diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7466 @@
+{
+  "best_metric": 0.7750564813613892,
+  "best_model_checkpoint": "temp/checkpoint-30660",
+  "epoch": 10.0,
+  "eval_steps": 500,
+  "global_step": 30660,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042,
+      "loss": 4.34,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00092,
+      "loss": 2.3611,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000999313949689644,
+      "loss": 1.8959,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009984972231296962,
+      "loss": 1.7016,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009976804965697486,
+      "loss": 1.4917,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009968637700098007,
+      "loss": 1.3633,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996047043449853,
+      "loss": 1.3293,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952303168899053,
+      "loss": 1.3843,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009944135903299576,
+      "loss": 1.4711,
+      "step": 225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935968637700097,
+      "loss": 1.4568,
+      "step": 250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000992780137210062,
+      "loss": 1.2875,
+      "step": 275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009919634106501144,
+      "loss": 1.3972,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009911466840901667,
+      "loss": 1.2847,
+      "step": 325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009903299575302188,
+      "loss": 1.43,
+      "step": 350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009895132309702711,
+      "loss": 1.306,
+      "step": 375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009886965044103235,
+      "loss": 1.3228,
+      "step": 400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009878797778503758,
+      "loss": 1.3421,
+      "step": 425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000987063051290428,
+      "loss": 1.3667,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009862463247304802,
+      "loss": 1.2756,
+      "step": 475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009854295981705325,
+      "loss": 1.2254,
+      "step": 500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009846128716105849,
+      "loss": 1.2671,
+      "step": 525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009837961450506372,
+      "loss": 1.3432,
+      "step": 550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009829794184906893,
+      "loss": 1.2765,
+      "step": 575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009821626919307416,
+      "loss": 1.1244,
+      "step": 600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000981345965370794,
+      "loss": 1.3241,
+      "step": 625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000980529238810846,
+      "loss": 1.3232,
+      "step": 650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009797125122508984,
+      "loss": 1.3353,
+      "step": 675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009788957856909507,
+      "loss": 1.2559,
+      "step": 700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000978079059131003,
+      "loss": 1.2461,
+      "step": 725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009772623325710553,
+      "loss": 1.2864,
+      "step": 750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009764456060111076,
+      "loss": 1.2687,
+      "step": 775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009756288794511598,
+      "loss": 1.2791,
+      "step": 800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009748121528912121,
+      "loss": 1.2676,
+      "step": 825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009739954263312644,
+      "loss": 1.4354,
+      "step": 850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009731786997713165,
+      "loss": 1.2231,
+      "step": 875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009723619732113688,
+      "loss": 1.2008,
+      "step": 900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009715452466514211,
+      "loss": 1.2522,
+      "step": 925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009707285200914734,
+      "loss": 1.2547,
+      "step": 950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009699117935315257,
+      "loss": 1.2101,
+      "step": 975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009690950669715779,
+      "loss": 1.2404,
+      "step": 1000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009682783404116302,
+      "loss": 1.3073,
+      "step": 1025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009674616138516826,
+      "loss": 1.1727,
+      "step": 1050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009666448872917347,
+      "loss": 1.5219,
+      "step": 1075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000965828160731787,
+      "loss": 1.3067,
+      "step": 1100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009650114341718393,
+      "loss": 1.2617,
+      "step": 1125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009641947076118915,
+      "loss": 1.1939,
+      "step": 1150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009633779810519439,
+      "loss": 1.2605,
+      "step": 1175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009625612544919961,
+      "loss": 1.2543,
+      "step": 1200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009617445279320484,
+      "loss": 1.255,
+      "step": 1225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009609278013721007,
+      "loss": 1.2582,
+      "step": 1250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009601110748121528,
+      "loss": 1.2397,
+      "step": 1275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009592943482522051,
+      "loss": 1.2944,
+      "step": 1300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009584776216922575,
+      "loss": 1.1534,
+      "step": 1325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009576608951323097,
+      "loss": 1.2319,
+      "step": 1350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000956844168572362,
+      "loss": 1.2521,
+      "step": 1375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009560274420124143,
+      "loss": 1.224,
+      "step": 1400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009552107154524665,
+      "loss": 1.2421,
+      "step": 1425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009543939888925189,
+      "loss": 1.1581,
+      "step": 1450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009535772623325711,
+      "loss": 1.2528,
+      "step": 1475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009527605357726233,
+      "loss": 1.141,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009519438092126756,
+      "loss": 1.1966,
+      "step": 1525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009511270826527278,
+      "loss": 1.1853,
+      "step": 1550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009503103560927802,
+      "loss": 1.1506,
+      "step": 1575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009494936295328325,
+      "loss": 1.2139,
+      "step": 1600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009486769029728847,
+      "loss": 1.2569,
+      "step": 1625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000947860176412937,
+      "loss": 1.2148,
+      "step": 1650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009470434498529893,
+      "loss": 1.1866,
+      "step": 1675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009462267232930414,
+      "loss": 1.228,
+      "step": 1700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0009454099967330938,
+      "loss": 1.2117,
+      "step": 1725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000944593270173146,
+      "loss": 1.2629,
+      "step": 1750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0009437765436131983,
+      "loss": 1.2341,
+      "step": 1775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0009429598170532506,
+      "loss": 1.1796,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0009421430904933028,
+      "loss": 1.2528,
+      "step": 1825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0009413263639333552,
+      "loss": 1.1945,
+      "step": 1850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0009405096373734075,
+      "loss": 1.2459,
+      "step": 1875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0009396929108134596,
+      "loss": 1.104,
+      "step": 1900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0009388761842535119,
+      "loss": 1.1864,
+      "step": 1925
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0009380594576935642,
+      "loss": 1.129,
+      "step": 1950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0009372427311336165,
+      "loss": 1.2213,
+      "step": 1975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0009364260045736688,
+      "loss": 1.1773,
+      "step": 2000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0009356419470761189,
+      "loss": 1.2489,
+      "step": 2025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0009348252205161712,
+      "loss": 1.1637,
+      "step": 2050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0009340084939562234,
+      "loss": 1.2635,
+      "step": 2075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0009331917673962757,
+      "loss": 1.2673,
+      "step": 2100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0009323750408363281,
+      "loss": 1.2722,
+      "step": 2125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0009315583142763803,
+      "loss": 1.1882,
+      "step": 2150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0009307415877164326,
+      "loss": 1.1911,
+      "step": 2175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0009299248611564848,
+      "loss": 1.2546,
+      "step": 2200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0009291081345965371,
+      "loss": 1.1473,
+      "step": 2225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0009282914080365894,
+      "loss": 1.2054,
+      "step": 2250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0009274746814766416,
+      "loss": 1.1305,
+      "step": 2275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0009266579549166939,
+      "loss": 1.2261,
+      "step": 2300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0009258412283567462,
+      "loss": 1.1463,
+      "step": 2325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0009250245017967984,
+      "loss": 1.2333,
+      "step": 2350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0009242077752368508,
+      "loss": 1.1699,
+      "step": 2375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0009233910486769031,
+      "loss": 1.2033,
+      "step": 2400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0009225743221169553,
+      "loss": 1.1624,
+      "step": 2425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0009217575955570075,
+      "loss": 1.1374,
+      "step": 2450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0009209408689970597,
+      "loss": 1.214,
+      "step": 2475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000920124142437112,
+      "loss": 1.2615,
+      "step": 2500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0009193074158771644,
+      "loss": 1.2242,
+      "step": 2525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0009184906893172166,
+      "loss": 1.2355,
+      "step": 2550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0009176739627572689,
+      "loss": 1.2533,
+      "step": 2575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0009168572361973212,
+      "loss": 1.1497,
+      "step": 2600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0009160405096373734,
+      "loss": 1.1764,
+      "step": 2625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0009152237830774257,
+      "loss": 1.2131,
+      "step": 2650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000914407056517478,
+      "loss": 1.2165,
+      "step": 2675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0009135903299575302,
+      "loss": 1.2087,
+      "step": 2700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0009127736033975825,
+      "loss": 1.2069,
+      "step": 2725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0009119568768376347,
+      "loss": 1.1186,
+      "step": 2750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0009111401502776871,
+      "loss": 1.1936,
+      "step": 2775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0009103234237177394,
+      "loss": 1.1926,
+      "step": 2800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0009095066971577916,
+      "loss": 1.175,
+      "step": 2825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0009086899705978438,
+      "loss": 1.2437,
+      "step": 2850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0009078732440378961,
+      "loss": 1.156,
+      "step": 2875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0009070565174779483,
+      "loss": 1.2573,
+      "step": 2900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0009062397909180007,
+      "loss": 1.1603,
+      "step": 2925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000905423064358053,
+      "loss": 1.2515,
+      "step": 2950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0009046063377981052,
+      "loss": 1.173,
+      "step": 2975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0009037896112381575,
+      "loss": 1.1754,
+      "step": 3000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0009029728846782097,
+      "loss": 1.1455,
+      "step": 3025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0009021561581182621,
+      "loss": 1.1903,
+      "step": 3050
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.0751750469207764,
+      "eval_runtime": 957.3865,
+      "eval_samples_per_second": 1.045,
+      "eval_steps_per_second": 0.131,
+      "eval_wer": 14.025925580527836,
+      "step": 3066
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0009013394315583143,
+      "loss": 1.0824,
+      "step": 3075
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0009005227049983665,
+      "loss": 1.0884,
+      "step": 3100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0008997059784384188,
+      "loss": 1.0469,
+      "step": 3125
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0008988892518784711,
+      "loss": 1.1158,
+      "step": 3150
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0008980725253185234,
+      "loss": 1.1601,
+      "step": 3175
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0008972557987585757,
+      "loss": 1.1254,
+      "step": 3200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000896439072198628,
+      "loss": 1.083,
+      "step": 3225
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0008956223456386802,
+      "loss": 1.437,
+      "step": 3250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0008948056190787324,
+      "loss": 1.1374,
+      "step": 3275
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0008939888925187846,
+      "loss": 1.1438,
+      "step": 3300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000893172165958837,
+      "loss": 1.1136,
+      "step": 3325
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0008923554393988893,
+      "loss": 1.0219,
+      "step": 3350
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0008915387128389415,
+      "loss": 1.1524,
+      "step": 3375
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0008907219862789938,
+      "loss": 1.1272,
+      "step": 3400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0008899052597190462,
+      "loss": 1.1296,
+      "step": 3425
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0008890885331590984,
+      "loss": 1.168,
+      "step": 3450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0008882718065991506,
+      "loss": 1.167,
+      "step": 3475
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0008874550800392029,
+      "loss": 1.127,
+      "step": 3500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0008866383534792551,
+      "loss": 1.1062,
+      "step": 3525
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0008858216269193074,
+      "loss": 1.1726,
+      "step": 3550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0008850049003593597,
+      "loss": 1.1571,
+      "step": 3575
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000884188173799412,
+      "loss": 1.0593,
+      "step": 3600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0008833714472394643,
+      "loss": 1.1024,
+      "step": 3625
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0008825547206795165,
+      "loss": 1.1431,
+      "step": 3650
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0008817379941195687,
+      "loss": 1.1414,
+      "step": 3675
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0008809212675596211,
+      "loss": 1.1188,
+      "step": 3700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0008801045409996733,
+      "loss": 1.1289,
+      "step": 3725
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0008792878144397256,
+      "loss": 1.12,
+      "step": 3750
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0008784710878797779,
+      "loss": 1.0852,
+      "step": 3775
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0008776543613198301,
+      "loss": 1.123,
+      "step": 3800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0008768376347598825,
+      "loss": 1.2024,
+      "step": 3825
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0008760209081999347,
+      "loss": 1.1671,
+      "step": 3850
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000875204181639987,
+      "loss": 1.1649,
+      "step": 3875
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0008743874550800392,
+      "loss": 1.1101,
+      "step": 3900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0008735707285200914,
+      "loss": 1.1336,
+      "step": 3925
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0008727540019601437,
+      "loss": 1.1928,
+      "step": 3950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0008719372754001961,
+      "loss": 1.1132,
+      "step": 3975
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0008711205488402483,
+      "loss": 1.1441,
+      "step": 4000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0008703038222803006,
+      "loss": 1.0876,
+      "step": 4025
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0008694870957203529,
+      "loss": 1.1469,
+      "step": 4050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0008686703691604051,
+      "loss": 1.0801,
+      "step": 4075
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0008678536426004574,
+      "loss": 1.2046,
+      "step": 4100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0008670369160405096,
+      "loss": 1.16,
+      "step": 4125
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0008662201894805619,
+      "loss": 1.0832,
+      "step": 4150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0008654034629206142,
+      "loss": 1.0785,
+      "step": 4175
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0008645867363606664,
+      "loss": 1.1291,
+      "step": 4200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0008637700098007188,
+      "loss": 1.1209,
+      "step": 4225
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0008629532832407711,
+      "loss": 1.1116,
+      "step": 4250
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0008621365566808233,
+      "loss": 1.1478,
+      "step": 4275
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0008613198301208755,
+      "loss": 1.1547,
+      "step": 4300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0008605031035609278,
+      "loss": 1.1724,
+      "step": 4325
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00085968637700098,
+      "loss": 1.2658,
+      "step": 4350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0008588696504410324,
+      "loss": 1.0794,
+      "step": 4375
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0008580529238810846,
+      "loss": 1.1891,
+      "step": 4400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0008572361973211369,
+      "loss": 1.207,
+      "step": 4425
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0008564194707611892,
+      "loss": 1.0941,
+      "step": 4450
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0008556027442012414,
+      "loss": 1.155,
+      "step": 4475
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0008547860176412938,
+      "loss": 1.1785,
+      "step": 4500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000853969291081346,
+      "loss": 1.0882,
+      "step": 4525
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0008531525645213982,
+      "loss": 1.1544,
+      "step": 4550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0008523358379614505,
+      "loss": 1.0764,
+      "step": 4575
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0008515191114015028,
+      "loss": 1.099,
+      "step": 4600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0008507023848415551,
+      "loss": 1.0939,
+      "step": 4625
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0008498856582816074,
+      "loss": 1.057,
+      "step": 4650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0008490689317216596,
+      "loss": 1.2006,
+      "step": 4675
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0008482522051617119,
+      "loss": 1.1362,
+      "step": 4700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0008474354786017641,
+      "loss": 1.1804,
+      "step": 4725
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0008466187520418163,
+      "loss": 1.2434,
+      "step": 4750
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0008458020254818687,
+      "loss": 1.2353,
+      "step": 4775
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000844985298921921,
+      "loss": 1.1643,
+      "step": 4800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0008441685723619732,
+      "loss": 1.114,
+      "step": 4825
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0008433518458020255,
+      "loss": 1.1803,
+      "step": 4850
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0008425351192420779,
+      "loss": 1.0897,
+      "step": 4875
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0008417183926821301,
+      "loss": 1.0807,
+      "step": 4900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0008409016661221823,
+      "loss": 1.1861,
+      "step": 4925
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0008400849395622345,
+      "loss": 1.1627,
+      "step": 4950
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0008392682130022868,
+      "loss": 1.1307,
+      "step": 4975
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0008384514864423391,
+      "loss": 1.1836,
+      "step": 5000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0008376347598823914,
+      "loss": 1.1408,
+      "step": 5025
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0008368180333224437,
+      "loss": 1.1182,
+      "step": 5050
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000836001306762496,
+      "loss": 1.0922,
+      "step": 5075
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0008351845802025482,
+      "loss": 1.1128,
+      "step": 5100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0008343678536426004,
+      "loss": 1.0759,
+      "step": 5125
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0008335511270826528,
+      "loss": 1.1109,
+      "step": 5150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.000832734400522705,
+      "loss": 1.0652,
+      "step": 5175
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0008319176739627573,
+      "loss": 1.1616,
+      "step": 5200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0008311009474028095,
+      "loss": 1.125,
+      "step": 5225
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0008302842208428618,
+      "loss": 1.1628,
+      "step": 5250
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0008294674942829142,
+      "loss": 1.1775,
+      "step": 5275
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0008286507677229664,
+      "loss": 1.1475,
+      "step": 5300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0008278340411630187,
+      "loss": 1.1174,
+      "step": 5325
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0008270173146030709,
+      "loss": 1.1714,
+      "step": 5350
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0008262005880431231,
+      "loss": 1.0809,
+      "step": 5375
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0008253838614831754,
+      "loss": 1.1005,
+      "step": 5400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0008245671349232278,
+      "loss": 1.1567,
+      "step": 5425
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00082375040836328,
+      "loss": 1.1455,
+      "step": 5450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0008229336818033323,
+      "loss": 1.0424,
+      "step": 5475
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0008221169552433845,
+      "loss": 1.1255,
+      "step": 5500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0008213002286834368,
+      "loss": 1.0574,
+      "step": 5525
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000820483502123489,
+      "loss": 1.0636,
+      "step": 5550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0008196667755635413,
+      "loss": 1.1092,
+      "step": 5575
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0008188500490035936,
+      "loss": 1.104,
+      "step": 5600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0008180333224436459,
+      "loss": 1.1652,
+      "step": 5625
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0008172165958836981,
+      "loss": 1.084,
+      "step": 5650
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0008163998693237505,
+      "loss": 1.0867,
+      "step": 5675
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0008155831427638028,
+      "loss": 1.0894,
+      "step": 5700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000814766416203855,
+      "loss": 1.1023,
+      "step": 5725
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0008139496896439072,
+      "loss": 1.1705,
+      "step": 5750
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0008131329630839594,
+      "loss": 1.1505,
+      "step": 5775
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0008123162365240117,
+      "loss": 1.0699,
+      "step": 5800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0008114995099640641,
+      "loss": 1.1623,
+      "step": 5825
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0008106827834041163,
+      "loss": 1.104,
+      "step": 5850
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0008098660568441686,
+      "loss": 1.087,
+      "step": 5875
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0008090493302842209,
+      "loss": 1.1056,
+      "step": 5900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0008082326037242731,
+      "loss": 1.094,
+      "step": 5925
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0008074158771643254,
+      "loss": 1.0792,
+      "step": 5950
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0008065991506043777,
+      "loss": 1.1011,
+      "step": 5975
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0008057824240444299,
+      "loss": 1.077,
+      "step": 6000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0008049656974844822,
+      "loss": 1.1078,
+      "step": 6025
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0008041489709245344,
+      "loss": 1.0974,
+      "step": 6050
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0008033322443645868,
+      "loss": 1.1687,
+      "step": 6075
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0008025155178046391,
+      "loss": 1.1413,
+      "step": 6100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0008016987912446913,
+      "loss": 1.1263,
+      "step": 6125
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.996329128742218,
+      "eval_runtime": 962.9173,
+      "eval_samples_per_second": 1.039,
+      "eval_steps_per_second": 0.13,
+      "eval_wer": 11.694488482700736,
+      "step": 6132
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0008008820646847436,
+      "loss": 1.0098,
+      "step": 6150
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0008000653381247958,
+      "loss": 1.065,
+      "step": 6175
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.000799248611564848,
+      "loss": 0.9865,
+      "step": 6200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0007984318850049004,
+      "loss": 1.1061,
+      "step": 6225
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0007976151584449527,
+      "loss": 1.148,
+      "step": 6250
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0007967984318850049,
+      "loss": 1.003,
+      "step": 6275
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0007959817053250572,
+      "loss": 1.0571,
+      "step": 6300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0007951649787651094,
+      "loss": 1.036,
+      "step": 6325
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0007943482522051618,
+      "loss": 1.0794,
+      "step": 6350
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000793531525645214,
+      "loss": 1.0093,
+      "step": 6375
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0007927147990852662,
+      "loss": 1.0416,
+      "step": 6400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0007918980725253185,
+      "loss": 1.0552,
+      "step": 6425
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0007910813459653708,
+      "loss": 1.061,
+      "step": 6450
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.000790264619405423,
+      "loss": 1.0884,
+      "step": 6475
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0007894478928454754,
+      "loss": 1.042,
+      "step": 6500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0007886311662855277,
+      "loss": 1.0047,
+      "step": 6525
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0007878144397255799,
+      "loss": 1.0589,
+      "step": 6550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0007869977131656321,
+      "loss": 1.108,
+      "step": 6575
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0007861809866056843,
+      "loss": 1.0201,
+      "step": 6600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0007853642600457367,
+      "loss": 1.0152,
+      "step": 6625
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.000784547533485789,
+      "loss": 1.049,
+      "step": 6650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0007837308069258412,
+      "loss": 1.1016,
+      "step": 6675
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0007829140803658935,
+      "loss": 1.057,
+      "step": 6700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0007820973538059458,
+      "loss": 0.9956,
+      "step": 6725
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0007812806272459981,
+      "loss": 1.0933,
+      "step": 6750
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0007804639006860504,
+      "loss": 1.1044,
+      "step": 6775
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0007796471741261026,
+      "loss": 1.064,
+      "step": 6800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0007788304475661548,
+      "loss": 1.0255,
+      "step": 6825
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0007780137210062071,
+      "loss": 1.0522,
+      "step": 6850
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0007771969944462594,
+      "loss": 1.0389,
+      "step": 6875
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0007763802678863117,
+      "loss": 1.0294,
+      "step": 6900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000775563541326364,
+      "loss": 1.0551,
+      "step": 6925
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0007747468147664162,
+      "loss": 1.0042,
+      "step": 6950
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0007739300882064685,
+      "loss": 1.0697,
+      "step": 6975
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0007731133616465207,
+      "loss": 1.1156,
+      "step": 7000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.000772296635086573,
+      "loss": 1.0384,
+      "step": 7025
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0007714799085266253,
+      "loss": 1.0968,
+      "step": 7050
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0007706631819666776,
+      "loss": 1.0474,
+      "step": 7075
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0007698464554067298,
+      "loss": 1.0105,
+      "step": 7100
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0007690297288467821,
+      "loss": 1.1244,
+      "step": 7125
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0007682130022868345,
+      "loss": 1.0803,
+      "step": 7150
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0007673962757268867,
+      "loss": 1.0919,
+      "step": 7175
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0007665795491669389,
+      "loss": 1.1078,
+      "step": 7200
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0007657628226069911,
+      "loss": 1.0405,
+      "step": 7225
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0007649460960470434,
+      "loss": 1.1196,
+      "step": 7250
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0007641293694870958,
+      "loss": 1.1042,
+      "step": 7275
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.000763312642927148,
+      "loss": 1.0317,
+      "step": 7300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0007624959163672003,
+      "loss": 1.0347,
+      "step": 7325
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0007616791898072526,
+      "loss": 1.0893,
+      "step": 7350
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0007608624632473048,
+      "loss": 1.1129,
+      "step": 7375
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.000760045736687357,
+      "loss": 1.0961,
+      "step": 7400
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0007592290101274094,
+      "loss": 1.0584,
+      "step": 7425
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0007584122835674616,
+      "loss": 1.0908,
+      "step": 7450
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0007575955570075139,
+      "loss": 1.1904,
+      "step": 7475
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0007567788304475661,
+      "loss": 1.0264,
+      "step": 7500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0007559621038876184,
+      "loss": 1.1083,
+      "step": 7525
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0007551453773276708,
+      "loss": 1.0739,
+      "step": 7550
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.000754328650767723,
+      "loss": 1.0946,
+      "step": 7575
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0007535119242077753,
+      "loss": 1.0595,
+      "step": 7600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0007526951976478275,
+      "loss": 0.985,
+      "step": 7625
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0007518784710878797,
+      "loss": 1.0613,
+      "step": 7650
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0007510617445279321,
+      "loss": 1.0642,
+      "step": 7675
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0007502450179679844,
+      "loss": 1.119,
+      "step": 7700
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0007494282914080366,
+      "loss": 1.0649,
+      "step": 7725
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0007486115648480889,
+      "loss": 1.0372,
+      "step": 7750
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0007477948382881411,
+      "loss": 1.1075,
+      "step": 7775
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0007469781117281935,
+      "loss": 1.1406,
+      "step": 7800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0007461613851682457,
+      "loss": 1.0612,
+      "step": 7825
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0007453446586082979,
+      "loss": 1.1374,
+      "step": 7850
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0007445279320483502,
+      "loss": 1.1272,
+      "step": 7875
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0007437112054884025,
+      "loss": 1.0771,
+      "step": 7900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0007428944789284547,
+      "loss": 1.0746,
+      "step": 7925
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0007420777523685071,
+      "loss": 1.0436,
+      "step": 7950
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0007412610258085594,
+      "loss": 1.0137,
+      "step": 7975
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0007404442992486116,
+      "loss": 1.039,
+      "step": 8000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0007396275726886638,
+      "loss": 1.0621,
+      "step": 8025
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000738810846128716,
+      "loss": 1.0593,
+      "step": 8050
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0007379941195687684,
+      "loss": 1.0682,
+      "step": 8075
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0007371773930088207,
+      "loss": 1.1118,
+      "step": 8100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0007363606664488729,
+      "loss": 1.0247,
+      "step": 8125
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0007355439398889252,
+      "loss": 1.0238,
+      "step": 8150
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0007347272133289775,
+      "loss": 0.9952,
+      "step": 8175
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0007339104867690298,
+      "loss": 1.1454,
+      "step": 8200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0007330937602090821,
+      "loss": 1.1397,
+      "step": 8225
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0007322770336491343,
+      "loss": 1.047,
+      "step": 8250
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0007314603070891865,
+      "loss": 1.0736,
+      "step": 8275
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0007306435805292388,
+      "loss": 1.0854,
+      "step": 8300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.000729826853969291,
+      "loss": 1.0001,
+      "step": 8325
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0007290101274093434,
+      "loss": 1.1361,
+      "step": 8350
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0007281934008493957,
+      "loss": 1.1009,
+      "step": 8375
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0007273766742894479,
+      "loss": 1.0611,
+      "step": 8400
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0007265599477295002,
+      "loss": 1.0593,
+      "step": 8425
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0007257432211695524,
+      "loss": 1.0397,
+      "step": 8450
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0007249264946096047,
+      "loss": 1.0726,
+      "step": 8475
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.000724109768049657,
+      "loss": 1.0879,
+      "step": 8500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0007232930414897093,
+      "loss": 1.0216,
+      "step": 8525
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0007224763149297615,
+      "loss": 0.965,
+      "step": 8550
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0007216595883698138,
+      "loss": 1.0499,
+      "step": 8575
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0007208428618098661,
+      "loss": 1.0893,
+      "step": 8600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0007200261352499184,
+      "loss": 1.0375,
+      "step": 8625
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0007192094086899706,
+      "loss": 0.9769,
+      "step": 8650
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0007183926821300228,
+      "loss": 1.0305,
+      "step": 8675
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0007175759555700751,
+      "loss": 1.0713,
+      "step": 8700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0007167592290101275,
+      "loss": 1.0575,
+      "step": 8725
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0007159425024501797,
+      "loss": 1.0231,
+      "step": 8750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.000715125775890232,
+      "loss": 1.0553,
+      "step": 8775
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0007143090493302843,
+      "loss": 1.0258,
+      "step": 8800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0007134923227703365,
+      "loss": 1.0678,
+      "step": 8825
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007127082652727866,
+      "loss": 1.1467,
+      "step": 8850
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000711891538712839,
+      "loss": 1.074,
+      "step": 8875
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007110748121528913,
+      "loss": 1.1102,
+      "step": 8900
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007102580855929435,
+      "loss": 1.0164,
+      "step": 8925
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007094413590329958,
+      "loss": 1.0392,
+      "step": 8950
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000708624632473048,
+      "loss": 1.1001,
+      "step": 8975
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007078079059131003,
+      "loss": 1.1367,
+      "step": 9000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007069911793531526,
+      "loss": 1.0218,
+      "step": 9025
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007061744527932048,
+      "loss": 1.0521,
+      "step": 9050
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007053577262332571,
+      "loss": 1.0717,
+      "step": 9075
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007045409996733094,
+      "loss": 1.0535,
+      "step": 9100
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007037242731133617,
+      "loss": 1.0772,
+      "step": 9125
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.000702907546553414,
+      "loss": 1.0168,
+      "step": 9150
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007020908199934663,
+      "loss": 1.0555,
+      "step": 9175
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.9617839455604553,
+      "eval_runtime": 961.3646,
+      "eval_samples_per_second": 1.04,
+      "eval_steps_per_second": 0.13,
+      "eval_wer": 8.094749603655693,
+      "step": 9198
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007012740934335184,
+      "loss": 1.1186,
+      "step": 9200
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007004573668735707,
+      "loss": 1.0314,
+      "step": 9225
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000699640640313623,
+      "loss": 0.9924,
+      "step": 9250
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0006988239137536753,
+      "loss": 0.9701,
+      "step": 9275
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0006980071871937276,
+      "loss": 0.9698,
+      "step": 9300
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0006971904606337798,
+      "loss": 0.962,
+      "step": 9325
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006963737340738321,
+      "loss": 0.9654,
+      "step": 9350
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006955570075138844,
+      "loss": 0.964,
+      "step": 9375
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006947402809539366,
+      "loss": 0.986,
+      "step": 9400
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006939235543939889,
+      "loss": 0.9084,
+      "step": 9425
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006931068278340412,
+      "loss": 0.9674,
+      "step": 9450
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006922901012740934,
+      "loss": 0.9632,
+      "step": 9475
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006914733747141457,
+      "loss": 1.002,
+      "step": 9500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006906566481541981,
+      "loss": 0.9838,
+      "step": 9525
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006898399215942503,
+      "loss": 0.9844,
+      "step": 9550
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006890231950343026,
+      "loss": 0.9847,
+      "step": 9575
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006882064684743547,
+      "loss": 1.0419,
+      "step": 9600
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.000687389741914407,
+      "loss": 0.989,
+      "step": 9625
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006865730153544593,
+      "loss": 0.9602,
+      "step": 9650
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006857562887945116,
+      "loss": 0.9614,
+      "step": 9675
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006849395622345639,
+      "loss": 0.9817,
+      "step": 9700
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006841228356746162,
+      "loss": 1.0453,
+      "step": 9725
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006833061091146684,
+      "loss": 0.9884,
+      "step": 9750
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006824893825547207,
+      "loss": 0.9821,
+      "step": 9775
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.000681672655994773,
+      "loss": 0.9874,
+      "step": 9800
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006808559294348252,
+      "loss": 0.99,
+      "step": 9825
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006800392028748775,
+      "loss": 1.0586,
+      "step": 9850
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0006792224763149297,
+      "loss": 1.0218,
+      "step": 9875
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.000678405749754982,
+      "loss": 0.9834,
+      "step": 9900
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0006775890231950344,
+      "loss": 1.0296,
+      "step": 9925
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0006767722966350866,
+      "loss": 1.0014,
+      "step": 9950
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0006759555700751389,
+      "loss": 0.9946,
+      "step": 9975
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0006751388435151912,
+      "loss": 1.0409,
+      "step": 10000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0006743221169552433,
+      "loss": 1.0256,
+      "step": 10025
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0006735053903952957,
+      "loss": 1.0599,
+      "step": 10050
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.000672688663835348,
+      "loss": 1.092,
+      "step": 10075
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0006718719372754002,
+      "loss": 1.0112,
+      "step": 10100
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0006710552107154525,
+      "loss": 0.99,
+      "step": 10125
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0006702384841555047,
+      "loss": 0.9849,
+      "step": 10150
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.000669421757595557,
+      "loss": 0.9633,
+      "step": 10175
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0006686050310356094,
+      "loss": 1.0148,
+      "step": 10200
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0006677883044756615,
+      "loss": 1.0537,
+      "step": 10225
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0006669715779157138,
+      "loss": 1.0297,
+      "step": 10250
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0006661548513557661,
+      "loss": 0.9652,
+      "step": 10275
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0006653381247958183,
+      "loss": 1.0758,
+      "step": 10300
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0006645213982358707,
+      "loss": 0.9836,
+      "step": 10325
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.000663704671675923,
+      "loss": 1.0076,
+      "step": 10350
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0006628879451159752,
+      "loss": 1.0426,
+      "step": 10375
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0006620712185560275,
+      "loss": 1.0186,
+      "step": 10400
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0006612544919960796,
+      "loss": 0.9763,
+      "step": 10425
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.000660437765436132,
+      "loss": 0.9896,
+      "step": 10450
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0006596210388761843,
+      "loss": 1.0314,
+      "step": 10475
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0006588043123162365,
+      "loss": 1.0192,
+      "step": 10500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0006579875857562888,
+      "loss": 1.0586,
+      "step": 10525
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0006571708591963411,
+      "loss": 1.062,
+      "step": 10550
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0006563541326363933,
+      "loss": 1.0372,
+      "step": 10575
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0006555374060764457,
+      "loss": 0.9983,
+      "step": 10600
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.000654720679516498,
+      "loss": 0.9564,
+      "step": 10625
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0006539039529565501,
+      "loss": 1.0326,
+      "step": 10650
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0006530872263966024,
+      "loss": 1.0218,
+      "step": 10675
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0006522704998366546,
+      "loss": 0.9739,
+      "step": 10700
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.000651453773276707,
+      "loss": 0.9456,
+      "step": 10725
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0006506370467167593,
+      "loss": 0.9507,
+      "step": 10750
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0006498203201568115,
+      "loss": 1.0446,
+      "step": 10775
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0006490035935968638,
+      "loss": 1.0757,
+      "step": 10800
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0006481868670369161,
+      "loss": 0.9148,
+      "step": 10825
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0006473701404769683,
+      "loss": 1.0086,
+      "step": 10850
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0006465534139170206,
+      "loss": 1.0549,
+      "step": 10875
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0006457366873570729,
+      "loss": 0.9891,
+      "step": 10900
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0006449199607971251,
+      "loss": 1.0153,
+      "step": 10925
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0006441032342371774,
+      "loss": 1.0154,
+      "step": 10950
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0006432865076772296,
+      "loss": 1.0767,
+      "step": 10975
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.000642469781117282,
+      "loss": 1.0556,
+      "step": 11000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0006416530545573343,
+      "loss": 1.0148,
+      "step": 11025
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0006408363279973864,
+      "loss": 1.043,
+      "step": 11050
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0006400196014374387,
+      "loss": 1.0198,
+      "step": 11075
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.000639202874877491,
+      "loss": 0.9839,
+      "step": 11100
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0006383861483175433,
+      "loss": 1.0209,
+      "step": 11125
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0006375694217575956,
+      "loss": 1.0115,
+      "step": 11150
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0006367526951976479,
+      "loss": 1.0739,
+      "step": 11175
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0006359359686377001,
+      "loss": 1.0323,
+      "step": 11200
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0006351192420777524,
+      "loss": 0.9769,
+      "step": 11225
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0006343025155178046,
+      "loss": 1.0332,
+      "step": 11250
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0006334857889578569,
+      "loss": 0.9927,
+      "step": 11275
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0006326690623979092,
+      "loss": 1.0269,
+      "step": 11300
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0006318523358379614,
+      "loss": 1.0341,
+      "step": 11325
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0006310356092780137,
+      "loss": 0.9852,
+      "step": 11350
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0006302188827180661,
+      "loss": 1.08,
+      "step": 11375
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0006294021561581183,
+      "loss": 1.0056,
+      "step": 11400
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0006285854295981706,
+      "loss": 1.0044,
+      "step": 11425
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0006277687030382229,
+      "loss": 0.9931,
+      "step": 11450
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.000626951976478275,
+      "loss": 1.0077,
+      "step": 11475
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0006261352499183273,
+      "loss": 1.0072,
+      "step": 11500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0006253185233583796,
+      "loss": 0.975,
+      "step": 11525
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0006245017967984319,
+      "loss": 0.9875,
+      "step": 11550
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0006236850702384842,
+      "loss": 0.9519,
+      "step": 11575
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0006228683436785364,
+      "loss": 1.0437,
+      "step": 11600
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0006220516171185887,
+      "loss": 0.9725,
+      "step": 11625
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0006212348905586411,
+      "loss": 1.0102,
+      "step": 11650
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0006204181639986932,
+      "loss": 1.072,
+      "step": 11675
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0006196014374387455,
+      "loss": 1.0051,
+      "step": 11700
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0006187847108787978,
+      "loss": 0.9997,
+      "step": 11725
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00061796798431885,
+      "loss": 1.0409,
+      "step": 11750
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0006171512577589024,
+      "loss": 1.0285,
+      "step": 11775
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0006163345311989546,
+      "loss": 1.0697,
+      "step": 11800
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0006155178046390069,
+      "loss": 1.0923,
+      "step": 11825
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0006147010780790592,
+      "loss": 1.0785,
+      "step": 11850
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0006138843515191113,
+      "loss": 0.9977,
+      "step": 11875
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0006130676249591636,
+      "loss": 1.0268,
+      "step": 11900
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.000612250898399216,
+      "loss": 1.022,
+      "step": 11925
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0006114341718392682,
+      "loss": 0.9922,
+      "step": 11950
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0006106174452793205,
+      "loss": 1.0643,
+      "step": 11975
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0006098007187193728,
+      "loss": 1.0241,
+      "step": 12000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.000608983992159425,
+      "loss": 0.9835,
+      "step": 12025
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0006081672655994774,
+      "loss": 0.9964,
+      "step": 12050
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0006073505390395295,
+      "loss": 0.9939,
+      "step": 12075
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0006065338124795818,
+      "loss": 1.0776,
+      "step": 12100
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0006057170859196341,
+      "loss": 1.0002,
+      "step": 12125
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0006049003593596863,
+      "loss": 1.053,
+      "step": 12150
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0006040836327997387,
+      "loss": 0.9921,
+      "step": 12175
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.000603266906239791,
+      "loss": 1.0601,
+      "step": 12200
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0006024501796798432,
+      "loss": 1.0023,
+      "step": 12225
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0006016334531198955,
+      "loss": 1.0065,
+      "step": 12250
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 0.9367652535438538,
+      "eval_runtime": 929.0448,
+      "eval_samples_per_second": 1.076,
+      "eval_steps_per_second": 0.135,
+      "eval_wer": 10.500792688613261,
+      "step": 12264
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0006008167265599478,
+      "loss": 0.9477,
+      "step": 12275
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0006,
+      "loss": 0.9413,
+      "step": 12300
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0005991832734400523,
+      "loss": 0.999,
+      "step": 12325
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0005983665468801045,
+      "loss": 0.9664,
+      "step": 12350
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0005975498203201568,
+      "loss": 0.9532,
+      "step": 12375
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0005967330937602091,
+      "loss": 0.9227,
+      "step": 12400
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0005959163672002613,
+      "loss": 0.916,
+      "step": 12425
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0005950996406403137,
+      "loss": 0.9664,
+      "step": 12450
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.000594282914080366,
+      "loss": 0.9344,
+      "step": 12475
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0005934661875204181,
+      "loss": 0.9729,
+      "step": 12500
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0005926494609604704,
+      "loss": 0.9404,
+      "step": 12525
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0005918327344005227,
+      "loss": 0.9351,
+      "step": 12550
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.000591016007840575,
+      "loss": 0.9692,
+      "step": 12575
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0005901992812806273,
+      "loss": 0.9366,
+      "step": 12600
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0005893825547206795,
+      "loss": 0.9594,
+      "step": 12625
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0005885658281607318,
+      "loss": 0.9101,
+      "step": 12650
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0005877491016007841,
+      "loss": 0.9675,
+      "step": 12675
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0005869323750408362,
+      "loss": 0.9396,
+      "step": 12700
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0005861156484808886,
+      "loss": 0.9851,
+      "step": 12725
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0005852989219209409,
+      "loss": 0.9424,
+      "step": 12750
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0005844821953609931,
+      "loss": 0.9211,
+      "step": 12775
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0005836654688010454,
+      "loss": 0.9562,
+      "step": 12800
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0005828487422410978,
+      "loss": 0.9599,
+      "step": 12825
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00058203201568115,
+      "loss": 0.9343,
+      "step": 12850
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0005812152891212023,
+      "loss": 0.9537,
+      "step": 12875
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0005803985625612544,
+      "loss": 0.9199,
+      "step": 12900
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0005795818360013067,
+      "loss": 0.9185,
+      "step": 12925
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.000578765109441359,
+      "loss": 0.9382,
+      "step": 12950
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0005779483828814113,
+      "loss": 0.9689,
+      "step": 12975
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0005771316563214636,
+      "loss": 0.9916,
+      "step": 13000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0005763149297615159,
+      "loss": 0.9593,
+      "step": 13025
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0005754982032015681,
+      "loss": 0.9291,
+      "step": 13050
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0005747141457040182,
+      "loss": 1.0283,
+      "step": 13075
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0005738974191440706,
+      "loss": 0.9577,
+      "step": 13100
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0005730806925841229,
+      "loss": 0.9594,
+      "step": 13125
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0005722639660241751,
+      "loss": 0.9411,
+      "step": 13150
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0005714472394642274,
+      "loss": 0.9607,
+      "step": 13175
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0005706305129042797,
+      "loss": 0.9902,
+      "step": 13200
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.000569813786344332,
+      "loss": 0.979,
+      "step": 13225
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0005689970597843842,
+      "loss": 0.9485,
+      "step": 13250
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0005681803332244365,
+      "loss": 0.9492,
+      "step": 13275
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0005673636066644887,
+      "loss": 0.9136,
+      "step": 13300
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.000566546880104541,
+      "loss": 0.9569,
+      "step": 13325
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0005657301535445932,
+      "loss": 0.9184,
+      "step": 13350
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0005649134269846456,
+      "loss": 0.9555,
+      "step": 13375
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0005640967004246979,
+      "loss": 0.9633,
+      "step": 13400
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0005632799738647501,
+      "loss": 0.9738,
+      "step": 13425
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0005624632473048023,
+      "loss": 0.9897,
+      "step": 13450
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0005616465207448546,
+      "loss": 1.0043,
+      "step": 13475
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0005608297941849069,
+      "loss": 0.9531,
+      "step": 13500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0005600130676249592,
+      "loss": 0.9441,
+      "step": 13525
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0005591963410650115,
+      "loss": 0.9505,
+      "step": 13550
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0005583796145050637,
+      "loss": 0.9985,
+      "step": 13575
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.000557562887945116,
+      "loss": 0.939,
+      "step": 13600
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0005567461613851682,
+      "loss": 1.0095,
+      "step": 13625
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0005559294348252205,
+      "loss": 0.9849,
+      "step": 13650
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0005551127082652728,
+      "loss": 0.9495,
+      "step": 13675
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.000554295981705325,
+      "loss": 0.9426,
+      "step": 13700
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0005534792551453773,
+      "loss": 0.9301,
+      "step": 13725
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0005526625285854296,
+      "loss": 0.9652,
+      "step": 13750
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0005518458020254819,
+      "loss": 1.0232,
+      "step": 13775
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0005510290754655342,
+      "loss": 0.9095,
+      "step": 13800
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0005502123489055865,
+      "loss": 1.0281,
+      "step": 13825
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0005493956223456387,
+      "loss": 0.9334,
+      "step": 13850
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0005485788957856909,
+      "loss": 0.9301,
+      "step": 13875
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0005477621692257432,
+      "loss": 0.9297,
+      "step": 13900
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0005469454426657955,
+      "loss": 0.9506,
+      "step": 13925
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0005461287161058478,
+      "loss": 0.949,
+      "step": 13950
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0005453119895459,
+      "loss": 0.9812,
+      "step": 13975
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0005444952629859523,
+      "loss": 0.9207,
+      "step": 14000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0005436785364260047,
+      "loss": 0.9633,
+      "step": 14025
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0005428618098660569,
+      "loss": 0.9787,
+      "step": 14050
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0005420450833061091,
+      "loss": 0.9924,
+      "step": 14075
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0005412283567461614,
+      "loss": 0.9027,
+      "step": 14100
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0005404116301862136,
+      "loss": 0.9882,
+      "step": 14125
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.000539594903626266,
+      "loss": 0.9808,
+      "step": 14150
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0005387781770663182,
+      "loss": 0.9603,
+      "step": 14175
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0005379614505063705,
+      "loss": 1.01,
+      "step": 14200
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0005371447239464228,
+      "loss": 1.0208,
+      "step": 14225
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.000536327997386475,
+      "loss": 0.988,
+      "step": 14250
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0005355112708265272,
+      "loss": 0.9962,
+      "step": 14275
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0005346945442665796,
+      "loss": 1.0374,
+      "step": 14300
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0005338778177066318,
+      "loss": 0.9575,
+      "step": 14325
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0005330610911466841,
+      "loss": 0.933,
+      "step": 14350
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0005322443645867364,
+      "loss": 0.9409,
+      "step": 14375
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0005314276380267886,
+      "loss": 0.9495,
+      "step": 14400
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.000530610911466841,
+      "loss": 1.0001,
+      "step": 14425
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0005297941849068932,
+      "loss": 1.0216,
+      "step": 14450
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0005290101274093434,
+      "loss": 0.9431,
+      "step": 14475
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0005281934008493956,
+      "loss": 0.9686,
+      "step": 14500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0005273766742894479,
+      "loss": 0.9477,
+      "step": 14525
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0005265599477295002,
+      "loss": 0.9446,
+      "step": 14550
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0005257432211695525,
+      "loss": 0.9808,
+      "step": 14575
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0005249264946096047,
+      "loss": 0.9629,
+      "step": 14600
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0005241097680496569,
+      "loss": 0.9066,
+      "step": 14625
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0005232930414897092,
+      "loss": 0.9627,
+      "step": 14650
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0005224763149297615,
+      "loss": 0.9822,
+      "step": 14675
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0005216595883698138,
+      "loss": 0.9371,
+      "step": 14700
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0005208428618098661,
+      "loss": 1.0388,
+      "step": 14725
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0005200261352499184,
+      "loss": 0.9602,
+      "step": 14750
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0005192094086899706,
+      "loss": 0.968,
+      "step": 14775
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0005183926821300229,
+      "loss": 0.9375,
+      "step": 14800
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0005175759555700752,
+      "loss": 0.9769,
+      "step": 14825
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0005167592290101274,
+      "loss": 0.9493,
+      "step": 14850
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0005159425024501797,
+      "loss": 0.9451,
+      "step": 14875
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0005151257758902319,
+      "loss": 0.9599,
+      "step": 14900
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0005143090493302842,
+      "loss": 0.9742,
+      "step": 14925
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0005134923227703365,
+      "loss": 0.9618,
+      "step": 14950
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0005126755962103888,
+      "loss": 1.0095,
+      "step": 14975
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0005118588696504411,
+      "loss": 0.9383,
+      "step": 15000
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0005110421430904933,
+      "loss": 0.9605,
+      "step": 15025
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0005102254165305455,
+      "loss": 0.9599,
+      "step": 15050
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0005094086899705978,
+      "loss": 0.9536,
+      "step": 15075
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0005085919634106502,
+      "loss": 0.9486,
+      "step": 15100
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0005077752368507024,
+      "loss": 0.9549,
+      "step": 15125
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0005069585102907547,
+      "loss": 0.9811,
+      "step": 15150
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0005061417837308069,
+      "loss": 0.9321,
+      "step": 15175
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0005053250571708592,
+      "loss": 0.9358,
+      "step": 15200
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0005045083306109115,
+      "loss": 0.952,
+      "step": 15225
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0005036916040509637,
+      "loss": 0.9713,
+      "step": 15250
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.000502874877491016,
+      "loss": 0.9529,
+      "step": 15275
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0005020581509310683,
+      "loss": 1.0018,
+      "step": 15300
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0005012414243711205,
+      "loss": 0.9907,
+      "step": 15325
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.8856672644615173,
+      "eval_runtime": 957.3834,
+      "eval_samples_per_second": 1.045,
+      "eval_steps_per_second": 0.131,
+      "eval_wer": 6.835773570829059,
+      "step": 15330
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0005004246978111729,
+      "loss": 0.8702,
+      "step": 15350
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0004996079712512251,
+      "loss": 0.8384,
+      "step": 15375
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0004987912446912774,
+      "loss": 0.8731,
+      "step": 15400
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0004979745181313297,
+      "loss": 0.9217,
+      "step": 15425
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0004971577915713819,
+      "loss": 0.885,
+      "step": 15450
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0004963410650114341,
+      "loss": 0.9239,
+      "step": 15475
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0004955243384514865,
+      "loss": 0.9232,
+      "step": 15500
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0004947076118915388,
+      "loss": 0.899,
+      "step": 15525
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.000493890885331591,
+      "loss": 0.9134,
+      "step": 15550
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0004930741587716432,
+      "loss": 0.8702,
+      "step": 15575
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0004922574322116955,
+      "loss": 0.9255,
+      "step": 15600
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0004914407056517479,
+      "loss": 0.8858,
+      "step": 15625
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0004906239790918001,
+      "loss": 0.874,
+      "step": 15650
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0004898072525318523,
+      "loss": 0.9004,
+      "step": 15675
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0004889905259719046,
+      "loss": 0.8701,
+      "step": 15700
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0004881737994119569,
+      "loss": 0.9114,
+      "step": 15725
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00048735707285200915,
+      "loss": 0.8608,
+      "step": 15750
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00048657301535445937,
+      "loss": 0.9103,
+      "step": 15775
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00048575628879451164,
+      "loss": 0.9099,
+      "step": 15800
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0004849722312969618,
+      "loss": 0.9542,
+      "step": 15825
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00048415550473701407,
+      "loss": 0.9748,
+      "step": 15850
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00048333877817706634,
+      "loss": 0.8518,
+      "step": 15875
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0004825220516171186,
+      "loss": 0.9075,
+      "step": 15900
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0004817053250571709,
+      "loss": 0.9004,
+      "step": 15925
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00048088859849722314,
+      "loss": 0.9015,
+      "step": 15950
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0004800718719372754,
+      "loss": 0.9192,
+      "step": 15975
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0004792551453773277,
+      "loss": 0.8597,
+      "step": 16000
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00047843841881737995,
+      "loss": 0.8859,
+      "step": 16025
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0004776216922574322,
+      "loss": 0.924,
+      "step": 16050
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0004768049656974845,
+      "loss": 0.8799,
+      "step": 16075
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00047598823913753676,
+      "loss": 0.9405,
+      "step": 16100
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0004751715125775891,
+      "loss": 0.8748,
+      "step": 16125
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0004743547860176413,
+      "loss": 0.9115,
+      "step": 16150
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00047353805945769356,
+      "loss": 0.9161,
+      "step": 16175
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00047272133289774583,
+      "loss": 0.9297,
+      "step": 16200
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00047190460633779815,
+      "loss": 0.9021,
+      "step": 16225
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00047108787977785037,
+      "loss": 0.8839,
+      "step": 16250
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00047027115321790264,
+      "loss": 0.9481,
+      "step": 16275
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0004694544266579549,
+      "loss": 0.8923,
+      "step": 16300
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00046863770009800723,
+      "loss": 0.8888,
+      "step": 16325
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00046782097353805944,
+      "loss": 0.9642,
+      "step": 16350
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0004670042469781117,
+      "loss": 0.8957,
+      "step": 16375
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00046618752041816403,
+      "loss": 0.8488,
+      "step": 16400
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0004653707938582163,
+      "loss": 0.9738,
+      "step": 16425
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00046455406729826857,
+      "loss": 0.9539,
+      "step": 16450
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0004637373407383208,
+      "loss": 0.8878,
+      "step": 16475
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0004629206141783731,
+      "loss": 0.9682,
+      "step": 16500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0004621038876184254,
+      "loss": 0.9023,
+      "step": 16525
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00046128716105847765,
+      "loss": 0.8979,
+      "step": 16550
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00046047043449852986,
+      "loss": 0.9489,
+      "step": 16575
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0004596537079385822,
+      "loss": 0.9734,
+      "step": 16600
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00045883698137863445,
+      "loss": 0.9354,
+      "step": 16625
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0004580202548186867,
+      "loss": 0.8973,
+      "step": 16650
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.000457203528258739,
+      "loss": 0.8845,
+      "step": 16675
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00045638680169879126,
+      "loss": 0.9092,
+      "step": 16700
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00045557007513884353,
+      "loss": 0.9234,
+      "step": 16725
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0004547533485788958,
+      "loss": 0.904,
+      "step": 16750
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00045393662201894807,
+      "loss": 0.9256,
+      "step": 16775
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00045311989545900034,
+      "loss": 1.0002,
+      "step": 16800
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0004523031688990526,
+      "loss": 0.9155,
+      "step": 16825
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00045148644233910487,
+      "loss": 0.8999,
+      "step": 16850
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00045066971577915714,
+      "loss": 0.8973,
+      "step": 16875
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0004498529892192094,
+      "loss": 0.9038,
+      "step": 16900
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0004490362626592617,
+      "loss": 0.963,
+      "step": 16925
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.000448219536099314,
+      "loss": 0.9102,
+      "step": 16950
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0004474028095393662,
+      "loss": 0.9021,
+      "step": 16975
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0004465860829794185,
+      "loss": 0.9318,
+      "step": 17000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00044576935641947075,
+      "loss": 0.9126,
+      "step": 17025
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0004449526298595231,
+      "loss": 0.9213,
+      "step": 17050
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0004441359032995753,
+      "loss": 0.8985,
+      "step": 17075
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00044331917673962756,
+      "loss": 0.9006,
+      "step": 17100
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00044250245017967983,
+      "loss": 0.8745,
+      "step": 17125
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00044168572361973215,
+      "loss": 0.9848,
+      "step": 17150
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00044086899705978437,
+      "loss": 0.8967,
+      "step": 17175
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00044005227049983664,
+      "loss": 0.972,
+      "step": 17200
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00043923554393988896,
+      "loss": 0.9553,
+      "step": 17225
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0004384188173799412,
+      "loss": 0.9128,
+      "step": 17250
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0004376020908199935,
+      "loss": 0.9333,
+      "step": 17275
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0004367853642600457,
+      "loss": 0.9197,
+      "step": 17300
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00043596863770009803,
+      "loss": 0.9185,
+      "step": 17325
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0004351519111401503,
+      "loss": 0.8902,
+      "step": 17350
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00043433518458020257,
+      "loss": 0.9017,
+      "step": 17375
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0004335184580202548,
+      "loss": 0.8876,
+      "step": 17400
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0004327017314603071,
+      "loss": 0.8792,
+      "step": 17425
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0004318850049003594,
+      "loss": 0.8989,
+      "step": 17450
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00043106827834041165,
+      "loss": 0.8694,
+      "step": 17475
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0004302515517804639,
+      "loss": 0.8945,
+      "step": 17500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0004294348252205162,
+      "loss": 0.8897,
+      "step": 17525
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00042861809866056845,
+      "loss": 0.925,
+      "step": 17550
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0004278013721006207,
+      "loss": 0.9442,
+      "step": 17575
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.000426984645540673,
+      "loss": 0.9025,
+      "step": 17600
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00042616791898072526,
+      "loss": 0.8805,
+      "step": 17625
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00042535119242077753,
+      "loss": 0.9071,
+      "step": 17650
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0004245344658608298,
+      "loss": 0.9194,
+      "step": 17675
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00042371773930088206,
+      "loss": 0.8833,
+      "step": 17700
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00042290101274093433,
+      "loss": 0.9591,
+      "step": 17725
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0004220842861809866,
+      "loss": 0.9122,
+      "step": 17750
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0004212675596210389,
+      "loss": 0.9193,
+      "step": 17775
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00042045083306109114,
+      "loss": 0.9287,
+      "step": 17800
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0004196341065011434,
+      "loss": 0.9433,
+      "step": 17825
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0004188173799411957,
+      "loss": 0.8985,
+      "step": 17850
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.000418000653381248,
+      "loss": 0.9895,
+      "step": 17875
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0004171839268213002,
+      "loss": 0.9005,
+      "step": 17900
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0004163672002613525,
+      "loss": 0.9078,
+      "step": 17925
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00041555047370140475,
+      "loss": 0.8996,
+      "step": 17950
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0004147337471414571,
+      "loss": 0.9454,
+      "step": 17975
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00041391702058150934,
+      "loss": 0.9536,
+      "step": 18000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00041310029402156156,
+      "loss": 0.908,
+      "step": 18025
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0004122835674616139,
+      "loss": 0.9281,
+      "step": 18050
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00041146684090166615,
+      "loss": 0.9143,
+      "step": 18075
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0004106501143417184,
+      "loss": 0.9711,
+      "step": 18100
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00040983338778177063,
+      "loss": 0.9095,
+      "step": 18125
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00040901666122182296,
+      "loss": 0.9171,
+      "step": 18150
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0004081999346618752,
+      "loss": 0.9006,
+      "step": 18175
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0004073832081019275,
+      "loss": 0.8891,
+      "step": 18200
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0004065664815419797,
+      "loss": 0.9096,
+      "step": 18225
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00040574975498203203,
+      "loss": 0.9266,
+      "step": 18250
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0004049330284220843,
+      "loss": 0.8653,
+      "step": 18275
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00040411630186213657,
+      "loss": 0.9335,
+      "step": 18300
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00040329957530218884,
+      "loss": 0.9124,
+      "step": 18325
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0004024828487422411,
+      "loss": 0.8719,
+      "step": 18350
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0004016661221822934,
+      "loss": 0.8971,
+      "step": 18375
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 0.8379027247428894,
+      "eval_runtime": 924.2552,
+      "eval_samples_per_second": 1.082,
+      "eval_steps_per_second": 0.135,
+      "eval_wer": 3.4785041499580336,
+      "step": 18396
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00040084939562234564,
+      "loss": 0.859,
+      "step": 18400
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0004000326690623979,
+      "loss": 0.8687,
+      "step": 18425
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0003992159425024502,
+      "loss": 0.9106,
+      "step": 18450
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00039839921594250245,
+      "loss": 0.8549,
+      "step": 18475
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0003975824893825547,
+      "loss": 0.8658,
+      "step": 18500
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.000396765762822607,
+      "loss": 0.884,
+      "step": 18525
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00039594903626265926,
+      "loss": 0.8732,
+      "step": 18550
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0003951323097027115,
+      "loss": 0.865,
+      "step": 18575
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00039431558314276385,
+      "loss": 0.8562,
+      "step": 18600
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00039349885658281606,
+      "loss": 0.8829,
+      "step": 18625
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00039268213002286833,
+      "loss": 0.841,
+      "step": 18650
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0003918654034629206,
+      "loss": 0.8788,
+      "step": 18675
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0003910486769029729,
+      "loss": 0.8661,
+      "step": 18700
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0003902319503430252,
+      "loss": 0.8646,
+      "step": 18725
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0003894152237830774,
+      "loss": 0.8711,
+      "step": 18750
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0003885984972231297,
+      "loss": 0.867,
+      "step": 18775
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.000387781770663182,
+      "loss": 0.8376,
+      "step": 18800
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00038696504410323427,
+      "loss": 0.8376,
+      "step": 18825
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0003861483175432865,
+      "loss": 0.868,
+      "step": 18850
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0003853315909833388,
+      "loss": 0.87,
+      "step": 18875
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0003845148644233911,
+      "loss": 0.8565,
+      "step": 18900
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00038369813786344334,
+      "loss": 0.8978,
+      "step": 18925
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00038288141130349556,
+      "loss": 0.8337,
+      "step": 18950
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0003820646847435479,
+      "loss": 0.8872,
+      "step": 18975
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00038124795818360015,
+      "loss": 0.8982,
+      "step": 19000
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0003804312316236524,
+      "loss": 0.8826,
+      "step": 19025
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0003796145050637047,
+      "loss": 0.8575,
+      "step": 19050
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00037879777850375696,
+      "loss": 0.8727,
+      "step": 19075
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0003779810519438092,
+      "loss": 0.8679,
+      "step": 19100
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0003771643253838615,
+      "loss": 0.8943,
+      "step": 19125
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00037634759882391376,
+      "loss": 0.8655,
+      "step": 19150
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00037553087226396603,
+      "loss": 0.8608,
+      "step": 19175
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0003747141457040183,
+      "loss": 0.8632,
+      "step": 19200
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00037389741914407057,
+      "loss": 0.9066,
+      "step": 19225
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00037308069258412284,
+      "loss": 0.8252,
+      "step": 19250
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0003722639660241751,
+      "loss": 0.845,
+      "step": 19275
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0003714472394642274,
+      "loss": 0.8753,
+      "step": 19300
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0003706305129042797,
+      "loss": 0.8675,
+      "step": 19325
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0003698137863443319,
+      "loss": 0.8494,
+      "step": 19350
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0003689970597843842,
+      "loss": 0.8422,
+      "step": 19375
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00036818033322443645,
+      "loss": 0.8895,
+      "step": 19400
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00036736360666448877,
+      "loss": 0.8138,
+      "step": 19425
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00036654688010454104,
+      "loss": 0.8782,
+      "step": 19450
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00036573015354459326,
+      "loss": 0.898,
+      "step": 19475
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0003649134269846455,
+      "loss": 0.8641,
+      "step": 19500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00036409670042469785,
+      "loss": 0.9043,
+      "step": 19525
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0003632799738647501,
+      "loss": 0.8375,
+      "step": 19550
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00036246324730480233,
+      "loss": 0.8787,
+      "step": 19575
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00036164652074485465,
+      "loss": 0.8633,
+      "step": 19600
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0003608297941849069,
+      "loss": 0.8185,
+      "step": 19625
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0003600130676249592,
+      "loss": 0.8812,
+      "step": 19650
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0003591963410650114,
+      "loss": 0.8775,
+      "step": 19675
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00035837961450506373,
+      "loss": 0.8696,
+      "step": 19700
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.000357562887945116,
+      "loss": 0.8679,
+      "step": 19725
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00035674616138516827,
+      "loss": 0.8761,
+      "step": 19750
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0003559294348252205,
+      "loss": 0.8419,
+      "step": 19775
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0003551127082652728,
+      "loss": 0.819,
+      "step": 19800
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00035429598170532507,
+      "loss": 0.8853,
+      "step": 19825
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00035347925514537734,
+      "loss": 0.8783,
+      "step": 19850
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0003526625285854296,
+      "loss": 0.8378,
+      "step": 19875
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0003518458020254819,
+      "loss": 0.8761,
+      "step": 19900
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00035102907546553415,
+      "loss": 0.8666,
+      "step": 19925
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0003502123489055864,
+      "loss": 0.8914,
+      "step": 19950
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0003493956223456387,
+      "loss": 0.8462,
+      "step": 19975
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00034857889578569095,
+      "loss": 0.8955,
+      "step": 20000
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0003477621692257432,
+      "loss": 0.877,
+      "step": 20025
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0003469454426657955,
+      "loss": 0.9005,
+      "step": 20050
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00034612871610584776,
+      "loss": 0.8345,
+      "step": 20075
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00034531198954590003,
+      "loss": 0.8579,
+      "step": 20100
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0003444952629859523,
+      "loss": 0.9066,
+      "step": 20125
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0003436785364260046,
+      "loss": 0.9039,
+      "step": 20150
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00034286180986605684,
+      "loss": 0.8852,
+      "step": 20175
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0003420450833061091,
+      "loss": 0.8837,
+      "step": 20200
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00034122835674616137,
+      "loss": 0.8865,
+      "step": 20225
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0003404116301862137,
+      "loss": 0.8624,
+      "step": 20250
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00033959490362626596,
+      "loss": 0.8882,
+      "step": 20275
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0003387781770663182,
+      "loss": 0.8593,
+      "step": 20300
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00033796145050637045,
+      "loss": 0.8604,
+      "step": 20325
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00033714472394642277,
+      "loss": 0.8864,
+      "step": 20350
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00033632799738647504,
+      "loss": 0.8714,
+      "step": 20375
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00033551127082652725,
+      "loss": 0.865,
+      "step": 20400
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0003346945442665796,
+      "loss": 0.8669,
+      "step": 20425
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00033387781770663185,
+      "loss": 0.8539,
+      "step": 20450
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0003330610911466841,
+      "loss": 0.8656,
+      "step": 20475
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00033224436458673633,
+      "loss": 0.8657,
+      "step": 20500
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00033142763802678865,
+      "loss": 0.8555,
+      "step": 20525
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0003306109114668409,
+      "loss": 0.8165,
+      "step": 20550
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0003297941849068932,
+      "loss": 0.8965,
+      "step": 20575
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0003289774583469454,
+      "loss": 0.875,
+      "step": 20600
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00032816073178699773,
+      "loss": 0.8652,
+      "step": 20625
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00032734400522705,
+      "loss": 0.8254,
+      "step": 20650
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00032652727866710226,
+      "loss": 0.8958,
+      "step": 20675
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00032571055210715453,
+      "loss": 0.8497,
+      "step": 20700
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0003248938255472068,
+      "loss": 0.9016,
+      "step": 20725
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00032407709898725907,
+      "loss": 0.8201,
+      "step": 20750
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00032326037242731134,
+      "loss": 0.8374,
+      "step": 20775
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0003224436458673636,
+      "loss": 0.8702,
+      "step": 20800
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0003216269193074159,
+      "loss": 0.8392,
+      "step": 20825
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00032081019274746815,
+      "loss": 0.8368,
+      "step": 20850
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0003199934661875204,
+      "loss": 0.8286,
+      "step": 20875
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0003191767396275727,
+      "loss": 0.8603,
+      "step": 20900
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00031836001306762495,
+      "loss": 0.8719,
+      "step": 20925
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0003175432865076772,
+      "loss": 0.9351,
+      "step": 20950
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00031672655994772954,
+      "loss": 0.8594,
+      "step": 20975
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0003159098333877818,
+      "loss": 0.8628,
+      "step": 21000
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00031509310682783403,
+      "loss": 0.8722,
+      "step": 21025
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0003142763802678863,
+      "loss": 0.9016,
+      "step": 21050
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0003134596537079386,
+      "loss": 0.8682,
+      "step": 21075
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0003126429271479909,
+      "loss": 0.8727,
+      "step": 21100
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0003118262005880431,
+      "loss": 0.9259,
+      "step": 21125
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00031100947402809537,
+      "loss": 0.8866,
+      "step": 21150
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0003101927474681477,
+      "loss": 0.8884,
+      "step": 21175
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00030937602090819996,
+      "loss": 0.8934,
+      "step": 21200
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0003085592943482522,
+      "loss": 0.8695,
+      "step": 21225
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0003077425677883045,
+      "loss": 0.9416,
+      "step": 21250
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00030692584122835677,
+      "loss": 0.8355,
+      "step": 21275
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00030610911466840904,
+      "loss": 0.9009,
+      "step": 21300
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00030529238810846125,
+      "loss": 0.8709,
+      "step": 21325
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.0003044756615485136,
+      "loss": 0.8767,
+      "step": 21350
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00030365893498856584,
+      "loss": 0.8528,
+      "step": 21375
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.0003028422084286181,
+      "loss": 0.8456,
+      "step": 21400
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00030202548186867033,
+      "loss": 0.8351,
+      "step": 21425
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00030120875530872265,
+      "loss": 0.858,
+      "step": 21450
+    },
+    {
+      "epoch": 7.0,
+      "eval_loss": 0.812373161315918,
+      "eval_runtime": 934.4652,
+      "eval_samples_per_second": 1.07,
+      "eval_steps_per_second": 0.134,
+      "eval_wer": 2.2754826074792502,
+      "step": 21462
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0003003920287487749,
+      "loss": 0.8563,
+      "step": 21475
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.0002995753021888272,
+      "loss": 0.825,
+      "step": 21500
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.00029875857562887946,
+      "loss": 0.8714,
+      "step": 21525
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0002979418490689317,
+      "loss": 0.7912,
+      "step": 21550
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.000297125122508984,
+      "loss": 0.8249,
+      "step": 21575
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00029630839594903626,
+      "loss": 0.8296,
+      "step": 21600
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00029549166938908853,
+      "loss": 0.8464,
+      "step": 21625
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0002946749428291408,
+      "loss": 0.8263,
+      "step": 21650
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.00029385821626919307,
+      "loss": 0.834,
+      "step": 21675
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00029304148970924534,
+      "loss": 0.8206,
+      "step": 21700
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00029222476314929766,
+      "loss": 0.8296,
+      "step": 21725
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.0002914080365893499,
+      "loss": 0.8253,
+      "step": 21750
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00029059131002940214,
+      "loss": 0.8362,
+      "step": 21775
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00028977458346945447,
+      "loss": 0.8686,
+      "step": 21800
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.00028895785690950674,
+      "loss": 0.8181,
+      "step": 21825
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00028814113034955895,
+      "loss": 0.7907,
+      "step": 21850
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.0002873244037896112,
+      "loss": 0.8777,
+      "step": 21875
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00028650767722966354,
+      "loss": 0.88,
+      "step": 21900
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.0002856909506697158,
+      "loss": 0.8507,
+      "step": 21925
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.000284874224109768,
+      "loss": 0.8222,
+      "step": 21950
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.0002840574975498203,
+      "loss": 0.802,
+      "step": 21975
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0002832407709898726,
+      "loss": 0.8305,
+      "step": 22000
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0002824240444299249,
+      "loss": 0.8268,
+      "step": 22025
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.0002816073178699771,
+      "loss": 0.8203,
+      "step": 22050
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0002807905913100294,
+      "loss": 0.828,
+      "step": 22075
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.0002799738647500817,
+      "loss": 0.8244,
+      "step": 22100
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00027915713819013396,
+      "loss": 0.8224,
+      "step": 22125
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.0002783404116301862,
+      "loss": 0.8037,
+      "step": 22150
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.0002775236850702385,
+      "loss": 0.8378,
+      "step": 22175
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00027670695851029077,
+      "loss": 0.8376,
+      "step": 22200
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00027589023195034304,
+      "loss": 0.8312,
+      "step": 22225
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.0002750735053903953,
+      "loss": 0.8517,
+      "step": 22250
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.0002742567788304476,
+      "loss": 0.8209,
+      "step": 22275
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00027344005227049984,
+      "loss": 0.7937,
+      "step": 22300
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.0002726233257105521,
+      "loss": 0.8252,
+      "step": 22325
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.0002718065991506044,
+      "loss": 0.8369,
+      "step": 22350
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00027098987259065665,
+      "loss": 0.8664,
+      "step": 22375
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.0002701731460307089,
+      "loss": 0.8301,
+      "step": 22400
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.0002693564194707612,
+      "loss": 0.8137,
+      "step": 22425
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00026853969291081346,
+      "loss": 0.799,
+      "step": 22450
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.0002677229663508657,
+      "loss": 0.8315,
+      "step": 22475
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.000266906239790918,
+      "loss": 0.8476,
+      "step": 22500
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0002660895132309703,
+      "loss": 0.8102,
+      "step": 22525
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0002652727866710226,
+      "loss": 0.8072,
+      "step": 22550
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.0002644560601110748,
+      "loss": 0.8209,
+      "step": 22575
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.00026363933355112707,
+      "loss": 0.8533,
+      "step": 22600
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.0002628226069911794,
+      "loss": 0.8236,
+      "step": 22625
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.00026200588043123166,
+      "loss": 0.8425,
+      "step": 22650
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.0002611891538712839,
+      "loss": 0.8203,
+      "step": 22675
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.00026037242731133614,
+      "loss": 0.8542,
+      "step": 22700
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.00025955570075138847,
+      "loss": 0.8355,
+      "step": 22725
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00025873897419144073,
+      "loss": 0.8262,
+      "step": 22750
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00025792224763149295,
+      "loss": 0.8016,
+      "step": 22775
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00025710552107154527,
+      "loss": 0.8357,
+      "step": 22800
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00025628879451159754,
+      "loss": 0.8348,
+      "step": 22825
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.0002554720679516498,
+      "loss": 0.8382,
+      "step": 22850
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.000254655341391702,
+      "loss": 0.8915,
+      "step": 22875
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00025383861483175435,
+      "loss": 0.8582,
+      "step": 22900
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.0002530218882718066,
+      "loss": 0.8403,
+      "step": 22925
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0002522051617118589,
+      "loss": 0.8239,
+      "step": 22950
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0002513884351519111,
+      "loss": 0.8343,
+      "step": 22975
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0002505717085919634,
+      "loss": 0.8199,
+      "step": 23000
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.0002497549820320157,
+      "loss": 0.8383,
+      "step": 23025
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.00024893825547206796,
+      "loss": 0.8285,
+      "step": 23050
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.00024812152891212023,
+      "loss": 0.8192,
+      "step": 23075
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0002473048023521725,
+      "loss": 0.8531,
+      "step": 23100
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00024648807579222477,
+      "loss": 0.8463,
+      "step": 23125
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.00024567134923227704,
+      "loss": 0.8328,
+      "step": 23150
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.0002448546226723293,
+      "loss": 0.8301,
+      "step": 23175
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.00024403789611238157,
+      "loss": 0.7949,
+      "step": 23200
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.00024322116955243384,
+      "loss": 0.8713,
+      "step": 23225
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.00024240444299248614,
+      "loss": 0.8304,
+      "step": 23250
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.00024158771643253838,
+      "loss": 0.8424,
+      "step": 23275
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.00024077098987259067,
+      "loss": 0.8135,
+      "step": 23300
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.00023995426331264294,
+      "loss": 0.8487,
+      "step": 23325
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.0002391375367526952,
+      "loss": 0.79,
+      "step": 23350
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.00023832081019274748,
+      "loss": 0.8142,
+      "step": 23375
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.00023750408363279975,
+      "loss": 0.8285,
+      "step": 23400
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.00023668735707285202,
+      "loss": 0.8223,
+      "step": 23425
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.0002358706305129043,
+      "loss": 0.8366,
+      "step": 23450
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.00023505390395295656,
+      "loss": 0.8234,
+      "step": 23475
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.00023423717739300882,
+      "loss": 0.8345,
+      "step": 23500
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.0002334204508330611,
+      "loss": 0.7979,
+      "step": 23525
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.00023260372427311336,
+      "loss": 0.8581,
+      "step": 23550
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.00023178699771316563,
+      "loss": 0.8349,
+      "step": 23575
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.00023097027115321793,
+      "loss": 0.799,
+      "step": 23600
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.00023015354459327017,
+      "loss": 0.8198,
+      "step": 23625
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.00022933681803332246,
+      "loss": 0.7925,
+      "step": 23650
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.0002285200914733747,
+      "loss": 0.8696,
+      "step": 23675
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.000227703364913427,
+      "loss": 0.8114,
+      "step": 23700
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.00022688663835347924,
+      "loss": 0.8593,
+      "step": 23725
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.00022606991179353154,
+      "loss": 0.8285,
+      "step": 23750
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.00022525318523358378,
+      "loss": 0.8469,
+      "step": 23775
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.00022443645867363608,
+      "loss": 0.8801,
+      "step": 23800
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.00022361973211368832,
+      "loss": 0.8604,
+      "step": 23825
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.00022280300555374061,
+      "loss": 0.8238,
+      "step": 23850
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.00022198627899379288,
+      "loss": 0.8033,
+      "step": 23875
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.00022116955243384515,
+      "loss": 0.8529,
+      "step": 23900
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.00022035282587389742,
+      "loss": 0.8373,
+      "step": 23925
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.0002195360993139497,
+      "loss": 0.8218,
+      "step": 23950
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.00021871937275400196,
+      "loss": 0.8206,
+      "step": 23975
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.00021790264619405423,
+      "loss": 0.8293,
+      "step": 24000
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.00021708591963410652,
+      "loss": 0.8172,
+      "step": 24025
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.00021626919307415876,
+      "loss": 0.8329,
+      "step": 24050
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.00021545246651421106,
+      "loss": 0.8494,
+      "step": 24075
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.0002146357399542633,
+      "loss": 0.8686,
+      "step": 24100
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.0002138190133943156,
+      "loss": 0.8416,
+      "step": 24125
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.00021300228683436787,
+      "loss": 0.8036,
+      "step": 24150
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.00021218556027442014,
+      "loss": 0.8794,
+      "step": 24175
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.0002113688337144724,
+      "loss": 0.7951,
+      "step": 24200
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.00021055210715452467,
+      "loss": 0.8146,
+      "step": 24225
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.00020973538059457694,
+      "loss": 0.8773,
+      "step": 24250
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.0002089186540346292,
+      "loss": 0.8265,
+      "step": 24275
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.00020810192747468148,
+      "loss": 0.8296,
+      "step": 24300
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.00020728520091473375,
+      "loss": 0.8041,
+      "step": 24325
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.00020646847435478602,
+      "loss": 0.8803,
+      "step": 24350
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.00020565174779483829,
+      "loss": 0.8693,
+      "step": 24375
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.00020483502123489055,
+      "loss": 0.856,
+      "step": 24400
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.00020401829467494285,
+      "loss": 0.7898,
+      "step": 24425
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.0002032015681149951,
+      "loss": 0.8292,
+      "step": 24450
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.0002023848415550474,
+      "loss": 0.8262,
+      "step": 24475
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.00020156811499509963,
+      "loss": 0.824,
+      "step": 24500
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.00020075138843515193,
+      "loss": 0.8035,
+      "step": 24525
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 0.7898829579353333,
+      "eval_runtime": 928.0218,
+      "eval_samples_per_second": 1.078,
+      "eval_steps_per_second": 0.135,
+      "eval_wer": 0.9045975939569151,
+      "step": 24528
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 0.00019993466187520417,
+      "loss": 0.7917,
+      "step": 24550
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 0.00019911793531525646,
+      "loss": 0.8023,
+      "step": 24575
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 0.0001983012087553087,
+      "loss": 0.8186,
+      "step": 24600
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.000197484482195361,
+      "loss": 0.8236,
+      "step": 24625
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 0.00019666775563541327,
+      "loss": 0.787,
+      "step": 24650
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.00019585102907546554,
+      "loss": 0.8501,
+      "step": 24675
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.0001950343025155178,
+      "loss": 0.8028,
+      "step": 24700
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.00019421757595557008,
+      "loss": 0.7872,
+      "step": 24725
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 0.00019340084939562237,
+      "loss": 0.7791,
+      "step": 24750
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 0.0001925841228356746,
+      "loss": 0.7923,
+      "step": 24775
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.0001917673962757269,
+      "loss": 0.7652,
+      "step": 24800
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 0.00019095066971577915,
+      "loss": 0.7888,
+      "step": 24825
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 0.00019013394315583145,
+      "loss": 0.8608,
+      "step": 24850
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 0.0001893172165958837,
+      "loss": 0.8283,
+      "step": 24875
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.00018850049003593598,
+      "loss": 0.7833,
+      "step": 24900
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 0.00018768376347598825,
+      "loss": 0.8041,
+      "step": 24925
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.00018686703691604052,
+      "loss": 0.7879,
+      "step": 24950
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.0001860503103560928,
+      "loss": 0.7992,
+      "step": 24975
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.00018523358379614506,
+      "loss": 0.8202,
+      "step": 25000
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 0.00018441685723619733,
+      "loss": 0.7904,
+      "step": 25025
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 0.0001836001306762496,
+      "loss": 0.7857,
+      "step": 25050
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 0.00018278340411630187,
+      "loss": 0.832,
+      "step": 25075
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.00018196667755635413,
+      "loss": 0.802,
+      "step": 25100
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.0001811499509964064,
+      "loss": 0.8337,
+      "step": 25125
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 0.00018033322443645867,
+      "loss": 0.8053,
+      "step": 25150
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 0.00017951649787651094,
+      "loss": 0.8313,
+      "step": 25175
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 0.00017869977131656324,
+      "loss": 0.8284,
+      "step": 25200
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 0.00017788304475661548,
+      "loss": 0.82,
+      "step": 25225
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.00017706631819666777,
+      "loss": 0.8016,
+      "step": 25250
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.00017624959163672002,
+      "loss": 0.8122,
+      "step": 25275
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 0.0001754328650767723,
+      "loss": 0.8121,
+      "step": 25300
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 0.00017461613851682455,
+      "loss": 0.8173,
+      "step": 25325
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.00017379941195687685,
+      "loss": 0.838,
+      "step": 25350
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 0.0001729826853969291,
+      "loss": 0.7872,
+      "step": 25375
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 0.0001721659588369814,
+      "loss": 0.8176,
+      "step": 25400
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 0.00017134923227703363,
+      "loss": 0.8363,
+      "step": 25425
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.00017053250571708592,
+      "loss": 0.8252,
+      "step": 25450
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 0.0001697157791571382,
+      "loss": 0.8278,
+      "step": 25475
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 0.00016889905259719046,
+      "loss": 0.8216,
+      "step": 25500
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.00016808232603724276,
+      "loss": 0.7899,
+      "step": 25525
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.000167265599477295,
+      "loss": 0.7876,
+      "step": 25550
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 0.0001664488729173473,
+      "loss": 0.8127,
+      "step": 25575
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 0.00016563214635739954,
+      "loss": 0.7943,
+      "step": 25600
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.00016481541979745183,
+      "loss": 0.8343,
+      "step": 25625
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 0.00016399869323750407,
+      "loss": 0.7859,
+      "step": 25650
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 0.00016318196667755637,
+      "loss": 0.7914,
+      "step": 25675
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.0001623652401176086,
+      "loss": 0.7867,
+      "step": 25700
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.0001615485135576609,
+      "loss": 0.8232,
+      "step": 25725
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 0.00016073178699771318,
+      "loss": 0.8661,
+      "step": 25750
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 0.00015991506043776545,
+      "loss": 0.7897,
+      "step": 25775
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 0.00015909833387781771,
+      "loss": 0.7948,
+      "step": 25800
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 0.00015828160731786998,
+      "loss": 0.8237,
+      "step": 25825
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 0.00015746488075792225,
+      "loss": 0.7961,
+      "step": 25850
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 0.00015664815419797452,
+      "loss": 0.8382,
+      "step": 25875
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.0001558314276380268,
+      "loss": 0.8154,
+      "step": 25900
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 0.00015501470107807906,
+      "loss": 0.8529,
+      "step": 25925
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 0.00015419797451813133,
+      "loss": 0.8029,
+      "step": 25950
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 0.0001533812479581836,
+      "loss": 0.7945,
+      "step": 25975
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 0.00015256452139823586,
+      "loss": 0.8124,
+      "step": 26000
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 0.00015174779483828816,
+      "loss": 0.8042,
+      "step": 26025
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 0.0001509310682783404,
+      "loss": 0.7905,
+      "step": 26050
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 0.0001501143417183927,
+      "loss": 0.813,
+      "step": 26075
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 0.00014929761515844494,
+      "loss": 0.7823,
+      "step": 26100
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 0.00014848088859849723,
+      "loss": 0.7981,
+      "step": 26125
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 0.0001476968311009474,
+      "loss": 0.7958,
+      "step": 26150
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 0.00014688010454099967,
+      "loss": 0.82,
+      "step": 26175
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.00014606337798105196,
+      "loss": 0.8012,
+      "step": 26200
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.0001452466514211042,
+      "loss": 0.7772,
+      "step": 26225
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 0.0001444299248611565,
+      "loss": 0.8139,
+      "step": 26250
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 0.00014361319830120874,
+      "loss": 0.8207,
+      "step": 26275
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 0.00014279647174126104,
+      "loss": 0.8665,
+      "step": 26300
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.00014197974518131328,
+      "loss": 0.8271,
+      "step": 26325
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.00014116301862136558,
+      "loss": 0.8198,
+      "step": 26350
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 0.00014034629206141782,
+      "loss": 0.7893,
+      "step": 26375
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 0.0001395295655014701,
+      "loss": 0.8025,
+      "step": 26400
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 0.00013871283894152238,
+      "loss": 0.8238,
+      "step": 26425
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 0.00013789611238157465,
+      "loss": 0.7924,
+      "step": 26450
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 0.00013707938582162695,
+      "loss": 0.7637,
+      "step": 26475
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 0.0001362626592616792,
+      "loss": 0.82,
+      "step": 26500
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 0.00013544593270173148,
+      "loss": 0.7799,
+      "step": 26525
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 0.00013462920614178373,
+      "loss": 0.8069,
+      "step": 26550
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 0.00013381247958183602,
+      "loss": 0.8108,
+      "step": 26575
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 0.00013299575302188826,
+      "loss": 0.7857,
+      "step": 26600
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 0.00013217902646194056,
+      "loss": 0.8306,
+      "step": 26625
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 0.0001313622999019928,
+      "loss": 0.7965,
+      "step": 26650
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 0.0001305455733420451,
+      "loss": 0.8243,
+      "step": 26675
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 0.00012972884678209734,
+      "loss": 0.7843,
+      "step": 26700
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 0.00012891212022214963,
+      "loss": 0.7699,
+      "step": 26725
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 0.0001280953936622019,
+      "loss": 0.8122,
+      "step": 26750
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 0.00012727866710225417,
+      "loss": 0.815,
+      "step": 26775
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 0.00012646194054230644,
+      "loss": 0.7969,
+      "step": 26800
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 0.0001256452139823587,
+      "loss": 0.8181,
+      "step": 26825
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 0.00012482848742241098,
+      "loss": 0.8133,
+      "step": 26850
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 0.00012401176086246325,
+      "loss": 0.7942,
+      "step": 26875
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 0.00012319503430251551,
+      "loss": 0.8325,
+      "step": 26900
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 0.00012237830774256778,
+      "loss": 0.818,
+      "step": 26925
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 0.00012156158118262007,
+      "loss": 0.8149,
+      "step": 26950
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 0.00012074485462267233,
+      "loss": 0.8104,
+      "step": 26975
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 0.0001199281280627246,
+      "loss": 0.7673,
+      "step": 27000
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 0.00011911140150277687,
+      "loss": 0.8226,
+      "step": 27025
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 0.00011829467494282914,
+      "loss": 0.7903,
+      "step": 27050
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 0.00011747794838288141,
+      "loss": 0.7812,
+      "step": 27075
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 0.00011666122182293368,
+      "loss": 0.7588,
+      "step": 27100
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 0.00011584449526298596,
+      "loss": 0.7999,
+      "step": 27125
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 0.00011502776870303823,
+      "loss": 0.8294,
+      "step": 27150
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 0.0001142110421430905,
+      "loss": 0.7698,
+      "step": 27175
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 0.00011339431558314277,
+      "loss": 0.8146,
+      "step": 27200
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 0.00011257758902319504,
+      "loss": 0.8294,
+      "step": 27225
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 0.0001117608624632473,
+      "loss": 0.7958,
+      "step": 27250
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 0.00011094413590329957,
+      "loss": 0.8342,
+      "step": 27275
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 0.00011012740934335184,
+      "loss": 0.7979,
+      "step": 27300
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 0.00010931068278340411,
+      "loss": 0.8213,
+      "step": 27325
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 0.00010849395622345638,
+      "loss": 0.7923,
+      "step": 27350
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 0.00010767722966350866,
+      "loss": 0.789,
+      "step": 27375
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 0.00010686050310356093,
+      "loss": 0.8077,
+      "step": 27400
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 0.0001060437765436132,
+      "loss": 0.786,
+      "step": 27425
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 0.00010522704998366547,
+      "loss": 0.7545,
+      "step": 27450
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 0.00010441032342371774,
+      "loss": 0.8159,
+      "step": 27475
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 0.00010359359686377,
+      "loss": 0.7748,
+      "step": 27500
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 0.00010277687030382227,
+      "loss": 0.8255,
+      "step": 27525
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 0.00010196014374387456,
+      "loss": 0.8138,
+      "step": 27550
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 0.00010114341718392683,
+      "loss": 0.8034,
+      "step": 27575
+    },
+    {
+      "epoch": 9.0,
+      "eval_loss": 0.7793058753013611,
+      "eval_runtime": 987.3495,
+      "eval_samples_per_second": 1.013,
+      "eval_steps_per_second": 0.127,
+      "eval_wer": 0.2611209549566353,
+      "step": 27594
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 0.0001003266906239791,
+      "loss": 0.8083,
+      "step": 27600
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 9.950996406403136e-05,
+      "loss": 0.8269,
+      "step": 27625
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 9.869323750408365e-05,
+      "loss": 0.7802,
+      "step": 27650
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 9.787651094413591e-05,
+      "loss": 0.8287,
+      "step": 27675
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 9.705978438418818e-05,
+      "loss": 0.799,
+      "step": 27700
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 9.624305782424045e-05,
+      "loss": 0.8113,
+      "step": 27725
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 9.542633126429272e-05,
+      "loss": 0.7731,
+      "step": 27750
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 9.460960470434499e-05,
+      "loss": 0.8259,
+      "step": 27775
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 9.379287814439726e-05,
+      "loss": 0.7874,
+      "step": 27800
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 9.297615158444953e-05,
+      "loss": 0.7963,
+      "step": 27825
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 9.21594250245018e-05,
+      "loss": 0.8033,
+      "step": 27850
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 9.134269846455406e-05,
+      "loss": 0.7849,
+      "step": 27875
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 9.052597190460633e-05,
+      "loss": 0.7614,
+      "step": 27900
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 8.970924534465862e-05,
+      "loss": 0.8092,
+      "step": 27925
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 8.889251878471088e-05,
+      "loss": 0.8077,
+      "step": 27950
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 8.807579222476315e-05,
+      "loss": 0.745,
+      "step": 27975
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 8.725906566481542e-05,
+      "loss": 0.7826,
+      "step": 28000
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 8.644233910486769e-05,
+      "loss": 0.7996,
+      "step": 28025
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 8.562561254491996e-05,
+      "loss": 0.7794,
+      "step": 28050
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 8.480888598497223e-05,
+      "loss": 0.7683,
+      "step": 28075
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 8.39921594250245e-05,
+      "loss": 0.7636,
+      "step": 28100
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 8.317543286507677e-05,
+      "loss": 0.8064,
+      "step": 28125
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 8.235870630512903e-05,
+      "loss": 0.7704,
+      "step": 28150
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 8.154197974518132e-05,
+      "loss": 0.7917,
+      "step": 28175
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 8.072525318523359e-05,
+      "loss": 0.7519,
+      "step": 28200
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 7.990852662528585e-05,
+      "loss": 0.7986,
+      "step": 28225
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 7.909180006533812e-05,
+      "loss": 0.8103,
+      "step": 28250
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 7.827507350539039e-05,
+      "loss": 0.8082,
+      "step": 28275
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 7.745834694544267e-05,
+      "loss": 0.8105,
+      "step": 28300
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 7.664162038549494e-05,
+      "loss": 0.7817,
+      "step": 28325
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 7.582489382554721e-05,
+      "loss": 0.7758,
+      "step": 28350
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 7.500816726559948e-05,
+      "loss": 0.8288,
+      "step": 28375
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 7.419144070565175e-05,
+      "loss": 0.8038,
+      "step": 28400
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 7.337471414570402e-05,
+      "loss": 0.7801,
+      "step": 28425
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 7.25579875857563e-05,
+      "loss": 0.8335,
+      "step": 28450
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 7.174126102580857e-05,
+      "loss": 0.7795,
+      "step": 28475
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 7.092453446586084e-05,
+      "loss": 0.773,
+      "step": 28500
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 7.01078079059131e-05,
+      "loss": 0.8057,
+      "step": 28525
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 6.929108134596538e-05,
+      "loss": 0.8015,
+      "step": 28550
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 6.847435478601764e-05,
+      "loss": 0.7955,
+      "step": 28575
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 6.765762822606991e-05,
+      "loss": 0.7673,
+      "step": 28600
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 6.684090166612218e-05,
+      "loss": 0.8372,
+      "step": 28625
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 6.602417510617445e-05,
+      "loss": 0.791,
+      "step": 28650
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 6.520744854622672e-05,
+      "loss": 0.7455,
+      "step": 28675
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 6.439072198627899e-05,
+      "loss": 0.8192,
+      "step": 28700
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 6.357399542633127e-05,
+      "loss": 0.7503,
+      "step": 28725
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 6.275726886638354e-05,
+      "loss": 0.7876,
+      "step": 28750
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 6.194054230643581e-05,
+      "loss": 0.7727,
+      "step": 28775
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 6.112381574648808e-05,
+      "loss": 0.7724,
+      "step": 28800
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 6.0307089186540345e-05,
+      "loss": 0.7848,
+      "step": 28825
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.9490362626592614e-05,
+      "loss": 0.7885,
+      "step": 28850
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.867363606664489e-05,
+      "loss": 0.7819,
+      "step": 28875
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.785690950669716e-05,
+      "loss": 0.7699,
+      "step": 28900
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.7040182946749434e-05,
+      "loss": 0.8049,
+      "step": 28925
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.62234563868017e-05,
+      "loss": 0.8224,
+      "step": 28950
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.540672982685397e-05,
+      "loss": 0.7652,
+      "step": 28975
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.459000326690624e-05,
+      "loss": 0.824,
+      "step": 29000
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.377327670695851e-05,
+      "loss": 0.7782,
+      "step": 29025
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.2956550147010785e-05,
+      "loss": 0.7705,
+      "step": 29050
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.213982358706305e-05,
+      "loss": 0.771,
+      "step": 29075
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.132309702711532e-05,
+      "loss": 0.781,
+      "step": 29100
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.050637046716759e-05,
+      "loss": 0.8164,
+      "step": 29125
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.968964390721986e-05,
+      "loss": 0.8296,
+      "step": 29150
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.8872917347272135e-05,
+      "loss": 0.7553,
+      "step": 29175
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.8056190787324404e-05,
+      "loss": 0.8039,
+      "step": 29200
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.723946422737667e-05,
+      "loss": 0.7857,
+      "step": 29225
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.642273766742895e-05,
+      "loss": 0.8134,
+      "step": 29250
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.560601110748122e-05,
+      "loss": 0.7865,
+      "step": 29275
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.4789284547533486e-05,
+      "loss": 0.8112,
+      "step": 29300
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.397255798758576e-05,
+      "loss": 0.8086,
+      "step": 29325
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.315583142763803e-05,
+      "loss": 0.794,
+      "step": 29350
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.23391048676903e-05,
+      "loss": 0.8027,
+      "step": 29375
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.152237830774257e-05,
+      "loss": 0.802,
+      "step": 29400
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.0705651747794836e-05,
+      "loss": 0.7727,
+      "step": 29425
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.988892518784711e-05,
+      "loss": 0.7935,
+      "step": 29450
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.907219862789938e-05,
+      "loss": 0.7929,
+      "step": 29475
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.825547206795165e-05,
+      "loss": 0.7946,
+      "step": 29500
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.743874550800392e-05,
+      "loss": 0.7881,
+      "step": 29525
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.662201894805619e-05,
+      "loss": 0.7713,
+      "step": 29550
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.580529238810846e-05,
+      "loss": 0.814,
+      "step": 29575
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.498856582816073e-05,
+      "loss": 0.7921,
+      "step": 29600
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.417183926821301e-05,
+      "loss": 0.747,
+      "step": 29625
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.3355112708265276e-05,
+      "loss": 0.7977,
+      "step": 29650
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.2538386148317544e-05,
+      "loss": 0.7985,
+      "step": 29675
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.172165958836981e-05,
+      "loss": 0.7921,
+      "step": 29700
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.090493302842208e-05,
+      "loss": 0.7724,
+      "step": 29725
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.0088206468474357e-05,
+      "loss": 0.7708,
+      "step": 29750
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.9271479908526626e-05,
+      "loss": 0.7724,
+      "step": 29775
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 2.8454753348578895e-05,
+      "loss": 0.8304,
+      "step": 29800
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.7638026788631167e-05,
+      "loss": 0.7765,
+      "step": 29825
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.6821300228683436e-05,
+      "loss": 0.8349,
+      "step": 29850
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.6004573668735708e-05,
+      "loss": 0.8128,
+      "step": 29875
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.5187847108787977e-05,
+      "loss": 0.7528,
+      "step": 29900
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.437112054884025e-05,
+      "loss": 0.8442,
+      "step": 29925
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.355439398889252e-05,
+      "loss": 0.7797,
+      "step": 29950
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.273766742894479e-05,
+      "loss": 0.8008,
+      "step": 29975
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.192094086899706e-05,
+      "loss": 0.7834,
+      "step": 30000
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.110421430904933e-05,
+      "loss": 0.774,
+      "step": 30025
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.02874877491016e-05,
+      "loss": 0.8219,
+      "step": 30050
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.9470761189153872e-05,
+      "loss": 0.8094,
+      "step": 30075
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.8654034629206144e-05,
+      "loss": 0.7588,
+      "step": 30100
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.7837308069258413e-05,
+      "loss": 0.7986,
+      "step": 30125
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.7020581509310685e-05,
+      "loss": 0.7798,
+      "step": 30150
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.6203854949362954e-05,
+      "loss": 0.8109,
+      "step": 30175
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.5387128389415226e-05,
+      "loss": 0.809,
+      "step": 30200
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.4570401829467494e-05,
+      "loss": 0.8146,
+      "step": 30225
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.3753675269519765e-05,
+      "loss": 0.7787,
+      "step": 30250
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.2936948709572035e-05,
+      "loss": 0.7642,
+      "step": 30275
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.2120222149624306e-05,
+      "loss": 0.8149,
+      "step": 30300
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.1303495589676576e-05,
+      "loss": 0.7651,
+      "step": 30325
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.0486769029728847e-05,
+      "loss": 0.8262,
+      "step": 30350
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 9.670042469781117e-06,
+      "loss": 0.8193,
+      "step": 30375
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 8.853315909833388e-06,
+      "loss": 0.7813,
+      "step": 30400
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 8.03658934988566e-06,
+      "loss": 0.7949,
+      "step": 30425
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 7.2198627899379294e-06,
+      "loss": 0.8129,
+      "step": 30450
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 6.403136229990199e-06,
+      "loss": 0.833,
+      "step": 30475
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.58640967004247e-06,
+      "loss": 0.7824,
+      "step": 30500
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.76968311009474e-06,
+      "loss": 0.7823,
+      "step": 30525
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 3.952956550147011e-06,
+      "loss": 0.8411,
+      "step": 30550
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.1362299901992813e-06,
+      "loss": 0.8127,
+      "step": 30575
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.3195034302515518e-06,
+      "loss": 0.8256,
+      "step": 30600
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.5027768703038222e-06,
+      "loss": 0.7862,
+      "step": 30625
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 6.860503103560928e-07,
+      "loss": 0.7774,
+      "step": 30650
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 0.7750564813613892,
+      "eval_runtime": 942.8578,
+      "eval_samples_per_second": 1.061,
+      "eval_steps_per_second": 0.133,
+      "eval_wer": 0.11190898069570084,
+      "step": 30660
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 30660,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 2.636131979132928e+20,
+  "trial_name": null,
+  "trial_params": null
+}