diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4657 @@
+{
+  "best_metric": 13.61759339287429,
+  "best_model_checkpoint": "weights/whisper-small-vi\\checkpoint-1000",
+  "epoch": 1.5175946125391255,
+  "eval_steps": 1000,
+  "global_step": 16000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0023712415820923835,
+      "grad_norm": 4.243698596954346,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 0.1841,
+      "step": 25
+    },
+    {
+      "epoch": 0.004742483164184767,
+      "grad_norm": 18.725072860717773,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.1768,
+      "step": 50
+    },
+    {
+      "epoch": 0.00711372474627715,
+      "grad_norm": 4.0062947273254395,
+      "learning_rate": 1.48e-06,
+      "loss": 0.1891,
+      "step": 75
+    },
+    {
+      "epoch": 0.009484966328369534,
+      "grad_norm": 9.787406921386719,
+      "learning_rate": 1.98e-06,
+      "loss": 0.2035,
+      "step": 100
+    },
+    {
+      "epoch": 0.011856207910461918,
+      "grad_norm": 7.350272178649902,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 0.1894,
+      "step": 125
+    },
+    {
+      "epoch": 0.0142274494925543,
+      "grad_norm": 6.18973445892334,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.1834,
+      "step": 150
+    },
+    {
+      "epoch": 0.016598691074646683,
+      "grad_norm": 7.800885200500488,
+      "learning_rate": 3.48e-06,
+      "loss": 0.2039,
+      "step": 175
+    },
+    {
+      "epoch": 0.018969932656739068,
+      "grad_norm": 7.546644687652588,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.2302,
+      "step": 200
+    },
+    {
+      "epoch": 0.021341174238831452,
+      "grad_norm": 8.287006378173828,
+      "learning_rate": 4.48e-06,
+      "loss": 0.243,
+      "step": 225
+    },
+    {
+      "epoch": 0.023712415820923836,
+      "grad_norm": 5.540294647216797,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.2015,
+      "step": 250
+    },
+    {
+      "epoch": 0.02608365740301622,
+      "grad_norm": 10.512287139892578,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 0.2165,
+      "step": 275
+    },
+    {
+      "epoch": 0.0284548989851086,
+      "grad_norm": 5.343227863311768,
+      "learning_rate": 5.98e-06,
+      "loss": 0.1845,
+      "step": 300
+    },
+    {
+      "epoch": 0.030826140567200986,
+      "grad_norm": 8.020576477050781,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 0.1921,
+      "step": 325
+    },
+    {
+      "epoch": 0.03319738214929337,
+      "grad_norm": 11.32276725769043,
+      "learning_rate": 6.98e-06,
+      "loss": 0.2009,
+      "step": 350
+    },
+    {
+      "epoch": 0.03556862373138575,
+      "grad_norm": 7.940517425537109,
+      "learning_rate": 7.48e-06,
+      "loss": 0.2111,
+      "step": 375
+    },
+    {
+      "epoch": 0.037939865313478136,
+      "grad_norm": 6.660912036895752,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.2495,
+      "step": 400
+    },
+    {
+      "epoch": 0.04031110689557052,
+      "grad_norm": 5.362044811248779,
+      "learning_rate": 8.48e-06,
+      "loss": 0.2016,
+      "step": 425
+    },
+    {
+      "epoch": 0.042682348477662904,
+      "grad_norm": 10.02092170715332,
+      "learning_rate": 8.98e-06,
+      "loss": 0.2319,
+      "step": 450
+    },
+    {
+      "epoch": 0.04505359005975529,
+      "grad_norm": 9.783544540405273,
+      "learning_rate": 9.48e-06,
+      "loss": 0.2037,
+      "step": 475
+    },
+    {
+      "epoch": 0.04742483164184767,
+      "grad_norm": 8.47091007232666,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.2357,
+      "step": 500
+    },
+    {
+      "epoch": 0.04979607322394006,
+      "grad_norm": 8.72714900970459,
+      "learning_rate": 9.99229014745093e-06,
+      "loss": 0.2152,
+      "step": 525
+    },
+    {
+      "epoch": 0.05216731480603244,
+      "grad_norm": 7.690199851989746,
+      "learning_rate": 9.98425905104565e-06,
+      "loss": 0.265,
+      "step": 550
+    },
+    {
+      "epoch": 0.05453855638812482,
+      "grad_norm": 6.1832051277160645,
+      "learning_rate": 9.976227954640369e-06,
+      "loss": 0.2535,
+      "step": 575
+    },
+    {
+      "epoch": 0.0569097979702172,
+      "grad_norm": 7.757453441619873,
+      "learning_rate": 9.968196858235088e-06,
+      "loss": 0.2208,
+      "step": 600
+    },
+    {
+      "epoch": 0.05928103955230959,
+      "grad_norm": 6.8661909103393555,
+      "learning_rate": 9.960165761829806e-06,
+      "loss": 0.1958,
+      "step": 625
+    },
+    {
+      "epoch": 0.06165228113440197,
+      "grad_norm": 7.996158123016357,
+      "learning_rate": 9.952134665424525e-06,
+      "loss": 0.2013,
+      "step": 650
+    },
+    {
+      "epoch": 0.06402352271649436,
+      "grad_norm": 10.544328689575195,
+      "learning_rate": 9.944103569019243e-06,
+      "loss": 0.2457,
+      "step": 675
+    },
+    {
+      "epoch": 0.06639476429858673,
+      "grad_norm": 10.344038963317871,
+      "learning_rate": 9.936072472613962e-06,
+      "loss": 0.2417,
+      "step": 700
+    },
+    {
+      "epoch": 0.06876600588067912,
+      "grad_norm": 10.921828269958496,
+      "learning_rate": 9.928041376208682e-06,
+      "loss": 0.2269,
+      "step": 725
+    },
+    {
+      "epoch": 0.0711372474627715,
+      "grad_norm": 7.314992427825928,
+      "learning_rate": 9.9200102798034e-06,
+      "loss": 0.2232,
+      "step": 750
+    },
+    {
+      "epoch": 0.07350848904486389,
+      "grad_norm": 9.345252990722656,
+      "learning_rate": 9.911979183398119e-06,
+      "loss": 0.2492,
+      "step": 775
+    },
+    {
+      "epoch": 0.07587973062695627,
+      "grad_norm": 6.190252780914307,
+      "learning_rate": 9.903948086992836e-06,
+      "loss": 0.2268,
+      "step": 800
+    },
+    {
+      "epoch": 0.07825097220904866,
+      "grad_norm": 9.122097969055176,
+      "learning_rate": 9.895916990587556e-06,
+      "loss": 0.2364,
+      "step": 825
+    },
+    {
+      "epoch": 0.08062221379114104,
+      "grad_norm": 6.245472431182861,
+      "learning_rate": 9.887885894182274e-06,
+      "loss": 0.2476,
+      "step": 850
+    },
+    {
+      "epoch": 0.08299345537323342,
+      "grad_norm": 6.766631603240967,
+      "learning_rate": 9.879854797776993e-06,
+      "loss": 0.2545,
+      "step": 875
+    },
+    {
+      "epoch": 0.08536469695532581,
+      "grad_norm": 13.654972076416016,
+      "learning_rate": 9.871823701371712e-06,
+      "loss": 0.2387,
+      "step": 900
+    },
+    {
+      "epoch": 0.08773593853741819,
+      "grad_norm": 6.007532119750977,
+      "learning_rate": 9.863792604966432e-06,
+      "loss": 0.2138,
+      "step": 925
+    },
+    {
+      "epoch": 0.09010718011951058,
+      "grad_norm": 3.846298933029175,
+      "learning_rate": 9.85576150856115e-06,
+      "loss": 0.2341,
+      "step": 950
+    },
+    {
+      "epoch": 0.09247842170160296,
+      "grad_norm": 7.410999298095703,
+      "learning_rate": 9.847730412155869e-06,
+      "loss": 0.2325,
+      "step": 975
+    },
+    {
+      "epoch": 0.09484966328369535,
+      "grad_norm": 3.528373956680298,
+      "learning_rate": 9.839699315750587e-06,
+      "loss": 0.2047,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09484966328369535,
+      "eval_loss": 0.23825648427009583,
+      "eval_runtime": 570.6842,
+      "eval_samples_per_second": 2.218,
+      "eval_steps_per_second": 0.279,
+      "eval_wer": 13.61759339287429,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09722090486578773,
+      "grad_norm": 9.557660102844238,
+      "learning_rate": 9.831668219345306e-06,
+      "loss": 0.2551,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09959214644788011,
+      "grad_norm": 4.750800132751465,
+      "learning_rate": 9.823637122940024e-06,
+      "loss": 0.2255,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1019633880299725,
+      "grad_norm": 14.495481491088867,
+      "learning_rate": 9.815606026534743e-06,
+      "loss": 0.2538,
+      "step": 1075
+    },
+    {
+      "epoch": 0.10433462961206488,
+      "grad_norm": 7.538572311401367,
+      "learning_rate": 9.807574930129462e-06,
+      "loss": 0.2069,
+      "step": 1100
+    },
+    {
+      "epoch": 0.10670587119415727,
+      "grad_norm": 7.318747043609619,
+      "learning_rate": 9.799543833724182e-06,
+      "loss": 0.2529,
+      "step": 1125
+    },
+    {
+      "epoch": 0.10907711277624964,
+      "grad_norm": 6.736952781677246,
+      "learning_rate": 9.7915127373189e-06,
+      "loss": 0.2533,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11144835435834202,
+      "grad_norm": 8.421292304992676,
+      "learning_rate": 9.783481640913617e-06,
+      "loss": 0.266,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1138195959404344,
+      "grad_norm": 7.647660255432129,
+      "learning_rate": 9.775450544508337e-06,
+      "loss": 0.2098,
+      "step": 1200
+    },
+    {
+      "epoch": 0.11619083752252679,
+      "grad_norm": 7.508166790008545,
+      "learning_rate": 9.767419448103056e-06,
+      "loss": 0.2487,
+      "step": 1225
+    },
+    {
+      "epoch": 0.11856207910461918,
+      "grad_norm": 11.438481330871582,
+      "learning_rate": 9.759388351697775e-06,
+      "loss": 0.2269,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12093332068671156,
+      "grad_norm": 12.222189903259277,
+      "learning_rate": 9.751357255292493e-06,
+      "loss": 0.2605,
+      "step": 1275
+    },
+    {
+      "epoch": 0.12330456226880394,
+      "grad_norm": 5.241353511810303,
+      "learning_rate": 9.743326158887213e-06,
+      "loss": 0.2371,
+      "step": 1300
+    },
+    {
+      "epoch": 0.12567580385089633,
+      "grad_norm": 8.221159934997559,
+      "learning_rate": 9.73529506248193e-06,
+      "loss": 0.2388,
+      "step": 1325
+    },
+    {
+      "epoch": 0.12804704543298873,
+      "grad_norm": 6.663714408874512,
+      "learning_rate": 9.72726396607665e-06,
+      "loss": 0.25,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1304182870150811,
+      "grad_norm": 7.624959945678711,
+      "learning_rate": 9.719232869671367e-06,
+      "loss": 0.2688,
+      "step": 1375
+    },
+    {
+      "epoch": 0.13278952859717347,
+      "grad_norm": 5.731607437133789,
+      "learning_rate": 9.711201773266087e-06,
+      "loss": 0.213,
+      "step": 1400
+    },
+    {
+      "epoch": 0.13516077017926587,
+      "grad_norm": 11.201618194580078,
+      "learning_rate": 9.703170676860806e-06,
+      "loss": 0.2282,
+      "step": 1425
+    },
+    {
+      "epoch": 0.13753201176135824,
+      "grad_norm": 7.8266520500183105,
+      "learning_rate": 9.695139580455526e-06,
+      "loss": 0.2354,
+      "step": 1450
+    },
+    {
+      "epoch": 0.13990325334345063,
+      "grad_norm": 12.77951717376709,
+      "learning_rate": 9.687108484050243e-06,
+      "loss": 0.2223,
+      "step": 1475
+    },
+    {
+      "epoch": 0.142274494925543,
+      "grad_norm": 10.189809799194336,
+      "learning_rate": 9.679077387644961e-06,
+      "loss": 0.2032,
+      "step": 1500
+    },
+    {
+      "epoch": 0.1446457365076354,
+      "grad_norm": 7.438467979431152,
+      "learning_rate": 9.67104629123968e-06,
+      "loss": 0.2479,
+      "step": 1525
+    },
+    {
+      "epoch": 0.14701697808972777,
+      "grad_norm": 9.758583068847656,
+      "learning_rate": 9.6630151948344e-06,
+      "loss": 0.2674,
+      "step": 1550
+    },
+    {
+      "epoch": 0.14938821967182017,
+      "grad_norm": 6.636728763580322,
+      "learning_rate": 9.65498409842912e-06,
+      "loss": 0.2647,
+      "step": 1575
+    },
+    {
+      "epoch": 0.15175946125391254,
+      "grad_norm": 7.328827857971191,
+      "learning_rate": 9.646953002023837e-06,
+      "loss": 0.226,
+      "step": 1600
+    },
+    {
+      "epoch": 0.15413070283600494,
+      "grad_norm": 6.227550983428955,
+      "learning_rate": 9.638921905618556e-06,
+      "loss": 0.2673,
+      "step": 1625
+    },
+    {
+      "epoch": 0.1565019444180973,
+      "grad_norm": 7.557720184326172,
+      "learning_rate": 9.630890809213274e-06,
+      "loss": 0.2726,
+      "step": 1650
+    },
+    {
+      "epoch": 0.1588731860001897,
+      "grad_norm": 8.988253593444824,
+      "learning_rate": 9.622859712807993e-06,
+      "loss": 0.227,
+      "step": 1675
+    },
+    {
+      "epoch": 0.16124442758228208,
+      "grad_norm": 4.943378925323486,
+      "learning_rate": 9.614828616402711e-06,
+      "loss": 0.2505,
+      "step": 1700
+    },
+    {
+      "epoch": 0.16361566916437448,
+      "grad_norm": 5.867473602294922,
+      "learning_rate": 9.60679751999743e-06,
+      "loss": 0.2441,
+      "step": 1725
+    },
+    {
+      "epoch": 0.16598691074646685,
+      "grad_norm": 8.801599502563477,
+      "learning_rate": 9.59876642359215e-06,
+      "loss": 0.2284,
+      "step": 1750
+    },
+    {
+      "epoch": 0.16835815232855925,
+      "grad_norm": 8.777979850769043,
+      "learning_rate": 9.59073532718687e-06,
+      "loss": 0.2322,
+      "step": 1775
+    },
+    {
+      "epoch": 0.17072939391065162,
+      "grad_norm": 6.238548278808594,
+      "learning_rate": 9.582704230781587e-06,
+      "loss": 0.2148,
+      "step": 1800
+    },
+    {
+      "epoch": 0.173100635492744,
+      "grad_norm": 5.301454067230225,
+      "learning_rate": 9.574673134376305e-06,
+      "loss": 0.2107,
+      "step": 1825
+    },
+    {
+      "epoch": 0.17547187707483639,
+      "grad_norm": 9.7332124710083,
+      "learning_rate": 9.566642037971024e-06,
+      "loss": 0.2605,
+      "step": 1850
+    },
+    {
+      "epoch": 0.17784311865692876,
+      "grad_norm": 7.469912052154541,
+      "learning_rate": 9.558610941565744e-06,
+      "loss": 0.2178,
+      "step": 1875
+    },
+    {
+      "epoch": 0.18021436023902115,
+      "grad_norm": 6.883511543273926,
+      "learning_rate": 9.550579845160463e-06,
+      "loss": 0.2634,
+      "step": 1900
+    },
+    {
+      "epoch": 0.18258560182111352,
+      "grad_norm": 7.4876556396484375,
+      "learning_rate": 9.54254874875518e-06,
+      "loss": 0.2622,
+      "step": 1925
+    },
+    {
+      "epoch": 0.18495684340320592,
+      "grad_norm": 9.996825218200684,
+      "learning_rate": 9.5345176523499e-06,
+      "loss": 0.2421,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1873280849852983,
+      "grad_norm": 6.896517276763916,
+      "learning_rate": 9.526486555944618e-06,
+      "loss": 0.2789,
+      "step": 1975
+    },
+    {
+      "epoch": 0.1896993265673907,
+      "grad_norm": 7.97552490234375,
+      "learning_rate": 9.518455459539337e-06,
+      "loss": 0.2306,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1896993265673907,
+      "eval_loss": 0.23884236812591553,
+      "eval_runtime": 478.7023,
+      "eval_samples_per_second": 2.645,
+      "eval_steps_per_second": 0.332,
+      "eval_wer": 14.102564102564102,
+      "step": 2000
+    },
+    {
+      "epoch": 0.19207056814948306,
+      "grad_norm": 10.05894660949707,
+      "learning_rate": 9.510745606990268e-06,
+      "loss": 0.2219,
+      "step": 2025
+    },
+    {
+      "epoch": 0.19444180973157546,
+      "grad_norm": 7.940761566162109,
+      "learning_rate": 9.502714510584986e-06,
+      "loss": 0.1989,
+      "step": 2050
+    },
+    {
+      "epoch": 0.19681305131366783,
+      "grad_norm": 5.554014205932617,
+      "learning_rate": 9.494683414179705e-06,
+      "loss": 0.2362,
+      "step": 2075
+    },
+    {
+      "epoch": 0.19918429289576023,
+      "grad_norm": 11.077298164367676,
+      "learning_rate": 9.486652317774423e-06,
+      "loss": 0.2714,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2015555344778526,
+      "grad_norm": 5.0622124671936035,
+      "learning_rate": 9.478621221369142e-06,
+      "loss": 0.2842,
+      "step": 2125
+    },
+    {
+      "epoch": 0.203926776059945,
+      "grad_norm": 8.84643840789795,
+      "learning_rate": 9.470590124963861e-06,
+      "loss": 0.2859,
+      "step": 2150
+    },
+    {
+      "epoch": 0.20629801764203737,
+      "grad_norm": 6.987701416015625,
+      "learning_rate": 9.46255902855858e-06,
+      "loss": 0.2613,
+      "step": 2175
+    },
+    {
+      "epoch": 0.20866925922412977,
+      "grad_norm": 6.213912010192871,
+      "learning_rate": 9.454527932153299e-06,
+      "loss": 0.2609,
+      "step": 2200
+    },
+    {
+      "epoch": 0.21104050080622214,
+      "grad_norm": 8.363004684448242,
+      "learning_rate": 9.446496835748016e-06,
+      "loss": 0.2347,
+      "step": 2225
+    },
+    {
+      "epoch": 0.21341174238831453,
+      "grad_norm": 10.02326774597168,
+      "learning_rate": 9.438465739342736e-06,
+      "loss": 0.2096,
+      "step": 2250
+    },
+    {
+      "epoch": 0.2157829839704069,
+      "grad_norm": 7.619383811950684,
+      "learning_rate": 9.430434642937455e-06,
+      "loss": 0.2404,
+      "step": 2275
+    },
+    {
+      "epoch": 0.21815422555249928,
+      "grad_norm": 14.356492042541504,
+      "learning_rate": 9.422403546532173e-06,
+      "loss": 0.2426,
+      "step": 2300
+    },
+    {
+      "epoch": 0.22052546713459167,
+      "grad_norm": 8.412384986877441,
+      "learning_rate": 9.414372450126892e-06,
+      "loss": 0.2554,
+      "step": 2325
+    },
+    {
+      "epoch": 0.22289670871668404,
+      "grad_norm": 11.35639762878418,
+      "learning_rate": 9.406341353721612e-06,
+      "loss": 0.2334,
+      "step": 2350
+    },
+    {
+      "epoch": 0.22526795029877644,
+      "grad_norm": 9.665873527526855,
+      "learning_rate": 9.39831025731633e-06,
+      "loss": 0.24,
+      "step": 2375
+    },
+    {
+      "epoch": 0.2276391918808688,
+      "grad_norm": 11.83912467956543,
+      "learning_rate": 9.390279160911049e-06,
+      "loss": 0.1851,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2300104334629612,
+      "grad_norm": 7.5992608070373535,
+      "learning_rate": 9.382569308361978e-06,
+      "loss": 0.2536,
+      "step": 2425
+    },
+    {
+      "epoch": 0.23238167504505358,
+      "grad_norm": 9.059174537658691,
+      "learning_rate": 9.374538211956697e-06,
+      "loss": 0.2271,
+      "step": 2450
+    },
+    {
+      "epoch": 0.23475291662714598,
+      "grad_norm": 5.181990623474121,
+      "learning_rate": 9.366507115551415e-06,
+      "loss": 0.2235,
+      "step": 2475
+    },
+    {
+      "epoch": 0.23712415820923835,
+      "grad_norm": 7.1707611083984375,
+      "learning_rate": 9.358476019146134e-06,
+      "loss": 0.2382,
+      "step": 2500
+    },
+    {
+      "epoch": 0.23949539979133075,
+      "grad_norm": 11.292250633239746,
+      "learning_rate": 9.350444922740854e-06,
+      "loss": 0.2572,
+      "step": 2525
+    },
+    {
+      "epoch": 0.24186664137342312,
+      "grad_norm": 5.159794807434082,
+      "learning_rate": 9.342413826335573e-06,
+      "loss": 0.2527,
+      "step": 2550
+    },
+    {
+      "epoch": 0.24423788295551552,
+      "grad_norm": 8.422658920288086,
+      "learning_rate": 9.33438272993029e-06,
+      "loss": 0.2192,
+      "step": 2575
+    },
+    {
+      "epoch": 0.2466091245376079,
+      "grad_norm": 6.054739952087402,
+      "learning_rate": 9.32635163352501e-06,
+      "loss": 0.2055,
+      "step": 2600
+    },
+    {
+      "epoch": 0.24898036611970029,
+      "grad_norm": 6.057258605957031,
+      "learning_rate": 9.318320537119728e-06,
+      "loss": 0.2519,
+      "step": 2625
+    },
+    {
+      "epoch": 0.25135160770179266,
+      "grad_norm": 5.51326847076416,
+      "learning_rate": 9.310289440714447e-06,
+      "loss": 0.2164,
+      "step": 2650
+    },
+    {
+      "epoch": 0.25372284928388505,
+      "grad_norm": 6.998528003692627,
+      "learning_rate": 9.302258344309167e-06,
+      "loss": 0.2309,
+      "step": 2675
+    },
+    {
+      "epoch": 0.25609409086597745,
+      "grad_norm": 6.977015972137451,
+      "learning_rate": 9.294227247903884e-06,
+      "loss": 0.2225,
+      "step": 2700
+    },
+    {
+      "epoch": 0.2584653324480698,
+      "grad_norm": 6.228749752044678,
+      "learning_rate": 9.286196151498604e-06,
+      "loss": 0.2314,
+      "step": 2725
+    },
+    {
+      "epoch": 0.2608365740301622,
+      "grad_norm": 8.625155448913574,
+      "learning_rate": 9.278165055093323e-06,
+      "loss": 0.2653,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2632078156122546,
+      "grad_norm": 8.030765533447266,
+      "learning_rate": 9.27013395868804e-06,
+      "loss": 0.2269,
+      "step": 2775
+    },
+    {
+      "epoch": 0.26557905719434693,
+      "grad_norm": 6.238178730010986,
+      "learning_rate": 9.262102862282758e-06,
+      "loss": 0.2742,
+      "step": 2800
+    },
+    {
+      "epoch": 0.26795029877643933,
+      "grad_norm": 7.385952949523926,
+      "learning_rate": 9.254071765877478e-06,
+      "loss": 0.2205,
+      "step": 2825
+    },
+    {
+      "epoch": 0.27032154035853173,
+      "grad_norm": 6.59016227722168,
+      "learning_rate": 9.246040669472197e-06,
+      "loss": 0.1748,
+      "step": 2850
+    },
+    {
+      "epoch": 0.27269278194062413,
+      "grad_norm": 9.493818283081055,
+      "learning_rate": 9.238009573066917e-06,
+      "loss": 0.248,
+      "step": 2875
+    },
+    {
+      "epoch": 0.27506402352271647,
+      "grad_norm": 8.83185863494873,
+      "learning_rate": 9.229978476661634e-06,
+      "loss": 0.2511,
+      "step": 2900
+    },
+    {
+      "epoch": 0.27743526510480887,
+      "grad_norm": 8.106586456298828,
+      "learning_rate": 9.221947380256354e-06,
+      "loss": 0.2796,
+      "step": 2925
+    },
+    {
+      "epoch": 0.27980650668690127,
+      "grad_norm": 10.219332695007324,
+      "learning_rate": 9.213916283851071e-06,
+      "loss": 0.2303,
+      "step": 2950
+    },
+    {
+      "epoch": 0.28217774826899367,
+      "grad_norm": 8.04117202758789,
+      "learning_rate": 9.205885187445791e-06,
+      "loss": 0.2251,
+      "step": 2975
+    },
+    {
+      "epoch": 0.284548989851086,
+      "grad_norm": 10.147671699523926,
+      "learning_rate": 9.19785409104051e-06,
+      "loss": 0.2344,
+      "step": 3000
+    },
+    {
+      "epoch": 0.284548989851086,
+      "eval_loss": 0.23593533039093018,
+      "eval_runtime": 479.7648,
+      "eval_samples_per_second": 2.639,
+      "eval_steps_per_second": 0.331,
+      "eval_wer": 14.06895227119946,
+      "step": 3000
+    },
+    {
+      "epoch": 0.2869202314331784,
+      "grad_norm": 7.848628044128418,
+      "learning_rate": 9.189822994635228e-06,
+      "loss": 0.2304,
+      "step": 3025
+    },
+    {
+      "epoch": 0.2892914730152708,
+      "grad_norm": 7.655786037445068,
+      "learning_rate": 9.181791898229947e-06,
+      "loss": 0.2508,
+      "step": 3050
+    },
+    {
+      "epoch": 0.2916627145973632,
+      "grad_norm": 8.65014362335205,
+      "learning_rate": 9.173760801824667e-06,
+      "loss": 0.2781,
+      "step": 3075
+    },
+    {
+      "epoch": 0.29403395617945555,
+      "grad_norm": 7.906963348388672,
+      "learning_rate": 9.165729705419384e-06,
+      "loss": 0.1927,
+      "step": 3100
+    },
+    {
+      "epoch": 0.29640519776154794,
+      "grad_norm": 9.74050235748291,
+      "learning_rate": 9.157698609014102e-06,
+      "loss": 0.2221,
+      "step": 3125
+    },
+    {
+      "epoch": 0.29877643934364034,
+      "grad_norm": 5.777608871459961,
+      "learning_rate": 9.149667512608822e-06,
+      "loss": 0.21,
+      "step": 3150
+    },
+    {
+      "epoch": 0.30114768092573274,
+      "grad_norm": 7.260315418243408,
+      "learning_rate": 9.141636416203541e-06,
+      "loss": 0.2605,
+      "step": 3175
+    },
+    {
+      "epoch": 0.3035189225078251,
+      "grad_norm": 7.249190807342529,
+      "learning_rate": 9.13360531979826e-06,
+      "loss": 0.2494,
+      "step": 3200
+    },
+    {
+      "epoch": 0.3058901640899175,
+      "grad_norm": 11.440152168273926,
+      "learning_rate": 9.125574223392978e-06,
+      "loss": 0.2854,
+      "step": 3225
+    },
+    {
+      "epoch": 0.3082614056720099,
+      "grad_norm": 9.915044784545898,
+      "learning_rate": 9.117543126987697e-06,
+      "loss": 0.2429,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3106326472541022,
+      "grad_norm": 6.068655967712402,
+      "learning_rate": 9.109512030582415e-06,
+      "loss": 0.2192,
+      "step": 3275
+    },
+    {
+      "epoch": 0.3130038888361946,
+      "grad_norm": 6.529017925262451,
+      "learning_rate": 9.101480934177135e-06,
+      "loss": 0.2473,
+      "step": 3300
+    },
+    {
+      "epoch": 0.315375130418287,
+      "grad_norm": 7.932959079742432,
+      "learning_rate": 9.093449837771854e-06,
+      "loss": 0.2257,
+      "step": 3325
+    },
+    {
+      "epoch": 0.3177463720003794,
+      "grad_norm": 9.91701602935791,
+      "learning_rate": 9.085418741366572e-06,
+      "loss": 0.2501,
+      "step": 3350
+    },
+    {
+      "epoch": 0.32011761358247176,
+      "grad_norm": 5.059365272521973,
+      "learning_rate": 9.077387644961291e-06,
+      "loss": 0.2183,
+      "step": 3375
+    },
+    {
+      "epoch": 0.32248885516456416,
+      "grad_norm": 6.630497932434082,
+      "learning_rate": 9.06935654855601e-06,
+      "loss": 0.2419,
+      "step": 3400
+    },
+    {
+      "epoch": 0.32486009674665656,
+      "grad_norm": 7.9866838455200195,
+      "learning_rate": 9.061325452150728e-06,
+      "loss": 0.2151,
+      "step": 3425
+    },
+    {
+      "epoch": 0.32723133832874896,
+      "grad_norm": 7.101008892059326,
+      "learning_rate": 9.053294355745446e-06,
+      "loss": 0.2153,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3296025799108413,
+      "grad_norm": 6.151196479797363,
+      "learning_rate": 9.045263259340165e-06,
+      "loss": 0.2194,
+      "step": 3475
+    },
+    {
+      "epoch": 0.3319738214929337,
+      "grad_norm": 9.240913391113281,
+      "learning_rate": 9.037232162934885e-06,
+      "loss": 0.258,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3343450630750261,
+      "grad_norm": 6.897662162780762,
+      "learning_rate": 9.029201066529604e-06,
+      "loss": 0.2524,
+      "step": 3525
+    },
+    {
+      "epoch": 0.3367163046571185,
+      "grad_norm": 8.628865242004395,
+      "learning_rate": 9.021169970124322e-06,
+      "loss": 0.238,
+      "step": 3550
+    },
+    {
+      "epoch": 0.33908754623921084,
+      "grad_norm": 11.23631763458252,
+      "learning_rate": 9.013138873719041e-06,
+      "loss": 0.2407,
+      "step": 3575
+    },
+    {
+      "epoch": 0.34145878782130323,
+      "grad_norm": 10.063039779663086,
+      "learning_rate": 9.005107777313759e-06,
+      "loss": 0.2542,
+      "step": 3600
+    },
+    {
+      "epoch": 0.34383002940339563,
+      "grad_norm": 7.6819748878479,
+      "learning_rate": 8.997076680908478e-06,
+      "loss": 0.2432,
+      "step": 3625
+    },
+    {
+      "epoch": 0.346201270985488,
+      "grad_norm": 10.960393905639648,
+      "learning_rate": 8.989045584503198e-06,
+      "loss": 0.2398,
+      "step": 3650
+    },
+    {
+      "epoch": 0.3485725125675804,
+      "grad_norm": 8.009682655334473,
+      "learning_rate": 8.981014488097915e-06,
+      "loss": 0.2007,
+      "step": 3675
+    },
+    {
+      "epoch": 0.35094375414967277,
+      "grad_norm": 7.910845756530762,
+      "learning_rate": 8.972983391692635e-06,
+      "loss": 0.2059,
+      "step": 3700
+    },
+    {
+      "epoch": 0.35331499573176517,
+      "grad_norm": 5.531894683837891,
+      "learning_rate": 8.964952295287354e-06,
+      "loss": 0.2093,
+      "step": 3725
+    },
+    {
+      "epoch": 0.3556862373138575,
+      "grad_norm": 10.481830596923828,
+      "learning_rate": 8.956921198882072e-06,
+      "loss": 0.221,
+      "step": 3750
+    },
+    {
+      "epoch": 0.3580574788959499,
+      "grad_norm": 10.431686401367188,
+      "learning_rate": 8.94889010247679e-06,
+      "loss": 0.2498,
+      "step": 3775
+    },
+    {
+      "epoch": 0.3604287204780423,
+      "grad_norm": 7.973873615264893,
+      "learning_rate": 8.940859006071509e-06,
+      "loss": 0.2912,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3627999620601347,
+      "grad_norm": 7.270717144012451,
+      "learning_rate": 8.932827909666228e-06,
+      "loss": 0.2475,
+      "step": 3825
+    },
+    {
+      "epoch": 0.36517120364222705,
+      "grad_norm": 9.13284683227539,
+      "learning_rate": 8.924796813260948e-06,
+      "loss": 0.218,
+      "step": 3850
+    },
+    {
+      "epoch": 0.36754244522431945,
+      "grad_norm": 8.543096542358398,
+      "learning_rate": 8.916765716855666e-06,
+      "loss": 0.2471,
+      "step": 3875
+    },
+    {
+      "epoch": 0.36991368680641185,
+      "grad_norm": 4.368423938751221,
+      "learning_rate": 8.908734620450385e-06,
+      "loss": 0.2697,
+      "step": 3900
+    },
+    {
+      "epoch": 0.37228492838850424,
+      "grad_norm": 8.543111801147461,
+      "learning_rate": 8.900703524045103e-06,
+      "loss": 0.2658,
+      "step": 3925
+    },
+    {
+      "epoch": 0.3746561699705966,
+      "grad_norm": 4.0211663246154785,
+      "learning_rate": 8.892672427639822e-06,
+      "loss": 0.2122,
+      "step": 3950
+    },
+    {
+      "epoch": 0.377027411552689,
+      "grad_norm": 9.078948974609375,
+      "learning_rate": 8.88464133123454e-06,
+      "loss": 0.2455,
+      "step": 3975
+    },
+    {
+      "epoch": 0.3793986531347814,
+      "grad_norm": 7.649990558624268,
+      "learning_rate": 8.876610234829259e-06,
+      "loss": 0.2294,
+      "step": 4000
+    },
+    {
+      "epoch": 0.3793986531347814,
+      "eval_loss": 0.23183076083660126,
+      "eval_runtime": 483.1887,
+      "eval_samples_per_second": 2.62,
+      "eval_steps_per_second": 0.329,
+      "eval_wer": 13.680015365408623,
+      "step": 4000
+    },
+    {
+      "epoch": 0.3817698947168738,
+      "grad_norm": 8.352575302124023,
+      "learning_rate": 8.868579138423979e-06,
+      "loss": 0.2096,
+      "step": 4025
+    },
+    {
+      "epoch": 0.3841411362989661,
+      "grad_norm": 8.847747802734375,
+      "learning_rate": 8.860548042018698e-06,
+      "loss": 0.2445,
+      "step": 4050
+    },
+    {
+      "epoch": 0.3865123778810585,
+      "grad_norm": 8.746859550476074,
+      "learning_rate": 8.852516945613416e-06,
+      "loss": 0.2489,
+      "step": 4075
+    },
+    {
+      "epoch": 0.3888836194631509,
+      "grad_norm": 7.650919437408447,
+      "learning_rate": 8.844485849208135e-06,
+      "loss": 0.2015,
+      "step": 4100
+    },
+    {
+      "epoch": 0.39125486104524326,
+      "grad_norm": 9.161053657531738,
+      "learning_rate": 8.836454752802853e-06,
+      "loss": 0.2279,
+      "step": 4125
+    },
+    {
+      "epoch": 0.39362610262733566,
+      "grad_norm": 9.58436393737793,
+      "learning_rate": 8.828423656397572e-06,
+      "loss": 0.2276,
+      "step": 4150
+    },
+    {
+      "epoch": 0.39599734420942806,
+      "grad_norm": 6.949397087097168,
+      "learning_rate": 8.820392559992292e-06,
+      "loss": 0.2476,
+      "step": 4175
+    },
+    {
+      "epoch": 0.39836858579152046,
+      "grad_norm": 8.661158561706543,
+      "learning_rate": 8.81236146358701e-06,
+      "loss": 0.2403,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4007398273736128,
+      "grad_norm": 5.287428379058838,
+      "learning_rate": 8.804330367181729e-06,
+      "loss": 0.2268,
+      "step": 4225
+    },
+    {
+      "epoch": 0.4031110689557052,
+      "grad_norm": 3.7870798110961914,
+      "learning_rate": 8.796299270776446e-06,
+      "loss": 0.2567,
+      "step": 4250
+    },
+    {
+      "epoch": 0.4054823105377976,
+      "grad_norm": 9.175024032592773,
+      "learning_rate": 8.788268174371166e-06,
+      "loss": 0.2364,
+      "step": 4275
+    },
+    {
+      "epoch": 0.40785355211989,
+      "grad_norm": 8.662485122680664,
+      "learning_rate": 8.780237077965883e-06,
+      "loss": 0.2483,
+      "step": 4300
+    },
+    {
+      "epoch": 0.41022479370198234,
+      "grad_norm": 6.981433391571045,
+      "learning_rate": 8.772205981560603e-06,
+      "loss": 0.2671,
+      "step": 4325
+    },
+    {
+      "epoch": 0.41259603528407474,
+      "grad_norm": 7.031622409820557,
+      "learning_rate": 8.764174885155322e-06,
+      "loss": 0.21,
+      "step": 4350
+    },
+    {
+      "epoch": 0.41496727686616713,
+      "grad_norm": 8.828649520874023,
+      "learning_rate": 8.756143788750042e-06,
+      "loss": 0.2477,
+      "step": 4375
+    },
+    {
+      "epoch": 0.41733851844825953,
+      "grad_norm": 7.427188396453857,
+      "learning_rate": 8.74811269234476e-06,
+      "loss": 0.2329,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4197097600303519,
+      "grad_norm": 5.831352233886719,
+      "learning_rate": 8.740081595939479e-06,
+      "loss": 0.2094,
+      "step": 4425
+    },
+    {
+      "epoch": 0.4220810016124443,
+      "grad_norm": 14.921052932739258,
+      "learning_rate": 8.732050499534196e-06,
+      "loss": 0.2123,
+      "step": 4450
+    },
+    {
+      "epoch": 0.42445224319453667,
+      "grad_norm": 5.82997465133667,
+      "learning_rate": 8.724019403128916e-06,
+      "loss": 0.2299,
+      "step": 4475
+    },
+    {
+      "epoch": 0.42682348477662907,
+      "grad_norm": 8.217269897460938,
+      "learning_rate": 8.715988306723635e-06,
+      "loss": 0.2327,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4291947263587214,
+      "grad_norm": 7.112634658813477,
+      "learning_rate": 8.707957210318353e-06,
+      "loss": 0.2541,
+      "step": 4525
+    },
+    {
+      "epoch": 0.4315659679408138,
+      "grad_norm": 6.622945785522461,
+      "learning_rate": 8.699926113913072e-06,
+      "loss": 0.2033,
+      "step": 4550
+    },
+    {
+      "epoch": 0.4339372095229062,
+      "grad_norm": 11.053245544433594,
+      "learning_rate": 8.691895017507792e-06,
+      "loss": 0.2353,
+      "step": 4575
+    },
+    {
+      "epoch": 0.43630845110499855,
+      "grad_norm": 6.637090682983398,
+      "learning_rate": 8.68386392110251e-06,
+      "loss": 0.2282,
+      "step": 4600
+    },
+    {
+      "epoch": 0.43867969268709095,
+      "grad_norm": 5.031044006347656,
+      "learning_rate": 8.675832824697227e-06,
+      "loss": 0.2466,
+      "step": 4625
+    },
+    {
+      "epoch": 0.44105093426918335,
+      "grad_norm": 5.068697452545166,
+      "learning_rate": 8.667801728291947e-06,
+      "loss": 0.2681,
+      "step": 4650
+    },
+    {
+      "epoch": 0.44342217585127575,
+      "grad_norm": 5.92368745803833,
+      "learning_rate": 8.659770631886666e-06,
+      "loss": 0.2555,
+      "step": 4675
+    },
+    {
+      "epoch": 0.4457934174333681,
+      "grad_norm": 7.328484058380127,
+      "learning_rate": 8.651739535481385e-06,
+      "loss": 0.2444,
+      "step": 4700
+    },
+    {
+      "epoch": 0.4481646590154605,
+      "grad_norm": 6.549984931945801,
+      "learning_rate": 8.643708439076103e-06,
+      "loss": 0.2471,
+      "step": 4725
+    },
+    {
+      "epoch": 0.4505359005975529,
+      "grad_norm": 7.688663959503174,
+      "learning_rate": 8.635677342670822e-06,
+      "loss": 0.2525,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4529071421796453,
+      "grad_norm": 11.712684631347656,
+      "learning_rate": 8.62764624626554e-06,
+      "loss": 0.2567,
+      "step": 4775
+    },
+    {
+      "epoch": 0.4552783837617376,
+      "grad_norm": 7.224384307861328,
+      "learning_rate": 8.61961514986026e-06,
+      "loss": 0.2077,
+      "step": 4800
+    },
+    {
+      "epoch": 0.45764962534383,
+      "grad_norm": 6.971681118011475,
+      "learning_rate": 8.611584053454979e-06,
+      "loss": 0.2784,
+      "step": 4825
+    },
+    {
+      "epoch": 0.4600208669259224,
+      "grad_norm": 4.564589500427246,
+      "learning_rate": 8.603552957049697e-06,
+      "loss": 0.2336,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4623921085080148,
+      "grad_norm": 6.923882484436035,
+      "learning_rate": 8.595521860644416e-06,
+      "loss": 0.2109,
+      "step": 4875
+    },
+    {
+      "epoch": 0.46476335009010716,
+      "grad_norm": 6.198056697845459,
+      "learning_rate": 8.587490764239136e-06,
+      "loss": 0.2159,
+      "step": 4900
+    },
+    {
+      "epoch": 0.46713459167219956,
+      "grad_norm": 10.406079292297363,
+      "learning_rate": 8.579459667833853e-06,
+      "loss": 0.2409,
+      "step": 4925
+    },
+    {
+      "epoch": 0.46950583325429196,
+      "grad_norm": 9.45881175994873,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.2466,
+      "step": 4950
+    },
+    {
+      "epoch": 0.47187707483638436,
+      "grad_norm": 8.945230484008789,
+      "learning_rate": 8.56339747502329e-06,
+      "loss": 0.2539,
+      "step": 4975
+    },
+    {
+      "epoch": 0.4742483164184767,
+      "grad_norm": 7.036405563354492,
+      "learning_rate": 8.55536637861801e-06,
+      "loss": 0.2164,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4742483164184767,
+      "eval_loss": 0.23614493012428284,
+      "eval_runtime": 519.2865,
+      "eval_samples_per_second": 2.438,
+      "eval_steps_per_second": 0.306,
+      "eval_wer": 17.991933160472488,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4766195580005691,
+      "grad_norm": 8.915026664733887,
+      "learning_rate": 8.547335282212729e-06,
+      "loss": 0.2376,
+      "step": 5025
+    },
+    {
+      "epoch": 0.4789907995826615,
+      "grad_norm": 4.80546236038208,
+      "learning_rate": 8.539304185807447e-06,
+      "loss": 0.1811,
+      "step": 5050
+    },
+    {
+      "epoch": 0.48136204116475384,
+      "grad_norm": 6.2160210609436035,
+      "learning_rate": 8.531273089402166e-06,
+      "loss": 0.2248,
+      "step": 5075
+    },
+    {
+      "epoch": 0.48373328274684624,
+      "grad_norm": 5.965505123138428,
+      "learning_rate": 8.523241992996884e-06,
+      "loss": 0.2138,
+      "step": 5100
+    },
+    {
+      "epoch": 0.48610452432893864,
+      "grad_norm": 5.956753730773926,
+      "learning_rate": 8.515210896591603e-06,
+      "loss": 0.1774,
+      "step": 5125
+    },
+    {
+      "epoch": 0.48847576591103103,
+      "grad_norm": 9.409728050231934,
+      "learning_rate": 8.507179800186323e-06,
+      "loss": 0.2703,
+      "step": 5150
+    },
+    {
+      "epoch": 0.4908470074931234,
+      "grad_norm": 8.576353073120117,
+      "learning_rate": 8.49914870378104e-06,
+      "loss": 0.1976,
+      "step": 5175
+    },
+    {
+      "epoch": 0.4932182490752158,
+      "grad_norm": 9.11634635925293,
+      "learning_rate": 8.49111760737576e-06,
+      "loss": 0.2511,
+      "step": 5200
+    },
+    {
+      "epoch": 0.4955894906573082,
+      "grad_norm": 12.834346771240234,
+      "learning_rate": 8.48308651097048e-06,
+      "loss": 0.2716,
+      "step": 5225
+    },
+    {
+      "epoch": 0.49796073223940057,
+      "grad_norm": 11.0597505569458,
+      "learning_rate": 8.475055414565197e-06,
+      "loss": 0.2496,
+      "step": 5250
+    },
+    {
+      "epoch": 0.500331973821493,
+      "grad_norm": 5.973386287689209,
+      "learning_rate": 8.467024318159915e-06,
+      "loss": 0.2026,
+      "step": 5275
+    },
+    {
+      "epoch": 0.5027032154035853,
+      "grad_norm": 6.129185199737549,
+      "learning_rate": 8.458993221754634e-06,
+      "loss": 0.2138,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5050744569856777,
+      "grad_norm": 7.220257759094238,
+      "learning_rate": 8.450962125349353e-06,
+      "loss": 0.2401,
+      "step": 5325
+    },
+    {
+      "epoch": 0.5074456985677701,
+      "grad_norm": 8.916608810424805,
+      "learning_rate": 8.442931028944073e-06,
+      "loss": 0.2189,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5098169401498625,
+      "grad_norm": 6.470343112945557,
+      "learning_rate": 8.43489993253879e-06,
+      "loss": 0.2425,
+      "step": 5375
+    },
+    {
+      "epoch": 0.5121881817319549,
+      "grad_norm": 5.5481486320495605,
+      "learning_rate": 8.42686883613351e-06,
+      "loss": 0.2084,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5145594233140472,
+      "grad_norm": 12.080617904663086,
+      "learning_rate": 8.418837739728228e-06,
+      "loss": 0.2226,
+      "step": 5425
+    },
+    {
+      "epoch": 0.5169306648961396,
+      "grad_norm": 6.952599048614502,
+      "learning_rate": 8.410806643322947e-06,
+      "loss": 0.2561,
+      "step": 5450
+    },
+    {
+      "epoch": 0.519301906478232,
+      "grad_norm": 5.153127193450928,
+      "learning_rate": 8.402775546917666e-06,
+      "loss": 0.2209,
+      "step": 5475
+    },
+    {
+      "epoch": 0.5216731480603244,
+      "grad_norm": 9.593254089355469,
+      "learning_rate": 8.394744450512384e-06,
+      "loss": 0.2374,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5240443896424167,
+      "grad_norm": 8.538336753845215,
+      "learning_rate": 8.386713354107104e-06,
+      "loss": 0.2615,
+      "step": 5525
+    },
+    {
+      "epoch": 0.5264156312245092,
+      "grad_norm": 7.285602569580078,
+      "learning_rate": 8.378682257701823e-06,
+      "loss": 0.2166,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5287868728066015,
+      "grad_norm": 7.6555585861206055,
+      "learning_rate": 8.370972405152752e-06,
+      "loss": 0.1629,
+      "step": 5575
+    },
+    {
+      "epoch": 0.5311581143886939,
+      "grad_norm": 8.01225471496582,
+      "learning_rate": 8.362941308747471e-06,
+      "loss": 0.2538,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5335293559707863,
+      "grad_norm": 8.645890235900879,
+      "learning_rate": 8.35491021234219e-06,
+      "loss": 0.2718,
+      "step": 5625
+    },
+    {
+      "epoch": 0.5359005975528787,
+      "grad_norm": 7.651273727416992,
+      "learning_rate": 8.346879115936908e-06,
+      "loss": 0.2211,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5382718391349711,
+      "grad_norm": 11.032370567321777,
+      "learning_rate": 8.338848019531626e-06,
+      "loss": 0.1996,
+      "step": 5675
+    },
+    {
+      "epoch": 0.5406430807170635,
+      "grad_norm": 7.57820463180542,
+      "learning_rate": 8.330816923126346e-06,
+      "loss": 0.2211,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5430143222991558,
+      "grad_norm": 5.820690155029297,
+      "learning_rate": 8.322785826721065e-06,
+      "loss": 0.2076,
+      "step": 5725
+    },
+    {
+      "epoch": 0.5453855638812483,
+      "grad_norm": 7.3265204429626465,
+      "learning_rate": 8.314754730315784e-06,
+      "loss": 0.2346,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5477568054633406,
+      "grad_norm": 7.54530143737793,
+      "learning_rate": 8.306723633910502e-06,
+      "loss": 0.234,
+      "step": 5775
+    },
+    {
+      "epoch": 0.5501280470454329,
+      "grad_norm": 6.74851131439209,
+      "learning_rate": 8.298692537505221e-06,
+      "loss": 0.2454,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5524992886275254,
+      "grad_norm": 6.331222057342529,
+      "learning_rate": 8.29066144109994e-06,
+      "loss": 0.2641,
+      "step": 5825
+    },
+    {
+      "epoch": 0.5548705302096177,
+      "grad_norm": 7.0079731941223145,
+      "learning_rate": 8.282630344694659e-06,
+      "loss": 0.2343,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5572417717917102,
+      "grad_norm": 8.573423385620117,
+      "learning_rate": 8.274599248289376e-06,
+      "loss": 0.2169,
+      "step": 5875
+    },
+    {
+      "epoch": 0.5596130133738025,
+      "grad_norm": 8.464674949645996,
+      "learning_rate": 8.266568151884096e-06,
+      "loss": 0.2127,
+      "step": 5900
+    },
+    {
+      "epoch": 0.5619842549558949,
+      "grad_norm": 6.330426216125488,
+      "learning_rate": 8.258537055478815e-06,
+      "loss": 0.2087,
+      "step": 5925
+    },
+    {
+      "epoch": 0.5643554965379873,
+      "grad_norm": 6.594757080078125,
+      "learning_rate": 8.250505959073534e-06,
+      "loss": 0.192,
+      "step": 5950
+    },
+    {
+      "epoch": 0.5667267381200797,
+      "grad_norm": 5.730506896972656,
+      "learning_rate": 8.242474862668252e-06,
+      "loss": 0.2222,
+      "step": 5975
+    },
+    {
+      "epoch": 0.569097979702172,
+      "grad_norm": 6.290910243988037,
+      "learning_rate": 8.23444376626297e-06,
+      "loss": 0.235,
+      "step": 6000
+    },
+    {
+      "epoch": 0.569097979702172,
+      "eval_loss": 0.22957296669483185,
+      "eval_runtime": 516.889,
+      "eval_samples_per_second": 2.449,
+      "eval_steps_per_second": 0.308,
+      "eval_wer": 17.574186113511956,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5714692212842645,
+      "grad_norm": 9.003548622131348,
+      "learning_rate": 8.22641266985769e-06,
+      "loss": 0.2516,
+      "step": 6025
+    },
+    {
+      "epoch": 0.5738404628663568,
+      "grad_norm": 5.0031890869140625,
+      "learning_rate": 8.218381573452409e-06,
+      "loss": 0.2661,
+      "step": 6050
+    },
+    {
+      "epoch": 0.5762117044484492,
+      "grad_norm": 7.2725419998168945,
+      "learning_rate": 8.210350477047128e-06,
+      "loss": 0.2241,
+      "step": 6075
+    },
+    {
+      "epoch": 0.5785829460305416,
+      "grad_norm": 6.713600158691406,
+      "learning_rate": 8.202319380641846e-06,
+      "loss": 0.2457,
+      "step": 6100
+    },
+    {
+      "epoch": 0.580954187612634,
+      "grad_norm": 5.608094215393066,
+      "learning_rate": 8.194288284236565e-06,
+      "loss": 0.2153,
+      "step": 6125
+    },
+    {
+      "epoch": 0.5833254291947264,
+      "grad_norm": 12.03117561340332,
+      "learning_rate": 8.186257187831283e-06,
+      "loss": 0.2465,
+      "step": 6150
+    },
+    {
+      "epoch": 0.5856966707768188,
+      "grad_norm": 4.534688472747803,
+      "learning_rate": 8.178226091426002e-06,
+      "loss": 0.2131,
+      "step": 6175
+    },
+    {
+      "epoch": 0.5880679123589111,
+      "grad_norm": 7.247952461242676,
+      "learning_rate": 8.17019499502072e-06,
+      "loss": 0.2174,
+      "step": 6200
+    },
+    {
+      "epoch": 0.5904391539410035,
+      "grad_norm": 4.587021350860596,
+      "learning_rate": 8.16216389861544e-06,
+      "loss": 0.197,
+      "step": 6225
+    },
+    {
+      "epoch": 0.5928103955230959,
+      "grad_norm": 8.814271926879883,
+      "learning_rate": 8.154132802210159e-06,
+      "loss": 0.2595,
+      "step": 6250
+    },
+    {
+      "epoch": 0.5951816371051882,
+      "grad_norm": 4.947065353393555,
+      "learning_rate": 8.146101705804878e-06,
+      "loss": 0.1998,
+      "step": 6275
+    },
+    {
+      "epoch": 0.5975528786872807,
+      "grad_norm": 6.412868022918701,
+      "learning_rate": 8.138070609399596e-06,
+      "loss": 0.2034,
+      "step": 6300
+    },
+    {
+      "epoch": 0.599924120269373,
+      "grad_norm": 12.337003707885742,
+      "learning_rate": 8.130039512994314e-06,
+      "loss": 0.2466,
+      "step": 6325
+    },
+    {
+      "epoch": 0.6022953618514655,
+      "grad_norm": 6.576595306396484,
+      "learning_rate": 8.122008416589033e-06,
+      "loss": 0.2196,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6046666034335578,
+      "grad_norm": 6.821414947509766,
+      "learning_rate": 8.113977320183752e-06,
+      "loss": 0.211,
+      "step": 6375
+    },
+    {
+      "epoch": 0.6070378450156502,
+      "grad_norm": 5.0727219581604,
+      "learning_rate": 8.105946223778472e-06,
+      "loss": 0.2174,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6094090865977426,
+      "grad_norm": 5.923619270324707,
+      "learning_rate": 8.09791512737319e-06,
+      "loss": 0.2347,
+      "step": 6425
+    },
+    {
+      "epoch": 0.611780328179835,
+      "grad_norm": 9.480695724487305,
+      "learning_rate": 8.089884030967909e-06,
+      "loss": 0.2139,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6141515697619273,
+      "grad_norm": 7.231909275054932,
+      "learning_rate": 8.081852934562627e-06,
+      "loss": 0.218,
+      "step": 6475
+    },
+    {
+      "epoch": 0.6165228113440198,
+      "grad_norm": 7.82000732421875,
+      "learning_rate": 8.073821838157346e-06,
+      "loss": 0.2413,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6188940529261121,
+      "grad_norm": 6.023988246917725,
+      "learning_rate": 8.065790741752064e-06,
+      "loss": 0.2644,
+      "step": 6525
+    },
+    {
+      "epoch": 0.6212652945082044,
+      "grad_norm": 3.8875234127044678,
+      "learning_rate": 8.057759645346783e-06,
+      "loss": 0.2439,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6236365360902969,
+      "grad_norm": 12.519841194152832,
+      "learning_rate": 8.049728548941503e-06,
+      "loss": 0.2092,
+      "step": 6575
+    },
+    {
+      "epoch": 0.6260077776723892,
+      "grad_norm": 6.534714221954346,
+      "learning_rate": 8.041697452536222e-06,
+      "loss": 0.2572,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6283790192544817,
+      "grad_norm": 4.25216817855835,
+      "learning_rate": 8.03366635613094e-06,
+      "loss": 0.2509,
+      "step": 6625
+    },
+    {
+      "epoch": 0.630750260836574,
+      "grad_norm": 9.793882369995117,
+      "learning_rate": 8.025635259725657e-06,
+      "loss": 0.2332,
+      "step": 6650
+    },
+    {
+      "epoch": 0.6331215024186664,
+      "grad_norm": 6.981718063354492,
+      "learning_rate": 8.017604163320377e-06,
+      "loss": 0.2266,
+      "step": 6675
+    },
+    {
+      "epoch": 0.6354927440007588,
+      "grad_norm": 11.162854194641113,
+      "learning_rate": 8.009573066915096e-06,
+      "loss": 0.2127,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6378639855828512,
+      "grad_norm": 10.495009422302246,
+      "learning_rate": 8.001541970509816e-06,
+      "loss": 0.2808,
+      "step": 6725
+    },
+    {
+      "epoch": 0.6402352271649435,
+      "grad_norm": 5.085358619689941,
+      "learning_rate": 7.993510874104533e-06,
+      "loss": 0.2252,
+      "step": 6750
+    },
+    {
+      "epoch": 0.642606468747036,
+      "grad_norm": 6.271093845367432,
+      "learning_rate": 7.985479777699253e-06,
+      "loss": 0.2011,
+      "step": 6775
+    },
+    {
+      "epoch": 0.6449777103291283,
+      "grad_norm": 9.27597713470459,
+      "learning_rate": 7.97744868129397e-06,
+      "loss": 0.2376,
+      "step": 6800
+    },
+    {
+      "epoch": 0.6473489519112207,
+      "grad_norm": 5.758539199829102,
+      "learning_rate": 7.96941758488869e-06,
+      "loss": 0.2185,
+      "step": 6825
+    },
+    {
+      "epoch": 0.6497201934933131,
+      "grad_norm": 6.621229648590088,
+      "learning_rate": 7.961386488483407e-06,
+      "loss": 0.2424,
+      "step": 6850
+    },
+    {
+      "epoch": 0.6520914350754055,
+      "grad_norm": 5.7187981605529785,
+      "learning_rate": 7.953355392078127e-06,
+      "loss": 0.1937,
+      "step": 6875
+    },
+    {
+      "epoch": 0.6544626766574979,
+      "grad_norm": 7.257383346557617,
+      "learning_rate": 7.945324295672846e-06,
+      "loss": 0.1941,
+      "step": 6900
+    },
+    {
+      "epoch": 0.6568339182395903,
+      "grad_norm": 6.857747554779053,
+      "learning_rate": 7.937293199267566e-06,
+      "loss": 0.1955,
+      "step": 6925
+    },
+    {
+      "epoch": 0.6592051598216826,
+      "grad_norm": 6.411887168884277,
+      "learning_rate": 7.929262102862283e-06,
+      "loss": 0.1863,
+      "step": 6950
+    },
+    {
+      "epoch": 0.661576401403775,
+      "grad_norm": 4.437167167663574,
+      "learning_rate": 7.921231006457003e-06,
+      "loss": 0.2279,
+      "step": 6975
+    },
+    {
+      "epoch": 0.6639476429858674,
+      "grad_norm": 4.998786926269531,
+      "learning_rate": 7.91319991005172e-06,
+      "loss": 0.2147,
+      "step": 7000
+    },
+    {
+      "epoch": 0.6639476429858674,
+      "eval_loss": 0.22273053228855133,
+      "eval_runtime": 538.6875,
+      "eval_samples_per_second": 2.35,
+      "eval_steps_per_second": 0.295,
+      "eval_wer": 19.864592336502447,
+      "step": 7000
+    },
+    {
+      "epoch": 0.6663188845679597,
+      "grad_norm": 4.569517612457275,
+      "learning_rate": 7.90516881364644e-06,
+      "loss": 0.2388,
+      "step": 7025
+    },
+    {
+      "epoch": 0.6686901261500522,
+      "grad_norm": 5.319979667663574,
+      "learning_rate": 7.89713771724116e-06,
+      "loss": 0.2313,
+      "step": 7050
+    },
+    {
+      "epoch": 0.6710613677321445,
+      "grad_norm": 5.558248519897461,
+      "learning_rate": 7.889106620835877e-06,
+      "loss": 0.2409,
+      "step": 7075
+    },
+    {
+      "epoch": 0.673432609314237,
+      "grad_norm": 9.056896209716797,
+      "learning_rate": 7.881075524430596e-06,
+      "loss": 0.2322,
+      "step": 7100
+    },
+    {
+      "epoch": 0.6758038508963293,
+      "grad_norm": 7.935134410858154,
+      "learning_rate": 7.873044428025314e-06,
+      "loss": 0.2417,
+      "step": 7125
+    },
+    {
+      "epoch": 0.6781750924784217,
+      "grad_norm": 6.591653823852539,
+      "learning_rate": 7.865013331620033e-06,
+      "loss": 0.2492,
+      "step": 7150
+    },
+    {
+      "epoch": 0.6805463340605141,
+      "grad_norm": 7.399641513824463,
+      "learning_rate": 7.856982235214751e-06,
+      "loss": 0.2703,
+      "step": 7175
+    },
+    {
+      "epoch": 0.6829175756426065,
+      "grad_norm": 7.3584208488464355,
+      "learning_rate": 7.84895113880947e-06,
+      "loss": 0.2416,
+      "step": 7200
+    },
+    {
+      "epoch": 0.6852888172246988,
+      "grad_norm": 14.287632942199707,
+      "learning_rate": 7.84092004240419e-06,
+      "loss": 0.2047,
+      "step": 7225
+    },
+    {
+      "epoch": 0.6876600588067913,
+      "grad_norm": 13.036450386047363,
+      "learning_rate": 7.83288894599891e-06,
+      "loss": 0.2376,
+      "step": 7250
+    },
+    {
+      "epoch": 0.6900313003888836,
+      "grad_norm": 10.043025016784668,
+      "learning_rate": 7.824857849593627e-06,
+      "loss": 0.2191,
+      "step": 7275
+    },
+    {
+      "epoch": 0.692402541970976,
+      "grad_norm": 9.027446746826172,
+      "learning_rate": 7.816826753188346e-06,
+      "loss": 0.2295,
+      "step": 7300
+    },
+    {
+      "epoch": 0.6947737835530684,
+      "grad_norm": 6.668909072875977,
+      "learning_rate": 7.808795656783064e-06,
+      "loss": 0.2383,
+      "step": 7325
+    },
+    {
+      "epoch": 0.6971450251351607,
+      "grad_norm": 13.295737266540527,
+      "learning_rate": 7.800764560377784e-06,
+      "loss": 0.2166,
+      "step": 7350
+    },
+    {
+      "epoch": 0.6995162667172532,
+      "grad_norm": 7.885791301727295,
+      "learning_rate": 7.792733463972501e-06,
+      "loss": 0.2511,
+      "step": 7375
+    },
+    {
+      "epoch": 0.7018875082993455,
+      "grad_norm": 6.995784759521484,
+      "learning_rate": 7.78470236756722e-06,
+      "loss": 0.2409,
+      "step": 7400
+    },
+    {
+      "epoch": 0.7042587498814379,
+      "grad_norm": 10.582818031311035,
+      "learning_rate": 7.77667127116194e-06,
+      "loss": 0.2231,
+      "step": 7425
+    },
+    {
+      "epoch": 0.7066299914635303,
+      "grad_norm": 6.680695056915283,
+      "learning_rate": 7.76864017475666e-06,
+      "loss": 0.2343,
+      "step": 7450
+    },
+    {
+      "epoch": 0.7090012330456227,
+      "grad_norm": 3.8850960731506348,
+      "learning_rate": 7.760609078351377e-06,
+      "loss": 0.2181,
+      "step": 7475
+    },
+    {
+      "epoch": 0.711372474627715,
+      "grad_norm": 9.75331974029541,
+      "learning_rate": 7.752577981946095e-06,
+      "loss": 0.2195,
+      "step": 7500
+    },
+    {
+      "epoch": 0.7137437162098075,
+      "grad_norm": 8.908821105957031,
+      "learning_rate": 7.744546885540814e-06,
+      "loss": 0.2307,
+      "step": 7525
+    },
+    {
+      "epoch": 0.7161149577918998,
+      "grad_norm": 7.855329513549805,
+      "learning_rate": 7.736515789135534e-06,
+      "loss": 0.2046,
+      "step": 7550
+    },
+    {
+      "epoch": 0.7184861993739923,
+      "grad_norm": 7.9686198234558105,
+      "learning_rate": 7.728484692730253e-06,
+      "loss": 0.2521,
+      "step": 7575
+    },
+    {
+      "epoch": 0.7208574409560846,
+      "grad_norm": 3.51355242729187,
+      "learning_rate": 7.72045359632497e-06,
+      "loss": 0.2172,
+      "step": 7600
+    },
+    {
+      "epoch": 0.723228682538177,
+      "grad_norm": 5.622837543487549,
+      "learning_rate": 7.71242249991969e-06,
+      "loss": 0.2024,
+      "step": 7625
+    },
+    {
+      "epoch": 0.7255999241202694,
+      "grad_norm": 4.206454753875732,
+      "learning_rate": 7.704391403514408e-06,
+      "loss": 0.2203,
+      "step": 7650
+    },
+    {
+      "epoch": 0.7279711657023618,
+      "grad_norm": 6.0628275871276855,
+      "learning_rate": 7.696360307109127e-06,
+      "loss": 0.2134,
+      "step": 7675
+    },
+    {
+      "epoch": 0.7303424072844541,
+      "grad_norm": 8.967851638793945,
+      "learning_rate": 7.688329210703845e-06,
+      "loss": 0.2354,
+      "step": 7700
+    },
+    {
+      "epoch": 0.7327136488665466,
+      "grad_norm": 7.647063255310059,
+      "learning_rate": 7.680298114298564e-06,
+      "loss": 0.2242,
+      "step": 7725
+    },
+    {
+      "epoch": 0.7350848904486389,
+      "grad_norm": 5.125141620635986,
+      "learning_rate": 7.672267017893284e-06,
+      "loss": 0.2381,
+      "step": 7750
+    },
+    {
+      "epoch": 0.7374561320307312,
+      "grad_norm": 7.5472092628479,
+      "learning_rate": 7.664235921488003e-06,
+      "loss": 0.197,
+      "step": 7775
+    },
+    {
+      "epoch": 0.7398273736128237,
+      "grad_norm": 8.719536781311035,
+      "learning_rate": 7.656204825082721e-06,
+      "loss": 0.2327,
+      "step": 7800
+    },
+    {
+      "epoch": 0.742198615194916,
+      "grad_norm": 7.150578022003174,
+      "learning_rate": 7.648173728677439e-06,
+      "loss": 0.22,
+      "step": 7825
+    },
+    {
+      "epoch": 0.7445698567770085,
+      "grad_norm": 6.205988883972168,
+      "learning_rate": 7.640142632272158e-06,
+      "loss": 0.2051,
+      "step": 7850
+    },
+    {
+      "epoch": 0.7469410983591008,
+      "grad_norm": 9.157383918762207,
+      "learning_rate": 7.632111535866877e-06,
+      "loss": 0.1998,
+      "step": 7875
+    },
+    {
+      "epoch": 0.7493123399411932,
+      "grad_norm": 8.229233741760254,
+      "learning_rate": 7.624080439461596e-06,
+      "loss": 0.2213,
+      "step": 7900
+    },
+    {
+      "epoch": 0.7516835815232856,
+      "grad_norm": 7.646115779876709,
+      "learning_rate": 7.6160493430563145e-06,
+      "loss": 0.218,
+      "step": 7925
+    },
+    {
+      "epoch": 0.754054823105378,
+      "grad_norm": 6.914640426635742,
+      "learning_rate": 7.608018246651033e-06,
+      "loss": 0.1995,
+      "step": 7950
+    },
+    {
+      "epoch": 0.7564260646874703,
+      "grad_norm": 5.177979469299316,
+      "learning_rate": 7.5999871502457525e-06,
+      "loss": 0.2548,
+      "step": 7975
+    },
+    {
+      "epoch": 0.7587973062695628,
+      "grad_norm": 7.25577449798584,
+      "learning_rate": 7.591956053840471e-06,
+      "loss": 0.1909,
+      "step": 8000
+    },
+    {
+      "epoch": 0.7587973062695628,
+      "eval_loss": 0.21843287348747253,
+      "eval_runtime": 610.3779,
+      "eval_samples_per_second": 2.074,
+      "eval_steps_per_second": 0.26,
+      "eval_wer": 32.589071353116296,
+      "step": 8000
+    },
+    {
+      "epoch": 0.7611685478516551,
+      "grad_norm": 4.782321453094482,
+      "learning_rate": 7.58392495743519e-06,
+      "loss": 0.2018,
+      "step": 8025
+    },
+    {
+      "epoch": 0.7635397894337476,
+      "grad_norm": 8.032735824584961,
+      "learning_rate": 7.575893861029908e-06,
+      "loss": 0.2271,
+      "step": 8050
+    },
+    {
+      "epoch": 0.7659110310158399,
+      "grad_norm": 8.738744735717773,
+      "learning_rate": 7.5678627646246275e-06,
+      "loss": 0.2383,
+      "step": 8075
+    },
+    {
+      "epoch": 0.7682822725979322,
+      "grad_norm": 9.34449291229248,
+      "learning_rate": 7.559831668219346e-06,
+      "loss": 0.2013,
+      "step": 8100
+    },
+    {
+      "epoch": 0.7706535141800247,
+      "grad_norm": 6.86797571182251,
+      "learning_rate": 7.5518005718140655e-06,
+      "loss": 0.2367,
+      "step": 8125
+    },
+    {
+      "epoch": 0.773024755762117,
+      "grad_norm": 8.582794189453125,
+      "learning_rate": 7.543769475408783e-06,
+      "loss": 0.2312,
+      "step": 8150
+    },
+    {
+      "epoch": 0.7753959973442094,
+      "grad_norm": 7.473124980926514,
+      "learning_rate": 7.535738379003502e-06,
+      "loss": 0.2607,
+      "step": 8175
+    },
+    {
+      "epoch": 0.7777672389263018,
+      "grad_norm": 4.819562911987305,
+      "learning_rate": 7.527707282598221e-06,
+      "loss": 0.2099,
+      "step": 8200
+    },
+    {
+      "epoch": 0.7801384805083942,
+      "grad_norm": 6.569749355316162,
+      "learning_rate": 7.5196761861929405e-06,
+      "loss": 0.2232,
+      "step": 8225
+    },
+    {
+      "epoch": 0.7825097220904865,
+      "grad_norm": 7.714107036590576,
+      "learning_rate": 7.511645089787658e-06,
+      "loss": 0.2345,
+      "step": 8250
+    },
+    {
+      "epoch": 0.784880963672579,
+      "grad_norm": 8.967876434326172,
+      "learning_rate": 7.503613993382377e-06,
+      "loss": 0.2145,
+      "step": 8275
+    },
+    {
+      "epoch": 0.7872522052546713,
+      "grad_norm": 10.014185905456543,
+      "learning_rate": 7.495582896977096e-06,
+      "loss": 0.2101,
+      "step": 8300
+    },
+    {
+      "epoch": 0.7896234468367638,
+      "grad_norm": 7.594282150268555,
+      "learning_rate": 7.487551800571815e-06,
+      "loss": 0.2381,
+      "step": 8325
+    },
+    {
+      "epoch": 0.7919946884188561,
+      "grad_norm": 5.880343914031982,
+      "learning_rate": 7.479520704166533e-06,
+      "loss": 0.1992,
+      "step": 8350
+    },
+    {
+      "epoch": 0.7943659300009485,
+      "grad_norm": 4.198406219482422,
+      "learning_rate": 7.471489607761252e-06,
+      "loss": 0.2071,
+      "step": 8375
+    },
+    {
+      "epoch": 0.7967371715830409,
+      "grad_norm": 6.5477213859558105,
+      "learning_rate": 7.463458511355971e-06,
+      "loss": 0.1994,
+      "step": 8400
+    },
+    {
+      "epoch": 0.7991084131651333,
+      "grad_norm": 7.467320919036865,
+      "learning_rate": 7.45542741495069e-06,
+      "loss": 0.2125,
+      "step": 8425
+    },
+    {
+      "epoch": 0.8014796547472256,
+      "grad_norm": 6.896966457366943,
+      "learning_rate": 7.4473963185454075e-06,
+      "loss": 0.1927,
+      "step": 8450
+    },
+    {
+      "epoch": 0.803850896329318,
+      "grad_norm": 6.3800368309021,
+      "learning_rate": 7.439365222140127e-06,
+      "loss": 0.2614,
+      "step": 8475
+    },
+    {
+      "epoch": 0.8062221379114104,
+      "grad_norm": 7.89002799987793,
+      "learning_rate": 7.431334125734846e-06,
+      "loss": 0.2302,
+      "step": 8500
+    },
+    {
+      "epoch": 0.8085933794935029,
+      "grad_norm": 8.861137390136719,
+      "learning_rate": 7.423303029329565e-06,
+      "loss": 0.2377,
+      "step": 8525
+    },
+    {
+      "epoch": 0.8109646210755952,
+      "grad_norm": 4.580018520355225,
+      "learning_rate": 7.415271932924284e-06,
+      "loss": 0.2015,
+      "step": 8550
+    },
+    {
+      "epoch": 0.8133358626576875,
+      "grad_norm": 6.245753288269043,
+      "learning_rate": 7.407240836519002e-06,
+      "loss": 0.2228,
+      "step": 8575
+    },
+    {
+      "epoch": 0.81570710423978,
+      "grad_norm": 7.133525371551514,
+      "learning_rate": 7.3992097401137205e-06,
+      "loss": 0.2369,
+      "step": 8600
+    },
+    {
+      "epoch": 0.8180783458218723,
+      "grad_norm": 3.831495761871338,
+      "learning_rate": 7.39117864370844e-06,
+      "loss": 0.1972,
+      "step": 8625
+    },
+    {
+      "epoch": 0.8204495874039647,
+      "grad_norm": 7.945802211761475,
+      "learning_rate": 7.3831475473031585e-06,
+      "loss": 0.2245,
+      "step": 8650
+    },
+    {
+      "epoch": 0.8228208289860571,
+      "grad_norm": 6.700128078460693,
+      "learning_rate": 7.375116450897877e-06,
+      "loss": 0.1953,
+      "step": 8675
+    },
+    {
+      "epoch": 0.8251920705681495,
+      "grad_norm": 6.118048191070557,
+      "learning_rate": 7.367085354492596e-06,
+      "loss": 0.1983,
+      "step": 8700
+    },
+    {
+      "epoch": 0.8275633121502418,
+      "grad_norm": 9.640941619873047,
+      "learning_rate": 7.359054258087315e-06,
+      "loss": 0.2425,
+      "step": 8725
+    },
+    {
+      "epoch": 0.8299345537323343,
+      "grad_norm": 13.098782539367676,
+      "learning_rate": 7.3510231616820335e-06,
+      "loss": 0.2369,
+      "step": 8750
+    },
+    {
+      "epoch": 0.8323057953144266,
+      "grad_norm": 7.9155449867248535,
+      "learning_rate": 7.342992065276752e-06,
+      "loss": 0.2413,
+      "step": 8775
+    },
+    {
+      "epoch": 0.8346770368965191,
+      "grad_norm": 6.480501651763916,
+      "learning_rate": 7.334960968871471e-06,
+      "loss": 0.23,
+      "step": 8800
+    },
+    {
+      "epoch": 0.8370482784786114,
+      "grad_norm": 5.60577917098999,
+      "learning_rate": 7.32692987246619e-06,
+      "loss": 0.2358,
+      "step": 8825
+    },
+    {
+      "epoch": 0.8394195200607037,
+      "grad_norm": 5.705789089202881,
+      "learning_rate": 7.318898776060909e-06,
+      "loss": 0.2235,
+      "step": 8850
+    },
+    {
+      "epoch": 0.8417907616427962,
+      "grad_norm": 7.375821590423584,
+      "learning_rate": 7.310867679655628e-06,
+      "loss": 0.1994,
+      "step": 8875
+    },
+    {
+      "epoch": 0.8441620032248885,
+      "grad_norm": 4.602260112762451,
+      "learning_rate": 7.302836583250346e-06,
+      "loss": 0.2199,
+      "step": 8900
+    },
+    {
+      "epoch": 0.8465332448069809,
+      "grad_norm": 7.766406536102295,
+      "learning_rate": 7.294805486845064e-06,
+      "loss": 0.2083,
+      "step": 8925
+    },
+    {
+      "epoch": 0.8489044863890733,
+      "grad_norm": 6.205694198608398,
+      "learning_rate": 7.286774390439784e-06,
+      "loss": 0.1812,
+      "step": 8950
+    },
+    {
+      "epoch": 0.8512757279711657,
+      "grad_norm": 7.249185085296631,
+      "learning_rate": 7.278743294034503e-06,
+      "loss": 0.203,
+      "step": 8975
+    },
+    {
+      "epoch": 0.8536469695532581,
+      "grad_norm": 6.586127281188965,
+      "learning_rate": 7.270712197629221e-06,
+      "loss": 0.2099,
+      "step": 9000
+    },
+    {
+      "epoch": 0.8536469695532581,
+      "eval_loss": 0.2207770198583603,
+      "eval_runtime": 560.3626,
+      "eval_samples_per_second": 2.259,
+      "eval_steps_per_second": 0.284,
+      "eval_wer": 29.117449342168445,
+      "step": 9000
+    },
+    {
+      "epoch": 0.8560182111353505,
+      "grad_norm": 7.582109451293945,
+      "learning_rate": 7.262681101223939e-06,
+      "loss": 0.2333,
+      "step": 9025
+    },
+    {
+      "epoch": 0.8583894527174428,
+      "grad_norm": 5.324299335479736,
+      "learning_rate": 7.254650004818659e-06,
+      "loss": 0.209,
+      "step": 9050
+    },
+    {
+      "epoch": 0.8607606942995353,
+      "grad_norm": 8.337318420410156,
+      "learning_rate": 7.246618908413377e-06,
+      "loss": 0.2274,
+      "step": 9075
+    },
+    {
+      "epoch": 0.8631319358816276,
+      "grad_norm": 4.812930107116699,
+      "learning_rate": 7.238587812008096e-06,
+      "loss": 0.2088,
+      "step": 9100
+    },
+    {
+      "epoch": 0.86550317746372,
+      "grad_norm": 8.623558044433594,
+      "learning_rate": 7.230556715602814e-06,
+      "loss": 0.231,
+      "step": 9125
+    },
+    {
+      "epoch": 0.8678744190458124,
+      "grad_norm": 9.329276084899902,
+      "learning_rate": 7.222525619197534e-06,
+      "loss": 0.2154,
+      "step": 9150
+    },
+    {
+      "epoch": 0.8702456606279048,
+      "grad_norm": 6.28425407409668,
+      "learning_rate": 7.214494522792252e-06,
+      "loss": 0.2227,
+      "step": 9175
+    },
+    {
+      "epoch": 0.8726169022099971,
+      "grad_norm": 10.491592407226562,
+      "learning_rate": 7.20646342638697e-06,
+      "loss": 0.2007,
+      "step": 9200
+    },
+    {
+      "epoch": 0.8749881437920896,
+      "grad_norm": 4.43834114074707,
+      "learning_rate": 7.198432329981689e-06,
+      "loss": 0.1883,
+      "step": 9225
+    },
+    {
+      "epoch": 0.8773593853741819,
+      "grad_norm": 7.208180904388428,
+      "learning_rate": 7.190401233576409e-06,
+      "loss": 0.2217,
+      "step": 9250
+    },
+    {
+      "epoch": 0.8797306269562744,
+      "grad_norm": 3.9081575870513916,
+      "learning_rate": 7.182370137171127e-06,
+      "loss": 0.2064,
+      "step": 9275
+    },
+    {
+      "epoch": 0.8821018685383667,
+      "grad_norm": 5.184658527374268,
+      "learning_rate": 7.174339040765847e-06,
+      "loss": 0.219,
+      "step": 9300
+    },
+    {
+      "epoch": 0.884473110120459,
+      "grad_norm": 10.200240135192871,
+      "learning_rate": 7.1663079443605645e-06,
+      "loss": 0.2278,
+      "step": 9325
+    },
+    {
+      "epoch": 0.8868443517025515,
+      "grad_norm": 6.337146282196045,
+      "learning_rate": 7.158276847955283e-06,
+      "loss": 0.1616,
+      "step": 9350
+    },
+    {
+      "epoch": 0.8892155932846438,
+      "grad_norm": 9.013226509094238,
+      "learning_rate": 7.1502457515500024e-06,
+      "loss": 0.2164,
+      "step": 9375
+    },
+    {
+      "epoch": 0.8915868348667362,
+      "grad_norm": 7.484189033508301,
+      "learning_rate": 7.142214655144721e-06,
+      "loss": 0.2295,
+      "step": 9400
+    },
+    {
+      "epoch": 0.8939580764488286,
+      "grad_norm": 8.29464054107666,
+      "learning_rate": 7.1341835587394395e-06,
+      "loss": 0.2083,
+      "step": 9425
+    },
+    {
+      "epoch": 0.896329318030921,
+      "grad_norm": 6.764057159423828,
+      "learning_rate": 7.126152462334158e-06,
+      "loss": 0.2362,
+      "step": 9450
+    },
+    {
+      "epoch": 0.8987005596130134,
+      "grad_norm": 8.269946098327637,
+      "learning_rate": 7.1181213659288775e-06,
+      "loss": 0.2336,
+      "step": 9475
+    },
+    {
+      "epoch": 0.9010718011951058,
+      "grad_norm": 45.17983627319336,
+      "learning_rate": 7.110090269523596e-06,
+      "loss": 0.1842,
+      "step": 9500
+    },
+    {
+      "epoch": 0.9034430427771981,
+      "grad_norm": 6.396184921264648,
+      "learning_rate": 7.102059173118315e-06,
+      "loss": 0.2176,
+      "step": 9525
+    },
+    {
+      "epoch": 0.9058142843592906,
+      "grad_norm": 8.392062187194824,
+      "learning_rate": 7.094028076713033e-06,
+      "loss": 0.2032,
+      "step": 9550
+    },
+    {
+      "epoch": 0.9081855259413829,
+      "grad_norm": 7.554737091064453,
+      "learning_rate": 7.0859969803077525e-06,
+      "loss": 0.2174,
+      "step": 9575
+    },
+    {
+      "epoch": 0.9105567675234753,
+      "grad_norm": 4.99600076675415,
+      "learning_rate": 7.077965883902471e-06,
+      "loss": 0.2017,
+      "step": 9600
+    },
+    {
+      "epoch": 0.9129280091055677,
+      "grad_norm": 7.299047946929932,
+      "learning_rate": 7.0699347874971905e-06,
+      "loss": 0.2406,
+      "step": 9625
+    },
+    {
+      "epoch": 0.91529925068766,
+      "grad_norm": 4.908593654632568,
+      "learning_rate": 7.061903691091908e-06,
+      "loss": 0.2456,
+      "step": 9650
+    },
+    {
+      "epoch": 0.9176704922697524,
+      "grad_norm": 5.818723678588867,
+      "learning_rate": 7.053872594686627e-06,
+      "loss": 0.1964,
+      "step": 9675
+    },
+    {
+      "epoch": 0.9200417338518448,
+      "grad_norm": 4.100748538970947,
+      "learning_rate": 7.045841498281346e-06,
+      "loss": 0.1899,
+      "step": 9700
+    },
+    {
+      "epoch": 0.9224129754339372,
+      "grad_norm": 6.137265205383301,
+      "learning_rate": 7.038131645732276e-06,
+      "loss": 0.2709,
+      "step": 9725
+    },
+    {
+      "epoch": 0.9247842170160296,
+      "grad_norm": 4.024059295654297,
+      "learning_rate": 7.0301005493269945e-06,
+      "loss": 0.1773,
+      "step": 9750
+    },
+    {
+      "epoch": 0.927155458598122,
+      "grad_norm": 4.895248889923096,
+      "learning_rate": 7.022069452921714e-06,
+      "loss": 0.2164,
+      "step": 9775
+    },
+    {
+      "epoch": 0.9295267001802143,
+      "grad_norm": 4.847474098205566,
+      "learning_rate": 7.0140383565164325e-06,
+      "loss": 0.2139,
+      "step": 9800
+    },
+    {
+      "epoch": 0.9318979417623068,
+      "grad_norm": 7.41204833984375,
+      "learning_rate": 7.00600726011115e-06,
+      "loss": 0.192,
+      "step": 9825
+    },
+    {
+      "epoch": 0.9342691833443991,
+      "grad_norm": 7.142247200012207,
+      "learning_rate": 6.99797616370587e-06,
+      "loss": 0.2149,
+      "step": 9850
+    },
+    {
+      "epoch": 0.9366404249264915,
+      "grad_norm": 7.4694132804870605,
+      "learning_rate": 6.989945067300589e-06,
+      "loss": 0.1743,
+      "step": 9875
+    },
+    {
+      "epoch": 0.9390116665085839,
+      "grad_norm": 4.399720191955566,
+      "learning_rate": 6.9819139708953076e-06,
+      "loss": 0.1863,
+      "step": 9900
+    },
+    {
+      "epoch": 0.9413829080906763,
+      "grad_norm": 6.795614242553711,
+      "learning_rate": 6.973882874490025e-06,
+      "loss": 0.2048,
+      "step": 9925
+    },
+    {
+      "epoch": 0.9437541496727687,
+      "grad_norm": 6.58676815032959,
+      "learning_rate": 6.965851778084745e-06,
+      "loss": 0.1831,
+      "step": 9950
+    },
+    {
+      "epoch": 0.9461253912548611,
+      "grad_norm": 5.277420520782471,
+      "learning_rate": 6.957820681679463e-06,
+      "loss": 0.2051,
+      "step": 9975
+    },
+    {
+      "epoch": 0.9484966328369534,
+      "grad_norm": 6.933620929718018,
+      "learning_rate": 6.949789585274183e-06,
+      "loss": 0.1788,
+      "step": 10000
+    },
+    {
+      "epoch": 0.9484966328369534,
+      "eval_loss": 0.21403691172599792,
+      "eval_runtime": 623.1408,
+      "eval_samples_per_second": 2.032,
+      "eval_steps_per_second": 0.255,
+      "eval_wer": 45.486411216748294,
+      "step": 10000
+    },
+    {
+      "epoch": 0.9508678744190459,
+      "grad_norm": 15.806493759155273,
+      "learning_rate": 6.9417584888689e-06,
+      "loss": 0.2154,
+      "step": 10025
+    },
+    {
+      "epoch": 0.9532391160011382,
+      "grad_norm": 4.786881446838379,
+      "learning_rate": 6.93372739246362e-06,
+      "loss": 0.1989,
+      "step": 10050
+    },
+    {
+      "epoch": 0.9556103575832305,
+      "grad_norm": 5.720562934875488,
+      "learning_rate": 6.925696296058338e-06,
+      "loss": 0.1868,
+      "step": 10075
+    },
+    {
+      "epoch": 0.957981599165323,
+      "grad_norm": 7.773390293121338,
+      "learning_rate": 6.917665199653058e-06,
+      "loss": 0.2461,
+      "step": 10100
+    },
+    {
+      "epoch": 0.9603528407474153,
+      "grad_norm": 6.787060260772705,
+      "learning_rate": 6.909634103247776e-06,
+      "loss": 0.2037,
+      "step": 10125
+    },
+    {
+      "epoch": 0.9627240823295077,
+      "grad_norm": 6.018710136413574,
+      "learning_rate": 6.901603006842495e-06,
+      "loss": 0.1903,
+      "step": 10150
+    },
+    {
+      "epoch": 0.9650953239116001,
+      "grad_norm": 5.613012790679932,
+      "learning_rate": 6.893571910437213e-06,
+      "loss": 0.2026,
+      "step": 10175
+    },
+    {
+      "epoch": 0.9674665654936925,
+      "grad_norm": 7.66652250289917,
+      "learning_rate": 6.885540814031933e-06,
+      "loss": 0.2319,
+      "step": 10200
+    },
+    {
+      "epoch": 0.9698378070757849,
+      "grad_norm": 5.4878411293029785,
+      "learning_rate": 6.877509717626651e-06,
+      "loss": 0.1656,
+      "step": 10225
+    },
+    {
+      "epoch": 0.9722090486578773,
+      "grad_norm": 5.760538578033447,
+      "learning_rate": 6.869478621221369e-06,
+      "loss": 0.2308,
+      "step": 10250
+    },
+    {
+      "epoch": 0.9745802902399696,
+      "grad_norm": 6.5572919845581055,
+      "learning_rate": 6.861447524816088e-06,
+      "loss": 0.2121,
+      "step": 10275
+    },
+    {
+      "epoch": 0.9769515318220621,
+      "grad_norm": 5.401267051696777,
+      "learning_rate": 6.853416428410807e-06,
+      "loss": 0.207,
+      "step": 10300
+    },
+    {
+      "epoch": 0.9793227734041544,
+      "grad_norm": 7.6488871574401855,
+      "learning_rate": 6.845385332005526e-06,
+      "loss": 0.2126,
+      "step": 10325
+    },
+    {
+      "epoch": 0.9816940149862468,
+      "grad_norm": 5.2823662757873535,
+      "learning_rate": 6.837354235600244e-06,
+      "loss": 0.1937,
+      "step": 10350
+    },
+    {
+      "epoch": 0.9840652565683392,
+      "grad_norm": 3.183607339859009,
+      "learning_rate": 6.8293231391949634e-06,
+      "loss": 0.188,
+      "step": 10375
+    },
+    {
+      "epoch": 0.9864364981504316,
+      "grad_norm": 6.274131774902344,
+      "learning_rate": 6.821292042789682e-06,
+      "loss": 0.262,
+      "step": 10400
+    },
+    {
+      "epoch": 0.988807739732524,
+      "grad_norm": 7.148483753204346,
+      "learning_rate": 6.813260946384401e-06,
+      "loss": 0.1966,
+      "step": 10425
+    },
+    {
+      "epoch": 0.9911789813146163,
+      "grad_norm": 6.260406494140625,
+      "learning_rate": 6.805229849979119e-06,
+      "loss": 0.2239,
+      "step": 10450
+    },
+    {
+      "epoch": 0.9935502228967087,
+      "grad_norm": 9.445612907409668,
+      "learning_rate": 6.7971987535738385e-06,
+      "loss": 0.1942,
+      "step": 10475
+    },
+    {
+      "epoch": 0.9959214644788011,
+      "grad_norm": 7.5293474197387695,
+      "learning_rate": 6.789167657168557e-06,
+      "loss": 0.2131,
+      "step": 10500
+    },
+    {
+      "epoch": 0.9982927060608935,
+      "grad_norm": 5.564962387084961,
+      "learning_rate": 6.7811365607632765e-06,
+      "loss": 0.2211,
+      "step": 10525
+    },
+    {
+      "epoch": 1.000663947642986,
+      "grad_norm": 7.690740585327148,
+      "learning_rate": 6.773105464357995e-06,
+      "loss": 0.1954,
+      "step": 10550
+    },
+    {
+      "epoch": 1.0030351892250782,
+      "grad_norm": 7.644260406494141,
+      "learning_rate": 6.765074367952713e-06,
+      "loss": 0.1167,
+      "step": 10575
+    },
+    {
+      "epoch": 1.0054064308071706,
+      "grad_norm": 6.853944301605225,
+      "learning_rate": 6.757043271547432e-06,
+      "loss": 0.1095,
+      "step": 10600
+    },
+    {
+      "epoch": 1.007777672389263,
+      "grad_norm": 7.964993953704834,
+      "learning_rate": 6.7490121751421515e-06,
+      "loss": 0.1041,
+      "step": 10625
+    },
+    {
+      "epoch": 1.0101489139713553,
+      "grad_norm": 10.254597663879395,
+      "learning_rate": 6.74098107873687e-06,
+      "loss": 0.1259,
+      "step": 10650
+    },
+    {
+      "epoch": 1.0125201555534478,
+      "grad_norm": 7.004283905029297,
+      "learning_rate": 6.732949982331588e-06,
+      "loss": 0.0998,
+      "step": 10675
+    },
+    {
+      "epoch": 1.0148913971355402,
+      "grad_norm": 5.59473991394043,
+      "learning_rate": 6.724918885926307e-06,
+      "loss": 0.1055,
+      "step": 10700
+    },
+    {
+      "epoch": 1.0172626387176325,
+      "grad_norm": 5.191290855407715,
+      "learning_rate": 6.716887789521026e-06,
+      "loss": 0.1066,
+      "step": 10725
+    },
+    {
+      "epoch": 1.019633880299725,
+      "grad_norm": 6.007319927215576,
+      "learning_rate": 6.708856693115745e-06,
+      "loss": 0.1033,
+      "step": 10750
+    },
+    {
+      "epoch": 1.0220051218818174,
+      "grad_norm": 4.007826805114746,
+      "learning_rate": 6.700825596710463e-06,
+      "loss": 0.0991,
+      "step": 10775
+    },
+    {
+      "epoch": 1.0243763634639098,
+      "grad_norm": 5.704361438751221,
+      "learning_rate": 6.692794500305182e-06,
+      "loss": 0.121,
+      "step": 10800
+    },
+    {
+      "epoch": 1.026747605046002,
+      "grad_norm": 5.501503944396973,
+      "learning_rate": 6.684763403899901e-06,
+      "loss": 0.1233,
+      "step": 10825
+    },
+    {
+      "epoch": 1.0291188466280945,
+      "grad_norm": 2.979905843734741,
+      "learning_rate": 6.67673230749462e-06,
+      "loss": 0.1252,
+      "step": 10850
+    },
+    {
+      "epoch": 1.031490088210187,
+      "grad_norm": 6.167366981506348,
+      "learning_rate": 6.668701211089339e-06,
+      "loss": 0.1056,
+      "step": 10875
+    },
+    {
+      "epoch": 1.0338613297922792,
+      "grad_norm": 7.7317609786987305,
+      "learning_rate": 6.660670114684057e-06,
+      "loss": 0.1127,
+      "step": 10900
+    },
+    {
+      "epoch": 1.0362325713743716,
+      "grad_norm": 7.994433879852295,
+      "learning_rate": 6.652639018278776e-06,
+      "loss": 0.1132,
+      "step": 10925
+    },
+    {
+      "epoch": 1.038603812956464,
+      "grad_norm": 5.057292938232422,
+      "learning_rate": 6.644607921873495e-06,
+      "loss": 0.1056,
+      "step": 10950
+    },
+    {
+      "epoch": 1.0409750545385563,
+      "grad_norm": 5.171298980712891,
+      "learning_rate": 6.636576825468214e-06,
+      "loss": 0.1163,
+      "step": 10975
+    },
+    {
+      "epoch": 1.0433462961206488,
+      "grad_norm": 7.75325870513916,
+      "learning_rate": 6.6285457290629315e-06,
+      "loss": 0.125,
+      "step": 11000
+    },
+    {
+      "epoch": 1.0433462961206488,
+      "eval_loss": 0.21488377451896667,
+      "eval_runtime": 552.8301,
+      "eval_samples_per_second": 2.29,
+      "eval_steps_per_second": 0.288,
+      "eval_wer": 25.247287045039858,
+      "step": 11000
+    },
+    {
+      "epoch": 1.0457175377027412,
+      "grad_norm": 6.812617778778076,
+      "learning_rate": 6.620514632657651e-06,
+      "loss": 0.124,
+      "step": 11025
+    },
+    {
+      "epoch": 1.0480887792848335,
+      "grad_norm": 5.972348213195801,
+      "learning_rate": 6.6124835362523694e-06,
+      "loss": 0.1259,
+      "step": 11050
+    },
+    {
+      "epoch": 1.050460020866926,
+      "grad_norm": 7.062375545501709,
+      "learning_rate": 6.604452439847089e-06,
+      "loss": 0.1128,
+      "step": 11075
+    },
+    {
+      "epoch": 1.0528312624490184,
+      "grad_norm": 7.601444244384766,
+      "learning_rate": 6.5964213434418065e-06,
+      "loss": 0.1224,
+      "step": 11100
+    },
+    {
+      "epoch": 1.0552025040311106,
+      "grad_norm": 4.812473773956299,
+      "learning_rate": 6.588390247036526e-06,
+      "loss": 0.1072,
+      "step": 11125
+    },
+    {
+      "epoch": 1.057573745613203,
+      "grad_norm": 4.50166130065918,
+      "learning_rate": 6.5803591506312445e-06,
+      "loss": 0.1134,
+      "step": 11150
+    },
+    {
+      "epoch": 1.0599449871952955,
+      "grad_norm": 3.7743520736694336,
+      "learning_rate": 6.572649298082174e-06,
+      "loss": 0.1383,
+      "step": 11175
+    },
+    {
+      "epoch": 1.0623162287773877,
+      "grad_norm": 7.0009331703186035,
+      "learning_rate": 6.564618201676894e-06,
+      "loss": 0.1248,
+      "step": 11200
+    },
+    {
+      "epoch": 1.0646874703594802,
+      "grad_norm": 5.16038179397583,
+      "learning_rate": 6.556587105271612e-06,
+      "loss": 0.1208,
+      "step": 11225
+    },
+    {
+      "epoch": 1.0670587119415726,
+      "grad_norm": 6.29218864440918,
+      "learning_rate": 6.548556008866332e-06,
+      "loss": 0.1432,
+      "step": 11250
+    },
+    {
+      "epoch": 1.0694299535236649,
+      "grad_norm": 3.430107355117798,
+      "learning_rate": 6.540524912461049e-06,
+      "loss": 0.1222,
+      "step": 11275
+    },
+    {
+      "epoch": 1.0718011951057573,
+      "grad_norm": 7.517146587371826,
+      "learning_rate": 6.532493816055768e-06,
+      "loss": 0.1417,
+      "step": 11300
+    },
+    {
+      "epoch": 1.0741724366878498,
+      "grad_norm": 13.580179214477539,
+      "learning_rate": 6.524462719650487e-06,
+      "loss": 0.1478,
+      "step": 11325
+    },
+    {
+      "epoch": 1.0765436782699422,
+      "grad_norm": 5.569423675537109,
+      "learning_rate": 6.516431623245206e-06,
+      "loss": 0.1653,
+      "step": 11350
+    },
+    {
+      "epoch": 1.0789149198520345,
+      "grad_norm": 9.024543762207031,
+      "learning_rate": 6.5084005268399245e-06,
+      "loss": 0.1315,
+      "step": 11375
+    },
+    {
+      "epoch": 1.081286161434127,
+      "grad_norm": 3.7613959312438965,
+      "learning_rate": 6.500369430434643e-06,
+      "loss": 0.1026,
+      "step": 11400
+    },
+    {
+      "epoch": 1.0836574030162194,
+      "grad_norm": 5.981335639953613,
+      "learning_rate": 6.492338334029362e-06,
+      "loss": 0.0916,
+      "step": 11425
+    },
+    {
+      "epoch": 1.0860286445983116,
+      "grad_norm": 5.735662937164307,
+      "learning_rate": 6.484307237624081e-06,
+      "loss": 0.111,
+      "step": 11450
+    },
+    {
+      "epoch": 1.088399886180404,
+      "grad_norm": 2.677314043045044,
+      "learning_rate": 6.4762761412188e-06,
+      "loss": 0.1018,
+      "step": 11475
+    },
+    {
+      "epoch": 1.0907711277624965,
+      "grad_norm": 5.175323009490967,
+      "learning_rate": 6.468245044813518e-06,
+      "loss": 0.1275,
+      "step": 11500
+    },
+    {
+      "epoch": 1.0931423693445887,
+      "grad_norm": 4.597871780395508,
+      "learning_rate": 6.4602139484082375e-06,
+      "loss": 0.1098,
+      "step": 11525
+    },
+    {
+      "epoch": 1.0955136109266812,
+      "grad_norm": 3.1632707118988037,
+      "learning_rate": 6.452182852002956e-06,
+      "loss": 0.0969,
+      "step": 11550
+    },
+    {
+      "epoch": 1.0978848525087737,
+      "grad_norm": 5.567389011383057,
+      "learning_rate": 6.444151755597675e-06,
+      "loss": 0.1232,
+      "step": 11575
+    },
+    {
+      "epoch": 1.1002560940908659,
+      "grad_norm": 3.456010103225708,
+      "learning_rate": 6.436120659192393e-06,
+      "loss": 0.1239,
+      "step": 11600
+    },
+    {
+      "epoch": 1.1026273356729583,
+      "grad_norm": 3.5885462760925293,
+      "learning_rate": 6.428089562787112e-06,
+      "loss": 0.1358,
+      "step": 11625
+    },
+    {
+      "epoch": 1.1049985772550508,
+      "grad_norm": 5.118311405181885,
+      "learning_rate": 6.420058466381831e-06,
+      "loss": 0.112,
+      "step": 11650
+    },
+    {
+      "epoch": 1.107369818837143,
+      "grad_norm": 3.1795310974121094,
+      "learning_rate": 6.4120273699765505e-06,
+      "loss": 0.1127,
+      "step": 11675
+    },
+    {
+      "epoch": 1.1097410604192355,
+      "grad_norm": 3.9411444664001465,
+      "learning_rate": 6.403996273571268e-06,
+      "loss": 0.1127,
+      "step": 11700
+    },
+    {
+      "epoch": 1.112112302001328,
+      "grad_norm": 7.419850826263428,
+      "learning_rate": 6.395965177165987e-06,
+      "loss": 0.1269,
+      "step": 11725
+    },
+    {
+      "epoch": 1.1144835435834204,
+      "grad_norm": 5.062563896179199,
+      "learning_rate": 6.387934080760706e-06,
+      "loss": 0.1108,
+      "step": 11750
+    },
+    {
+      "epoch": 1.1168547851655126,
+      "grad_norm": 8.336091041564941,
+      "learning_rate": 6.379902984355425e-06,
+      "loss": 0.1442,
+      "step": 11775
+    },
+    {
+      "epoch": 1.119226026747605,
+      "grad_norm": 2.9433302879333496,
+      "learning_rate": 6.371871887950144e-06,
+      "loss": 0.1313,
+      "step": 11800
+    },
+    {
+      "epoch": 1.1215972683296975,
+      "grad_norm": 3.0976691246032715,
+      "learning_rate": 6.363840791544862e-06,
+      "loss": 0.1139,
+      "step": 11825
+    },
+    {
+      "epoch": 1.1239685099117898,
+      "grad_norm": 2.8985209465026855,
+      "learning_rate": 6.355809695139581e-06,
+      "loss": 0.1147,
+      "step": 11850
+    },
+    {
+      "epoch": 1.1263397514938822,
+      "grad_norm": 5.217418193817139,
+      "learning_rate": 6.3477785987343e-06,
+      "loss": 0.0958,
+      "step": 11875
+    },
+    {
+      "epoch": 1.1287109930759747,
+      "grad_norm": 5.1537604331970215,
+      "learning_rate": 6.339747502329019e-06,
+      "loss": 0.1465,
+      "step": 11900
+    },
+    {
+      "epoch": 1.131082234658067,
+      "grad_norm": 7.060305595397949,
+      "learning_rate": 6.331716405923737e-06,
+      "loss": 0.145,
+      "step": 11925
+    },
+    {
+      "epoch": 1.1334534762401594,
+      "grad_norm": 4.791098117828369,
+      "learning_rate": 6.323685309518456e-06,
+      "loss": 0.113,
+      "step": 11950
+    },
+    {
+      "epoch": 1.1358247178222518,
+      "grad_norm": 6.57902193069458,
+      "learning_rate": 6.315654213113175e-06,
+      "loss": 0.115,
+      "step": 11975
+    },
+    {
+      "epoch": 1.138195959404344,
+      "grad_norm": 5.277735710144043,
+      "learning_rate": 6.307623116707894e-06,
+      "loss": 0.113,
+      "step": 12000
+    },
+    {
+      "epoch": 1.138195959404344,
+      "eval_loss": 0.2158912867307663,
+      "eval_runtime": 574.1987,
+      "eval_samples_per_second": 2.205,
+      "eval_steps_per_second": 0.277,
+      "eval_wer": 28.87736483242101,
+      "step": 12000
+    },
+    {
+      "epoch": 1.1405672009864365,
+      "grad_norm": 4.220649719238281,
+      "learning_rate": 6.299592020302612e-06,
+      "loss": 0.1332,
+      "step": 12025
+    },
+    {
+      "epoch": 1.142938442568529,
+      "grad_norm": 3.2870547771453857,
+      "learning_rate": 6.2915609238973305e-06,
+      "loss": 0.1016,
+      "step": 12050
+    },
+    {
+      "epoch": 1.1453096841506212,
+      "grad_norm": 3.2950220108032227,
+      "learning_rate": 6.28352982749205e-06,
+      "loss": 0.1381,
+      "step": 12075
+    },
+    {
+      "epoch": 1.1476809257327136,
+      "grad_norm": 4.546053409576416,
+      "learning_rate": 6.275498731086768e-06,
+      "loss": 0.1149,
+      "step": 12100
+    },
+    {
+      "epoch": 1.150052167314806,
+      "grad_norm": 2.389995574951172,
+      "learning_rate": 6.267467634681488e-06,
+      "loss": 0.1006,
+      "step": 12125
+    },
+    {
+      "epoch": 1.1524234088968983,
+      "grad_norm": 6.161362171173096,
+      "learning_rate": 6.2594365382762055e-06,
+      "loss": 0.1267,
+      "step": 12150
+    },
+    {
+      "epoch": 1.1547946504789908,
+      "grad_norm": 4.0460710525512695,
+      "learning_rate": 6.251405441870925e-06,
+      "loss": 0.1471,
+      "step": 12175
+    },
+    {
+      "epoch": 1.1571658920610832,
+      "grad_norm": 4.770595073699951,
+      "learning_rate": 6.2433743454656435e-06,
+      "loss": 0.1207,
+      "step": 12200
+    },
+    {
+      "epoch": 1.1595371336431755,
+      "grad_norm": 5.524403095245361,
+      "learning_rate": 6.235343249060363e-06,
+      "loss": 0.1031,
+      "step": 12225
+    },
+    {
+      "epoch": 1.161908375225268,
+      "grad_norm": 3.3218834400177,
+      "learning_rate": 6.2273121526550806e-06,
+      "loss": 0.147,
+      "step": 12250
+    },
+    {
+      "epoch": 1.1642796168073604,
+      "grad_norm": 4.686155796051025,
+      "learning_rate": 6.2192810562498e-06,
+      "loss": 0.102,
+      "step": 12275
+    },
+    {
+      "epoch": 1.1666508583894526,
+      "grad_norm": 5.86762809753418,
+      "learning_rate": 6.2112499598445185e-06,
+      "loss": 0.1291,
+      "step": 12300
+    },
+    {
+      "epoch": 1.169022099971545,
+      "grad_norm": 5.153403282165527,
+      "learning_rate": 6.203218863439238e-06,
+      "loss": 0.1104,
+      "step": 12325
+    },
+    {
+      "epoch": 1.1713933415536375,
+      "grad_norm": 6.292455673217773,
+      "learning_rate": 6.195187767033956e-06,
+      "loss": 0.0963,
+      "step": 12350
+    },
+    {
+      "epoch": 1.17376458313573,
+      "grad_norm": 6.645671844482422,
+      "learning_rate": 6.187156670628674e-06,
+      "loss": 0.1238,
+      "step": 12375
+    },
+    {
+      "epoch": 1.1761358247178222,
+      "grad_norm": 4.748604774475098,
+      "learning_rate": 6.179125574223394e-06,
+      "loss": 0.126,
+      "step": 12400
+    },
+    {
+      "epoch": 1.1785070662999146,
+      "grad_norm": 4.2247748374938965,
+      "learning_rate": 6.171094477818112e-06,
+      "loss": 0.0806,
+      "step": 12425
+    },
+    {
+      "epoch": 1.180878307882007,
+      "grad_norm": 9.339821815490723,
+      "learning_rate": 6.163063381412831e-06,
+      "loss": 0.1351,
+      "step": 12450
+    },
+    {
+      "epoch": 1.1832495494640993,
+      "grad_norm": 3.476694107055664,
+      "learning_rate": 6.155032285007549e-06,
+      "loss": 0.1037,
+      "step": 12475
+    },
+    {
+      "epoch": 1.1856207910461918,
+      "grad_norm": 9.010672569274902,
+      "learning_rate": 6.147001188602269e-06,
+      "loss": 0.1099,
+      "step": 12500
+    },
+    {
+      "epoch": 1.1879920326282842,
+      "grad_norm": 4.160516262054443,
+      "learning_rate": 6.138970092196987e-06,
+      "loss": 0.1105,
+      "step": 12525
+    },
+    {
+      "epoch": 1.1903632742103765,
+      "grad_norm": 6.462663650512695,
+      "learning_rate": 6.130938995791707e-06,
+      "loss": 0.1085,
+      "step": 12550
+    },
+    {
+      "epoch": 1.192734515792469,
+      "grad_norm": 3.9349684715270996,
+      "learning_rate": 6.122907899386424e-06,
+      "loss": 0.1422,
+      "step": 12575
+    },
+    {
+      "epoch": 1.1951057573745614,
+      "grad_norm": 3.4753317832946777,
+      "learning_rate": 6.114876802981144e-06,
+      "loss": 0.1413,
+      "step": 12600
+    },
+    {
+      "epoch": 1.1974769989566538,
+      "grad_norm": 7.593752384185791,
+      "learning_rate": 6.106845706575862e-06,
+      "loss": 0.1145,
+      "step": 12625
+    },
+    {
+      "epoch": 1.199848240538746,
+      "grad_norm": 7.460291862487793,
+      "learning_rate": 6.098814610170582e-06,
+      "loss": 0.1172,
+      "step": 12650
+    },
+    {
+      "epoch": 1.2022194821208385,
+      "grad_norm": 2.9689126014709473,
+      "learning_rate": 6.090783513765299e-06,
+      "loss": 0.1041,
+      "step": 12675
+    },
+    {
+      "epoch": 1.204590723702931,
+      "grad_norm": 4.123712062835693,
+      "learning_rate": 6.082752417360018e-06,
+      "loss": 0.1087,
+      "step": 12700
+    },
+    {
+      "epoch": 1.2069619652850232,
+      "grad_norm": 3.49700665473938,
+      "learning_rate": 6.074721320954737e-06,
+      "loss": 0.124,
+      "step": 12725
+    },
+    {
+      "epoch": 1.2093332068671157,
+      "grad_norm": 8.247212409973145,
+      "learning_rate": 6.066690224549457e-06,
+      "loss": 0.1299,
+      "step": 12750
+    },
+    {
+      "epoch": 1.211704448449208,
+      "grad_norm": 6.524272441864014,
+      "learning_rate": 6.058659128144174e-06,
+      "loss": 0.1211,
+      "step": 12775
+    },
+    {
+      "epoch": 1.2140756900313003,
+      "grad_norm": 9.503357887268066,
+      "learning_rate": 6.050628031738893e-06,
+      "loss": 0.1359,
+      "step": 12800
+    },
+    {
+      "epoch": 1.2164469316133928,
+      "grad_norm": 5.396457672119141,
+      "learning_rate": 6.042596935333612e-06,
+      "loss": 0.1058,
+      "step": 12825
+    },
+    {
+      "epoch": 1.2188181731954852,
+      "grad_norm": 2.986445426940918,
+      "learning_rate": 6.034565838928331e-06,
+      "loss": 0.1106,
+      "step": 12850
+    },
+    {
+      "epoch": 1.2211894147775775,
+      "grad_norm": 6.511775493621826,
+      "learning_rate": 6.02653474252305e-06,
+      "loss": 0.1366,
+      "step": 12875
+    },
+    {
+      "epoch": 1.22356065635967,
+      "grad_norm": 4.643221855163574,
+      "learning_rate": 6.018503646117768e-06,
+      "loss": 0.0988,
+      "step": 12900
+    },
+    {
+      "epoch": 1.2259318979417624,
+      "grad_norm": 5.791896820068359,
+      "learning_rate": 6.010472549712487e-06,
+      "loss": 0.1292,
+      "step": 12925
+    },
+    {
+      "epoch": 1.2283031395238546,
+      "grad_norm": 4.876262187957764,
+      "learning_rate": 6.002441453307206e-06,
+      "loss": 0.1271,
+      "step": 12950
+    },
+    {
+      "epoch": 1.230674381105947,
+      "grad_norm": 3.594116449356079,
+      "learning_rate": 5.994410356901925e-06,
+      "loss": 0.1165,
+      "step": 12975
+    },
+    {
+      "epoch": 1.2330456226880395,
+      "grad_norm": 3.887071132659912,
+      "learning_rate": 5.986379260496643e-06,
+      "loss": 0.0986,
+      "step": 13000
+    },
+    {
+      "epoch": 1.2330456226880395,
+      "eval_loss": 0.21452254056930542,
+      "eval_runtime": 505.9784,
+      "eval_samples_per_second": 2.502,
+      "eval_steps_per_second": 0.314,
+      "eval_wer": 15.78315567079612,
+      "step": 13000
+    },
+    {
+      "epoch": 1.2354168642701318,
+      "grad_norm": 6.891282558441162,
+      "learning_rate": 5.9783481640913625e-06,
+      "loss": 0.1243,
+      "step": 13025
+    },
+    {
+      "epoch": 1.2377881058522242,
+      "grad_norm": 6.045233726501465,
+      "learning_rate": 5.970317067686081e-06,
+      "loss": 0.141,
+      "step": 13050
+    },
+    {
+      "epoch": 1.2401593474343167,
+      "grad_norm": 4.684844017028809,
+      "learning_rate": 5.9622859712808004e-06,
+      "loss": 0.1192,
+      "step": 13075
+    },
+    {
+      "epoch": 1.242530589016409,
+      "grad_norm": 3.5737416744232178,
+      "learning_rate": 5.954254874875518e-06,
+      "loss": 0.1014,
+      "step": 13100
+    },
+    {
+      "epoch": 1.2449018305985013,
+      "grad_norm": 3.9428131580352783,
+      "learning_rate": 5.946223778470237e-06,
+      "loss": 0.1293,
+      "step": 13125
+    },
+    {
+      "epoch": 1.2472730721805938,
+      "grad_norm": 4.936779022216797,
+      "learning_rate": 5.938192682064956e-06,
+      "loss": 0.1054,
+      "step": 13150
+    },
+    {
+      "epoch": 1.249644313762686,
+      "grad_norm": 5.83259916305542,
+      "learning_rate": 5.930161585659675e-06,
+      "loss": 0.0962,
+      "step": 13175
+    },
+    {
+      "epoch": 1.2520155553447785,
+      "grad_norm": 3.2616126537323,
+      "learning_rate": 5.922130489254394e-06,
+      "loss": 0.1125,
+      "step": 13200
+    },
+    {
+      "epoch": 1.254386796926871,
+      "grad_norm": 4.956861972808838,
+      "learning_rate": 5.914099392849112e-06,
+      "loss": 0.1328,
+      "step": 13225
+    },
+    {
+      "epoch": 1.2567580385089632,
+      "grad_norm": 5.116754531860352,
+      "learning_rate": 5.906068296443831e-06,
+      "loss": 0.1087,
+      "step": 13250
+    },
+    {
+      "epoch": 1.2591292800910556,
+      "grad_norm": 2.1782491207122803,
+      "learning_rate": 5.89803720003855e-06,
+      "loss": 0.1068,
+      "step": 13275
+    },
+    {
+      "epoch": 1.261500521673148,
+      "grad_norm": 5.729766368865967,
+      "learning_rate": 5.890006103633269e-06,
+      "loss": 0.1136,
+      "step": 13300
+    },
+    {
+      "epoch": 1.2638717632552403,
+      "grad_norm": 5.741735935211182,
+      "learning_rate": 5.881975007227987e-06,
+      "loss": 0.1103,
+      "step": 13325
+    },
+    {
+      "epoch": 1.2662430048373328,
+      "grad_norm": 4.753970623016357,
+      "learning_rate": 5.873943910822706e-06,
+      "loss": 0.1025,
+      "step": 13350
+    },
+    {
+      "epoch": 1.2686142464194252,
+      "grad_norm": 6.524509429931641,
+      "learning_rate": 5.865912814417425e-06,
+      "loss": 0.1228,
+      "step": 13375
+    },
+    {
+      "epoch": 1.2709854880015177,
+      "grad_norm": 4.528171539306641,
+      "learning_rate": 5.857881718012144e-06,
+      "loss": 0.1262,
+      "step": 13400
+    },
+    {
+      "epoch": 1.27335672958361,
+      "grad_norm": 6.253600120544434,
+      "learning_rate": 5.849850621606862e-06,
+      "loss": 0.0991,
+      "step": 13425
+    },
+    {
+      "epoch": 1.2757279711657024,
+      "grad_norm": 3.7937471866607666,
+      "learning_rate": 5.84181952520158e-06,
+      "loss": 0.1034,
+      "step": 13450
+    },
+    {
+      "epoch": 1.2780992127477948,
+      "grad_norm": 6.3199310302734375,
+      "learning_rate": 5.8337884287963e-06,
+      "loss": 0.1162,
+      "step": 13475
+    },
+    {
+      "epoch": 1.2804704543298873,
+      "grad_norm": 4.21429967880249,
+      "learning_rate": 5.825757332391019e-06,
+      "loss": 0.1005,
+      "step": 13500
+    },
+    {
+      "epoch": 1.2828416959119795,
+      "grad_norm": 8.968953132629395,
+      "learning_rate": 5.817726235985737e-06,
+      "loss": 0.1094,
+      "step": 13525
+    },
+    {
+      "epoch": 1.285212937494072,
+      "grad_norm": 4.219638347625732,
+      "learning_rate": 5.8096951395804555e-06,
+      "loss": 0.1007,
+      "step": 13550
+    },
+    {
+      "epoch": 1.2875841790761644,
+      "grad_norm": 4.303260803222656,
+      "learning_rate": 5.801664043175175e-06,
+      "loss": 0.1143,
+      "step": 13575
+    },
+    {
+      "epoch": 1.2899554206582566,
+      "grad_norm": 9.81352710723877,
+      "learning_rate": 5.793632946769893e-06,
+      "loss": 0.1217,
+      "step": 13600
+    },
+    {
+      "epoch": 1.292326662240349,
+      "grad_norm": 5.257843494415283,
+      "learning_rate": 5.785601850364613e-06,
+      "loss": 0.0905,
+      "step": 13625
+    },
+    {
+      "epoch": 1.2946979038224415,
+      "grad_norm": 6.5194597244262695,
+      "learning_rate": 5.7775707539593305e-06,
+      "loss": 0.0947,
+      "step": 13650
+    },
+    {
+      "epoch": 1.2970691454045338,
+      "grad_norm": 4.982461452484131,
+      "learning_rate": 5.76953965755405e-06,
+      "loss": 0.1114,
+      "step": 13675
+    },
+    {
+      "epoch": 1.2994403869866262,
+      "grad_norm": 8.524166107177734,
+      "learning_rate": 5.7615085611487685e-06,
+      "loss": 0.1087,
+      "step": 13700
+    },
+    {
+      "epoch": 1.3018116285687187,
+      "grad_norm": 10.688188552856445,
+      "learning_rate": 5.753477464743488e-06,
+      "loss": 0.1146,
+      "step": 13725
+    },
+    {
+      "epoch": 1.304182870150811,
+      "grad_norm": 3.5015459060668945,
+      "learning_rate": 5.745446368338206e-06,
+      "loss": 0.1036,
+      "step": 13750
+    },
+    {
+      "epoch": 1.3065541117329034,
+      "grad_norm": 6.868871212005615,
+      "learning_rate": 5.737415271932925e-06,
+      "loss": 0.1232,
+      "step": 13775
+    },
+    {
+      "epoch": 1.3089253533149958,
+      "grad_norm": 4.6430840492248535,
+      "learning_rate": 5.7293841755276435e-06,
+      "loss": 0.1043,
+      "step": 13800
+    },
+    {
+      "epoch": 1.311296594897088,
+      "grad_norm": 5.601919651031494,
+      "learning_rate": 5.721353079122363e-06,
+      "loss": 0.1512,
+      "step": 13825
+    },
+    {
+      "epoch": 1.3136678364791805,
+      "grad_norm": 2.6990678310394287,
+      "learning_rate": 5.713321982717081e-06,
+      "loss": 0.1317,
+      "step": 13850
+    },
+    {
+      "epoch": 1.316039078061273,
+      "grad_norm": 7.8757524490356445,
+      "learning_rate": 5.705290886311799e-06,
+      "loss": 0.1446,
+      "step": 13875
+    },
+    {
+      "epoch": 1.3184103196433652,
+      "grad_norm": 5.1340837478637695,
+      "learning_rate": 5.697259789906519e-06,
+      "loss": 0.0972,
+      "step": 13900
+    },
+    {
+      "epoch": 1.3207815612254576,
+      "grad_norm": 5.61411714553833,
+      "learning_rate": 5.689228693501237e-06,
+      "loss": 0.0987,
+      "step": 13925
+    },
+    {
+      "epoch": 1.32315280280755,
+      "grad_norm": 5.238297462463379,
+      "learning_rate": 5.6811975970959565e-06,
+      "loss": 0.1397,
+      "step": 13950
+    },
+    {
+      "epoch": 1.3255240443896423,
+      "grad_norm": 9.023843765258789,
+      "learning_rate": 5.673166500690674e-06,
+      "loss": 0.109,
+      "step": 13975
+    },
+    {
+      "epoch": 1.3278952859717348,
+      "grad_norm": 3.5052764415740967,
+      "learning_rate": 5.665135404285394e-06,
+      "loss": 0.1105,
+      "step": 14000
+    },
+    {
+      "epoch": 1.3278952859717348,
+      "eval_loss": 0.21087968349456787,
+      "eval_runtime": 519.3229,
+      "eval_samples_per_second": 2.438,
+      "eval_steps_per_second": 0.306,
+      "eval_wer": 15.658311725727456,
+      "step": 14000
+    },
+    {
+      "epoch": 1.3302665275538272,
+      "grad_norm": 5.8544840812683105,
+      "learning_rate": 5.657104307880112e-06,
+      "loss": 0.1405,
+      "step": 14025
+    },
+    {
+      "epoch": 1.3326377691359195,
+      "grad_norm": 4.098303318023682,
+      "learning_rate": 5.649073211474832e-06,
+      "loss": 0.1097,
+      "step": 14050
+    },
+    {
+      "epoch": 1.335009010718012,
+      "grad_norm": 3.10842227935791,
+      "learning_rate": 5.641042115069549e-06,
+      "loss": 0.1242,
+      "step": 14075
+    },
+    {
+      "epoch": 1.3373802523001044,
+      "grad_norm": 4.459357738494873,
+      "learning_rate": 5.633011018664269e-06,
+      "loss": 0.1036,
+      "step": 14100
+    },
+    {
+      "epoch": 1.3397514938821966,
+      "grad_norm": 4.551081657409668,
+      "learning_rate": 5.624979922258987e-06,
+      "loss": 0.1092,
+      "step": 14125
+    },
+    {
+      "epoch": 1.342122735464289,
+      "grad_norm": 4.360960006713867,
+      "learning_rate": 5.616948825853707e-06,
+      "loss": 0.105,
+      "step": 14150
+    },
+    {
+      "epoch": 1.3444939770463815,
+      "grad_norm": 8.497123718261719,
+      "learning_rate": 5.608917729448424e-06,
+      "loss": 0.0894,
+      "step": 14175
+    },
+    {
+      "epoch": 1.3468652186284737,
+      "grad_norm": 3.021643877029419,
+      "learning_rate": 5.600886633043143e-06,
+      "loss": 0.1086,
+      "step": 14200
+    },
+    {
+      "epoch": 1.3492364602105662,
+      "grad_norm": 6.0381245613098145,
+      "learning_rate": 5.592855536637862e-06,
+      "loss": 0.1155,
+      "step": 14225
+    },
+    {
+      "epoch": 1.3516077017926587,
+      "grad_norm": 5.288522243499756,
+      "learning_rate": 5.584824440232582e-06,
+      "loss": 0.1044,
+      "step": 14250
+    },
+    {
+      "epoch": 1.3539789433747509,
+      "grad_norm": 6.401548862457275,
+      "learning_rate": 5.5767933438273e-06,
+      "loss": 0.1429,
+      "step": 14275
+    },
+    {
+      "epoch": 1.3563501849568433,
+      "grad_norm": 4.481237888336182,
+      "learning_rate": 5.568762247422018e-06,
+      "loss": 0.1044,
+      "step": 14300
+    },
+    {
+      "epoch": 1.3587214265389358,
+      "grad_norm": 3.4766366481781006,
+      "learning_rate": 5.560731151016737e-06,
+      "loss": 0.1003,
+      "step": 14325
+    },
+    {
+      "epoch": 1.3610926681210282,
+      "grad_norm": 2.493058443069458,
+      "learning_rate": 5.552700054611456e-06,
+      "loss": 0.1219,
+      "step": 14350
+    },
+    {
+      "epoch": 1.3634639097031205,
+      "grad_norm": 4.772429466247559,
+      "learning_rate": 5.544668958206175e-06,
+      "loss": 0.1139,
+      "step": 14375
+    },
+    {
+      "epoch": 1.365835151285213,
+      "grad_norm": 25.902294158935547,
+      "learning_rate": 5.536637861800893e-06,
+      "loss": 0.1226,
+      "step": 14400
+    },
+    {
+      "epoch": 1.3682063928673054,
+      "grad_norm": 2.4243063926696777,
+      "learning_rate": 5.5286067653956124e-06,
+      "loss": 0.1212,
+      "step": 14425
+    },
+    {
+      "epoch": 1.3705776344493978,
+      "grad_norm": 4.881016731262207,
+      "learning_rate": 5.520575668990331e-06,
+      "loss": 0.0936,
+      "step": 14450
+    },
+    {
+      "epoch": 1.37294887603149,
+      "grad_norm": 5.078170299530029,
+      "learning_rate": 5.51254457258505e-06,
+      "loss": 0.1199,
+      "step": 14475
+    },
+    {
+      "epoch": 1.3753201176135825,
+      "grad_norm": 3.8469927310943604,
+      "learning_rate": 5.504513476179768e-06,
+      "loss": 0.1261,
+      "step": 14500
+    },
+    {
+      "epoch": 1.377691359195675,
+      "grad_norm": 4.4348673820495605,
+      "learning_rate": 5.4964823797744875e-06,
+      "loss": 0.1003,
+      "step": 14525
+    },
+    {
+      "epoch": 1.3800626007777672,
+      "grad_norm": 5.1838250160217285,
+      "learning_rate": 5.488451283369206e-06,
+      "loss": 0.11,
+      "step": 14550
+    },
+    {
+      "epoch": 1.3824338423598597,
+      "grad_norm": 6.766570091247559,
+      "learning_rate": 5.4804201869639254e-06,
+      "loss": 0.1118,
+      "step": 14575
+    },
+    {
+      "epoch": 1.3848050839419521,
+      "grad_norm": 4.994007587432861,
+      "learning_rate": 5.472389090558643e-06,
+      "loss": 0.11,
+      "step": 14600
+    },
+    {
+      "epoch": 1.3871763255240444,
+      "grad_norm": 2.069736957550049,
+      "learning_rate": 5.464357994153362e-06,
+      "loss": 0.1004,
+      "step": 14625
+    },
+    {
+      "epoch": 1.3895475671061368,
+      "grad_norm": 6.938327312469482,
+      "learning_rate": 5.456326897748081e-06,
+      "loss": 0.1309,
+      "step": 14650
+    },
+    {
+      "epoch": 1.3919188086882293,
+      "grad_norm": 2.083895206451416,
+      "learning_rate": 5.4482958013428e-06,
+      "loss": 0.1228,
+      "step": 14675
+    },
+    {
+      "epoch": 1.3942900502703215,
+      "grad_norm": 6.256675720214844,
+      "learning_rate": 5.440264704937519e-06,
+      "loss": 0.1089,
+      "step": 14700
+    },
+    {
+      "epoch": 1.396661291852414,
+      "grad_norm": 4.159014701843262,
+      "learning_rate": 5.432233608532237e-06,
+      "loss": 0.1134,
+      "step": 14725
+    },
+    {
+      "epoch": 1.3990325334345064,
+      "grad_norm": 7.001519203186035,
+      "learning_rate": 5.424202512126956e-06,
+      "loss": 0.1255,
+      "step": 14750
+    },
+    {
+      "epoch": 1.4014037750165986,
+      "grad_norm": 4.936473846435547,
+      "learning_rate": 5.416171415721675e-06,
+      "loss": 0.097,
+      "step": 14775
+    },
+    {
+      "epoch": 1.403775016598691,
+      "grad_norm": 8.277131080627441,
+      "learning_rate": 5.408140319316394e-06,
+      "loss": 0.1469,
+      "step": 14800
+    },
+    {
+      "epoch": 1.4061462581807835,
+      "grad_norm": 4.345362663269043,
+      "learning_rate": 5.400109222911112e-06,
+      "loss": 0.0988,
+      "step": 14825
+    },
+    {
+      "epoch": 1.4085174997628758,
+      "grad_norm": 4.8023905754089355,
+      "learning_rate": 5.392078126505831e-06,
+      "loss": 0.1373,
+      "step": 14850
+    },
+    {
+      "epoch": 1.4108887413449682,
+      "grad_norm": 4.75295352935791,
+      "learning_rate": 5.38404703010055e-06,
+      "loss": 0.0965,
+      "step": 14875
+    },
+    {
+      "epoch": 1.4132599829270607,
+      "grad_norm": 3.85670804977417,
+      "learning_rate": 5.376015933695269e-06,
+      "loss": 0.1235,
+      "step": 14900
+    },
+    {
+      "epoch": 1.415631224509153,
+      "grad_norm": 6.758788585662842,
+      "learning_rate": 5.367984837289987e-06,
+      "loss": 0.0878,
+      "step": 14925
+    },
+    {
+      "epoch": 1.4180024660912454,
+      "grad_norm": 3.1618828773498535,
+      "learning_rate": 5.359953740884705e-06,
+      "loss": 0.1131,
+      "step": 14950
+    },
+    {
+      "epoch": 1.4203737076733378,
+      "grad_norm": 6.409102916717529,
+      "learning_rate": 5.351922644479425e-06,
+      "loss": 0.1059,
+      "step": 14975
+    },
+    {
+      "epoch": 1.42274494925543,
+      "grad_norm": 15.774487495422363,
+      "learning_rate": 5.343891548074143e-06,
+      "loss": 0.1179,
+      "step": 15000
+    },
+    {
+      "epoch": 1.42274494925543,
+      "eval_loss": 0.21161529421806335,
+      "eval_runtime": 492.5235,
+      "eval_samples_per_second": 2.57,
+      "eval_steps_per_second": 0.323,
+      "eval_wer": 13.86247959281667,
+      "step": 15000
+    },
+    {
+      "epoch": 1.4251161908375225,
+      "grad_norm": 6.858580112457275,
+      "learning_rate": 5.335860451668863e-06,
+      "loss": 0.1265,
+      "step": 15025
+    },
+    {
+      "epoch": 1.427487432419615,
+      "grad_norm": 5.119607925415039,
+      "learning_rate": 5.3278293552635805e-06,
+      "loss": 0.1202,
+      "step": 15050
+    },
+    {
+      "epoch": 1.4298586740017072,
+      "grad_norm": 3.0737967491149902,
+      "learning_rate": 5.3197982588583e-06,
+      "loss": 0.1152,
+      "step": 15075
+    },
+    {
+      "epoch": 1.4322299155837996,
+      "grad_norm": 5.1812214851379395,
+      "learning_rate": 5.3117671624530184e-06,
+      "loss": 0.1048,
+      "step": 15100
+    },
+    {
+      "epoch": 1.434601157165892,
+      "grad_norm": 5.984945774078369,
+      "learning_rate": 5.303736066047738e-06,
+      "loss": 0.1231,
+      "step": 15125
+    },
+    {
+      "epoch": 1.4369723987479843,
+      "grad_norm": 4.989310264587402,
+      "learning_rate": 5.2957049696424555e-06,
+      "loss": 0.1261,
+      "step": 15150
+    },
+    {
+      "epoch": 1.4393436403300768,
+      "grad_norm": 3.379992961883545,
+      "learning_rate": 5.287673873237175e-06,
+      "loss": 0.1317,
+      "step": 15175
+    },
+    {
+      "epoch": 1.4417148819121692,
+      "grad_norm": 4.174716472625732,
+      "learning_rate": 5.279964020688106e-06,
+      "loss": 0.1199,
+      "step": 15200
+    },
+    {
+      "epoch": 1.4440861234942615,
+      "grad_norm": 5.499051570892334,
+      "learning_rate": 5.271932924282823e-06,
+      "loss": 0.1176,
+      "step": 15225
+    },
+    {
+      "epoch": 1.446457365076354,
+      "grad_norm": 8.469459533691406,
+      "learning_rate": 5.263901827877542e-06,
+      "loss": 0.1168,
+      "step": 15250
+    },
+    {
+      "epoch": 1.4488286066584464,
+      "grad_norm": 5.149110317230225,
+      "learning_rate": 5.255870731472261e-06,
+      "loss": 0.1092,
+      "step": 15275
+    },
+    {
+      "epoch": 1.4511998482405388,
+      "grad_norm": 7.932348728179932,
+      "learning_rate": 5.24783963506698e-06,
+      "loss": 0.1195,
+      "step": 15300
+    },
+    {
+      "epoch": 1.453571089822631,
+      "grad_norm": 7.8235015869140625,
+      "learning_rate": 5.239808538661698e-06,
+      "loss": 0.1402,
+      "step": 15325
+    },
+    {
+      "epoch": 1.4559423314047235,
+      "grad_norm": 6.649631977081299,
+      "learning_rate": 5.231777442256417e-06,
+      "loss": 0.1248,
+      "step": 15350
+    },
+    {
+      "epoch": 1.458313572986816,
+      "grad_norm": 4.304548740386963,
+      "learning_rate": 5.223746345851136e-06,
+      "loss": 0.1053,
+      "step": 15375
+    },
+    {
+      "epoch": 1.4606848145689084,
+      "grad_norm": 4.6666741371154785,
+      "learning_rate": 5.215715249445855e-06,
+      "loss": 0.1138,
+      "step": 15400
+    },
+    {
+      "epoch": 1.4630560561510007,
+      "grad_norm": 26.738964080810547,
+      "learning_rate": 5.2076841530405734e-06,
+      "loss": 0.1227,
+      "step": 15425
+    },
+    {
+      "epoch": 1.465427297733093,
+      "grad_norm": 9.68618392944336,
+      "learning_rate": 5.199653056635292e-06,
+      "loss": 0.1159,
+      "step": 15450
+    },
+    {
+      "epoch": 1.4677985393151856,
+      "grad_norm": 6.386533260345459,
+      "learning_rate": 5.191621960230011e-06,
+      "loss": 0.1349,
+      "step": 15475
+    },
+    {
+      "epoch": 1.4701697808972778,
+      "grad_norm": 10.31882095336914,
+      "learning_rate": 5.18359086382473e-06,
+      "loss": 0.1173,
+      "step": 15500
+    },
+    {
+      "epoch": 1.4725410224793702,
+      "grad_norm": 6.594086647033691,
+      "learning_rate": 5.175559767419448e-06,
+      "loss": 0.1118,
+      "step": 15525
+    },
+    {
+      "epoch": 1.4749122640614627,
+      "grad_norm": 3.3293492794036865,
+      "learning_rate": 5.167528671014167e-06,
+      "loss": 0.119,
+      "step": 15550
+    },
+    {
+      "epoch": 1.477283505643555,
+      "grad_norm": 8.750438690185547,
+      "learning_rate": 5.159497574608886e-06,
+      "loss": 0.1189,
+      "step": 15575
+    },
+    {
+      "epoch": 1.4796547472256474,
+      "grad_norm": 5.673712253570557,
+      "learning_rate": 5.151466478203605e-06,
+      "loss": 0.1132,
+      "step": 15600
+    },
+    {
+      "epoch": 1.4820259888077398,
+      "grad_norm": 3.8876192569732666,
+      "learning_rate": 5.143435381798324e-06,
+      "loss": 0.1224,
+      "step": 15625
+    },
+    {
+      "epoch": 1.484397230389832,
+      "grad_norm": 3.829383134841919,
+      "learning_rate": 5.135404285393042e-06,
+      "loss": 0.1093,
+      "step": 15650
+    },
+    {
+      "epoch": 1.4867684719719245,
+      "grad_norm": 6.885310173034668,
+      "learning_rate": 5.127373188987761e-06,
+      "loss": 0.1222,
+      "step": 15675
+    },
+    {
+      "epoch": 1.489139713554017,
+      "grad_norm": 4.36471700668335,
+      "learning_rate": 5.11934209258248e-06,
+      "loss": 0.1113,
+      "step": 15700
+    },
+    {
+      "epoch": 1.4915109551361092,
+      "grad_norm": 9.012696266174316,
+      "learning_rate": 5.111310996177199e-06,
+      "loss": 0.1142,
+      "step": 15725
+    },
+    {
+      "epoch": 1.4938821967182017,
+      "grad_norm": 5.295135974884033,
+      "learning_rate": 5.103279899771917e-06,
+      "loss": 0.1021,
+      "step": 15750
+    },
+    {
+      "epoch": 1.4962534383002941,
+      "grad_norm": 5.390635013580322,
+      "learning_rate": 5.095248803366636e-06,
+      "loss": 0.1075,
+      "step": 15775
+    },
+    {
+      "epoch": 1.4986246798823863,
+      "grad_norm": 4.967793941497803,
+      "learning_rate": 5.087217706961355e-06,
+      "loss": 0.1503,
+      "step": 15800
+    },
+    {
+      "epoch": 1.5009959214644788,
+      "grad_norm": 3.1676905155181885,
+      "learning_rate": 5.079186610556074e-06,
+      "loss": 0.114,
+      "step": 15825
+    },
+    {
+      "epoch": 1.5033671630465713,
+      "grad_norm": 4.194946765899658,
+      "learning_rate": 5.071155514150791e-06,
+      "loss": 0.1158,
+      "step": 15850
+    },
+    {
+      "epoch": 1.5057384046286635,
+      "grad_norm": 5.673632621765137,
+      "learning_rate": 5.063124417745511e-06,
+      "loss": 0.1304,
+      "step": 15875
+    },
+    {
+      "epoch": 1.508109646210756,
+      "grad_norm": 4.258605480194092,
+      "learning_rate": 5.05509332134023e-06,
+      "loss": 0.0995,
+      "step": 15900
+    },
+    {
+      "epoch": 1.5104808877928484,
+      "grad_norm": 9.870444297790527,
+      "learning_rate": 5.047062224934949e-06,
+      "loss": 0.0934,
+      "step": 15925
+    },
+    {
+      "epoch": 1.5128521293749406,
+      "grad_norm": 4.676215648651123,
+      "learning_rate": 5.039031128529668e-06,
+      "loss": 0.142,
+      "step": 15950
+    },
+    {
+      "epoch": 1.515223370957033,
+      "grad_norm": 5.100001335144043,
+      "learning_rate": 5.031000032124386e-06,
+      "loss": 0.1196,
+      "step": 15975
+    },
+    {
+      "epoch": 1.5175946125391255,
+      "grad_norm": 4.015578269958496,
+      "learning_rate": 5.022968935719104e-06,
+      "loss": 0.1024,
+      "step": 16000
+    },
+    {
+      "epoch": 1.5175946125391255,
+      "eval_loss": 0.20574869215488434,
+      "eval_runtime": 496.3633,
+      "eval_samples_per_second": 2.551,
+      "eval_steps_per_second": 0.32,
+      "eval_wer": 14.525112839719581,
+      "step": 16000
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 31629,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.693719972855808e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}