate": 4.4569522107335535e-05,
+      "loss": 0.1794,
+      "step": 17780
+    },
+    {
+      "epoch": 1.4664433216168717,
+      "grad_norm": 0.6661089062690735,
+      "learning_rate": 4.455161307034631e-05,
+      "loss": 0.1812,
+      "step": 17800
+    },
+    {
+      "epoch": 1.468090949033392,
+      "grad_norm": 0.6958649754524231,
+      "learning_rate": 4.4533678161774894e-05,
+      "loss": 0.1714,
+      "step": 17820
+    },
+    {
+      "epoch": 1.4697385764499122,
+      "grad_norm": 0.3765020966529846,
+      "learning_rate": 4.45157174053536e-05,
+      "loss": 0.176,
+      "step": 17840
+    },
+    {
+      "epoch": 1.4713862038664323,
+      "grad_norm": 1.378657579421997,
+      "learning_rate": 4.4497730824848885e-05,
+      "loss": 0.1645,
+      "step": 17860
+    },
+    {
+      "epoch": 1.4730338312829525,
+      "grad_norm": 0.8017684817314148,
+      "learning_rate": 4.447971844406143e-05,
+      "loss": 0.1795,
+      "step": 17880
+    },
+    {
+      "epoch": 1.4746814586994728,
+      "grad_norm": 1.0388306379318237,
+      "learning_rate": 4.4461680286826024e-05,
+      "loss": 0.1941,
+      "step": 17900
+    },
+    {
+      "epoch": 1.476329086115993,
+      "grad_norm": 0.5598440766334534,
+      "learning_rate": 4.4443616377011586e-05,
+      "loss": 0.1879,
+      "step": 17920
+    },
+    {
+      "epoch": 1.4779767135325133,
+      "grad_norm": 0.18948721885681152,
+      "learning_rate": 4.4425526738521096e-05,
+      "loss": 0.1837,
+      "step": 17940
+    },
+    {
+      "epoch": 1.4796243409490333,
+      "grad_norm": 1.4350675344467163,
+      "learning_rate": 4.440741139529159e-05,
+      "loss": 0.1578,
+      "step": 17960
+    },
+    {
+      "epoch": 1.4812719683655535,
+      "grad_norm": 0.7614567279815674,
+      "learning_rate": 4.4389270371294104e-05,
+      "loss": 0.1878,
+      "step": 17980
+    },
+    {
+      "epoch": 1.4829195957820738,
+      "grad_norm": 0.5294657945632935,
+      "learning_rate": 4.437110369053368e-05,
+      "loss": 0.1792,
+      "step": 18000
+    },
+    {
+      "epoch": 1.4829195957820738,
+      "eval_loss": 0.539612352848053,
+      "eval_runtime": 824.8651,
+      "eval_samples_per_second": 26.415,
+      "eval_steps_per_second": 6.605,
+      "eval_wer": 0.2317739899166627,
+      "step": 18000
+    },
+    {
+      "epoch": 1.484567223198594,
+      "grad_norm": 0.9461752772331238,
+      "learning_rate": 4.4352911377049266e-05,
+      "loss": 0.1759,
+      "step": 18020
+    },
+    {
+      "epoch": 1.4862148506151143,
+      "grad_norm": 0.35723596811294556,
+      "learning_rate": 4.433469345491379e-05,
+      "loss": 0.1883,
+      "step": 18040
+    },
+    {
+      "epoch": 1.4878624780316345,
+      "grad_norm": 1.2138187885284424,
+      "learning_rate": 4.4316449948234007e-05,
+      "loss": 0.1486,
+      "step": 18060
+    },
+    {
+      "epoch": 1.4895101054481548,
+      "grad_norm": 0.9856829643249512,
+      "learning_rate": 4.4298180881150575e-05,
+      "loss": 0.1985,
+      "step": 18080
+    },
+    {
+      "epoch": 1.4911577328646748,
+      "grad_norm": 0.697496771812439,
+      "learning_rate": 4.4279886277837945e-05,
+      "loss": 0.1799,
+      "step": 18100
+    },
+    {
+      "epoch": 1.492805360281195,
+      "grad_norm": 0.8957895040512085,
+      "learning_rate": 4.426156616250436e-05,
+      "loss": 0.1755,
+      "step": 18120
+    },
+    {
+      "epoch": 1.4944529876977153,
+      "grad_norm": 0.3005205988883972,
+      "learning_rate": 4.424322055939183e-05,
+      "loss": 0.1754,
+      "step": 18140
+    },
+    {
+      "epoch": 1.4961006151142355,
+      "grad_norm": 0.5195316672325134,
+      "learning_rate": 4.42248494927761e-05,
+      "loss": 0.1566,
+      "step": 18160
+    },
+    {
+      "epoch": 1.4977482425307558,
+      "grad_norm": 1.898149013519287,
+      "learning_rate": 4.42064529869666e-05,
+      "loss": 0.177,
+      "step": 18180
+    },
+    {
+      "epoch": 1.4993958699472758,
+      "grad_norm": 2.6191961765289307,
+      "learning_rate": 4.4188031066306406e-05,
+      "loss": 0.1794,
+      "step": 18200
+    },
+    {
+      "epoch": 1.501043497363796,
+      "grad_norm": 0.5942957997322083,
+      "learning_rate": 4.416958375517225e-05,
+      "loss": 0.1805,
+      "step": 18220
+    },
+    {
+      "epoch": 1.5026911247803163,
+      "grad_norm": 0.2722390294075012,
+      "learning_rate": 4.415111107797445e-05,
+      "loss": 0.1704,
+      "step": 18240
+    },
+    {
+      "epoch": 1.5043387521968365,
+      "grad_norm": 1.5523074865341187,
+      "learning_rate": 4.41326130591569e-05,
+      "loss": 0.1683,
+      "step": 18260
+    },
+    {
+      "epoch": 1.5059863796133568,
+      "grad_norm": 0.6534819006919861,
+      "learning_rate": 4.4114089723197e-05,
+      "loss": 0.1734,
+      "step": 18280
+    },
+    {
+      "epoch": 1.507634007029877,
+      "grad_norm": 0.3791787922382355,
+      "learning_rate": 4.409554109460568e-05,
+      "loss": 0.1713,
+      "step": 18300
+    },
+    {
+      "epoch": 1.5092816344463973,
+      "grad_norm": 0.8552284836769104,
+      "learning_rate": 4.407696719792734e-05,
+      "loss": 0.1816,
+      "step": 18320
+    },
+    {
+      "epoch": 1.5109292618629175,
+      "grad_norm": 1.9957807064056396,
+      "learning_rate": 4.405836805773979e-05,
+      "loss": 0.1676,
+      "step": 18340
+    },
+    {
+      "epoch": 1.5125768892794376,
+      "grad_norm": 3.502870559692383,
+      "learning_rate": 4.4039743698654244e-05,
+      "loss": 0.1557,
+      "step": 18360
+    },
+    {
+      "epoch": 1.5142245166959578,
+      "grad_norm": 2.0844368934631348,
+      "learning_rate": 4.402109414531532e-05,
+      "loss": 0.1744,
+      "step": 18380
+    },
+    {
+      "epoch": 1.515872144112478,
+      "grad_norm": 2.32145357131958,
+      "learning_rate": 4.4002419422400954e-05,
+      "loss": 0.1853,
+      "step": 18400
+    },
+    {
+      "epoch": 1.517519771528998,
+      "grad_norm": 4.515986919403076,
+      "learning_rate": 4.398371955462237e-05,
+      "loss": 0.1737,
+      "step": 18420
+    },
+    {
+      "epoch": 1.5191673989455183,
+      "grad_norm": 0.3069842457771301,
+      "learning_rate": 4.396499456672409e-05,
+      "loss": 0.1863,
+      "step": 18440
+    },
+    {
+      "epoch": 1.5208150263620386,
+      "grad_norm": 2.3054933547973633,
+      "learning_rate": 4.394624448348388e-05,
+      "loss": 0.1503,
+      "step": 18460
+    },
+    {
+      "epoch": 1.5224626537785588,
+      "grad_norm": 0.7396153211593628,
+      "learning_rate": 4.392746932971269e-05,
+      "loss": 0.1882,
+      "step": 18480
+    },
+    {
+      "epoch": 1.524110281195079,
+      "grad_norm": 0.6996922492980957,
+      "learning_rate": 4.390866913025466e-05,
+      "loss": 0.1812,
+      "step": 18500
+    },
+    {
+      "epoch": 1.5257579086115993,
+      "grad_norm": 0.8116775751113892,
+      "learning_rate": 4.3889843909987064e-05,
+      "loss": 0.182,
+      "step": 18520
+    },
+    {
+      "epoch": 1.5274055360281196,
+      "grad_norm": 0.23936544358730316,
+      "learning_rate": 4.387099369382029e-05,
+      "loss": 0.1837,
+      "step": 18540
+    },
+    {
+      "epoch": 1.5290531634446398,
+      "grad_norm": 0.7272351980209351,
+      "learning_rate": 4.3852118506697795e-05,
+      "loss": 0.1601,
+      "step": 18560
+    },
+    {
+      "epoch": 1.53070079086116,
+      "grad_norm": 0.8234082460403442,
+      "learning_rate": 4.3833218373596094e-05,
+      "loss": 0.1628,
+      "step": 18580
+    },
+    {
+      "epoch": 1.53234841827768,
+      "grad_norm": 0.8772994875907898,
+      "learning_rate": 4.38142933195247e-05,
+      "loss": 0.1683,
+      "step": 18600
+    },
+    {
+      "epoch": 1.5339960456942003,
+      "grad_norm": 0.4307045340538025,
+      "learning_rate": 4.379534336952609e-05,
+      "loss": 0.1983,
+      "step": 18620
+    },
+    {
+      "epoch": 1.5356436731107206,
+      "grad_norm": 0.24042223393917084,
+      "learning_rate": 4.377636854867572e-05,
+      "loss": 0.163,
+      "step": 18640
+    },
+    {
+      "epoch": 1.5372913005272406,
+      "grad_norm": 1.3200021982192993,
+      "learning_rate": 4.3757368882081905e-05,
+      "loss": 0.1574,
+      "step": 18660
+    },
+    {
+      "epoch": 1.5389389279437609,
+      "grad_norm": 0.6988914608955383,
+      "learning_rate": 4.37383443948859e-05,
+      "loss": 0.1656,
+      "step": 18680
+    },
+    {
+      "epoch": 1.540586555360281,
+      "grad_norm": 0.46211034059524536,
+      "learning_rate": 4.371929511226175e-05,
+      "loss": 0.1843,
+      "step": 18700
+    },
+    {
+      "epoch": 1.5422341827768014,
+      "grad_norm": 0.7007827758789062,
+      "learning_rate": 4.370022105941633e-05,
+      "loss": 0.1642,
+      "step": 18720
+    },
+    {
+      "epoch": 1.5438818101933216,
+      "grad_norm": 0.3869803249835968,
+      "learning_rate": 4.3681122261589305e-05,
+      "loss": 0.1561,
+      "step": 18740
+    },
+    {
+      "epoch": 1.5455294376098418,
+      "grad_norm": 0.945253312587738,
+      "learning_rate": 4.366199874405306e-05,
+      "loss": 0.161,
+      "step": 18760
+    },
+    {
+      "epoch": 1.547177065026362,
+      "grad_norm": 1.2405415773391724,
+      "learning_rate": 4.36428505321127e-05,
+      "loss": 0.1671,
+      "step": 18780
+    },
+    {
+      "epoch": 1.5488246924428823,
+      "grad_norm": 0.9262678027153015,
+      "learning_rate": 4.3623677651106025e-05,
+      "loss": 0.1819,
+      "step": 18800
+    },
+    {
+      "epoch": 1.5504723198594026,
+      "grad_norm": 0.5617028474807739,
+      "learning_rate": 4.360448012640345e-05,
+      "loss": 0.1823,
+      "step": 18820
+    },
+    {
+      "epoch": 1.5521199472759226,
+      "grad_norm": 0.24927659332752228,
+      "learning_rate": 4.358525798340801e-05,
+      "loss": 0.1802,
+      "step": 18840
+    },
+    {
+      "epoch": 1.5537675746924429,
+      "grad_norm": 1.1216681003570557,
+      "learning_rate": 4.356601124755532e-05,
+      "loss": 0.142,
+      "step": 18860
+    },
+    {
+      "epoch": 1.5554152021089631,
+      "grad_norm": 1.321250319480896,
+      "learning_rate": 4.3546739944313544e-05,
+      "loss": 0.1849,
+      "step": 18880
+    },
+    {
+      "epoch": 1.5570628295254831,
+      "grad_norm": 1.2743161916732788,
+      "learning_rate": 4.352744409918332e-05,
+      "loss": 0.1802,
+      "step": 18900
+    },
+    {
+      "epoch": 1.5587104569420034,
+      "grad_norm": 0.6473543047904968,
+      "learning_rate": 4.350812373769781e-05,
+      "loss": 0.1807,
+      "step": 18920
+    },
+    {
+      "epoch": 1.5603580843585236,
+      "grad_norm": 0.6407447457313538,
+      "learning_rate": 4.348877888542259e-05,
+      "loss": 0.1825,
+      "step": 18940
+    },
+    {
+      "epoch": 1.5620057117750439,
+      "grad_norm": 1.5329558849334717,
+      "learning_rate": 4.346940956795564e-05,
+      "loss": 0.1555,
+      "step": 18960
+    },
+    {
+      "epoch": 1.5636533391915641,
+      "grad_norm": 0.9263864755630493,
+      "learning_rate": 4.345001581092733e-05,
+      "loss": 0.1847,
+      "step": 18980
+    },
+    {
+      "epoch": 1.5653009666080844,
+      "grad_norm": 0.6662928462028503,
+      "learning_rate": 4.343059764000035e-05,
+      "loss": 0.1947,
+      "step": 19000
+    },
+    {
+      "epoch": 1.5669485940246046,
+      "grad_norm": 3.533243417739868,
+      "learning_rate": 4.3411155080869704e-05,
+      "loss": 0.189,
+      "step": 19020
+    },
+    {
+      "epoch": 1.5685962214411249,
+      "grad_norm": 0.34724509716033936,
+      "learning_rate": 4.339168815926268e-05,
+      "loss": 0.1751,
+      "step": 19040
+    },
+    {
+      "epoch": 1.5702438488576451,
+      "grad_norm": 1.1154448986053467,
+      "learning_rate": 4.337219690093879e-05,
+      "loss": 0.1615,
+      "step": 19060
+    },
+    {
+      "epoch": 1.5718914762741654,
+      "grad_norm": 2.445345163345337,
+      "learning_rate": 4.335268133168974e-05,
+      "loss": 0.1927,
+      "step": 19080
+    },
+    {
+      "epoch": 1.5735391036906854,
+      "grad_norm": 1.5590624809265137,
+      "learning_rate": 4.3333141477339426e-05,
+      "loss": 0.1985,
+      "step": 19100
+    },
+    {
+      "epoch": 1.5751867311072056,
+      "grad_norm": 1.7810287475585938,
+      "learning_rate": 4.3313577363743874e-05,
+      "loss": 0.1825,
+      "step": 19120
+    },
+    {
+      "epoch": 1.5768343585237259,
+      "grad_norm": 0.18706871569156647,
+      "learning_rate": 4.32939890167912e-05,
+      "loss": 0.1884,
+      "step": 19140
+    },
+    {
+      "epoch": 1.578481985940246,
+      "grad_norm": 5.887953281402588,
+      "learning_rate": 4.327437646240161e-05,
+      "loss": 0.1584,
+      "step": 19160
+    },
+    {
+      "epoch": 1.5801296133567662,
+      "grad_norm": 0.7244459390640259,
+      "learning_rate": 4.32547397265273e-05,
+      "loss": 0.1821,
+      "step": 19180
+    },
+    {
+      "epoch": 1.5817772407732864,
+      "grad_norm": 0.7429137229919434,
+      "learning_rate": 4.323507883515252e-05,
+      "loss": 0.18,
+      "step": 19200
+    },
+    {
+      "epoch": 1.5834248681898067,
+      "grad_norm": 0.573156476020813,
+      "learning_rate": 4.321539381429343e-05,
+      "loss": 0.1777,
+      "step": 19220
+    },
+    {
+      "epoch": 1.585072495606327,
+      "grad_norm": 0.19367003440856934,
+      "learning_rate": 4.3195684689998155e-05,
+      "loss": 0.1765,
+      "step": 19240
+    },
+    {
+      "epoch": 1.5867201230228472,
+      "grad_norm": 1.1158256530761719,
+      "learning_rate": 4.31759514883467e-05,
+      "loss": 0.1667,
+      "step": 19260
+    },
+    {
+      "epoch": 1.5883677504393674,
+      "grad_norm": 0.7506932616233826,
+      "learning_rate": 4.3156194235450936e-05,
+      "loss": 0.1657,
+      "step": 19280
+    },
+    {
+      "epoch": 1.5900153778558876,
+      "grad_norm": 4.906134605407715,
+      "learning_rate": 4.313641295745455e-05,
+      "loss": 0.1837,
+      "step": 19300
+    },
+    {
+      "epoch": 1.591663005272408,
+      "grad_norm": 2.754065990447998,
+      "learning_rate": 4.311660768053304e-05,
+      "loss": 0.1786,
+      "step": 19320
+    },
+    {
+      "epoch": 1.593310632688928,
+      "grad_norm": 0.19174990057945251,
+      "learning_rate": 4.309677843089364e-05,
+      "loss": 0.1813,
+      "step": 19340
+    },
+    {
+      "epoch": 1.5949582601054482,
+      "grad_norm": 4.260095119476318,
+      "learning_rate": 4.307692523477531e-05,
+      "loss": 0.1542,
+      "step": 19360
+    },
+    {
+      "epoch": 1.5966058875219684,
+      "grad_norm": 1.4479669332504272,
+      "learning_rate": 4.305704811844871e-05,
+      "loss": 0.1803,
+      "step": 19380
+    },
+    {
+      "epoch": 1.5982535149384884,
+      "grad_norm": 1.0454448461532593,
+      "learning_rate": 4.303714710821613e-05,
+      "loss": 0.1877,
+      "step": 19400
+    },
+    {
+      "epoch": 1.5999011423550087,
+      "grad_norm": 1.397001028060913,
+      "learning_rate": 4.301722223041151e-05,
+      "loss": 0.182,
+      "step": 19420
+    },
+    {
+      "epoch": 1.601548769771529,
+      "grad_norm": 0.5425103902816772,
+      "learning_rate": 4.299827151317217e-05,
+      "loss": 0.1793,
+      "step": 19440
+    },
+    {
+      "epoch": 1.6031963971880492,
+      "grad_norm": 1.6720526218414307,
+      "learning_rate": 4.297830016946455e-05,
+      "loss": 0.1804,
+      "step": 19460
+    },
+    {
+      "epoch": 1.6048440246045694,
+      "grad_norm": 0.6204736232757568,
+      "learning_rate": 4.295830503605383e-05,
+      "loss": 0.1906,
+      "step": 19480
+    },
+    {
+      "epoch": 1.6064916520210897,
+      "grad_norm": 2.814265727996826,
+      "learning_rate": 4.293828613939845e-05,
+      "loss": 0.1687,
+      "step": 19500
+    },
+    {
+      "epoch": 1.60813927943761,
+      "grad_norm": 2.273043394088745,
+      "learning_rate": 4.2918243505988354e-05,
+      "loss": 0.1837,
+      "step": 19520
+    },
+    {
+      "epoch": 1.6097869068541302,
+      "grad_norm": 1.504699945449829,
+      "learning_rate": 4.289817716234485e-05,
+      "loss": 0.1715,
+      "step": 19540
+    },
+    {
+      "epoch": 1.6114345342706504,
+      "grad_norm": 2.6365151405334473,
+      "learning_rate": 4.287808713502063e-05,
+      "loss": 0.1655,
+      "step": 19560
+    },
+    {
+      "epoch": 1.6130821616871704,
+      "grad_norm": 0.9557230472564697,
+      "learning_rate": 4.2857973450599744e-05,
+      "loss": 0.1628,
+      "step": 19580
+    },
+    {
+      "epoch": 1.6147297891036907,
+      "grad_norm": 0.9700766801834106,
+      "learning_rate": 4.283783613569752e-05,
+      "loss": 0.1912,
+      "step": 19600
+    },
+    {
+      "epoch": 1.616377416520211,
+      "grad_norm": 1.2748225927352905,
+      "learning_rate": 4.281767521696057e-05,
+      "loss": 0.1788,
+      "step": 19620
+    },
+    {
+      "epoch": 1.618025043936731,
+      "grad_norm": 0.2875102460384369,
+      "learning_rate": 4.279749072106675e-05,
+      "loss": 0.1726,
+      "step": 19640
+    },
+    {
+      "epoch": 1.6196726713532512,
+      "grad_norm": 2.1846017837524414,
+      "learning_rate": 4.2777282674725086e-05,
+      "loss": 0.1544,
+      "step": 19660
+    },
+    {
+      "epoch": 1.6213202987697715,
+      "grad_norm": 1.5533403158187866,
+      "learning_rate": 4.2757051104675786e-05,
+      "loss": 0.1621,
+      "step": 19680
+    },
+    {
+      "epoch": 1.6229679261862917,
+      "grad_norm": 6.069032669067383,
+      "learning_rate": 4.273780934867807e-05,
+      "loss": 0.1855,
+      "step": 19700
+    },
+    {
+      "epoch": 1.624615553602812,
+      "grad_norm": 1.1619160175323486,
+      "learning_rate": 4.271753198442825e-05,
+      "loss": 0.1821,
+      "step": 19720
+    },
+    {
+      "epoch": 1.6262631810193322,
+      "grad_norm": 0.8923582434654236,
+      "learning_rate": 4.269723117553562e-05,
+      "loss": 0.1861,
+      "step": 19740
+    },
+    {
+      "epoch": 1.6279108084358525,
+      "grad_norm": 4.761159420013428,
+      "learning_rate": 4.2676906948863134e-05,
+      "loss": 0.147,
+      "step": 19760
+    },
+    {
+      "epoch": 1.6295584358523727,
+      "grad_norm": 3.785452127456665,
+      "learning_rate": 4.265655933130473e-05,
+      "loss": 0.1789,
+      "step": 19780
+    },
+    {
+      "epoch": 1.631206063268893,
+      "grad_norm": 7.1180806159973145,
+      "learning_rate": 4.26361883497853e-05,
+      "loss": 0.1867,
+      "step": 19800
+    },
+    {
+      "epoch": 1.632853690685413,
+      "grad_norm": 0.7360396981239319,
+      "learning_rate": 4.2615794031260655e-05,
+      "loss": 0.1789,
+      "step": 19820
+    },
+    {
+      "epoch": 1.6345013181019332,
+      "grad_norm": 0.38821306824684143,
+      "learning_rate": 4.259537640271749e-05,
+      "loss": 0.1824,
+      "step": 19840
+    },
+    {
+      "epoch": 1.6361489455184535,
+      "grad_norm": 1.545905590057373,
+      "learning_rate": 4.2574935491173325e-05,
+      "loss": 0.1436,
+      "step": 19860
+    },
+    {
+      "epoch": 1.6377965729349735,
+      "grad_norm": 2.462200880050659,
+      "learning_rate": 4.2554471323676505e-05,
+      "loss": 0.1876,
+      "step": 19880
+    },
+    {
+      "epoch": 1.6394442003514937,
+      "grad_norm": 1.7179991006851196,
+      "learning_rate": 4.253398392730614e-05,
+      "loss": 0.1744,
+      "step": 19900
+    },
+    {
+      "epoch": 1.641091827768014,
+      "grad_norm": 1.1609716415405273,
+      "learning_rate": 4.2513473329172106e-05,
+      "loss": 0.1865,
+      "step": 19920
+    },
+    {
+      "epoch": 1.6427394551845342,
+      "grad_norm": 0.445671945810318,
+      "learning_rate": 4.249293955641493e-05,
+      "loss": 0.1842,
+      "step": 19940
+    },
+    {
+      "epoch": 1.6443870826010545,
+      "grad_norm": 1.0819984674453735,
+      "learning_rate": 4.2472382636205845e-05,
+      "loss": 0.1583,
+      "step": 19960
+    },
+    {
+      "epoch": 1.6460347100175747,
+      "grad_norm": 2.5188121795654297,
+      "learning_rate": 4.24518025957467e-05,
+      "loss": 0.1861,
+      "step": 19980
+    },
+    {
+      "epoch": 1.647682337434095,
+      "grad_norm": 0.8718958497047424,
+      "learning_rate": 4.243119946226996e-05,
+      "loss": 0.1842,
+      "step": 20000
+    },
+    {
+      "epoch": 1.647682337434095,
+      "eval_loss": 0.5287063121795654,
+      "eval_runtime": 251.8074,
+      "eval_samples_per_second": 86.53,
+      "eval_steps_per_second": 21.636,
+      "eval_wer": 0.23113573314148306,
+      "step": 20000
+    },
+    {
+      "epoch": 1.6493299648506152,
+      "grad_norm": 1.2543200254440308,
+      "learning_rate": 4.2410573263038583e-05,
+      "loss": 0.1729,
+      "step": 20020
+    },
+    {
+      "epoch": 1.6509775922671355,
+      "grad_norm": 0.2878034710884094,
+      "learning_rate": 4.238992402534613e-05,
+      "loss": 0.1789,
+      "step": 20040
+    },
+    {
+      "epoch": 1.6526252196836555,
+      "grad_norm": 2.4613733291625977,
+      "learning_rate": 4.236925177651658e-05,
+      "loss": 0.1631,
+      "step": 20060
+    },
+    {
+      "epoch": 1.6542728471001757,
+      "grad_norm": 0.9280514121055603,
+      "learning_rate": 4.234855654390442e-05,
+      "loss": 0.1741,
+      "step": 20080
+    },
+    {
+      "epoch": 1.655920474516696,
+      "grad_norm": 1.553101897239685,
+      "learning_rate": 4.232783835489451e-05,
+      "loss": 0.1665,
+      "step": 20100
+    },
+    {
+      "epoch": 1.657568101933216,
+      "grad_norm": 0.5047245621681213,
+      "learning_rate": 4.23070972369021e-05,
+      "loss": 0.1882,
+      "step": 20120
+    },
+    {
+      "epoch": 1.6592157293497363,
+      "grad_norm": 0.32549336552619934,
+      "learning_rate": 4.228633321737276e-05,
+      "loss": 0.1716,
+      "step": 20140
+    },
+    {
+      "epoch": 1.6608633567662565,
+      "grad_norm": 2.7314305305480957,
+      "learning_rate": 4.226554632378241e-05,
+      "loss": 0.1551,
+      "step": 20160
+    },
+    {
+      "epoch": 1.6625109841827768,
+      "grad_norm": 1.0868523120880127,
+      "learning_rate": 4.2244736583637214e-05,
+      "loss": 0.1674,
+      "step": 20180
+    },
+    {
+      "epoch": 1.664158611599297,
+      "grad_norm": 0.6131607294082642,
+      "learning_rate": 4.222390402447353e-05,
+      "loss": 0.1832,
+      "step": 20200
+    },
+    {
+      "epoch": 1.6658062390158173,
+      "grad_norm": 0.8340827226638794,
+      "learning_rate": 4.220304867385798e-05,
+      "loss": 0.1695,
+      "step": 20220
+    },
+    {
+      "epoch": 1.6674538664323375,
+      "grad_norm": 0.22361576557159424,
+      "learning_rate": 4.21821705593873e-05,
+      "loss": 0.1935,
+      "step": 20240
+    },
+    {
+      "epoch": 1.6691014938488578,
+      "grad_norm": 1.1089187860488892,
+      "learning_rate": 4.216126970868836e-05,
+      "loss": 0.1536,
+      "step": 20260
+    },
+    {
+      "epoch": 1.670749121265378,
+      "grad_norm": 0.504047691822052,
+      "learning_rate": 4.214034614941811e-05,
+      "loss": 0.1758,
+      "step": 20280
+    },
+    {
+      "epoch": 1.672396748681898,
+      "grad_norm": 0.8742786049842834,
+      "learning_rate": 4.211939990926355e-05,
+      "loss": 0.1843,
+      "step": 20300
+    },
+    {
+      "epoch": 1.6740443760984183,
+      "grad_norm": 0.856685221195221,
+      "learning_rate": 4.209843101594171e-05,
+      "loss": 0.1708,
+      "step": 20320
+    },
+    {
+      "epoch": 1.6756920035149385,
+      "grad_norm": 0.20526465773582458,
+      "learning_rate": 4.2077439497199576e-05,
+      "loss": 0.1699,
+      "step": 20340
+    },
+    {
+      "epoch": 1.6773396309314585,
+      "grad_norm": 0.9350545406341553,
+      "learning_rate": 4.2056425380814074e-05,
+      "loss": 0.1503,
+      "step": 20360
+    },
+    {
+      "epoch": 1.6789872583479788,
+      "grad_norm": 2.845334529876709,
+      "learning_rate": 4.203538869459204e-05,
+      "loss": 0.1708,
+      "step": 20380
+    },
+    {
+      "epoch": 1.680634885764499,
+      "grad_norm": 1.2838850021362305,
+      "learning_rate": 4.2014329466370176e-05,
+      "loss": 0.18,
+      "step": 20400
+    },
+    {
+      "epoch": 1.6822825131810193,
+      "grad_norm": 1.8057399988174438,
+      "learning_rate": 4.1993247724015e-05,
+      "loss": 0.173,
+      "step": 20420
+    },
+    {
+      "epoch": 1.6839301405975395,
+      "grad_norm": 0.3127855658531189,
+      "learning_rate": 4.1972143495422844e-05,
+      "loss": 0.1728,
+      "step": 20440
+    },
+    {
+      "epoch": 1.6855777680140598,
+      "grad_norm": 2.700399398803711,
+      "learning_rate": 4.195101680851977e-05,
+      "loss": 0.1578,
+      "step": 20460
+    },
+    {
+      "epoch": 1.68722539543058,
+      "grad_norm": 0.7250802516937256,
+      "learning_rate": 4.192986769126158e-05,
+      "loss": 0.1823,
+      "step": 20480
+    },
+    {
+      "epoch": 1.6888730228471003,
+      "grad_norm": 1.1070244312286377,
+      "learning_rate": 4.1908696171633734e-05,
+      "loss": 0.189,
+      "step": 20500
+    },
+    {
+      "epoch": 1.6905206502636205,
+      "grad_norm": 1.0246353149414062,
+      "learning_rate": 4.188750227765135e-05,
+      "loss": 0.1902,
+      "step": 20520
+    },
+    {
+      "epoch": 1.6921682776801406,
+      "grad_norm": 0.6496437191963196,
+      "learning_rate": 4.186628603735916e-05,
+      "loss": 0.1747,
+      "step": 20540
+    },
+    {
+      "epoch": 1.6938159050966608,
+      "grad_norm": 0.8663806319236755,
+      "learning_rate": 4.184504747883145e-05,
+      "loss": 0.1519,
+      "step": 20560
+    },
+    {
+      "epoch": 1.695463532513181,
+      "grad_norm": 0.7036964893341064,
+      "learning_rate": 4.182378663017205e-05,
+      "loss": 0.1829,
+      "step": 20580
+    },
+    {
+      "epoch": 1.697111159929701,
+      "grad_norm": 0.7158936858177185,
+      "learning_rate": 4.180250351951427e-05,
+      "loss": 0.1672,
+      "step": 20600
+    },
+    {
+      "epoch": 1.6987587873462213,
+      "grad_norm": 0.5382769703865051,
+      "learning_rate": 4.17811981750209e-05,
+      "loss": 0.1804,
+      "step": 20620
+    },
+    {
+      "epoch": 1.7004064147627416,
+      "grad_norm": 0.5654181838035583,
+      "learning_rate": 4.175987062488413e-05,
+      "loss": 0.1585,
+      "step": 20640
+    },
+    {
+      "epoch": 1.7020540421792618,
+      "grad_norm": 1.4323805570602417,
+      "learning_rate": 4.1738520897325544e-05,
+      "loss": 0.1532,
+      "step": 20660
+    },
+    {
+      "epoch": 1.703701669595782,
+      "grad_norm": 0.7951692938804626,
+      "learning_rate": 4.171714902059608e-05,
+      "loss": 0.1755,
+      "step": 20680
+    },
+    {
+      "epoch": 1.7053492970123023,
+      "grad_norm": 0.7378861308097839,
+      "learning_rate": 4.169575502297597e-05,
+      "loss": 0.1814,
+      "step": 20700
+    },
+    {
+      "epoch": 1.7069969244288226,
+      "grad_norm": 0.6902081966400146,
+      "learning_rate": 4.1674338932774724e-05,
+      "loss": 0.1639,
+      "step": 20720
+    },
+    {
+      "epoch": 1.7086445518453428,
+      "grad_norm": 0.16693931818008423,
+      "learning_rate": 4.1652900778331076e-05,
+      "loss": 0.1788,
+      "step": 20740
+    },
+    {
+      "epoch": 1.710292179261863,
+      "grad_norm": 2.9214699268341064,
+      "learning_rate": 4.1631440588012974e-05,
+      "loss": 0.1562,
+      "step": 20760
+    },
+    {
+      "epoch": 1.711939806678383,
+      "grad_norm": 0.8093401193618774,
+      "learning_rate": 4.1609958390217516e-05,
+      "loss": 0.1714,
+      "step": 20780
+    },
+    {
+      "epoch": 1.7135874340949033,
+      "grad_norm": 1.0199005603790283,
+      "learning_rate": 4.158845421337093e-05,
+      "loss": 0.1869,
+      "step": 20800
+    },
+    {
+      "epoch": 1.7152350615114236,
+      "grad_norm": 0.5902179479598999,
+      "learning_rate": 4.1566928085928504e-05,
+      "loss": 0.1868,
+      "step": 20820
+    },
+    {
+      "epoch": 1.7168826889279436,
+      "grad_norm": 0.30310845375061035,
+      "learning_rate": 4.154538003637459e-05,
+      "loss": 0.1851,
+      "step": 20840
+    },
+    {
+      "epoch": 1.7185303163444638,
+      "grad_norm": 4.026705741882324,
+      "learning_rate": 4.1523810093222547e-05,
+      "loss": 0.1667,
+      "step": 20860
+    },
+    {
+      "epoch": 1.720177943760984,
+      "grad_norm": 0.82432621717453,
+      "learning_rate": 4.1502218285014714e-05,
+      "loss": 0.1785,
+      "step": 20880
+    },
+    {
+      "epoch": 1.7218255711775043,
+      "grad_norm": 1.0857402086257935,
+      "learning_rate": 4.148060464032234e-05,
+      "loss": 0.1867,
+      "step": 20900
+    },
+    {
+      "epoch": 1.7234731985940246,
+      "grad_norm": 1.142042636871338,
+      "learning_rate": 4.145896918774558e-05,
+      "loss": 0.1744,
+      "step": 20920
+    },
+    {
+      "epoch": 1.7251208260105448,
+      "grad_norm": 0.4160640835762024,
+      "learning_rate": 4.143731195591345e-05,
+      "loss": 0.1801,
+      "step": 20940
+    },
+    {
+      "epoch": 1.726768453427065,
+      "grad_norm": 1.8969520330429077,
+      "learning_rate": 4.141563297348378e-05,
+      "loss": 0.1536,
+      "step": 20960
+    },
+    {
+      "epoch": 1.7284160808435853,
+      "grad_norm": 0.7605903744697571,
+      "learning_rate": 4.1393932269143206e-05,
+      "loss": 0.1777,
+      "step": 20980
+    },
+    {
+      "epoch": 1.7300637082601056,
+      "grad_norm": 1.4995478391647339,
+      "learning_rate": 4.137220987160705e-05,
+      "loss": 0.184,
+      "step": 21000
+    },
+    {
+      "epoch": 1.7317113356766256,
+      "grad_norm": 1.7079527378082275,
+      "learning_rate": 4.13504658096194e-05,
+      "loss": 0.1799,
+      "step": 21020
+    },
+    {
+      "epoch": 1.7333589630931459,
+      "grad_norm": 0.3677631616592407,
+      "learning_rate": 4.132870011195298e-05,
+      "loss": 0.1718,
+      "step": 21040
+    },
+    {
+      "epoch": 1.735006590509666,
+      "grad_norm": 5.215697288513184,
+      "learning_rate": 4.1306912807409134e-05,
+      "loss": 0.1515,
+      "step": 21060
+    },
+    {
+      "epoch": 1.7366542179261861,
+      "grad_norm": 2.6421728134155273,
+      "learning_rate": 4.1285103924817836e-05,
+      "loss": 0.1877,
+      "step": 21080
+    },
+    {
+      "epoch": 1.7383018453427064,
+      "grad_norm": 2.3778457641601562,
+      "learning_rate": 4.126327349303758e-05,
+      "loss": 0.1818,
+      "step": 21100
+    },
+    {
+      "epoch": 1.7399494727592266,
+      "grad_norm": 2.7478296756744385,
+      "learning_rate": 4.124142154095539e-05,
+      "loss": 0.1791,
+      "step": 21120
+    },
+    {
+      "epoch": 1.7415971001757469,
+      "grad_norm": 0.5207391381263733,
+      "learning_rate": 4.121954809748675e-05,
+      "loss": 0.1644,
+      "step": 21140
+    },
+    {
+      "epoch": 1.7432447275922671,
+      "grad_norm": 3.1791322231292725,
+      "learning_rate": 4.119765319157559e-05,
+      "loss": 0.1652,
+      "step": 21160
+    },
+    {
+      "epoch": 1.7448923550087874,
+      "grad_norm": 4.9425764083862305,
+      "learning_rate": 4.117573685219427e-05,
+      "loss": 0.1841,
+      "step": 21180
+    },
+    {
+      "epoch": 1.7465399824253076,
+      "grad_norm": 0.698239266872406,
+      "learning_rate": 4.115379910834345e-05,
+      "loss": 0.1775,
+      "step": 21200
+    },
+    {
+      "epoch": 1.7481876098418279,
+      "grad_norm": 3.20509934425354,
+      "learning_rate": 4.113183998905218e-05,
+      "loss": 0.1703,
+      "step": 21220
+    },
+    {
+      "epoch": 1.749835237258348,
+      "grad_norm": 0.2055424153804779,
+      "learning_rate": 4.110985952337775e-05,
+      "loss": 0.1679,
+      "step": 21240
+    },
+    {
+      "epoch": 1.7514828646748684,
+      "grad_norm": 2.4991447925567627,
+      "learning_rate": 4.108785774040571e-05,
+      "loss": 0.1528,
+      "step": 21260
+    },
+    {
+      "epoch": 1.7531304920913884,
+      "grad_norm": 1.0346381664276123,
+      "learning_rate": 4.106583466924981e-05,
+      "loss": 0.1735,
+      "step": 21280
+    },
+    {
+      "epoch": 1.7547781195079086,
+      "grad_norm": 1.605350136756897,
+      "learning_rate": 4.104379033905199e-05,
+      "loss": 0.1765,
+      "step": 21300
+    },
+    {
+      "epoch": 1.7564257469244289,
+      "grad_norm": 0.4309673309326172,
+      "learning_rate": 4.102172477898232e-05,
+      "loss": 0.1841,
+      "step": 21320
+    },
+    {
+      "epoch": 1.758073374340949,
+      "grad_norm": 0.5647534132003784,
+      "learning_rate": 4.099963801823892e-05,
+      "loss": 0.1606,
+      "step": 21340
+    },
+    {
+      "epoch": 1.7597210017574691,
+      "grad_norm": 2.7488176822662354,
+      "learning_rate": 4.0977530086048025e-05,
+      "loss": 0.1562,
+      "step": 21360
+    },
+    {
+      "epoch": 1.7613686291739894,
+      "grad_norm": 0.6944543719291687,
+      "learning_rate": 4.095540101166383e-05,
+      "loss": 0.1757,
+      "step": 21380
+    },
+    {
+      "epoch": 1.7630162565905096,
+      "grad_norm": 2.5942211151123047,
+      "learning_rate": 4.093325082436854e-05,
+      "loss": 0.1852,
+      "step": 21400
+    },
+    {
+      "epoch": 1.76466388400703,
+      "grad_norm": 0.6985669732093811,
+      "learning_rate": 4.09110795534723e-05,
+      "loss": 0.1801,
+      "step": 21420
+    },
+    {
+      "epoch": 1.7663115114235501,
+      "grad_norm": 0.5788651704788208,
+      "learning_rate": 4.088888722831312e-05,
+      "loss": 0.1841,
+      "step": 21440
+    },
+    {
+      "epoch": 1.7679591388400704,
+      "grad_norm": 1.5852128267288208,
+      "learning_rate": 4.0866673878256925e-05,
+      "loss": 0.1632,
+      "step": 21460
+    },
+    {
+      "epoch": 1.7696067662565906,
+      "grad_norm": 1.0001317262649536,
+      "learning_rate": 4.0844439532697386e-05,
+      "loss": 0.1828,
+      "step": 21480
+    },
+    {
+      "epoch": 1.7712543936731109,
+      "grad_norm": 3.1319148540496826,
+      "learning_rate": 4.0822184221056036e-05,
+      "loss": 0.1797,
+      "step": 21500
+    },
+    {
+      "epoch": 1.772902021089631,
+      "grad_norm": 1.2414968013763428,
+      "learning_rate": 4.0799907972782105e-05,
+      "loss": 0.172,
+      "step": 21520
+    },
+    {
+      "epoch": 1.7745496485061512,
+      "grad_norm": 0.37049078941345215,
+      "learning_rate": 4.077761081735252e-05,
+      "loss": 0.1836,
+      "step": 21540
+    },
+    {
+      "epoch": 1.7761972759226714,
+      "grad_norm": 2.273507595062256,
+      "learning_rate": 4.0755292784271896e-05,
+      "loss": 0.1663,
+      "step": 21560
+    },
+    {
+      "epoch": 1.7778449033391914,
+      "grad_norm": 1.669340968132019,
+      "learning_rate": 4.073295390307248e-05,
+      "loss": 0.1789,
+      "step": 21580
+    },
+    {
+      "epoch": 1.7794925307557117,
+      "grad_norm": 0.8448110222816467,
+      "learning_rate": 4.071059420331408e-05,
+      "loss": 0.1927,
+      "step": 21600
+    },
+    {
+      "epoch": 1.781140158172232,
+      "grad_norm": 2.4922406673431396,
+      "learning_rate": 4.0688213714584076e-05,
+      "loss": 0.1852,
+      "step": 21620
+    },
+    {
+      "epoch": 1.7827877855887522,
+      "grad_norm": 0.36666399240493774,
+      "learning_rate": 4.066581246649734e-05,
+      "loss": 0.1869,
+      "step": 21640
+    },
+    {
+      "epoch": 1.7844354130052724,
+      "grad_norm": 1.3697789907455444,
+      "learning_rate": 4.0643390488696225e-05,
+      "loss": 0.159,
+      "step": 21660
+    },
+    {
+      "epoch": 1.7860830404217927,
+      "grad_norm": 0.6621468663215637,
+      "learning_rate": 4.062094781085051e-05,
+      "loss": 0.1827,
+      "step": 21680
+    },
+    {
+      "epoch": 1.787730667838313,
+      "grad_norm": 3.3689136505126953,
+      "learning_rate": 4.059848446265736e-05,
+      "loss": 0.1829,
+      "step": 21700
+    },
+    {
+      "epoch": 1.7893782952548332,
+      "grad_norm": 1.2748357057571411,
+      "learning_rate": 4.0576000473841295e-05,
+      "loss": 0.1814,
+      "step": 21720
+    },
+    {
+      "epoch": 1.7910259226713534,
+      "grad_norm": 0.7753732800483704,
+      "learning_rate": 4.0553495874154167e-05,
+      "loss": 0.1837,
+      "step": 21740
+    },
+    {
+      "epoch": 1.7926735500878734,
+      "grad_norm": 4.699821472167969,
+      "learning_rate": 4.053097069337507e-05,
+      "loss": 0.1608,
+      "step": 21760
+    },
+    {
+      "epoch": 1.7943211775043937,
+      "grad_norm": 0.6081763505935669,
+      "learning_rate": 4.050842496131034e-05,
+      "loss": 0.1802,
+      "step": 21780
+    },
+    {
+      "epoch": 1.795968804920914,
+      "grad_norm": 1.5527398586273193,
+      "learning_rate": 4.0485858707793534e-05,
+      "loss": 0.186,
+      "step": 21800
+    },
+    {
+      "epoch": 1.797616432337434,
+      "grad_norm": 0.7068532109260559,
+      "learning_rate": 4.046327196268533e-05,
+      "loss": 0.1653,
+      "step": 21820
+    },
+    {
+      "epoch": 1.7992640597539542,
+      "grad_norm": 1.8269671201705933,
+      "learning_rate": 4.044066475587353e-05,
+      "loss": 0.1799,
+      "step": 21840
+    },
+    {
+      "epoch": 1.8009116871704745,
+      "grad_norm": 3.3052139282226562,
+      "learning_rate": 4.041803711727303e-05,
+      "loss": 0.1502,
+      "step": 21860
+    },
+    {
+      "epoch": 1.8025593145869947,
+      "grad_norm": 0.9614394307136536,
+      "learning_rate": 4.039538907682574e-05,
+      "loss": 0.1739,
+      "step": 21880
+    },
+    {
+      "epoch": 1.804206942003515,
+      "grad_norm": 1.014877438545227,
+      "learning_rate": 4.0372720664500575e-05,
+      "loss": 0.1829,
+      "step": 21900
+    },
+    {
+      "epoch": 1.8058545694200352,
+      "grad_norm": 1.322003960609436,
+      "learning_rate": 4.035003191029341e-05,
+      "loss": 0.1935,
+      "step": 21920
+    },
+    {
+      "epoch": 1.8075021968365554,
+      "grad_norm": 0.6618562340736389,
+      "learning_rate": 4.0328458779473224e-05,
+      "loss": 0.1741,
+      "step": 21940
+    },
+    {
+      "epoch": 1.8091498242530757,
+      "grad_norm": 1.402084469795227,
+      "learning_rate": 4.0305730444973686e-05,
+      "loss": 0.1621,
+      "step": 21960
+    },
+    {
+      "epoch": 1.810797451669596,
+      "grad_norm": 1.3006558418273926,
+      "learning_rate": 4.0282981857236654e-05,
+      "loss": 0.1674,
+      "step": 21980
+    },
+    {
+      "epoch": 1.812445079086116,
+      "grad_norm": 1.6449670791625977,
+      "learning_rate": 4.026021304636408e-05,
+      "loss": 0.1845,
+      "step": 22000
+    },
+    {
+      "epoch": 1.812445079086116,
+      "eval_loss": 0.5395076274871826,
+      "eval_runtime": 246.9697,
+      "eval_samples_per_second": 88.225,
+      "eval_steps_per_second": 22.059,
+      "eval_wer": 0.2258113279380107,
+      "step": 22000
+    },
+    {
+      "epoch": 1.8140927065026362,
+      "grad_norm": 0.5984182953834534,
+      "learning_rate": 4.0237424042484694e-05,
+      "loss": 0.18,
+      "step": 22020
+    },
+    {
+      "epoch": 1.8157403339191565,
+      "grad_norm": 1.0855306386947632,
+      "learning_rate": 4.0214614875753955e-05,
+      "loss": 0.1651,
+      "step": 22040
+    },
+    {
+      "epoch": 1.8173879613356765,
+      "grad_norm": 1.4949283599853516,
+      "learning_rate": 4.0191785576353956e-05,
+      "loss": 0.1563,
+      "step": 22060
+    },
+    {
+      "epoch": 1.8190355887521967,
+      "grad_norm": 0.9411268830299377,
+      "learning_rate": 4.016893617449349e-05,
+      "loss": 0.1807,
+      "step": 22080
+    },
+    {
+      "epoch": 1.820683216168717,
+      "grad_norm": 3.1195433139801025,
+      "learning_rate": 4.014606670040791e-05,
+      "loss": 0.1611,
+      "step": 22100
+    },
+    {
+      "epoch": 1.8223308435852372,
+      "grad_norm": 1.749937653541565,
+      "learning_rate": 4.012317718435915e-05,
+      "loss": 0.1809,
+      "step": 22120
+    },
+    {
+      "epoch": 1.8239784710017575,
+      "grad_norm": 0.41568905115127563,
+      "learning_rate": 4.010026765663566e-05,
+      "loss": 0.1754,
+      "step": 22140
+    },
+    {
+      "epoch": 1.8256260984182777,
+      "grad_norm": 1.6782418489456177,
+      "learning_rate": 4.007733814755235e-05,
+      "loss": 0.1577,
+      "step": 22160
+    },
+    {
+      "epoch": 1.827273725834798,
+      "grad_norm": 1.470269799232483,
+      "learning_rate": 4.005438868745059e-05,
+      "loss": 0.1804,
+      "step": 22180
+    },
+    {
+      "epoch": 1.8289213532513182,
+      "grad_norm": 0.8983897566795349,
+      "learning_rate": 4.003141930669818e-05,
+      "loss": 0.1771,
+      "step": 22200
+    },
+    {
+      "epoch": 1.8305689806678385,
+      "grad_norm": 0.6398609280586243,
+      "learning_rate": 4.000843003568921e-05,
+      "loss": 0.1937,
+      "step": 22220
+    },
+    {
+      "epoch": 1.8322166080843585,
+      "grad_norm": 0.20628367364406586,
+      "learning_rate": 3.998542090484414e-05,
+      "loss": 0.1804,
+      "step": 22240
+    },
+    {
+      "epoch": 1.8338642355008787,
+      "grad_norm": 1.640540361404419,
+      "learning_rate": 3.9962391944609704e-05,
+      "loss": 0.1576,
+      "step": 22260
+    },
+    {
+      "epoch": 1.835511862917399,
+      "grad_norm": 0.5554460883140564,
+      "learning_rate": 3.9939343185458856e-05,
+      "loss": 0.1629,
+      "step": 22280
+    },
+    {
+      "epoch": 1.837159490333919,
+      "grad_norm": 0.5767085552215576,
+      "learning_rate": 3.991627465789076e-05,
+      "loss": 0.1779,
+      "step": 22300
+    },
+    {
+      "epoch": 1.8388071177504393,
+      "grad_norm": 0.3550276458263397,
+      "learning_rate": 3.989318639243075e-05,
+      "loss": 0.1812,
+      "step": 22320
+    },
+    {
+      "epoch": 1.8404547451669595,
+      "grad_norm": 0.23671790957450867,
+      "learning_rate": 3.9870078419630276e-05,
+      "loss": 0.1717,
+      "step": 22340
+    },
+    {
+      "epoch": 1.8421023725834798,
+      "grad_norm": 1.2688475847244263,
+      "learning_rate": 3.984695077006683e-05,
+      "loss": 0.1606,
+      "step": 22360
+    },
+    {
+      "epoch": 1.84375,
+      "grad_norm": 2.13102388381958,
+      "learning_rate": 3.982380347434398e-05,
+      "loss": 0.1652,
+      "step": 22380
+    },
+    {
+      "epoch": 1.8453976274165202,
+      "grad_norm": 0.6820785403251648,
+      "learning_rate": 3.980063656309128e-05,
+      "loss": 0.1832,
+      "step": 22400
+    },
+    {
+      "epoch": 1.8470452548330405,
+      "grad_norm": 1.0797230005264282,
+      "learning_rate": 3.9777450066964236e-05,
+      "loss": 0.1673,
+      "step": 22420
+    },
+    {
+      "epoch": 1.8486928822495607,
+      "grad_norm": 0.22749900817871094,
+      "learning_rate": 3.9754244016644286e-05,
+      "loss": 0.1935,
+      "step": 22440
+    },
+    {
+      "epoch": 1.850340509666081,
+      "grad_norm": 21.42913246154785,
+      "learning_rate": 3.973101844283871e-05,
+      "loss": 0.1616,
+      "step": 22460
+    },
+    {
+      "epoch": 1.851988137082601,
+      "grad_norm": 1.0109410285949707,
+      "learning_rate": 3.970777337628066e-05,
+      "loss": 0.1733,
+      "step": 22480
+    },
+    {
+      "epoch": 1.8536357644991213,
+      "grad_norm": 1.8864178657531738,
+      "learning_rate": 3.968450884772905e-05,
+      "loss": 0.191,
+      "step": 22500
+    },
+    {
+      "epoch": 1.8552833919156415,
+      "grad_norm": 1.0088257789611816,
+      "learning_rate": 3.9661224887968564e-05,
+      "loss": 0.179,
+      "step": 22520
+    },
+    {
+      "epoch": 1.8569310193321615,
+      "grad_norm": 0.27392879128456116,
+      "learning_rate": 3.9637921527809594e-05,
+      "loss": 0.1883,
+      "step": 22540
+    },
+    {
+      "epoch": 1.8585786467486818,
+      "grad_norm": 1.0473995208740234,
+      "learning_rate": 3.9614598798088205e-05,
+      "loss": 0.1568,
+      "step": 22560
+    },
+    {
+      "epoch": 1.860226274165202,
+      "grad_norm": 2.516446352005005,
+      "learning_rate": 3.959125672966609e-05,
+      "loss": 0.179,
+      "step": 22580
+    },
+    {
+      "epoch": 1.8618739015817223,
+      "grad_norm": 0.5339066982269287,
+      "learning_rate": 3.956789535343054e-05,
+      "loss": 0.1886,
+      "step": 22600
+    },
+    {
+      "epoch": 1.8635215289982425,
+      "grad_norm": 3.4075698852539062,
+      "learning_rate": 3.954451470029439e-05,
+      "loss": 0.1723,
+      "step": 22620
+    },
+    {
+      "epoch": 1.8651691564147628,
+      "grad_norm": 0.4262460172176361,
+      "learning_rate": 3.952111480119597e-05,
+      "loss": 0.1774,
+      "step": 22640
+    },
+    {
+      "epoch": 1.866816783831283,
+      "grad_norm": 1.0418535470962524,
+      "learning_rate": 3.94976956870991e-05,
+      "loss": 0.1613,
+      "step": 22660
+    },
+    {
+      "epoch": 1.8684644112478033,
+      "grad_norm": 1.1704654693603516,
+      "learning_rate": 3.947425738899302e-05,
+      "loss": 0.1879,
+      "step": 22680
+    },
+    {
+      "epoch": 1.8701120386643235,
+      "grad_norm": 1.1728986501693726,
+      "learning_rate": 3.9450799937892324e-05,
+      "loss": 0.1655,
+      "step": 22700
+    },
+    {
+      "epoch": 1.8717596660808435,
+      "grad_norm": 1.5375456809997559,
+      "learning_rate": 3.9427323364837e-05,
+      "loss": 0.1865,
+      "step": 22720
+    },
+    {
+      "epoch": 1.8734072934973638,
+      "grad_norm": 0.16443948447704315,
+      "learning_rate": 3.940382770089232e-05,
+      "loss": 0.179,
+      "step": 22740
+    },
+    {
+      "epoch": 1.875054920913884,
+      "grad_norm": 13.896778106689453,
+      "learning_rate": 3.9380312977148795e-05,
+      "loss": 0.1495,
+      "step": 22760
+    },
+    {
+      "epoch": 1.876702548330404,
+      "grad_norm": 2.690234422683716,
+      "learning_rate": 3.935677922472218e-05,
+      "loss": 0.1715,
+      "step": 22780
+    },
+    {
+      "epoch": 1.8783501757469243,
+      "grad_norm": 2.224005937576294,
+      "learning_rate": 3.933322647475342e-05,
+      "loss": 0.1753,
+      "step": 22800
+    },
+    {
+      "epoch": 1.8799978031634446,
+      "grad_norm": 1.617680311203003,
+      "learning_rate": 3.9309654758408556e-05,
+      "loss": 0.1773,
+      "step": 22820
+    },
+    {
+      "epoch": 1.8816454305799648,
+      "grad_norm": 0.17472465336322784,
+      "learning_rate": 3.928606410687877e-05,
+      "loss": 0.174,
+      "step": 22840
+    },
+    {
+      "epoch": 1.883293057996485,
+      "grad_norm": 2.1024489402770996,
+      "learning_rate": 3.926245455138029e-05,
+      "loss": 0.1486,
+      "step": 22860
+    },
+    {
+      "epoch": 1.8849406854130053,
+      "grad_norm": 3.535928726196289,
+      "learning_rate": 3.923882612315434e-05,
+      "loss": 0.1917,
+      "step": 22880
+    },
+    {
+      "epoch": 1.8865883128295255,
+      "grad_norm": 1.4998210668563843,
+      "learning_rate": 3.9215178853467135e-05,
+      "loss": 0.1786,
+      "step": 22900
+    },
+    {
+      "epoch": 1.8882359402460458,
+      "grad_norm": 1.4746694564819336,
+      "learning_rate": 3.9191512773609806e-05,
+      "loss": 0.1731,
+      "step": 22920
+    },
+    {
+      "epoch": 1.889883567662566,
+      "grad_norm": 0.23350854218006134,
+      "learning_rate": 3.91678279148984e-05,
+      "loss": 0.1719,
+      "step": 22940
+    },
+    {
+      "epoch": 1.891531195079086,
+      "grad_norm": 1.7336878776550293,
+      "learning_rate": 3.9144124308673784e-05,
+      "loss": 0.1495,
+      "step": 22960
+    },
+    {
+      "epoch": 1.8931788224956063,
+      "grad_norm": 1.0310170650482178,
+      "learning_rate": 3.912040198630166e-05,
+      "loss": 0.1682,
+      "step": 22980
+    },
+    {
+      "epoch": 1.8948264499121266,
+      "grad_norm": 0.8213064074516296,
+      "learning_rate": 3.9096660979172464e-05,
+      "loss": 0.1812,
+      "step": 23000
+    },
+    {
+      "epoch": 1.8964740773286466,
+      "grad_norm": 0.4071774184703827,
+      "learning_rate": 3.907290131870139e-05,
+      "loss": 0.1697,
+      "step": 23020
+    },
+    {
+      "epoch": 1.8981217047451668,
+      "grad_norm": 0.33335015177726746,
+      "learning_rate": 3.9049123036328296e-05,
+      "loss": 0.1645,
+      "step": 23040
+    },
+    {
+      "epoch": 1.899769332161687,
+      "grad_norm": 1.0531188249588013,
+      "learning_rate": 3.9025326163517676e-05,
+      "loss": 0.1484,
+      "step": 23060
+    },
+    {
+      "epoch": 1.9014169595782073,
+      "grad_norm": 1.8591293096542358,
+      "learning_rate": 3.900151073175864e-05,
+      "loss": 0.1911,
+      "step": 23080
+    },
+    {
+      "epoch": 1.9030645869947276,
+      "grad_norm": 1.4219475984573364,
+      "learning_rate": 3.897767677256485e-05,
+      "loss": 0.1809,
+      "step": 23100
+    },
+    {
+      "epoch": 1.9047122144112478,
+      "grad_norm": 1.1440541744232178,
+      "learning_rate": 3.8955017379019396e-05,
+      "loss": 0.174,
+      "step": 23120
+    },
+    {
+      "epoch": 1.906359841827768,
+      "grad_norm": 0.2897416353225708,
+      "learning_rate": 3.8931147382061806e-05,
+      "loss": 0.176,
+      "step": 23140
+    },
+    {
+      "epoch": 1.9080074692442883,
+      "grad_norm": 1.9240251779556274,
+      "learning_rate": 3.890725895077744e-05,
+      "loss": 0.1549,
+      "step": 23160
+    },
+    {
+      "epoch": 1.9096550966608086,
+      "grad_norm": 0.8471350073814392,
+      "learning_rate": 3.888335211677656e-05,
+      "loss": 0.1813,
+      "step": 23180
+    },
+    {
+      "epoch": 1.9113027240773286,
+      "grad_norm": 1.2786263227462769,
+      "learning_rate": 3.8859426911693756e-05,
+      "loss": 0.1754,
+      "step": 23200
+    },
+    {
+      "epoch": 1.9129503514938488,
+      "grad_norm": 2.0185673236846924,
+      "learning_rate": 3.8835483367187974e-05,
+      "loss": 0.1832,
+      "step": 23220
+    },
+    {
+      "epoch": 1.914597978910369,
+      "grad_norm": 0.2411476969718933,
+      "learning_rate": 3.8811521514942374e-05,
+      "loss": 0.1714,
+      "step": 23240
+    },
+    {
+      "epoch": 1.9162456063268891,
+      "grad_norm": 1.0712230205535889,
+      "learning_rate": 3.8787541386664374e-05,
+      "loss": 0.1563,
+      "step": 23260
+    },
+    {
+      "epoch": 1.9178932337434094,
+      "grad_norm": 0.9587976932525635,
+      "learning_rate": 3.876354301408558e-05,
+      "loss": 0.178,
+      "step": 23280
+    },
+    {
+      "epoch": 1.9195408611599296,
+      "grad_norm": 1.298622965812683,
+      "learning_rate": 3.873952642896171e-05,
+      "loss": 0.1695,
+      "step": 23300
+    },
+    {
+      "epoch": 1.9211884885764499,
+      "grad_norm": 4.192885398864746,
+      "learning_rate": 3.871549166307261e-05,
+      "loss": 0.1721,
+      "step": 23320
+    },
+    {
+      "epoch": 1.92283611599297,
+      "grad_norm": 0.1895458996295929,
+      "learning_rate": 3.869143874822218e-05,
+      "loss": 0.1775,
+      "step": 23340
+    },
+    {
+      "epoch": 1.9244837434094904,
+      "grad_norm": 1.721372127532959,
+      "learning_rate": 3.8667367716238326e-05,
+      "loss": 0.1597,
+      "step": 23360
+    },
+    {
+      "epoch": 1.9261313708260106,
+      "grad_norm": 4.6658525466918945,
+      "learning_rate": 3.864327859897291e-05,
+      "loss": 0.1728,
+      "step": 23380
+    },
+    {
+      "epoch": 1.9277789982425309,
+      "grad_norm": 0.3837490379810333,
+      "learning_rate": 3.861917142830177e-05,
+      "loss": 0.1788,
+      "step": 23400
+    },
+    {
+      "epoch": 1.929426625659051,
+      "grad_norm": 0.4308772087097168,
+      "learning_rate": 3.859504623612461e-05,
+      "loss": 0.1747,
+      "step": 23420
+    },
+    {
+      "epoch": 1.9310742530755711,
+      "grad_norm": 0.3830674886703491,
+      "learning_rate": 3.857090305436496e-05,
+      "loss": 0.1703,
+      "step": 23440
+    },
+    {
+      "epoch": 1.9327218804920914,
+      "grad_norm": 0.9429712891578674,
+      "learning_rate": 3.854674191497019e-05,
+      "loss": 0.1552,
+      "step": 23460
+    },
+    {
+      "epoch": 1.9343695079086116,
+      "grad_norm": 0.5584468245506287,
+      "learning_rate": 3.85225628499114e-05,
+      "loss": 0.1817,
+      "step": 23480
+    },
+    {
+      "epoch": 1.9360171353251316,
+      "grad_norm": 1.4192808866500854,
+      "learning_rate": 3.849836589118345e-05,
+      "loss": 0.1776,
+      "step": 23500
+    },
+    {
+      "epoch": 1.937664762741652,
+      "grad_norm": 0.453254371881485,
+      "learning_rate": 3.847415107080485e-05,
+      "loss": 0.1721,
+      "step": 23520
+    },
+    {
+      "epoch": 1.9393123901581721,
+      "grad_norm": 0.37701448798179626,
+      "learning_rate": 3.8449918420817734e-05,
+      "loss": 0.1724,
+      "step": 23540
+    },
+    {
+      "epoch": 1.9409600175746924,
+      "grad_norm": 0.9709746241569519,
+      "learning_rate": 3.8425667973287847e-05,
+      "loss": 0.1512,
+      "step": 23560
+    },
+    {
+      "epoch": 1.9426076449912126,
+      "grad_norm": 3.227597951889038,
+      "learning_rate": 3.8401399760304504e-05,
+      "loss": 0.1702,
+      "step": 23580
+    },
+    {
+      "epoch": 1.9442552724077329,
+      "grad_norm": 1.5262757539749146,
+      "learning_rate": 3.837711381398048e-05,
+      "loss": 0.1773,
+      "step": 23600
+    },
+    {
+      "epoch": 1.9459028998242531,
+      "grad_norm": 0.9164407849311829,
+      "learning_rate": 3.835281016645206e-05,
+      "loss": 0.168,
+      "step": 23620
+    },
+    {
+      "epoch": 1.9475505272407734,
+      "grad_norm": 0.3824036121368408,
+      "learning_rate": 3.832848884987892e-05,
+      "loss": 0.178,
+      "step": 23640
+    },
+    {
+      "epoch": 1.9491981546572936,
+      "grad_norm": 0.8727893233299255,
+      "learning_rate": 3.8304149896444145e-05,
+      "loss": 0.1525,
+      "step": 23660
+    },
+    {
+      "epoch": 1.9508457820738139,
+      "grad_norm": 1.189348578453064,
+      "learning_rate": 3.827979333835413e-05,
+      "loss": 0.1655,
+      "step": 23680
+    },
+    {
+      "epoch": 1.952493409490334,
+      "grad_norm": 0.6089633107185364,
+      "learning_rate": 3.825541920783858e-05,
+      "loss": 0.179,
+      "step": 23700
+    },
+    {
+      "epoch": 1.9541410369068541,
+      "grad_norm": 5.746613025665283,
+      "learning_rate": 3.8231027537150464e-05,
+      "loss": 0.1729,
+      "step": 23720
+    },
+    {
+      "epoch": 1.9557886643233744,
+      "grad_norm": 0.3469756841659546,
+      "learning_rate": 3.820661835856594e-05,
+      "loss": 0.2064,
+      "step": 23740
+    },
+    {
+      "epoch": 1.9574362917398944,
+      "grad_norm": 1.2260953187942505,
+      "learning_rate": 3.818219170438434e-05,
+      "loss": 0.1463,
+      "step": 23760
+    },
+    {
+      "epoch": 1.9590839191564147,
+      "grad_norm": 0.5917554497718811,
+      "learning_rate": 3.815774760692813e-05,
+      "loss": 0.1633,
+      "step": 23780
+    },
+    {
+      "epoch": 1.960731546572935,
+      "grad_norm": 0.7345995903015137,
+      "learning_rate": 3.8133286098542835e-05,
+      "loss": 0.1662,
+      "step": 23800
+    },
+    {
+      "epoch": 1.9623791739894552,
+      "grad_norm": 0.6015161275863647,
+      "learning_rate": 3.810880721159704e-05,
+      "loss": 0.1782,
+      "step": 23820
+    },
+    {
+      "epoch": 1.9640268014059754,
+      "grad_norm": 0.1269834190607071,
+      "learning_rate": 3.808431097848233e-05,
+      "loss": 0.177,
+      "step": 23840
+    },
+    {
+      "epoch": 1.9656744288224957,
+      "grad_norm": 3.330901861190796,
+      "learning_rate": 3.805979743161322e-05,
+      "loss": 0.1551,
+      "step": 23860
+    },
+    {
+      "epoch": 1.967322056239016,
+      "grad_norm": 1.1339266300201416,
+      "learning_rate": 3.8035266603427144e-05,
+      "loss": 0.1705,
+      "step": 23880
+    },
+    {
+      "epoch": 1.9689696836555362,
+      "grad_norm": 0.560189962387085,
+      "learning_rate": 3.801071852638443e-05,
+      "loss": 0.1862,
+      "step": 23900
+    },
+    {
+      "epoch": 1.9706173110720564,
+      "grad_norm": 0.4056016206741333,
+      "learning_rate": 3.798615323296819e-05,
+      "loss": 0.1811,
+      "step": 23920
+    },
+    {
+      "epoch": 1.9722649384885764,
+      "grad_norm": 0.24118812382221222,
+      "learning_rate": 3.796157075568433e-05,
+      "loss": 0.1726,
+      "step": 23940
+    },
+    {
+      "epoch": 1.9739125659050967,
+      "grad_norm": 1.474798560142517,
+      "learning_rate": 3.793697112706153e-05,
+      "loss": 0.1446,
+      "step": 23960
+    },
+    {
+      "epoch": 1.975560193321617,
+      "grad_norm": 1.19919753074646,
+      "learning_rate": 3.791235437965109e-05,
+      "loss": 0.1664,
+      "step": 23980
+    },
+    {
+      "epoch": 1.977207820738137,
+      "grad_norm": 0.38849976658821106,
+      "learning_rate": 3.788772054602705e-05,
+      "loss": 0.1769,
+      "step": 24000
+    },
+    {
+      "epoch": 1.977207820738137,
+      "eval_loss": 0.5249012112617493,
+      "eval_runtime": 254.2293,
+      "eval_samples_per_second": 85.706,
+      "eval_steps_per_second": 21.429,
+      "eval_wer": 0.22245208175285466,
+      "step": 24000
+    }
+  ],
+  "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": 1.2933992650885111e+21,
+  "train_batch_size": 24,
+  "trial_name": null,
+  "trial_params": null
+}