diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14433 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.6899028305872412,
+  "eval_steps": 100,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0008449514152936206,
+      "grad_norm": 1.3914175033569336,
+      "learning_rate": 4.99991550485847e-05,
+      "loss": 0.0856,
+      "step": 2
+    },
+    {
+      "epoch": 0.0016899028305872412,
+      "grad_norm": 1.3959804773330688,
+      "learning_rate": 4.999831009716942e-05,
+      "loss": 0.0982,
+      "step": 4
+    },
+    {
+      "epoch": 0.0025348542458808617,
+      "grad_norm": 1.4670644998550415,
+      "learning_rate": 4.999746514575412e-05,
+      "loss": 0.0918,
+      "step": 6
+    },
+    {
+      "epoch": 0.0033798056611744824,
+      "grad_norm": 2.241302728652954,
+      "learning_rate": 4.999662019433883e-05,
+      "loss": 0.1214,
+      "step": 8
+    },
+    {
+      "epoch": 0.004224757076468103,
+      "grad_norm": 2.0347797870635986,
+      "learning_rate": 4.999577524292353e-05,
+      "loss": 0.1111,
+      "step": 10
+    },
+    {
+      "epoch": 0.005069708491761723,
+      "grad_norm": 1.1951134204864502,
+      "learning_rate": 4.999493029150824e-05,
+      "loss": 0.0817,
+      "step": 12
+    },
+    {
+      "epoch": 0.005914659907055344,
+      "grad_norm": 1.2443560361862183,
+      "learning_rate": 4.9994085340092946e-05,
+      "loss": 0.0903,
+      "step": 14
+    },
+    {
+      "epoch": 0.006759611322348965,
+      "grad_norm": 1.4616291522979736,
+      "learning_rate": 4.999324038867765e-05,
+      "loss": 0.0968,
+      "step": 16
+    },
+    {
+      "epoch": 0.0076045627376425855,
+      "grad_norm": 1.999237298965454,
+      "learning_rate": 4.999239543726236e-05,
+      "loss": 0.1221,
+      "step": 18
+    },
+    {
+      "epoch": 0.008449514152936205,
+      "grad_norm": 1.4877630472183228,
+      "learning_rate": 4.999155048584707e-05,
+      "loss": 0.1137,
+      "step": 20
+    },
+    {
+      "epoch": 0.009294465568229826,
+      "grad_norm": 1.4436724185943604,
+      "learning_rate": 4.999070553443177e-05,
+      "loss": 0.1129,
+      "step": 22
+    },
+    {
+      "epoch": 0.010139416983523447,
+      "grad_norm": 2.1934313774108887,
+      "learning_rate": 4.998986058301648e-05,
+      "loss": 0.1052,
+      "step": 24
+    },
+    {
+      "epoch": 0.010984368398817067,
+      "grad_norm": 1.8377941846847534,
+      "learning_rate": 4.998901563160118e-05,
+      "loss": 0.1381,
+      "step": 26
+    },
+    {
+      "epoch": 0.011829319814110688,
+      "grad_norm": 2.0095016956329346,
+      "learning_rate": 4.998817068018589e-05,
+      "loss": 0.0852,
+      "step": 28
+    },
+    {
+      "epoch": 0.012674271229404309,
+      "grad_norm": 3.2389862537384033,
+      "learning_rate": 4.99873257287706e-05,
+      "loss": 0.1161,
+      "step": 30
+    },
+    {
+      "epoch": 0.01351922264469793,
+      "grad_norm": 2.737915277481079,
+      "learning_rate": 4.9986480777355304e-05,
+      "loss": 0.1713,
+      "step": 32
+    },
+    {
+      "epoch": 0.01436417405999155,
+      "grad_norm": 1.6804777383804321,
+      "learning_rate": 4.998563582594001e-05,
+      "loss": 0.1234,
+      "step": 34
+    },
+    {
+      "epoch": 0.015209125475285171,
+      "grad_norm": 1.8924976587295532,
+      "learning_rate": 4.998479087452472e-05,
+      "loss": 0.0826,
+      "step": 36
+    },
+    {
+      "epoch": 0.01605407689057879,
+      "grad_norm": 1.8899160623550415,
+      "learning_rate": 4.998394592310942e-05,
+      "loss": 0.1116,
+      "step": 38
+    },
+    {
+      "epoch": 0.01689902830587241,
+      "grad_norm": 1.7245818376541138,
+      "learning_rate": 4.998310097169413e-05,
+      "loss": 0.0863,
+      "step": 40
+    },
+    {
+      "epoch": 0.017743979721166033,
+      "grad_norm": 1.1223626136779785,
+      "learning_rate": 4.998225602027883e-05,
+      "loss": 0.0823,
+      "step": 42
+    },
+    {
+      "epoch": 0.018588931136459652,
+      "grad_norm": 2.4463398456573486,
+      "learning_rate": 4.998141106886354e-05,
+      "loss": 0.1465,
+      "step": 44
+    },
+    {
+      "epoch": 0.019433882551753275,
+      "grad_norm": 1.4499152898788452,
+      "learning_rate": 4.998056611744825e-05,
+      "loss": 0.1043,
+      "step": 46
+    },
+    {
+      "epoch": 0.020278833967046894,
+      "grad_norm": 2.1130013465881348,
+      "learning_rate": 4.9979721166032955e-05,
+      "loss": 0.153,
+      "step": 48
+    },
+    {
+      "epoch": 0.021123785382340516,
+      "grad_norm": 1.4567383527755737,
+      "learning_rate": 4.997887621461766e-05,
+      "loss": 0.1481,
+      "step": 50
+    },
+    {
+      "epoch": 0.021968736797634135,
+      "grad_norm": 1.4559706449508667,
+      "learning_rate": 4.997803126320237e-05,
+      "loss": 0.0989,
+      "step": 52
+    },
+    {
+      "epoch": 0.022813688212927757,
+      "grad_norm": 1.9101276397705078,
+      "learning_rate": 4.997718631178707e-05,
+      "loss": 0.1458,
+      "step": 54
+    },
+    {
+      "epoch": 0.023658639628221376,
+      "grad_norm": 1.50411856174469,
+      "learning_rate": 4.9976341360371784e-05,
+      "loss": 0.0991,
+      "step": 56
+    },
+    {
+      "epoch": 0.024503591043515,
+      "grad_norm": 1.0242832899093628,
+      "learning_rate": 4.9975496408956484e-05,
+      "loss": 0.0766,
+      "step": 58
+    },
+    {
+      "epoch": 0.025348542458808618,
+      "grad_norm": 2.2057089805603027,
+      "learning_rate": 4.997465145754119e-05,
+      "loss": 0.1176,
+      "step": 60
+    },
+    {
+      "epoch": 0.02619349387410224,
+      "grad_norm": 1.3259694576263428,
+      "learning_rate": 4.99738065061259e-05,
+      "loss": 0.1631,
+      "step": 62
+    },
+    {
+      "epoch": 0.02703844528939586,
+      "grad_norm": 1.2836776971817017,
+      "learning_rate": 4.9972961554710606e-05,
+      "loss": 0.08,
+      "step": 64
+    },
+    {
+      "epoch": 0.02788339670468948,
+      "grad_norm": 1.250314712524414,
+      "learning_rate": 4.997211660329531e-05,
+      "loss": 0.0944,
+      "step": 66
+    },
+    {
+      "epoch": 0.0287283481199831,
+      "grad_norm": 1.753688097000122,
+      "learning_rate": 4.997127165188002e-05,
+      "loss": 0.133,
+      "step": 68
+    },
+    {
+      "epoch": 0.029573299535276723,
+      "grad_norm": 1.5159249305725098,
+      "learning_rate": 4.997042670046472e-05,
+      "loss": 0.0975,
+      "step": 70
+    },
+    {
+      "epoch": 0.030418250950570342,
+      "grad_norm": 1.78278386592865,
+      "learning_rate": 4.9969581749049434e-05,
+      "loss": 0.1093,
+      "step": 72
+    },
+    {
+      "epoch": 0.03126320236586396,
+      "grad_norm": 1.725652813911438,
+      "learning_rate": 4.9968736797634135e-05,
+      "loss": 0.1074,
+      "step": 74
+    },
+    {
+      "epoch": 0.03210815378115758,
+      "grad_norm": 1.6478575468063354,
+      "learning_rate": 4.996789184621884e-05,
+      "loss": 0.1279,
+      "step": 76
+    },
+    {
+      "epoch": 0.032953105196451206,
+      "grad_norm": 1.2018022537231445,
+      "learning_rate": 4.996704689480355e-05,
+      "loss": 0.0749,
+      "step": 78
+    },
+    {
+      "epoch": 0.03379805661174482,
+      "grad_norm": 0.8992096185684204,
+      "learning_rate": 4.9966201943388256e-05,
+      "loss": 0.0656,
+      "step": 80
+    },
+    {
+      "epoch": 0.034643008027038444,
+      "grad_norm": 2.0089211463928223,
+      "learning_rate": 4.9965356991972964e-05,
+      "loss": 0.1403,
+      "step": 82
+    },
+    {
+      "epoch": 0.035487959442332066,
+      "grad_norm": 1.9890964031219482,
+      "learning_rate": 4.996451204055767e-05,
+      "loss": 0.0917,
+      "step": 84
+    },
+    {
+      "epoch": 0.03633291085762569,
+      "grad_norm": 1.6564279794692993,
+      "learning_rate": 4.996366708914237e-05,
+      "loss": 0.127,
+      "step": 86
+    },
+    {
+      "epoch": 0.037177862272919304,
+      "grad_norm": 2.2462258338928223,
+      "learning_rate": 4.9962822137727085e-05,
+      "loss": 0.1866,
+      "step": 88
+    },
+    {
+      "epoch": 0.03802281368821293,
+      "grad_norm": 2.442894697189331,
+      "learning_rate": 4.9961977186311786e-05,
+      "loss": 0.096,
+      "step": 90
+    },
+    {
+      "epoch": 0.03886776510350655,
+      "grad_norm": 1.7473067045211792,
+      "learning_rate": 4.996113223489649e-05,
+      "loss": 0.1065,
+      "step": 92
+    },
+    {
+      "epoch": 0.03971271651880017,
+      "grad_norm": 1.4837480783462524,
+      "learning_rate": 4.99602872834812e-05,
+      "loss": 0.0878,
+      "step": 94
+    },
+    {
+      "epoch": 0.04055766793409379,
+      "grad_norm": 0.9949487447738647,
+      "learning_rate": 4.995944233206591e-05,
+      "loss": 0.0795,
+      "step": 96
+    },
+    {
+      "epoch": 0.04140261934938741,
+      "grad_norm": 1.6431764364242554,
+      "learning_rate": 4.9958597380650614e-05,
+      "loss": 0.1184,
+      "step": 98
+    },
+    {
+      "epoch": 0.04224757076468103,
+      "grad_norm": 1.0905040502548218,
+      "learning_rate": 4.995775242923532e-05,
+      "loss": 0.086,
+      "step": 100
+    },
+    {
+      "epoch": 0.04224757076468103,
+      "eval_accuracy": 0.6362303845490602,
+      "eval_cer": 0.11421438464320686,
+      "eval_loss": 0.22910405695438385,
+      "eval_runtime": 887.866,
+      "eval_samples_per_second": 13.063,
+      "eval_steps_per_second": 0.409,
+      "step": 100
+    },
+    {
+      "epoch": 0.043092522179974654,
+      "grad_norm": 1.3749598264694214,
+      "learning_rate": 4.995690747782002e-05,
+      "loss": 0.1197,
+      "step": 102
+    },
+    {
+      "epoch": 0.04393747359526827,
+      "grad_norm": 1.572817325592041,
+      "learning_rate": 4.9956062526404736e-05,
+      "loss": 0.0924,
+      "step": 104
+    },
+    {
+      "epoch": 0.04478242501056189,
+      "grad_norm": 1.5053339004516602,
+      "learning_rate": 4.9955217574989436e-05,
+      "loss": 0.0926,
+      "step": 106
+    },
+    {
+      "epoch": 0.045627376425855515,
+      "grad_norm": 1.816696047782898,
+      "learning_rate": 4.995437262357415e-05,
+      "loss": 0.143,
+      "step": 108
+    },
+    {
+      "epoch": 0.04647232784114914,
+      "grad_norm": 1.8459056615829468,
+      "learning_rate": 4.995352767215885e-05,
+      "loss": 0.1229,
+      "step": 110
+    },
+    {
+      "epoch": 0.04731727925644275,
+      "grad_norm": 1.3267372846603394,
+      "learning_rate": 4.995268272074356e-05,
+      "loss": 0.0874,
+      "step": 112
+    },
+    {
+      "epoch": 0.048162230671736375,
+      "grad_norm": 1.7917542457580566,
+      "learning_rate": 4.9951837769328265e-05,
+      "loss": 0.1485,
+      "step": 114
+    },
+    {
+      "epoch": 0.04900718208703,
+      "grad_norm": 1.7345539331436157,
+      "learning_rate": 4.995099281791297e-05,
+      "loss": 0.09,
+      "step": 116
+    },
+    {
+      "epoch": 0.04985213350232361,
+      "grad_norm": 1.5779085159301758,
+      "learning_rate": 4.995014786649768e-05,
+      "loss": 0.1262,
+      "step": 118
+    },
+    {
+      "epoch": 0.050697084917617236,
+      "grad_norm": 1.391013503074646,
+      "learning_rate": 4.994930291508239e-05,
+      "loss": 0.1145,
+      "step": 120
+    },
+    {
+      "epoch": 0.05154203633291086,
+      "grad_norm": 1.6034536361694336,
+      "learning_rate": 4.994845796366709e-05,
+      "loss": 0.0881,
+      "step": 122
+    },
+    {
+      "epoch": 0.05238698774820448,
+      "grad_norm": 1.5294889211654663,
+      "learning_rate": 4.99476130122518e-05,
+      "loss": 0.0793,
+      "step": 124
+    },
+    {
+      "epoch": 0.053231939163498096,
+      "grad_norm": 1.197351336479187,
+      "learning_rate": 4.99467680608365e-05,
+      "loss": 0.091,
+      "step": 126
+    },
+    {
+      "epoch": 0.05407689057879172,
+      "grad_norm": 1.323849081993103,
+      "learning_rate": 4.994592310942121e-05,
+      "loss": 0.0888,
+      "step": 128
+    },
+    {
+      "epoch": 0.05492184199408534,
+      "grad_norm": 1.7171475887298584,
+      "learning_rate": 4.9945078158005916e-05,
+      "loss": 0.1436,
+      "step": 130
+    },
+    {
+      "epoch": 0.05576679340937896,
+      "grad_norm": 1.7138563394546509,
+      "learning_rate": 4.994423320659062e-05,
+      "loss": 0.09,
+      "step": 132
+    },
+    {
+      "epoch": 0.05661174482467258,
+      "grad_norm": 1.6496665477752686,
+      "learning_rate": 4.994338825517533e-05,
+      "loss": 0.1415,
+      "step": 134
+    },
+    {
+      "epoch": 0.0574566962399662,
+      "grad_norm": 1.3270729780197144,
+      "learning_rate": 4.994254330376004e-05,
+      "loss": 0.1073,
+      "step": 136
+    },
+    {
+      "epoch": 0.058301647655259824,
+      "grad_norm": 1.2084906101226807,
+      "learning_rate": 4.994169835234474e-05,
+      "loss": 0.0861,
+      "step": 138
+    },
+    {
+      "epoch": 0.059146599070553446,
+      "grad_norm": 1.885323405265808,
+      "learning_rate": 4.994085340092945e-05,
+      "loss": 0.138,
+      "step": 140
+    },
+    {
+      "epoch": 0.05999155048584706,
+      "grad_norm": 1.3070194721221924,
+      "learning_rate": 4.994000844951415e-05,
+      "loss": 0.0725,
+      "step": 142
+    },
+    {
+      "epoch": 0.060836501901140684,
+      "grad_norm": 1.494089961051941,
+      "learning_rate": 4.993916349809886e-05,
+      "loss": 0.1076,
+      "step": 144
+    },
+    {
+      "epoch": 0.061681453316434307,
+      "grad_norm": 1.5598512887954712,
+      "learning_rate": 4.993831854668357e-05,
+      "loss": 0.1068,
+      "step": 146
+    },
+    {
+      "epoch": 0.06252640473172792,
+      "grad_norm": 2.4986348152160645,
+      "learning_rate": 4.9937473595268274e-05,
+      "loss": 0.1439,
+      "step": 148
+    },
+    {
+      "epoch": 0.06337135614702155,
+      "grad_norm": 1.6942882537841797,
+      "learning_rate": 4.993662864385298e-05,
+      "loss": 0.0694,
+      "step": 150
+    },
+    {
+      "epoch": 0.06421630756231517,
+      "grad_norm": 1.8440594673156738,
+      "learning_rate": 4.993578369243769e-05,
+      "loss": 0.1387,
+      "step": 152
+    },
+    {
+      "epoch": 0.06506125897760878,
+      "grad_norm": 1.647815227508545,
+      "learning_rate": 4.993493874102239e-05,
+      "loss": 0.0997,
+      "step": 154
+    },
+    {
+      "epoch": 0.06590621039290241,
+      "grad_norm": 1.5283992290496826,
+      "learning_rate": 4.99340937896071e-05,
+      "loss": 0.0803,
+      "step": 156
+    },
+    {
+      "epoch": 0.06675116180819603,
+      "grad_norm": 1.7466899156570435,
+      "learning_rate": 4.99332488381918e-05,
+      "loss": 0.1309,
+      "step": 158
+    },
+    {
+      "epoch": 0.06759611322348964,
+      "grad_norm": 1.4426119327545166,
+      "learning_rate": 4.993240388677651e-05,
+      "loss": 0.107,
+      "step": 160
+    },
+    {
+      "epoch": 0.06844106463878327,
+      "grad_norm": 1.6664427518844604,
+      "learning_rate": 4.993155893536122e-05,
+      "loss": 0.1763,
+      "step": 162
+    },
+    {
+      "epoch": 0.06928601605407689,
+      "grad_norm": 1.8783625364303589,
+      "learning_rate": 4.9930713983945925e-05,
+      "loss": 0.0942,
+      "step": 164
+    },
+    {
+      "epoch": 0.07013096746937052,
+      "grad_norm": 1.3738420009613037,
+      "learning_rate": 4.992986903253063e-05,
+      "loss": 0.1128,
+      "step": 166
+    },
+    {
+      "epoch": 0.07097591888466413,
+      "grad_norm": 1.2888435125350952,
+      "learning_rate": 4.992902408111534e-05,
+      "loss": 0.08,
+      "step": 168
+    },
+    {
+      "epoch": 0.07182087029995775,
+      "grad_norm": 0.921699047088623,
+      "learning_rate": 4.992817912970004e-05,
+      "loss": 0.1028,
+      "step": 170
+    },
+    {
+      "epoch": 0.07266582171525138,
+      "grad_norm": 1.8012775182724,
+      "learning_rate": 4.9927334178284754e-05,
+      "loss": 0.1012,
+      "step": 172
+    },
+    {
+      "epoch": 0.07351077313054499,
+      "grad_norm": 1.061102271080017,
+      "learning_rate": 4.9926489226869454e-05,
+      "loss": 0.1341,
+      "step": 174
+    },
+    {
+      "epoch": 0.07435572454583861,
+      "grad_norm": 1.1755284070968628,
+      "learning_rate": 4.992564427545416e-05,
+      "loss": 0.0944,
+      "step": 176
+    },
+    {
+      "epoch": 0.07520067596113224,
+      "grad_norm": 1.4010897874832153,
+      "learning_rate": 4.992479932403887e-05,
+      "loss": 0.0736,
+      "step": 178
+    },
+    {
+      "epoch": 0.07604562737642585,
+      "grad_norm": 1.391518235206604,
+      "learning_rate": 4.9923954372623576e-05,
+      "loss": 0.0622,
+      "step": 180
+    },
+    {
+      "epoch": 0.07689057879171948,
+      "grad_norm": 1.2992569208145142,
+      "learning_rate": 4.992310942120828e-05,
+      "loss": 0.0982,
+      "step": 182
+    },
+    {
+      "epoch": 0.0777355302070131,
+      "grad_norm": 1.4223698377609253,
+      "learning_rate": 4.992226446979299e-05,
+      "loss": 0.098,
+      "step": 184
+    },
+    {
+      "epoch": 0.07858048162230671,
+      "grad_norm": 1.5545382499694824,
+      "learning_rate": 4.992141951837769e-05,
+      "loss": 0.171,
+      "step": 186
+    },
+    {
+      "epoch": 0.07942543303760034,
+      "grad_norm": 2.143627882003784,
+      "learning_rate": 4.9920574566962404e-05,
+      "loss": 0.1158,
+      "step": 188
+    },
+    {
+      "epoch": 0.08027038445289396,
+      "grad_norm": 1.5633560419082642,
+      "learning_rate": 4.9919729615547105e-05,
+      "loss": 0.0865,
+      "step": 190
+    },
+    {
+      "epoch": 0.08111533586818757,
+      "grad_norm": 1.6810214519500732,
+      "learning_rate": 4.991888466413181e-05,
+      "loss": 0.0905,
+      "step": 192
+    },
+    {
+      "epoch": 0.0819602872834812,
+      "grad_norm": 1.0320245027542114,
+      "learning_rate": 4.991803971271652e-05,
+      "loss": 0.0748,
+      "step": 194
+    },
+    {
+      "epoch": 0.08280523869877482,
+      "grad_norm": 1.9010602235794067,
+      "learning_rate": 4.9917194761301226e-05,
+      "loss": 0.1433,
+      "step": 196
+    },
+    {
+      "epoch": 0.08365019011406843,
+      "grad_norm": 2.0801427364349365,
+      "learning_rate": 4.9916349809885934e-05,
+      "loss": 0.1286,
+      "step": 198
+    },
+    {
+      "epoch": 0.08449514152936206,
+      "grad_norm": 1.6263489723205566,
+      "learning_rate": 4.991550485847064e-05,
+      "loss": 0.097,
+      "step": 200
+    },
+    {
+      "epoch": 0.08449514152936206,
+      "eval_accuracy": 0.6549405069839628,
+      "eval_cer": 0.10543680584348562,
+      "eval_loss": 0.21902646124362946,
+      "eval_runtime": 857.1192,
+      "eval_samples_per_second": 13.531,
+      "eval_steps_per_second": 0.424,
+      "step": 200
+    },
+    {
+      "epoch": 0.08534009294465568,
+      "grad_norm": 1.3326414823532104,
+      "learning_rate": 4.991465990705534e-05,
+      "loss": 0.1085,
+      "step": 202
+    },
+    {
+      "epoch": 0.08618504435994931,
+      "grad_norm": 1.6693871021270752,
+      "learning_rate": 4.9913814955640055e-05,
+      "loss": 0.1202,
+      "step": 204
+    },
+    {
+      "epoch": 0.08702999577524292,
+      "grad_norm": 1.5326752662658691,
+      "learning_rate": 4.9912970004224756e-05,
+      "loss": 0.1062,
+      "step": 206
+    },
+    {
+      "epoch": 0.08787494719053654,
+      "grad_norm": 2.1729543209075928,
+      "learning_rate": 4.991212505280947e-05,
+      "loss": 0.1443,
+      "step": 208
+    },
+    {
+      "epoch": 0.08871989860583017,
+      "grad_norm": 1.343558669090271,
+      "learning_rate": 4.991128010139417e-05,
+      "loss": 0.1014,
+      "step": 210
+    },
+    {
+      "epoch": 0.08956485002112378,
+      "grad_norm": 1.2149333953857422,
+      "learning_rate": 4.991043514997888e-05,
+      "loss": 0.09,
+      "step": 212
+    },
+    {
+      "epoch": 0.0904098014364174,
+      "grad_norm": 1.063955307006836,
+      "learning_rate": 4.9909590198563584e-05,
+      "loss": 0.1024,
+      "step": 214
+    },
+    {
+      "epoch": 0.09125475285171103,
+      "grad_norm": 1.5514568090438843,
+      "learning_rate": 4.990874524714829e-05,
+      "loss": 0.0901,
+      "step": 216
+    },
+    {
+      "epoch": 0.09209970426700465,
+      "grad_norm": 1.8420741558074951,
+      "learning_rate": 4.9907900295733e-05,
+      "loss": 0.1084,
+      "step": 218
+    },
+    {
+      "epoch": 0.09294465568229827,
+      "grad_norm": 1.7071248292922974,
+      "learning_rate": 4.9907055344317706e-05,
+      "loss": 0.0838,
+      "step": 220
+    },
+    {
+      "epoch": 0.09378960709759189,
+      "grad_norm": 1.5390372276306152,
+      "learning_rate": 4.9906210392902406e-05,
+      "loss": 0.1133,
+      "step": 222
+    },
+    {
+      "epoch": 0.0946345585128855,
+      "grad_norm": 1.734567403793335,
+      "learning_rate": 4.990536544148712e-05,
+      "loss": 0.0862,
+      "step": 224
+    },
+    {
+      "epoch": 0.09547950992817913,
+      "grad_norm": 0.9405034780502319,
+      "learning_rate": 4.990452049007182e-05,
+      "loss": 0.0941,
+      "step": 226
+    },
+    {
+      "epoch": 0.09632446134347275,
+      "grad_norm": 1.2817813158035278,
+      "learning_rate": 4.990367553865653e-05,
+      "loss": 0.073,
+      "step": 228
+    },
+    {
+      "epoch": 0.09716941275876637,
+      "grad_norm": 2.075343132019043,
+      "learning_rate": 4.9902830587241235e-05,
+      "loss": 0.132,
+      "step": 230
+    },
+    {
+      "epoch": 0.09801436417406,
+      "grad_norm": 1.2112518548965454,
+      "learning_rate": 4.990198563582594e-05,
+      "loss": 0.0918,
+      "step": 232
+    },
+    {
+      "epoch": 0.09885931558935361,
+      "grad_norm": 0.846361517906189,
+      "learning_rate": 4.990114068441065e-05,
+      "loss": 0.063,
+      "step": 234
+    },
+    {
+      "epoch": 0.09970426700464723,
+      "grad_norm": 1.6504607200622559,
+      "learning_rate": 4.990029573299536e-05,
+      "loss": 0.0915,
+      "step": 236
+    },
+    {
+      "epoch": 0.10054921841994086,
+      "grad_norm": 1.5142329931259155,
+      "learning_rate": 4.989945078158006e-05,
+      "loss": 0.1099,
+      "step": 238
+    },
+    {
+      "epoch": 0.10139416983523447,
+      "grad_norm": 2.2613418102264404,
+      "learning_rate": 4.989860583016477e-05,
+      "loss": 0.1245,
+      "step": 240
+    },
+    {
+      "epoch": 0.1022391212505281,
+      "grad_norm": 2.134218454360962,
+      "learning_rate": 4.989776087874947e-05,
+      "loss": 0.1116,
+      "step": 242
+    },
+    {
+      "epoch": 0.10308407266582172,
+      "grad_norm": 1.6032145023345947,
+      "learning_rate": 4.989691592733418e-05,
+      "loss": 0.085,
+      "step": 244
+    },
+    {
+      "epoch": 0.10392902408111533,
+      "grad_norm": 1.2227692604064941,
+      "learning_rate": 4.9896070975918886e-05,
+      "loss": 0.0886,
+      "step": 246
+    },
+    {
+      "epoch": 0.10477397549640896,
+      "grad_norm": 2.2352206707000732,
+      "learning_rate": 4.989522602450359e-05,
+      "loss": 0.1255,
+      "step": 248
+    },
+    {
+      "epoch": 0.10561892691170258,
+      "grad_norm": 0.9914106726646423,
+      "learning_rate": 4.98943810730883e-05,
+      "loss": 0.0814,
+      "step": 250
+    },
+    {
+      "epoch": 0.10646387832699619,
+      "grad_norm": 2.0763142108917236,
+      "learning_rate": 4.989353612167301e-05,
+      "loss": 0.0949,
+      "step": 252
+    },
+    {
+      "epoch": 0.10730882974228982,
+      "grad_norm": 1.0693895816802979,
+      "learning_rate": 4.989269117025771e-05,
+      "loss": 0.0659,
+      "step": 254
+    },
+    {
+      "epoch": 0.10815378115758344,
+      "grad_norm": 1.8804770708084106,
+      "learning_rate": 4.989184621884242e-05,
+      "loss": 0.1019,
+      "step": 256
+    },
+    {
+      "epoch": 0.10899873257287707,
+      "grad_norm": 1.8545231819152832,
+      "learning_rate": 4.989100126742712e-05,
+      "loss": 0.1008,
+      "step": 258
+    },
+    {
+      "epoch": 0.10984368398817068,
+      "grad_norm": 1.1952356100082397,
+      "learning_rate": 4.989015631601183e-05,
+      "loss": 0.1363,
+      "step": 260
+    },
+    {
+      "epoch": 0.1106886354034643,
+      "grad_norm": 1.715725064277649,
+      "learning_rate": 4.988931136459654e-05,
+      "loss": 0.1249,
+      "step": 262
+    },
+    {
+      "epoch": 0.11153358681875793,
+      "grad_norm": 1.5759917497634888,
+      "learning_rate": 4.9888466413181244e-05,
+      "loss": 0.0784,
+      "step": 264
+    },
+    {
+      "epoch": 0.11237853823405154,
+      "grad_norm": 1.252929925918579,
+      "learning_rate": 4.988762146176595e-05,
+      "loss": 0.073,
+      "step": 266
+    },
+    {
+      "epoch": 0.11322348964934516,
+      "grad_norm": 1.0508784055709839,
+      "learning_rate": 4.988677651035066e-05,
+      "loss": 0.0835,
+      "step": 268
+    },
+    {
+      "epoch": 0.11406844106463879,
+      "grad_norm": 1.2772610187530518,
+      "learning_rate": 4.988593155893536e-05,
+      "loss": 0.0727,
+      "step": 270
+    },
+    {
+      "epoch": 0.1149133924799324,
+      "grad_norm": 1.5705053806304932,
+      "learning_rate": 4.988508660752007e-05,
+      "loss": 0.1377,
+      "step": 272
+    },
+    {
+      "epoch": 0.11575834389522602,
+      "grad_norm": 2.1583268642425537,
+      "learning_rate": 4.988424165610477e-05,
+      "loss": 0.1126,
+      "step": 274
+    },
+    {
+      "epoch": 0.11660329531051965,
+      "grad_norm": 1.3954452276229858,
+      "learning_rate": 4.988339670468948e-05,
+      "loss": 0.0855,
+      "step": 276
+    },
+    {
+      "epoch": 0.11744824672581326,
+      "grad_norm": 2.0817723274230957,
+      "learning_rate": 4.988255175327419e-05,
+      "loss": 0.1132,
+      "step": 278
+    },
+    {
+      "epoch": 0.11829319814110689,
+      "grad_norm": 1.833322525024414,
+      "learning_rate": 4.9881706801858895e-05,
+      "loss": 0.1117,
+      "step": 280
+    },
+    {
+      "epoch": 0.11913814955640051,
+      "grad_norm": 1.5989530086517334,
+      "learning_rate": 4.98808618504436e-05,
+      "loss": 0.1265,
+      "step": 282
+    },
+    {
+      "epoch": 0.11998310097169412,
+      "grad_norm": 1.8421043157577515,
+      "learning_rate": 4.988001689902831e-05,
+      "loss": 0.1372,
+      "step": 284
+    },
+    {
+      "epoch": 0.12082805238698775,
+      "grad_norm": 1.897548794746399,
+      "learning_rate": 4.987917194761301e-05,
+      "loss": 0.1103,
+      "step": 286
+    },
+    {
+      "epoch": 0.12167300380228137,
+      "grad_norm": 1.4291915893554688,
+      "learning_rate": 4.9878326996197724e-05,
+      "loss": 0.0847,
+      "step": 288
+    },
+    {
+      "epoch": 0.12251795521757498,
+      "grad_norm": 1.1558469533920288,
+      "learning_rate": 4.9877482044782424e-05,
+      "loss": 0.0889,
+      "step": 290
+    },
+    {
+      "epoch": 0.12336290663286861,
+      "grad_norm": 1.7630038261413574,
+      "learning_rate": 4.987663709336713e-05,
+      "loss": 0.126,
+      "step": 292
+    },
+    {
+      "epoch": 0.12420785804816223,
+      "grad_norm": 1.0064209699630737,
+      "learning_rate": 4.987579214195184e-05,
+      "loss": 0.086,
+      "step": 294
+    },
+    {
+      "epoch": 0.12505280946345584,
+      "grad_norm": 1.59539794921875,
+      "learning_rate": 4.9874947190536546e-05,
+      "loss": 0.1184,
+      "step": 296
+    },
+    {
+      "epoch": 0.12589776087874946,
+      "grad_norm": 1.7932195663452148,
+      "learning_rate": 4.987410223912125e-05,
+      "loss": 0.1376,
+      "step": 298
+    },
+    {
+      "epoch": 0.1267427122940431,
+      "grad_norm": 1.8148212432861328,
+      "learning_rate": 4.987325728770596e-05,
+      "loss": 0.081,
+      "step": 300
+    },
+    {
+      "epoch": 0.1267427122940431,
+      "eval_accuracy": 0.6697706501120883,
+      "eval_cer": 0.096782730665067,
+      "eval_loss": 0.2164023518562317,
+      "eval_runtime": 849.0391,
+      "eval_samples_per_second": 13.66,
+      "eval_steps_per_second": 0.428,
+      "step": 300
+    },
+    {
+      "epoch": 0.12758766370933672,
+      "grad_norm": 1.354212760925293,
+      "learning_rate": 4.987241233629066e-05,
+      "loss": 0.1006,
+      "step": 302
+    },
+    {
+      "epoch": 0.12843261512463033,
+      "grad_norm": 2.113490104675293,
+      "learning_rate": 4.9871567384875374e-05,
+      "loss": 0.1404,
+      "step": 304
+    },
+    {
+      "epoch": 0.12927756653992395,
+      "grad_norm": 1.6982043981552124,
+      "learning_rate": 4.9870722433460075e-05,
+      "loss": 0.2001,
+      "step": 306
+    },
+    {
+      "epoch": 0.13012251795521756,
+      "grad_norm": 1.6756244897842407,
+      "learning_rate": 4.986987748204479e-05,
+      "loss": 0.085,
+      "step": 308
+    },
+    {
+      "epoch": 0.1309674693705112,
+      "grad_norm": 1.2098358869552612,
+      "learning_rate": 4.986903253062949e-05,
+      "loss": 0.0603,
+      "step": 310
+    },
+    {
+      "epoch": 0.13181242078580482,
+      "grad_norm": 1.7302310466766357,
+      "learning_rate": 4.9868187579214196e-05,
+      "loss": 0.0816,
+      "step": 312
+    },
+    {
+      "epoch": 0.13265737220109844,
+      "grad_norm": 1.3803563117980957,
+      "learning_rate": 4.9867342627798904e-05,
+      "loss": 0.0795,
+      "step": 314
+    },
+    {
+      "epoch": 0.13350232361639205,
+      "grad_norm": 1.4253759384155273,
+      "learning_rate": 4.986649767638361e-05,
+      "loss": 0.1219,
+      "step": 316
+    },
+    {
+      "epoch": 0.13434727503168567,
+      "grad_norm": 1.8579727411270142,
+      "learning_rate": 4.986565272496832e-05,
+      "loss": 0.117,
+      "step": 318
+    },
+    {
+      "epoch": 0.13519222644697929,
+      "grad_norm": 1.3731677532196045,
+      "learning_rate": 4.9864807773553025e-05,
+      "loss": 0.0697,
+      "step": 320
+    },
+    {
+      "epoch": 0.13603717786227293,
+      "grad_norm": 1.7941423654556274,
+      "learning_rate": 4.9863962822137726e-05,
+      "loss": 0.1499,
+      "step": 322
+    },
+    {
+      "epoch": 0.13688212927756654,
+      "grad_norm": 1.8242301940917969,
+      "learning_rate": 4.986311787072244e-05,
+      "loss": 0.1148,
+      "step": 324
+    },
+    {
+      "epoch": 0.13772708069286016,
+      "grad_norm": 1.8088881969451904,
+      "learning_rate": 4.986227291930714e-05,
+      "loss": 0.1117,
+      "step": 326
+    },
+    {
+      "epoch": 0.13857203210815378,
+      "grad_norm": 1.3591686487197876,
+      "learning_rate": 4.986142796789185e-05,
+      "loss": 0.1057,
+      "step": 328
+    },
+    {
+      "epoch": 0.1394169835234474,
+      "grad_norm": 1.6867038011550903,
+      "learning_rate": 4.9860583016476554e-05,
+      "loss": 0.1065,
+      "step": 330
+    },
+    {
+      "epoch": 0.14026193493874103,
+      "grad_norm": 1.2499358654022217,
+      "learning_rate": 4.985973806506126e-05,
+      "loss": 0.1251,
+      "step": 332
+    },
+    {
+      "epoch": 0.14110688635403465,
+      "grad_norm": 1.7259184122085571,
+      "learning_rate": 4.985889311364597e-05,
+      "loss": 0.1066,
+      "step": 334
+    },
+    {
+      "epoch": 0.14195183776932827,
+      "grad_norm": 2.0314226150512695,
+      "learning_rate": 4.9858048162230676e-05,
+      "loss": 0.117,
+      "step": 336
+    },
+    {
+      "epoch": 0.14279678918462188,
+      "grad_norm": 1.4193428754806519,
+      "learning_rate": 4.9857203210815376e-05,
+      "loss": 0.129,
+      "step": 338
+    },
+    {
+      "epoch": 0.1436417405999155,
+      "grad_norm": 1.639891266822815,
+      "learning_rate": 4.985635825940009e-05,
+      "loss": 0.1269,
+      "step": 340
+    },
+    {
+      "epoch": 0.1444866920152091,
+      "grad_norm": 1.1235016584396362,
+      "learning_rate": 4.985551330798479e-05,
+      "loss": 0.0722,
+      "step": 342
+    },
+    {
+      "epoch": 0.14533164343050275,
+      "grad_norm": 1.941998839378357,
+      "learning_rate": 4.98546683565695e-05,
+      "loss": 0.085,
+      "step": 344
+    },
+    {
+      "epoch": 0.14617659484579637,
+      "grad_norm": 1.3687087297439575,
+      "learning_rate": 4.9853823405154205e-05,
+      "loss": 0.0776,
+      "step": 346
+    },
+    {
+      "epoch": 0.14702154626108999,
+      "grad_norm": 1.7349680662155151,
+      "learning_rate": 4.985297845373891e-05,
+      "loss": 0.1329,
+      "step": 348
+    },
+    {
+      "epoch": 0.1478664976763836,
+      "grad_norm": 1.4035015106201172,
+      "learning_rate": 4.985213350232362e-05,
+      "loss": 0.1047,
+      "step": 350
+    },
+    {
+      "epoch": 0.14871144909167722,
+      "grad_norm": 1.2989908456802368,
+      "learning_rate": 4.985128855090833e-05,
+      "loss": 0.0821,
+      "step": 352
+    },
+    {
+      "epoch": 0.14955640050697086,
+      "grad_norm": 1.9474072456359863,
+      "learning_rate": 4.985044359949303e-05,
+      "loss": 0.1334,
+      "step": 354
+    },
+    {
+      "epoch": 0.15040135192226448,
+      "grad_norm": 1.348775863647461,
+      "learning_rate": 4.984959864807774e-05,
+      "loss": 0.0856,
+      "step": 356
+    },
+    {
+      "epoch": 0.1512463033375581,
+      "grad_norm": 1.348597764968872,
+      "learning_rate": 4.984875369666244e-05,
+      "loss": 0.093,
+      "step": 358
+    },
+    {
+      "epoch": 0.1520912547528517,
+      "grad_norm": 1.7595020532608032,
+      "learning_rate": 4.984790874524715e-05,
+      "loss": 0.1192,
+      "step": 360
+    },
+    {
+      "epoch": 0.15293620616814532,
+      "grad_norm": 1.1924686431884766,
+      "learning_rate": 4.9847063793831856e-05,
+      "loss": 0.0749,
+      "step": 362
+    },
+    {
+      "epoch": 0.15378115758343897,
+      "grad_norm": 1.7564600706100464,
+      "learning_rate": 4.984621884241656e-05,
+      "loss": 0.0952,
+      "step": 364
+    },
+    {
+      "epoch": 0.15462610899873258,
+      "grad_norm": 1.3915026187896729,
+      "learning_rate": 4.984537389100127e-05,
+      "loss": 0.1268,
+      "step": 366
+    },
+    {
+      "epoch": 0.1554710604140262,
+      "grad_norm": 1.8819456100463867,
+      "learning_rate": 4.984452893958598e-05,
+      "loss": 0.0854,
+      "step": 368
+    },
+    {
+      "epoch": 0.1563160118293198,
+      "grad_norm": 1.6254719495773315,
+      "learning_rate": 4.984368398817068e-05,
+      "loss": 0.084,
+      "step": 370
+    },
+    {
+      "epoch": 0.15716096324461343,
+      "grad_norm": 2.196483850479126,
+      "learning_rate": 4.984283903675539e-05,
+      "loss": 0.1089,
+      "step": 372
+    },
+    {
+      "epoch": 0.15800591465990704,
+      "grad_norm": 1.471126675605774,
+      "learning_rate": 4.984199408534009e-05,
+      "loss": 0.0832,
+      "step": 374
+    },
+    {
+      "epoch": 0.1588508660752007,
+      "grad_norm": 1.1505743265151978,
+      "learning_rate": 4.98411491339248e-05,
+      "loss": 0.076,
+      "step": 376
+    },
+    {
+      "epoch": 0.1596958174904943,
+      "grad_norm": 1.4914796352386475,
+      "learning_rate": 4.984030418250951e-05,
+      "loss": 0.1304,
+      "step": 378
+    },
+    {
+      "epoch": 0.16054076890578792,
+      "grad_norm": 1.9942247867584229,
+      "learning_rate": 4.9839459231094214e-05,
+      "loss": 0.1227,
+      "step": 380
+    },
+    {
+      "epoch": 0.16138572032108153,
+      "grad_norm": 1.6885210275650024,
+      "learning_rate": 4.983861427967892e-05,
+      "loss": 0.0926,
+      "step": 382
+    },
+    {
+      "epoch": 0.16223067173637515,
+      "grad_norm": 1.2002671957015991,
+      "learning_rate": 4.983776932826363e-05,
+      "loss": 0.1011,
+      "step": 384
+    },
+    {
+      "epoch": 0.1630756231516688,
+      "grad_norm": 1.6890459060668945,
+      "learning_rate": 4.983692437684833e-05,
+      "loss": 0.1243,
+      "step": 386
+    },
+    {
+      "epoch": 0.1639205745669624,
+      "grad_norm": 1.4184560775756836,
+      "learning_rate": 4.983607942543304e-05,
+      "loss": 0.0848,
+      "step": 388
+    },
+    {
+      "epoch": 0.16476552598225602,
+      "grad_norm": 2.200415849685669,
+      "learning_rate": 4.983523447401774e-05,
+      "loss": 0.0779,
+      "step": 390
+    },
+    {
+      "epoch": 0.16561047739754964,
+      "grad_norm": 1.5030500888824463,
+      "learning_rate": 4.983438952260245e-05,
+      "loss": 0.0828,
+      "step": 392
+    },
+    {
+      "epoch": 0.16645542881284325,
+      "grad_norm": 1.2970515489578247,
+      "learning_rate": 4.983354457118716e-05,
+      "loss": 0.0889,
+      "step": 394
+    },
+    {
+      "epoch": 0.16730038022813687,
+      "grad_norm": 1.576575517654419,
+      "learning_rate": 4.9832699619771865e-05,
+      "loss": 0.0942,
+      "step": 396
+    },
+    {
+      "epoch": 0.1681453316434305,
+      "grad_norm": 1.6719400882720947,
+      "learning_rate": 4.983185466835657e-05,
+      "loss": 0.1166,
+      "step": 398
+    },
+    {
+      "epoch": 0.16899028305872413,
+      "grad_norm": 1.6408090591430664,
+      "learning_rate": 4.983100971694128e-05,
+      "loss": 0.1293,
+      "step": 400
+    },
+    {
+      "epoch": 0.16899028305872413,
+      "eval_accuracy": 0.6733919641317468,
+      "eval_cer": 0.09762961263971347,
+      "eval_loss": 0.20648193359375,
+      "eval_runtime": 861.8594,
+      "eval_samples_per_second": 13.457,
+      "eval_steps_per_second": 0.421,
+      "step": 400
+    },
+    {
+      "epoch": 0.16983523447401774,
+      "grad_norm": 2.1278653144836426,
+      "learning_rate": 4.983016476552598e-05,
+      "loss": 0.1321,
+      "step": 402
+    },
+    {
+      "epoch": 0.17068018588931136,
+      "grad_norm": 1.8531532287597656,
+      "learning_rate": 4.9829319814110694e-05,
+      "loss": 0.1042,
+      "step": 404
+    },
+    {
+      "epoch": 0.17152513730460497,
+      "grad_norm": 1.587566614151001,
+      "learning_rate": 4.9828474862695394e-05,
+      "loss": 0.1147,
+      "step": 406
+    },
+    {
+      "epoch": 0.17237008871989862,
+      "grad_norm": 2.2086760997772217,
+      "learning_rate": 4.982762991128011e-05,
+      "loss": 0.1246,
+      "step": 408
+    },
+    {
+      "epoch": 0.17321504013519223,
+      "grad_norm": 1.2952370643615723,
+      "learning_rate": 4.982678495986481e-05,
+      "loss": 0.1073,
+      "step": 410
+    },
+    {
+      "epoch": 0.17405999155048585,
+      "grad_norm": 1.7142947912216187,
+      "learning_rate": 4.9825940008449516e-05,
+      "loss": 0.0869,
+      "step": 412
+    },
+    {
+      "epoch": 0.17490494296577946,
+      "grad_norm": 1.9402838945388794,
+      "learning_rate": 4.982509505703422e-05,
+      "loss": 0.1126,
+      "step": 414
+    },
+    {
+      "epoch": 0.17574989438107308,
+      "grad_norm": 2.2303125858306885,
+      "learning_rate": 4.982425010561893e-05,
+      "loss": 0.1515,
+      "step": 416
+    },
+    {
+      "epoch": 0.1765948457963667,
+      "grad_norm": 1.8753235340118408,
+      "learning_rate": 4.982340515420364e-05,
+      "loss": 0.1059,
+      "step": 418
+    },
+    {
+      "epoch": 0.17743979721166034,
+      "grad_norm": 1.825201392173767,
+      "learning_rate": 4.9822560202788344e-05,
+      "loss": 0.1604,
+      "step": 420
+    },
+    {
+      "epoch": 0.17828474862695395,
+      "grad_norm": 2.0175940990448,
+      "learning_rate": 4.9821715251373045e-05,
+      "loss": 0.1078,
+      "step": 422
+    },
+    {
+      "epoch": 0.17912970004224757,
+      "grad_norm": 1.2101908922195435,
+      "learning_rate": 4.982087029995776e-05,
+      "loss": 0.1017,
+      "step": 424
+    },
+    {
+      "epoch": 0.17997465145754118,
+      "grad_norm": 1.1782002449035645,
+      "learning_rate": 4.982002534854246e-05,
+      "loss": 0.0722,
+      "step": 426
+    },
+    {
+      "epoch": 0.1808196028728348,
+      "grad_norm": 1.684949517250061,
+      "learning_rate": 4.9819180397127167e-05,
+      "loss": 0.1096,
+      "step": 428
+    },
+    {
+      "epoch": 0.18166455428812844,
+      "grad_norm": 1.9654364585876465,
+      "learning_rate": 4.9818335445711874e-05,
+      "loss": 0.0896,
+      "step": 430
+    },
+    {
+      "epoch": 0.18250950570342206,
+      "grad_norm": 1.5413470268249512,
+      "learning_rate": 4.981749049429658e-05,
+      "loss": 0.0781,
+      "step": 432
+    },
+    {
+      "epoch": 0.18335445711871567,
+      "grad_norm": 2.4538686275482178,
+      "learning_rate": 4.981664554288129e-05,
+      "loss": 0.1153,
+      "step": 434
+    },
+    {
+      "epoch": 0.1841994085340093,
+      "grad_norm": 1.3073904514312744,
+      "learning_rate": 4.9815800591465995e-05,
+      "loss": 0.0847,
+      "step": 436
+    },
+    {
+      "epoch": 0.1850443599493029,
+      "grad_norm": 1.651445746421814,
+      "learning_rate": 4.9814955640050696e-05,
+      "loss": 0.1313,
+      "step": 438
+    },
+    {
+      "epoch": 0.18588931136459655,
+      "grad_norm": 1.7214906215667725,
+      "learning_rate": 4.981411068863541e-05,
+      "loss": 0.1008,
+      "step": 440
+    },
+    {
+      "epoch": 0.18673426277989016,
+      "grad_norm": 1.0988144874572754,
+      "learning_rate": 4.981326573722011e-05,
+      "loss": 0.1162,
+      "step": 442
+    },
+    {
+      "epoch": 0.18757921419518378,
+      "grad_norm": 2.5338993072509766,
+      "learning_rate": 4.981242078580482e-05,
+      "loss": 0.1522,
+      "step": 444
+    },
+    {
+      "epoch": 0.1884241656104774,
+      "grad_norm": 1.2699946165084839,
+      "learning_rate": 4.9811575834389525e-05,
+      "loss": 0.0943,
+      "step": 446
+    },
+    {
+      "epoch": 0.189269117025771,
+      "grad_norm": 2.2291219234466553,
+      "learning_rate": 4.981073088297423e-05,
+      "loss": 0.0993,
+      "step": 448
+    },
+    {
+      "epoch": 0.19011406844106463,
+      "grad_norm": 1.102462649345398,
+      "learning_rate": 4.980988593155894e-05,
+      "loss": 0.0959,
+      "step": 450
+    },
+    {
+      "epoch": 0.19095901985635827,
+      "grad_norm": 1.1297458410263062,
+      "learning_rate": 4.9809040980143646e-05,
+      "loss": 0.0748,
+      "step": 452
+    },
+    {
+      "epoch": 0.19180397127165189,
+      "grad_norm": 1.316767930984497,
+      "learning_rate": 4.9808196028728347e-05,
+      "loss": 0.0729,
+      "step": 454
+    },
+    {
+      "epoch": 0.1926489226869455,
+      "grad_norm": 1.3369216918945312,
+      "learning_rate": 4.980735107731306e-05,
+      "loss": 0.0896,
+      "step": 456
+    },
+    {
+      "epoch": 0.19349387410223912,
+      "grad_norm": 1.6894314289093018,
+      "learning_rate": 4.980650612589776e-05,
+      "loss": 0.1607,
+      "step": 458
+    },
+    {
+      "epoch": 0.19433882551753273,
+      "grad_norm": 1.727819800376892,
+      "learning_rate": 4.980566117448247e-05,
+      "loss": 0.1024,
+      "step": 460
+    },
+    {
+      "epoch": 0.19518377693282637,
+      "grad_norm": 2.435275077819824,
+      "learning_rate": 4.9804816223067175e-05,
+      "loss": 0.1779,
+      "step": 462
+    },
+    {
+      "epoch": 0.19602872834812,
+      "grad_norm": 1.8471691608428955,
+      "learning_rate": 4.980397127165188e-05,
+      "loss": 0.0943,
+      "step": 464
+    },
+    {
+      "epoch": 0.1968736797634136,
+      "grad_norm": 1.471346139907837,
+      "learning_rate": 4.980312632023659e-05,
+      "loss": 0.1148,
+      "step": 466
+    },
+    {
+      "epoch": 0.19771863117870722,
+      "grad_norm": 2.621778726577759,
+      "learning_rate": 4.98022813688213e-05,
+      "loss": 0.1145,
+      "step": 468
+    },
+    {
+      "epoch": 0.19856358259400084,
+      "grad_norm": 1.4826730489730835,
+      "learning_rate": 4.9801436417406e-05,
+      "loss": 0.1011,
+      "step": 470
+    },
+    {
+      "epoch": 0.19940853400929445,
+      "grad_norm": 1.8840138912200928,
+      "learning_rate": 4.980059146599071e-05,
+      "loss": 0.0823,
+      "step": 472
+    },
+    {
+      "epoch": 0.2002534854245881,
+      "grad_norm": 1.329613208770752,
+      "learning_rate": 4.979974651457541e-05,
+      "loss": 0.0794,
+      "step": 474
+    },
+    {
+      "epoch": 0.2010984368398817,
+      "grad_norm": 1.6798653602600098,
+      "learning_rate": 4.979890156316012e-05,
+      "loss": 0.068,
+      "step": 476
+    },
+    {
+      "epoch": 0.20194338825517533,
+      "grad_norm": 2.3119256496429443,
+      "learning_rate": 4.9798056611744826e-05,
+      "loss": 0.1738,
+      "step": 478
+    },
+    {
+      "epoch": 0.20278833967046894,
+      "grad_norm": 1.5852372646331787,
+      "learning_rate": 4.979721166032953e-05,
+      "loss": 0.1466,
+      "step": 480
+    },
+    {
+      "epoch": 0.20363329108576256,
+      "grad_norm": 1.3242956399917603,
+      "learning_rate": 4.979636670891424e-05,
+      "loss": 0.0747,
+      "step": 482
+    },
+    {
+      "epoch": 0.2044782425010562,
+      "grad_norm": 1.8334121704101562,
+      "learning_rate": 4.979552175749895e-05,
+      "loss": 0.0832,
+      "step": 484
+    },
+    {
+      "epoch": 0.20532319391634982,
+      "grad_norm": 1.5589213371276855,
+      "learning_rate": 4.979467680608365e-05,
+      "loss": 0.0939,
+      "step": 486
+    },
+    {
+      "epoch": 0.20616814533164343,
+      "grad_norm": 0.6316360831260681,
+      "learning_rate": 4.979383185466836e-05,
+      "loss": 0.0623,
+      "step": 488
+    },
+    {
+      "epoch": 0.20701309674693705,
+      "grad_norm": 1.5444834232330322,
+      "learning_rate": 4.979298690325306e-05,
+      "loss": 0.0954,
+      "step": 490
+    },
+    {
+      "epoch": 0.20785804816223066,
+      "grad_norm": 2.2183780670166016,
+      "learning_rate": 4.979214195183777e-05,
+      "loss": 0.1144,
+      "step": 492
+    },
+    {
+      "epoch": 0.20870299957752428,
+      "grad_norm": 1.4162744283676147,
+      "learning_rate": 4.979129700042248e-05,
+      "loss": 0.0713,
+      "step": 494
+    },
+    {
+      "epoch": 0.20954795099281792,
+      "grad_norm": 1.2704912424087524,
+      "learning_rate": 4.9790452049007184e-05,
+      "loss": 0.0556,
+      "step": 496
+    },
+    {
+      "epoch": 0.21039290240811154,
+      "grad_norm": 1.9707940816879272,
+      "learning_rate": 4.978960709759189e-05,
+      "loss": 0.0829,
+      "step": 498
+    },
+    {
+      "epoch": 0.21123785382340515,
+      "grad_norm": 1.1645714044570923,
+      "learning_rate": 4.97887621461766e-05,
+      "loss": 0.0997,
+      "step": 500
+    },
+    {
+      "epoch": 0.21123785382340515,
+      "eval_accuracy": 0.6679599931022591,
+      "eval_cer": 0.10002028988064257,
+      "eval_loss": 0.21607132256031036,
+      "eval_runtime": 889.5025,
+      "eval_samples_per_second": 13.039,
+      "eval_steps_per_second": 0.408,
+      "step": 500
+    },
+    {
+      "epoch": 0.21208280523869877,
+      "grad_norm": 1.2636017799377441,
+      "learning_rate": 4.97879171947613e-05,
+      "loss": 0.0853,
+      "step": 502
+    },
+    {
+      "epoch": 0.21292775665399238,
+      "grad_norm": 1.3952999114990234,
+      "learning_rate": 4.978707224334601e-05,
+      "loss": 0.0827,
+      "step": 504
+    },
+    {
+      "epoch": 0.21377270806928603,
+      "grad_norm": 1.4890190362930298,
+      "learning_rate": 4.978622729193071e-05,
+      "loss": 0.0793,
+      "step": 506
+    },
+    {
+      "epoch": 0.21461765948457964,
+      "grad_norm": 3.3428499698638916,
+      "learning_rate": 4.978538234051543e-05,
+      "loss": 0.0804,
+      "step": 508
+    },
+    {
+      "epoch": 0.21546261089987326,
+      "grad_norm": 1.8637378215789795,
+      "learning_rate": 4.978453738910013e-05,
+      "loss": 0.1163,
+      "step": 510
+    },
+    {
+      "epoch": 0.21630756231516687,
+      "grad_norm": 1.4218493700027466,
+      "learning_rate": 4.9783692437684835e-05,
+      "loss": 0.0926,
+      "step": 512
+    },
+    {
+      "epoch": 0.2171525137304605,
+      "grad_norm": 1.2058361768722534,
+      "learning_rate": 4.978284748626954e-05,
+      "loss": 0.0877,
+      "step": 514
+    },
+    {
+      "epoch": 0.21799746514575413,
+      "grad_norm": 1.6888997554779053,
+      "learning_rate": 4.978200253485425e-05,
+      "loss": 0.1208,
+      "step": 516
+    },
+    {
+      "epoch": 0.21884241656104775,
+      "grad_norm": 1.4230191707611084,
+      "learning_rate": 4.9781157583438957e-05,
+      "loss": 0.1062,
+      "step": 518
+    },
+    {
+      "epoch": 0.21968736797634136,
+      "grad_norm": 1.6900297403335571,
+      "learning_rate": 4.9780312632023664e-05,
+      "loss": 0.1051,
+      "step": 520
+    },
+    {
+      "epoch": 0.22053231939163498,
+      "grad_norm": 1.9866880178451538,
+      "learning_rate": 4.9779467680608364e-05,
+      "loss": 0.144,
+      "step": 522
+    },
+    {
+      "epoch": 0.2213772708069286,
+      "grad_norm": 1.5090681314468384,
+      "learning_rate": 4.977862272919308e-05,
+      "loss": 0.1688,
+      "step": 524
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 1.5939511060714722,
+      "learning_rate": 4.977777777777778e-05,
+      "loss": 0.1095,
+      "step": 526
+    },
+    {
+      "epoch": 0.22306717363751585,
+      "grad_norm": 1.6898002624511719,
+      "learning_rate": 4.9776932826362486e-05,
+      "loss": 0.0933,
+      "step": 528
+    },
+    {
+      "epoch": 0.22391212505280947,
+      "grad_norm": 1.6217966079711914,
+      "learning_rate": 4.977608787494719e-05,
+      "loss": 0.0906,
+      "step": 530
+    },
+    {
+      "epoch": 0.22475707646810308,
+      "grad_norm": 1.187659740447998,
+      "learning_rate": 4.97752429235319e-05,
+      "loss": 0.1091,
+      "step": 532
+    },
+    {
+      "epoch": 0.2256020278833967,
+      "grad_norm": 1.255968689918518,
+      "learning_rate": 4.977439797211661e-05,
+      "loss": 0.0659,
+      "step": 534
+    },
+    {
+      "epoch": 0.22644697929869032,
+      "grad_norm": 1.5511211156845093,
+      "learning_rate": 4.9773553020701315e-05,
+      "loss": 0.0802,
+      "step": 536
+    },
+    {
+      "epoch": 0.22729193071398396,
+      "grad_norm": 1.7793858051300049,
+      "learning_rate": 4.9772708069286015e-05,
+      "loss": 0.1129,
+      "step": 538
+    },
+    {
+      "epoch": 0.22813688212927757,
+      "grad_norm": 1.4437144994735718,
+      "learning_rate": 4.977186311787073e-05,
+      "loss": 0.1,
+      "step": 540
+    },
+    {
+      "epoch": 0.2289818335445712,
+      "grad_norm": 1.4379544258117676,
+      "learning_rate": 4.977101816645543e-05,
+      "loss": 0.0864,
+      "step": 542
+    },
+    {
+      "epoch": 0.2298267849598648,
+      "grad_norm": 0.8294060826301575,
+      "learning_rate": 4.9770173215040137e-05,
+      "loss": 0.0763,
+      "step": 544
+    },
+    {
+      "epoch": 0.23067173637515842,
+      "grad_norm": 2.189450740814209,
+      "learning_rate": 4.9769328263624844e-05,
+      "loss": 0.123,
+      "step": 546
+    },
+    {
+      "epoch": 0.23151668779045204,
+      "grad_norm": 1.3774739503860474,
+      "learning_rate": 4.976848331220955e-05,
+      "loss": 0.0839,
+      "step": 548
+    },
+    {
+      "epoch": 0.23236163920574568,
+      "grad_norm": 1.9243252277374268,
+      "learning_rate": 4.976763836079426e-05,
+      "loss": 0.1049,
+      "step": 550
+    },
+    {
+      "epoch": 0.2332065906210393,
+      "grad_norm": 1.1769413948059082,
+      "learning_rate": 4.9766793409378965e-05,
+      "loss": 0.0834,
+      "step": 552
+    },
+    {
+      "epoch": 0.2340515420363329,
+      "grad_norm": 1.6378949880599976,
+      "learning_rate": 4.9765948457963666e-05,
+      "loss": 0.1071,
+      "step": 554
+    },
+    {
+      "epoch": 0.23489649345162653,
+      "grad_norm": 1.901498556137085,
+      "learning_rate": 4.976510350654838e-05,
+      "loss": 0.1007,
+      "step": 556
+    },
+    {
+      "epoch": 0.23574144486692014,
+      "grad_norm": 1.9260848760604858,
+      "learning_rate": 4.976425855513308e-05,
+      "loss": 0.2061,
+      "step": 558
+    },
+    {
+      "epoch": 0.23658639628221378,
+      "grad_norm": 3.253631353378296,
+      "learning_rate": 4.976341360371779e-05,
+      "loss": 0.1726,
+      "step": 560
+    },
+    {
+      "epoch": 0.2374313476975074,
+      "grad_norm": 1.5378361940383911,
+      "learning_rate": 4.9762568652302495e-05,
+      "loss": 0.0962,
+      "step": 562
+    },
+    {
+      "epoch": 0.23827629911280102,
+      "grad_norm": 1.3367713689804077,
+      "learning_rate": 4.97617237008872e-05,
+      "loss": 0.1047,
+      "step": 564
+    },
+    {
+      "epoch": 0.23912125052809463,
+      "grad_norm": 1.2774410247802734,
+      "learning_rate": 4.976087874947191e-05,
+      "loss": 0.0745,
+      "step": 566
+    },
+    {
+      "epoch": 0.23996620194338825,
+      "grad_norm": 1.8468585014343262,
+      "learning_rate": 4.9760033798056616e-05,
+      "loss": 0.1207,
+      "step": 568
+    },
+    {
+      "epoch": 0.24081115335868186,
+      "grad_norm": 2.125532388687134,
+      "learning_rate": 4.9759188846641317e-05,
+      "loss": 0.1321,
+      "step": 570
+    },
+    {
+      "epoch": 0.2416561047739755,
+      "grad_norm": 1.7013835906982422,
+      "learning_rate": 4.975834389522603e-05,
+      "loss": 0.0941,
+      "step": 572
+    },
+    {
+      "epoch": 0.24250105618926912,
+      "grad_norm": 1.4416042566299438,
+      "learning_rate": 4.975749894381073e-05,
+      "loss": 0.1195,
+      "step": 574
+    },
+    {
+      "epoch": 0.24334600760456274,
+      "grad_norm": 1.7456398010253906,
+      "learning_rate": 4.975665399239544e-05,
+      "loss": 0.1035,
+      "step": 576
+    },
+    {
+      "epoch": 0.24419095901985635,
+      "grad_norm": 1.6715686321258545,
+      "learning_rate": 4.9755809040980145e-05,
+      "loss": 0.083,
+      "step": 578
+    },
+    {
+      "epoch": 0.24503591043514997,
+      "grad_norm": 1.738721251487732,
+      "learning_rate": 4.975496408956485e-05,
+      "loss": 0.0842,
+      "step": 580
+    },
+    {
+      "epoch": 0.2458808618504436,
+      "grad_norm": 1.4637415409088135,
+      "learning_rate": 4.975411913814956e-05,
+      "loss": 0.1192,
+      "step": 582
+    },
+    {
+      "epoch": 0.24672581326573723,
+      "grad_norm": 1.5448874235153198,
+      "learning_rate": 4.975327418673427e-05,
+      "loss": 0.1092,
+      "step": 584
+    },
+    {
+      "epoch": 0.24757076468103084,
+      "grad_norm": 1.877724051475525,
+      "learning_rate": 4.975242923531897e-05,
+      "loss": 0.1203,
+      "step": 586
+    },
+    {
+      "epoch": 0.24841571609632446,
+      "grad_norm": 1.81662917137146,
+      "learning_rate": 4.975158428390368e-05,
+      "loss": 0.1101,
+      "step": 588
+    },
+    {
+      "epoch": 0.24926066751161807,
+      "grad_norm": 1.8778841495513916,
+      "learning_rate": 4.975073933248838e-05,
+      "loss": 0.105,
+      "step": 590
+    },
+    {
+      "epoch": 0.2501056189269117,
+      "grad_norm": 1.7268378734588623,
+      "learning_rate": 4.974989438107309e-05,
+      "loss": 0.1164,
+      "step": 592
+    },
+    {
+      "epoch": 0.2509505703422053,
+      "grad_norm": 1.972676396369934,
+      "learning_rate": 4.9749049429657796e-05,
+      "loss": 0.1177,
+      "step": 594
+    },
+    {
+      "epoch": 0.2517955217574989,
+      "grad_norm": 1.2417385578155518,
+      "learning_rate": 4.97482044782425e-05,
+      "loss": 0.0752,
+      "step": 596
+    },
+    {
+      "epoch": 0.2526404731727926,
+      "grad_norm": 1.4219918251037598,
+      "learning_rate": 4.974735952682721e-05,
+      "loss": 0.0786,
+      "step": 598
+    },
+    {
+      "epoch": 0.2534854245880862,
+      "grad_norm": 1.9878331422805786,
+      "learning_rate": 4.974651457541192e-05,
+      "loss": 0.1277,
+      "step": 600
+    },
+    {
+      "epoch": 0.2534854245880862,
+      "eval_accuracy": 0.6739092947059838,
+      "eval_cer": 0.09566237638610761,
+      "eval_loss": 0.20857658982276917,
+      "eval_runtime": 850.0665,
+      "eval_samples_per_second": 13.644,
+      "eval_steps_per_second": 0.427,
+      "step": 600
+    },
+    {
+      "epoch": 0.2543303760033798,
+      "grad_norm": 2.0730292797088623,
+      "learning_rate": 4.974566962399662e-05,
+      "loss": 0.1183,
+      "step": 602
+    },
+    {
+      "epoch": 0.25517532741867344,
+      "grad_norm": 1.7861765623092651,
+      "learning_rate": 4.974482467258133e-05,
+      "loss": 0.1055,
+      "step": 604
+    },
+    {
+      "epoch": 0.25602027883396705,
+      "grad_norm": 1.1870955228805542,
+      "learning_rate": 4.974397972116603e-05,
+      "loss": 0.1028,
+      "step": 606
+    },
+    {
+      "epoch": 0.25686523024926067,
+      "grad_norm": 1.5035452842712402,
+      "learning_rate": 4.9743134769750747e-05,
+      "loss": 0.0945,
+      "step": 608
+    },
+    {
+      "epoch": 0.2577101816645543,
+      "grad_norm": 1.4320887327194214,
+      "learning_rate": 4.974228981833545e-05,
+      "loss": 0.0896,
+      "step": 610
+    },
+    {
+      "epoch": 0.2585551330798479,
+      "grad_norm": 1.3751976490020752,
+      "learning_rate": 4.9741444866920154e-05,
+      "loss": 0.0968,
+      "step": 612
+    },
+    {
+      "epoch": 0.2594000844951415,
+      "grad_norm": 1.5161865949630737,
+      "learning_rate": 4.974059991550486e-05,
+      "loss": 0.1359,
+      "step": 614
+    },
+    {
+      "epoch": 0.26024503591043513,
+      "grad_norm": 1.5288583040237427,
+      "learning_rate": 4.973975496408957e-05,
+      "loss": 0.0897,
+      "step": 616
+    },
+    {
+      "epoch": 0.26108998732572875,
+      "grad_norm": 2.264397382736206,
+      "learning_rate": 4.9738910012674276e-05,
+      "loss": 0.1312,
+      "step": 618
+    },
+    {
+      "epoch": 0.2619349387410224,
+      "grad_norm": 1.3001281023025513,
+      "learning_rate": 4.973806506125898e-05,
+      "loss": 0.0848,
+      "step": 620
+    },
+    {
+      "epoch": 0.26277989015631603,
+      "grad_norm": 2.159705638885498,
+      "learning_rate": 4.973722010984368e-05,
+      "loss": 0.0851,
+      "step": 622
+    },
+    {
+      "epoch": 0.26362484157160965,
+      "grad_norm": 1.870424747467041,
+      "learning_rate": 4.97363751584284e-05,
+      "loss": 0.0959,
+      "step": 624
+    },
+    {
+      "epoch": 0.26446979298690326,
+      "grad_norm": 1.2810477018356323,
+      "learning_rate": 4.97355302070131e-05,
+      "loss": 0.0888,
+      "step": 626
+    },
+    {
+      "epoch": 0.2653147444021969,
+      "grad_norm": 2.051835775375366,
+      "learning_rate": 4.9734685255597805e-05,
+      "loss": 0.0992,
+      "step": 628
+    },
+    {
+      "epoch": 0.2661596958174905,
+      "grad_norm": 2.29409122467041,
+      "learning_rate": 4.973384030418251e-05,
+      "loss": 0.1117,
+      "step": 630
+    },
+    {
+      "epoch": 0.2670046472327841,
+      "grad_norm": 1.814590573310852,
+      "learning_rate": 4.973299535276722e-05,
+      "loss": 0.0997,
+      "step": 632
+    },
+    {
+      "epoch": 0.2678495986480777,
+      "grad_norm": 1.616007685661316,
+      "learning_rate": 4.9732150401351927e-05,
+      "loss": 0.1106,
+      "step": 634
+    },
+    {
+      "epoch": 0.26869455006337134,
+      "grad_norm": 1.9944722652435303,
+      "learning_rate": 4.9731305449936634e-05,
+      "loss": 0.1214,
+      "step": 636
+    },
+    {
+      "epoch": 0.26953950147866496,
+      "grad_norm": 1.2416284084320068,
+      "learning_rate": 4.9730460498521334e-05,
+      "loss": 0.1131,
+      "step": 638
+    },
+    {
+      "epoch": 0.27038445289395857,
+      "grad_norm": 0.6368843913078308,
+      "learning_rate": 4.972961554710605e-05,
+      "loss": 0.043,
+      "step": 640
+    },
+    {
+      "epoch": 0.27122940430925224,
+      "grad_norm": 1.266077995300293,
+      "learning_rate": 4.972877059569075e-05,
+      "loss": 0.1076,
+      "step": 642
+    },
+    {
+      "epoch": 0.27207435572454586,
+      "grad_norm": 1.4216665029525757,
+      "learning_rate": 4.9727925644275456e-05,
+      "loss": 0.1004,
+      "step": 644
+    },
+    {
+      "epoch": 0.2729193071398395,
+      "grad_norm": 1.3999009132385254,
+      "learning_rate": 4.972708069286016e-05,
+      "loss": 0.0958,
+      "step": 646
+    },
+    {
+      "epoch": 0.2737642585551331,
+      "grad_norm": 0.8566420078277588,
+      "learning_rate": 4.972623574144487e-05,
+      "loss": 0.1033,
+      "step": 648
+    },
+    {
+      "epoch": 0.2746092099704267,
+      "grad_norm": 1.770908236503601,
+      "learning_rate": 4.972539079002958e-05,
+      "loss": 0.1356,
+      "step": 650
+    },
+    {
+      "epoch": 0.2754541613857203,
+      "grad_norm": 1.3873233795166016,
+      "learning_rate": 4.9724545838614285e-05,
+      "loss": 0.1131,
+      "step": 652
+    },
+    {
+      "epoch": 0.27629911280101394,
+      "grad_norm": 1.4224236011505127,
+      "learning_rate": 4.9723700887198985e-05,
+      "loss": 0.0794,
+      "step": 654
+    },
+    {
+      "epoch": 0.27714406421630755,
+      "grad_norm": 1.4874037504196167,
+      "learning_rate": 4.97228559357837e-05,
+      "loss": 0.0949,
+      "step": 656
+    },
+    {
+      "epoch": 0.27798901563160117,
+      "grad_norm": 1.5964959859848022,
+      "learning_rate": 4.97220109843684e-05,
+      "loss": 0.0936,
+      "step": 658
+    },
+    {
+      "epoch": 0.2788339670468948,
+      "grad_norm": 1.5574475526809692,
+      "learning_rate": 4.9721166032953107e-05,
+      "loss": 0.0992,
+      "step": 660
+    },
+    {
+      "epoch": 0.2796789184621884,
+      "grad_norm": 3.1248722076416016,
+      "learning_rate": 4.9720321081537814e-05,
+      "loss": 0.0757,
+      "step": 662
+    },
+    {
+      "epoch": 0.28052386987748207,
+      "grad_norm": 2.060107707977295,
+      "learning_rate": 4.971947613012252e-05,
+      "loss": 0.0923,
+      "step": 664
+    },
+    {
+      "epoch": 0.2813688212927757,
+      "grad_norm": 1.112054467201233,
+      "learning_rate": 4.971863117870723e-05,
+      "loss": 0.0541,
+      "step": 666
+    },
+    {
+      "epoch": 0.2822137727080693,
+      "grad_norm": 1.5512161254882812,
+      "learning_rate": 4.9717786227291935e-05,
+      "loss": 0.0921,
+      "step": 668
+    },
+    {
+      "epoch": 0.2830587241233629,
+      "grad_norm": 1.544339895248413,
+      "learning_rate": 4.9716941275876636e-05,
+      "loss": 0.0877,
+      "step": 670
+    },
+    {
+      "epoch": 0.28390367553865653,
+      "grad_norm": 1.6371303796768188,
+      "learning_rate": 4.971609632446135e-05,
+      "loss": 0.1396,
+      "step": 672
+    },
+    {
+      "epoch": 0.28474862695395015,
+      "grad_norm": 1.3973644971847534,
+      "learning_rate": 4.971525137304605e-05,
+      "loss": 0.1194,
+      "step": 674
+    },
+    {
+      "epoch": 0.28559357836924376,
+      "grad_norm": 1.4875513315200806,
+      "learning_rate": 4.971440642163076e-05,
+      "loss": 0.0841,
+      "step": 676
+    },
+    {
+      "epoch": 0.2864385297845374,
+      "grad_norm": 1.3536832332611084,
+      "learning_rate": 4.9713561470215465e-05,
+      "loss": 0.0736,
+      "step": 678
+    },
+    {
+      "epoch": 0.287283481199831,
+      "grad_norm": 1.8549797534942627,
+      "learning_rate": 4.971271651880017e-05,
+      "loss": 0.1192,
+      "step": 680
+    },
+    {
+      "epoch": 0.2881284326151246,
+      "grad_norm": 2.0601296424865723,
+      "learning_rate": 4.971187156738488e-05,
+      "loss": 0.1393,
+      "step": 682
+    },
+    {
+      "epoch": 0.2889733840304182,
+      "grad_norm": 1.577346920967102,
+      "learning_rate": 4.9711026615969586e-05,
+      "loss": 0.072,
+      "step": 684
+    },
+    {
+      "epoch": 0.2898183354457119,
+      "grad_norm": 1.3023242950439453,
+      "learning_rate": 4.9710181664554287e-05,
+      "loss": 0.0936,
+      "step": 686
+    },
+    {
+      "epoch": 0.2906632868610055,
+      "grad_norm": 1.5268125534057617,
+      "learning_rate": 4.9709336713139e-05,
+      "loss": 0.1015,
+      "step": 688
+    },
+    {
+      "epoch": 0.2915082382762991,
+      "grad_norm": 1.6719399690628052,
+      "learning_rate": 4.97084917617237e-05,
+      "loss": 0.0853,
+      "step": 690
+    },
+    {
+      "epoch": 0.29235318969159274,
+      "grad_norm": 1.622419834136963,
+      "learning_rate": 4.970764681030841e-05,
+      "loss": 0.0808,
+      "step": 692
+    },
+    {
+      "epoch": 0.29319814110688636,
+      "grad_norm": 1.5559837818145752,
+      "learning_rate": 4.9706801858893115e-05,
+      "loss": 0.1144,
+      "step": 694
+    },
+    {
+      "epoch": 0.29404309252217997,
+      "grad_norm": 1.5186535120010376,
+      "learning_rate": 4.970595690747782e-05,
+      "loss": 0.1343,
+      "step": 696
+    },
+    {
+      "epoch": 0.2948880439374736,
+      "grad_norm": 1.011915683746338,
+      "learning_rate": 4.970511195606253e-05,
+      "loss": 0.064,
+      "step": 698
+    },
+    {
+      "epoch": 0.2957329953527672,
+      "grad_norm": 1.7633938789367676,
+      "learning_rate": 4.970426700464724e-05,
+      "loss": 0.1294,
+      "step": 700
+    },
+    {
+      "epoch": 0.2957329953527672,
+      "eval_accuracy": 0.6749439558544577,
+      "eval_cer": 0.09162204363206507,
+      "eval_loss": 0.20422282814979553,
+      "eval_runtime": 857.9136,
+      "eval_samples_per_second": 13.519,
+      "eval_steps_per_second": 0.423,
+      "step": 700
+    },
+    {
+      "epoch": 0.2965779467680608,
+      "grad_norm": 1.6819132566452026,
+      "learning_rate": 4.970342205323194e-05,
+      "loss": 0.0934,
+      "step": 702
+    },
+    {
+      "epoch": 0.29742289818335443,
+      "grad_norm": 1.0647895336151123,
+      "learning_rate": 4.970257710181665e-05,
+      "loss": 0.1007,
+      "step": 704
+    },
+    {
+      "epoch": 0.2982678495986481,
+      "grad_norm": 2.1595728397369385,
+      "learning_rate": 4.970173215040135e-05,
+      "loss": 0.1345,
+      "step": 706
+    },
+    {
+      "epoch": 0.2991128010139417,
+      "grad_norm": 1.0478488206863403,
+      "learning_rate": 4.9700887198986066e-05,
+      "loss": 0.0608,
+      "step": 708
+    },
+    {
+      "epoch": 0.29995775242923534,
+      "grad_norm": 1.2893446683883667,
+      "learning_rate": 4.9700042247570766e-05,
+      "loss": 0.1021,
+      "step": 710
+    },
+    {
+      "epoch": 0.30080270384452895,
+      "grad_norm": 2.40004563331604,
+      "learning_rate": 4.969919729615547e-05,
+      "loss": 0.1251,
+      "step": 712
+    },
+    {
+      "epoch": 0.30164765525982257,
+      "grad_norm": 1.4895504713058472,
+      "learning_rate": 4.969835234474018e-05,
+      "loss": 0.101,
+      "step": 714
+    },
+    {
+      "epoch": 0.3024926066751162,
+      "grad_norm": 1.2066045999526978,
+      "learning_rate": 4.969750739332489e-05,
+      "loss": 0.1069,
+      "step": 716
+    },
+    {
+      "epoch": 0.3033375580904098,
+      "grad_norm": 2.0504655838012695,
+      "learning_rate": 4.9696662441909595e-05,
+      "loss": 0.1287,
+      "step": 718
+    },
+    {
+      "epoch": 0.3041825095057034,
+      "grad_norm": 2.006098508834839,
+      "learning_rate": 4.96958174904943e-05,
+      "loss": 0.1442,
+      "step": 720
+    },
+    {
+      "epoch": 0.30502746092099703,
+      "grad_norm": 1.9753419160842896,
+      "learning_rate": 4.9694972539079e-05,
+      "loss": 0.1023,
+      "step": 722
+    },
+    {
+      "epoch": 0.30587241233629064,
+      "grad_norm": 1.803053617477417,
+      "learning_rate": 4.9694127587663717e-05,
+      "loss": 0.1168,
+      "step": 724
+    },
+    {
+      "epoch": 0.30671736375158426,
+      "grad_norm": 1.8871780633926392,
+      "learning_rate": 4.969328263624842e-05,
+      "loss": 0.12,
+      "step": 726
+    },
+    {
+      "epoch": 0.30756231516687793,
+      "grad_norm": 1.7384098768234253,
+      "learning_rate": 4.9692437684833124e-05,
+      "loss": 0.1312,
+      "step": 728
+    },
+    {
+      "epoch": 0.30840726658217155,
+      "grad_norm": 1.6403692960739136,
+      "learning_rate": 4.969159273341783e-05,
+      "loss": 0.0897,
+      "step": 730
+    },
+    {
+      "epoch": 0.30925221799746516,
+      "grad_norm": 1.7307875156402588,
+      "learning_rate": 4.969074778200254e-05,
+      "loss": 0.0944,
+      "step": 732
+    },
+    {
+      "epoch": 0.3100971694127588,
+      "grad_norm": 1.8397547006607056,
+      "learning_rate": 4.9689902830587246e-05,
+      "loss": 0.0745,
+      "step": 734
+    },
+    {
+      "epoch": 0.3109421208280524,
+      "grad_norm": 1.6901930570602417,
+      "learning_rate": 4.968905787917195e-05,
+      "loss": 0.1161,
+      "step": 736
+    },
+    {
+      "epoch": 0.311787072243346,
+      "grad_norm": 1.9397594928741455,
+      "learning_rate": 4.9688212927756653e-05,
+      "loss": 0.0964,
+      "step": 738
+    },
+    {
+      "epoch": 0.3126320236586396,
+      "grad_norm": 1.4291322231292725,
+      "learning_rate": 4.968736797634137e-05,
+      "loss": 0.0987,
+      "step": 740
+    },
+    {
+      "epoch": 0.31347697507393324,
+      "grad_norm": 1.0510236024856567,
+      "learning_rate": 4.968652302492607e-05,
+      "loss": 0.0831,
+      "step": 742
+    },
+    {
+      "epoch": 0.31432192648922685,
+      "grad_norm": 1.2731682062149048,
+      "learning_rate": 4.9685678073510775e-05,
+      "loss": 0.0792,
+      "step": 744
+    },
+    {
+      "epoch": 0.31516687790452047,
+      "grad_norm": 1.3949739933013916,
+      "learning_rate": 4.968483312209548e-05,
+      "loss": 0.0904,
+      "step": 746
+    },
+    {
+      "epoch": 0.3160118293198141,
+      "grad_norm": 1.6889134645462036,
+      "learning_rate": 4.968398817068019e-05,
+      "loss": 0.0752,
+      "step": 748
+    },
+    {
+      "epoch": 0.31685678073510776,
+      "grad_norm": 1.578761339187622,
+      "learning_rate": 4.9683143219264897e-05,
+      "loss": 0.0781,
+      "step": 750
+    },
+    {
+      "epoch": 0.3177017321504014,
+      "grad_norm": 1.338944435119629,
+      "learning_rate": 4.9682298267849604e-05,
+      "loss": 0.0753,
+      "step": 752
+    },
+    {
+      "epoch": 0.318546683565695,
+      "grad_norm": 1.8360531330108643,
+      "learning_rate": 4.9681453316434304e-05,
+      "loss": 0.1355,
+      "step": 754
+    },
+    {
+      "epoch": 0.3193916349809886,
+      "grad_norm": 1.851449966430664,
+      "learning_rate": 4.968060836501902e-05,
+      "loss": 0.1004,
+      "step": 756
+    },
+    {
+      "epoch": 0.3202365863962822,
+      "grad_norm": 1.6657130718231201,
+      "learning_rate": 4.967976341360372e-05,
+      "loss": 0.1208,
+      "step": 758
+    },
+    {
+      "epoch": 0.32108153781157583,
+      "grad_norm": 1.2685651779174805,
+      "learning_rate": 4.9678918462188426e-05,
+      "loss": 0.0878,
+      "step": 760
+    },
+    {
+      "epoch": 0.32192648922686945,
+      "grad_norm": 1.2503762245178223,
+      "learning_rate": 4.967807351077313e-05,
+      "loss": 0.084,
+      "step": 762
+    },
+    {
+      "epoch": 0.32277144064216307,
+      "grad_norm": 1.9573127031326294,
+      "learning_rate": 4.967722855935784e-05,
+      "loss": 0.1035,
+      "step": 764
+    },
+    {
+      "epoch": 0.3236163920574567,
+      "grad_norm": 1.7501864433288574,
+      "learning_rate": 4.967638360794255e-05,
+      "loss": 0.1081,
+      "step": 766
+    },
+    {
+      "epoch": 0.3244613434727503,
+      "grad_norm": 1.4179309606552124,
+      "learning_rate": 4.9675538656527255e-05,
+      "loss": 0.0809,
+      "step": 768
+    },
+    {
+      "epoch": 0.3253062948880439,
+      "grad_norm": 1.814603328704834,
+      "learning_rate": 4.9674693705111955e-05,
+      "loss": 0.1207,
+      "step": 770
+    },
+    {
+      "epoch": 0.3261512463033376,
+      "grad_norm": 2.027031660079956,
+      "learning_rate": 4.967384875369667e-05,
+      "loss": 0.1055,
+      "step": 772
+    },
+    {
+      "epoch": 0.3269961977186312,
+      "grad_norm": 1.419382095336914,
+      "learning_rate": 4.967300380228137e-05,
+      "loss": 0.1301,
+      "step": 774
+    },
+    {
+      "epoch": 0.3278411491339248,
+      "grad_norm": 1.9750804901123047,
+      "learning_rate": 4.9672158850866077e-05,
+      "loss": 0.0979,
+      "step": 776
+    },
+    {
+      "epoch": 0.32868610054921843,
+      "grad_norm": 1.274034857749939,
+      "learning_rate": 4.9671313899450784e-05,
+      "loss": 0.1249,
+      "step": 778
+    },
+    {
+      "epoch": 0.32953105196451205,
+      "grad_norm": 1.4384434223175049,
+      "learning_rate": 4.967046894803549e-05,
+      "loss": 0.0958,
+      "step": 780
+    },
+    {
+      "epoch": 0.33037600337980566,
+      "grad_norm": 1.5350993871688843,
+      "learning_rate": 4.96696239966202e-05,
+      "loss": 0.104,
+      "step": 782
+    },
+    {
+      "epoch": 0.3312209547950993,
+      "grad_norm": 1.434937596321106,
+      "learning_rate": 4.9668779045204905e-05,
+      "loss": 0.1108,
+      "step": 784
+    },
+    {
+      "epoch": 0.3320659062103929,
+      "grad_norm": 1.7014148235321045,
+      "learning_rate": 4.9667934093789606e-05,
+      "loss": 0.0979,
+      "step": 786
+    },
+    {
+      "epoch": 0.3329108576256865,
+      "grad_norm": 2.0817980766296387,
+      "learning_rate": 4.966708914237432e-05,
+      "loss": 0.1755,
+      "step": 788
+    },
+    {
+      "epoch": 0.3337558090409801,
+      "grad_norm": 1.4785884618759155,
+      "learning_rate": 4.966624419095902e-05,
+      "loss": 0.1047,
+      "step": 790
+    },
+    {
+      "epoch": 0.33460076045627374,
+      "grad_norm": 1.2489787340164185,
+      "learning_rate": 4.966539923954373e-05,
+      "loss": 0.0788,
+      "step": 792
+    },
+    {
+      "epoch": 0.3354457118715674,
+      "grad_norm": 1.1976639032363892,
+      "learning_rate": 4.9664554288128435e-05,
+      "loss": 0.0901,
+      "step": 794
+    },
+    {
+      "epoch": 0.336290663286861,
+      "grad_norm": 1.2119868993759155,
+      "learning_rate": 4.966370933671314e-05,
+      "loss": 0.0854,
+      "step": 796
+    },
+    {
+      "epoch": 0.33713561470215464,
+      "grad_norm": 1.3082143068313599,
+      "learning_rate": 4.966286438529785e-05,
+      "loss": 0.0958,
+      "step": 798
+    },
+    {
+      "epoch": 0.33798056611744826,
+      "grad_norm": 1.2046393156051636,
+      "learning_rate": 4.9662019433882556e-05,
+      "loss": 0.1144,
+      "step": 800
+    },
+    {
+      "epoch": 0.33798056611744826,
+      "eval_accuracy": 0.6734781858941197,
+      "eval_cer": 0.09568884144781531,
+      "eval_loss": 0.20647239685058594,
+      "eval_runtime": 849.7943,
+      "eval_samples_per_second": 13.648,
+      "eval_steps_per_second": 0.427,
+      "step": 800
+    },
+    {
+      "epoch": 0.33882551753274187,
+      "grad_norm": 1.3277101516723633,
+      "learning_rate": 4.966117448246726e-05,
+      "loss": 0.1023,
+      "step": 802
+    },
+    {
+      "epoch": 0.3396704689480355,
+      "grad_norm": 2.174595832824707,
+      "learning_rate": 4.966032953105197e-05,
+      "loss": 0.1265,
+      "step": 804
+    },
+    {
+      "epoch": 0.3405154203633291,
+      "grad_norm": 1.4048463106155396,
+      "learning_rate": 4.965948457963667e-05,
+      "loss": 0.0777,
+      "step": 806
+    },
+    {
+      "epoch": 0.3413603717786227,
+      "grad_norm": 1.428268313407898,
+      "learning_rate": 4.9658639628221385e-05,
+      "loss": 0.0767,
+      "step": 808
+    },
+    {
+      "epoch": 0.34220532319391633,
+      "grad_norm": 1.6813422441482544,
+      "learning_rate": 4.9657794676806085e-05,
+      "loss": 0.088,
+      "step": 810
+    },
+    {
+      "epoch": 0.34305027460920995,
+      "grad_norm": 1.722402811050415,
+      "learning_rate": 4.965694972539079e-05,
+      "loss": 0.135,
+      "step": 812
+    },
+    {
+      "epoch": 0.34389522602450356,
+      "grad_norm": 1.913205862045288,
+      "learning_rate": 4.96561047739755e-05,
+      "loss": 0.084,
+      "step": 814
+    },
+    {
+      "epoch": 0.34474017743979724,
+      "grad_norm": 2.381237506866455,
+      "learning_rate": 4.965525982256021e-05,
+      "loss": 0.0957,
+      "step": 816
+    },
+    {
+      "epoch": 0.34558512885509085,
+      "grad_norm": 1.3818658590316772,
+      "learning_rate": 4.9654414871144914e-05,
+      "loss": 0.0889,
+      "step": 818
+    },
+    {
+      "epoch": 0.34643008027038447,
+      "grad_norm": 2.095109701156616,
+      "learning_rate": 4.965356991972962e-05,
+      "loss": 0.1397,
+      "step": 820
+    },
+    {
+      "epoch": 0.3472750316856781,
+      "grad_norm": 1.6586017608642578,
+      "learning_rate": 4.965272496831432e-05,
+      "loss": 0.105,
+      "step": 822
+    },
+    {
+      "epoch": 0.3481199831009717,
+      "grad_norm": 0.987585723400116,
+      "learning_rate": 4.9651880016899036e-05,
+      "loss": 0.0574,
+      "step": 824
+    },
+    {
+      "epoch": 0.3489649345162653,
+      "grad_norm": 1.1450375318527222,
+      "learning_rate": 4.9651035065483736e-05,
+      "loss": 0.1137,
+      "step": 826
+    },
+    {
+      "epoch": 0.34980988593155893,
+      "grad_norm": 1.1953158378601074,
+      "learning_rate": 4.9650190114068443e-05,
+      "loss": 0.0888,
+      "step": 828
+    },
+    {
+      "epoch": 0.35065483734685254,
+      "grad_norm": 1.4406365156173706,
+      "learning_rate": 4.964934516265315e-05,
+      "loss": 0.0587,
+      "step": 830
+    },
+    {
+      "epoch": 0.35149978876214616,
+      "grad_norm": 2.095817804336548,
+      "learning_rate": 4.964850021123786e-05,
+      "loss": 0.0986,
+      "step": 832
+    },
+    {
+      "epoch": 0.3523447401774398,
+      "grad_norm": 1.560104489326477,
+      "learning_rate": 4.9647655259822565e-05,
+      "loss": 0.0834,
+      "step": 834
+    },
+    {
+      "epoch": 0.3531896915927334,
+      "grad_norm": 1.1767041683197021,
+      "learning_rate": 4.964681030840727e-05,
+      "loss": 0.0685,
+      "step": 836
+    },
+    {
+      "epoch": 0.35403464300802706,
+      "grad_norm": 1.3894157409667969,
+      "learning_rate": 4.964596535699197e-05,
+      "loss": 0.075,
+      "step": 838
+    },
+    {
+      "epoch": 0.3548795944233207,
+      "grad_norm": 2.41341233253479,
+      "learning_rate": 4.9645120405576687e-05,
+      "loss": 0.1638,
+      "step": 840
+    },
+    {
+      "epoch": 0.3557245458386143,
+      "grad_norm": 1.4480561017990112,
+      "learning_rate": 4.964427545416139e-05,
+      "loss": 0.0849,
+      "step": 842
+    },
+    {
+      "epoch": 0.3565694972539079,
+      "grad_norm": 1.3235077857971191,
+      "learning_rate": 4.9643430502746094e-05,
+      "loss": 0.0826,
+      "step": 844
+    },
+    {
+      "epoch": 0.3574144486692015,
+      "grad_norm": 1.5331876277923584,
+      "learning_rate": 4.96425855513308e-05,
+      "loss": 0.1023,
+      "step": 846
+    },
+    {
+      "epoch": 0.35825940008449514,
+      "grad_norm": 1.0759912729263306,
+      "learning_rate": 4.964174059991551e-05,
+      "loss": 0.1031,
+      "step": 848
+    },
+    {
+      "epoch": 0.35910435149978875,
+      "grad_norm": 2.00351881980896,
+      "learning_rate": 4.9640895648500216e-05,
+      "loss": 0.1153,
+      "step": 850
+    },
+    {
+      "epoch": 0.35994930291508237,
+      "grad_norm": 1.0817115306854248,
+      "learning_rate": 4.964005069708492e-05,
+      "loss": 0.0852,
+      "step": 852
+    },
+    {
+      "epoch": 0.360794254330376,
+      "grad_norm": 2.143852472305298,
+      "learning_rate": 4.9639205745669623e-05,
+      "loss": 0.1136,
+      "step": 854
+    },
+    {
+      "epoch": 0.3616392057456696,
+      "grad_norm": 1.8244491815567017,
+      "learning_rate": 4.963836079425434e-05,
+      "loss": 0.0983,
+      "step": 856
+    },
+    {
+      "epoch": 0.36248415716096327,
+      "grad_norm": 1.4860317707061768,
+      "learning_rate": 4.963751584283904e-05,
+      "loss": 0.0958,
+      "step": 858
+    },
+    {
+      "epoch": 0.3633291085762569,
+      "grad_norm": 2.0593981742858887,
+      "learning_rate": 4.9636670891423745e-05,
+      "loss": 0.101,
+      "step": 860
+    },
+    {
+      "epoch": 0.3641740599915505,
+      "grad_norm": 1.366678237915039,
+      "learning_rate": 4.963582594000845e-05,
+      "loss": 0.0713,
+      "step": 862
+    },
+    {
+      "epoch": 0.3650190114068441,
+      "grad_norm": 1.5878137350082397,
+      "learning_rate": 4.963498098859316e-05,
+      "loss": 0.0652,
+      "step": 864
+    },
+    {
+      "epoch": 0.36586396282213773,
+      "grad_norm": 1.733601689338684,
+      "learning_rate": 4.9634136037177867e-05,
+      "loss": 0.0886,
+      "step": 866
+    },
+    {
+      "epoch": 0.36670891423743135,
+      "grad_norm": 1.2537791728973389,
+      "learning_rate": 4.9633291085762574e-05,
+      "loss": 0.0853,
+      "step": 868
+    },
+    {
+      "epoch": 0.36755386565272496,
+      "grad_norm": 1.5744574069976807,
+      "learning_rate": 4.9632446134347274e-05,
+      "loss": 0.0706,
+      "step": 870
+    },
+    {
+      "epoch": 0.3683988170680186,
+      "grad_norm": 2.3552136421203613,
+      "learning_rate": 4.963160118293199e-05,
+      "loss": 0.1153,
+      "step": 872
+    },
+    {
+      "epoch": 0.3692437684833122,
+      "grad_norm": 1.0820305347442627,
+      "learning_rate": 4.963075623151669e-05,
+      "loss": 0.0671,
+      "step": 874
+    },
+    {
+      "epoch": 0.3700887198986058,
+      "grad_norm": 1.6050610542297363,
+      "learning_rate": 4.9629911280101396e-05,
+      "loss": 0.0934,
+      "step": 876
+    },
+    {
+      "epoch": 0.3709336713138994,
+      "grad_norm": 1.5215171575546265,
+      "learning_rate": 4.96290663286861e-05,
+      "loss": 0.1586,
+      "step": 878
+    },
+    {
+      "epoch": 0.3717786227291931,
+      "grad_norm": 1.4043138027191162,
+      "learning_rate": 4.962822137727081e-05,
+      "loss": 0.0879,
+      "step": 880
+    },
+    {
+      "epoch": 0.3726235741444867,
+      "grad_norm": 0.8804981112480164,
+      "learning_rate": 4.962737642585552e-05,
+      "loss": 0.0489,
+      "step": 882
+    },
+    {
+      "epoch": 0.37346852555978033,
+      "grad_norm": 1.3882102966308594,
+      "learning_rate": 4.9626531474440225e-05,
+      "loss": 0.0761,
+      "step": 884
+    },
+    {
+      "epoch": 0.37431347697507394,
+      "grad_norm": 1.4841866493225098,
+      "learning_rate": 4.9625686523024925e-05,
+      "loss": 0.0844,
+      "step": 886
+    },
+    {
+      "epoch": 0.37515842839036756,
+      "grad_norm": 1.3954799175262451,
+      "learning_rate": 4.962484157160964e-05,
+      "loss": 0.0709,
+      "step": 888
+    },
+    {
+      "epoch": 0.3760033798056612,
+      "grad_norm": 1.3372573852539062,
+      "learning_rate": 4.962399662019434e-05,
+      "loss": 0.096,
+      "step": 890
+    },
+    {
+      "epoch": 0.3768483312209548,
+      "grad_norm": 1.2923667430877686,
+      "learning_rate": 4.962315166877905e-05,
+      "loss": 0.1023,
+      "step": 892
+    },
+    {
+      "epoch": 0.3776932826362484,
+      "grad_norm": 1.5196459293365479,
+      "learning_rate": 4.9622306717363754e-05,
+      "loss": 0.115,
+      "step": 894
+    },
+    {
+      "epoch": 0.378538234051542,
+      "grad_norm": 1.83379065990448,
+      "learning_rate": 4.962146176594846e-05,
+      "loss": 0.0993,
+      "step": 896
+    },
+    {
+      "epoch": 0.37938318546683564,
+      "grad_norm": 1.5020735263824463,
+      "learning_rate": 4.962061681453317e-05,
+      "loss": 0.0853,
+      "step": 898
+    },
+    {
+      "epoch": 0.38022813688212925,
+      "grad_norm": 1.7137893438339233,
+      "learning_rate": 4.9619771863117875e-05,
+      "loss": 0.1109,
+      "step": 900
+    },
+    {
+      "epoch": 0.38022813688212925,
+      "eval_accuracy": 0.6778754957751336,
+      "eval_cer": 0.09363338832185043,
+      "eval_loss": 0.20783965289592743,
+      "eval_runtime": 856.7083,
+      "eval_samples_per_second": 13.538,
+      "eval_steps_per_second": 0.424,
+      "step": 900
+    },
+    {
+      "epoch": 0.3810730882974229,
+      "grad_norm": 1.7347830533981323,
+      "learning_rate": 4.9618926911702576e-05,
+      "loss": 0.1199,
+      "step": 902
+    },
+    {
+      "epoch": 0.38191803971271654,
+      "grad_norm": 1.9052618741989136,
+      "learning_rate": 4.961808196028729e-05,
+      "loss": 0.1172,
+      "step": 904
+    },
+    {
+      "epoch": 0.38276299112801015,
+      "grad_norm": 1.5322483777999878,
+      "learning_rate": 4.961723700887199e-05,
+      "loss": 0.0881,
+      "step": 906
+    },
+    {
+      "epoch": 0.38360794254330377,
+      "grad_norm": 0.9071521759033203,
+      "learning_rate": 4.9616392057456704e-05,
+      "loss": 0.0793,
+      "step": 908
+    },
+    {
+      "epoch": 0.3844528939585974,
+      "grad_norm": 1.6952403783798218,
+      "learning_rate": 4.9615547106041405e-05,
+      "loss": 0.0917,
+      "step": 910
+    },
+    {
+      "epoch": 0.385297845373891,
+      "grad_norm": 1.9736824035644531,
+      "learning_rate": 4.961470215462611e-05,
+      "loss": 0.09,
+      "step": 912
+    },
+    {
+      "epoch": 0.3861427967891846,
+      "grad_norm": 1.999121904373169,
+      "learning_rate": 4.961385720321082e-05,
+      "loss": 0.1251,
+      "step": 914
+    },
+    {
+      "epoch": 0.38698774820447823,
+      "grad_norm": 1.406522512435913,
+      "learning_rate": 4.9613012251795526e-05,
+      "loss": 0.0969,
+      "step": 916
+    },
+    {
+      "epoch": 0.38783269961977185,
+      "grad_norm": 1.3701584339141846,
+      "learning_rate": 4.9612167300380233e-05,
+      "loss": 0.0932,
+      "step": 918
+    },
+    {
+      "epoch": 0.38867765103506546,
+      "grad_norm": 1.0016272068023682,
+      "learning_rate": 4.961132234896494e-05,
+      "loss": 0.0597,
+      "step": 920
+    },
+    {
+      "epoch": 0.3895226024503591,
+      "grad_norm": 1.3503810167312622,
+      "learning_rate": 4.961047739754964e-05,
+      "loss": 0.071,
+      "step": 922
+    },
+    {
+      "epoch": 0.39036755386565275,
+      "grad_norm": 1.599910855293274,
+      "learning_rate": 4.9609632446134355e-05,
+      "loss": 0.099,
+      "step": 924
+    },
+    {
+      "epoch": 0.39121250528094637,
+      "grad_norm": 1.4374080896377563,
+      "learning_rate": 4.9608787494719055e-05,
+      "loss": 0.0792,
+      "step": 926
+    },
+    {
+      "epoch": 0.39205745669624,
+      "grad_norm": 2.009080648422241,
+      "learning_rate": 4.960794254330376e-05,
+      "loss": 0.1313,
+      "step": 928
+    },
+    {
+      "epoch": 0.3929024081115336,
+      "grad_norm": 1.5988420248031616,
+      "learning_rate": 4.960709759188847e-05,
+      "loss": 0.1013,
+      "step": 930
+    },
+    {
+      "epoch": 0.3937473595268272,
+      "grad_norm": 1.9358876943588257,
+      "learning_rate": 4.960625264047318e-05,
+      "loss": 0.1015,
+      "step": 932
+    },
+    {
+      "epoch": 0.3945923109421208,
+      "grad_norm": 1.4540034532546997,
+      "learning_rate": 4.9605407689057884e-05,
+      "loss": 0.0727,
+      "step": 934
+    },
+    {
+      "epoch": 0.39543726235741444,
+      "grad_norm": 1.400652289390564,
+      "learning_rate": 4.960456273764259e-05,
+      "loss": 0.0867,
+      "step": 936
+    },
+    {
+      "epoch": 0.39628221377270806,
+      "grad_norm": 1.7808191776275635,
+      "learning_rate": 4.960371778622729e-05,
+      "loss": 0.1269,
+      "step": 938
+    },
+    {
+      "epoch": 0.3971271651880017,
+      "grad_norm": 1.5522500276565552,
+      "learning_rate": 4.9602872834812006e-05,
+      "loss": 0.1115,
+      "step": 940
+    },
+    {
+      "epoch": 0.3979721166032953,
+      "grad_norm": 1.5305380821228027,
+      "learning_rate": 4.9602027883396706e-05,
+      "loss": 0.0993,
+      "step": 942
+    },
+    {
+      "epoch": 0.3988170680185889,
+      "grad_norm": 1.0456280708312988,
+      "learning_rate": 4.9601182931981413e-05,
+      "loss": 0.0775,
+      "step": 944
+    },
+    {
+      "epoch": 0.3996620194338826,
+      "grad_norm": 1.5103769302368164,
+      "learning_rate": 4.960033798056612e-05,
+      "loss": 0.0801,
+      "step": 946
+    },
+    {
+      "epoch": 0.4005069708491762,
+      "grad_norm": 1.217235803604126,
+      "learning_rate": 4.959949302915083e-05,
+      "loss": 0.0951,
+      "step": 948
+    },
+    {
+      "epoch": 0.4013519222644698,
+      "grad_norm": 1.5151118040084839,
+      "learning_rate": 4.9598648077735535e-05,
+      "loss": 0.1016,
+      "step": 950
+    },
+    {
+      "epoch": 0.4021968736797634,
+      "grad_norm": 1.443966269493103,
+      "learning_rate": 4.959780312632024e-05,
+      "loss": 0.1283,
+      "step": 952
+    },
+    {
+      "epoch": 0.40304182509505704,
+      "grad_norm": 1.539581537246704,
+      "learning_rate": 4.959695817490494e-05,
+      "loss": 0.115,
+      "step": 954
+    },
+    {
+      "epoch": 0.40388677651035065,
+      "grad_norm": 1.4787349700927734,
+      "learning_rate": 4.959611322348966e-05,
+      "loss": 0.1022,
+      "step": 956
+    },
+    {
+      "epoch": 0.40473172792564427,
+      "grad_norm": 2.0702366828918457,
+      "learning_rate": 4.959526827207436e-05,
+      "loss": 0.1072,
+      "step": 958
+    },
+    {
+      "epoch": 0.4055766793409379,
+      "grad_norm": 1.1393963098526,
+      "learning_rate": 4.9594423320659064e-05,
+      "loss": 0.1219,
+      "step": 960
+    },
+    {
+      "epoch": 0.4064216307562315,
+      "grad_norm": 0.837650716304779,
+      "learning_rate": 4.959357836924377e-05,
+      "loss": 0.0657,
+      "step": 962
+    },
+    {
+      "epoch": 0.4072665821715251,
+      "grad_norm": 1.5785096883773804,
+      "learning_rate": 4.959273341782848e-05,
+      "loss": 0.0883,
+      "step": 964
+    },
+    {
+      "epoch": 0.40811153358681873,
+      "grad_norm": 1.83567214012146,
+      "learning_rate": 4.9591888466413186e-05,
+      "loss": 0.0854,
+      "step": 966
+    },
+    {
+      "epoch": 0.4089564850021124,
+      "grad_norm": 1.9943451881408691,
+      "learning_rate": 4.959104351499789e-05,
+      "loss": 0.133,
+      "step": 968
+    },
+    {
+      "epoch": 0.409801436417406,
+      "grad_norm": 1.8926469087600708,
+      "learning_rate": 4.9590198563582593e-05,
+      "loss": 0.1113,
+      "step": 970
+    },
+    {
+      "epoch": 0.41064638783269963,
+      "grad_norm": 1.2133026123046875,
+      "learning_rate": 4.958935361216731e-05,
+      "loss": 0.0637,
+      "step": 972
+    },
+    {
+      "epoch": 0.41149133924799325,
+      "grad_norm": 1.2940943241119385,
+      "learning_rate": 4.958850866075201e-05,
+      "loss": 0.0829,
+      "step": 974
+    },
+    {
+      "epoch": 0.41233629066328686,
+      "grad_norm": 1.3455214500427246,
+      "learning_rate": 4.9587663709336715e-05,
+      "loss": 0.0953,
+      "step": 976
+    },
+    {
+      "epoch": 0.4131812420785805,
+      "grad_norm": 1.8440299034118652,
+      "learning_rate": 4.958681875792142e-05,
+      "loss": 0.1046,
+      "step": 978
+    },
+    {
+      "epoch": 0.4140261934938741,
+      "grad_norm": 1.57143235206604,
+      "learning_rate": 4.958597380650613e-05,
+      "loss": 0.1112,
+      "step": 980
+    },
+    {
+      "epoch": 0.4148711449091677,
+      "grad_norm": 1.4068124294281006,
+      "learning_rate": 4.958512885509084e-05,
+      "loss": 0.0852,
+      "step": 982
+    },
+    {
+      "epoch": 0.4157160963244613,
+      "grad_norm": 1.538854718208313,
+      "learning_rate": 4.9584283903675544e-05,
+      "loss": 0.0878,
+      "step": 984
+    },
+    {
+      "epoch": 0.41656104773975494,
+      "grad_norm": 0.881242573261261,
+      "learning_rate": 4.9583438952260244e-05,
+      "loss": 0.0786,
+      "step": 986
+    },
+    {
+      "epoch": 0.41740599915504856,
+      "grad_norm": 1.5759536027908325,
+      "learning_rate": 4.958259400084496e-05,
+      "loss": 0.0756,
+      "step": 988
+    },
+    {
+      "epoch": 0.41825095057034223,
+      "grad_norm": 1.311062216758728,
+      "learning_rate": 4.958174904942966e-05,
+      "loss": 0.0629,
+      "step": 990
+    },
+    {
+      "epoch": 0.41909590198563584,
+      "grad_norm": 1.1126432418823242,
+      "learning_rate": 4.9580904098014366e-05,
+      "loss": 0.0562,
+      "step": 992
+    },
+    {
+      "epoch": 0.41994085340092946,
+      "grad_norm": 1.553884744644165,
+      "learning_rate": 4.958005914659907e-05,
+      "loss": 0.1244,
+      "step": 994
+    },
+    {
+      "epoch": 0.4207858048162231,
+      "grad_norm": 2.2669692039489746,
+      "learning_rate": 4.957921419518378e-05,
+      "loss": 0.098,
+      "step": 996
+    },
+    {
+      "epoch": 0.4216307562315167,
+      "grad_norm": 1.335403323173523,
+      "learning_rate": 4.957836924376849e-05,
+      "loss": 0.0727,
+      "step": 998
+    },
+    {
+      "epoch": 0.4224757076468103,
+      "grad_norm": 3.8670544624328613,
+      "learning_rate": 4.9577524292353195e-05,
+      "loss": 0.1018,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4224757076468103,
+      "eval_accuracy": 0.6790826004483531,
+      "eval_cer": 0.09158675688312147,
+      "eval_loss": 0.20456762611865997,
+      "eval_runtime": 853.1585,
+      "eval_samples_per_second": 13.594,
+      "eval_steps_per_second": 0.425,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4233206590621039,
+      "grad_norm": 1.5914243459701538,
+      "learning_rate": 4.9576679340937895e-05,
+      "loss": 0.0926,
+      "step": 1002
+    },
+    {
+      "epoch": 0.42416561047739754,
+      "grad_norm": 2.1932218074798584,
+      "learning_rate": 4.957583438952261e-05,
+      "loss": 0.1381,
+      "step": 1004
+    },
+    {
+      "epoch": 0.42501056189269115,
+      "grad_norm": 1.216361165046692,
+      "learning_rate": 4.957498943810731e-05,
+      "loss": 0.0675,
+      "step": 1006
+    },
+    {
+      "epoch": 0.42585551330798477,
+      "grad_norm": 1.8609379529953003,
+      "learning_rate": 4.9574144486692023e-05,
+      "loss": 0.0891,
+      "step": 1008
+    },
+    {
+      "epoch": 0.42670046472327844,
+      "grad_norm": 1.5109552145004272,
+      "learning_rate": 4.9573299535276724e-05,
+      "loss": 0.0534,
+      "step": 1010
+    },
+    {
+      "epoch": 0.42754541613857205,
+      "grad_norm": 2.247575521469116,
+      "learning_rate": 4.957245458386143e-05,
+      "loss": 0.1475,
+      "step": 1012
+    },
+    {
+      "epoch": 0.42839036755386567,
+      "grad_norm": 1.9921953678131104,
+      "learning_rate": 4.957160963244614e-05,
+      "loss": 0.1092,
+      "step": 1014
+    },
+    {
+      "epoch": 0.4292353189691593,
+      "grad_norm": 1.308531641960144,
+      "learning_rate": 4.9570764681030845e-05,
+      "loss": 0.048,
+      "step": 1016
+    },
+    {
+      "epoch": 0.4300802703844529,
+      "grad_norm": 1.459812045097351,
+      "learning_rate": 4.956991972961555e-05,
+      "loss": 0.0829,
+      "step": 1018
+    },
+    {
+      "epoch": 0.4309252217997465,
+      "grad_norm": 1.1062469482421875,
+      "learning_rate": 4.956907477820026e-05,
+      "loss": 0.0761,
+      "step": 1020
+    },
+    {
+      "epoch": 0.43177017321504013,
+      "grad_norm": 0.9959312081336975,
+      "learning_rate": 4.956822982678496e-05,
+      "loss": 0.0776,
+      "step": 1022
+    },
+    {
+      "epoch": 0.43261512463033375,
+      "grad_norm": 2.045827627182007,
+      "learning_rate": 4.9567384875369674e-05,
+      "loss": 0.0805,
+      "step": 1024
+    },
+    {
+      "epoch": 0.43346007604562736,
+      "grad_norm": 1.678009271621704,
+      "learning_rate": 4.9566539923954375e-05,
+      "loss": 0.129,
+      "step": 1026
+    },
+    {
+      "epoch": 0.434305027460921,
+      "grad_norm": 1.8472874164581299,
+      "learning_rate": 4.956569497253908e-05,
+      "loss": 0.0689,
+      "step": 1028
+    },
+    {
+      "epoch": 0.4351499788762146,
+      "grad_norm": 1.6969095468521118,
+      "learning_rate": 4.956485002112379e-05,
+      "loss": 0.104,
+      "step": 1030
+    },
+    {
+      "epoch": 0.43599493029150826,
+      "grad_norm": 1.5743629932403564,
+      "learning_rate": 4.9564005069708496e-05,
+      "loss": 0.0823,
+      "step": 1032
+    },
+    {
+      "epoch": 0.4368398817068019,
+      "grad_norm": 1.9706130027770996,
+      "learning_rate": 4.9563160118293203e-05,
+      "loss": 0.1057,
+      "step": 1034
+    },
+    {
+      "epoch": 0.4376848331220955,
+      "grad_norm": 1.8118597269058228,
+      "learning_rate": 4.956231516687791e-05,
+      "loss": 0.0937,
+      "step": 1036
+    },
+    {
+      "epoch": 0.4385297845373891,
+      "grad_norm": 1.5070606470108032,
+      "learning_rate": 4.956147021546261e-05,
+      "loss": 0.0813,
+      "step": 1038
+    },
+    {
+      "epoch": 0.4393747359526827,
+      "grad_norm": 2.170083999633789,
+      "learning_rate": 4.9560625264047325e-05,
+      "loss": 0.1135,
+      "step": 1040
+    },
+    {
+      "epoch": 0.44021968736797634,
+      "grad_norm": 1.5932613611221313,
+      "learning_rate": 4.9559780312632025e-05,
+      "loss": 0.1101,
+      "step": 1042
+    },
+    {
+      "epoch": 0.44106463878326996,
+      "grad_norm": 1.2873493432998657,
+      "learning_rate": 4.955893536121673e-05,
+      "loss": 0.0622,
+      "step": 1044
+    },
+    {
+      "epoch": 0.4419095901985636,
+      "grad_norm": 1.2306898832321167,
+      "learning_rate": 4.955809040980144e-05,
+      "loss": 0.0947,
+      "step": 1046
+    },
+    {
+      "epoch": 0.4427545416138572,
+      "grad_norm": 1.1455901861190796,
+      "learning_rate": 4.955724545838615e-05,
+      "loss": 0.0694,
+      "step": 1048
+    },
+    {
+      "epoch": 0.4435994930291508,
+      "grad_norm": 1.363356351852417,
+      "learning_rate": 4.9556400506970854e-05,
+      "loss": 0.0776,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 2.466238260269165,
+      "learning_rate": 4.955555555555556e-05,
+      "loss": 0.1503,
+      "step": 1052
+    },
+    {
+      "epoch": 0.4452893958597381,
+      "grad_norm": 1.633049488067627,
+      "learning_rate": 4.955471060414026e-05,
+      "loss": 0.0902,
+      "step": 1054
+    },
+    {
+      "epoch": 0.4461343472750317,
+      "grad_norm": 1.3308897018432617,
+      "learning_rate": 4.9553865652724976e-05,
+      "loss": 0.1079,
+      "step": 1056
+    },
+    {
+      "epoch": 0.4469792986903253,
+      "grad_norm": 1.2919471263885498,
+      "learning_rate": 4.9553020701309676e-05,
+      "loss": 0.0943,
+      "step": 1058
+    },
+    {
+      "epoch": 0.44782425010561894,
+      "grad_norm": 2.11273193359375,
+      "learning_rate": 4.9552175749894383e-05,
+      "loss": 0.1257,
+      "step": 1060
+    },
+    {
+      "epoch": 0.44866920152091255,
+      "grad_norm": 1.6662590503692627,
+      "learning_rate": 4.955133079847909e-05,
+      "loss": 0.1103,
+      "step": 1062
+    },
+    {
+      "epoch": 0.44951415293620617,
+      "grad_norm": 1.2448116540908813,
+      "learning_rate": 4.95504858470638e-05,
+      "loss": 0.1195,
+      "step": 1064
+    },
+    {
+      "epoch": 0.4503591043514998,
+      "grad_norm": 1.083648920059204,
+      "learning_rate": 4.9549640895648505e-05,
+      "loss": 0.0719,
+      "step": 1066
+    },
+    {
+      "epoch": 0.4512040557667934,
+      "grad_norm": 1.6844433546066284,
+      "learning_rate": 4.954879594423321e-05,
+      "loss": 0.1019,
+      "step": 1068
+    },
+    {
+      "epoch": 0.452049007182087,
+      "grad_norm": 1.9984877109527588,
+      "learning_rate": 4.954795099281791e-05,
+      "loss": 0.1353,
+      "step": 1070
+    },
+    {
+      "epoch": 0.45289395859738063,
+      "grad_norm": 1.462343692779541,
+      "learning_rate": 4.954710604140263e-05,
+      "loss": 0.1035,
+      "step": 1072
+    },
+    {
+      "epoch": 0.45373891001267425,
+      "grad_norm": 2.0531160831451416,
+      "learning_rate": 4.954626108998733e-05,
+      "loss": 0.1279,
+      "step": 1074
+    },
+    {
+      "epoch": 0.4545838614279679,
+      "grad_norm": 3.241274833679199,
+      "learning_rate": 4.9545416138572034e-05,
+      "loss": 0.0799,
+      "step": 1076
+    },
+    {
+      "epoch": 0.45542881284326153,
+      "grad_norm": 1.5598374605178833,
+      "learning_rate": 4.954457118715674e-05,
+      "loss": 0.0958,
+      "step": 1078
+    },
+    {
+      "epoch": 0.45627376425855515,
+      "grad_norm": 0.751557469367981,
+      "learning_rate": 4.954372623574145e-05,
+      "loss": 0.0542,
+      "step": 1080
+    },
+    {
+      "epoch": 0.45711871567384876,
+      "grad_norm": 1.6809083223342896,
+      "learning_rate": 4.9542881284326156e-05,
+      "loss": 0.0907,
+      "step": 1082
+    },
+    {
+      "epoch": 0.4579636670891424,
+      "grad_norm": 1.2016775608062744,
+      "learning_rate": 4.954203633291086e-05,
+      "loss": 0.0635,
+      "step": 1084
+    },
+    {
+      "epoch": 0.458808618504436,
+      "grad_norm": 1.912942886352539,
+      "learning_rate": 4.9541191381495563e-05,
+      "loss": 0.1076,
+      "step": 1086
+    },
+    {
+      "epoch": 0.4596535699197296,
+      "grad_norm": 1.3507989645004272,
+      "learning_rate": 4.954034643008027e-05,
+      "loss": 0.0906,
+      "step": 1088
+    },
+    {
+      "epoch": 0.4604985213350232,
+      "grad_norm": 1.4633934497833252,
+      "learning_rate": 4.953950147866498e-05,
+      "loss": 0.1085,
+      "step": 1090
+    },
+    {
+      "epoch": 0.46134347275031684,
+      "grad_norm": 1.7697800397872925,
+      "learning_rate": 4.9538656527249685e-05,
+      "loss": 0.1049,
+      "step": 1092
+    },
+    {
+      "epoch": 0.46218842416561046,
+      "grad_norm": 1.7392349243164062,
+      "learning_rate": 4.953781157583439e-05,
+      "loss": 0.1521,
+      "step": 1094
+    },
+    {
+      "epoch": 0.46303337558090407,
+      "grad_norm": 1.0418444871902466,
+      "learning_rate": 4.953696662441909e-05,
+      "loss": 0.0685,
+      "step": 1096
+    },
+    {
+      "epoch": 0.46387832699619774,
+      "grad_norm": 1.2857134342193604,
+      "learning_rate": 4.953612167300381e-05,
+      "loss": 0.1004,
+      "step": 1098
+    },
+    {
+      "epoch": 0.46472327841149136,
+      "grad_norm": 1.4575908184051514,
+      "learning_rate": 4.953527672158851e-05,
+      "loss": 0.0841,
+      "step": 1100
+    },
+    {
+      "epoch": 0.46472327841149136,
+      "eval_accuracy": 0.6908087601310571,
+      "eval_cer": 0.0895048386954489,
+      "eval_loss": 0.19633953273296356,
+      "eval_runtime": 863.5349,
+      "eval_samples_per_second": 13.431,
+      "eval_steps_per_second": 0.42,
+      "step": 1100
+    },
+    {
+      "epoch": 0.465568229826785,
+      "grad_norm": 1.182370901107788,
+      "learning_rate": 4.9534431770173214e-05,
+      "loss": 0.085,
+      "step": 1102
+    },
+    {
+      "epoch": 0.4664131812420786,
+      "grad_norm": 1.535293698310852,
+      "learning_rate": 4.953358681875792e-05,
+      "loss": 0.077,
+      "step": 1104
+    },
+    {
+      "epoch": 0.4672581326573722,
+      "grad_norm": 1.1662472486495972,
+      "learning_rate": 4.953274186734263e-05,
+      "loss": 0.0968,
+      "step": 1106
+    },
+    {
+      "epoch": 0.4681030840726658,
+      "grad_norm": 3.085517406463623,
+      "learning_rate": 4.9531896915927336e-05,
+      "loss": 0.1435,
+      "step": 1108
+    },
+    {
+      "epoch": 0.46894803548795944,
+      "grad_norm": 1.6203645467758179,
+      "learning_rate": 4.953105196451204e-05,
+      "loss": 0.0944,
+      "step": 1110
+    },
+    {
+      "epoch": 0.46979298690325305,
+      "grad_norm": 1.4693485498428345,
+      "learning_rate": 4.9530207013096744e-05,
+      "loss": 0.0955,
+      "step": 1112
+    },
+    {
+      "epoch": 0.47063793831854667,
+      "grad_norm": 1.4041072130203247,
+      "learning_rate": 4.952936206168146e-05,
+      "loss": 0.0675,
+      "step": 1114
+    },
+    {
+      "epoch": 0.4714828897338403,
+      "grad_norm": 2.4356982707977295,
+      "learning_rate": 4.952851711026616e-05,
+      "loss": 0.0835,
+      "step": 1116
+    },
+    {
+      "epoch": 0.4723278411491339,
+      "grad_norm": 1.7264307737350464,
+      "learning_rate": 4.952767215885087e-05,
+      "loss": 0.1218,
+      "step": 1118
+    },
+    {
+      "epoch": 0.47317279256442757,
+      "grad_norm": 1.329499363899231,
+      "learning_rate": 4.952682720743557e-05,
+      "loss": 0.0959,
+      "step": 1120
+    },
+    {
+      "epoch": 0.4740177439797212,
+      "grad_norm": 1.0061687231063843,
+      "learning_rate": 4.952598225602028e-05,
+      "loss": 0.052,
+      "step": 1122
+    },
+    {
+      "epoch": 0.4748626953950148,
+      "grad_norm": 1.739823818206787,
+      "learning_rate": 4.952513730460499e-05,
+      "loss": 0.0817,
+      "step": 1124
+    },
+    {
+      "epoch": 0.4757076468103084,
+      "grad_norm": 1.7504117488861084,
+      "learning_rate": 4.9524292353189694e-05,
+      "loss": 0.0709,
+      "step": 1126
+    },
+    {
+      "epoch": 0.47655259822560203,
+      "grad_norm": 2.0645530223846436,
+      "learning_rate": 4.9523447401774394e-05,
+      "loss": 0.1052,
+      "step": 1128
+    },
+    {
+      "epoch": 0.47739754964089565,
+      "grad_norm": 1.7341753244400024,
+      "learning_rate": 4.952260245035911e-05,
+      "loss": 0.0761,
+      "step": 1130
+    },
+    {
+      "epoch": 0.47824250105618926,
+      "grad_norm": 1.0064074993133545,
+      "learning_rate": 4.952175749894381e-05,
+      "loss": 0.0867,
+      "step": 1132
+    },
+    {
+      "epoch": 0.4790874524714829,
+      "grad_norm": 2.0180881023406982,
+      "learning_rate": 4.952091254752852e-05,
+      "loss": 0.0961,
+      "step": 1134
+    },
+    {
+      "epoch": 0.4799324038867765,
+      "grad_norm": 1.345908761024475,
+      "learning_rate": 4.952006759611322e-05,
+      "loss": 0.1018,
+      "step": 1136
+    },
+    {
+      "epoch": 0.4807773553020701,
+      "grad_norm": 1.6427843570709229,
+      "learning_rate": 4.951922264469793e-05,
+      "loss": 0.1179,
+      "step": 1138
+    },
+    {
+      "epoch": 0.4816223067173637,
+      "grad_norm": 1.294453501701355,
+      "learning_rate": 4.951837769328264e-05,
+      "loss": 0.0795,
+      "step": 1140
+    },
+    {
+      "epoch": 0.4824672581326574,
+      "grad_norm": 1.0479216575622559,
+      "learning_rate": 4.9517532741867345e-05,
+      "loss": 0.075,
+      "step": 1142
+    },
+    {
+      "epoch": 0.483312209547951,
+      "grad_norm": 1.1413781642913818,
+      "learning_rate": 4.951668779045205e-05,
+      "loss": 0.0705,
+      "step": 1144
+    },
+    {
+      "epoch": 0.4841571609632446,
+      "grad_norm": 1.2139525413513184,
+      "learning_rate": 4.951584283903676e-05,
+      "loss": 0.075,
+      "step": 1146
+    },
+    {
+      "epoch": 0.48500211237853824,
+      "grad_norm": 1.3668715953826904,
+      "learning_rate": 4.951499788762146e-05,
+      "loss": 0.1006,
+      "step": 1148
+    },
+    {
+      "epoch": 0.48584706379383186,
+      "grad_norm": 1.5356974601745605,
+      "learning_rate": 4.9514152936206173e-05,
+      "loss": 0.0996,
+      "step": 1150
+    },
+    {
+      "epoch": 0.4866920152091255,
+      "grad_norm": 1.1368151903152466,
+      "learning_rate": 4.9513307984790874e-05,
+      "loss": 0.1125,
+      "step": 1152
+    },
+    {
+      "epoch": 0.4875369666244191,
+      "grad_norm": 1.661220908164978,
+      "learning_rate": 4.951246303337558e-05,
+      "loss": 0.0705,
+      "step": 1154
+    },
+    {
+      "epoch": 0.4883819180397127,
+      "grad_norm": 1.7183247804641724,
+      "learning_rate": 4.951161808196029e-05,
+      "loss": 0.086,
+      "step": 1156
+    },
+    {
+      "epoch": 0.4892268694550063,
+      "grad_norm": 1.1112664937973022,
+      "learning_rate": 4.9510773130544995e-05,
+      "loss": 0.0401,
+      "step": 1158
+    },
+    {
+      "epoch": 0.49007182087029993,
+      "grad_norm": 2.460999011993408,
+      "learning_rate": 4.95099281791297e-05,
+      "loss": 0.1188,
+      "step": 1160
+    },
+    {
+      "epoch": 0.4909167722855936,
+      "grad_norm": 1.5830085277557373,
+      "learning_rate": 4.950908322771441e-05,
+      "loss": 0.0991,
+      "step": 1162
+    },
+    {
+      "epoch": 0.4917617237008872,
+      "grad_norm": 1.55359947681427,
+      "learning_rate": 4.950823827629911e-05,
+      "loss": 0.0971,
+      "step": 1164
+    },
+    {
+      "epoch": 0.49260667511618084,
+      "grad_norm": 1.919572114944458,
+      "learning_rate": 4.9507393324883824e-05,
+      "loss": 0.0943,
+      "step": 1166
+    },
+    {
+      "epoch": 0.49345162653147445,
+      "grad_norm": 1.5410270690917969,
+      "learning_rate": 4.9506548373468525e-05,
+      "loss": 0.1009,
+      "step": 1168
+    },
+    {
+      "epoch": 0.49429657794676807,
+      "grad_norm": 1.9407888650894165,
+      "learning_rate": 4.950570342205323e-05,
+      "loss": 0.1136,
+      "step": 1170
+    },
+    {
+      "epoch": 0.4951415293620617,
+      "grad_norm": 1.451253056526184,
+      "learning_rate": 4.950485847063794e-05,
+      "loss": 0.0933,
+      "step": 1172
+    },
+    {
+      "epoch": 0.4959864807773553,
+      "grad_norm": 1.2166279554367065,
+      "learning_rate": 4.9504013519222646e-05,
+      "loss": 0.0745,
+      "step": 1174
+    },
+    {
+      "epoch": 0.4968314321926489,
+      "grad_norm": 1.455151081085205,
+      "learning_rate": 4.9503168567807353e-05,
+      "loss": 0.0863,
+      "step": 1176
+    },
+    {
+      "epoch": 0.49767638360794253,
+      "grad_norm": 1.4487229585647583,
+      "learning_rate": 4.950232361639206e-05,
+      "loss": 0.1104,
+      "step": 1178
+    },
+    {
+      "epoch": 0.49852133502323615,
+      "grad_norm": 1.898914098739624,
+      "learning_rate": 4.950147866497676e-05,
+      "loss": 0.0923,
+      "step": 1180
+    },
+    {
+      "epoch": 0.49936628643852976,
+      "grad_norm": 1.0237014293670654,
+      "learning_rate": 4.9500633713561475e-05,
+      "loss": 0.0685,
+      "step": 1182
+    },
+    {
+      "epoch": 0.5002112378538234,
+      "grad_norm": 2.124537944793701,
+      "learning_rate": 4.9499788762146176e-05,
+      "loss": 0.0698,
+      "step": 1184
+    },
+    {
+      "epoch": 0.501056189269117,
+      "grad_norm": 1.1420694589614868,
+      "learning_rate": 4.949894381073088e-05,
+      "loss": 0.0652,
+      "step": 1186
+    },
+    {
+      "epoch": 0.5019011406844106,
+      "grad_norm": 1.4151241779327393,
+      "learning_rate": 4.949809885931559e-05,
+      "loss": 0.0701,
+      "step": 1188
+    },
+    {
+      "epoch": 0.5027460920997042,
+      "grad_norm": 1.3990213871002197,
+      "learning_rate": 4.94972539079003e-05,
+      "loss": 0.0737,
+      "step": 1190
+    },
+    {
+      "epoch": 0.5035910435149978,
+      "grad_norm": 1.4167402982711792,
+      "learning_rate": 4.9496408956485004e-05,
+      "loss": 0.0812,
+      "step": 1192
+    },
+    {
+      "epoch": 0.5044359949302915,
+      "grad_norm": 1.370712399482727,
+      "learning_rate": 4.949556400506971e-05,
+      "loss": 0.0744,
+      "step": 1194
+    },
+    {
+      "epoch": 0.5052809463455852,
+      "grad_norm": 1.3453139066696167,
+      "learning_rate": 4.949471905365441e-05,
+      "loss": 0.0754,
+      "step": 1196
+    },
+    {
+      "epoch": 0.5061258977608788,
+      "grad_norm": 1.5076264142990112,
+      "learning_rate": 4.9493874102239126e-05,
+      "loss": 0.1056,
+      "step": 1198
+    },
+    {
+      "epoch": 0.5069708491761724,
+      "grad_norm": 1.5745254755020142,
+      "learning_rate": 4.9493029150823826e-05,
+      "loss": 0.1156,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5069708491761724,
+      "eval_accuracy": 0.6875323331608898,
+      "eval_cer": 0.0944185184858456,
+      "eval_loss": 0.20057915151119232,
+      "eval_runtime": 866.1613,
+      "eval_samples_per_second": 13.39,
+      "eval_steps_per_second": 0.419,
+      "step": 1200
+    },
+    {
+      "epoch": 0.507815800591466,
+      "grad_norm": 1.8199948072433472,
+      "learning_rate": 4.9492184199408534e-05,
+      "loss": 0.0937,
+      "step": 1202
+    },
+    {
+      "epoch": 0.5086607520067596,
+      "grad_norm": 1.4204751253128052,
+      "learning_rate": 4.949133924799324e-05,
+      "loss": 0.0713,
+      "step": 1204
+    },
+    {
+      "epoch": 0.5095057034220533,
+      "grad_norm": 1.4452489614486694,
+      "learning_rate": 4.949049429657795e-05,
+      "loss": 0.0746,
+      "step": 1206
+    },
+    {
+      "epoch": 0.5103506548373469,
+      "grad_norm": 1.4586948156356812,
+      "learning_rate": 4.9489649345162655e-05,
+      "loss": 0.0855,
+      "step": 1208
+    },
+    {
+      "epoch": 0.5111956062526405,
+      "grad_norm": 1.7247511148452759,
+      "learning_rate": 4.948880439374736e-05,
+      "loss": 0.116,
+      "step": 1210
+    },
+    {
+      "epoch": 0.5120405576679341,
+      "grad_norm": 1.3063023090362549,
+      "learning_rate": 4.948795944233206e-05,
+      "loss": 0.1308,
+      "step": 1212
+    },
+    {
+      "epoch": 0.5128855090832277,
+      "grad_norm": 1.9612244367599487,
+      "learning_rate": 4.948711449091678e-05,
+      "loss": 0.1132,
+      "step": 1214
+    },
+    {
+      "epoch": 0.5137304604985213,
+      "grad_norm": 1.2515413761138916,
+      "learning_rate": 4.948626953950148e-05,
+      "loss": 0.0853,
+      "step": 1216
+    },
+    {
+      "epoch": 0.514575411913815,
+      "grad_norm": 1.5823932886123657,
+      "learning_rate": 4.948542458808619e-05,
+      "loss": 0.0938,
+      "step": 1218
+    },
+    {
+      "epoch": 0.5154203633291086,
+      "grad_norm": 1.1870125532150269,
+      "learning_rate": 4.948457963667089e-05,
+      "loss": 0.0729,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5162653147444022,
+      "grad_norm": 1.5410425662994385,
+      "learning_rate": 4.94837346852556e-05,
+      "loss": 0.1073,
+      "step": 1222
+    },
+    {
+      "epoch": 0.5171102661596958,
+      "grad_norm": 1.2500526905059814,
+      "learning_rate": 4.9482889733840306e-05,
+      "loss": 0.0997,
+      "step": 1224
+    },
+    {
+      "epoch": 0.5179552175749894,
+      "grad_norm": 1.1845496892929077,
+      "learning_rate": 4.948204478242501e-05,
+      "loss": 0.0758,
+      "step": 1226
+    },
+    {
+      "epoch": 0.518800168990283,
+      "grad_norm": 1.2934046983718872,
+      "learning_rate": 4.9481199831009714e-05,
+      "loss": 0.0725,
+      "step": 1228
+    },
+    {
+      "epoch": 0.5196451204055766,
+      "grad_norm": 1.3043361902236938,
+      "learning_rate": 4.948035487959443e-05,
+      "loss": 0.0793,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5204900718208703,
+      "grad_norm": 1.6191684007644653,
+      "learning_rate": 4.947950992817913e-05,
+      "loss": 0.1012,
+      "step": 1232
+    },
+    {
+      "epoch": 0.5213350232361639,
+      "grad_norm": 1.4432382583618164,
+      "learning_rate": 4.947866497676384e-05,
+      "loss": 0.0988,
+      "step": 1234
+    },
+    {
+      "epoch": 0.5221799746514575,
+      "grad_norm": 2.0699760913848877,
+      "learning_rate": 4.947782002534854e-05,
+      "loss": 0.1041,
+      "step": 1236
+    },
+    {
+      "epoch": 0.5230249260667512,
+      "grad_norm": 1.1539101600646973,
+      "learning_rate": 4.947697507393325e-05,
+      "loss": 0.0571,
+      "step": 1238
+    },
+    {
+      "epoch": 0.5238698774820448,
+      "grad_norm": 1.6627494096755981,
+      "learning_rate": 4.947613012251796e-05,
+      "loss": 0.1198,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5247148288973384,
+      "grad_norm": 1.9634501934051514,
+      "learning_rate": 4.9475285171102664e-05,
+      "loss": 0.0775,
+      "step": 1242
+    },
+    {
+      "epoch": 0.5255597803126321,
+      "grad_norm": 2.4786336421966553,
+      "learning_rate": 4.947444021968737e-05,
+      "loss": 0.1651,
+      "step": 1244
+    },
+    {
+      "epoch": 0.5264047317279257,
+      "grad_norm": 1.4655958414077759,
+      "learning_rate": 4.947359526827208e-05,
+      "loss": 0.1079,
+      "step": 1246
+    },
+    {
+      "epoch": 0.5272496831432193,
+      "grad_norm": 1.3418774604797363,
+      "learning_rate": 4.947275031685678e-05,
+      "loss": 0.0967,
+      "step": 1248
+    },
+    {
+      "epoch": 0.5280946345585129,
+      "grad_norm": 0.9843947887420654,
+      "learning_rate": 4.947190536544149e-05,
+      "loss": 0.0791,
+      "step": 1250
+    },
+    {
+      "epoch": 0.5289395859738065,
+      "grad_norm": 1.4920053482055664,
+      "learning_rate": 4.947106041402619e-05,
+      "loss": 0.1008,
+      "step": 1252
+    },
+    {
+      "epoch": 0.5297845373891001,
+      "grad_norm": 1.4727956056594849,
+      "learning_rate": 4.94702154626109e-05,
+      "loss": 0.0876,
+      "step": 1254
+    },
+    {
+      "epoch": 0.5306294888043938,
+      "grad_norm": 1.8977808952331543,
+      "learning_rate": 4.946937051119561e-05,
+      "loss": 0.1361,
+      "step": 1256
+    },
+    {
+      "epoch": 0.5314744402196874,
+      "grad_norm": 1.456107258796692,
+      "learning_rate": 4.9468525559780315e-05,
+      "loss": 0.0878,
+      "step": 1258
+    },
+    {
+      "epoch": 0.532319391634981,
+      "grad_norm": 1.4788284301757812,
+      "learning_rate": 4.946768060836502e-05,
+      "loss": 0.0822,
+      "step": 1260
+    },
+    {
+      "epoch": 0.5331643430502746,
+      "grad_norm": 1.1848583221435547,
+      "learning_rate": 4.946683565694973e-05,
+      "loss": 0.0886,
+      "step": 1262
+    },
+    {
+      "epoch": 0.5340092944655682,
+      "grad_norm": 0.9302415251731873,
+      "learning_rate": 4.946599070553443e-05,
+      "loss": 0.0481,
+      "step": 1264
+    },
+    {
+      "epoch": 0.5348542458808618,
+      "grad_norm": 1.38499915599823,
+      "learning_rate": 4.9465145754119144e-05,
+      "loss": 0.0893,
+      "step": 1266
+    },
+    {
+      "epoch": 0.5356991972961554,
+      "grad_norm": 1.4074949026107788,
+      "learning_rate": 4.9464300802703844e-05,
+      "loss": 0.1123,
+      "step": 1268
+    },
+    {
+      "epoch": 0.5365441487114491,
+      "grad_norm": 1.7731876373291016,
+      "learning_rate": 4.946345585128855e-05,
+      "loss": 0.1557,
+      "step": 1270
+    },
+    {
+      "epoch": 0.5373891001267427,
+      "grad_norm": 1.3012690544128418,
+      "learning_rate": 4.946261089987326e-05,
+      "loss": 0.0454,
+      "step": 1272
+    },
+    {
+      "epoch": 0.5382340515420363,
+      "grad_norm": 1.2259783744812012,
+      "learning_rate": 4.9461765948457966e-05,
+      "loss": 0.0802,
+      "step": 1274
+    },
+    {
+      "epoch": 0.5390790029573299,
+      "grad_norm": 1.516466736793518,
+      "learning_rate": 4.946092099704267e-05,
+      "loss": 0.0518,
+      "step": 1276
+    },
+    {
+      "epoch": 0.5399239543726235,
+      "grad_norm": 1.8392707109451294,
+      "learning_rate": 4.946007604562738e-05,
+      "loss": 0.1264,
+      "step": 1278
+    },
+    {
+      "epoch": 0.5407689057879171,
+      "grad_norm": 1.0790072679519653,
+      "learning_rate": 4.945923109421208e-05,
+      "loss": 0.0821,
+      "step": 1280
+    },
+    {
+      "epoch": 0.5416138572032109,
+      "grad_norm": 1.1289983987808228,
+      "learning_rate": 4.9458386142796794e-05,
+      "loss": 0.1059,
+      "step": 1282
+    },
+    {
+      "epoch": 0.5424588086185045,
+      "grad_norm": 1.5571544170379639,
+      "learning_rate": 4.9457541191381495e-05,
+      "loss": 0.0883,
+      "step": 1284
+    },
+    {
+      "epoch": 0.5433037600337981,
+      "grad_norm": 1.0887531042099,
+      "learning_rate": 4.94566962399662e-05,
+      "loss": 0.0886,
+      "step": 1286
+    },
+    {
+      "epoch": 0.5441487114490917,
+      "grad_norm": 1.6409960985183716,
+      "learning_rate": 4.945585128855091e-05,
+      "loss": 0.1336,
+      "step": 1288
+    },
+    {
+      "epoch": 0.5449936628643853,
+      "grad_norm": 1.6622674465179443,
+      "learning_rate": 4.9455006337135616e-05,
+      "loss": 0.114,
+      "step": 1290
+    },
+    {
+      "epoch": 0.545838614279679,
+      "grad_norm": 1.7212952375411987,
+      "learning_rate": 4.9454161385720324e-05,
+      "loss": 0.1141,
+      "step": 1292
+    },
+    {
+      "epoch": 0.5466835656949726,
+      "grad_norm": 1.6510391235351562,
+      "learning_rate": 4.945331643430503e-05,
+      "loss": 0.1123,
+      "step": 1294
+    },
+    {
+      "epoch": 0.5475285171102662,
+      "grad_norm": 2.1044700145721436,
+      "learning_rate": 4.945247148288973e-05,
+      "loss": 0.0899,
+      "step": 1296
+    },
+    {
+      "epoch": 0.5483734685255598,
+      "grad_norm": 1.7959705591201782,
+      "learning_rate": 4.9451626531474445e-05,
+      "loss": 0.1213,
+      "step": 1298
+    },
+    {
+      "epoch": 0.5492184199408534,
+      "grad_norm": 0.6972851753234863,
+      "learning_rate": 4.9450781580059146e-05,
+      "loss": 0.0517,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5492184199408534,
+      "eval_accuracy": 0.7086566649422315,
+      "eval_cer": 0.08514692520091392,
+      "eval_loss": 0.189493328332901,
+      "eval_runtime": 866.8131,
+      "eval_samples_per_second": 13.38,
+      "eval_steps_per_second": 0.419,
+      "step": 1300
+    },
+    {
+      "epoch": 0.550063371356147,
+      "grad_norm": 1.3948755264282227,
+      "learning_rate": 4.944993662864385e-05,
+      "loss": 0.058,
+      "step": 1302
+    },
+    {
+      "epoch": 0.5509083227714406,
+      "grad_norm": 1.276227593421936,
+      "learning_rate": 4.944909167722856e-05,
+      "loss": 0.0629,
+      "step": 1304
+    },
+    {
+      "epoch": 0.5517532741867343,
+      "grad_norm": 1.310956597328186,
+      "learning_rate": 4.944824672581327e-05,
+      "loss": 0.1241,
+      "step": 1306
+    },
+    {
+      "epoch": 0.5525982256020279,
+      "grad_norm": 1.3557322025299072,
+      "learning_rate": 4.9447401774397974e-05,
+      "loss": 0.081,
+      "step": 1308
+    },
+    {
+      "epoch": 0.5534431770173215,
+      "grad_norm": 1.0258203744888306,
+      "learning_rate": 4.944655682298268e-05,
+      "loss": 0.0719,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5542881284326151,
+      "grad_norm": 1.1905845403671265,
+      "learning_rate": 4.944571187156738e-05,
+      "loss": 0.0712,
+      "step": 1312
+    },
+    {
+      "epoch": 0.5551330798479087,
+      "grad_norm": 1.176316499710083,
+      "learning_rate": 4.9444866920152096e-05,
+      "loss": 0.0587,
+      "step": 1314
+    },
+    {
+      "epoch": 0.5559780312632023,
+      "grad_norm": 9.898603439331055,
+      "learning_rate": 4.9444021968736796e-05,
+      "loss": 0.1039,
+      "step": 1316
+    },
+    {
+      "epoch": 0.556822982678496,
+      "grad_norm": 1.3090765476226807,
+      "learning_rate": 4.944317701732151e-05,
+      "loss": 0.0823,
+      "step": 1318
+    },
+    {
+      "epoch": 0.5576679340937896,
+      "grad_norm": 1.292490839958191,
+      "learning_rate": 4.944233206590621e-05,
+      "loss": 0.0615,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5585128855090832,
+      "grad_norm": 1.8449876308441162,
+      "learning_rate": 4.944148711449092e-05,
+      "loss": 0.0872,
+      "step": 1322
+    },
+    {
+      "epoch": 0.5593578369243768,
+      "grad_norm": 1.1056883335113525,
+      "learning_rate": 4.9440642163075625e-05,
+      "loss": 0.059,
+      "step": 1324
+    },
+    {
+      "epoch": 0.5602027883396705,
+      "grad_norm": 1.3612993955612183,
+      "learning_rate": 4.943979721166033e-05,
+      "loss": 0.0717,
+      "step": 1326
+    },
+    {
+      "epoch": 0.5610477397549641,
+      "grad_norm": 1.218751311302185,
+      "learning_rate": 4.943895226024503e-05,
+      "loss": 0.0785,
+      "step": 1328
+    },
+    {
+      "epoch": 0.5618926911702578,
+      "grad_norm": 1.318162441253662,
+      "learning_rate": 4.943810730882975e-05,
+      "loss": 0.0926,
+      "step": 1330
+    },
+    {
+      "epoch": 0.5627376425855514,
+      "grad_norm": 1.3321706056594849,
+      "learning_rate": 4.943726235741445e-05,
+      "loss": 0.0892,
+      "step": 1332
+    },
+    {
+      "epoch": 0.563582594000845,
+      "grad_norm": 1.7704854011535645,
+      "learning_rate": 4.943641740599916e-05,
+      "loss": 0.1265,
+      "step": 1334
+    },
+    {
+      "epoch": 0.5644275454161386,
+      "grad_norm": 1.5705130100250244,
+      "learning_rate": 4.943557245458386e-05,
+      "loss": 0.1006,
+      "step": 1336
+    },
+    {
+      "epoch": 0.5652724968314322,
+      "grad_norm": 1.3921126127243042,
+      "learning_rate": 4.943472750316857e-05,
+      "loss": 0.0755,
+      "step": 1338
+    },
+    {
+      "epoch": 0.5661174482467258,
+      "grad_norm": 1.9573496580123901,
+      "learning_rate": 4.9433882551753276e-05,
+      "loss": 0.1235,
+      "step": 1340
+    },
+    {
+      "epoch": 0.5669623996620194,
+      "grad_norm": 1.466693639755249,
+      "learning_rate": 4.943303760033798e-05,
+      "loss": 0.0701,
+      "step": 1342
+    },
+    {
+      "epoch": 0.5678073510773131,
+      "grad_norm": 0.9446965456008911,
+      "learning_rate": 4.943219264892269e-05,
+      "loss": 0.1052,
+      "step": 1344
+    },
+    {
+      "epoch": 0.5686523024926067,
+      "grad_norm": 1.8884165287017822,
+      "learning_rate": 4.94313476975074e-05,
+      "loss": 0.0726,
+      "step": 1346
+    },
+    {
+      "epoch": 0.5694972539079003,
+      "grad_norm": 2.2107183933258057,
+      "learning_rate": 4.94305027460921e-05,
+      "loss": 0.1353,
+      "step": 1348
+    },
+    {
+      "epoch": 0.5703422053231939,
+      "grad_norm": 1.5481027364730835,
+      "learning_rate": 4.942965779467681e-05,
+      "loss": 0.0856,
+      "step": 1350
+    },
+    {
+      "epoch": 0.5711871567384875,
+      "grad_norm": 1.549285650253296,
+      "learning_rate": 4.942881284326151e-05,
+      "loss": 0.1037,
+      "step": 1352
+    },
+    {
+      "epoch": 0.5720321081537811,
+      "grad_norm": 1.2403219938278198,
+      "learning_rate": 4.942796789184622e-05,
+      "loss": 0.0832,
+      "step": 1354
+    },
+    {
+      "epoch": 0.5728770595690748,
+      "grad_norm": 0.9897046089172363,
+      "learning_rate": 4.942712294043093e-05,
+      "loss": 0.0664,
+      "step": 1356
+    },
+    {
+      "epoch": 0.5737220109843684,
+      "grad_norm": 1.9358055591583252,
+      "learning_rate": 4.9426277989015634e-05,
+      "loss": 0.0824,
+      "step": 1358
+    },
+    {
+      "epoch": 0.574566962399662,
+      "grad_norm": 1.8482145071029663,
+      "learning_rate": 4.942543303760034e-05,
+      "loss": 0.1046,
+      "step": 1360
+    },
+    {
+      "epoch": 0.5754119138149556,
+      "grad_norm": 1.4269449710845947,
+      "learning_rate": 4.942458808618505e-05,
+      "loss": 0.0983,
+      "step": 1362
+    },
+    {
+      "epoch": 0.5762568652302492,
+      "grad_norm": 1.4967262744903564,
+      "learning_rate": 4.942374313476975e-05,
+      "loss": 0.0965,
+      "step": 1364
+    },
+    {
+      "epoch": 0.5771018166455428,
+      "grad_norm": 1.8186310529708862,
+      "learning_rate": 4.942289818335446e-05,
+      "loss": 0.092,
+      "step": 1366
+    },
+    {
+      "epoch": 0.5779467680608364,
+      "grad_norm": 1.330729603767395,
+      "learning_rate": 4.942205323193916e-05,
+      "loss": 0.0979,
+      "step": 1368
+    },
+    {
+      "epoch": 0.5787917194761302,
+      "grad_norm": 1.4221312999725342,
+      "learning_rate": 4.942120828052387e-05,
+      "loss": 0.0516,
+      "step": 1370
+    },
+    {
+      "epoch": 0.5796366708914238,
+      "grad_norm": 1.670209288597107,
+      "learning_rate": 4.942036332910858e-05,
+      "loss": 0.0888,
+      "step": 1372
+    },
+    {
+      "epoch": 0.5804816223067174,
+      "grad_norm": 1.2513757944107056,
+      "learning_rate": 4.9419518377693285e-05,
+      "loss": 0.106,
+      "step": 1374
+    },
+    {
+      "epoch": 0.581326573722011,
+      "grad_norm": 0.760837197303772,
+      "learning_rate": 4.941867342627799e-05,
+      "loss": 0.0722,
+      "step": 1376
+    },
+    {
+      "epoch": 0.5821715251373046,
+      "grad_norm": 1.0159239768981934,
+      "learning_rate": 4.94178284748627e-05,
+      "loss": 0.0757,
+      "step": 1378
+    },
+    {
+      "epoch": 0.5830164765525983,
+      "grad_norm": 1.9194375276565552,
+      "learning_rate": 4.94169835234474e-05,
+      "loss": 0.0796,
+      "step": 1380
+    },
+    {
+      "epoch": 0.5838614279678919,
+      "grad_norm": 1.4433456659317017,
+      "learning_rate": 4.9416138572032114e-05,
+      "loss": 0.076,
+      "step": 1382
+    },
+    {
+      "epoch": 0.5847063793831855,
+      "grad_norm": 1.4545176029205322,
+      "learning_rate": 4.9415293620616814e-05,
+      "loss": 0.0777,
+      "step": 1384
+    },
+    {
+      "epoch": 0.5855513307984791,
+      "grad_norm": 0.9777308106422424,
+      "learning_rate": 4.941444866920152e-05,
+      "loss": 0.0548,
+      "step": 1386
+    },
+    {
+      "epoch": 0.5863962822137727,
+      "grad_norm": 1.5005024671554565,
+      "learning_rate": 4.941360371778623e-05,
+      "loss": 0.1143,
+      "step": 1388
+    },
+    {
+      "epoch": 0.5872412336290663,
+      "grad_norm": 1.571602463722229,
+      "learning_rate": 4.9412758766370936e-05,
+      "loss": 0.0929,
+      "step": 1390
+    },
+    {
+      "epoch": 0.5880861850443599,
+      "grad_norm": 1.8419545888900757,
+      "learning_rate": 4.941191381495564e-05,
+      "loss": 0.1039,
+      "step": 1392
+    },
+    {
+      "epoch": 0.5889311364596536,
+      "grad_norm": 1.9891055822372437,
+      "learning_rate": 4.941106886354035e-05,
+      "loss": 0.0944,
+      "step": 1394
+    },
+    {
+      "epoch": 0.5897760878749472,
+      "grad_norm": 1.0187172889709473,
+      "learning_rate": 4.941022391212505e-05,
+      "loss": 0.1026,
+      "step": 1396
+    },
+    {
+      "epoch": 0.5906210392902408,
+      "grad_norm": 1.8273286819458008,
+      "learning_rate": 4.9409378960709764e-05,
+      "loss": 0.1627,
+      "step": 1398
+    },
+    {
+      "epoch": 0.5914659907055344,
+      "grad_norm": 1.9127320051193237,
+      "learning_rate": 4.9408534009294465e-05,
+      "loss": 0.0871,
+      "step": 1400
+    },
+    {
+      "epoch": 0.5914659907055344,
+      "eval_accuracy": 0.6934816347646146,
+      "eval_cer": 0.08890496396340763,
+      "eval_loss": 0.19768354296684265,
+      "eval_runtime": 860.1969,
+      "eval_samples_per_second": 13.483,
+      "eval_steps_per_second": 0.422,
+      "step": 1400
+    },
+    {
+      "epoch": 0.592310942120828,
+      "grad_norm": 1.432771921157837,
+      "learning_rate": 4.940768905787917e-05,
+      "loss": 0.12,
+      "step": 1402
+    },
+    {
+      "epoch": 0.5931558935361216,
+      "grad_norm": 1.443217158317566,
+      "learning_rate": 4.940684410646388e-05,
+      "loss": 0.0944,
+      "step": 1404
+    },
+    {
+      "epoch": 0.5940008449514153,
+      "grad_norm": 1.263504981994629,
+      "learning_rate": 4.9405999155048586e-05,
+      "loss": 0.0651,
+      "step": 1406
+    },
+    {
+      "epoch": 0.5948457963667089,
+      "grad_norm": 1.656178593635559,
+      "learning_rate": 4.9405154203633294e-05,
+      "loss": 0.1205,
+      "step": 1408
+    },
+    {
+      "epoch": 0.5956907477820025,
+      "grad_norm": 1.8223587274551392,
+      "learning_rate": 4.9404309252218e-05,
+      "loss": 0.0985,
+      "step": 1410
+    },
+    {
+      "epoch": 0.5965356991972962,
+      "grad_norm": 1.6137700080871582,
+      "learning_rate": 4.94034643008027e-05,
+      "loss": 0.1026,
+      "step": 1412
+    },
+    {
+      "epoch": 0.5973806506125898,
+      "grad_norm": 1.2118284702301025,
+      "learning_rate": 4.9402619349387415e-05,
+      "loss": 0.1008,
+      "step": 1414
+    },
+    {
+      "epoch": 0.5982256020278834,
+      "grad_norm": 1.541785717010498,
+      "learning_rate": 4.9401774397972116e-05,
+      "loss": 0.1128,
+      "step": 1416
+    },
+    {
+      "epoch": 0.5990705534431771,
+      "grad_norm": 1.5790534019470215,
+      "learning_rate": 4.940092944655683e-05,
+      "loss": 0.1197,
+      "step": 1418
+    },
+    {
+      "epoch": 0.5999155048584707,
+      "grad_norm": 2.461165189743042,
+      "learning_rate": 4.940008449514153e-05,
+      "loss": 0.1317,
+      "step": 1420
+    },
+    {
+      "epoch": 0.6007604562737643,
+      "grad_norm": 1.9079132080078125,
+      "learning_rate": 4.939923954372624e-05,
+      "loss": 0.0902,
+      "step": 1422
+    },
+    {
+      "epoch": 0.6016054076890579,
+      "grad_norm": 1.5427287817001343,
+      "learning_rate": 4.9398394592310944e-05,
+      "loss": 0.0782,
+      "step": 1424
+    },
+    {
+      "epoch": 0.6024503591043515,
+      "grad_norm": 1.3272346258163452,
+      "learning_rate": 4.939754964089565e-05,
+      "loss": 0.0631,
+      "step": 1426
+    },
+    {
+      "epoch": 0.6032953105196451,
+      "grad_norm": 1.1318906545639038,
+      "learning_rate": 4.939670468948035e-05,
+      "loss": 0.0485,
+      "step": 1428
+    },
+    {
+      "epoch": 0.6041402619349387,
+      "grad_norm": 2.0333144664764404,
+      "learning_rate": 4.9395859738065066e-05,
+      "loss": 0.1254,
+      "step": 1430
+    },
+    {
+      "epoch": 0.6049852133502324,
+      "grad_norm": 1.9606693983078003,
+      "learning_rate": 4.9395014786649766e-05,
+      "loss": 0.1414,
+      "step": 1432
+    },
+    {
+      "epoch": 0.605830164765526,
+      "grad_norm": 1.3295642137527466,
+      "learning_rate": 4.939416983523448e-05,
+      "loss": 0.0697,
+      "step": 1434
+    },
+    {
+      "epoch": 0.6066751161808196,
+      "grad_norm": 1.227980613708496,
+      "learning_rate": 4.939332488381918e-05,
+      "loss": 0.0723,
+      "step": 1436
+    },
+    {
+      "epoch": 0.6075200675961132,
+      "grad_norm": 1.926112413406372,
+      "learning_rate": 4.939247993240389e-05,
+      "loss": 0.136,
+      "step": 1438
+    },
+    {
+      "epoch": 0.6083650190114068,
+      "grad_norm": 1.1189000606536865,
+      "learning_rate": 4.9391634980988595e-05,
+      "loss": 0.0709,
+      "step": 1440
+    },
+    {
+      "epoch": 0.6092099704267004,
+      "grad_norm": 1.3692779541015625,
+      "learning_rate": 4.93907900295733e-05,
+      "loss": 0.0687,
+      "step": 1442
+    },
+    {
+      "epoch": 0.6100549218419941,
+      "grad_norm": 1.1951595544815063,
+      "learning_rate": 4.938994507815801e-05,
+      "loss": 0.0743,
+      "step": 1444
+    },
+    {
+      "epoch": 0.6108998732572877,
+      "grad_norm": 1.0123379230499268,
+      "learning_rate": 4.938910012674272e-05,
+      "loss": 0.0779,
+      "step": 1446
+    },
+    {
+      "epoch": 0.6117448246725813,
+      "grad_norm": 1.1633543968200684,
+      "learning_rate": 4.938825517532742e-05,
+      "loss": 0.0938,
+      "step": 1448
+    },
+    {
+      "epoch": 0.6125897760878749,
+      "grad_norm": 1.457838773727417,
+      "learning_rate": 4.938741022391213e-05,
+      "loss": 0.0932,
+      "step": 1450
+    },
+    {
+      "epoch": 0.6134347275031685,
+      "grad_norm": 1.9920662641525269,
+      "learning_rate": 4.938656527249683e-05,
+      "loss": 0.1435,
+      "step": 1452
+    },
+    {
+      "epoch": 0.6142796789184621,
+      "grad_norm": 1.591558814048767,
+      "learning_rate": 4.938572032108154e-05,
+      "loss": 0.0985,
+      "step": 1454
+    },
+    {
+      "epoch": 0.6151246303337559,
+      "grad_norm": 1.4729034900665283,
+      "learning_rate": 4.9384875369666246e-05,
+      "loss": 0.1336,
+      "step": 1456
+    },
+    {
+      "epoch": 0.6159695817490495,
+      "grad_norm": 1.2558763027191162,
+      "learning_rate": 4.938403041825095e-05,
+      "loss": 0.0699,
+      "step": 1458
+    },
+    {
+      "epoch": 0.6168145331643431,
+      "grad_norm": 1.9105608463287354,
+      "learning_rate": 4.938318546683566e-05,
+      "loss": 0.1193,
+      "step": 1460
+    },
+    {
+      "epoch": 0.6176594845796367,
+      "grad_norm": 1.7269306182861328,
+      "learning_rate": 4.938234051542037e-05,
+      "loss": 0.0777,
+      "step": 1462
+    },
+    {
+      "epoch": 0.6185044359949303,
+      "grad_norm": 1.195877194404602,
+      "learning_rate": 4.938149556400507e-05,
+      "loss": 0.0568,
+      "step": 1464
+    },
+    {
+      "epoch": 0.6193493874102239,
+      "grad_norm": 1.5723657608032227,
+      "learning_rate": 4.938065061258978e-05,
+      "loss": 0.1203,
+      "step": 1466
+    },
+    {
+      "epoch": 0.6201943388255176,
+      "grad_norm": 0.8827709555625916,
+      "learning_rate": 4.937980566117448e-05,
+      "loss": 0.0517,
+      "step": 1468
+    },
+    {
+      "epoch": 0.6210392902408112,
+      "grad_norm": 1.3201031684875488,
+      "learning_rate": 4.937896070975919e-05,
+      "loss": 0.0948,
+      "step": 1470
+    },
+    {
+      "epoch": 0.6218842416561048,
+      "grad_norm": 0.9939825534820557,
+      "learning_rate": 4.93781157583439e-05,
+      "loss": 0.0537,
+      "step": 1472
+    },
+    {
+      "epoch": 0.6227291930713984,
+      "grad_norm": 1.3091474771499634,
+      "learning_rate": 4.9377270806928604e-05,
+      "loss": 0.0888,
+      "step": 1474
+    },
+    {
+      "epoch": 0.623574144486692,
+      "grad_norm": 1.2489527463912964,
+      "learning_rate": 4.937642585551331e-05,
+      "loss": 0.1017,
+      "step": 1476
+    },
+    {
+      "epoch": 0.6244190959019856,
+      "grad_norm": 1.2982828617095947,
+      "learning_rate": 4.937558090409802e-05,
+      "loss": 0.0805,
+      "step": 1478
+    },
+    {
+      "epoch": 0.6252640473172792,
+      "grad_norm": 1.9336445331573486,
+      "learning_rate": 4.937473595268272e-05,
+      "loss": 0.1048,
+      "step": 1480
+    },
+    {
+      "epoch": 0.6261089987325729,
+      "grad_norm": 1.5748566389083862,
+      "learning_rate": 4.937389100126743e-05,
+      "loss": 0.0779,
+      "step": 1482
+    },
+    {
+      "epoch": 0.6269539501478665,
+      "grad_norm": 1.3378995656967163,
+      "learning_rate": 4.937304604985213e-05,
+      "loss": 0.082,
+      "step": 1484
+    },
+    {
+      "epoch": 0.6277989015631601,
+      "grad_norm": 1.0639171600341797,
+      "learning_rate": 4.937220109843684e-05,
+      "loss": 0.051,
+      "step": 1486
+    },
+    {
+      "epoch": 0.6286438529784537,
+      "grad_norm": 1.6450079679489136,
+      "learning_rate": 4.937135614702155e-05,
+      "loss": 0.1282,
+      "step": 1488
+    },
+    {
+      "epoch": 0.6294888043937473,
+      "grad_norm": 1.433846354484558,
+      "learning_rate": 4.9370511195606255e-05,
+      "loss": 0.091,
+      "step": 1490
+    },
+    {
+      "epoch": 0.6303337558090409,
+      "grad_norm": 1.7269262075424194,
+      "learning_rate": 4.936966624419096e-05,
+      "loss": 0.0804,
+      "step": 1492
+    },
+    {
+      "epoch": 0.6311787072243346,
+      "grad_norm": 1.0868664979934692,
+      "learning_rate": 4.936882129277567e-05,
+      "loss": 0.0454,
+      "step": 1494
+    },
+    {
+      "epoch": 0.6320236586396282,
+      "grad_norm": 1.883876085281372,
+      "learning_rate": 4.936797634136037e-05,
+      "loss": 0.0747,
+      "step": 1496
+    },
+    {
+      "epoch": 0.6328686100549218,
+      "grad_norm": 1.3019837141036987,
+      "learning_rate": 4.9367131389945084e-05,
+      "loss": 0.0904,
+      "step": 1498
+    },
+    {
+      "epoch": 0.6337135614702155,
+      "grad_norm": 1.3057466745376587,
+      "learning_rate": 4.9366286438529784e-05,
+      "loss": 0.0998,
+      "step": 1500
+    },
+    {
+      "epoch": 0.6337135614702155,
+      "eval_accuracy": 0.6930505259527505,
+      "eval_cer": 0.08929311820178727,
+      "eval_loss": 0.19271579384803772,
+      "eval_runtime": 852.7052,
+      "eval_samples_per_second": 13.601,
+      "eval_steps_per_second": 0.426,
+      "step": 1500
+    },
+    {
+      "epoch": 0.6345585128855091,
+      "grad_norm": 1.0389184951782227,
+      "learning_rate": 4.936544148711449e-05,
+      "loss": 0.063,
+      "step": 1502
+    },
+    {
+      "epoch": 0.6354034643008027,
+      "grad_norm": 1.1744555234909058,
+      "learning_rate": 4.93645965356992e-05,
+      "loss": 0.0855,
+      "step": 1504
+    },
+    {
+      "epoch": 0.6362484157160964,
+      "grad_norm": 1.6268651485443115,
+      "learning_rate": 4.9363751584283906e-05,
+      "loss": 0.0518,
+      "step": 1506
+    },
+    {
+      "epoch": 0.63709336713139,
+      "grad_norm": 1.0523124933242798,
+      "learning_rate": 4.936290663286861e-05,
+      "loss": 0.0617,
+      "step": 1508
+    },
+    {
+      "epoch": 0.6379383185466836,
+      "grad_norm": 1.024778962135315,
+      "learning_rate": 4.936206168145332e-05,
+      "loss": 0.1021,
+      "step": 1510
+    },
+    {
+      "epoch": 0.6387832699619772,
+      "grad_norm": 1.15168297290802,
+      "learning_rate": 4.936121673003802e-05,
+      "loss": 0.076,
+      "step": 1512
+    },
+    {
+      "epoch": 0.6396282213772708,
+      "grad_norm": 1.4460582733154297,
+      "learning_rate": 4.9360371778622734e-05,
+      "loss": 0.0866,
+      "step": 1514
+    },
+    {
+      "epoch": 0.6404731727925644,
+      "grad_norm": 1.1238740682601929,
+      "learning_rate": 4.9359526827207435e-05,
+      "loss": 0.0708,
+      "step": 1516
+    },
+    {
+      "epoch": 0.641318124207858,
+      "grad_norm": 1.175075650215149,
+      "learning_rate": 4.935868187579215e-05,
+      "loss": 0.0893,
+      "step": 1518
+    },
+    {
+      "epoch": 0.6421630756231517,
+      "grad_norm": 1.2177211046218872,
+      "learning_rate": 4.935783692437685e-05,
+      "loss": 0.0703,
+      "step": 1520
+    },
+    {
+      "epoch": 0.6430080270384453,
+      "grad_norm": 1.4435173273086548,
+      "learning_rate": 4.9356991972961556e-05,
+      "loss": 0.0947,
+      "step": 1522
+    },
+    {
+      "epoch": 0.6438529784537389,
+      "grad_norm": 2.4668965339660645,
+      "learning_rate": 4.9356147021546264e-05,
+      "loss": 0.0882,
+      "step": 1524
+    },
+    {
+      "epoch": 0.6446979298690325,
+      "grad_norm": 1.096189260482788,
+      "learning_rate": 4.935530207013097e-05,
+      "loss": 0.0671,
+      "step": 1526
+    },
+    {
+      "epoch": 0.6455428812843261,
+      "grad_norm": 1.430290937423706,
+      "learning_rate": 4.935445711871567e-05,
+      "loss": 0.1009,
+      "step": 1528
+    },
+    {
+      "epoch": 0.6463878326996197,
+      "grad_norm": 1.5221188068389893,
+      "learning_rate": 4.9353612167300385e-05,
+      "loss": 0.0941,
+      "step": 1530
+    },
+    {
+      "epoch": 0.6472327841149134,
+      "grad_norm": 1.0545361042022705,
+      "learning_rate": 4.9352767215885086e-05,
+      "loss": 0.0741,
+      "step": 1532
+    },
+    {
+      "epoch": 0.648077735530207,
+      "grad_norm": 1.3876293897628784,
+      "learning_rate": 4.93519222644698e-05,
+      "loss": 0.0963,
+      "step": 1534
+    },
+    {
+      "epoch": 0.6489226869455006,
+      "grad_norm": 1.425647497177124,
+      "learning_rate": 4.93510773130545e-05,
+      "loss": 0.0768,
+      "step": 1536
+    },
+    {
+      "epoch": 0.6497676383607942,
+      "grad_norm": 2.1672873497009277,
+      "learning_rate": 4.935023236163921e-05,
+      "loss": 0.0723,
+      "step": 1538
+    },
+    {
+      "epoch": 0.6506125897760878,
+      "grad_norm": 1.7907317876815796,
+      "learning_rate": 4.9349387410223914e-05,
+      "loss": 0.0776,
+      "step": 1540
+    },
+    {
+      "epoch": 0.6514575411913816,
+      "grad_norm": 1.7650822401046753,
+      "learning_rate": 4.934854245880862e-05,
+      "loss": 0.0895,
+      "step": 1542
+    },
+    {
+      "epoch": 0.6523024926066752,
+      "grad_norm": 1.2962666749954224,
+      "learning_rate": 4.934769750739333e-05,
+      "loss": 0.0661,
+      "step": 1544
+    },
+    {
+      "epoch": 0.6531474440219688,
+      "grad_norm": 1.4323803186416626,
+      "learning_rate": 4.9346852555978036e-05,
+      "loss": 0.1123,
+      "step": 1546
+    },
+    {
+      "epoch": 0.6539923954372624,
+      "grad_norm": 1.3358235359191895,
+      "learning_rate": 4.9346007604562736e-05,
+      "loss": 0.0559,
+      "step": 1548
+    },
+    {
+      "epoch": 0.654837346852556,
+      "grad_norm": 1.4420747756958008,
+      "learning_rate": 4.934516265314745e-05,
+      "loss": 0.0939,
+      "step": 1550
+    },
+    {
+      "epoch": 0.6556822982678496,
+      "grad_norm": 1.780863881111145,
+      "learning_rate": 4.934431770173215e-05,
+      "loss": 0.1005,
+      "step": 1552
+    },
+    {
+      "epoch": 0.6565272496831432,
+      "grad_norm": 1.2848925590515137,
+      "learning_rate": 4.934347275031686e-05,
+      "loss": 0.0594,
+      "step": 1554
+    },
+    {
+      "epoch": 0.6573722010984369,
+      "grad_norm": 2.379051923751831,
+      "learning_rate": 4.9342627798901565e-05,
+      "loss": 0.1367,
+      "step": 1556
+    },
+    {
+      "epoch": 0.6582171525137305,
+      "grad_norm": 1.4313839673995972,
+      "learning_rate": 4.934178284748627e-05,
+      "loss": 0.065,
+      "step": 1558
+    },
+    {
+      "epoch": 0.6590621039290241,
+      "grad_norm": 0.9839615821838379,
+      "learning_rate": 4.934093789607098e-05,
+      "loss": 0.0803,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6599070553443177,
+      "grad_norm": 2.2462387084960938,
+      "learning_rate": 4.934009294465569e-05,
+      "loss": 0.1078,
+      "step": 1562
+    },
+    {
+      "epoch": 0.6607520067596113,
+      "grad_norm": 1.4986047744750977,
+      "learning_rate": 4.933924799324039e-05,
+      "loss": 0.0967,
+      "step": 1564
+    },
+    {
+      "epoch": 0.6615969581749049,
+      "grad_norm": 1.3797781467437744,
+      "learning_rate": 4.93384030418251e-05,
+      "loss": 0.0631,
+      "step": 1566
+    },
+    {
+      "epoch": 0.6624419095901986,
+      "grad_norm": 2.0769073963165283,
+      "learning_rate": 4.93375580904098e-05,
+      "loss": 0.1283,
+      "step": 1568
+    },
+    {
+      "epoch": 0.6632868610054922,
+      "grad_norm": 0.8822681307792664,
+      "learning_rate": 4.933671313899451e-05,
+      "loss": 0.0781,
+      "step": 1570
+    },
+    {
+      "epoch": 0.6641318124207858,
+      "grad_norm": 1.346742033958435,
+      "learning_rate": 4.9335868187579216e-05,
+      "loss": 0.0983,
+      "step": 1572
+    },
+    {
+      "epoch": 0.6649767638360794,
+      "grad_norm": 1.2557731866836548,
+      "learning_rate": 4.933502323616392e-05,
+      "loss": 0.0548,
+      "step": 1574
+    },
+    {
+      "epoch": 0.665821715251373,
+      "grad_norm": 1.5680410861968994,
+      "learning_rate": 4.933417828474863e-05,
+      "loss": 0.0855,
+      "step": 1576
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 1.2320771217346191,
+      "learning_rate": 4.933333333333334e-05,
+      "loss": 0.0845,
+      "step": 1578
+    },
+    {
+      "epoch": 0.6675116180819602,
+      "grad_norm": 1.3170132637023926,
+      "learning_rate": 4.933248838191804e-05,
+      "loss": 0.067,
+      "step": 1580
+    },
+    {
+      "epoch": 0.6683565694972539,
+      "grad_norm": 0.8890075087547302,
+      "learning_rate": 4.933164343050275e-05,
+      "loss": 0.055,
+      "step": 1582
+    },
+    {
+      "epoch": 0.6692015209125475,
+      "grad_norm": 1.6186269521713257,
+      "learning_rate": 4.933079847908745e-05,
+      "loss": 0.0718,
+      "step": 1584
+    },
+    {
+      "epoch": 0.6700464723278412,
+      "grad_norm": 1.8913078308105469,
+      "learning_rate": 4.932995352767216e-05,
+      "loss": 0.1134,
+      "step": 1586
+    },
+    {
+      "epoch": 0.6708914237431348,
+      "grad_norm": 1.2571996450424194,
+      "learning_rate": 4.932910857625687e-05,
+      "loss": 0.0947,
+      "step": 1588
+    },
+    {
+      "epoch": 0.6717363751584284,
+      "grad_norm": 1.6301320791244507,
+      "learning_rate": 4.9328263624841574e-05,
+      "loss": 0.0976,
+      "step": 1590
+    },
+    {
+      "epoch": 0.672581326573722,
+      "grad_norm": 1.32038414478302,
+      "learning_rate": 4.932741867342628e-05,
+      "loss": 0.0967,
+      "step": 1592
+    },
+    {
+      "epoch": 0.6734262779890157,
+      "grad_norm": 0.9501339197158813,
+      "learning_rate": 4.932657372201099e-05,
+      "loss": 0.0651,
+      "step": 1594
+    },
+    {
+      "epoch": 0.6742712294043093,
+      "grad_norm": 1.1755475997924805,
+      "learning_rate": 4.932572877059569e-05,
+      "loss": 0.0383,
+      "step": 1596
+    },
+    {
+      "epoch": 0.6751161808196029,
+      "grad_norm": 1.402435302734375,
+      "learning_rate": 4.93248838191804e-05,
+      "loss": 0.0969,
+      "step": 1598
+    },
+    {
+      "epoch": 0.6759611322348965,
+      "grad_norm": 1.6371697187423706,
+      "learning_rate": 4.93240388677651e-05,
+      "loss": 0.1062,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6759611322348965,
+      "eval_accuracy": 0.6960682876357993,
+      "eval_cer": 0.08430886491350335,
+      "eval_loss": 0.1908927857875824,
+      "eval_runtime": 855.2921,
+      "eval_samples_per_second": 13.56,
+      "eval_steps_per_second": 0.424,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6768060836501901,
+      "grad_norm": 1.6198582649230957,
+      "learning_rate": 4.932319391634981e-05,
+      "loss": 0.0832,
+      "step": 1602
+    },
+    {
+      "epoch": 0.6776510350654837,
+      "grad_norm": 1.237261414527893,
+      "learning_rate": 4.932234896493452e-05,
+      "loss": 0.0604,
+      "step": 1604
+    },
+    {
+      "epoch": 0.6784959864807774,
+      "grad_norm": 1.5254098176956177,
+      "learning_rate": 4.9321504013519225e-05,
+      "loss": 0.0925,
+      "step": 1606
+    },
+    {
+      "epoch": 0.679340937896071,
+      "grad_norm": 1.858588457107544,
+      "learning_rate": 4.932065906210393e-05,
+      "loss": 0.1063,
+      "step": 1608
+    },
+    {
+      "epoch": 0.6801858893113646,
+      "grad_norm": 1.6307566165924072,
+      "learning_rate": 4.931981411068864e-05,
+      "loss": 0.1097,
+      "step": 1610
+    },
+    {
+      "epoch": 0.6810308407266582,
+      "grad_norm": 1.245123267173767,
+      "learning_rate": 4.931896915927334e-05,
+      "loss": 0.0468,
+      "step": 1612
+    },
+    {
+      "epoch": 0.6818757921419518,
+      "grad_norm": 1.2134346961975098,
+      "learning_rate": 4.9318124207858054e-05,
+      "loss": 0.0527,
+      "step": 1614
+    },
+    {
+      "epoch": 0.6827207435572454,
+      "grad_norm": 1.724666953086853,
+      "learning_rate": 4.9317279256442754e-05,
+      "loss": 0.067,
+      "step": 1616
+    },
+    {
+      "epoch": 0.683565694972539,
+      "grad_norm": 1.303871750831604,
+      "learning_rate": 4.931643430502747e-05,
+      "loss": 0.089,
+      "step": 1618
+    },
+    {
+      "epoch": 0.6844106463878327,
+      "grad_norm": 1.2575643062591553,
+      "learning_rate": 4.931558935361217e-05,
+      "loss": 0.0684,
+      "step": 1620
+    },
+    {
+      "epoch": 0.6852555978031263,
+      "grad_norm": 0.8384171724319458,
+      "learning_rate": 4.9314744402196876e-05,
+      "loss": 0.0726,
+      "step": 1622
+    },
+    {
+      "epoch": 0.6861005492184199,
+      "grad_norm": 1.2416189908981323,
+      "learning_rate": 4.931389945078158e-05,
+      "loss": 0.0519,
+      "step": 1624
+    },
+    {
+      "epoch": 0.6869455006337135,
+      "grad_norm": 1.111448884010315,
+      "learning_rate": 4.931305449936629e-05,
+      "loss": 0.0773,
+      "step": 1626
+    },
+    {
+      "epoch": 0.6877904520490071,
+      "grad_norm": 1.8492799997329712,
+      "learning_rate": 4.931220954795099e-05,
+      "loss": 0.0861,
+      "step": 1628
+    },
+    {
+      "epoch": 0.6886354034643009,
+      "grad_norm": 1.129862666130066,
+      "learning_rate": 4.9311364596535704e-05,
+      "loss": 0.063,
+      "step": 1630
+    },
+    {
+      "epoch": 0.6894803548795945,
+      "grad_norm": 1.5507149696350098,
+      "learning_rate": 4.9310519645120405e-05,
+      "loss": 0.0924,
+      "step": 1632
+    },
+    {
+      "epoch": 0.6903253062948881,
+      "grad_norm": 1.3672925233840942,
+      "learning_rate": 4.930967469370512e-05,
+      "loss": 0.098,
+      "step": 1634
+    },
+    {
+      "epoch": 0.6911702577101817,
+      "grad_norm": 1.8510890007019043,
+      "learning_rate": 4.930882974228982e-05,
+      "loss": 0.0784,
+      "step": 1636
+    },
+    {
+      "epoch": 0.6920152091254753,
+      "grad_norm": 1.4934600591659546,
+      "learning_rate": 4.9307984790874526e-05,
+      "loss": 0.1049,
+      "step": 1638
+    },
+    {
+      "epoch": 0.6928601605407689,
+      "grad_norm": 1.453052282333374,
+      "learning_rate": 4.9307139839459234e-05,
+      "loss": 0.0774,
+      "step": 1640
+    },
+    {
+      "epoch": 0.6937051119560625,
+      "grad_norm": 2.2687666416168213,
+      "learning_rate": 4.930629488804394e-05,
+      "loss": 0.1204,
+      "step": 1642
+    },
+    {
+      "epoch": 0.6945500633713562,
+      "grad_norm": 1.256612777709961,
+      "learning_rate": 4.930544993662865e-05,
+      "loss": 0.0879,
+      "step": 1644
+    },
+    {
+      "epoch": 0.6953950147866498,
+      "grad_norm": 1.2979470491409302,
+      "learning_rate": 4.9304604985213355e-05,
+      "loss": 0.0789,
+      "step": 1646
+    },
+    {
+      "epoch": 0.6962399662019434,
+      "grad_norm": 1.1799136400222778,
+      "learning_rate": 4.9303760033798056e-05,
+      "loss": 0.0836,
+      "step": 1648
+    },
+    {
+      "epoch": 0.697084917617237,
+      "grad_norm": 1.8143103122711182,
+      "learning_rate": 4.930291508238277e-05,
+      "loss": 0.0853,
+      "step": 1650
+    },
+    {
+      "epoch": 0.6979298690325306,
+      "grad_norm": 1.3416088819503784,
+      "learning_rate": 4.930207013096747e-05,
+      "loss": 0.0847,
+      "step": 1652
+    },
+    {
+      "epoch": 0.6987748204478242,
+      "grad_norm": 1.4958585500717163,
+      "learning_rate": 4.930122517955218e-05,
+      "loss": 0.0858,
+      "step": 1654
+    },
+    {
+      "epoch": 0.6996197718631179,
+      "grad_norm": 1.0670164823532104,
+      "learning_rate": 4.9300380228136884e-05,
+      "loss": 0.069,
+      "step": 1656
+    },
+    {
+      "epoch": 0.7004647232784115,
+      "grad_norm": 1.2606033086776733,
+      "learning_rate": 4.929953527672159e-05,
+      "loss": 0.0869,
+      "step": 1658
+    },
+    {
+      "epoch": 0.7013096746937051,
+      "grad_norm": 1.3816112279891968,
+      "learning_rate": 4.92986903253063e-05,
+      "loss": 0.0723,
+      "step": 1660
+    },
+    {
+      "epoch": 0.7021546261089987,
+      "grad_norm": 1.619861125946045,
+      "learning_rate": 4.9297845373891006e-05,
+      "loss": 0.1112,
+      "step": 1662
+    },
+    {
+      "epoch": 0.7029995775242923,
+      "grad_norm": 1.5590943098068237,
+      "learning_rate": 4.9297000422475706e-05,
+      "loss": 0.1013,
+      "step": 1664
+    },
+    {
+      "epoch": 0.7038445289395859,
+      "grad_norm": 1.2259608507156372,
+      "learning_rate": 4.929615547106042e-05,
+      "loss": 0.0855,
+      "step": 1666
+    },
+    {
+      "epoch": 0.7046894803548795,
+      "grad_norm": 1.231600284576416,
+      "learning_rate": 4.929531051964512e-05,
+      "loss": 0.0624,
+      "step": 1668
+    },
+    {
+      "epoch": 0.7055344317701732,
+      "grad_norm": 1.3199141025543213,
+      "learning_rate": 4.929446556822983e-05,
+      "loss": 0.0742,
+      "step": 1670
+    },
+    {
+      "epoch": 0.7063793831854668,
+      "grad_norm": 1.259050726890564,
+      "learning_rate": 4.9293620616814535e-05,
+      "loss": 0.0468,
+      "step": 1672
+    },
+    {
+      "epoch": 0.7072243346007605,
+      "grad_norm": 1.5264647006988525,
+      "learning_rate": 4.929277566539924e-05,
+      "loss": 0.0859,
+      "step": 1674
+    },
+    {
+      "epoch": 0.7080692860160541,
+      "grad_norm": 1.6180810928344727,
+      "learning_rate": 4.929193071398395e-05,
+      "loss": 0.0636,
+      "step": 1676
+    },
+    {
+      "epoch": 0.7089142374313477,
+      "grad_norm": 0.8615092635154724,
+      "learning_rate": 4.929108576256866e-05,
+      "loss": 0.0533,
+      "step": 1678
+    },
+    {
+      "epoch": 0.7097591888466414,
+      "grad_norm": 1.568564534187317,
+      "learning_rate": 4.929024081115336e-05,
+      "loss": 0.0794,
+      "step": 1680
+    },
+    {
+      "epoch": 0.710604140261935,
+      "grad_norm": 1.2899255752563477,
+      "learning_rate": 4.928939585973807e-05,
+      "loss": 0.0938,
+      "step": 1682
+    },
+    {
+      "epoch": 0.7114490916772286,
+      "grad_norm": 1.375060796737671,
+      "learning_rate": 4.928855090832277e-05,
+      "loss": 0.0741,
+      "step": 1684
+    },
+    {
+      "epoch": 0.7122940430925222,
+      "grad_norm": 1.205748200416565,
+      "learning_rate": 4.928770595690748e-05,
+      "loss": 0.0844,
+      "step": 1686
+    },
+    {
+      "epoch": 0.7131389945078158,
+      "grad_norm": 1.0106422901153564,
+      "learning_rate": 4.9286861005492186e-05,
+      "loss": 0.0594,
+      "step": 1688
+    },
+    {
+      "epoch": 0.7139839459231094,
+      "grad_norm": 1.668855905532837,
+      "learning_rate": 4.928601605407689e-05,
+      "loss": 0.1166,
+      "step": 1690
+    },
+    {
+      "epoch": 0.714828897338403,
+      "grad_norm": 1.094852089881897,
+      "learning_rate": 4.92851711026616e-05,
+      "loss": 0.0687,
+      "step": 1692
+    },
+    {
+      "epoch": 0.7156738487536967,
+      "grad_norm": 1.2307512760162354,
+      "learning_rate": 4.928432615124631e-05,
+      "loss": 0.0645,
+      "step": 1694
+    },
+    {
+      "epoch": 0.7165188001689903,
+      "grad_norm": 1.497880220413208,
+      "learning_rate": 4.928348119983101e-05,
+      "loss": 0.0998,
+      "step": 1696
+    },
+    {
+      "epoch": 0.7173637515842839,
+      "grad_norm": 1.546085238456726,
+      "learning_rate": 4.928263624841572e-05,
+      "loss": 0.0825,
+      "step": 1698
+    },
+    {
+      "epoch": 0.7182087029995775,
+      "grad_norm": 1.2614907026290894,
+      "learning_rate": 4.928179129700042e-05,
+      "loss": 0.0985,
+      "step": 1700
+    },
+    {
+      "epoch": 0.7182087029995775,
+      "eval_accuracy": 0.7092602172788411,
+      "eval_cer": 0.08464408902846758,
+      "eval_loss": 0.18686163425445557,
+      "eval_runtime": 849.9029,
+      "eval_samples_per_second": 13.646,
+      "eval_steps_per_second": 0.427,
+      "step": 1700
+    },
+    {
+      "epoch": 0.7190536544148711,
+      "grad_norm": 1.5614362955093384,
+      "learning_rate": 4.928094634558513e-05,
+      "loss": 0.0727,
+      "step": 1702
+    },
+    {
+      "epoch": 0.7198986058301647,
+      "grad_norm": 1.0478379726409912,
+      "learning_rate": 4.928010139416984e-05,
+      "loss": 0.0629,
+      "step": 1704
+    },
+    {
+      "epoch": 0.7207435572454584,
+      "grad_norm": 1.7387198209762573,
+      "learning_rate": 4.9279256442754544e-05,
+      "loss": 0.097,
+      "step": 1706
+    },
+    {
+      "epoch": 0.721588508660752,
+      "grad_norm": 1.8205493688583374,
+      "learning_rate": 4.927841149133925e-05,
+      "loss": 0.1223,
+      "step": 1708
+    },
+    {
+      "epoch": 0.7224334600760456,
+      "grad_norm": 1.3796732425689697,
+      "learning_rate": 4.927756653992396e-05,
+      "loss": 0.0864,
+      "step": 1710
+    },
+    {
+      "epoch": 0.7232784114913392,
+      "grad_norm": 1.7345277070999146,
+      "learning_rate": 4.927672158850866e-05,
+      "loss": 0.0995,
+      "step": 1712
+    },
+    {
+      "epoch": 0.7241233629066328,
+      "grad_norm": 0.9681912660598755,
+      "learning_rate": 4.927587663709337e-05,
+      "loss": 0.05,
+      "step": 1714
+    },
+    {
+      "epoch": 0.7249683143219265,
+      "grad_norm": 1.7644003629684448,
+      "learning_rate": 4.927503168567807e-05,
+      "loss": 0.1208,
+      "step": 1716
+    },
+    {
+      "epoch": 0.7258132657372202,
+      "grad_norm": 1.760344386100769,
+      "learning_rate": 4.927418673426279e-05,
+      "loss": 0.08,
+      "step": 1718
+    },
+    {
+      "epoch": 0.7266582171525138,
+      "grad_norm": 1.5169976949691772,
+      "learning_rate": 4.927334178284749e-05,
+      "loss": 0.0945,
+      "step": 1720
+    },
+    {
+      "epoch": 0.7275031685678074,
+      "grad_norm": 1.2515987157821655,
+      "learning_rate": 4.9272496831432195e-05,
+      "loss": 0.0964,
+      "step": 1722
+    },
+    {
+      "epoch": 0.728348119983101,
+      "grad_norm": 1.1804972887039185,
+      "learning_rate": 4.92716518800169e-05,
+      "loss": 0.0945,
+      "step": 1724
+    },
+    {
+      "epoch": 0.7291930713983946,
+      "grad_norm": 1.7442156076431274,
+      "learning_rate": 4.927080692860161e-05,
+      "loss": 0.1006,
+      "step": 1726
+    },
+    {
+      "epoch": 0.7300380228136882,
+      "grad_norm": 1.4811662435531616,
+      "learning_rate": 4.926996197718631e-05,
+      "loss": 0.1162,
+      "step": 1728
+    },
+    {
+      "epoch": 0.7308829742289819,
+      "grad_norm": 1.6214262247085571,
+      "learning_rate": 4.9269117025771024e-05,
+      "loss": 0.0813,
+      "step": 1730
+    },
+    {
+      "epoch": 0.7317279256442755,
+      "grad_norm": 1.4832850694656372,
+      "learning_rate": 4.9268272074355724e-05,
+      "loss": 0.1104,
+      "step": 1732
+    },
+    {
+      "epoch": 0.7325728770595691,
+      "grad_norm": 1.1491117477416992,
+      "learning_rate": 4.926742712294044e-05,
+      "loss": 0.0714,
+      "step": 1734
+    },
+    {
+      "epoch": 0.7334178284748627,
+      "grad_norm": 1.9180891513824463,
+      "learning_rate": 4.926658217152514e-05,
+      "loss": 0.0781,
+      "step": 1736
+    },
+    {
+      "epoch": 0.7342627798901563,
+      "grad_norm": 1.390456199645996,
+      "learning_rate": 4.9265737220109846e-05,
+      "loss": 0.1352,
+      "step": 1738
+    },
+    {
+      "epoch": 0.7351077313054499,
+      "grad_norm": 1.0331450700759888,
+      "learning_rate": 4.926489226869455e-05,
+      "loss": 0.0808,
+      "step": 1740
+    },
+    {
+      "epoch": 0.7359526827207435,
+      "grad_norm": 1.2622566223144531,
+      "learning_rate": 4.926404731727926e-05,
+      "loss": 0.055,
+      "step": 1742
+    },
+    {
+      "epoch": 0.7367976341360372,
+      "grad_norm": 1.1740683317184448,
+      "learning_rate": 4.926320236586397e-05,
+      "loss": 0.0493,
+      "step": 1744
+    },
+    {
+      "epoch": 0.7376425855513308,
+      "grad_norm": 1.8595513105392456,
+      "learning_rate": 4.9262357414448674e-05,
+      "loss": 0.1211,
+      "step": 1746
+    },
+    {
+      "epoch": 0.7384875369666244,
+      "grad_norm": 1.0683369636535645,
+      "learning_rate": 4.9261512463033375e-05,
+      "loss": 0.0583,
+      "step": 1748
+    },
+    {
+      "epoch": 0.739332488381918,
+      "grad_norm": 1.4157065153121948,
+      "learning_rate": 4.926066751161809e-05,
+      "loss": 0.0554,
+      "step": 1750
+    },
+    {
+      "epoch": 0.7401774397972116,
+      "grad_norm": 1.5190364122390747,
+      "learning_rate": 4.925982256020279e-05,
+      "loss": 0.0631,
+      "step": 1752
+    },
+    {
+      "epoch": 0.7410223912125052,
+      "grad_norm": 2.315302848815918,
+      "learning_rate": 4.9258977608787496e-05,
+      "loss": 0.1047,
+      "step": 1754
+    },
+    {
+      "epoch": 0.7418673426277989,
+      "grad_norm": 0.8665867447853088,
+      "learning_rate": 4.9258132657372204e-05,
+      "loss": 0.0612,
+      "step": 1756
+    },
+    {
+      "epoch": 0.7427122940430925,
+      "grad_norm": 2.6190311908721924,
+      "learning_rate": 4.925728770595691e-05,
+      "loss": 0.1227,
+      "step": 1758
+    },
+    {
+      "epoch": 0.7435572454583862,
+      "grad_norm": 1.6784873008728027,
+      "learning_rate": 4.925644275454162e-05,
+      "loss": 0.0606,
+      "step": 1760
+    },
+    {
+      "epoch": 0.7444021968736798,
+      "grad_norm": 1.4590693712234497,
+      "learning_rate": 4.9255597803126325e-05,
+      "loss": 0.0942,
+      "step": 1762
+    },
+    {
+      "epoch": 0.7452471482889734,
+      "grad_norm": 1.7136425971984863,
+      "learning_rate": 4.9254752851711026e-05,
+      "loss": 0.1141,
+      "step": 1764
+    },
+    {
+      "epoch": 0.746092099704267,
+      "grad_norm": 0.9946573376655579,
+      "learning_rate": 4.925390790029574e-05,
+      "loss": 0.0359,
+      "step": 1766
+    },
+    {
+      "epoch": 0.7469370511195607,
+      "grad_norm": 1.1567507982254028,
+      "learning_rate": 4.925306294888044e-05,
+      "loss": 0.055,
+      "step": 1768
+    },
+    {
+      "epoch": 0.7477820025348543,
+      "grad_norm": 1.3129973411560059,
+      "learning_rate": 4.925221799746515e-05,
+      "loss": 0.0654,
+      "step": 1770
+    },
+    {
+      "epoch": 0.7486269539501479,
+      "grad_norm": 2.021648645401001,
+      "learning_rate": 4.9251373046049854e-05,
+      "loss": 0.0965,
+      "step": 1772
+    },
+    {
+      "epoch": 0.7494719053654415,
+      "grad_norm": 1.6443380117416382,
+      "learning_rate": 4.925052809463456e-05,
+      "loss": 0.1043,
+      "step": 1774
+    },
+    {
+      "epoch": 0.7503168567807351,
+      "grad_norm": 1.7316606044769287,
+      "learning_rate": 4.924968314321927e-05,
+      "loss": 0.0941,
+      "step": 1776
+    },
+    {
+      "epoch": 0.7511618081960287,
+      "grad_norm": 1.0968530178070068,
+      "learning_rate": 4.9248838191803976e-05,
+      "loss": 0.0676,
+      "step": 1778
+    },
+    {
+      "epoch": 0.7520067596113224,
+      "grad_norm": 0.9221205115318298,
+      "learning_rate": 4.9247993240388676e-05,
+      "loss": 0.0763,
+      "step": 1780
+    },
+    {
+      "epoch": 0.752851711026616,
+      "grad_norm": 0.9875777363777161,
+      "learning_rate": 4.924714828897339e-05,
+      "loss": 0.0691,
+      "step": 1782
+    },
+    {
+      "epoch": 0.7536966624419096,
+      "grad_norm": 1.3123608827590942,
+      "learning_rate": 4.924630333755809e-05,
+      "loss": 0.0727,
+      "step": 1784
+    },
+    {
+      "epoch": 0.7545416138572032,
+      "grad_norm": 1.5449692010879517,
+      "learning_rate": 4.92454583861428e-05,
+      "loss": 0.0764,
+      "step": 1786
+    },
+    {
+      "epoch": 0.7553865652724968,
+      "grad_norm": 1.8921825885772705,
+      "learning_rate": 4.9244613434727505e-05,
+      "loss": 0.1129,
+      "step": 1788
+    },
+    {
+      "epoch": 0.7562315166877904,
+      "grad_norm": 1.119535207748413,
+      "learning_rate": 4.924376848331221e-05,
+      "loss": 0.0859,
+      "step": 1790
+    },
+    {
+      "epoch": 0.757076468103084,
+      "grad_norm": 1.4250285625457764,
+      "learning_rate": 4.924292353189692e-05,
+      "loss": 0.085,
+      "step": 1792
+    },
+    {
+      "epoch": 0.7579214195183777,
+      "grad_norm": 1.9496543407440186,
+      "learning_rate": 4.924207858048163e-05,
+      "loss": 0.0963,
+      "step": 1794
+    },
+    {
+      "epoch": 0.7587663709336713,
+      "grad_norm": 1.6314023733139038,
+      "learning_rate": 4.924123362906633e-05,
+      "loss": 0.079,
+      "step": 1796
+    },
+    {
+      "epoch": 0.7596113223489649,
+      "grad_norm": 1.3706735372543335,
+      "learning_rate": 4.924038867765104e-05,
+      "loss": 0.0813,
+      "step": 1798
+    },
+    {
+      "epoch": 0.7604562737642585,
+      "grad_norm": 1.6105067729949951,
+      "learning_rate": 4.923954372623574e-05,
+      "loss": 0.0793,
+      "step": 1800
+    },
+    {
+      "epoch": 0.7604562737642585,
+      "eval_accuracy": 0.7114157613381618,
+      "eval_cer": 0.08386778055170832,
+      "eval_loss": 0.18643230199813843,
+      "eval_runtime": 864.6046,
+      "eval_samples_per_second": 13.414,
+      "eval_steps_per_second": 0.42,
+      "step": 1800
+    },
+    {
+      "epoch": 0.7613012251795521,
+      "grad_norm": 1.3744542598724365,
+      "learning_rate": 4.923869877482045e-05,
+      "loss": 0.1043,
+      "step": 1802
+    },
+    {
+      "epoch": 0.7621461765948458,
+      "grad_norm": 1.2851905822753906,
+      "learning_rate": 4.9237853823405156e-05,
+      "loss": 0.0903,
+      "step": 1804
+    },
+    {
+      "epoch": 0.7629911280101395,
+      "grad_norm": 1.7337363958358765,
+      "learning_rate": 4.923700887198986e-05,
+      "loss": 0.0664,
+      "step": 1806
+    },
+    {
+      "epoch": 0.7638360794254331,
+      "grad_norm": 1.3106553554534912,
+      "learning_rate": 4.923616392057457e-05,
+      "loss": 0.0589,
+      "step": 1808
+    },
+    {
+      "epoch": 0.7646810308407267,
+      "grad_norm": 1.8125030994415283,
+      "learning_rate": 4.923531896915928e-05,
+      "loss": 0.1044,
+      "step": 1810
+    },
+    {
+      "epoch": 0.7655259822560203,
+      "grad_norm": 0.759692907333374,
+      "learning_rate": 4.923447401774398e-05,
+      "loss": 0.0423,
+      "step": 1812
+    },
+    {
+      "epoch": 0.7663709336713139,
+      "grad_norm": 1.4425475597381592,
+      "learning_rate": 4.923362906632869e-05,
+      "loss": 0.0742,
+      "step": 1814
+    },
+    {
+      "epoch": 0.7672158850866075,
+      "grad_norm": 1.7777689695358276,
+      "learning_rate": 4.923278411491339e-05,
+      "loss": 0.1033,
+      "step": 1816
+    },
+    {
+      "epoch": 0.7680608365019012,
+      "grad_norm": 1.4362133741378784,
+      "learning_rate": 4.9231939163498106e-05,
+      "loss": 0.0826,
+      "step": 1818
+    },
+    {
+      "epoch": 0.7689057879171948,
+      "grad_norm": 1.4595433473587036,
+      "learning_rate": 4.923109421208281e-05,
+      "loss": 0.0929,
+      "step": 1820
+    },
+    {
+      "epoch": 0.7697507393324884,
+      "grad_norm": 1.1807221174240112,
+      "learning_rate": 4.9230249260667514e-05,
+      "loss": 0.0803,
+      "step": 1822
+    },
+    {
+      "epoch": 0.770595690747782,
+      "grad_norm": 1.3756366968154907,
+      "learning_rate": 4.922940430925222e-05,
+      "loss": 0.0933,
+      "step": 1824
+    },
+    {
+      "epoch": 0.7714406421630756,
+      "grad_norm": 1.903717041015625,
+      "learning_rate": 4.922855935783693e-05,
+      "loss": 0.1241,
+      "step": 1826
+    },
+    {
+      "epoch": 0.7722855935783692,
+      "grad_norm": 1.8950796127319336,
+      "learning_rate": 4.922771440642163e-05,
+      "loss": 0.113,
+      "step": 1828
+    },
+    {
+      "epoch": 0.7731305449936628,
+      "grad_norm": 1.7587209939956665,
+      "learning_rate": 4.922686945500634e-05,
+      "loss": 0.0748,
+      "step": 1830
+    },
+    {
+      "epoch": 0.7739754964089565,
+      "grad_norm": 1.414786458015442,
+      "learning_rate": 4.922602450359104e-05,
+      "loss": 0.0958,
+      "step": 1832
+    },
+    {
+      "epoch": 0.7748204478242501,
+      "grad_norm": 1.7570750713348389,
+      "learning_rate": 4.922517955217576e-05,
+      "loss": 0.0989,
+      "step": 1834
+    },
+    {
+      "epoch": 0.7756653992395437,
+      "grad_norm": 0.9731429815292358,
+      "learning_rate": 4.922433460076046e-05,
+      "loss": 0.0659,
+      "step": 1836
+    },
+    {
+      "epoch": 0.7765103506548373,
+      "grad_norm": 1.2279034852981567,
+      "learning_rate": 4.9223489649345165e-05,
+      "loss": 0.0802,
+      "step": 1838
+    },
+    {
+      "epoch": 0.7773553020701309,
+      "grad_norm": 1.8844351768493652,
+      "learning_rate": 4.922264469792987e-05,
+      "loss": 0.1129,
+      "step": 1840
+    },
+    {
+      "epoch": 0.7782002534854245,
+      "grad_norm": 1.544244408607483,
+      "learning_rate": 4.922179974651458e-05,
+      "loss": 0.0857,
+      "step": 1842
+    },
+    {
+      "epoch": 0.7790452049007182,
+      "grad_norm": 1.417275071144104,
+      "learning_rate": 4.9220954795099286e-05,
+      "loss": 0.0735,
+      "step": 1844
+    },
+    {
+      "epoch": 0.7798901563160119,
+      "grad_norm": 1.0909690856933594,
+      "learning_rate": 4.9220109843683994e-05,
+      "loss": 0.0635,
+      "step": 1846
+    },
+    {
+      "epoch": 0.7807351077313055,
+      "grad_norm": 1.1615931987762451,
+      "learning_rate": 4.9219264892268694e-05,
+      "loss": 0.0632,
+      "step": 1848
+    },
+    {
+      "epoch": 0.7815800591465991,
+      "grad_norm": 1.4405494928359985,
+      "learning_rate": 4.921841994085341e-05,
+      "loss": 0.0848,
+      "step": 1850
+    },
+    {
+      "epoch": 0.7824250105618927,
+      "grad_norm": 0.6920841932296753,
+      "learning_rate": 4.921757498943811e-05,
+      "loss": 0.0404,
+      "step": 1852
+    },
+    {
+      "epoch": 0.7832699619771863,
+      "grad_norm": 0.9411987662315369,
+      "learning_rate": 4.9216730038022816e-05,
+      "loss": 0.0662,
+      "step": 1854
+    },
+    {
+      "epoch": 0.78411491339248,
+      "grad_norm": 1.6385960578918457,
+      "learning_rate": 4.921588508660752e-05,
+      "loss": 0.0484,
+      "step": 1856
+    },
+    {
+      "epoch": 0.7849598648077736,
+      "grad_norm": 1.655723214149475,
+      "learning_rate": 4.921504013519223e-05,
+      "loss": 0.1107,
+      "step": 1858
+    },
+    {
+      "epoch": 0.7858048162230672,
+      "grad_norm": 1.5613646507263184,
+      "learning_rate": 4.921419518377694e-05,
+      "loss": 0.0997,
+      "step": 1860
+    },
+    {
+      "epoch": 0.7866497676383608,
+      "grad_norm": 1.729367971420288,
+      "learning_rate": 4.9213350232361644e-05,
+      "loss": 0.0823,
+      "step": 1862
+    },
+    {
+      "epoch": 0.7874947190536544,
+      "grad_norm": 1.5075145959854126,
+      "learning_rate": 4.9212505280946345e-05,
+      "loss": 0.0687,
+      "step": 1864
+    },
+    {
+      "epoch": 0.788339670468948,
+      "grad_norm": 1.4084213972091675,
+      "learning_rate": 4.921166032953106e-05,
+      "loss": 0.0671,
+      "step": 1866
+    },
+    {
+      "epoch": 0.7891846218842417,
+      "grad_norm": 1.0190798044204712,
+      "learning_rate": 4.921081537811576e-05,
+      "loss": 0.0727,
+      "step": 1868
+    },
+    {
+      "epoch": 0.7900295732995353,
+      "grad_norm": 0.8679783940315247,
+      "learning_rate": 4.9209970426700466e-05,
+      "loss": 0.0804,
+      "step": 1870
+    },
+    {
+      "epoch": 0.7908745247148289,
+      "grad_norm": 1.3874335289001465,
+      "learning_rate": 4.9209125475285174e-05,
+      "loss": 0.0753,
+      "step": 1872
+    },
+    {
+      "epoch": 0.7917194761301225,
+      "grad_norm": 1.33711576461792,
+      "learning_rate": 4.920828052386988e-05,
+      "loss": 0.0826,
+      "step": 1874
+    },
+    {
+      "epoch": 0.7925644275454161,
+      "grad_norm": 1.362648606300354,
+      "learning_rate": 4.920743557245459e-05,
+      "loss": 0.0591,
+      "step": 1876
+    },
+    {
+      "epoch": 0.7934093789607097,
+      "grad_norm": 1.0619179010391235,
+      "learning_rate": 4.9206590621039295e-05,
+      "loss": 0.0495,
+      "step": 1878
+    },
+    {
+      "epoch": 0.7942543303760033,
+      "grad_norm": 2.31254506111145,
+      "learning_rate": 4.9205745669623996e-05,
+      "loss": 0.0985,
+      "step": 1880
+    },
+    {
+      "epoch": 0.795099281791297,
+      "grad_norm": 1.3235636949539185,
+      "learning_rate": 4.920490071820871e-05,
+      "loss": 0.1215,
+      "step": 1882
+    },
+    {
+      "epoch": 0.7959442332065906,
+      "grad_norm": 1.0667047500610352,
+      "learning_rate": 4.920405576679341e-05,
+      "loss": 0.0401,
+      "step": 1884
+    },
+    {
+      "epoch": 0.7967891846218842,
+      "grad_norm": 1.7155953645706177,
+      "learning_rate": 4.920321081537812e-05,
+      "loss": 0.087,
+      "step": 1886
+    },
+    {
+      "epoch": 0.7976341360371778,
+      "grad_norm": 1.1526762247085571,
+      "learning_rate": 4.9202365863962824e-05,
+      "loss": 0.0752,
+      "step": 1888
+    },
+    {
+      "epoch": 0.7984790874524715,
+      "grad_norm": 1.039081335067749,
+      "learning_rate": 4.920152091254753e-05,
+      "loss": 0.0882,
+      "step": 1890
+    },
+    {
+      "epoch": 0.7993240388677652,
+      "grad_norm": 1.4192873239517212,
+      "learning_rate": 4.920067596113224e-05,
+      "loss": 0.0692,
+      "step": 1892
+    },
+    {
+      "epoch": 0.8001689902830588,
+      "grad_norm": 0.9547431468963623,
+      "learning_rate": 4.9199831009716946e-05,
+      "loss": 0.0972,
+      "step": 1894
+    },
+    {
+      "epoch": 0.8010139416983524,
+      "grad_norm": 1.382888674736023,
+      "learning_rate": 4.9198986058301646e-05,
+      "loss": 0.1152,
+      "step": 1896
+    },
+    {
+      "epoch": 0.801858893113646,
+      "grad_norm": 1.5938488245010376,
+      "learning_rate": 4.919814110688636e-05,
+      "loss": 0.0875,
+      "step": 1898
+    },
+    {
+      "epoch": 0.8027038445289396,
+      "grad_norm": 1.211908221244812,
+      "learning_rate": 4.919729615547106e-05,
+      "loss": 0.0808,
+      "step": 1900
+    },
+    {
+      "epoch": 0.8027038445289396,
+      "eval_accuracy": 0.7140886359717192,
+      "eval_cer": 0.08334730100479018,
+      "eval_loss": 0.18288320302963257,
+      "eval_runtime": 852.7471,
+      "eval_samples_per_second": 13.601,
+      "eval_steps_per_second": 0.426,
+      "step": 1900
+    },
+    {
+      "epoch": 0.8035487959442332,
+      "grad_norm": 1.273587703704834,
+      "learning_rate": 4.919645120405577e-05,
+      "loss": 0.0567,
+      "step": 1902
+    },
+    {
+      "epoch": 0.8043937473595268,
+      "grad_norm": 1.4771486520767212,
+      "learning_rate": 4.9195606252640475e-05,
+      "loss": 0.0687,
+      "step": 1904
+    },
+    {
+      "epoch": 0.8052386987748205,
+      "grad_norm": 1.6391814947128296,
+      "learning_rate": 4.919476130122518e-05,
+      "loss": 0.0936,
+      "step": 1906
+    },
+    {
+      "epoch": 0.8060836501901141,
+      "grad_norm": 1.199585199356079,
+      "learning_rate": 4.919391634980989e-05,
+      "loss": 0.0855,
+      "step": 1908
+    },
+    {
+      "epoch": 0.8069286016054077,
+      "grad_norm": 1.0377155542373657,
+      "learning_rate": 4.91930713983946e-05,
+      "loss": 0.0669,
+      "step": 1910
+    },
+    {
+      "epoch": 0.8077735530207013,
+      "grad_norm": 1.5404986143112183,
+      "learning_rate": 4.91922264469793e-05,
+      "loss": 0.1058,
+      "step": 1912
+    },
+    {
+      "epoch": 0.8086185044359949,
+      "grad_norm": 1.612430214881897,
+      "learning_rate": 4.919138149556401e-05,
+      "loss": 0.074,
+      "step": 1914
+    },
+    {
+      "epoch": 0.8094634558512885,
+      "grad_norm": 1.8857512474060059,
+      "learning_rate": 4.919053654414871e-05,
+      "loss": 0.0916,
+      "step": 1916
+    },
+    {
+      "epoch": 0.8103084072665822,
+      "grad_norm": 1.3807179927825928,
+      "learning_rate": 4.9189691592733426e-05,
+      "loss": 0.0728,
+      "step": 1918
+    },
+    {
+      "epoch": 0.8111533586818758,
+      "grad_norm": 1.6425131559371948,
+      "learning_rate": 4.9188846641318126e-05,
+      "loss": 0.1021,
+      "step": 1920
+    },
+    {
+      "epoch": 0.8119983100971694,
+      "grad_norm": 1.344103217124939,
+      "learning_rate": 4.918800168990283e-05,
+      "loss": 0.0515,
+      "step": 1922
+    },
+    {
+      "epoch": 0.812843261512463,
+      "grad_norm": 1.3962315320968628,
+      "learning_rate": 4.918715673848754e-05,
+      "loss": 0.0619,
+      "step": 1924
+    },
+    {
+      "epoch": 0.8136882129277566,
+      "grad_norm": 1.2917231321334839,
+      "learning_rate": 4.918631178707225e-05,
+      "loss": 0.0795,
+      "step": 1926
+    },
+    {
+      "epoch": 0.8145331643430502,
+      "grad_norm": 2.180860996246338,
+      "learning_rate": 4.918546683565695e-05,
+      "loss": 0.1103,
+      "step": 1928
+    },
+    {
+      "epoch": 0.8153781157583438,
+      "grad_norm": 2.4335546493530273,
+      "learning_rate": 4.918462188424166e-05,
+      "loss": 0.1115,
+      "step": 1930
+    },
+    {
+      "epoch": 0.8162230671736375,
+      "grad_norm": 1.399709939956665,
+      "learning_rate": 4.918377693282636e-05,
+      "loss": 0.0957,
+      "step": 1932
+    },
+    {
+      "epoch": 0.8170680185889312,
+      "grad_norm": 1.5264992713928223,
+      "learning_rate": 4.9182931981411076e-05,
+      "loss": 0.0728,
+      "step": 1934
+    },
+    {
+      "epoch": 0.8179129700042248,
+      "grad_norm": 1.3386390209197998,
+      "learning_rate": 4.918208702999578e-05,
+      "loss": 0.0642,
+      "step": 1936
+    },
+    {
+      "epoch": 0.8187579214195184,
+      "grad_norm": 2.438541889190674,
+      "learning_rate": 4.9181242078580484e-05,
+      "loss": 0.0949,
+      "step": 1938
+    },
+    {
+      "epoch": 0.819602872834812,
+      "grad_norm": 0.7417229413986206,
+      "learning_rate": 4.918039712716519e-05,
+      "loss": 0.029,
+      "step": 1940
+    },
+    {
+      "epoch": 0.8204478242501057,
+      "grad_norm": 1.8493363857269287,
+      "learning_rate": 4.91795521757499e-05,
+      "loss": 0.096,
+      "step": 1942
+    },
+    {
+      "epoch": 0.8212927756653993,
+      "grad_norm": 0.9756290912628174,
+      "learning_rate": 4.9178707224334606e-05,
+      "loss": 0.064,
+      "step": 1944
+    },
+    {
+      "epoch": 0.8221377270806929,
+      "grad_norm": 1.0960304737091064,
+      "learning_rate": 4.917786227291931e-05,
+      "loss": 0.0671,
+      "step": 1946
+    },
+    {
+      "epoch": 0.8229826784959865,
+      "grad_norm": 1.3161073923110962,
+      "learning_rate": 4.917701732150401e-05,
+      "loss": 0.0671,
+      "step": 1948
+    },
+    {
+      "epoch": 0.8238276299112801,
+      "grad_norm": 0.829484224319458,
+      "learning_rate": 4.917617237008873e-05,
+      "loss": 0.0633,
+      "step": 1950
+    },
+    {
+      "epoch": 0.8246725813265737,
+      "grad_norm": 1.1128253936767578,
+      "learning_rate": 4.917532741867343e-05,
+      "loss": 0.0658,
+      "step": 1952
+    },
+    {
+      "epoch": 0.8255175327418673,
+      "grad_norm": 1.2087801694869995,
+      "learning_rate": 4.9174482467258135e-05,
+      "loss": 0.0688,
+      "step": 1954
+    },
+    {
+      "epoch": 0.826362484157161,
+      "grad_norm": 1.6095219850540161,
+      "learning_rate": 4.917363751584284e-05,
+      "loss": 0.1243,
+      "step": 1956
+    },
+    {
+      "epoch": 0.8272074355724546,
+      "grad_norm": 1.4487024545669556,
+      "learning_rate": 4.917279256442755e-05,
+      "loss": 0.0848,
+      "step": 1958
+    },
+    {
+      "epoch": 0.8280523869877482,
+      "grad_norm": 1.430047869682312,
+      "learning_rate": 4.9171947613012256e-05,
+      "loss": 0.0731,
+      "step": 1960
+    },
+    {
+      "epoch": 0.8288973384030418,
+      "grad_norm": 1.377947449684143,
+      "learning_rate": 4.9171102661596964e-05,
+      "loss": 0.0692,
+      "step": 1962
+    },
+    {
+      "epoch": 0.8297422898183354,
+      "grad_norm": 1.4219231605529785,
+      "learning_rate": 4.9170257710181664e-05,
+      "loss": 0.0585,
+      "step": 1964
+    },
+    {
+      "epoch": 0.830587241233629,
+      "grad_norm": 1.647648811340332,
+      "learning_rate": 4.916941275876638e-05,
+      "loss": 0.0719,
+      "step": 1966
+    },
+    {
+      "epoch": 0.8314321926489227,
+      "grad_norm": 1.743295669555664,
+      "learning_rate": 4.916856780735108e-05,
+      "loss": 0.0811,
+      "step": 1968
+    },
+    {
+      "epoch": 0.8322771440642163,
+      "grad_norm": 1.3732768297195435,
+      "learning_rate": 4.9167722855935786e-05,
+      "loss": 0.0942,
+      "step": 1970
+    },
+    {
+      "epoch": 0.8331220954795099,
+      "grad_norm": 2.2335240840911865,
+      "learning_rate": 4.916687790452049e-05,
+      "loss": 0.1215,
+      "step": 1972
+    },
+    {
+      "epoch": 0.8339670468948035,
+      "grad_norm": 1.5069928169250488,
+      "learning_rate": 4.91660329531052e-05,
+      "loss": 0.0673,
+      "step": 1974
+    },
+    {
+      "epoch": 0.8348119983100971,
+      "grad_norm": 1.2649790048599243,
+      "learning_rate": 4.916518800168991e-05,
+      "loss": 0.0851,
+      "step": 1976
+    },
+    {
+      "epoch": 0.8356569497253908,
+      "grad_norm": 1.7199008464813232,
+      "learning_rate": 4.9164343050274614e-05,
+      "loss": 0.0936,
+      "step": 1978
+    },
+    {
+      "epoch": 0.8365019011406845,
+      "grad_norm": 0.6023948192596436,
+      "learning_rate": 4.9163498098859315e-05,
+      "loss": 0.0624,
+      "step": 1980
+    },
+    {
+      "epoch": 0.8373468525559781,
+      "grad_norm": 1.3942101001739502,
+      "learning_rate": 4.916265314744403e-05,
+      "loss": 0.0832,
+      "step": 1982
+    },
+    {
+      "epoch": 0.8381918039712717,
+      "grad_norm": 0.856122612953186,
+      "learning_rate": 4.916180819602873e-05,
+      "loss": 0.0485,
+      "step": 1984
+    },
+    {
+      "epoch": 0.8390367553865653,
+      "grad_norm": 1.3273183107376099,
+      "learning_rate": 4.9160963244613437e-05,
+      "loss": 0.0634,
+      "step": 1986
+    },
+    {
+      "epoch": 0.8398817068018589,
+      "grad_norm": 1.4347068071365356,
+      "learning_rate": 4.9160118293198144e-05,
+      "loss": 0.0599,
+      "step": 1988
+    },
+    {
+      "epoch": 0.8407266582171525,
+      "grad_norm": 1.5340876579284668,
+      "learning_rate": 4.915927334178285e-05,
+      "loss": 0.0607,
+      "step": 1990
+    },
+    {
+      "epoch": 0.8415716096324461,
+      "grad_norm": 1.6133233308792114,
+      "learning_rate": 4.915842839036756e-05,
+      "loss": 0.076,
+      "step": 1992
+    },
+    {
+      "epoch": 0.8424165610477398,
+      "grad_norm": 2.004894971847534,
+      "learning_rate": 4.9157583438952265e-05,
+      "loss": 0.1015,
+      "step": 1994
+    },
+    {
+      "epoch": 0.8432615124630334,
+      "grad_norm": 1.0740406513214111,
+      "learning_rate": 4.9156738487536966e-05,
+      "loss": 0.0576,
+      "step": 1996
+    },
+    {
+      "epoch": 0.844106463878327,
+      "grad_norm": 1.0599428415298462,
+      "learning_rate": 4.915589353612168e-05,
+      "loss": 0.0561,
+      "step": 1998
+    },
+    {
+      "epoch": 0.8449514152936206,
+      "grad_norm": 1.134386420249939,
+      "learning_rate": 4.915504858470638e-05,
+      "loss": 0.0546,
+      "step": 2000
+    },
+    {
+      "epoch": 0.8449514152936206,
+      "eval_accuracy": 0.7053802379720642,
+      "eval_cer": 0.0823592720343693,
+      "eval_loss": 0.18670396506786346,
+      "eval_runtime": 843.4731,
+      "eval_samples_per_second": 13.75,
+      "eval_steps_per_second": 0.43,
+      "step": 2000
+    },
+    {
+      "epoch": 0.8457963667089142,
+      "grad_norm": 1.8255170583724976,
+      "learning_rate": 4.915420363329109e-05,
+      "loss": 0.0661,
+      "step": 2002
+    },
+    {
+      "epoch": 0.8466413181242078,
+      "grad_norm": 1.5995157957077026,
+      "learning_rate": 4.9153358681875795e-05,
+      "loss": 0.0905,
+      "step": 2004
+    },
+    {
+      "epoch": 0.8474862695395015,
+      "grad_norm": 1.5706533193588257,
+      "learning_rate": 4.91525137304605e-05,
+      "loss": 0.1004,
+      "step": 2006
+    },
+    {
+      "epoch": 0.8483312209547951,
+      "grad_norm": 1.2460238933563232,
+      "learning_rate": 4.915166877904521e-05,
+      "loss": 0.0841,
+      "step": 2008
+    },
+    {
+      "epoch": 0.8491761723700887,
+      "grad_norm": 1.2849513292312622,
+      "learning_rate": 4.9150823827629916e-05,
+      "loss": 0.1101,
+      "step": 2010
+    },
+    {
+      "epoch": 0.8500211237853823,
+      "grad_norm": 1.5081242322921753,
+      "learning_rate": 4.9149978876214617e-05,
+      "loss": 0.0944,
+      "step": 2012
+    },
+    {
+      "epoch": 0.8508660752006759,
+      "grad_norm": 0.9104534387588501,
+      "learning_rate": 4.914913392479933e-05,
+      "loss": 0.0442,
+      "step": 2014
+    },
+    {
+      "epoch": 0.8517110266159695,
+      "grad_norm": 1.4931339025497437,
+      "learning_rate": 4.914828897338403e-05,
+      "loss": 0.0814,
+      "step": 2016
+    },
+    {
+      "epoch": 0.8525559780312632,
+      "grad_norm": 1.0569278001785278,
+      "learning_rate": 4.9147444021968745e-05,
+      "loss": 0.0601,
+      "step": 2018
+    },
+    {
+      "epoch": 0.8534009294465569,
+      "grad_norm": 1.3178728818893433,
+      "learning_rate": 4.9146599070553445e-05,
+      "loss": 0.0734,
+      "step": 2020
+    },
+    {
+      "epoch": 0.8542458808618505,
+      "grad_norm": 1.1889725923538208,
+      "learning_rate": 4.914575411913815e-05,
+      "loss": 0.0756,
+      "step": 2022
+    },
+    {
+      "epoch": 0.8550908322771441,
+      "grad_norm": 1.3995158672332764,
+      "learning_rate": 4.914490916772286e-05,
+      "loss": 0.0758,
+      "step": 2024
+    },
+    {
+      "epoch": 0.8559357836924377,
+      "grad_norm": 1.3304693698883057,
+      "learning_rate": 4.914406421630757e-05,
+      "loss": 0.1291,
+      "step": 2026
+    },
+    {
+      "epoch": 0.8567807351077313,
+      "grad_norm": 1.1206573247909546,
+      "learning_rate": 4.914321926489227e-05,
+      "loss": 0.0801,
+      "step": 2028
+    },
+    {
+      "epoch": 0.857625686523025,
+      "grad_norm": 1.374193787574768,
+      "learning_rate": 4.914237431347698e-05,
+      "loss": 0.0698,
+      "step": 2030
+    },
+    {
+      "epoch": 0.8584706379383186,
+      "grad_norm": 1.6789133548736572,
+      "learning_rate": 4.914152936206168e-05,
+      "loss": 0.0965,
+      "step": 2032
+    },
+    {
+      "epoch": 0.8593155893536122,
+      "grad_norm": 1.7472199201583862,
+      "learning_rate": 4.9140684410646396e-05,
+      "loss": 0.0747,
+      "step": 2034
+    },
+    {
+      "epoch": 0.8601605407689058,
+      "grad_norm": 1.2843263149261475,
+      "learning_rate": 4.9139839459231096e-05,
+      "loss": 0.13,
+      "step": 2036
+    },
+    {
+      "epoch": 0.8610054921841994,
+      "grad_norm": 1.7041819095611572,
+      "learning_rate": 4.91389945078158e-05,
+      "loss": 0.0828,
+      "step": 2038
+    },
+    {
+      "epoch": 0.861850443599493,
+      "grad_norm": 2.2484312057495117,
+      "learning_rate": 4.913814955640051e-05,
+      "loss": 0.0595,
+      "step": 2040
+    },
+    {
+      "epoch": 0.8626953950147866,
+      "grad_norm": 1.134190559387207,
+      "learning_rate": 4.913730460498522e-05,
+      "loss": 0.0963,
+      "step": 2042
+    },
+    {
+      "epoch": 0.8635403464300803,
+      "grad_norm": 1.1661795377731323,
+      "learning_rate": 4.9136459653569925e-05,
+      "loss": 0.0601,
+      "step": 2044
+    },
+    {
+      "epoch": 0.8643852978453739,
+      "grad_norm": 1.5845274925231934,
+      "learning_rate": 4.913561470215463e-05,
+      "loss": 0.0888,
+      "step": 2046
+    },
+    {
+      "epoch": 0.8652302492606675,
+      "grad_norm": 0.9580950140953064,
+      "learning_rate": 4.913476975073933e-05,
+      "loss": 0.0604,
+      "step": 2048
+    },
+    {
+      "epoch": 0.8660752006759611,
+      "grad_norm": 0.8986382484436035,
+      "learning_rate": 4.9133924799324047e-05,
+      "loss": 0.0651,
+      "step": 2050
+    },
+    {
+      "epoch": 0.8669201520912547,
+      "grad_norm": 1.660507321357727,
+      "learning_rate": 4.913307984790875e-05,
+      "loss": 0.098,
+      "step": 2052
+    },
+    {
+      "epoch": 0.8677651035065483,
+      "grad_norm": 1.333807349205017,
+      "learning_rate": 4.9132234896493454e-05,
+      "loss": 0.0846,
+      "step": 2054
+    },
+    {
+      "epoch": 0.868610054921842,
+      "grad_norm": 1.2735717296600342,
+      "learning_rate": 4.913138994507816e-05,
+      "loss": 0.0843,
+      "step": 2056
+    },
+    {
+      "epoch": 0.8694550063371356,
+      "grad_norm": 1.303557276725769,
+      "learning_rate": 4.913054499366287e-05,
+      "loss": 0.0682,
+      "step": 2058
+    },
+    {
+      "epoch": 0.8702999577524292,
+      "grad_norm": 0.9738208651542664,
+      "learning_rate": 4.9129700042247576e-05,
+      "loss": 0.0428,
+      "step": 2060
+    },
+    {
+      "epoch": 0.8711449091677228,
+      "grad_norm": 1.4468063116073608,
+      "learning_rate": 4.912885509083228e-05,
+      "loss": 0.0753,
+      "step": 2062
+    },
+    {
+      "epoch": 0.8719898605830165,
+      "grad_norm": 1.511555790901184,
+      "learning_rate": 4.912801013941698e-05,
+      "loss": 0.0601,
+      "step": 2064
+    },
+    {
+      "epoch": 0.8728348119983101,
+      "grad_norm": 1.3249778747558594,
+      "learning_rate": 4.91271651880017e-05,
+      "loss": 0.0742,
+      "step": 2066
+    },
+    {
+      "epoch": 0.8736797634136038,
+      "grad_norm": 1.4642529487609863,
+      "learning_rate": 4.91263202365864e-05,
+      "loss": 0.0613,
+      "step": 2068
+    },
+    {
+      "epoch": 0.8745247148288974,
+      "grad_norm": 1.8487999439239502,
+      "learning_rate": 4.9125475285171105e-05,
+      "loss": 0.097,
+      "step": 2070
+    },
+    {
+      "epoch": 0.875369666244191,
+      "grad_norm": 1.1964668035507202,
+      "learning_rate": 4.912463033375581e-05,
+      "loss": 0.0535,
+      "step": 2072
+    },
+    {
+      "epoch": 0.8762146176594846,
+      "grad_norm": 1.6669561862945557,
+      "learning_rate": 4.912378538234052e-05,
+      "loss": 0.0901,
+      "step": 2074
+    },
+    {
+      "epoch": 0.8770595690747782,
+      "grad_norm": 1.4297568798065186,
+      "learning_rate": 4.9122940430925227e-05,
+      "loss": 0.0731,
+      "step": 2076
+    },
+    {
+      "epoch": 0.8779045204900718,
+      "grad_norm": 1.4387907981872559,
+      "learning_rate": 4.9122095479509934e-05,
+      "loss": 0.0998,
+      "step": 2078
+    },
+    {
+      "epoch": 0.8787494719053655,
+      "grad_norm": 1.2306314706802368,
+      "learning_rate": 4.9121250528094634e-05,
+      "loss": 0.0688,
+      "step": 2080
+    },
+    {
+      "epoch": 0.8795944233206591,
+      "grad_norm": 1.894845724105835,
+      "learning_rate": 4.912040557667935e-05,
+      "loss": 0.0986,
+      "step": 2082
+    },
+    {
+      "epoch": 0.8804393747359527,
+      "grad_norm": 2.315152645111084,
+      "learning_rate": 4.911956062526405e-05,
+      "loss": 0.0945,
+      "step": 2084
+    },
+    {
+      "epoch": 0.8812843261512463,
+      "grad_norm": 2.2627370357513428,
+      "learning_rate": 4.9118715673848756e-05,
+      "loss": 0.0726,
+      "step": 2086
+    },
+    {
+      "epoch": 0.8821292775665399,
+      "grad_norm": 1.7095221281051636,
+      "learning_rate": 4.911787072243346e-05,
+      "loss": 0.1009,
+      "step": 2088
+    },
+    {
+      "epoch": 0.8829742289818335,
+      "grad_norm": 1.60263991355896,
+      "learning_rate": 4.911702577101817e-05,
+      "loss": 0.0422,
+      "step": 2090
+    },
+    {
+      "epoch": 0.8838191803971271,
+      "grad_norm": 0.9724740982055664,
+      "learning_rate": 4.911618081960288e-05,
+      "loss": 0.0434,
+      "step": 2092
+    },
+    {
+      "epoch": 0.8846641318124208,
+      "grad_norm": 1.3910263776779175,
+      "learning_rate": 4.9115335868187585e-05,
+      "loss": 0.0725,
+      "step": 2094
+    },
+    {
+      "epoch": 0.8855090832277144,
+      "grad_norm": 1.592474341392517,
+      "learning_rate": 4.9114490916772285e-05,
+      "loss": 0.0549,
+      "step": 2096
+    },
+    {
+      "epoch": 0.886354034643008,
+      "grad_norm": 0.9358974099159241,
+      "learning_rate": 4.9113645965357e-05,
+      "loss": 0.0485,
+      "step": 2098
+    },
+    {
+      "epoch": 0.8871989860583016,
+      "grad_norm": 1.6186065673828125,
+      "learning_rate": 4.91128010139417e-05,
+      "loss": 0.0803,
+      "step": 2100
+    },
+    {
+      "epoch": 0.8871989860583016,
+      "eval_accuracy": 0.7154681841696844,
+      "eval_cer": 0.08182997080021524,
+      "eval_loss": 0.18388408422470093,
+      "eval_runtime": 862.7282,
+      "eval_samples_per_second": 13.443,
+      "eval_steps_per_second": 0.421,
+      "step": 2100
+    },
+    {
+      "epoch": 0.8880439374735952,
+      "grad_norm": 1.2380942106246948,
+      "learning_rate": 4.9111956062526407e-05,
+      "loss": 0.0668,
+      "step": 2102
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 1.4649783372879028,
+      "learning_rate": 4.9111111111111114e-05,
+      "loss": 0.0652,
+      "step": 2104
+    },
+    {
+      "epoch": 0.8897338403041825,
+      "grad_norm": 1.243154525756836,
+      "learning_rate": 4.911026615969582e-05,
+      "loss": 0.0686,
+      "step": 2106
+    },
+    {
+      "epoch": 0.8905787917194762,
+      "grad_norm": 1.6419312953948975,
+      "learning_rate": 4.910942120828053e-05,
+      "loss": 0.0678,
+      "step": 2108
+    },
+    {
+      "epoch": 0.8914237431347698,
+      "grad_norm": 1.3396897315979004,
+      "learning_rate": 4.9108576256865235e-05,
+      "loss": 0.0713,
+      "step": 2110
+    },
+    {
+      "epoch": 0.8922686945500634,
+      "grad_norm": 1.093079686164856,
+      "learning_rate": 4.9107731305449936e-05,
+      "loss": 0.071,
+      "step": 2112
+    },
+    {
+      "epoch": 0.893113645965357,
+      "grad_norm": 1.6767911911010742,
+      "learning_rate": 4.910688635403465e-05,
+      "loss": 0.0732,
+      "step": 2114
+    },
+    {
+      "epoch": 0.8939585973806506,
+      "grad_norm": 1.8932477235794067,
+      "learning_rate": 4.910604140261935e-05,
+      "loss": 0.0712,
+      "step": 2116
+    },
+    {
+      "epoch": 0.8948035487959443,
+      "grad_norm": 1.5414156913757324,
+      "learning_rate": 4.9105196451204064e-05,
+      "loss": 0.0815,
+      "step": 2118
+    },
+    {
+      "epoch": 0.8956485002112379,
+      "grad_norm": 1.5718928575515747,
+      "learning_rate": 4.9104351499788765e-05,
+      "loss": 0.0805,
+      "step": 2120
+    },
+    {
+      "epoch": 0.8964934516265315,
+      "grad_norm": 1.5843585729599,
+      "learning_rate": 4.910350654837347e-05,
+      "loss": 0.0796,
+      "step": 2122
+    },
+    {
+      "epoch": 0.8973384030418251,
+      "grad_norm": 1.3187426328659058,
+      "learning_rate": 4.910266159695818e-05,
+      "loss": 0.0629,
+      "step": 2124
+    },
+    {
+      "epoch": 0.8981833544571187,
+      "grad_norm": 1.3708245754241943,
+      "learning_rate": 4.9101816645542886e-05,
+      "loss": 0.0706,
+      "step": 2126
+    },
+    {
+      "epoch": 0.8990283058724123,
+      "grad_norm": 1.081969976425171,
+      "learning_rate": 4.9100971694127587e-05,
+      "loss": 0.0749,
+      "step": 2128
+    },
+    {
+      "epoch": 0.899873257287706,
+      "grad_norm": 1.7093164920806885,
+      "learning_rate": 4.91001267427123e-05,
+      "loss": 0.1136,
+      "step": 2130
+    },
+    {
+      "epoch": 0.9007182087029996,
+      "grad_norm": 1.232445240020752,
+      "learning_rate": 4.9099281791297e-05,
+      "loss": 0.054,
+      "step": 2132
+    },
+    {
+      "epoch": 0.9015631601182932,
+      "grad_norm": 0.976844847202301,
+      "learning_rate": 4.9098436839881715e-05,
+      "loss": 0.0713,
+      "step": 2134
+    },
+    {
+      "epoch": 0.9024081115335868,
+      "grad_norm": 1.0007492303848267,
+      "learning_rate": 4.9097591888466415e-05,
+      "loss": 0.0991,
+      "step": 2136
+    },
+    {
+      "epoch": 0.9032530629488804,
+      "grad_norm": 1.6573784351348877,
+      "learning_rate": 4.909674693705112e-05,
+      "loss": 0.0958,
+      "step": 2138
+    },
+    {
+      "epoch": 0.904098014364174,
+      "grad_norm": 0.9490051865577698,
+      "learning_rate": 4.909590198563583e-05,
+      "loss": 0.0394,
+      "step": 2140
+    },
+    {
+      "epoch": 0.9049429657794676,
+      "grad_norm": 1.0081024169921875,
+      "learning_rate": 4.909505703422054e-05,
+      "loss": 0.0573,
+      "step": 2142
+    },
+    {
+      "epoch": 0.9057879171947613,
+      "grad_norm": 1.080897569656372,
+      "learning_rate": 4.9094212082805244e-05,
+      "loss": 0.0745,
+      "step": 2144
+    },
+    {
+      "epoch": 0.9066328686100549,
+      "grad_norm": 1.7212154865264893,
+      "learning_rate": 4.909336713138995e-05,
+      "loss": 0.0968,
+      "step": 2146
+    },
+    {
+      "epoch": 0.9074778200253485,
+      "grad_norm": 1.7466483116149902,
+      "learning_rate": 4.909252217997465e-05,
+      "loss": 0.101,
+      "step": 2148
+    },
+    {
+      "epoch": 0.9083227714406422,
+      "grad_norm": 1.1636947393417358,
+      "learning_rate": 4.9091677228559366e-05,
+      "loss": 0.0816,
+      "step": 2150
+    },
+    {
+      "epoch": 0.9091677228559358,
+      "grad_norm": 1.3398698568344116,
+      "learning_rate": 4.9090832277144066e-05,
+      "loss": 0.0766,
+      "step": 2152
+    },
+    {
+      "epoch": 0.9100126742712294,
+      "grad_norm": 1.0851556062698364,
+      "learning_rate": 4.908998732572877e-05,
+      "loss": 0.0523,
+      "step": 2154
+    },
+    {
+      "epoch": 0.9108576256865231,
+      "grad_norm": 1.4844582080841064,
+      "learning_rate": 4.908914237431348e-05,
+      "loss": 0.0495,
+      "step": 2156
+    },
+    {
+      "epoch": 0.9117025771018167,
+      "grad_norm": 1.3978465795516968,
+      "learning_rate": 4.908829742289819e-05,
+      "loss": 0.0908,
+      "step": 2158
+    },
+    {
+      "epoch": 0.9125475285171103,
+      "grad_norm": 1.431995153427124,
+      "learning_rate": 4.9087452471482895e-05,
+      "loss": 0.1416,
+      "step": 2160
+    },
+    {
+      "epoch": 0.9133924799324039,
+      "grad_norm": 0.8778975009918213,
+      "learning_rate": 4.90866075200676e-05,
+      "loss": 0.0601,
+      "step": 2162
+    },
+    {
+      "epoch": 0.9142374313476975,
+      "grad_norm": 1.388907551765442,
+      "learning_rate": 4.90857625686523e-05,
+      "loss": 0.0787,
+      "step": 2164
+    },
+    {
+      "epoch": 0.9150823827629911,
+      "grad_norm": 1.7531378269195557,
+      "learning_rate": 4.9084917617237017e-05,
+      "loss": 0.0747,
+      "step": 2166
+    },
+    {
+      "epoch": 0.9159273341782848,
+      "grad_norm": 1.6861788034439087,
+      "learning_rate": 4.908407266582172e-05,
+      "loss": 0.0863,
+      "step": 2168
+    },
+    {
+      "epoch": 0.9167722855935784,
+      "grad_norm": 1.4938262701034546,
+      "learning_rate": 4.9083227714406424e-05,
+      "loss": 0.0833,
+      "step": 2170
+    },
+    {
+      "epoch": 0.917617237008872,
+      "grad_norm": 1.2034574747085571,
+      "learning_rate": 4.908238276299113e-05,
+      "loss": 0.0638,
+      "step": 2172
+    },
+    {
+      "epoch": 0.9184621884241656,
+      "grad_norm": 1.2034447193145752,
+      "learning_rate": 4.908153781157583e-05,
+      "loss": 0.0532,
+      "step": 2174
+    },
+    {
+      "epoch": 0.9193071398394592,
+      "grad_norm": 1.0579421520233154,
+      "learning_rate": 4.9080692860160546e-05,
+      "loss": 0.0811,
+      "step": 2176
+    },
+    {
+      "epoch": 0.9201520912547528,
+      "grad_norm": 0.9848757386207581,
+      "learning_rate": 4.9079847908745246e-05,
+      "loss": 0.031,
+      "step": 2178
+    },
+    {
+      "epoch": 0.9209970426700465,
+      "grad_norm": 1.3882626295089722,
+      "learning_rate": 4.907900295732995e-05,
+      "loss": 0.0778,
+      "step": 2180
+    },
+    {
+      "epoch": 0.9218419940853401,
+      "grad_norm": 1.6235911846160889,
+      "learning_rate": 4.907815800591466e-05,
+      "loss": 0.0945,
+      "step": 2182
+    },
+    {
+      "epoch": 0.9226869455006337,
+      "grad_norm": 1.860224723815918,
+      "learning_rate": 4.907731305449937e-05,
+      "loss": 0.0859,
+      "step": 2184
+    },
+    {
+      "epoch": 0.9235318969159273,
+      "grad_norm": 1.091496229171753,
+      "learning_rate": 4.9076468103084075e-05,
+      "loss": 0.0622,
+      "step": 2186
+    },
+    {
+      "epoch": 0.9243768483312209,
+      "grad_norm": 1.7291980981826782,
+      "learning_rate": 4.907562315166878e-05,
+      "loss": 0.0749,
+      "step": 2188
+    },
+    {
+      "epoch": 0.9252217997465145,
+      "grad_norm": 1.0687997341156006,
+      "learning_rate": 4.907477820025348e-05,
+      "loss": 0.0507,
+      "step": 2190
+    },
+    {
+      "epoch": 0.9260667511618081,
+      "grad_norm": 0.9708640575408936,
+      "learning_rate": 4.9073933248838197e-05,
+      "loss": 0.0558,
+      "step": 2192
+    },
+    {
+      "epoch": 0.9269117025771019,
+      "grad_norm": 1.3645669221878052,
+      "learning_rate": 4.90730882974229e-05,
+      "loss": 0.0906,
+      "step": 2194
+    },
+    {
+      "epoch": 0.9277566539923955,
+      "grad_norm": 1.1811169385910034,
+      "learning_rate": 4.9072243346007604e-05,
+      "loss": 0.0693,
+      "step": 2196
+    },
+    {
+      "epoch": 0.9286016054076891,
+      "grad_norm": 1.0970830917358398,
+      "learning_rate": 4.907139839459231e-05,
+      "loss": 0.0704,
+      "step": 2198
+    },
+    {
+      "epoch": 0.9294465568229827,
+      "grad_norm": 1.5071057081222534,
+      "learning_rate": 4.907055344317702e-05,
+      "loss": 0.0932,
+      "step": 2200
+    },
+    {
+      "epoch": 0.9294465568229827,
+      "eval_accuracy": 0.7148646318330747,
+      "eval_cer": 0.0812124526937022,
+      "eval_loss": 0.1800081431865692,
+      "eval_runtime": 846.8242,
+      "eval_samples_per_second": 13.696,
+      "eval_steps_per_second": 0.429,
+      "step": 2200
+    },
+    {
+      "epoch": 0.9302915082382763,
+      "grad_norm": 1.3840326070785522,
+      "learning_rate": 4.9069708491761726e-05,
+      "loss": 0.0484,
+      "step": 2202
+    },
+    {
+      "epoch": 0.93113645965357,
+      "grad_norm": 1.4783692359924316,
+      "learning_rate": 4.906886354034643e-05,
+      "loss": 0.1473,
+      "step": 2204
+    },
+    {
+      "epoch": 0.9319814110688636,
+      "grad_norm": 1.3358536958694458,
+      "learning_rate": 4.906801858893113e-05,
+      "loss": 0.0753,
+      "step": 2206
+    },
+    {
+      "epoch": 0.9328263624841572,
+      "grad_norm": 0.773380696773529,
+      "learning_rate": 4.906717363751585e-05,
+      "loss": 0.0335,
+      "step": 2208
+    },
+    {
+      "epoch": 0.9336713138994508,
+      "grad_norm": 1.2706904411315918,
+      "learning_rate": 4.906632868610055e-05,
+      "loss": 0.0471,
+      "step": 2210
+    },
+    {
+      "epoch": 0.9345162653147444,
+      "grad_norm": 0.99266517162323,
+      "learning_rate": 4.9065483734685255e-05,
+      "loss": 0.067,
+      "step": 2212
+    },
+    {
+      "epoch": 0.935361216730038,
+      "grad_norm": 0.7559247612953186,
+      "learning_rate": 4.906463878326996e-05,
+      "loss": 0.0748,
+      "step": 2214
+    },
+    {
+      "epoch": 0.9362061681453316,
+      "grad_norm": 1.2688716650009155,
+      "learning_rate": 4.906379383185467e-05,
+      "loss": 0.0792,
+      "step": 2216
+    },
+    {
+      "epoch": 0.9370511195606253,
+      "grad_norm": 1.3983566761016846,
+      "learning_rate": 4.9062948880439377e-05,
+      "loss": 0.0737,
+      "step": 2218
+    },
+    {
+      "epoch": 0.9378960709759189,
+      "grad_norm": 1.3656744956970215,
+      "learning_rate": 4.9062103929024084e-05,
+      "loss": 0.0564,
+      "step": 2220
+    },
+    {
+      "epoch": 0.9387410223912125,
+      "grad_norm": 1.1833630800247192,
+      "learning_rate": 4.9061258977608784e-05,
+      "loss": 0.0567,
+      "step": 2222
+    },
+    {
+      "epoch": 0.9395859738065061,
+      "grad_norm": 1.6361331939697266,
+      "learning_rate": 4.90604140261935e-05,
+      "loss": 0.0693,
+      "step": 2224
+    },
+    {
+      "epoch": 0.9404309252217997,
+      "grad_norm": 1.5020231008529663,
+      "learning_rate": 4.90595690747782e-05,
+      "loss": 0.0655,
+      "step": 2226
+    },
+    {
+      "epoch": 0.9412758766370933,
+      "grad_norm": 1.5959280729293823,
+      "learning_rate": 4.9058724123362906e-05,
+      "loss": 0.1087,
+      "step": 2228
+    },
+    {
+      "epoch": 0.942120828052387,
+      "grad_norm": 1.0358680486679077,
+      "learning_rate": 4.905787917194761e-05,
+      "loss": 0.0493,
+      "step": 2230
+    },
+    {
+      "epoch": 0.9429657794676806,
+      "grad_norm": 1.8778531551361084,
+      "learning_rate": 4.905703422053232e-05,
+      "loss": 0.0577,
+      "step": 2232
+    },
+    {
+      "epoch": 0.9438107308829742,
+      "grad_norm": 2.4077320098876953,
+      "learning_rate": 4.905618926911703e-05,
+      "loss": 0.0651,
+      "step": 2234
+    },
+    {
+      "epoch": 0.9446556822982678,
+      "grad_norm": 1.2215629816055298,
+      "learning_rate": 4.9055344317701735e-05,
+      "loss": 0.0704,
+      "step": 2236
+    },
+    {
+      "epoch": 0.9455006337135615,
+      "grad_norm": 0.6339534521102905,
+      "learning_rate": 4.9054499366286435e-05,
+      "loss": 0.0603,
+      "step": 2238
+    },
+    {
+      "epoch": 0.9463455851288551,
+      "grad_norm": 1.2658920288085938,
+      "learning_rate": 4.905365441487115e-05,
+      "loss": 0.0615,
+      "step": 2240
+    },
+    {
+      "epoch": 0.9471905365441488,
+      "grad_norm": 0.8724871277809143,
+      "learning_rate": 4.905280946345585e-05,
+      "loss": 0.0728,
+      "step": 2242
+    },
+    {
+      "epoch": 0.9480354879594424,
+      "grad_norm": 1.5649269819259644,
+      "learning_rate": 4.905196451204056e-05,
+      "loss": 0.1156,
+      "step": 2244
+    },
+    {
+      "epoch": 0.948880439374736,
+      "grad_norm": 1.4007797241210938,
+      "learning_rate": 4.9051119560625264e-05,
+      "loss": 0.0734,
+      "step": 2246
+    },
+    {
+      "epoch": 0.9497253907900296,
+      "grad_norm": 0.7925835251808167,
+      "learning_rate": 4.905027460920997e-05,
+      "loss": 0.1002,
+      "step": 2248
+    },
+    {
+      "epoch": 0.9505703422053232,
+      "grad_norm": 0.9535337090492249,
+      "learning_rate": 4.904942965779468e-05,
+      "loss": 0.0518,
+      "step": 2250
+    },
+    {
+      "epoch": 0.9514152936206168,
+      "grad_norm": 0.983545184135437,
+      "learning_rate": 4.9048584706379385e-05,
+      "loss": 0.0521,
+      "step": 2252
+    },
+    {
+      "epoch": 0.9522602450359104,
+      "grad_norm": 1.7866032123565674,
+      "learning_rate": 4.904773975496409e-05,
+      "loss": 0.0574,
+      "step": 2254
+    },
+    {
+      "epoch": 0.9531051964512041,
+      "grad_norm": 1.5236202478408813,
+      "learning_rate": 4.90468948035488e-05,
+      "loss": 0.0679,
+      "step": 2256
+    },
+    {
+      "epoch": 0.9539501478664977,
+      "grad_norm": 0.9741945266723633,
+      "learning_rate": 4.90460498521335e-05,
+      "loss": 0.0574,
+      "step": 2258
+    },
+    {
+      "epoch": 0.9547950992817913,
+      "grad_norm": 1.6713435649871826,
+      "learning_rate": 4.9045204900718214e-05,
+      "loss": 0.0955,
+      "step": 2260
+    },
+    {
+      "epoch": 0.9556400506970849,
+      "grad_norm": 1.0833595991134644,
+      "learning_rate": 4.9044359949302915e-05,
+      "loss": 0.0803,
+      "step": 2262
+    },
+    {
+      "epoch": 0.9564850021123785,
+      "grad_norm": 1.2216488122940063,
+      "learning_rate": 4.904351499788762e-05,
+      "loss": 0.0859,
+      "step": 2264
+    },
+    {
+      "epoch": 0.9573299535276721,
+      "grad_norm": 1.2390130758285522,
+      "learning_rate": 4.904267004647233e-05,
+      "loss": 0.0506,
+      "step": 2266
+    },
+    {
+      "epoch": 0.9581749049429658,
+      "grad_norm": 1.330955982208252,
+      "learning_rate": 4.9041825095057036e-05,
+      "loss": 0.0526,
+      "step": 2268
+    },
+    {
+      "epoch": 0.9590198563582594,
+      "grad_norm": 1.5434483289718628,
+      "learning_rate": 4.904098014364174e-05,
+      "loss": 0.0539,
+      "step": 2270
+    },
+    {
+      "epoch": 0.959864807773553,
+      "grad_norm": 0.4926339387893677,
+      "learning_rate": 4.904013519222645e-05,
+      "loss": 0.051,
+      "step": 2272
+    },
+    {
+      "epoch": 0.9607097591888466,
+      "grad_norm": 1.3270699977874756,
+      "learning_rate": 4.903929024081115e-05,
+      "loss": 0.0641,
+      "step": 2274
+    },
+    {
+      "epoch": 0.9615547106041402,
+      "grad_norm": 1.2787952423095703,
+      "learning_rate": 4.9038445289395865e-05,
+      "loss": 0.0713,
+      "step": 2276
+    },
+    {
+      "epoch": 0.9623996620194338,
+      "grad_norm": 2.050325632095337,
+      "learning_rate": 4.9037600337980565e-05,
+      "loss": 0.0765,
+      "step": 2278
+    },
+    {
+      "epoch": 0.9632446134347274,
+      "grad_norm": 1.6342370510101318,
+      "learning_rate": 4.903675538656527e-05,
+      "loss": 0.1075,
+      "step": 2280
+    },
+    {
+      "epoch": 0.9640895648500212,
+      "grad_norm": 1.5692452192306519,
+      "learning_rate": 4.903591043514998e-05,
+      "loss": 0.0795,
+      "step": 2282
+    },
+    {
+      "epoch": 0.9649345162653148,
+      "grad_norm": 0.9670857787132263,
+      "learning_rate": 4.903506548373469e-05,
+      "loss": 0.0611,
+      "step": 2284
+    },
+    {
+      "epoch": 0.9657794676806084,
+      "grad_norm": 1.2690353393554688,
+      "learning_rate": 4.9034220532319394e-05,
+      "loss": 0.0779,
+      "step": 2286
+    },
+    {
+      "epoch": 0.966624419095902,
+      "grad_norm": 1.3158725500106812,
+      "learning_rate": 4.90333755809041e-05,
+      "loss": 0.0876,
+      "step": 2288
+    },
+    {
+      "epoch": 0.9674693705111956,
+      "grad_norm": 1.4484848976135254,
+      "learning_rate": 4.90325306294888e-05,
+      "loss": 0.0851,
+      "step": 2290
+    },
+    {
+      "epoch": 0.9683143219264893,
+      "grad_norm": 0.6505631804466248,
+      "learning_rate": 4.9031685678073516e-05,
+      "loss": 0.026,
+      "step": 2292
+    },
+    {
+      "epoch": 0.9691592733417829,
+      "grad_norm": 1.601660132408142,
+      "learning_rate": 4.9030840726658216e-05,
+      "loss": 0.0897,
+      "step": 2294
+    },
+    {
+      "epoch": 0.9700042247570765,
+      "grad_norm": 1.488582730293274,
+      "learning_rate": 4.9029995775242923e-05,
+      "loss": 0.1167,
+      "step": 2296
+    },
+    {
+      "epoch": 0.9708491761723701,
+      "grad_norm": 2.107656955718994,
+      "learning_rate": 4.902915082382763e-05,
+      "loss": 0.1119,
+      "step": 2298
+    },
+    {
+      "epoch": 0.9716941275876637,
+      "grad_norm": 0.8586990237236023,
+      "learning_rate": 4.902830587241234e-05,
+      "loss": 0.0527,
+      "step": 2300
+    },
+    {
+      "epoch": 0.9716941275876637,
+      "eval_accuracy": 0.7121917571995171,
+      "eval_cer": 0.08185643586192295,
+      "eval_loss": 0.1810568869113922,
+      "eval_runtime": 867.447,
+      "eval_samples_per_second": 13.37,
+      "eval_steps_per_second": 0.418,
+      "step": 2300
+    },
+    {
+      "epoch": 0.9725390790029573,
+      "grad_norm": 1.5183982849121094,
+      "learning_rate": 4.9027460920997045e-05,
+      "loss": 0.0819,
+      "step": 2302
+    },
+    {
+      "epoch": 0.973384030418251,
+      "grad_norm": 0.7407907247543335,
+      "learning_rate": 4.902661596958175e-05,
+      "loss": 0.0496,
+      "step": 2304
+    },
+    {
+      "epoch": 0.9742289818335446,
+      "grad_norm": 1.159289002418518,
+      "learning_rate": 4.902577101816645e-05,
+      "loss": 0.0925,
+      "step": 2306
+    },
+    {
+      "epoch": 0.9750739332488382,
+      "grad_norm": 1.0706584453582764,
+      "learning_rate": 4.9024926066751167e-05,
+      "loss": 0.0459,
+      "step": 2308
+    },
+    {
+      "epoch": 0.9759188846641318,
+      "grad_norm": 1.2856115102767944,
+      "learning_rate": 4.902408111533587e-05,
+      "loss": 0.1012,
+      "step": 2310
+    },
+    {
+      "epoch": 0.9767638360794254,
+      "grad_norm": 1.437662124633789,
+      "learning_rate": 4.9023236163920574e-05,
+      "loss": 0.0896,
+      "step": 2312
+    },
+    {
+      "epoch": 0.977608787494719,
+      "grad_norm": 1.2910943031311035,
+      "learning_rate": 4.902239121250528e-05,
+      "loss": 0.0656,
+      "step": 2314
+    },
+    {
+      "epoch": 0.9784537389100126,
+      "grad_norm": 1.636964201927185,
+      "learning_rate": 4.902154626108999e-05,
+      "loss": 0.0864,
+      "step": 2316
+    },
+    {
+      "epoch": 0.9792986903253063,
+      "grad_norm": 1.5599114894866943,
+      "learning_rate": 4.9020701309674696e-05,
+      "loss": 0.0813,
+      "step": 2318
+    },
+    {
+      "epoch": 0.9801436417405999,
+      "grad_norm": 1.6414740085601807,
+      "learning_rate": 4.90198563582594e-05,
+      "loss": 0.066,
+      "step": 2320
+    },
+    {
+      "epoch": 0.9809885931558935,
+      "grad_norm": 0.8238911032676697,
+      "learning_rate": 4.9019011406844103e-05,
+      "loss": 0.0533,
+      "step": 2322
+    },
+    {
+      "epoch": 0.9818335445711872,
+      "grad_norm": 1.3339018821716309,
+      "learning_rate": 4.901816645542882e-05,
+      "loss": 0.0765,
+      "step": 2324
+    },
+    {
+      "epoch": 0.9826784959864808,
+      "grad_norm": 1.6035016775131226,
+      "learning_rate": 4.901732150401352e-05,
+      "loss": 0.0885,
+      "step": 2326
+    },
+    {
+      "epoch": 0.9835234474017744,
+      "grad_norm": 0.8856274485588074,
+      "learning_rate": 4.9016476552598225e-05,
+      "loss": 0.0627,
+      "step": 2328
+    },
+    {
+      "epoch": 0.9843683988170681,
+      "grad_norm": 1.3794585466384888,
+      "learning_rate": 4.901563160118293e-05,
+      "loss": 0.072,
+      "step": 2330
+    },
+    {
+      "epoch": 0.9852133502323617,
+      "grad_norm": 1.3091775178909302,
+      "learning_rate": 4.901478664976764e-05,
+      "loss": 0.1265,
+      "step": 2332
+    },
+    {
+      "epoch": 0.9860583016476553,
+      "grad_norm": 1.0958503484725952,
+      "learning_rate": 4.9013941698352347e-05,
+      "loss": 0.0409,
+      "step": 2334
+    },
+    {
+      "epoch": 0.9869032530629489,
+      "grad_norm": 1.701104760169983,
+      "learning_rate": 4.9013096746937054e-05,
+      "loss": 0.0665,
+      "step": 2336
+    },
+    {
+      "epoch": 0.9877482044782425,
+      "grad_norm": 1.8106247186660767,
+      "learning_rate": 4.9012251795521754e-05,
+      "loss": 0.0842,
+      "step": 2338
+    },
+    {
+      "epoch": 0.9885931558935361,
+      "grad_norm": 1.0502506494522095,
+      "learning_rate": 4.901140684410647e-05,
+      "loss": 0.0682,
+      "step": 2340
+    },
+    {
+      "epoch": 0.9894381073088298,
+      "grad_norm": 1.2314811944961548,
+      "learning_rate": 4.901056189269117e-05,
+      "loss": 0.0712,
+      "step": 2342
+    },
+    {
+      "epoch": 0.9902830587241234,
+      "grad_norm": 1.5358272790908813,
+      "learning_rate": 4.900971694127588e-05,
+      "loss": 0.0675,
+      "step": 2344
+    },
+    {
+      "epoch": 0.991128010139417,
+      "grad_norm": 1.1753113269805908,
+      "learning_rate": 4.900887198986058e-05,
+      "loss": 0.07,
+      "step": 2346
+    },
+    {
+      "epoch": 0.9919729615547106,
+      "grad_norm": 1.636322021484375,
+      "learning_rate": 4.900802703844529e-05,
+      "loss": 0.0772,
+      "step": 2348
+    },
+    {
+      "epoch": 0.9928179129700042,
+      "grad_norm": 1.526485562324524,
+      "learning_rate": 4.900718208703e-05,
+      "loss": 0.0975,
+      "step": 2350
+    },
+    {
+      "epoch": 0.9936628643852978,
+      "grad_norm": 1.3800063133239746,
+      "learning_rate": 4.9006337135614705e-05,
+      "loss": 0.0984,
+      "step": 2352
+    },
+    {
+      "epoch": 0.9945078158005914,
+      "grad_norm": 1.46196711063385,
+      "learning_rate": 4.900549218419941e-05,
+      "loss": 0.0758,
+      "step": 2354
+    },
+    {
+      "epoch": 0.9953527672158851,
+      "grad_norm": 0.7219085693359375,
+      "learning_rate": 4.900464723278412e-05,
+      "loss": 0.0641,
+      "step": 2356
+    },
+    {
+      "epoch": 0.9961977186311787,
+      "grad_norm": 1.4330068826675415,
+      "learning_rate": 4.900380228136882e-05,
+      "loss": 0.1131,
+      "step": 2358
+    },
+    {
+      "epoch": 0.9970426700464723,
+      "grad_norm": 0.9048241972923279,
+      "learning_rate": 4.9002957329953533e-05,
+      "loss": 0.0586,
+      "step": 2360
+    },
+    {
+      "epoch": 0.9978876214617659,
+      "grad_norm": 1.220953106880188,
+      "learning_rate": 4.9002112378538234e-05,
+      "loss": 0.0737,
+      "step": 2362
+    },
+    {
+      "epoch": 0.9987325728770595,
+      "grad_norm": 1.8127233982086182,
+      "learning_rate": 4.900126742712294e-05,
+      "loss": 0.0779,
+      "step": 2364
+    },
+    {
+      "epoch": 0.9995775242923531,
+      "grad_norm": 0.9549395442008972,
+      "learning_rate": 4.900042247570765e-05,
+      "loss": 0.0507,
+      "step": 2366
+    },
+    {
+      "epoch": 1.0004224757076468,
+      "grad_norm": 1.0994274616241455,
+      "learning_rate": 4.8999577524292355e-05,
+      "loss": 0.0667,
+      "step": 2368
+    },
+    {
+      "epoch": 1.0012674271229405,
+      "grad_norm": 1.3673089742660522,
+      "learning_rate": 4.899873257287706e-05,
+      "loss": 0.0409,
+      "step": 2370
+    },
+    {
+      "epoch": 1.002112378538234,
+      "grad_norm": 1.2117745876312256,
+      "learning_rate": 4.899788762146177e-05,
+      "loss": 0.0487,
+      "step": 2372
+    },
+    {
+      "epoch": 1.0029573299535277,
+      "grad_norm": 1.5104575157165527,
+      "learning_rate": 4.899704267004647e-05,
+      "loss": 0.0739,
+      "step": 2374
+    },
+    {
+      "epoch": 1.0038022813688212,
+      "grad_norm": 1.4151710271835327,
+      "learning_rate": 4.8996197718631184e-05,
+      "loss": 0.0622,
+      "step": 2376
+    },
+    {
+      "epoch": 1.004647232784115,
+      "grad_norm": 0.9456071853637695,
+      "learning_rate": 4.8995352767215885e-05,
+      "loss": 0.0614,
+      "step": 2378
+    },
+    {
+      "epoch": 1.0054921841994084,
+      "grad_norm": 0.9996126890182495,
+      "learning_rate": 4.899450781580059e-05,
+      "loss": 0.0477,
+      "step": 2380
+    },
+    {
+      "epoch": 1.0063371356147022,
+      "grad_norm": 0.665332019329071,
+      "learning_rate": 4.89936628643853e-05,
+      "loss": 0.033,
+      "step": 2382
+    },
+    {
+      "epoch": 1.0071820870299957,
+      "grad_norm": 0.9107492566108704,
+      "learning_rate": 4.8992817912970006e-05,
+      "loss": 0.0541,
+      "step": 2384
+    },
+    {
+      "epoch": 1.0080270384452894,
+      "grad_norm": 1.0256426334381104,
+      "learning_rate": 4.8991972961554713e-05,
+      "loss": 0.0438,
+      "step": 2386
+    },
+    {
+      "epoch": 1.008871989860583,
+      "grad_norm": 1.5273648500442505,
+      "learning_rate": 4.899112801013942e-05,
+      "loss": 0.0486,
+      "step": 2388
+    },
+    {
+      "epoch": 1.0097169412758766,
+      "grad_norm": 0.8203524351119995,
+      "learning_rate": 4.899028305872412e-05,
+      "loss": 0.0556,
+      "step": 2390
+    },
+    {
+      "epoch": 1.0105618926911704,
+      "grad_norm": 1.0912214517593384,
+      "learning_rate": 4.8989438107308835e-05,
+      "loss": 0.0579,
+      "step": 2392
+    },
+    {
+      "epoch": 1.0114068441064639,
+      "grad_norm": 1.2179914712905884,
+      "learning_rate": 4.8988593155893535e-05,
+      "loss": 0.0423,
+      "step": 2394
+    },
+    {
+      "epoch": 1.0122517955217576,
+      "grad_norm": 0.6758914589881897,
+      "learning_rate": 4.898774820447824e-05,
+      "loss": 0.0311,
+      "step": 2396
+    },
+    {
+      "epoch": 1.013096746937051,
+      "grad_norm": 1.940477967262268,
+      "learning_rate": 4.898690325306295e-05,
+      "loss": 0.0979,
+      "step": 2398
+    },
+    {
+      "epoch": 1.0139416983523448,
+      "grad_norm": 1.4944899082183838,
+      "learning_rate": 4.898605830164766e-05,
+      "loss": 0.0532,
+      "step": 2400
+    },
+    {
+      "epoch": 1.0139416983523448,
+      "eval_accuracy": 0.7180548370408691,
+      "eval_cer": 0.07823954409520364,
+      "eval_loss": 0.17912158370018005,
+      "eval_runtime": 844.0986,
+      "eval_samples_per_second": 13.74,
+      "eval_steps_per_second": 0.43,
+      "step": 2400
+    },
+    {
+      "epoch": 1.0147866497676383,
+      "grad_norm": 1.0510189533233643,
+      "learning_rate": 4.8985213350232364e-05,
+      "loss": 0.0689,
+      "step": 2402
+    },
+    {
+      "epoch": 1.015631601182932,
+      "grad_norm": 0.9086995720863342,
+      "learning_rate": 4.898436839881707e-05,
+      "loss": 0.0534,
+      "step": 2404
+    },
+    {
+      "epoch": 1.0164765525982256,
+      "grad_norm": 1.3393218517303467,
+      "learning_rate": 4.898352344740177e-05,
+      "loss": 0.0418,
+      "step": 2406
+    },
+    {
+      "epoch": 1.0173215040135193,
+      "grad_norm": 0.9823423027992249,
+      "learning_rate": 4.8982678495986486e-05,
+      "loss": 0.0455,
+      "step": 2408
+    },
+    {
+      "epoch": 1.0181664554288128,
+      "grad_norm": 0.6783949136734009,
+      "learning_rate": 4.8981833544571186e-05,
+      "loss": 0.0495,
+      "step": 2410
+    },
+    {
+      "epoch": 1.0190114068441065,
+      "grad_norm": 1.0267900228500366,
+      "learning_rate": 4.8980988593155893e-05,
+      "loss": 0.0452,
+      "step": 2412
+    },
+    {
+      "epoch": 1.0198563582594,
+      "grad_norm": 0.7721602320671082,
+      "learning_rate": 4.89801436417406e-05,
+      "loss": 0.0577,
+      "step": 2414
+    },
+    {
+      "epoch": 1.0207013096746937,
+      "grad_norm": 1.2067053318023682,
+      "learning_rate": 4.897929869032531e-05,
+      "loss": 0.0488,
+      "step": 2416
+    },
+    {
+      "epoch": 1.0215462610899873,
+      "grad_norm": 1.3156932592391968,
+      "learning_rate": 4.8978453738910015e-05,
+      "loss": 0.0601,
+      "step": 2418
+    },
+    {
+      "epoch": 1.022391212505281,
+      "grad_norm": 1.6176562309265137,
+      "learning_rate": 4.897760878749472e-05,
+      "loss": 0.0538,
+      "step": 2420
+    },
+    {
+      "epoch": 1.0232361639205745,
+      "grad_norm": 1.0514973402023315,
+      "learning_rate": 4.897676383607942e-05,
+      "loss": 0.0583,
+      "step": 2422
+    },
+    {
+      "epoch": 1.0240811153358682,
+      "grad_norm": 1.4830445051193237,
+      "learning_rate": 4.8975918884664137e-05,
+      "loss": 0.0643,
+      "step": 2424
+    },
+    {
+      "epoch": 1.0249260667511617,
+      "grad_norm": 2.010371208190918,
+      "learning_rate": 4.897507393324884e-05,
+      "loss": 0.0437,
+      "step": 2426
+    },
+    {
+      "epoch": 1.0257710181664554,
+      "grad_norm": 0.8634136319160461,
+      "learning_rate": 4.8974228981833544e-05,
+      "loss": 0.0405,
+      "step": 2428
+    },
+    {
+      "epoch": 1.026615969581749,
+      "grad_norm": 1.5516853332519531,
+      "learning_rate": 4.897338403041825e-05,
+      "loss": 0.0441,
+      "step": 2430
+    },
+    {
+      "epoch": 1.0274609209970427,
+      "grad_norm": 1.314122200012207,
+      "learning_rate": 4.897253907900296e-05,
+      "loss": 0.0669,
+      "step": 2432
+    },
+    {
+      "epoch": 1.0283058724123364,
+      "grad_norm": 1.0754412412643433,
+      "learning_rate": 4.8971694127587666e-05,
+      "loss": 0.0333,
+      "step": 2434
+    },
+    {
+      "epoch": 1.02915082382763,
+      "grad_norm": 0.8331787586212158,
+      "learning_rate": 4.897084917617237e-05,
+      "loss": 0.066,
+      "step": 2436
+    },
+    {
+      "epoch": 1.0299957752429236,
+      "grad_norm": 1.0247435569763184,
+      "learning_rate": 4.8970004224757073e-05,
+      "loss": 0.0465,
+      "step": 2438
+    },
+    {
+      "epoch": 1.0308407266582171,
+      "grad_norm": 1.3126165866851807,
+      "learning_rate": 4.896915927334179e-05,
+      "loss": 0.0551,
+      "step": 2440
+    },
+    {
+      "epoch": 1.0316856780735109,
+      "grad_norm": 1.8739526271820068,
+      "learning_rate": 4.896831432192649e-05,
+      "loss": 0.0519,
+      "step": 2442
+    },
+    {
+      "epoch": 1.0325306294888044,
+      "grad_norm": 1.7847691774368286,
+      "learning_rate": 4.89674693705112e-05,
+      "loss": 0.057,
+      "step": 2444
+    },
+    {
+      "epoch": 1.033375580904098,
+      "grad_norm": 1.296621322631836,
+      "learning_rate": 4.89666244190959e-05,
+      "loss": 0.0829,
+      "step": 2446
+    },
+    {
+      "epoch": 1.0342205323193916,
+      "grad_norm": 1.0262562036514282,
+      "learning_rate": 4.896577946768061e-05,
+      "loss": 0.0656,
+      "step": 2448
+    },
+    {
+      "epoch": 1.0350654837346853,
+      "grad_norm": 1.4114569425582886,
+      "learning_rate": 4.896493451626532e-05,
+      "loss": 0.063,
+      "step": 2450
+    },
+    {
+      "epoch": 1.0359104351499788,
+      "grad_norm": 1.8110418319702148,
+      "learning_rate": 4.8964089564850024e-05,
+      "loss": 0.0508,
+      "step": 2452
+    },
+    {
+      "epoch": 1.0367553865652726,
+      "grad_norm": 1.9191076755523682,
+      "learning_rate": 4.896324461343473e-05,
+      "loss": 0.0761,
+      "step": 2454
+    },
+    {
+      "epoch": 1.037600337980566,
+      "grad_norm": 1.3961220979690552,
+      "learning_rate": 4.896239966201944e-05,
+      "loss": 0.061,
+      "step": 2456
+    },
+    {
+      "epoch": 1.0384452893958598,
+      "grad_norm": 0.8220006823539734,
+      "learning_rate": 4.896155471060414e-05,
+      "loss": 0.0524,
+      "step": 2458
+    },
+    {
+      "epoch": 1.0392902408111533,
+      "grad_norm": 1.0066295862197876,
+      "learning_rate": 4.896070975918885e-05,
+      "loss": 0.0468,
+      "step": 2460
+    },
+    {
+      "epoch": 1.040135192226447,
+      "grad_norm": 1.067622184753418,
+      "learning_rate": 4.895986480777355e-05,
+      "loss": 0.0476,
+      "step": 2462
+    },
+    {
+      "epoch": 1.0409801436417405,
+      "grad_norm": 2.0356531143188477,
+      "learning_rate": 4.895901985635826e-05,
+      "loss": 0.0969,
+      "step": 2464
+    },
+    {
+      "epoch": 1.0418250950570342,
+      "grad_norm": 2.8178772926330566,
+      "learning_rate": 4.895817490494297e-05,
+      "loss": 0.0532,
+      "step": 2466
+    },
+    {
+      "epoch": 1.0426700464723277,
+      "grad_norm": 2.3388311862945557,
+      "learning_rate": 4.8957329953527675e-05,
+      "loss": 0.0768,
+      "step": 2468
+    },
+    {
+      "epoch": 1.0435149978876215,
+      "grad_norm": 0.9694822430610657,
+      "learning_rate": 4.895648500211238e-05,
+      "loss": 0.0359,
+      "step": 2470
+    },
+    {
+      "epoch": 1.044359949302915,
+      "grad_norm": 1.434303879737854,
+      "learning_rate": 4.895564005069709e-05,
+      "loss": 0.0581,
+      "step": 2472
+    },
+    {
+      "epoch": 1.0452049007182087,
+      "grad_norm": 1.204118013381958,
+      "learning_rate": 4.895479509928179e-05,
+      "loss": 0.0881,
+      "step": 2474
+    },
+    {
+      "epoch": 1.0460498521335024,
+      "grad_norm": 1.4074668884277344,
+      "learning_rate": 4.8953950147866503e-05,
+      "loss": 0.0411,
+      "step": 2476
+    },
+    {
+      "epoch": 1.046894803548796,
+      "grad_norm": 1.492323637008667,
+      "learning_rate": 4.8953105196451204e-05,
+      "loss": 0.0401,
+      "step": 2478
+    },
+    {
+      "epoch": 1.0477397549640897,
+      "grad_norm": 1.0911036729812622,
+      "learning_rate": 4.895226024503591e-05,
+      "loss": 0.0441,
+      "step": 2480
+    },
+    {
+      "epoch": 1.0485847063793832,
+      "grad_norm": 1.6560875177383423,
+      "learning_rate": 4.895141529362062e-05,
+      "loss": 0.0521,
+      "step": 2482
+    },
+    {
+      "epoch": 1.049429657794677,
+      "grad_norm": 1.2704576253890991,
+      "learning_rate": 4.8950570342205325e-05,
+      "loss": 0.064,
+      "step": 2484
+    },
+    {
+      "epoch": 1.0502746092099704,
+      "grad_norm": 0.9503681063652039,
+      "learning_rate": 4.894972539079003e-05,
+      "loss": 0.0467,
+      "step": 2486
+    },
+    {
+      "epoch": 1.0511195606252641,
+      "grad_norm": 0.7836410403251648,
+      "learning_rate": 4.894888043937474e-05,
+      "loss": 0.0411,
+      "step": 2488
+    },
+    {
+      "epoch": 1.0519645120405576,
+      "grad_norm": 1.5865596532821655,
+      "learning_rate": 4.894803548795944e-05,
+      "loss": 0.0583,
+      "step": 2490
+    },
+    {
+      "epoch": 1.0528094634558514,
+      "grad_norm": 1.2086788415908813,
+      "learning_rate": 4.8947190536544154e-05,
+      "loss": 0.0408,
+      "step": 2492
+    },
+    {
+      "epoch": 1.0536544148711449,
+      "grad_norm": 0.8142328262329102,
+      "learning_rate": 4.8946345585128855e-05,
+      "loss": 0.0322,
+      "step": 2494
+    },
+    {
+      "epoch": 1.0544993662864386,
+      "grad_norm": 0.7533618211746216,
+      "learning_rate": 4.894550063371356e-05,
+      "loss": 0.0454,
+      "step": 2496
+    },
+    {
+      "epoch": 1.055344317701732,
+      "grad_norm": 0.9661786556243896,
+      "learning_rate": 4.894465568229827e-05,
+      "loss": 0.0479,
+      "step": 2498
+    },
+    {
+      "epoch": 1.0561892691170258,
+      "grad_norm": 1.2609660625457764,
+      "learning_rate": 4.8943810730882976e-05,
+      "loss": 0.0653,
+      "step": 2500
+    },
+    {
+      "epoch": 1.0561892691170258,
+      "eval_accuracy": 0.7173650629418865,
+      "eval_cer": 0.08063904302336865,
+      "eval_loss": 0.17944836616516113,
+      "eval_runtime": 861.1084,
+      "eval_samples_per_second": 13.469,
+      "eval_steps_per_second": 0.422,
+      "step": 2500
+    },
+    {
+      "epoch": 1.0570342205323193,
+      "grad_norm": 0.6538660526275635,
+      "learning_rate": 4.8942965779467683e-05,
+      "loss": 0.0483,
+      "step": 2502
+    },
+    {
+      "epoch": 1.057879171947613,
+      "grad_norm": 0.7036990523338318,
+      "learning_rate": 4.894212082805239e-05,
+      "loss": 0.0314,
+      "step": 2504
+    },
+    {
+      "epoch": 1.0587241233629066,
+      "grad_norm": 1.2298040390014648,
+      "learning_rate": 4.894127587663709e-05,
+      "loss": 0.0485,
+      "step": 2506
+    },
+    {
+      "epoch": 1.0595690747782003,
+      "grad_norm": 0.9623536467552185,
+      "learning_rate": 4.8940430925221805e-05,
+      "loss": 0.0521,
+      "step": 2508
+    },
+    {
+      "epoch": 1.0604140261934938,
+      "grad_norm": 1.2405459880828857,
+      "learning_rate": 4.8939585973806505e-05,
+      "loss": 0.0592,
+      "step": 2510
+    },
+    {
+      "epoch": 1.0612589776087875,
+      "grad_norm": 0.9672164916992188,
+      "learning_rate": 4.893874102239121e-05,
+      "loss": 0.0688,
+      "step": 2512
+    },
+    {
+      "epoch": 1.062103929024081,
+      "grad_norm": 1.3323535919189453,
+      "learning_rate": 4.893789607097592e-05,
+      "loss": 0.0475,
+      "step": 2514
+    },
+    {
+      "epoch": 1.0629488804393747,
+      "grad_norm": 0.9052666425704956,
+      "learning_rate": 4.893705111956063e-05,
+      "loss": 0.0275,
+      "step": 2516
+    },
+    {
+      "epoch": 1.0637938318546682,
+      "grad_norm": 0.8641655445098877,
+      "learning_rate": 4.8936206168145334e-05,
+      "loss": 0.0344,
+      "step": 2518
+    },
+    {
+      "epoch": 1.064638783269962,
+      "grad_norm": 1.5869804620742798,
+      "learning_rate": 4.893536121673004e-05,
+      "loss": 0.0674,
+      "step": 2520
+    },
+    {
+      "epoch": 1.0654837346852557,
+      "grad_norm": 1.136039137840271,
+      "learning_rate": 4.893451626531474e-05,
+      "loss": 0.0507,
+      "step": 2522
+    },
+    {
+      "epoch": 1.0663286861005492,
+      "grad_norm": 2.1299498081207275,
+      "learning_rate": 4.8933671313899456e-05,
+      "loss": 0.0856,
+      "step": 2524
+    },
+    {
+      "epoch": 1.067173637515843,
+      "grad_norm": 1.477407693862915,
+      "learning_rate": 4.8932826362484156e-05,
+      "loss": 0.0644,
+      "step": 2526
+    },
+    {
+      "epoch": 1.0680185889311364,
+      "grad_norm": 1.1373028755187988,
+      "learning_rate": 4.8931981411068863e-05,
+      "loss": 0.0342,
+      "step": 2528
+    },
+    {
+      "epoch": 1.0688635403464302,
+      "grad_norm": 1.0594611167907715,
+      "learning_rate": 4.893113645965357e-05,
+      "loss": 0.0743,
+      "step": 2530
+    },
+    {
+      "epoch": 1.0697084917617237,
+      "grad_norm": 1.1284602880477905,
+      "learning_rate": 4.893029150823828e-05,
+      "loss": 0.0621,
+      "step": 2532
+    },
+    {
+      "epoch": 1.0705534431770174,
+      "grad_norm": 1.341365933418274,
+      "learning_rate": 4.8929446556822985e-05,
+      "loss": 0.0596,
+      "step": 2534
+    },
+    {
+      "epoch": 1.071398394592311,
+      "grad_norm": 1.3680822849273682,
+      "learning_rate": 4.892860160540769e-05,
+      "loss": 0.0457,
+      "step": 2536
+    },
+    {
+      "epoch": 1.0722433460076046,
+      "grad_norm": 1.4528318643569946,
+      "learning_rate": 4.892775665399239e-05,
+      "loss": 0.0551,
+      "step": 2538
+    },
+    {
+      "epoch": 1.0730882974228981,
+      "grad_norm": 1.788142442703247,
+      "learning_rate": 4.892691170257711e-05,
+      "loss": 0.0783,
+      "step": 2540
+    },
+    {
+      "epoch": 1.0739332488381919,
+      "grad_norm": 1.1063508987426758,
+      "learning_rate": 4.892606675116181e-05,
+      "loss": 0.0448,
+      "step": 2542
+    },
+    {
+      "epoch": 1.0747782002534854,
+      "grad_norm": 1.275145173072815,
+      "learning_rate": 4.892522179974652e-05,
+      "loss": 0.0513,
+      "step": 2544
+    },
+    {
+      "epoch": 1.075623151668779,
+      "grad_norm": 1.1520663499832153,
+      "learning_rate": 4.892437684833122e-05,
+      "loss": 0.0384,
+      "step": 2546
+    },
+    {
+      "epoch": 1.0764681030840726,
+      "grad_norm": 1.0242836475372314,
+      "learning_rate": 4.892353189691593e-05,
+      "loss": 0.0444,
+      "step": 2548
+    },
+    {
+      "epoch": 1.0773130544993663,
+      "grad_norm": 1.0930004119873047,
+      "learning_rate": 4.8922686945500636e-05,
+      "loss": 0.0381,
+      "step": 2550
+    },
+    {
+      "epoch": 1.0781580059146598,
+      "grad_norm": 1.1523323059082031,
+      "learning_rate": 4.892184199408534e-05,
+      "loss": 0.0565,
+      "step": 2552
+    },
+    {
+      "epoch": 1.0790029573299535,
+      "grad_norm": 1.0468764305114746,
+      "learning_rate": 4.892099704267005e-05,
+      "loss": 0.0436,
+      "step": 2554
+    },
+    {
+      "epoch": 1.079847908745247,
+      "grad_norm": 1.1597340106964111,
+      "learning_rate": 4.892015209125476e-05,
+      "loss": 0.0477,
+      "step": 2556
+    },
+    {
+      "epoch": 1.0806928601605408,
+      "grad_norm": 1.3584051132202148,
+      "learning_rate": 4.891930713983946e-05,
+      "loss": 0.0652,
+      "step": 2558
+    },
+    {
+      "epoch": 1.0815378115758345,
+      "grad_norm": 0.9354948401451111,
+      "learning_rate": 4.891846218842417e-05,
+      "loss": 0.0463,
+      "step": 2560
+    },
+    {
+      "epoch": 1.082382762991128,
+      "grad_norm": 1.0756053924560547,
+      "learning_rate": 4.891761723700887e-05,
+      "loss": 0.0427,
+      "step": 2562
+    },
+    {
+      "epoch": 1.0832277144064215,
+      "grad_norm": 1.3494528532028198,
+      "learning_rate": 4.891677228559358e-05,
+      "loss": 0.0559,
+      "step": 2564
+    },
+    {
+      "epoch": 1.0840726658217152,
+      "grad_norm": 3.9756555557250977,
+      "learning_rate": 4.891592733417829e-05,
+      "loss": 0.0671,
+      "step": 2566
+    },
+    {
+      "epoch": 1.084917617237009,
+      "grad_norm": 1.4587855339050293,
+      "learning_rate": 4.8915082382762994e-05,
+      "loss": 0.042,
+      "step": 2568
+    },
+    {
+      "epoch": 1.0857625686523025,
+      "grad_norm": 1.1794617176055908,
+      "learning_rate": 4.89142374313477e-05,
+      "loss": 0.0722,
+      "step": 2570
+    },
+    {
+      "epoch": 1.0866075200675962,
+      "grad_norm": 1.3769999742507935,
+      "learning_rate": 4.891339247993241e-05,
+      "loss": 0.0846,
+      "step": 2572
+    },
+    {
+      "epoch": 1.0874524714828897,
+      "grad_norm": 0.7593065500259399,
+      "learning_rate": 4.891254752851711e-05,
+      "loss": 0.0436,
+      "step": 2574
+    },
+    {
+      "epoch": 1.0882974228981834,
+      "grad_norm": 1.1376982927322388,
+      "learning_rate": 4.891170257710182e-05,
+      "loss": 0.048,
+      "step": 2576
+    },
+    {
+      "epoch": 1.089142374313477,
+      "grad_norm": 0.9180907011032104,
+      "learning_rate": 4.891085762568652e-05,
+      "loss": 0.0582,
+      "step": 2578
+    },
+    {
+      "epoch": 1.0899873257287707,
+      "grad_norm": 1.5679020881652832,
+      "learning_rate": 4.891001267427123e-05,
+      "loss": 0.0565,
+      "step": 2580
+    },
+    {
+      "epoch": 1.0908322771440642,
+      "grad_norm": 1.0620578527450562,
+      "learning_rate": 4.890916772285594e-05,
+      "loss": 0.0823,
+      "step": 2582
+    },
+    {
+      "epoch": 1.091677228559358,
+      "grad_norm": 0.9624361395835876,
+      "learning_rate": 4.8908322771440645e-05,
+      "loss": 0.0332,
+      "step": 2584
+    },
+    {
+      "epoch": 1.0925221799746514,
+      "grad_norm": 1.1361215114593506,
+      "learning_rate": 4.890747782002535e-05,
+      "loss": 0.0579,
+      "step": 2586
+    },
+    {
+      "epoch": 1.0933671313899451,
+      "grad_norm": 1.0327388048171997,
+      "learning_rate": 4.890663286861006e-05,
+      "loss": 0.0408,
+      "step": 2588
+    },
+    {
+      "epoch": 1.0942120828052386,
+      "grad_norm": 1.140649676322937,
+      "learning_rate": 4.890578791719476e-05,
+      "loss": 0.0414,
+      "step": 2590
+    },
+    {
+      "epoch": 1.0950570342205324,
+      "grad_norm": 1.0419838428497314,
+      "learning_rate": 4.8904942965779473e-05,
+      "loss": 0.0517,
+      "step": 2592
+    },
+    {
+      "epoch": 1.0959019856358259,
+      "grad_norm": 0.7673367261886597,
+      "learning_rate": 4.8904098014364174e-05,
+      "loss": 0.0279,
+      "step": 2594
+    },
+    {
+      "epoch": 1.0967469370511196,
+      "grad_norm": 1.4603475332260132,
+      "learning_rate": 4.890325306294888e-05,
+      "loss": 0.0661,
+      "step": 2596
+    },
+    {
+      "epoch": 1.097591888466413,
+      "grad_norm": 0.6877373456954956,
+      "learning_rate": 4.890240811153359e-05,
+      "loss": 0.0309,
+      "step": 2598
+    },
+    {
+      "epoch": 1.0984368398817068,
+      "grad_norm": 0.6919780969619751,
+      "learning_rate": 4.8901563160118295e-05,
+      "loss": 0.045,
+      "step": 2600
+    },
+    {
+      "epoch": 1.0984368398817068,
+      "eval_accuracy": 0.7184859458527332,
+      "eval_cer": 0.07980098273595808,
+      "eval_loss": 0.1835366040468216,
+      "eval_runtime": 857.7507,
+      "eval_samples_per_second": 13.521,
+      "eval_steps_per_second": 0.423,
+      "step": 2600
+    },
+    {
+      "epoch": 1.0992817912970003,
+      "grad_norm": 1.1517709493637085,
+      "learning_rate": 4.8900718208703e-05,
+      "loss": 0.0561,
+      "step": 2602
+    },
+    {
+      "epoch": 1.100126742712294,
+      "grad_norm": 1.0358161926269531,
+      "learning_rate": 4.889987325728771e-05,
+      "loss": 0.046,
+      "step": 2604
+    },
+    {
+      "epoch": 1.1009716941275878,
+      "grad_norm": 1.4217817783355713,
+      "learning_rate": 4.889902830587241e-05,
+      "loss": 0.073,
+      "step": 2606
+    },
+    {
+      "epoch": 1.1018166455428813,
+      "grad_norm": 1.2074991464614868,
+      "learning_rate": 4.8898183354457124e-05,
+      "loss": 0.0871,
+      "step": 2608
+    },
+    {
+      "epoch": 1.102661596958175,
+      "grad_norm": 2.3143765926361084,
+      "learning_rate": 4.8897338403041825e-05,
+      "loss": 0.0497,
+      "step": 2610
+    },
+    {
+      "epoch": 1.1035065483734685,
+      "grad_norm": 1.1069965362548828,
+      "learning_rate": 4.889649345162653e-05,
+      "loss": 0.0511,
+      "step": 2612
+    },
+    {
+      "epoch": 1.1043514997887622,
+      "grad_norm": 0.9317350387573242,
+      "learning_rate": 4.889564850021124e-05,
+      "loss": 0.0506,
+      "step": 2614
+    },
+    {
+      "epoch": 1.1051964512040557,
+      "grad_norm": 0.9004279971122742,
+      "learning_rate": 4.8894803548795946e-05,
+      "loss": 0.0338,
+      "step": 2616
+    },
+    {
+      "epoch": 1.1060414026193495,
+      "grad_norm": 2.0052106380462646,
+      "learning_rate": 4.8893958597380653e-05,
+      "loss": 0.0945,
+      "step": 2618
+    },
+    {
+      "epoch": 1.106886354034643,
+      "grad_norm": 1.1178474426269531,
+      "learning_rate": 4.889311364596536e-05,
+      "loss": 0.0474,
+      "step": 2620
+    },
+    {
+      "epoch": 1.1077313054499367,
+      "grad_norm": 0.9016062617301941,
+      "learning_rate": 4.889226869455006e-05,
+      "loss": 0.0391,
+      "step": 2622
+    },
+    {
+      "epoch": 1.1085762568652302,
+      "grad_norm": 1.6335241794586182,
+      "learning_rate": 4.8891423743134775e-05,
+      "loss": 0.0487,
+      "step": 2624
+    },
+    {
+      "epoch": 1.109421208280524,
+      "grad_norm": 1.228885293006897,
+      "learning_rate": 4.8890578791719475e-05,
+      "loss": 0.0529,
+      "step": 2626
+    },
+    {
+      "epoch": 1.1102661596958174,
+      "grad_norm": 0.7745740413665771,
+      "learning_rate": 4.888973384030418e-05,
+      "loss": 0.0231,
+      "step": 2628
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 1.8813176155090332,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 0.054,
+      "step": 2630
+    },
+    {
+      "epoch": 1.1119560625264047,
+      "grad_norm": 1.0765502452850342,
+      "learning_rate": 4.88880439374736e-05,
+      "loss": 0.0518,
+      "step": 2632
+    },
+    {
+      "epoch": 1.1128010139416984,
+      "grad_norm": 1.6047254800796509,
+      "learning_rate": 4.8887198986058304e-05,
+      "loss": 0.0632,
+      "step": 2634
+    },
+    {
+      "epoch": 1.113645965356992,
+      "grad_norm": 0.8298768997192383,
+      "learning_rate": 4.888635403464301e-05,
+      "loss": 0.031,
+      "step": 2636
+    },
+    {
+      "epoch": 1.1144909167722856,
+      "grad_norm": 0.9682705402374268,
+      "learning_rate": 4.888550908322771e-05,
+      "loss": 0.0421,
+      "step": 2638
+    },
+    {
+      "epoch": 1.1153358681875791,
+      "grad_norm": 0.9020190834999084,
+      "learning_rate": 4.8884664131812426e-05,
+      "loss": 0.0587,
+      "step": 2640
+    },
+    {
+      "epoch": 1.1161808196028729,
+      "grad_norm": 1.5844128131866455,
+      "learning_rate": 4.8883819180397126e-05,
+      "loss": 0.0633,
+      "step": 2642
+    },
+    {
+      "epoch": 1.1170257710181664,
+      "grad_norm": 1.0159263610839844,
+      "learning_rate": 4.888297422898184e-05,
+      "loss": 0.0635,
+      "step": 2644
+    },
+    {
+      "epoch": 1.11787072243346,
+      "grad_norm": 1.2040449380874634,
+      "learning_rate": 4.888212927756654e-05,
+      "loss": 0.0546,
+      "step": 2646
+    },
+    {
+      "epoch": 1.1187156738487536,
+      "grad_norm": 1.333282232284546,
+      "learning_rate": 4.888128432615125e-05,
+      "loss": 0.0415,
+      "step": 2648
+    },
+    {
+      "epoch": 1.1195606252640473,
+      "grad_norm": 0.9611084461212158,
+      "learning_rate": 4.8880439374735955e-05,
+      "loss": 0.0467,
+      "step": 2650
+    },
+    {
+      "epoch": 1.120405576679341,
+      "grad_norm": 1.6842930316925049,
+      "learning_rate": 4.887959442332066e-05,
+      "loss": 0.0573,
+      "step": 2652
+    },
+    {
+      "epoch": 1.1212505280946345,
+      "grad_norm": 1.4707005023956299,
+      "learning_rate": 4.887874947190537e-05,
+      "loss": 0.0691,
+      "step": 2654
+    },
+    {
+      "epoch": 1.1220954795099283,
+      "grad_norm": 1.7009764909744263,
+      "learning_rate": 4.887790452049008e-05,
+      "loss": 0.0526,
+      "step": 2656
+    },
+    {
+      "epoch": 1.1229404309252218,
+      "grad_norm": 0.4141417145729065,
+      "learning_rate": 4.887705956907478e-05,
+      "loss": 0.0406,
+      "step": 2658
+    },
+    {
+      "epoch": 1.1237853823405155,
+      "grad_norm": 1.2767736911773682,
+      "learning_rate": 4.887621461765949e-05,
+      "loss": 0.0599,
+      "step": 2660
+    },
+    {
+      "epoch": 1.124630333755809,
+      "grad_norm": 1.0069574117660522,
+      "learning_rate": 4.887536966624419e-05,
+      "loss": 0.0733,
+      "step": 2662
+    },
+    {
+      "epoch": 1.1254752851711027,
+      "grad_norm": 1.3213729858398438,
+      "learning_rate": 4.88745247148289e-05,
+      "loss": 0.0641,
+      "step": 2664
+    },
+    {
+      "epoch": 1.1263202365863962,
+      "grad_norm": 0.9671632051467896,
+      "learning_rate": 4.8873679763413606e-05,
+      "loss": 0.065,
+      "step": 2666
+    },
+    {
+      "epoch": 1.12716518800169,
+      "grad_norm": 1.066148042678833,
+      "learning_rate": 4.887283481199831e-05,
+      "loss": 0.0689,
+      "step": 2668
+    },
+    {
+      "epoch": 1.1280101394169835,
+      "grad_norm": 1.6266218423843384,
+      "learning_rate": 4.887198986058302e-05,
+      "loss": 0.0444,
+      "step": 2670
+    },
+    {
+      "epoch": 1.1288550908322772,
+      "grad_norm": 0.7963671088218689,
+      "learning_rate": 4.887114490916773e-05,
+      "loss": 0.0423,
+      "step": 2672
+    },
+    {
+      "epoch": 1.1297000422475707,
+      "grad_norm": 5.003674507141113,
+      "learning_rate": 4.887029995775243e-05,
+      "loss": 0.0406,
+      "step": 2674
+    },
+    {
+      "epoch": 1.1305449936628644,
+      "grad_norm": 1.4064865112304688,
+      "learning_rate": 4.886945500633714e-05,
+      "loss": 0.0567,
+      "step": 2676
+    },
+    {
+      "epoch": 1.131389945078158,
+      "grad_norm": 1.2468315362930298,
+      "learning_rate": 4.886861005492184e-05,
+      "loss": 0.0517,
+      "step": 2678
+    },
+    {
+      "epoch": 1.1322348964934517,
+      "grad_norm": 1.068264365196228,
+      "learning_rate": 4.886776510350655e-05,
+      "loss": 0.0442,
+      "step": 2680
+    },
+    {
+      "epoch": 1.1330798479087452,
+      "grad_norm": 1.4150062799453735,
+      "learning_rate": 4.886692015209126e-05,
+      "loss": 0.0397,
+      "step": 2682
+    },
+    {
+      "epoch": 1.133924799324039,
+      "grad_norm": 1.4678819179534912,
+      "learning_rate": 4.8866075200675964e-05,
+      "loss": 0.061,
+      "step": 2684
+    },
+    {
+      "epoch": 1.1347697507393324,
+      "grad_norm": 1.3416491746902466,
+      "learning_rate": 4.886523024926067e-05,
+      "loss": 0.0644,
+      "step": 2686
+    },
+    {
+      "epoch": 1.1356147021546261,
+      "grad_norm": 1.5595964193344116,
+      "learning_rate": 4.886438529784538e-05,
+      "loss": 0.051,
+      "step": 2688
+    },
+    {
+      "epoch": 1.1364596535699198,
+      "grad_norm": 0.7982243895530701,
+      "learning_rate": 4.886354034643008e-05,
+      "loss": 0.056,
+      "step": 2690
+    },
+    {
+      "epoch": 1.1373046049852134,
+      "grad_norm": 1.517142415046692,
+      "learning_rate": 4.886269539501479e-05,
+      "loss": 0.0659,
+      "step": 2692
+    },
+    {
+      "epoch": 1.1381495564005069,
+      "grad_norm": 0.7119234800338745,
+      "learning_rate": 4.886185044359949e-05,
+      "loss": 0.0539,
+      "step": 2694
+    },
+    {
+      "epoch": 1.1389945078158006,
+      "grad_norm": 1.0239163637161255,
+      "learning_rate": 4.88610054921842e-05,
+      "loss": 0.0475,
+      "step": 2696
+    },
+    {
+      "epoch": 1.1398394592310943,
+      "grad_norm": 1.3597328662872314,
+      "learning_rate": 4.886016054076891e-05,
+      "loss": 0.0515,
+      "step": 2698
+    },
+    {
+      "epoch": 1.1406844106463878,
+      "grad_norm": 0.914358377456665,
+      "learning_rate": 4.8859315589353615e-05,
+      "loss": 0.0496,
+      "step": 2700
+    },
+    {
+      "epoch": 1.1406844106463878,
+      "eval_accuracy": 0.7282290050008622,
+      "eval_cer": 0.07597237047557716,
+      "eval_loss": 0.1734546571969986,
+      "eval_runtime": 866.2492,
+      "eval_samples_per_second": 13.389,
+      "eval_steps_per_second": 0.419,
+      "step": 2700
+    },
+    {
+      "epoch": 1.1415293620616815,
+      "grad_norm": 1.1831682920455933,
+      "learning_rate": 4.885847063793832e-05,
+      "loss": 0.0741,
+      "step": 2702
+    },
+    {
+      "epoch": 1.142374313476975,
+      "grad_norm": 0.8696686029434204,
+      "learning_rate": 4.885762568652303e-05,
+      "loss": 0.0446,
+      "step": 2704
+    },
+    {
+      "epoch": 1.1432192648922688,
+      "grad_norm": 0.9323683977127075,
+      "learning_rate": 4.885678073510773e-05,
+      "loss": 0.0605,
+      "step": 2706
+    },
+    {
+      "epoch": 1.1440642163075623,
+      "grad_norm": 1.124110460281372,
+      "learning_rate": 4.8855935783692443e-05,
+      "loss": 0.036,
+      "step": 2708
+    },
+    {
+      "epoch": 1.144909167722856,
+      "grad_norm": 1.1443071365356445,
+      "learning_rate": 4.8855090832277144e-05,
+      "loss": 0.035,
+      "step": 2710
+    },
+    {
+      "epoch": 1.1457541191381495,
+      "grad_norm": 0.8278305530548096,
+      "learning_rate": 4.885424588086185e-05,
+      "loss": 0.0369,
+      "step": 2712
+    },
+    {
+      "epoch": 1.1465990705534432,
+      "grad_norm": 1.1934250593185425,
+      "learning_rate": 4.885340092944656e-05,
+      "loss": 0.0539,
+      "step": 2714
+    },
+    {
+      "epoch": 1.1474440219687367,
+      "grad_norm": 1.858482003211975,
+      "learning_rate": 4.8852555978031265e-05,
+      "loss": 0.0523,
+      "step": 2716
+    },
+    {
+      "epoch": 1.1482889733840305,
+      "grad_norm": 1.45504891872406,
+      "learning_rate": 4.885171102661597e-05,
+      "loss": 0.0532,
+      "step": 2718
+    },
+    {
+      "epoch": 1.149133924799324,
+      "grad_norm": 1.0459725856781006,
+      "learning_rate": 4.885086607520068e-05,
+      "loss": 0.0487,
+      "step": 2720
+    },
+    {
+      "epoch": 1.1499788762146177,
+      "grad_norm": 2.038820743560791,
+      "learning_rate": 4.885002112378538e-05,
+      "loss": 0.0513,
+      "step": 2722
+    },
+    {
+      "epoch": 1.1508238276299112,
+      "grad_norm": 0.9835671186447144,
+      "learning_rate": 4.8849176172370094e-05,
+      "loss": 0.0536,
+      "step": 2724
+    },
+    {
+      "epoch": 1.151668779045205,
+      "grad_norm": 1.1258163452148438,
+      "learning_rate": 4.8848331220954795e-05,
+      "loss": 0.0583,
+      "step": 2726
+    },
+    {
+      "epoch": 1.1525137304604984,
+      "grad_norm": 0.7174440026283264,
+      "learning_rate": 4.88474862695395e-05,
+      "loss": 0.0283,
+      "step": 2728
+    },
+    {
+      "epoch": 1.1533586818757922,
+      "grad_norm": 1.195196509361267,
+      "learning_rate": 4.884664131812421e-05,
+      "loss": 0.0518,
+      "step": 2730
+    },
+    {
+      "epoch": 1.1542036332910857,
+      "grad_norm": 0.956576943397522,
+      "learning_rate": 4.8845796366708916e-05,
+      "loss": 0.0562,
+      "step": 2732
+    },
+    {
+      "epoch": 1.1550485847063794,
+      "grad_norm": 1.3628367185592651,
+      "learning_rate": 4.8844951415293623e-05,
+      "loss": 0.0548,
+      "step": 2734
+    },
+    {
+      "epoch": 1.1558935361216731,
+      "grad_norm": 0.6316198110580444,
+      "learning_rate": 4.884410646387833e-05,
+      "loss": 0.0293,
+      "step": 2736
+    },
+    {
+      "epoch": 1.1567384875369666,
+      "grad_norm": 0.6486905217170715,
+      "learning_rate": 4.884326151246303e-05,
+      "loss": 0.043,
+      "step": 2738
+    },
+    {
+      "epoch": 1.1575834389522601,
+      "grad_norm": 1.5592443943023682,
+      "learning_rate": 4.8842416561047745e-05,
+      "loss": 0.0589,
+      "step": 2740
+    },
+    {
+      "epoch": 1.1584283903675539,
+      "grad_norm": 1.3737517595291138,
+      "learning_rate": 4.8841571609632446e-05,
+      "loss": 0.0638,
+      "step": 2742
+    },
+    {
+      "epoch": 1.1592733417828476,
+      "grad_norm": 0.8782877922058105,
+      "learning_rate": 4.884072665821716e-05,
+      "loss": 0.0582,
+      "step": 2744
+    },
+    {
+      "epoch": 1.160118293198141,
+      "grad_norm": 0.960125207901001,
+      "learning_rate": 4.883988170680186e-05,
+      "loss": 0.0655,
+      "step": 2746
+    },
+    {
+      "epoch": 1.1609632446134348,
+      "grad_norm": 1.4179805517196655,
+      "learning_rate": 4.883903675538657e-05,
+      "loss": 0.0614,
+      "step": 2748
+    },
+    {
+      "epoch": 1.1618081960287283,
+      "grad_norm": 0.8707264065742493,
+      "learning_rate": 4.8838191803971274e-05,
+      "loss": 0.0385,
+      "step": 2750
+    },
+    {
+      "epoch": 1.162653147444022,
+      "grad_norm": 1.0426108837127686,
+      "learning_rate": 4.883734685255598e-05,
+      "loss": 0.07,
+      "step": 2752
+    },
+    {
+      "epoch": 1.1634980988593155,
+      "grad_norm": 1.4466257095336914,
+      "learning_rate": 4.883650190114069e-05,
+      "loss": 0.0584,
+      "step": 2754
+    },
+    {
+      "epoch": 1.1643430502746093,
+      "grad_norm": 1.018989086151123,
+      "learning_rate": 4.8835656949725396e-05,
+      "loss": 0.0623,
+      "step": 2756
+    },
+    {
+      "epoch": 1.1651880016899028,
+      "grad_norm": 1.6744930744171143,
+      "learning_rate": 4.8834811998310096e-05,
+      "loss": 0.0807,
+      "step": 2758
+    },
+    {
+      "epoch": 1.1660329531051965,
+      "grad_norm": 1.2810227870941162,
+      "learning_rate": 4.883396704689481e-05,
+      "loss": 0.0462,
+      "step": 2760
+    },
+    {
+      "epoch": 1.16687790452049,
+      "grad_norm": 0.7232649326324463,
+      "learning_rate": 4.883312209547951e-05,
+      "loss": 0.0473,
+      "step": 2762
+    },
+    {
+      "epoch": 1.1677228559357837,
+      "grad_norm": 1.4324157238006592,
+      "learning_rate": 4.883227714406422e-05,
+      "loss": 0.0736,
+      "step": 2764
+    },
+    {
+      "epoch": 1.1685678073510772,
+      "grad_norm": 0.6245022416114807,
+      "learning_rate": 4.8831432192648925e-05,
+      "loss": 0.0281,
+      "step": 2766
+    },
+    {
+      "epoch": 1.169412758766371,
+      "grad_norm": 1.720564842224121,
+      "learning_rate": 4.883058724123363e-05,
+      "loss": 0.0831,
+      "step": 2768
+    },
+    {
+      "epoch": 1.1702577101816645,
+      "grad_norm": 1.7815865278244019,
+      "learning_rate": 4.882974228981834e-05,
+      "loss": 0.0448,
+      "step": 2770
+    },
+    {
+      "epoch": 1.1711026615969582,
+      "grad_norm": 1.1956509351730347,
+      "learning_rate": 4.882889733840305e-05,
+      "loss": 0.0696,
+      "step": 2772
+    },
+    {
+      "epoch": 1.1719476130122517,
+      "grad_norm": 0.9493592381477356,
+      "learning_rate": 4.882805238698775e-05,
+      "loss": 0.0671,
+      "step": 2774
+    },
+    {
+      "epoch": 1.1727925644275454,
+      "grad_norm": 0.7172707319259644,
+      "learning_rate": 4.882720743557246e-05,
+      "loss": 0.0495,
+      "step": 2776
+    },
+    {
+      "epoch": 1.173637515842839,
+      "grad_norm": 1.3545876741409302,
+      "learning_rate": 4.882636248415716e-05,
+      "loss": 0.0609,
+      "step": 2778
+    },
+    {
+      "epoch": 1.1744824672581327,
+      "grad_norm": 0.9758241772651672,
+      "learning_rate": 4.882551753274187e-05,
+      "loss": 0.0454,
+      "step": 2780
+    },
+    {
+      "epoch": 1.1753274186734264,
+      "grad_norm": 1.241162657737732,
+      "learning_rate": 4.8824672581326576e-05,
+      "loss": 0.0447,
+      "step": 2782
+    },
+    {
+      "epoch": 1.1761723700887199,
+      "grad_norm": 1.0415271520614624,
+      "learning_rate": 4.882382762991128e-05,
+      "loss": 0.0389,
+      "step": 2784
+    },
+    {
+      "epoch": 1.1770173215040136,
+      "grad_norm": 1.916858196258545,
+      "learning_rate": 4.882298267849599e-05,
+      "loss": 0.0658,
+      "step": 2786
+    },
+    {
+      "epoch": 1.1778622729193071,
+      "grad_norm": 1.2773157358169556,
+      "learning_rate": 4.88221377270807e-05,
+      "loss": 0.0654,
+      "step": 2788
+    },
+    {
+      "epoch": 1.1787072243346008,
+      "grad_norm": 1.5616599321365356,
+      "learning_rate": 4.88212927756654e-05,
+      "loss": 0.0643,
+      "step": 2790
+    },
+    {
+      "epoch": 1.1795521757498943,
+      "grad_norm": 1.3588258028030396,
+      "learning_rate": 4.882044782425011e-05,
+      "loss": 0.0546,
+      "step": 2792
+    },
+    {
+      "epoch": 1.180397127165188,
+      "grad_norm": 0.8794555068016052,
+      "learning_rate": 4.881960287283481e-05,
+      "loss": 0.0537,
+      "step": 2794
+    },
+    {
+      "epoch": 1.1812420785804816,
+      "grad_norm": 1.2843623161315918,
+      "learning_rate": 4.881875792141952e-05,
+      "loss": 0.0349,
+      "step": 2796
+    },
+    {
+      "epoch": 1.1820870299957753,
+      "grad_norm": 1.0971851348876953,
+      "learning_rate": 4.881791297000423e-05,
+      "loss": 0.0529,
+      "step": 2798
+    },
+    {
+      "epoch": 1.1829319814110688,
+      "grad_norm": 1.3400644063949585,
+      "learning_rate": 4.8817068018588934e-05,
+      "loss": 0.0555,
+      "step": 2800
+    },
+    {
+      "epoch": 1.1829319814110688,
+      "eval_accuracy": 0.7257285738920504,
+      "eval_cer": 0.07709272475453656,
+      "eval_loss": 0.1766710728406906,
+      "eval_runtime": 853.1559,
+      "eval_samples_per_second": 13.594,
+      "eval_steps_per_second": 0.425,
+      "step": 2800
+    },
+    {
+      "epoch": 1.1837769328263625,
+      "grad_norm": 0.753764271736145,
+      "learning_rate": 4.881622306717364e-05,
+      "loss": 0.0612,
+      "step": 2802
+    },
+    {
+      "epoch": 1.184621884241656,
+      "grad_norm": 3.2383697032928467,
+      "learning_rate": 4.881537811575835e-05,
+      "loss": 0.0604,
+      "step": 2804
+    },
+    {
+      "epoch": 1.1854668356569498,
+      "grad_norm": 0.9122393727302551,
+      "learning_rate": 4.881453316434305e-05,
+      "loss": 0.0318,
+      "step": 2806
+    },
+    {
+      "epoch": 1.1863117870722433,
+      "grad_norm": 1.2476575374603271,
+      "learning_rate": 4.881368821292776e-05,
+      "loss": 0.0407,
+      "step": 2808
+    },
+    {
+      "epoch": 1.187156738487537,
+      "grad_norm": 1.0261362791061401,
+      "learning_rate": 4.881284326151246e-05,
+      "loss": 0.0529,
+      "step": 2810
+    },
+    {
+      "epoch": 1.1880016899028305,
+      "grad_norm": 1.278996467590332,
+      "learning_rate": 4.881199831009717e-05,
+      "loss": 0.0606,
+      "step": 2812
+    },
+    {
+      "epoch": 1.1888466413181242,
+      "grad_norm": 1.2511422634124756,
+      "learning_rate": 4.881115335868188e-05,
+      "loss": 0.0555,
+      "step": 2814
+    },
+    {
+      "epoch": 1.1896915927334177,
+      "grad_norm": 0.9766137599945068,
+      "learning_rate": 4.8810308407266585e-05,
+      "loss": 0.0435,
+      "step": 2816
+    },
+    {
+      "epoch": 1.1905365441487115,
+      "grad_norm": 1.5147414207458496,
+      "learning_rate": 4.880946345585129e-05,
+      "loss": 0.0418,
+      "step": 2818
+    },
+    {
+      "epoch": 1.1913814955640052,
+      "grad_norm": 1.3996652364730835,
+      "learning_rate": 4.8808618504436e-05,
+      "loss": 0.0914,
+      "step": 2820
+    },
+    {
+      "epoch": 1.1922264469792987,
+      "grad_norm": 1.1199573278427124,
+      "learning_rate": 4.88077735530207e-05,
+      "loss": 0.0549,
+      "step": 2822
+    },
+    {
+      "epoch": 1.1930713983945922,
+      "grad_norm": 0.946550726890564,
+      "learning_rate": 4.8806928601605414e-05,
+      "loss": 0.0356,
+      "step": 2824
+    },
+    {
+      "epoch": 1.193916349809886,
+      "grad_norm": 0.9064375162124634,
+      "learning_rate": 4.8806083650190114e-05,
+      "loss": 0.0669,
+      "step": 2826
+    },
+    {
+      "epoch": 1.1947613012251797,
+      "grad_norm": 1.4223860502243042,
+      "learning_rate": 4.880523869877482e-05,
+      "loss": 0.0581,
+      "step": 2828
+    },
+    {
+      "epoch": 1.1956062526404732,
+      "grad_norm": 1.123694658279419,
+      "learning_rate": 4.880439374735953e-05,
+      "loss": 0.0437,
+      "step": 2830
+    },
+    {
+      "epoch": 1.1964512040557669,
+      "grad_norm": 1.1511688232421875,
+      "learning_rate": 4.8803548795944236e-05,
+      "loss": 0.056,
+      "step": 2832
+    },
+    {
+      "epoch": 1.1972961554710604,
+      "grad_norm": 1.4350991249084473,
+      "learning_rate": 4.880270384452894e-05,
+      "loss": 0.044,
+      "step": 2834
+    },
+    {
+      "epoch": 1.1981411068863541,
+      "grad_norm": 1.211211919784546,
+      "learning_rate": 4.880185889311365e-05,
+      "loss": 0.0378,
+      "step": 2836
+    },
+    {
+      "epoch": 1.1989860583016476,
+      "grad_norm": 1.3658467531204224,
+      "learning_rate": 4.880101394169835e-05,
+      "loss": 0.0848,
+      "step": 2838
+    },
+    {
+      "epoch": 1.1998310097169413,
+      "grad_norm": 1.6708577871322632,
+      "learning_rate": 4.8800168990283064e-05,
+      "loss": 0.0461,
+      "step": 2840
+    },
+    {
+      "epoch": 1.2006759611322348,
+      "grad_norm": 1.1294584274291992,
+      "learning_rate": 4.8799324038867765e-05,
+      "loss": 0.0485,
+      "step": 2842
+    },
+    {
+      "epoch": 1.2015209125475286,
+      "grad_norm": 1.5034394264221191,
+      "learning_rate": 4.879847908745248e-05,
+      "loss": 0.0611,
+      "step": 2844
+    },
+    {
+      "epoch": 1.202365863962822,
+      "grad_norm": 1.4616371393203735,
+      "learning_rate": 4.879763413603718e-05,
+      "loss": 0.0576,
+      "step": 2846
+    },
+    {
+      "epoch": 1.2032108153781158,
+      "grad_norm": 1.160904049873352,
+      "learning_rate": 4.8796789184621886e-05,
+      "loss": 0.0682,
+      "step": 2848
+    },
+    {
+      "epoch": 1.2040557667934093,
+      "grad_norm": 0.753709614276886,
+      "learning_rate": 4.8795944233206594e-05,
+      "loss": 0.0333,
+      "step": 2850
+    },
+    {
+      "epoch": 1.204900718208703,
+      "grad_norm": 1.6820114850997925,
+      "learning_rate": 4.87950992817913e-05,
+      "loss": 0.0587,
+      "step": 2852
+    },
+    {
+      "epoch": 1.2057456696239965,
+      "grad_norm": 1.2073854207992554,
+      "learning_rate": 4.879425433037601e-05,
+      "loss": 0.0678,
+      "step": 2854
+    },
+    {
+      "epoch": 1.2065906210392903,
+      "grad_norm": 2.104133129119873,
+      "learning_rate": 4.8793409378960715e-05,
+      "loss": 0.0868,
+      "step": 2856
+    },
+    {
+      "epoch": 1.2074355724545838,
+      "grad_norm": 1.4604604244232178,
+      "learning_rate": 4.8792564427545416e-05,
+      "loss": 0.0587,
+      "step": 2858
+    },
+    {
+      "epoch": 1.2082805238698775,
+      "grad_norm": 1.0882490873336792,
+      "learning_rate": 4.879171947613013e-05,
+      "loss": 0.0493,
+      "step": 2860
+    },
+    {
+      "epoch": 1.209125475285171,
+      "grad_norm": 1.402060866355896,
+      "learning_rate": 4.879087452471483e-05,
+      "loss": 0.0895,
+      "step": 2862
+    },
+    {
+      "epoch": 1.2099704267004647,
+      "grad_norm": 1.4716256856918335,
+      "learning_rate": 4.879002957329954e-05,
+      "loss": 0.0625,
+      "step": 2864
+    },
+    {
+      "epoch": 1.2108153781157585,
+      "grad_norm": 1.0345948934555054,
+      "learning_rate": 4.8789184621884244e-05,
+      "loss": 0.0621,
+      "step": 2866
+    },
+    {
+      "epoch": 1.211660329531052,
+      "grad_norm": 1.3120019435882568,
+      "learning_rate": 4.878833967046895e-05,
+      "loss": 0.0606,
+      "step": 2868
+    },
+    {
+      "epoch": 1.2125052809463455,
+      "grad_norm": 1.2341971397399902,
+      "learning_rate": 4.878749471905366e-05,
+      "loss": 0.0595,
+      "step": 2870
+    },
+    {
+      "epoch": 1.2133502323616392,
+      "grad_norm": 1.1355195045471191,
+      "learning_rate": 4.8786649767638366e-05,
+      "loss": 0.0392,
+      "step": 2872
+    },
+    {
+      "epoch": 1.214195183776933,
+      "grad_norm": 1.4073996543884277,
+      "learning_rate": 4.8785804816223066e-05,
+      "loss": 0.048,
+      "step": 2874
+    },
+    {
+      "epoch": 1.2150401351922264,
+      "grad_norm": 0.8575471043586731,
+      "learning_rate": 4.878495986480778e-05,
+      "loss": 0.0436,
+      "step": 2876
+    },
+    {
+      "epoch": 1.2158850866075201,
+      "grad_norm": 1.1738005876541138,
+      "learning_rate": 4.878411491339248e-05,
+      "loss": 0.0434,
+      "step": 2878
+    },
+    {
+      "epoch": 1.2167300380228137,
+      "grad_norm": 1.412050724029541,
+      "learning_rate": 4.878326996197719e-05,
+      "loss": 0.0475,
+      "step": 2880
+    },
+    {
+      "epoch": 1.2175749894381074,
+      "grad_norm": 1.4315383434295654,
+      "learning_rate": 4.8782425010561895e-05,
+      "loss": 0.063,
+      "step": 2882
+    },
+    {
+      "epoch": 1.2184199408534009,
+      "grad_norm": 1.2847777605056763,
+      "learning_rate": 4.87815800591466e-05,
+      "loss": 0.0495,
+      "step": 2884
+    },
+    {
+      "epoch": 1.2192648922686946,
+      "grad_norm": 1.1097843647003174,
+      "learning_rate": 4.878073510773131e-05,
+      "loss": 0.0836,
+      "step": 2886
+    },
+    {
+      "epoch": 1.2201098436839881,
+      "grad_norm": 1.2086609601974487,
+      "learning_rate": 4.877989015631602e-05,
+      "loss": 0.069,
+      "step": 2888
+    },
+    {
+      "epoch": 1.2209547950992818,
+      "grad_norm": 1.2778347730636597,
+      "learning_rate": 4.877904520490072e-05,
+      "loss": 0.0595,
+      "step": 2890
+    },
+    {
+      "epoch": 1.2217997465145753,
+      "grad_norm": 1.0457631349563599,
+      "learning_rate": 4.877820025348543e-05,
+      "loss": 0.0322,
+      "step": 2892
+    },
+    {
+      "epoch": 1.222644697929869,
+      "grad_norm": 1.6547378301620483,
+      "learning_rate": 4.877735530207013e-05,
+      "loss": 0.0622,
+      "step": 2894
+    },
+    {
+      "epoch": 1.2234896493451626,
+      "grad_norm": 1.1878632307052612,
+      "learning_rate": 4.877651035065484e-05,
+      "loss": 0.0549,
+      "step": 2896
+    },
+    {
+      "epoch": 1.2243346007604563,
+      "grad_norm": 1.374286413192749,
+      "learning_rate": 4.8775665399239546e-05,
+      "loss": 0.0628,
+      "step": 2898
+    },
+    {
+      "epoch": 1.2251795521757498,
+      "grad_norm": 0.7572351098060608,
+      "learning_rate": 4.877482044782425e-05,
+      "loss": 0.0645,
+      "step": 2900
+    },
+    {
+      "epoch": 1.2251795521757498,
+      "eval_accuracy": 0.7215899292981549,
+      "eval_cer": 0.07695157775876214,
+      "eval_loss": 0.17909274995326996,
+      "eval_runtime": 861.6493,
+      "eval_samples_per_second": 13.46,
+      "eval_steps_per_second": 0.421,
+      "step": 2900
+    },
+    {
+      "epoch": 1.2260245035910435,
+      "grad_norm": 1.633081316947937,
+      "learning_rate": 4.877397549640896e-05,
+      "loss": 0.0579,
+      "step": 2902
+    },
+    {
+      "epoch": 1.226869455006337,
+      "grad_norm": 1.0226025581359863,
+      "learning_rate": 4.877313054499367e-05,
+      "loss": 0.0822,
+      "step": 2904
+    },
+    {
+      "epoch": 1.2277144064216308,
+      "grad_norm": 1.0272724628448486,
+      "learning_rate": 4.877228559357837e-05,
+      "loss": 0.0706,
+      "step": 2906
+    },
+    {
+      "epoch": 1.2285593578369243,
+      "grad_norm": 0.925186812877655,
+      "learning_rate": 4.877144064216308e-05,
+      "loss": 0.0578,
+      "step": 2908
+    },
+    {
+      "epoch": 1.229404309252218,
+      "grad_norm": 0.5941814184188843,
+      "learning_rate": 4.877059569074778e-05,
+      "loss": 0.0352,
+      "step": 2910
+    },
+    {
+      "epoch": 1.2302492606675117,
+      "grad_norm": 1.436691164970398,
+      "learning_rate": 4.876975073933249e-05,
+      "loss": 0.0637,
+      "step": 2912
+    },
+    {
+      "epoch": 1.2310942120828052,
+      "grad_norm": 0.9653565883636475,
+      "learning_rate": 4.87689057879172e-05,
+      "loss": 0.0374,
+      "step": 2914
+    },
+    {
+      "epoch": 1.231939163498099,
+      "grad_norm": 0.8369848132133484,
+      "learning_rate": 4.8768060836501904e-05,
+      "loss": 0.0431,
+      "step": 2916
+    },
+    {
+      "epoch": 1.2327841149133925,
+      "grad_norm": 1.2851033210754395,
+      "learning_rate": 4.876721588508661e-05,
+      "loss": 0.0618,
+      "step": 2918
+    },
+    {
+      "epoch": 1.2336290663286862,
+      "grad_norm": 1.213028073310852,
+      "learning_rate": 4.876637093367132e-05,
+      "loss": 0.0585,
+      "step": 2920
+    },
+    {
+      "epoch": 1.2344740177439797,
+      "grad_norm": 1.9548083543777466,
+      "learning_rate": 4.876552598225602e-05,
+      "loss": 0.0659,
+      "step": 2922
+    },
+    {
+      "epoch": 1.2353189691592734,
+      "grad_norm": 1.0520684719085693,
+      "learning_rate": 4.876468103084073e-05,
+      "loss": 0.0374,
+      "step": 2924
+    },
+    {
+      "epoch": 1.236163920574567,
+      "grad_norm": 0.8738290667533875,
+      "learning_rate": 4.876383607942543e-05,
+      "loss": 0.0727,
+      "step": 2926
+    },
+    {
+      "epoch": 1.2370088719898606,
+      "grad_norm": 1.4663364887237549,
+      "learning_rate": 4.876299112801014e-05,
+      "loss": 0.0471,
+      "step": 2928
+    },
+    {
+      "epoch": 1.2378538234051542,
+      "grad_norm": 1.1130000352859497,
+      "learning_rate": 4.876214617659485e-05,
+      "loss": 0.053,
+      "step": 2930
+    },
+    {
+      "epoch": 1.2386987748204479,
+      "grad_norm": 0.9483621120452881,
+      "learning_rate": 4.8761301225179555e-05,
+      "loss": 0.058,
+      "step": 2932
+    },
+    {
+      "epoch": 1.2395437262357414,
+      "grad_norm": 0.7923450469970703,
+      "learning_rate": 4.876045627376426e-05,
+      "loss": 0.0384,
+      "step": 2934
+    },
+    {
+      "epoch": 1.240388677651035,
+      "grad_norm": 1.8122419118881226,
+      "learning_rate": 4.875961132234897e-05,
+      "loss": 0.076,
+      "step": 2936
+    },
+    {
+      "epoch": 1.2412336290663286,
+      "grad_norm": 1.1781667470932007,
+      "learning_rate": 4.875876637093367e-05,
+      "loss": 0.053,
+      "step": 2938
+    },
+    {
+      "epoch": 1.2420785804816223,
+      "grad_norm": 1.2569353580474854,
+      "learning_rate": 4.8757921419518384e-05,
+      "loss": 0.0514,
+      "step": 2940
+    },
+    {
+      "epoch": 1.2429235318969158,
+      "grad_norm": 1.32847261428833,
+      "learning_rate": 4.8757076468103084e-05,
+      "loss": 0.0648,
+      "step": 2942
+    },
+    {
+      "epoch": 1.2437684833122096,
+      "grad_norm": 0.7314063310623169,
+      "learning_rate": 4.87562315166878e-05,
+      "loss": 0.0582,
+      "step": 2944
+    },
+    {
+      "epoch": 1.244613434727503,
+      "grad_norm": 1.6381207704544067,
+      "learning_rate": 4.87553865652725e-05,
+      "loss": 0.0582,
+      "step": 2946
+    },
+    {
+      "epoch": 1.2454583861427968,
+      "grad_norm": 1.02126145362854,
+      "learning_rate": 4.8754541613857206e-05,
+      "loss": 0.0705,
+      "step": 2948
+    },
+    {
+      "epoch": 1.2463033375580905,
+      "grad_norm": 1.4831101894378662,
+      "learning_rate": 4.875369666244191e-05,
+      "loss": 0.0784,
+      "step": 2950
+    },
+    {
+      "epoch": 1.247148288973384,
+      "grad_norm": 1.5562660694122314,
+      "learning_rate": 4.875285171102662e-05,
+      "loss": 0.064,
+      "step": 2952
+    },
+    {
+      "epoch": 1.2479932403886775,
+      "grad_norm": 0.9851226210594177,
+      "learning_rate": 4.875200675961133e-05,
+      "loss": 0.0513,
+      "step": 2954
+    },
+    {
+      "epoch": 1.2488381918039713,
+      "grad_norm": 1.7277909517288208,
+      "learning_rate": 4.8751161808196034e-05,
+      "loss": 0.0535,
+      "step": 2956
+    },
+    {
+      "epoch": 1.249683143219265,
+      "grad_norm": 1.9152166843414307,
+      "learning_rate": 4.8750316856780735e-05,
+      "loss": 0.0508,
+      "step": 2958
+    },
+    {
+      "epoch": 1.2505280946345585,
+      "grad_norm": 0.7767104506492615,
+      "learning_rate": 4.874947190536545e-05,
+      "loss": 0.0317,
+      "step": 2960
+    },
+    {
+      "epoch": 1.251373046049852,
+      "grad_norm": 1.0719561576843262,
+      "learning_rate": 4.874862695395015e-05,
+      "loss": 0.0338,
+      "step": 2962
+    },
+    {
+      "epoch": 1.2522179974651457,
+      "grad_norm": 0.9941785931587219,
+      "learning_rate": 4.8747782002534856e-05,
+      "loss": 0.0297,
+      "step": 2964
+    },
+    {
+      "epoch": 1.2530629488804395,
+      "grad_norm": 1.413793683052063,
+      "learning_rate": 4.8746937051119564e-05,
+      "loss": 0.0414,
+      "step": 2966
+    },
+    {
+      "epoch": 1.253907900295733,
+      "grad_norm": 0.8977073431015015,
+      "learning_rate": 4.874609209970427e-05,
+      "loss": 0.0697,
+      "step": 2968
+    },
+    {
+      "epoch": 1.2547528517110267,
+      "grad_norm": 1.1440623998641968,
+      "learning_rate": 4.874524714828898e-05,
+      "loss": 0.0727,
+      "step": 2970
+    },
+    {
+      "epoch": 1.2555978031263202,
+      "grad_norm": 0.648844301700592,
+      "learning_rate": 4.8744402196873685e-05,
+      "loss": 0.0411,
+      "step": 2972
+    },
+    {
+      "epoch": 1.256442754541614,
+      "grad_norm": 1.1983152627944946,
+      "learning_rate": 4.8743557245458386e-05,
+      "loss": 0.0414,
+      "step": 2974
+    },
+    {
+      "epoch": 1.2572877059569074,
+      "grad_norm": 1.0279157161712646,
+      "learning_rate": 4.87427122940431e-05,
+      "loss": 0.0489,
+      "step": 2976
+    },
+    {
+      "epoch": 1.2581326573722011,
+      "grad_norm": 1.3486169576644897,
+      "learning_rate": 4.87418673426278e-05,
+      "loss": 0.0751,
+      "step": 2978
+    },
+    {
+      "epoch": 1.2589776087874947,
+      "grad_norm": 0.9830628037452698,
+      "learning_rate": 4.874102239121251e-05,
+      "loss": 0.0492,
+      "step": 2980
+    },
+    {
+      "epoch": 1.2598225602027884,
+      "grad_norm": 1.5126888751983643,
+      "learning_rate": 4.8740177439797214e-05,
+      "loss": 0.0404,
+      "step": 2982
+    },
+    {
+      "epoch": 1.2606675116180819,
+      "grad_norm": 0.9845932126045227,
+      "learning_rate": 4.873933248838192e-05,
+      "loss": 0.077,
+      "step": 2984
+    },
+    {
+      "epoch": 1.2615124630333756,
+      "grad_norm": 0.8719518780708313,
+      "learning_rate": 4.873848753696663e-05,
+      "loss": 0.0355,
+      "step": 2986
+    },
+    {
+      "epoch": 1.2623574144486693,
+      "grad_norm": 1.3625831604003906,
+      "learning_rate": 4.8737642585551336e-05,
+      "loss": 0.0571,
+      "step": 2988
+    },
+    {
+      "epoch": 1.2632023658639628,
+      "grad_norm": 1.1835618019104004,
+      "learning_rate": 4.8736797634136036e-05,
+      "loss": 0.0423,
+      "step": 2990
+    },
+    {
+      "epoch": 1.2640473172792563,
+      "grad_norm": 1.0713012218475342,
+      "learning_rate": 4.873595268272075e-05,
+      "loss": 0.0571,
+      "step": 2992
+    },
+    {
+      "epoch": 1.26489226869455,
+      "grad_norm": 1.7218633890151978,
+      "learning_rate": 4.873510773130545e-05,
+      "loss": 0.0462,
+      "step": 2994
+    },
+    {
+      "epoch": 1.2657372201098438,
+      "grad_norm": 1.8238509893417358,
+      "learning_rate": 4.873426277989016e-05,
+      "loss": 0.0559,
+      "step": 2996
+    },
+    {
+      "epoch": 1.2665821715251373,
+      "grad_norm": 1.0371171236038208,
+      "learning_rate": 4.8733417828474865e-05,
+      "loss": 0.0469,
+      "step": 2998
+    },
+    {
+      "epoch": 1.2674271229404308,
+      "grad_norm": 1.3527776002883911,
+      "learning_rate": 4.873257287705957e-05,
+      "loss": 0.0402,
+      "step": 3000
+    },
+    {
+      "epoch": 1.2674271229404308,
+      "eval_accuracy": 0.7070184514571478,
+      "eval_cer": 0.0812742045043535,
+      "eval_loss": 0.18232469260692596,
+      "eval_runtime": 842.602,
+      "eval_samples_per_second": 13.765,
+      "eval_steps_per_second": 0.431,
+      "step": 3000
+    },
+    {
+      "epoch": 1.2682720743557245,
+      "grad_norm": 0.8487452268600464,
+      "learning_rate": 4.873172792564428e-05,
+      "loss": 0.0171,
+      "step": 3002
+    },
+    {
+      "epoch": 1.2691170257710183,
+      "grad_norm": 1.096564531326294,
+      "learning_rate": 4.873088297422899e-05,
+      "loss": 0.0425,
+      "step": 3004
+    },
+    {
+      "epoch": 1.2699619771863118,
+      "grad_norm": 1.2362899780273438,
+      "learning_rate": 4.873003802281369e-05,
+      "loss": 0.0334,
+      "step": 3006
+    },
+    {
+      "epoch": 1.2708069286016055,
+      "grad_norm": 1.4478521347045898,
+      "learning_rate": 4.87291930713984e-05,
+      "loss": 0.0666,
+      "step": 3008
+    },
+    {
+      "epoch": 1.271651880016899,
+      "grad_norm": 0.9289385080337524,
+      "learning_rate": 4.87283481199831e-05,
+      "loss": 0.0588,
+      "step": 3010
+    },
+    {
+      "epoch": 1.2724968314321927,
+      "grad_norm": 0.5512641668319702,
+      "learning_rate": 4.872750316856781e-05,
+      "loss": 0.0457,
+      "step": 3012
+    },
+    {
+      "epoch": 1.2733417828474862,
+      "grad_norm": 1.2245798110961914,
+      "learning_rate": 4.8726658217152516e-05,
+      "loss": 0.06,
+      "step": 3014
+    },
+    {
+      "epoch": 1.27418673426278,
+      "grad_norm": 1.5422766208648682,
+      "learning_rate": 4.872581326573722e-05,
+      "loss": 0.0587,
+      "step": 3016
+    },
+    {
+      "epoch": 1.2750316856780735,
+      "grad_norm": 1.526551604270935,
+      "learning_rate": 4.872496831432193e-05,
+      "loss": 0.0808,
+      "step": 3018
+    },
+    {
+      "epoch": 1.2758766370933672,
+      "grad_norm": 1.5560036897659302,
+      "learning_rate": 4.872412336290664e-05,
+      "loss": 0.0531,
+      "step": 3020
+    },
+    {
+      "epoch": 1.2767215885086607,
+      "grad_norm": 1.0509698390960693,
+      "learning_rate": 4.872327841149134e-05,
+      "loss": 0.0342,
+      "step": 3022
+    },
+    {
+      "epoch": 1.2775665399239544,
+      "grad_norm": 1.421242356300354,
+      "learning_rate": 4.872243346007605e-05,
+      "loss": 0.0536,
+      "step": 3024
+    },
+    {
+      "epoch": 1.278411491339248,
+      "grad_norm": 0.5734511613845825,
+      "learning_rate": 4.872158850866075e-05,
+      "loss": 0.0293,
+      "step": 3026
+    },
+    {
+      "epoch": 1.2792564427545416,
+      "grad_norm": 1.554356336593628,
+      "learning_rate": 4.872074355724546e-05,
+      "loss": 0.1014,
+      "step": 3028
+    },
+    {
+      "epoch": 1.2801013941698351,
+      "grad_norm": 0.9434201717376709,
+      "learning_rate": 4.871989860583017e-05,
+      "loss": 0.0332,
+      "step": 3030
+    },
+    {
+      "epoch": 1.2809463455851289,
+      "grad_norm": 1.089858055114746,
+      "learning_rate": 4.8719053654414874e-05,
+      "loss": 0.0408,
+      "step": 3032
+    },
+    {
+      "epoch": 1.2817912970004226,
+      "grad_norm": 1.2021937370300293,
+      "learning_rate": 4.871820870299958e-05,
+      "loss": 0.0588,
+      "step": 3034
+    },
+    {
+      "epoch": 1.282636248415716,
+      "grad_norm": 0.9895578622817993,
+      "learning_rate": 4.871736375158429e-05,
+      "loss": 0.0568,
+      "step": 3036
+    },
+    {
+      "epoch": 1.2834811998310096,
+      "grad_norm": 1.1666065454483032,
+      "learning_rate": 4.871651880016899e-05,
+      "loss": 0.0351,
+      "step": 3038
+    },
+    {
+      "epoch": 1.2843261512463033,
+      "grad_norm": 0.9048338532447815,
+      "learning_rate": 4.87156738487537e-05,
+      "loss": 0.0455,
+      "step": 3040
+    },
+    {
+      "epoch": 1.285171102661597,
+      "grad_norm": 1.392256259918213,
+      "learning_rate": 4.87148288973384e-05,
+      "loss": 0.0706,
+      "step": 3042
+    },
+    {
+      "epoch": 1.2860160540768906,
+      "grad_norm": 1.8354949951171875,
+      "learning_rate": 4.871398394592312e-05,
+      "loss": 0.0558,
+      "step": 3044
+    },
+    {
+      "epoch": 1.286861005492184,
+      "grad_norm": 1.7575472593307495,
+      "learning_rate": 4.871313899450782e-05,
+      "loss": 0.0918,
+      "step": 3046
+    },
+    {
+      "epoch": 1.2877059569074778,
+      "grad_norm": 1.5376323461532593,
+      "learning_rate": 4.8712294043092525e-05,
+      "loss": 0.0524,
+      "step": 3048
+    },
+    {
+      "epoch": 1.2885509083227715,
+      "grad_norm": 1.8223731517791748,
+      "learning_rate": 4.871144909167723e-05,
+      "loss": 0.0825,
+      "step": 3050
+    },
+    {
+      "epoch": 1.289395859738065,
+      "grad_norm": 0.7369791269302368,
+      "learning_rate": 4.871060414026194e-05,
+      "loss": 0.0392,
+      "step": 3052
+    },
+    {
+      "epoch": 1.2902408111533588,
+      "grad_norm": 1.4611024856567383,
+      "learning_rate": 4.8709759188846646e-05,
+      "loss": 0.0358,
+      "step": 3054
+    },
+    {
+      "epoch": 1.2910857625686523,
+      "grad_norm": 0.8670369386672974,
+      "learning_rate": 4.8708914237431354e-05,
+      "loss": 0.0487,
+      "step": 3056
+    },
+    {
+      "epoch": 1.291930713983946,
+      "grad_norm": 1.0213311910629272,
+      "learning_rate": 4.8708069286016054e-05,
+      "loss": 0.0559,
+      "step": 3058
+    },
+    {
+      "epoch": 1.2927756653992395,
+      "grad_norm": 1.72591233253479,
+      "learning_rate": 4.870722433460077e-05,
+      "loss": 0.0665,
+      "step": 3060
+    },
+    {
+      "epoch": 1.2936206168145332,
+      "grad_norm": 0.8167611360549927,
+      "learning_rate": 4.870637938318547e-05,
+      "loss": 0.0359,
+      "step": 3062
+    },
+    {
+      "epoch": 1.2944655682298267,
+      "grad_norm": 0.7950469255447388,
+      "learning_rate": 4.8705534431770176e-05,
+      "loss": 0.0254,
+      "step": 3064
+    },
+    {
+      "epoch": 1.2953105196451205,
+      "grad_norm": 0.8578920364379883,
+      "learning_rate": 4.870468948035488e-05,
+      "loss": 0.037,
+      "step": 3066
+    },
+    {
+      "epoch": 1.296155471060414,
+      "grad_norm": 1.4960883855819702,
+      "learning_rate": 4.870384452893959e-05,
+      "loss": 0.0682,
+      "step": 3068
+    },
+    {
+      "epoch": 1.2970004224757077,
+      "grad_norm": 1.0831998586654663,
+      "learning_rate": 4.87029995775243e-05,
+      "loss": 0.0592,
+      "step": 3070
+    },
+    {
+      "epoch": 1.2978453738910012,
+      "grad_norm": 0.7494056820869446,
+      "learning_rate": 4.8702154626109004e-05,
+      "loss": 0.0346,
+      "step": 3072
+    },
+    {
+      "epoch": 1.298690325306295,
+      "grad_norm": 1.0414024591445923,
+      "learning_rate": 4.8701309674693705e-05,
+      "loss": 0.0519,
+      "step": 3074
+    },
+    {
+      "epoch": 1.2995352767215884,
+      "grad_norm": 1.9897801876068115,
+      "learning_rate": 4.870046472327842e-05,
+      "loss": 0.0363,
+      "step": 3076
+    },
+    {
+      "epoch": 1.3003802281368821,
+      "grad_norm": 1.645512580871582,
+      "learning_rate": 4.869961977186312e-05,
+      "loss": 0.0679,
+      "step": 3078
+    },
+    {
+      "epoch": 1.3012251795521759,
+      "grad_norm": 0.9590452909469604,
+      "learning_rate": 4.8698774820447826e-05,
+      "loss": 0.0389,
+      "step": 3080
+    },
+    {
+      "epoch": 1.3020701309674694,
+      "grad_norm": 1.0363378524780273,
+      "learning_rate": 4.8697929869032534e-05,
+      "loss": 0.0662,
+      "step": 3082
+    },
+    {
+      "epoch": 1.3029150823827629,
+      "grad_norm": 0.8959032297134399,
+      "learning_rate": 4.869708491761724e-05,
+      "loss": 0.0451,
+      "step": 3084
+    },
+    {
+      "epoch": 1.3037600337980566,
+      "grad_norm": 1.110154628753662,
+      "learning_rate": 4.869623996620195e-05,
+      "loss": 0.0425,
+      "step": 3086
+    },
+    {
+      "epoch": 1.3046049852133503,
+      "grad_norm": 1.2252897024154663,
+      "learning_rate": 4.8695395014786655e-05,
+      "loss": 0.0492,
+      "step": 3088
+    },
+    {
+      "epoch": 1.3054499366286438,
+      "grad_norm": 0.8930359482765198,
+      "learning_rate": 4.8694550063371356e-05,
+      "loss": 0.0376,
+      "step": 3090
+    },
+    {
+      "epoch": 1.3062948880439373,
+      "grad_norm": 1.2637935876846313,
+      "learning_rate": 4.869370511195607e-05,
+      "loss": 0.0387,
+      "step": 3092
+    },
+    {
+      "epoch": 1.307139839459231,
+      "grad_norm": 1.6533139944076538,
+      "learning_rate": 4.869286016054077e-05,
+      "loss": 0.068,
+      "step": 3094
+    },
+    {
+      "epoch": 1.3079847908745248,
+      "grad_norm": 1.4910434484481812,
+      "learning_rate": 4.869201520912548e-05,
+      "loss": 0.0341,
+      "step": 3096
+    },
+    {
+      "epoch": 1.3088297422898183,
+      "grad_norm": 1.8592034578323364,
+      "learning_rate": 4.8691170257710184e-05,
+      "loss": 0.0765,
+      "step": 3098
+    },
+    {
+      "epoch": 1.309674693705112,
+      "grad_norm": 1.5364049673080444,
+      "learning_rate": 4.869032530629489e-05,
+      "loss": 0.0805,
+      "step": 3100
+    },
+    {
+      "epoch": 1.309674693705112,
+      "eval_accuracy": 0.7233143645456113,
+      "eval_cer": 0.07973040923807087,
+      "eval_loss": 0.183185413479805,
+      "eval_runtime": 858.546,
+      "eval_samples_per_second": 13.509,
+      "eval_steps_per_second": 0.423,
+      "step": 3100
+    },
+    {
+      "epoch": 1.3105196451204055,
+      "grad_norm": 0.9121136665344238,
+      "learning_rate": 4.86894803548796e-05,
+      "loss": 0.0553,
+      "step": 3102
+    },
+    {
+      "epoch": 1.3113645965356993,
+      "grad_norm": 0.8206782341003418,
+      "learning_rate": 4.8688635403464306e-05,
+      "loss": 0.0856,
+      "step": 3104
+    },
+    {
+      "epoch": 1.3122095479509928,
+      "grad_norm": 1.5402354001998901,
+      "learning_rate": 4.8687790452049006e-05,
+      "loss": 0.0776,
+      "step": 3106
+    },
+    {
+      "epoch": 1.3130544993662865,
+      "grad_norm": 1.8080055713653564,
+      "learning_rate": 4.868694550063372e-05,
+      "loss": 0.0838,
+      "step": 3108
+    },
+    {
+      "epoch": 1.31389945078158,
+      "grad_norm": 1.5559817552566528,
+      "learning_rate": 4.868610054921842e-05,
+      "loss": 0.0727,
+      "step": 3110
+    },
+    {
+      "epoch": 1.3147444021968737,
+      "grad_norm": 1.1048095226287842,
+      "learning_rate": 4.868525559780313e-05,
+      "loss": 0.042,
+      "step": 3112
+    },
+    {
+      "epoch": 1.3155893536121672,
+      "grad_norm": 1.1088711023330688,
+      "learning_rate": 4.8684410646387835e-05,
+      "loss": 0.0363,
+      "step": 3114
+    },
+    {
+      "epoch": 1.316434305027461,
+      "grad_norm": 0.7765578031539917,
+      "learning_rate": 4.868356569497254e-05,
+      "loss": 0.0496,
+      "step": 3116
+    },
+    {
+      "epoch": 1.3172792564427547,
+      "grad_norm": 1.4490267038345337,
+      "learning_rate": 4.868272074355725e-05,
+      "loss": 0.0701,
+      "step": 3118
+    },
+    {
+      "epoch": 1.3181242078580482,
+      "grad_norm": 1.2103753089904785,
+      "learning_rate": 4.868187579214196e-05,
+      "loss": 0.0698,
+      "step": 3120
+    },
+    {
+      "epoch": 1.3189691592733417,
+      "grad_norm": 0.9798583984375,
+      "learning_rate": 4.868103084072666e-05,
+      "loss": 0.0581,
+      "step": 3122
+    },
+    {
+      "epoch": 1.3198141106886354,
+      "grad_norm": 1.6193875074386597,
+      "learning_rate": 4.868018588931137e-05,
+      "loss": 0.0594,
+      "step": 3124
+    },
+    {
+      "epoch": 1.3206590621039291,
+      "grad_norm": 1.3513240814208984,
+      "learning_rate": 4.867934093789607e-05,
+      "loss": 0.0716,
+      "step": 3126
+    },
+    {
+      "epoch": 1.3215040135192226,
+      "grad_norm": 2.1795995235443115,
+      "learning_rate": 4.867849598648078e-05,
+      "loss": 0.0645,
+      "step": 3128
+    },
+    {
+      "epoch": 1.3223489649345161,
+      "grad_norm": 1.0579677820205688,
+      "learning_rate": 4.8677651035065486e-05,
+      "loss": 0.0576,
+      "step": 3130
+    },
+    {
+      "epoch": 1.3231939163498099,
+      "grad_norm": 1.3420937061309814,
+      "learning_rate": 4.867680608365019e-05,
+      "loss": 0.0651,
+      "step": 3132
+    },
+    {
+      "epoch": 1.3240388677651036,
+      "grad_norm": 1.300595760345459,
+      "learning_rate": 4.86759611322349e-05,
+      "loss": 0.0684,
+      "step": 3134
+    },
+    {
+      "epoch": 1.324883819180397,
+      "grad_norm": 1.1893776655197144,
+      "learning_rate": 4.867511618081961e-05,
+      "loss": 0.0447,
+      "step": 3136
+    },
+    {
+      "epoch": 1.3257287705956908,
+      "grad_norm": 0.800032913684845,
+      "learning_rate": 4.867427122940431e-05,
+      "loss": 0.0312,
+      "step": 3138
+    },
+    {
+      "epoch": 1.3265737220109843,
+      "grad_norm": 0.8352811932563782,
+      "learning_rate": 4.867342627798902e-05,
+      "loss": 0.0304,
+      "step": 3140
+    },
+    {
+      "epoch": 1.327418673426278,
+      "grad_norm": 1.032954454421997,
+      "learning_rate": 4.867258132657372e-05,
+      "loss": 0.05,
+      "step": 3142
+    },
+    {
+      "epoch": 1.3282636248415716,
+      "grad_norm": 1.1263511180877686,
+      "learning_rate": 4.8671736375158436e-05,
+      "loss": 0.0373,
+      "step": 3144
+    },
+    {
+      "epoch": 1.3291085762568653,
+      "grad_norm": 0.6103000044822693,
+      "learning_rate": 4.867089142374314e-05,
+      "loss": 0.0483,
+      "step": 3146
+    },
+    {
+      "epoch": 1.3299535276721588,
+      "grad_norm": 0.8234397768974304,
+      "learning_rate": 4.8670046472327844e-05,
+      "loss": 0.0406,
+      "step": 3148
+    },
+    {
+      "epoch": 1.3307984790874525,
+      "grad_norm": 1.4005287885665894,
+      "learning_rate": 4.866920152091255e-05,
+      "loss": 0.0543,
+      "step": 3150
+    },
+    {
+      "epoch": 1.331643430502746,
+      "grad_norm": 0.7845334410667419,
+      "learning_rate": 4.866835656949726e-05,
+      "loss": 0.0483,
+      "step": 3152
+    },
+    {
+      "epoch": 1.3324883819180398,
+      "grad_norm": 1.4690190553665161,
+      "learning_rate": 4.8667511618081966e-05,
+      "loss": 0.031,
+      "step": 3154
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 1.3612943887710571,
+      "learning_rate": 4.866666666666667e-05,
+      "loss": 0.0605,
+      "step": 3156
+    },
+    {
+      "epoch": 1.334178284748627,
+      "grad_norm": 1.120639681816101,
+      "learning_rate": 4.866582171525137e-05,
+      "loss": 0.0481,
+      "step": 3158
+    },
+    {
+      "epoch": 1.3350232361639205,
+      "grad_norm": 1.0624009370803833,
+      "learning_rate": 4.866497676383609e-05,
+      "loss": 0.0894,
+      "step": 3160
+    },
+    {
+      "epoch": 1.3358681875792142,
+      "grad_norm": 0.8806793689727783,
+      "learning_rate": 4.866413181242079e-05,
+      "loss": 0.0623,
+      "step": 3162
+    },
+    {
+      "epoch": 1.336713138994508,
+      "grad_norm": 1.1015374660491943,
+      "learning_rate": 4.8663286861005495e-05,
+      "loss": 0.0596,
+      "step": 3164
+    },
+    {
+      "epoch": 1.3375580904098014,
+      "grad_norm": 1.0393707752227783,
+      "learning_rate": 4.86624419095902e-05,
+      "loss": 0.0635,
+      "step": 3166
+    },
+    {
+      "epoch": 1.338403041825095,
+      "grad_norm": 1.5667744874954224,
+      "learning_rate": 4.866159695817491e-05,
+      "loss": 0.0555,
+      "step": 3168
+    },
+    {
+      "epoch": 1.3392479932403887,
+      "grad_norm": 0.5672454237937927,
+      "learning_rate": 4.8660752006759616e-05,
+      "loss": 0.0206,
+      "step": 3170
+    },
+    {
+      "epoch": 1.3400929446556824,
+      "grad_norm": 1.0931203365325928,
+      "learning_rate": 4.8659907055344324e-05,
+      "loss": 0.0463,
+      "step": 3172
+    },
+    {
+      "epoch": 1.340937896070976,
+      "grad_norm": 1.1247515678405762,
+      "learning_rate": 4.8659062103929024e-05,
+      "loss": 0.0466,
+      "step": 3174
+    },
+    {
+      "epoch": 1.3417828474862694,
+      "grad_norm": 1.6329139471054077,
+      "learning_rate": 4.865821715251374e-05,
+      "loss": 0.0686,
+      "step": 3176
+    },
+    {
+      "epoch": 1.3426277989015631,
+      "grad_norm": 0.6330051422119141,
+      "learning_rate": 4.865737220109844e-05,
+      "loss": 0.0517,
+      "step": 3178
+    },
+    {
+      "epoch": 1.3434727503168569,
+      "grad_norm": 1.0388518571853638,
+      "learning_rate": 4.8656527249683146e-05,
+      "loss": 0.0455,
+      "step": 3180
+    },
+    {
+      "epoch": 1.3443177017321504,
+      "grad_norm": 1.2583259344100952,
+      "learning_rate": 4.865568229826785e-05,
+      "loss": 0.0837,
+      "step": 3182
+    },
+    {
+      "epoch": 1.345162653147444,
+      "grad_norm": 1.0448977947235107,
+      "learning_rate": 4.865483734685256e-05,
+      "loss": 0.0481,
+      "step": 3184
+    },
+    {
+      "epoch": 1.3460076045627376,
+      "grad_norm": 1.3680537939071655,
+      "learning_rate": 4.865399239543727e-05,
+      "loss": 0.0615,
+      "step": 3186
+    },
+    {
+      "epoch": 1.3468525559780313,
+      "grad_norm": 1.131136417388916,
+      "learning_rate": 4.8653147444021974e-05,
+      "loss": 0.0402,
+      "step": 3188
+    },
+    {
+      "epoch": 1.3476975073933248,
+      "grad_norm": 1.0717076063156128,
+      "learning_rate": 4.8652302492606675e-05,
+      "loss": 0.0569,
+      "step": 3190
+    },
+    {
+      "epoch": 1.3485424588086186,
+      "grad_norm": 0.7371670603752136,
+      "learning_rate": 4.865145754119139e-05,
+      "loss": 0.0498,
+      "step": 3192
+    },
+    {
+      "epoch": 1.349387410223912,
+      "grad_norm": 1.0945547819137573,
+      "learning_rate": 4.865061258977609e-05,
+      "loss": 0.0543,
+      "step": 3194
+    },
+    {
+      "epoch": 1.3502323616392058,
+      "grad_norm": 1.342444658279419,
+      "learning_rate": 4.8649767638360796e-05,
+      "loss": 0.0763,
+      "step": 3196
+    },
+    {
+      "epoch": 1.3510773130544993,
+      "grad_norm": 0.6870439648628235,
+      "learning_rate": 4.8648922686945504e-05,
+      "loss": 0.0454,
+      "step": 3198
+    },
+    {
+      "epoch": 1.351922264469793,
+      "grad_norm": 1.172759771347046,
+      "learning_rate": 4.864807773553021e-05,
+      "loss": 0.055,
+      "step": 3200
+    },
+    {
+      "epoch": 1.351922264469793,
+      "eval_accuracy": 0.7313329884462838,
+      "eval_cer": 0.07825718746967544,
+      "eval_loss": 0.1792062371969223,
+      "eval_runtime": 861.651,
+      "eval_samples_per_second": 13.46,
+      "eval_steps_per_second": 0.421,
+      "step": 3200
+    },
+    {
+      "epoch": 1.3527672158850865,
+      "grad_norm": 1.0497750043869019,
+      "learning_rate": 4.864723278411492e-05,
+      "loss": 0.0348,
+      "step": 3202
+    },
+    {
+      "epoch": 1.3536121673003803,
+      "grad_norm": 0.7206516265869141,
+      "learning_rate": 4.8646387832699625e-05,
+      "loss": 0.0368,
+      "step": 3204
+    },
+    {
+      "epoch": 1.3544571187156738,
+      "grad_norm": 1.5473790168762207,
+      "learning_rate": 4.8645542881284326e-05,
+      "loss": 0.053,
+      "step": 3206
+    },
+    {
+      "epoch": 1.3553020701309675,
+      "grad_norm": 1.5096811056137085,
+      "learning_rate": 4.864469792986904e-05,
+      "loss": 0.0786,
+      "step": 3208
+    },
+    {
+      "epoch": 1.3561470215462612,
+      "grad_norm": 1.3632938861846924,
+      "learning_rate": 4.864385297845374e-05,
+      "loss": 0.08,
+      "step": 3210
+    },
+    {
+      "epoch": 1.3569919729615547,
+      "grad_norm": 1.0166387557983398,
+      "learning_rate": 4.864300802703845e-05,
+      "loss": 0.0559,
+      "step": 3212
+    },
+    {
+      "epoch": 1.3578369243768482,
+      "grad_norm": 1.3810648918151855,
+      "learning_rate": 4.8642163075623154e-05,
+      "loss": 0.0533,
+      "step": 3214
+    },
+    {
+      "epoch": 1.358681875792142,
+      "grad_norm": 1.6063209772109985,
+      "learning_rate": 4.864131812420786e-05,
+      "loss": 0.0802,
+      "step": 3216
+    },
+    {
+      "epoch": 1.3595268272074357,
+      "grad_norm": 0.754176676273346,
+      "learning_rate": 4.864047317279257e-05,
+      "loss": 0.056,
+      "step": 3218
+    },
+    {
+      "epoch": 1.3603717786227292,
+      "grad_norm": 0.9545078873634338,
+      "learning_rate": 4.8639628221377276e-05,
+      "loss": 0.0346,
+      "step": 3220
+    },
+    {
+      "epoch": 1.3612167300380227,
+      "grad_norm": 1.2821192741394043,
+      "learning_rate": 4.8638783269961976e-05,
+      "loss": 0.0522,
+      "step": 3222
+    },
+    {
+      "epoch": 1.3620616814533164,
+      "grad_norm": 1.6496917009353638,
+      "learning_rate": 4.863793831854669e-05,
+      "loss": 0.0552,
+      "step": 3224
+    },
+    {
+      "epoch": 1.3629066328686101,
+      "grad_norm": 0.9524357318878174,
+      "learning_rate": 4.863709336713139e-05,
+      "loss": 0.0443,
+      "step": 3226
+    },
+    {
+      "epoch": 1.3637515842839036,
+      "grad_norm": 2.012040138244629,
+      "learning_rate": 4.86362484157161e-05,
+      "loss": 0.106,
+      "step": 3228
+    },
+    {
+      "epoch": 1.3645965356991974,
+      "grad_norm": 1.7522271871566772,
+      "learning_rate": 4.8635403464300805e-05,
+      "loss": 0.0511,
+      "step": 3230
+    },
+    {
+      "epoch": 1.3654414871144909,
+      "grad_norm": 1.4558488130569458,
+      "learning_rate": 4.863455851288551e-05,
+      "loss": 0.0682,
+      "step": 3232
+    },
+    {
+      "epoch": 1.3662864385297846,
+      "grad_norm": 1.263836145401001,
+      "learning_rate": 4.863371356147022e-05,
+      "loss": 0.0678,
+      "step": 3234
+    },
+    {
+      "epoch": 1.367131389945078,
+      "grad_norm": 1.5037480592727661,
+      "learning_rate": 4.863286861005493e-05,
+      "loss": 0.0764,
+      "step": 3236
+    },
+    {
+      "epoch": 1.3679763413603718,
+      "grad_norm": 0.8802792429924011,
+      "learning_rate": 4.863202365863963e-05,
+      "loss": 0.0575,
+      "step": 3238
+    },
+    {
+      "epoch": 1.3688212927756653,
+      "grad_norm": 1.4338186979293823,
+      "learning_rate": 4.863117870722434e-05,
+      "loss": 0.0469,
+      "step": 3240
+    },
+    {
+      "epoch": 1.369666244190959,
+      "grad_norm": 1.3018105030059814,
+      "learning_rate": 4.863033375580904e-05,
+      "loss": 0.0542,
+      "step": 3242
+    },
+    {
+      "epoch": 1.3705111956062526,
+      "grad_norm": 0.8074570894241333,
+      "learning_rate": 4.8629488804393756e-05,
+      "loss": 0.0392,
+      "step": 3244
+    },
+    {
+      "epoch": 1.3713561470215463,
+      "grad_norm": 1.513663411140442,
+      "learning_rate": 4.8628643852978456e-05,
+      "loss": 0.0563,
+      "step": 3246
+    },
+    {
+      "epoch": 1.37220109843684,
+      "grad_norm": 1.381496787071228,
+      "learning_rate": 4.862779890156316e-05,
+      "loss": 0.0526,
+      "step": 3248
+    },
+    {
+      "epoch": 1.3730460498521335,
+      "grad_norm": 1.0638294219970703,
+      "learning_rate": 4.862695395014787e-05,
+      "loss": 0.0497,
+      "step": 3250
+    },
+    {
+      "epoch": 1.373891001267427,
+      "grad_norm": 1.7604972124099731,
+      "learning_rate": 4.862610899873258e-05,
+      "loss": 0.0656,
+      "step": 3252
+    },
+    {
+      "epoch": 1.3747359526827208,
+      "grad_norm": 0.8120830059051514,
+      "learning_rate": 4.8625264047317285e-05,
+      "loss": 0.033,
+      "step": 3254
+    },
+    {
+      "epoch": 1.3755809040980145,
+      "grad_norm": 0.7674900889396667,
+      "learning_rate": 4.862441909590199e-05,
+      "loss": 0.0427,
+      "step": 3256
+    },
+    {
+      "epoch": 1.376425855513308,
+      "grad_norm": 0.8798816800117493,
+      "learning_rate": 4.862357414448669e-05,
+      "loss": 0.0474,
+      "step": 3258
+    },
+    {
+      "epoch": 1.3772708069286015,
+      "grad_norm": 0.8720738887786865,
+      "learning_rate": 4.86227291930714e-05,
+      "loss": 0.053,
+      "step": 3260
+    },
+    {
+      "epoch": 1.3781157583438952,
+      "grad_norm": 1.4881258010864258,
+      "learning_rate": 4.862188424165611e-05,
+      "loss": 0.0565,
+      "step": 3262
+    },
+    {
+      "epoch": 1.378960709759189,
+      "grad_norm": 1.2955182790756226,
+      "learning_rate": 4.8621039290240814e-05,
+      "loss": 0.0409,
+      "step": 3264
+    },
+    {
+      "epoch": 1.3798056611744824,
+      "grad_norm": 0.8371695876121521,
+      "learning_rate": 4.862019433882552e-05,
+      "loss": 0.0453,
+      "step": 3266
+    },
+    {
+      "epoch": 1.380650612589776,
+      "grad_norm": 0.7356275916099548,
+      "learning_rate": 4.861934938741022e-05,
+      "loss": 0.0385,
+      "step": 3268
+    },
+    {
+      "epoch": 1.3814955640050697,
+      "grad_norm": 0.7049744725227356,
+      "learning_rate": 4.8618504435994936e-05,
+      "loss": 0.0311,
+      "step": 3270
+    },
+    {
+      "epoch": 1.3823405154203634,
+      "grad_norm": 1.2064480781555176,
+      "learning_rate": 4.8617659484579636e-05,
+      "loss": 0.0689,
+      "step": 3272
+    },
+    {
+      "epoch": 1.383185466835657,
+      "grad_norm": 1.173024296760559,
+      "learning_rate": 4.861681453316434e-05,
+      "loss": 0.0644,
+      "step": 3274
+    },
+    {
+      "epoch": 1.3840304182509506,
+      "grad_norm": 1.1009023189544678,
+      "learning_rate": 4.861596958174905e-05,
+      "loss": 0.0343,
+      "step": 3276
+    },
+    {
+      "epoch": 1.3848753696662441,
+      "grad_norm": 1.1866307258605957,
+      "learning_rate": 4.861512463033376e-05,
+      "loss": 0.0398,
+      "step": 3278
+    },
+    {
+      "epoch": 1.3857203210815379,
+      "grad_norm": 0.8218021988868713,
+      "learning_rate": 4.8614279678918465e-05,
+      "loss": 0.0483,
+      "step": 3280
+    },
+    {
+      "epoch": 1.3865652724968314,
+      "grad_norm": 1.1046007871627808,
+      "learning_rate": 4.861343472750317e-05,
+      "loss": 0.0676,
+      "step": 3282
+    },
+    {
+      "epoch": 1.387410223912125,
+      "grad_norm": 0.6497601866722107,
+      "learning_rate": 4.861258977608787e-05,
+      "loss": 0.0293,
+      "step": 3284
+    },
+    {
+      "epoch": 1.3882551753274186,
+      "grad_norm": 1.103437066078186,
+      "learning_rate": 4.8611744824672586e-05,
+      "loss": 0.0379,
+      "step": 3286
+    },
+    {
+      "epoch": 1.3891001267427123,
+      "grad_norm": 0.9892705678939819,
+      "learning_rate": 4.861089987325729e-05,
+      "loss": 0.0557,
+      "step": 3288
+    },
+    {
+      "epoch": 1.3899450781580058,
+      "grad_norm": 0.8333448767662048,
+      "learning_rate": 4.8610054921841994e-05,
+      "loss": 0.0275,
+      "step": 3290
+    },
+    {
+      "epoch": 1.3907900295732996,
+      "grad_norm": 1.19562828540802,
+      "learning_rate": 4.86092099704267e-05,
+      "loss": 0.0497,
+      "step": 3292
+    },
+    {
+      "epoch": 1.3916349809885933,
+      "grad_norm": 1.448594093322754,
+      "learning_rate": 4.860836501901141e-05,
+      "loss": 0.0694,
+      "step": 3294
+    },
+    {
+      "epoch": 1.3924799324038868,
+      "grad_norm": 1.5646843910217285,
+      "learning_rate": 4.8607520067596116e-05,
+      "loss": 0.0426,
+      "step": 3296
+    },
+    {
+      "epoch": 1.3933248838191803,
+      "grad_norm": 1.2052522897720337,
+      "learning_rate": 4.860667511618082e-05,
+      "loss": 0.0654,
+      "step": 3298
+    },
+    {
+      "epoch": 1.394169835234474,
+      "grad_norm": 0.7969152331352234,
+      "learning_rate": 4.860583016476552e-05,
+      "loss": 0.0585,
+      "step": 3300
+    },
+    {
+      "epoch": 1.394169835234474,
+      "eval_accuracy": 0.74098982583204,
+      "eval_cer": 0.07193203772153461,
+      "eval_loss": 0.17303836345672607,
+      "eval_runtime": 852.7734,
+      "eval_samples_per_second": 13.6,
+      "eval_steps_per_second": 0.426,
+      "step": 3300
+    },
+    {
+      "epoch": 1.3950147866497677,
+      "grad_norm": 1.1091243028640747,
+      "learning_rate": 4.860498521335024e-05,
+      "loss": 0.0598,
+      "step": 3302
+    },
+    {
+      "epoch": 1.3958597380650613,
+      "grad_norm": 1.6562272310256958,
+      "learning_rate": 4.860414026193494e-05,
+      "loss": 0.0402,
+      "step": 3304
+    },
+    {
+      "epoch": 1.3967046894803548,
+      "grad_norm": 0.8330667018890381,
+      "learning_rate": 4.8603295310519645e-05,
+      "loss": 0.0563,
+      "step": 3306
+    },
+    {
+      "epoch": 1.3975496408956485,
+      "grad_norm": 1.4728366136550903,
+      "learning_rate": 4.860245035910435e-05,
+      "loss": 0.0635,
+      "step": 3308
+    },
+    {
+      "epoch": 1.3983945923109422,
+      "grad_norm": 1.3313997983932495,
+      "learning_rate": 4.860160540768906e-05,
+      "loss": 0.0547,
+      "step": 3310
+    },
+    {
+      "epoch": 1.3992395437262357,
+      "grad_norm": 1.9861879348754883,
+      "learning_rate": 4.8600760456273766e-05,
+      "loss": 0.0567,
+      "step": 3312
+    },
+    {
+      "epoch": 1.4000844951415294,
+      "grad_norm": 0.7557132244110107,
+      "learning_rate": 4.8599915504858474e-05,
+      "loss": 0.0478,
+      "step": 3314
+    },
+    {
+      "epoch": 1.400929446556823,
+      "grad_norm": 0.5707297325134277,
+      "learning_rate": 4.8599070553443174e-05,
+      "loss": 0.0377,
+      "step": 3316
+    },
+    {
+      "epoch": 1.4017743979721167,
+      "grad_norm": 1.3128173351287842,
+      "learning_rate": 4.859822560202789e-05,
+      "loss": 0.0579,
+      "step": 3318
+    },
+    {
+      "epoch": 1.4026193493874102,
+      "grad_norm": 1.7858682870864868,
+      "learning_rate": 4.859738065061259e-05,
+      "loss": 0.0718,
+      "step": 3320
+    },
+    {
+      "epoch": 1.403464300802704,
+      "grad_norm": 1.2796276807785034,
+      "learning_rate": 4.8596535699197296e-05,
+      "loss": 0.0408,
+      "step": 3322
+    },
+    {
+      "epoch": 1.4043092522179974,
+      "grad_norm": 0.8803476095199585,
+      "learning_rate": 4.8595690747782e-05,
+      "loss": 0.0522,
+      "step": 3324
+    },
+    {
+      "epoch": 1.4051542036332911,
+      "grad_norm": 0.9975403547286987,
+      "learning_rate": 4.859484579636671e-05,
+      "loss": 0.0305,
+      "step": 3326
+    },
+    {
+      "epoch": 1.4059991550485846,
+      "grad_norm": 0.4852333962917328,
+      "learning_rate": 4.859400084495142e-05,
+      "loss": 0.021,
+      "step": 3328
+    },
+    {
+      "epoch": 1.4068441064638784,
+      "grad_norm": 1.1154181957244873,
+      "learning_rate": 4.8593155893536124e-05,
+      "loss": 0.0351,
+      "step": 3330
+    },
+    {
+      "epoch": 1.4076890578791719,
+      "grad_norm": 0.9479047656059265,
+      "learning_rate": 4.8592310942120825e-05,
+      "loss": 0.0219,
+      "step": 3332
+    },
+    {
+      "epoch": 1.4085340092944656,
+      "grad_norm": 0.5760020017623901,
+      "learning_rate": 4.859146599070554e-05,
+      "loss": 0.0527,
+      "step": 3334
+    },
+    {
+      "epoch": 1.409378960709759,
+      "grad_norm": 1.5132248401641846,
+      "learning_rate": 4.859062103929024e-05,
+      "loss": 0.05,
+      "step": 3336
+    },
+    {
+      "epoch": 1.4102239121250528,
+      "grad_norm": 1.0323017835617065,
+      "learning_rate": 4.8589776087874946e-05,
+      "loss": 0.044,
+      "step": 3338
+    },
+    {
+      "epoch": 1.4110688635403466,
+      "grad_norm": 1.1576844453811646,
+      "learning_rate": 4.8588931136459654e-05,
+      "loss": 0.0751,
+      "step": 3340
+    },
+    {
+      "epoch": 1.41191381495564,
+      "grad_norm": 1.1319127082824707,
+      "learning_rate": 4.858808618504436e-05,
+      "loss": 0.0598,
+      "step": 3342
+    },
+    {
+      "epoch": 1.4127587663709336,
+      "grad_norm": 1.139543056488037,
+      "learning_rate": 4.858724123362907e-05,
+      "loss": 0.0542,
+      "step": 3344
+    },
+    {
+      "epoch": 1.4136037177862273,
+      "grad_norm": 1.0183779001235962,
+      "learning_rate": 4.8586396282213775e-05,
+      "loss": 0.0385,
+      "step": 3346
+    },
+    {
+      "epoch": 1.414448669201521,
+      "grad_norm": 0.6774150729179382,
+      "learning_rate": 4.8585551330798476e-05,
+      "loss": 0.0225,
+      "step": 3348
+    },
+    {
+      "epoch": 1.4152936206168145,
+      "grad_norm": 1.525310754776001,
+      "learning_rate": 4.858470637938319e-05,
+      "loss": 0.071,
+      "step": 3350
+    },
+    {
+      "epoch": 1.416138572032108,
+      "grad_norm": 1.1040525436401367,
+      "learning_rate": 4.858386142796789e-05,
+      "loss": 0.0322,
+      "step": 3352
+    },
+    {
+      "epoch": 1.4169835234474017,
+      "grad_norm": 1.2062792778015137,
+      "learning_rate": 4.8583016476552604e-05,
+      "loss": 0.0679,
+      "step": 3354
+    },
+    {
+      "epoch": 1.4178284748626955,
+      "grad_norm": 1.0307128429412842,
+      "learning_rate": 4.8582171525137304e-05,
+      "loss": 0.0512,
+      "step": 3356
+    },
+    {
+      "epoch": 1.418673426277989,
+      "grad_norm": 1.3051033020019531,
+      "learning_rate": 4.858132657372201e-05,
+      "loss": 0.045,
+      "step": 3358
+    },
+    {
+      "epoch": 1.4195183776932827,
+      "grad_norm": 0.960086464881897,
+      "learning_rate": 4.858048162230672e-05,
+      "loss": 0.065,
+      "step": 3360
+    },
+    {
+      "epoch": 1.4203633291085762,
+      "grad_norm": 1.8996869325637817,
+      "learning_rate": 4.8579636670891426e-05,
+      "loss": 0.0925,
+      "step": 3362
+    },
+    {
+      "epoch": 1.42120828052387,
+      "grad_norm": 1.6697046756744385,
+      "learning_rate": 4.857879171947613e-05,
+      "loss": 0.0473,
+      "step": 3364
+    },
+    {
+      "epoch": 1.4220532319391634,
+      "grad_norm": 1.2368358373641968,
+      "learning_rate": 4.857794676806084e-05,
+      "loss": 0.0495,
+      "step": 3366
+    },
+    {
+      "epoch": 1.4228981833544572,
+      "grad_norm": 1.3032385110855103,
+      "learning_rate": 4.857710181664554e-05,
+      "loss": 0.042,
+      "step": 3368
+    },
+    {
+      "epoch": 1.4237431347697507,
+      "grad_norm": 1.1829075813293457,
+      "learning_rate": 4.8576256865230255e-05,
+      "loss": 0.0909,
+      "step": 3370
+    },
+    {
+      "epoch": 1.4245880861850444,
+      "grad_norm": 1.2332669496536255,
+      "learning_rate": 4.8575411913814955e-05,
+      "loss": 0.0592,
+      "step": 3372
+    },
+    {
+      "epoch": 1.425433037600338,
+      "grad_norm": 1.6120294332504272,
+      "learning_rate": 4.857456696239966e-05,
+      "loss": 0.0711,
+      "step": 3374
+    },
+    {
+      "epoch": 1.4262779890156316,
+      "grad_norm": 1.1956902742385864,
+      "learning_rate": 4.857372201098437e-05,
+      "loss": 0.04,
+      "step": 3376
+    },
+    {
+      "epoch": 1.4271229404309254,
+      "grad_norm": 1.5087355375289917,
+      "learning_rate": 4.857287705956908e-05,
+      "loss": 0.0488,
+      "step": 3378
+    },
+    {
+      "epoch": 1.4279678918462189,
+      "grad_norm": 1.6760591268539429,
+      "learning_rate": 4.8572032108153784e-05,
+      "loss": 0.0851,
+      "step": 3380
+    },
+    {
+      "epoch": 1.4288128432615124,
+      "grad_norm": 0.8411288857460022,
+      "learning_rate": 4.857118715673849e-05,
+      "loss": 0.023,
+      "step": 3382
+    },
+    {
+      "epoch": 1.429657794676806,
+      "grad_norm": 1.0701520442962646,
+      "learning_rate": 4.857034220532319e-05,
+      "loss": 0.0479,
+      "step": 3384
+    },
+    {
+      "epoch": 1.4305027460920998,
+      "grad_norm": 0.9319422841072083,
+      "learning_rate": 4.8569497253907906e-05,
+      "loss": 0.0686,
+      "step": 3386
+    },
+    {
+      "epoch": 1.4313476975073933,
+      "grad_norm": 0.8737354278564453,
+      "learning_rate": 4.8568652302492606e-05,
+      "loss": 0.0662,
+      "step": 3388
+    },
+    {
+      "epoch": 1.4321926489226868,
+      "grad_norm": 0.5453359484672546,
+      "learning_rate": 4.856780735107731e-05,
+      "loss": 0.0324,
+      "step": 3390
+    },
+    {
+      "epoch": 1.4330376003379806,
+      "grad_norm": 0.6740847229957581,
+      "learning_rate": 4.856696239966202e-05,
+      "loss": 0.0392,
+      "step": 3392
+    },
+    {
+      "epoch": 1.4338825517532743,
+      "grad_norm": 1.1385046243667603,
+      "learning_rate": 4.856611744824673e-05,
+      "loss": 0.0392,
+      "step": 3394
+    },
+    {
+      "epoch": 1.4347275031685678,
+      "grad_norm": 0.7803374528884888,
+      "learning_rate": 4.8565272496831435e-05,
+      "loss": 0.0372,
+      "step": 3396
+    },
+    {
+      "epoch": 1.4355724545838613,
+      "grad_norm": 0.8832899928092957,
+      "learning_rate": 4.856442754541614e-05,
+      "loss": 0.0406,
+      "step": 3398
+    },
+    {
+      "epoch": 1.436417405999155,
+      "grad_norm": 1.3527898788452148,
+      "learning_rate": 4.856358259400084e-05,
+      "loss": 0.048,
+      "step": 3400
+    },
+    {
+      "epoch": 1.436417405999155,
+      "eval_accuracy": 0.7294361096740818,
+      "eval_cer": 0.07786903323129582,
+      "eval_loss": 0.1757935732603073,
+      "eval_runtime": 862.8655,
+      "eval_samples_per_second": 13.441,
+      "eval_steps_per_second": 0.421,
+      "step": 3400
+    },
+    {
+      "epoch": 1.4372623574144487,
+      "grad_norm": 0.9552874565124512,
+      "learning_rate": 4.8562737642585556e-05,
+      "loss": 0.0644,
+      "step": 3402
+    },
+    {
+      "epoch": 1.4381073088297422,
+      "grad_norm": 1.3589330911636353,
+      "learning_rate": 4.856189269117026e-05,
+      "loss": 0.0489,
+      "step": 3404
+    },
+    {
+      "epoch": 1.438952260245036,
+      "grad_norm": 0.9378911852836609,
+      "learning_rate": 4.8561047739754964e-05,
+      "loss": 0.0302,
+      "step": 3406
+    },
+    {
+      "epoch": 1.4397972116603295,
+      "grad_norm": 1.4078298807144165,
+      "learning_rate": 4.856020278833967e-05,
+      "loss": 0.0621,
+      "step": 3408
+    },
+    {
+      "epoch": 1.4406421630756232,
+      "grad_norm": 0.6866748929023743,
+      "learning_rate": 4.855935783692438e-05,
+      "loss": 0.0289,
+      "step": 3410
+    },
+    {
+      "epoch": 1.4414871144909167,
+      "grad_norm": 1.534067988395691,
+      "learning_rate": 4.8558512885509086e-05,
+      "loss": 0.0658,
+      "step": 3412
+    },
+    {
+      "epoch": 1.4423320659062104,
+      "grad_norm": 1.2581876516342163,
+      "learning_rate": 4.855766793409379e-05,
+      "loss": 0.0338,
+      "step": 3414
+    },
+    {
+      "epoch": 1.443177017321504,
+      "grad_norm": 0.8533525466918945,
+      "learning_rate": 4.855682298267849e-05,
+      "loss": 0.0269,
+      "step": 3416
+    },
+    {
+      "epoch": 1.4440219687367977,
+      "grad_norm": 1.2948267459869385,
+      "learning_rate": 4.855597803126321e-05,
+      "loss": 0.0476,
+      "step": 3418
+    },
+    {
+      "epoch": 1.4448669201520912,
+      "grad_norm": 1.1799198389053345,
+      "learning_rate": 4.855513307984791e-05,
+      "loss": 0.0315,
+      "step": 3420
+    },
+    {
+      "epoch": 1.445711871567385,
+      "grad_norm": 1.5496565103530884,
+      "learning_rate": 4.8554288128432615e-05,
+      "loss": 0.0595,
+      "step": 3422
+    },
+    {
+      "epoch": 1.4465568229826786,
+      "grad_norm": 1.3309407234191895,
+      "learning_rate": 4.855344317701732e-05,
+      "loss": 0.0335,
+      "step": 3424
+    },
+    {
+      "epoch": 1.4474017743979721,
+      "grad_norm": 0.614343523979187,
+      "learning_rate": 4.855259822560203e-05,
+      "loss": 0.0232,
+      "step": 3426
+    },
+    {
+      "epoch": 1.4482467258132656,
+      "grad_norm": 1.4628639221191406,
+      "learning_rate": 4.8551753274186736e-05,
+      "loss": 0.0552,
+      "step": 3428
+    },
+    {
+      "epoch": 1.4490916772285594,
+      "grad_norm": 1.2969770431518555,
+      "learning_rate": 4.8550908322771444e-05,
+      "loss": 0.058,
+      "step": 3430
+    },
+    {
+      "epoch": 1.449936628643853,
+      "grad_norm": 1.3628487586975098,
+      "learning_rate": 4.8550063371356144e-05,
+      "loss": 0.0558,
+      "step": 3432
+    },
+    {
+      "epoch": 1.4507815800591466,
+      "grad_norm": 1.4558742046356201,
+      "learning_rate": 4.854921841994086e-05,
+      "loss": 0.0619,
+      "step": 3434
+    },
+    {
+      "epoch": 1.45162653147444,
+      "grad_norm": 1.390631079673767,
+      "learning_rate": 4.854837346852556e-05,
+      "loss": 0.0686,
+      "step": 3436
+    },
+    {
+      "epoch": 1.4524714828897338,
+      "grad_norm": 1.0281144380569458,
+      "learning_rate": 4.8547528517110266e-05,
+      "loss": 0.039,
+      "step": 3438
+    },
+    {
+      "epoch": 1.4533164343050275,
+      "grad_norm": 1.603379726409912,
+      "learning_rate": 4.854668356569497e-05,
+      "loss": 0.0877,
+      "step": 3440
+    },
+    {
+      "epoch": 1.454161385720321,
+      "grad_norm": 0.7130106091499329,
+      "learning_rate": 4.854583861427968e-05,
+      "loss": 0.0424,
+      "step": 3442
+    },
+    {
+      "epoch": 1.4550063371356148,
+      "grad_norm": 1.1394532918930054,
+      "learning_rate": 4.854499366286439e-05,
+      "loss": 0.0464,
+      "step": 3444
+    },
+    {
+      "epoch": 1.4558512885509083,
+      "grad_norm": 1.2217916250228882,
+      "learning_rate": 4.8544148711449094e-05,
+      "loss": 0.0582,
+      "step": 3446
+    },
+    {
+      "epoch": 1.456696239966202,
+      "grad_norm": 1.2456247806549072,
+      "learning_rate": 4.8543303760033795e-05,
+      "loss": 0.0477,
+      "step": 3448
+    },
+    {
+      "epoch": 1.4575411913814955,
+      "grad_norm": 0.7696735858917236,
+      "learning_rate": 4.854245880861851e-05,
+      "loss": 0.0342,
+      "step": 3450
+    },
+    {
+      "epoch": 1.4583861427967892,
+      "grad_norm": 1.327634334564209,
+      "learning_rate": 4.854161385720321e-05,
+      "loss": 0.0875,
+      "step": 3452
+    },
+    {
+      "epoch": 1.4592310942120827,
+      "grad_norm": 1.5110286474227905,
+      "learning_rate": 4.854076890578792e-05,
+      "loss": 0.057,
+      "step": 3454
+    },
+    {
+      "epoch": 1.4600760456273765,
+      "grad_norm": 1.0184332132339478,
+      "learning_rate": 4.8539923954372624e-05,
+      "loss": 0.0526,
+      "step": 3456
+    },
+    {
+      "epoch": 1.46092099704267,
+      "grad_norm": 1.3125816583633423,
+      "learning_rate": 4.853907900295733e-05,
+      "loss": 0.0311,
+      "step": 3458
+    },
+    {
+      "epoch": 1.4617659484579637,
+      "grad_norm": 1.00360906124115,
+      "learning_rate": 4.853823405154204e-05,
+      "loss": 0.0511,
+      "step": 3460
+    },
+    {
+      "epoch": 1.4626108998732572,
+      "grad_norm": 0.9407820701599121,
+      "learning_rate": 4.8537389100126745e-05,
+      "loss": 0.041,
+      "step": 3462
+    },
+    {
+      "epoch": 1.463455851288551,
+      "grad_norm": 1.2548154592514038,
+      "learning_rate": 4.853654414871145e-05,
+      "loss": 0.0588,
+      "step": 3464
+    },
+    {
+      "epoch": 1.4643008027038444,
+      "grad_norm": 1.1931437253952026,
+      "learning_rate": 4.853569919729616e-05,
+      "loss": 0.0372,
+      "step": 3466
+    },
+    {
+      "epoch": 1.4651457541191382,
+      "grad_norm": 1.3026175498962402,
+      "learning_rate": 4.853485424588086e-05,
+      "loss": 0.0482,
+      "step": 3468
+    },
+    {
+      "epoch": 1.465990705534432,
+      "grad_norm": 1.1066519021987915,
+      "learning_rate": 4.8534009294465574e-05,
+      "loss": 0.0604,
+      "step": 3470
+    },
+    {
+      "epoch": 1.4668356569497254,
+      "grad_norm": 1.1285970211029053,
+      "learning_rate": 4.8533164343050274e-05,
+      "loss": 0.0729,
+      "step": 3472
+    },
+    {
+      "epoch": 1.467680608365019,
+      "grad_norm": 1.0288052558898926,
+      "learning_rate": 4.853231939163498e-05,
+      "loss": 0.0488,
+      "step": 3474
+    },
+    {
+      "epoch": 1.4685255597803126,
+      "grad_norm": 1.7681716680526733,
+      "learning_rate": 4.853147444021969e-05,
+      "loss": 0.0572,
+      "step": 3476
+    },
+    {
+      "epoch": 1.4693705111956064,
+      "grad_norm": 1.4394867420196533,
+      "learning_rate": 4.8530629488804396e-05,
+      "loss": 0.0669,
+      "step": 3478
+    },
+    {
+      "epoch": 1.4702154626108999,
+      "grad_norm": 0.686235249042511,
+      "learning_rate": 4.85297845373891e-05,
+      "loss": 0.0246,
+      "step": 3480
+    },
+    {
+      "epoch": 1.4710604140261934,
+      "grad_norm": 2.047348976135254,
+      "learning_rate": 4.852893958597381e-05,
+      "loss": 0.0798,
+      "step": 3482
+    },
+    {
+      "epoch": 1.471905365441487,
+      "grad_norm": 1.1849197149276733,
+      "learning_rate": 4.852809463455851e-05,
+      "loss": 0.0487,
+      "step": 3484
+    },
+    {
+      "epoch": 1.4727503168567808,
+      "grad_norm": 0.973675012588501,
+      "learning_rate": 4.8527249683143225e-05,
+      "loss": 0.0483,
+      "step": 3486
+    },
+    {
+      "epoch": 1.4735952682720743,
+      "grad_norm": 1.173438549041748,
+      "learning_rate": 4.8526404731727925e-05,
+      "loss": 0.0497,
+      "step": 3488
+    },
+    {
+      "epoch": 1.474440219687368,
+      "grad_norm": 0.9411981701850891,
+      "learning_rate": 4.852555978031263e-05,
+      "loss": 0.0599,
+      "step": 3490
+    },
+    {
+      "epoch": 1.4752851711026616,
+      "grad_norm": 0.963691771030426,
+      "learning_rate": 4.852471482889734e-05,
+      "loss": 0.0372,
+      "step": 3492
+    },
+    {
+      "epoch": 1.4761301225179553,
+      "grad_norm": 1.085005283355713,
+      "learning_rate": 4.852386987748205e-05,
+      "loss": 0.0612,
+      "step": 3494
+    },
+    {
+      "epoch": 1.4769750739332488,
+      "grad_norm": 1.5302870273590088,
+      "learning_rate": 4.8523024926066754e-05,
+      "loss": 0.0519,
+      "step": 3496
+    },
+    {
+      "epoch": 1.4778200253485425,
+      "grad_norm": 1.8927478790283203,
+      "learning_rate": 4.852217997465146e-05,
+      "loss": 0.0782,
+      "step": 3498
+    },
+    {
+      "epoch": 1.478664976763836,
+      "grad_norm": 1.595861554145813,
+      "learning_rate": 4.852133502323616e-05,
+      "loss": 0.0766,
+      "step": 3500
+    },
+    {
+      "epoch": 1.478664976763836,
+      "eval_accuracy": 0.7130539748232454,
+      "eval_cer": 0.08314440219836446,
+      "eval_loss": 0.18499918282032013,
+      "eval_runtime": 847.7155,
+      "eval_samples_per_second": 13.681,
+      "eval_steps_per_second": 0.428,
+      "step": 3500
+    },
+    {
+      "epoch": 1.4795099281791297,
+      "grad_norm": 2.2448182106018066,
+      "learning_rate": 4.8520490071820876e-05,
+      "loss": 0.0531,
+      "step": 3502
+    },
+    {
+      "epoch": 1.4803548795944232,
+      "grad_norm": 1.2173892259597778,
+      "learning_rate": 4.8519645120405576e-05,
+      "loss": 0.0576,
+      "step": 3504
+    },
+    {
+      "epoch": 1.481199831009717,
+      "grad_norm": 0.8814436793327332,
+      "learning_rate": 4.851880016899028e-05,
+      "loss": 0.0401,
+      "step": 3506
+    },
+    {
+      "epoch": 1.4820447824250107,
+      "grad_norm": 1.2326481342315674,
+      "learning_rate": 4.851795521757499e-05,
+      "loss": 0.0289,
+      "step": 3508
+    },
+    {
+      "epoch": 1.4828897338403042,
+      "grad_norm": 1.1329704523086548,
+      "learning_rate": 4.85171102661597e-05,
+      "loss": 0.0597,
+      "step": 3510
+    },
+    {
+      "epoch": 1.4837346852555977,
+      "grad_norm": 1.6266759634017944,
+      "learning_rate": 4.8516265314744405e-05,
+      "loss": 0.0549,
+      "step": 3512
+    },
+    {
+      "epoch": 1.4845796366708914,
+      "grad_norm": 0.7869406342506409,
+      "learning_rate": 4.851542036332911e-05,
+      "loss": 0.0411,
+      "step": 3514
+    },
+    {
+      "epoch": 1.4854245880861852,
+      "grad_norm": 1.4602829217910767,
+      "learning_rate": 4.851457541191381e-05,
+      "loss": 0.0635,
+      "step": 3516
+    },
+    {
+      "epoch": 1.4862695395014787,
+      "grad_norm": 1.53984534740448,
+      "learning_rate": 4.8513730460498526e-05,
+      "loss": 0.062,
+      "step": 3518
+    },
+    {
+      "epoch": 1.4871144909167722,
+      "grad_norm": 1.5901530981063843,
+      "learning_rate": 4.851288550908323e-05,
+      "loss": 0.0504,
+      "step": 3520
+    },
+    {
+      "epoch": 1.487959442332066,
+      "grad_norm": 1.6808077096939087,
+      "learning_rate": 4.8512040557667934e-05,
+      "loss": 0.0728,
+      "step": 3522
+    },
+    {
+      "epoch": 1.4888043937473596,
+      "grad_norm": 1.1791267395019531,
+      "learning_rate": 4.851119560625264e-05,
+      "loss": 0.0568,
+      "step": 3524
+    },
+    {
+      "epoch": 1.4896493451626531,
+      "grad_norm": 0.7604468464851379,
+      "learning_rate": 4.851035065483735e-05,
+      "loss": 0.0348,
+      "step": 3526
+    },
+    {
+      "epoch": 1.4904942965779466,
+      "grad_norm": 0.5152177214622498,
+      "learning_rate": 4.8509505703422056e-05,
+      "loss": 0.0358,
+      "step": 3528
+    },
+    {
+      "epoch": 1.4913392479932404,
+      "grad_norm": 1.253456473350525,
+      "learning_rate": 4.850866075200676e-05,
+      "loss": 0.0637,
+      "step": 3530
+    },
+    {
+      "epoch": 1.492184199408534,
+      "grad_norm": 1.5630178451538086,
+      "learning_rate": 4.850781580059146e-05,
+      "loss": 0.0478,
+      "step": 3532
+    },
+    {
+      "epoch": 1.4930291508238276,
+      "grad_norm": 0.7693706154823303,
+      "learning_rate": 4.850697084917618e-05,
+      "loss": 0.0295,
+      "step": 3534
+    },
+    {
+      "epoch": 1.4938741022391213,
+      "grad_norm": 1.4552922248840332,
+      "learning_rate": 4.850612589776088e-05,
+      "loss": 0.0498,
+      "step": 3536
+    },
+    {
+      "epoch": 1.4947190536544148,
+      "grad_norm": 1.1829502582550049,
+      "learning_rate": 4.8505280946345585e-05,
+      "loss": 0.036,
+      "step": 3538
+    },
+    {
+      "epoch": 1.4955640050697085,
+      "grad_norm": 1.4774078130722046,
+      "learning_rate": 4.850443599493029e-05,
+      "loss": 0.0378,
+      "step": 3540
+    },
+    {
+      "epoch": 1.496408956485002,
+      "grad_norm": 1.580325961112976,
+      "learning_rate": 4.8503591043515e-05,
+      "loss": 0.0629,
+      "step": 3542
+    },
+    {
+      "epoch": 1.4972539079002958,
+      "grad_norm": 0.8769638538360596,
+      "learning_rate": 4.8502746092099707e-05,
+      "loss": 0.0229,
+      "step": 3544
+    },
+    {
+      "epoch": 1.4980988593155893,
+      "grad_norm": 0.45959189534187317,
+      "learning_rate": 4.8501901140684414e-05,
+      "loss": 0.035,
+      "step": 3546
+    },
+    {
+      "epoch": 1.498943810730883,
+      "grad_norm": 1.3305035829544067,
+      "learning_rate": 4.8501056189269114e-05,
+      "loss": 0.0617,
+      "step": 3548
+    },
+    {
+      "epoch": 1.4997887621461765,
+      "grad_norm": 0.7923805713653564,
+      "learning_rate": 4.850021123785383e-05,
+      "loss": 0.0263,
+      "step": 3550
+    },
+    {
+      "epoch": 1.5006337135614702,
+      "grad_norm": 1.4536417722702026,
+      "learning_rate": 4.849936628643853e-05,
+      "loss": 0.0513,
+      "step": 3552
+    },
+    {
+      "epoch": 1.501478664976764,
+      "grad_norm": 1.329981803894043,
+      "learning_rate": 4.849852133502324e-05,
+      "loss": 0.0615,
+      "step": 3554
+    },
+    {
+      "epoch": 1.5023236163920575,
+      "grad_norm": 0.9799890518188477,
+      "learning_rate": 4.849767638360794e-05,
+      "loss": 0.0483,
+      "step": 3556
+    },
+    {
+      "epoch": 1.503168567807351,
+      "grad_norm": 0.7838258743286133,
+      "learning_rate": 4.849683143219265e-05,
+      "loss": 0.0378,
+      "step": 3558
+    },
+    {
+      "epoch": 1.5040135192226447,
+      "grad_norm": 1.4174835681915283,
+      "learning_rate": 4.849598648077736e-05,
+      "loss": 0.057,
+      "step": 3560
+    },
+    {
+      "epoch": 1.5048584706379384,
+      "grad_norm": 1.955910325050354,
+      "learning_rate": 4.8495141529362065e-05,
+      "loss": 0.0573,
+      "step": 3562
+    },
+    {
+      "epoch": 1.505703422053232,
+      "grad_norm": 1.1473673582077026,
+      "learning_rate": 4.849429657794677e-05,
+      "loss": 0.0664,
+      "step": 3564
+    },
+    {
+      "epoch": 1.5065483734685254,
+      "grad_norm": 1.1155494451522827,
+      "learning_rate": 4.849345162653148e-05,
+      "loss": 0.0519,
+      "step": 3566
+    },
+    {
+      "epoch": 1.5073933248838192,
+      "grad_norm": 1.3899874687194824,
+      "learning_rate": 4.849260667511618e-05,
+      "loss": 0.0513,
+      "step": 3568
+    },
+    {
+      "epoch": 1.508238276299113,
+      "grad_norm": 1.4755295515060425,
+      "learning_rate": 4.849176172370089e-05,
+      "loss": 0.0808,
+      "step": 3570
+    },
+    {
+      "epoch": 1.5090832277144064,
+      "grad_norm": 2.025038242340088,
+      "learning_rate": 4.8490916772285594e-05,
+      "loss": 0.07,
+      "step": 3572
+    },
+    {
+      "epoch": 1.5099281791297,
+      "grad_norm": 1.5562915802001953,
+      "learning_rate": 4.84900718208703e-05,
+      "loss": 0.0648,
+      "step": 3574
+    },
+    {
+      "epoch": 1.5107731305449936,
+      "grad_norm": 1.214245319366455,
+      "learning_rate": 4.848922686945501e-05,
+      "loss": 0.031,
+      "step": 3576
+    },
+    {
+      "epoch": 1.5116180819602874,
+      "grad_norm": 1.378103256225586,
+      "learning_rate": 4.8488381918039715e-05,
+      "loss": 0.0728,
+      "step": 3578
+    },
+    {
+      "epoch": 1.5124630333755809,
+      "grad_norm": 1.2112326622009277,
+      "learning_rate": 4.848753696662442e-05,
+      "loss": 0.0583,
+      "step": 3580
+    },
+    {
+      "epoch": 1.5133079847908744,
+      "grad_norm": 0.9792726039886475,
+      "learning_rate": 4.848669201520913e-05,
+      "loss": 0.0547,
+      "step": 3582
+    },
+    {
+      "epoch": 1.5141529362061683,
+      "grad_norm": 0.9429163336753845,
+      "learning_rate": 4.848584706379383e-05,
+      "loss": 0.0405,
+      "step": 3584
+    },
+    {
+      "epoch": 1.5149978876214618,
+      "grad_norm": 1.9372706413269043,
+      "learning_rate": 4.8485002112378544e-05,
+      "loss": 0.0825,
+      "step": 3586
+    },
+    {
+      "epoch": 1.5158428390367553,
+      "grad_norm": 0.8863945007324219,
+      "learning_rate": 4.8484157160963245e-05,
+      "loss": 0.0645,
+      "step": 3588
+    },
+    {
+      "epoch": 1.516687790452049,
+      "grad_norm": 0.756432831287384,
+      "learning_rate": 4.848331220954795e-05,
+      "loss": 0.0399,
+      "step": 3590
+    },
+    {
+      "epoch": 1.5175327418673428,
+      "grad_norm": 0.6821669936180115,
+      "learning_rate": 4.848246725813266e-05,
+      "loss": 0.0479,
+      "step": 3592
+    },
+    {
+      "epoch": 1.5183776932826363,
+      "grad_norm": 0.9661495089530945,
+      "learning_rate": 4.8481622306717366e-05,
+      "loss": 0.0469,
+      "step": 3594
+    },
+    {
+      "epoch": 1.5192226446979298,
+      "grad_norm": 0.7377157807350159,
+      "learning_rate": 4.848077735530207e-05,
+      "loss": 0.0441,
+      "step": 3596
+    },
+    {
+      "epoch": 1.5200675961132235,
+      "grad_norm": 1.2411776781082153,
+      "learning_rate": 4.847993240388678e-05,
+      "loss": 0.0629,
+      "step": 3598
+    },
+    {
+      "epoch": 1.5209125475285172,
+      "grad_norm": 0.49103307723999023,
+      "learning_rate": 4.847908745247148e-05,
+      "loss": 0.0325,
+      "step": 3600
+    },
+    {
+      "epoch": 1.5209125475285172,
+      "eval_accuracy": 0.7358165200896706,
+      "eval_cer": 0.07696039944599804,
+      "eval_loss": 0.17656931281089783,
+      "eval_runtime": 856.6087,
+      "eval_samples_per_second": 13.539,
+      "eval_steps_per_second": 0.424,
+      "step": 3600
+    },
+    {
+      "epoch": 1.5217574989438107,
+      "grad_norm": 1.3446134328842163,
+      "learning_rate": 4.8478242501056195e-05,
+      "loss": 0.0377,
+      "step": 3602
+    },
+    {
+      "epoch": 1.5226024503591042,
+      "grad_norm": 1.5772300958633423,
+      "learning_rate": 4.8477397549640895e-05,
+      "loss": 0.0764,
+      "step": 3604
+    },
+    {
+      "epoch": 1.523447401774398,
+      "grad_norm": 1.2345671653747559,
+      "learning_rate": 4.84765525982256e-05,
+      "loss": 0.0713,
+      "step": 3606
+    },
+    {
+      "epoch": 1.5242923531896917,
+      "grad_norm": 0.7605233192443848,
+      "learning_rate": 4.847570764681031e-05,
+      "loss": 0.0466,
+      "step": 3608
+    },
+    {
+      "epoch": 1.5251373046049852,
+      "grad_norm": 1.6644865274429321,
+      "learning_rate": 4.847486269539502e-05,
+      "loss": 0.0703,
+      "step": 3610
+    },
+    {
+      "epoch": 1.5259822560202787,
+      "grad_norm": 1.4693437814712524,
+      "learning_rate": 4.8474017743979724e-05,
+      "loss": 0.041,
+      "step": 3612
+    },
+    {
+      "epoch": 1.5268272074355724,
+      "grad_norm": 0.45182543992996216,
+      "learning_rate": 4.847317279256443e-05,
+      "loss": 0.0364,
+      "step": 3614
+    },
+    {
+      "epoch": 1.5276721588508662,
+      "grad_norm": 0.8842322826385498,
+      "learning_rate": 4.847232784114913e-05,
+      "loss": 0.0299,
+      "step": 3616
+    },
+    {
+      "epoch": 1.5285171102661597,
+      "grad_norm": 1.3502392768859863,
+      "learning_rate": 4.8471482889733846e-05,
+      "loss": 0.0492,
+      "step": 3618
+    },
+    {
+      "epoch": 1.5293620616814532,
+      "grad_norm": 0.7896035313606262,
+      "learning_rate": 4.8470637938318546e-05,
+      "loss": 0.0429,
+      "step": 3620
+    },
+    {
+      "epoch": 1.530207013096747,
+      "grad_norm": 0.8707165122032166,
+      "learning_rate": 4.846979298690325e-05,
+      "loss": 0.0199,
+      "step": 3622
+    },
+    {
+      "epoch": 1.5310519645120406,
+      "grad_norm": 1.013792634010315,
+      "learning_rate": 4.846894803548796e-05,
+      "loss": 0.0599,
+      "step": 3624
+    },
+    {
+      "epoch": 1.5318969159273341,
+      "grad_norm": 1.2326784133911133,
+      "learning_rate": 4.846810308407267e-05,
+      "loss": 0.05,
+      "step": 3626
+    },
+    {
+      "epoch": 1.5327418673426279,
+      "grad_norm": 1.147342562675476,
+      "learning_rate": 4.8467258132657375e-05,
+      "loss": 0.0525,
+      "step": 3628
+    },
+    {
+      "epoch": 1.5335868187579216,
+      "grad_norm": 1.6748418807983398,
+      "learning_rate": 4.846641318124208e-05,
+      "loss": 0.0703,
+      "step": 3630
+    },
+    {
+      "epoch": 1.534431770173215,
+      "grad_norm": 0.6268454790115356,
+      "learning_rate": 4.846556822982678e-05,
+      "loss": 0.0372,
+      "step": 3632
+    },
+    {
+      "epoch": 1.5352767215885086,
+      "grad_norm": 0.7244894504547119,
+      "learning_rate": 4.8464723278411497e-05,
+      "loss": 0.0441,
+      "step": 3634
+    },
+    {
+      "epoch": 1.5361216730038023,
+      "grad_norm": 1.510006070137024,
+      "learning_rate": 4.84638783269962e-05,
+      "loss": 0.0542,
+      "step": 3636
+    },
+    {
+      "epoch": 1.536966624419096,
+      "grad_norm": 1.2709475755691528,
+      "learning_rate": 4.8463033375580904e-05,
+      "loss": 0.0383,
+      "step": 3638
+    },
+    {
+      "epoch": 1.5378115758343895,
+      "grad_norm": 1.0092389583587646,
+      "learning_rate": 4.846218842416561e-05,
+      "loss": 0.0452,
+      "step": 3640
+    },
+    {
+      "epoch": 1.538656527249683,
+      "grad_norm": 1.1898479461669922,
+      "learning_rate": 4.846134347275032e-05,
+      "loss": 0.0454,
+      "step": 3642
+    },
+    {
+      "epoch": 1.5395014786649768,
+      "grad_norm": 1.168468952178955,
+      "learning_rate": 4.8460498521335026e-05,
+      "loss": 0.0367,
+      "step": 3644
+    },
+    {
+      "epoch": 1.5403464300802705,
+      "grad_norm": 0.954703152179718,
+      "learning_rate": 4.845965356991973e-05,
+      "loss": 0.0181,
+      "step": 3646
+    },
+    {
+      "epoch": 1.541191381495564,
+      "grad_norm": 1.3118579387664795,
+      "learning_rate": 4.845880861850443e-05,
+      "loss": 0.0656,
+      "step": 3648
+    },
+    {
+      "epoch": 1.5420363329108575,
+      "grad_norm": 0.525249719619751,
+      "learning_rate": 4.845796366708915e-05,
+      "loss": 0.0246,
+      "step": 3650
+    },
+    {
+      "epoch": 1.5428812843261512,
+      "grad_norm": 1.0019251108169556,
+      "learning_rate": 4.845711871567385e-05,
+      "loss": 0.0469,
+      "step": 3652
+    },
+    {
+      "epoch": 1.543726235741445,
+      "grad_norm": 1.4187920093536377,
+      "learning_rate": 4.845627376425856e-05,
+      "loss": 0.0472,
+      "step": 3654
+    },
+    {
+      "epoch": 1.5445711871567385,
+      "grad_norm": 1.3827303647994995,
+      "learning_rate": 4.845542881284326e-05,
+      "loss": 0.0635,
+      "step": 3656
+    },
+    {
+      "epoch": 1.545416138572032,
+      "grad_norm": 1.0358022451400757,
+      "learning_rate": 4.845458386142797e-05,
+      "loss": 0.0414,
+      "step": 3658
+    },
+    {
+      "epoch": 1.5462610899873257,
+      "grad_norm": 1.110398530960083,
+      "learning_rate": 4.8453738910012677e-05,
+      "loss": 0.036,
+      "step": 3660
+    },
+    {
+      "epoch": 1.5471060414026194,
+      "grad_norm": 0.48275184631347656,
+      "learning_rate": 4.8452893958597384e-05,
+      "loss": 0.0356,
+      "step": 3662
+    },
+    {
+      "epoch": 1.547950992817913,
+      "grad_norm": 0.5242049694061279,
+      "learning_rate": 4.845204900718209e-05,
+      "loss": 0.0235,
+      "step": 3664
+    },
+    {
+      "epoch": 1.5487959442332064,
+      "grad_norm": 1.1783398389816284,
+      "learning_rate": 4.84512040557668e-05,
+      "loss": 0.0572,
+      "step": 3666
+    },
+    {
+      "epoch": 1.5496408956485002,
+      "grad_norm": 1.400747299194336,
+      "learning_rate": 4.84503591043515e-05,
+      "loss": 0.0575,
+      "step": 3668
+    },
+    {
+      "epoch": 1.5504858470637939,
+      "grad_norm": 1.717510461807251,
+      "learning_rate": 4.844951415293621e-05,
+      "loss": 0.0668,
+      "step": 3670
+    },
+    {
+      "epoch": 1.5513307984790874,
+      "grad_norm": 1.1266288757324219,
+      "learning_rate": 4.844866920152091e-05,
+      "loss": 0.0462,
+      "step": 3672
+    },
+    {
+      "epoch": 1.5521757498943811,
+      "grad_norm": 0.7779021263122559,
+      "learning_rate": 4.844782425010562e-05,
+      "loss": 0.046,
+      "step": 3674
+    },
+    {
+      "epoch": 1.5530207013096748,
+      "grad_norm": 1.299617886543274,
+      "learning_rate": 4.844697929869033e-05,
+      "loss": 0.0495,
+      "step": 3676
+    },
+    {
+      "epoch": 1.5538656527249683,
+      "grad_norm": 0.6191155910491943,
+      "learning_rate": 4.8446134347275035e-05,
+      "loss": 0.0419,
+      "step": 3678
+    },
+    {
+      "epoch": 1.5547106041402619,
+      "grad_norm": 1.5401053428649902,
+      "learning_rate": 4.844528939585974e-05,
+      "loss": 0.0492,
+      "step": 3680
+    },
+    {
+      "epoch": 1.5555555555555556,
+      "grad_norm": 1.5327844619750977,
+      "learning_rate": 4.844444444444445e-05,
+      "loss": 0.0692,
+      "step": 3682
+    },
+    {
+      "epoch": 1.5564005069708493,
+      "grad_norm": 0.7170135378837585,
+      "learning_rate": 4.844359949302915e-05,
+      "loss": 0.0327,
+      "step": 3684
+    },
+    {
+      "epoch": 1.5572454583861428,
+      "grad_norm": 1.2383910417556763,
+      "learning_rate": 4.844275454161386e-05,
+      "loss": 0.0454,
+      "step": 3686
+    },
+    {
+      "epoch": 1.5580904098014363,
+      "grad_norm": 1.0864990949630737,
+      "learning_rate": 4.8441909590198564e-05,
+      "loss": 0.0492,
+      "step": 3688
+    },
+    {
+      "epoch": 1.55893536121673,
+      "grad_norm": 1.1290335655212402,
+      "learning_rate": 4.844106463878327e-05,
+      "loss": 0.0443,
+      "step": 3690
+    },
+    {
+      "epoch": 1.5597803126320238,
+      "grad_norm": 1.9262590408325195,
+      "learning_rate": 4.844021968736798e-05,
+      "loss": 0.06,
+      "step": 3692
+    },
+    {
+      "epoch": 1.5606252640473173,
+      "grad_norm": 1.2128050327301025,
+      "learning_rate": 4.8439374735952685e-05,
+      "loss": 0.0707,
+      "step": 3694
+    },
+    {
+      "epoch": 1.5614702154626108,
+      "grad_norm": 0.8806398510932922,
+      "learning_rate": 4.843852978453739e-05,
+      "loss": 0.0451,
+      "step": 3696
+    },
+    {
+      "epoch": 1.5623151668779045,
+      "grad_norm": 1.6122541427612305,
+      "learning_rate": 4.84376848331221e-05,
+      "loss": 0.0485,
+      "step": 3698
+    },
+    {
+      "epoch": 1.5631601182931982,
+      "grad_norm": 1.153076171875,
+      "learning_rate": 4.84368398817068e-05,
+      "loss": 0.0568,
+      "step": 3700
+    },
+    {
+      "epoch": 1.5631601182931982,
+      "eval_accuracy": 0.7448698051388171,
+      "eval_cer": 0.07185264253641152,
+      "eval_loss": 0.16689425706863403,
+      "eval_runtime": 859.966,
+      "eval_samples_per_second": 13.487,
+      "eval_steps_per_second": 0.422,
+      "step": 3700
+    },
+    {
+      "epoch": 1.5640050697084917,
+      "grad_norm": 1.530321478843689,
+      "learning_rate": 4.8435994930291514e-05,
+      "loss": 0.0513,
+      "step": 3702
+    },
+    {
+      "epoch": 1.5648500211237852,
+      "grad_norm": 0.5067934989929199,
+      "learning_rate": 4.8435149978876215e-05,
+      "loss": 0.0333,
+      "step": 3704
+    },
+    {
+      "epoch": 1.565694972539079,
+      "grad_norm": 0.6863754987716675,
+      "learning_rate": 4.843430502746092e-05,
+      "loss": 0.0331,
+      "step": 3706
+    },
+    {
+      "epoch": 1.5665399239543727,
+      "grad_norm": 0.6762244701385498,
+      "learning_rate": 4.843346007604563e-05,
+      "loss": 0.0281,
+      "step": 3708
+    },
+    {
+      "epoch": 1.5673848753696662,
+      "grad_norm": 0.8880215287208557,
+      "learning_rate": 4.8432615124630336e-05,
+      "loss": 0.029,
+      "step": 3710
+    },
+    {
+      "epoch": 1.5682298267849597,
+      "grad_norm": 1.9256477355957031,
+      "learning_rate": 4.843177017321504e-05,
+      "loss": 0.0542,
+      "step": 3712
+    },
+    {
+      "epoch": 1.5690747782002537,
+      "grad_norm": 1.291383147239685,
+      "learning_rate": 4.843092522179975e-05,
+      "loss": 0.0429,
+      "step": 3714
+    },
+    {
+      "epoch": 1.5699197296155472,
+      "grad_norm": 0.8845472931861877,
+      "learning_rate": 4.843008027038445e-05,
+      "loss": 0.0295,
+      "step": 3716
+    },
+    {
+      "epoch": 1.5707646810308407,
+      "grad_norm": 1.2028437852859497,
+      "learning_rate": 4.8429235318969165e-05,
+      "loss": 0.0485,
+      "step": 3718
+    },
+    {
+      "epoch": 1.5716096324461344,
+      "grad_norm": 1.249176025390625,
+      "learning_rate": 4.8428390367553865e-05,
+      "loss": 0.0261,
+      "step": 3720
+    },
+    {
+      "epoch": 1.5724545838614281,
+      "grad_norm": 1.1442838907241821,
+      "learning_rate": 4.842754541613857e-05,
+      "loss": 0.0597,
+      "step": 3722
+    },
+    {
+      "epoch": 1.5732995352767216,
+      "grad_norm": 1.2954820394515991,
+      "learning_rate": 4.842670046472328e-05,
+      "loss": 0.047,
+      "step": 3724
+    },
+    {
+      "epoch": 1.5741444866920151,
+      "grad_norm": 1.1261757612228394,
+      "learning_rate": 4.842585551330799e-05,
+      "loss": 0.0375,
+      "step": 3726
+    },
+    {
+      "epoch": 1.5749894381073088,
+      "grad_norm": 0.8863781094551086,
+      "learning_rate": 4.8425010561892694e-05,
+      "loss": 0.0332,
+      "step": 3728
+    },
+    {
+      "epoch": 1.5758343895226026,
+      "grad_norm": 0.660490095615387,
+      "learning_rate": 4.84241656104774e-05,
+      "loss": 0.0293,
+      "step": 3730
+    },
+    {
+      "epoch": 1.576679340937896,
+      "grad_norm": 0.3680121600627899,
+      "learning_rate": 4.84233206590621e-05,
+      "loss": 0.0454,
+      "step": 3732
+    },
+    {
+      "epoch": 1.5775242923531896,
+      "grad_norm": 1.0620887279510498,
+      "learning_rate": 4.8422475707646816e-05,
+      "loss": 0.0517,
+      "step": 3734
+    },
+    {
+      "epoch": 1.5783692437684833,
+      "grad_norm": 0.967285692691803,
+      "learning_rate": 4.8421630756231516e-05,
+      "loss": 0.0304,
+      "step": 3736
+    },
+    {
+      "epoch": 1.579214195183777,
+      "grad_norm": 1.02311372756958,
+      "learning_rate": 4.842078580481622e-05,
+      "loss": 0.0482,
+      "step": 3738
+    },
+    {
+      "epoch": 1.5800591465990705,
+      "grad_norm": 1.0686031579971313,
+      "learning_rate": 4.841994085340093e-05,
+      "loss": 0.0298,
+      "step": 3740
+    },
+    {
+      "epoch": 1.580904098014364,
+      "grad_norm": 1.1448440551757812,
+      "learning_rate": 4.841909590198564e-05,
+      "loss": 0.0596,
+      "step": 3742
+    },
+    {
+      "epoch": 1.5817490494296578,
+      "grad_norm": 0.899435818195343,
+      "learning_rate": 4.8418250950570345e-05,
+      "loss": 0.0393,
+      "step": 3744
+    },
+    {
+      "epoch": 1.5825940008449515,
+      "grad_norm": 0.6797332167625427,
+      "learning_rate": 4.841740599915505e-05,
+      "loss": 0.0286,
+      "step": 3746
+    },
+    {
+      "epoch": 1.583438952260245,
+      "grad_norm": 1.0120704174041748,
+      "learning_rate": 4.841656104773975e-05,
+      "loss": 0.0597,
+      "step": 3748
+    },
+    {
+      "epoch": 1.5842839036755385,
+      "grad_norm": 1.1370409727096558,
+      "learning_rate": 4.8415716096324467e-05,
+      "loss": 0.0551,
+      "step": 3750
+    },
+    {
+      "epoch": 1.5851288550908322,
+      "grad_norm": 1.4799180030822754,
+      "learning_rate": 4.841487114490917e-05,
+      "loss": 0.03,
+      "step": 3752
+    },
+    {
+      "epoch": 1.585973806506126,
+      "grad_norm": 1.0635912418365479,
+      "learning_rate": 4.841402619349388e-05,
+      "loss": 0.0513,
+      "step": 3754
+    },
+    {
+      "epoch": 1.5868187579214195,
+      "grad_norm": 0.6243870854377747,
+      "learning_rate": 4.841318124207858e-05,
+      "loss": 0.0308,
+      "step": 3756
+    },
+    {
+      "epoch": 1.5876637093367132,
+      "grad_norm": 1.172440528869629,
+      "learning_rate": 4.841233629066329e-05,
+      "loss": 0.0428,
+      "step": 3758
+    },
+    {
+      "epoch": 1.588508660752007,
+      "grad_norm": 1.209961175918579,
+      "learning_rate": 4.8411491339247996e-05,
+      "loss": 0.0458,
+      "step": 3760
+    },
+    {
+      "epoch": 1.5893536121673004,
+      "grad_norm": 1.1312841176986694,
+      "learning_rate": 4.84106463878327e-05,
+      "loss": 0.054,
+      "step": 3762
+    },
+    {
+      "epoch": 1.590198563582594,
+      "grad_norm": 0.6913443207740784,
+      "learning_rate": 4.840980143641741e-05,
+      "loss": 0.0403,
+      "step": 3764
+    },
+    {
+      "epoch": 1.5910435149978877,
+      "grad_norm": 1.1605889797210693,
+      "learning_rate": 4.840895648500212e-05,
+      "loss": 0.0479,
+      "step": 3766
+    },
+    {
+      "epoch": 1.5918884664131814,
+      "grad_norm": 0.7447473406791687,
+      "learning_rate": 4.840811153358682e-05,
+      "loss": 0.0268,
+      "step": 3768
+    },
+    {
+      "epoch": 1.5927334178284749,
+      "grad_norm": 0.9783474206924438,
+      "learning_rate": 4.840726658217153e-05,
+      "loss": 0.0465,
+      "step": 3770
+    },
+    {
+      "epoch": 1.5935783692437684,
+      "grad_norm": 1.1488279104232788,
+      "learning_rate": 4.840642163075623e-05,
+      "loss": 0.0517,
+      "step": 3772
+    },
+    {
+      "epoch": 1.5944233206590621,
+      "grad_norm": 0.8444589376449585,
+      "learning_rate": 4.840557667934094e-05,
+      "loss": 0.0339,
+      "step": 3774
+    },
+    {
+      "epoch": 1.5952682720743558,
+      "grad_norm": 1.1645581722259521,
+      "learning_rate": 4.8404731727925647e-05,
+      "loss": 0.041,
+      "step": 3776
+    },
+    {
+      "epoch": 1.5961132234896493,
+      "grad_norm": 1.6231311559677124,
+      "learning_rate": 4.8403886776510354e-05,
+      "loss": 0.054,
+      "step": 3778
+    },
+    {
+      "epoch": 1.5969581749049429,
+      "grad_norm": 0.9748203754425049,
+      "learning_rate": 4.840304182509506e-05,
+      "loss": 0.0428,
+      "step": 3780
+    },
+    {
+      "epoch": 1.5978031263202366,
+      "grad_norm": 1.2018306255340576,
+      "learning_rate": 4.840219687367977e-05,
+      "loss": 0.0527,
+      "step": 3782
+    },
+    {
+      "epoch": 1.5986480777355303,
+      "grad_norm": 1.094373345375061,
+      "learning_rate": 4.840135192226447e-05,
+      "loss": 0.0628,
+      "step": 3784
+    },
+    {
+      "epoch": 1.5994930291508238,
+      "grad_norm": 0.814469575881958,
+      "learning_rate": 4.840050697084918e-05,
+      "loss": 0.0224,
+      "step": 3786
+    },
+    {
+      "epoch": 1.6003379805661173,
+      "grad_norm": 0.8022437691688538,
+      "learning_rate": 4.839966201943388e-05,
+      "loss": 0.0482,
+      "step": 3788
+    },
+    {
+      "epoch": 1.601182931981411,
+      "grad_norm": 1.218186855316162,
+      "learning_rate": 4.839881706801859e-05,
+      "loss": 0.0675,
+      "step": 3790
+    },
+    {
+      "epoch": 1.6020278833967048,
+      "grad_norm": 1.626511573791504,
+      "learning_rate": 4.83979721166033e-05,
+      "loss": 0.0712,
+      "step": 3792
+    },
+    {
+      "epoch": 1.6028728348119983,
+      "grad_norm": 1.63881516456604,
+      "learning_rate": 4.8397127165188005e-05,
+      "loss": 0.0836,
+      "step": 3794
+    },
+    {
+      "epoch": 1.6037177862272918,
+      "grad_norm": 1.0243483781814575,
+      "learning_rate": 4.839628221377271e-05,
+      "loss": 0.0255,
+      "step": 3796
+    },
+    {
+      "epoch": 1.6045627376425855,
+      "grad_norm": 1.4431846141815186,
+      "learning_rate": 4.839543726235742e-05,
+      "loss": 0.0573,
+      "step": 3798
+    },
+    {
+      "epoch": 1.6054076890578792,
+      "grad_norm": 1.132359266281128,
+      "learning_rate": 4.839459231094212e-05,
+      "loss": 0.0403,
+      "step": 3800
+    },
+    {
+      "epoch": 1.6054076890578792,
+      "eval_accuracy": 0.7347818589411967,
+      "eval_cer": 0.075275457183941,
+      "eval_loss": 0.17756511270999908,
+      "eval_runtime": 841.6089,
+      "eval_samples_per_second": 13.781,
+      "eval_steps_per_second": 0.431,
+      "step": 3800
+    },
+    {
+      "epoch": 1.6062526404731727,
+      "grad_norm": 1.0357937812805176,
+      "learning_rate": 4.839374735952683e-05,
+      "loss": 0.0542,
+      "step": 3802
+    },
+    {
+      "epoch": 1.6070975918884665,
+      "grad_norm": 0.7196958661079407,
+      "learning_rate": 4.8392902408111534e-05,
+      "loss": 0.0331,
+      "step": 3804
+    },
+    {
+      "epoch": 1.6079425433037602,
+      "grad_norm": 0.6513099670410156,
+      "learning_rate": 4.839205745669624e-05,
+      "loss": 0.0424,
+      "step": 3806
+    },
+    {
+      "epoch": 1.6087874947190537,
+      "grad_norm": 1.5637171268463135,
+      "learning_rate": 4.839121250528095e-05,
+      "loss": 0.0593,
+      "step": 3808
+    },
+    {
+      "epoch": 1.6096324461343472,
+      "grad_norm": 0.7417770028114319,
+      "learning_rate": 4.8390367553865655e-05,
+      "loss": 0.0378,
+      "step": 3810
+    },
+    {
+      "epoch": 1.610477397549641,
+      "grad_norm": 0.97124183177948,
+      "learning_rate": 4.838952260245036e-05,
+      "loss": 0.0405,
+      "step": 3812
+    },
+    {
+      "epoch": 1.6113223489649346,
+      "grad_norm": 1.1514919996261597,
+      "learning_rate": 4.838867765103507e-05,
+      "loss": 0.0416,
+      "step": 3814
+    },
+    {
+      "epoch": 1.6121673003802282,
+      "grad_norm": 1.0334582328796387,
+      "learning_rate": 4.838783269961977e-05,
+      "loss": 0.0359,
+      "step": 3816
+    },
+    {
+      "epoch": 1.6130122517955217,
+      "grad_norm": 0.6415423154830933,
+      "learning_rate": 4.8386987748204484e-05,
+      "loss": 0.0269,
+      "step": 3818
+    },
+    {
+      "epoch": 1.6138572032108154,
+      "grad_norm": 0.9257076382637024,
+      "learning_rate": 4.8386142796789185e-05,
+      "loss": 0.0433,
+      "step": 3820
+    },
+    {
+      "epoch": 1.614702154626109,
+      "grad_norm": 1.0483789443969727,
+      "learning_rate": 4.838529784537389e-05,
+      "loss": 0.0402,
+      "step": 3822
+    },
+    {
+      "epoch": 1.6155471060414026,
+      "grad_norm": 1.224219799041748,
+      "learning_rate": 4.83844528939586e-05,
+      "loss": 0.0515,
+      "step": 3824
+    },
+    {
+      "epoch": 1.6163920574566961,
+      "grad_norm": 2.190443992614746,
+      "learning_rate": 4.8383607942543306e-05,
+      "loss": 0.0708,
+      "step": 3826
+    },
+    {
+      "epoch": 1.6172370088719898,
+      "grad_norm": 0.5145219564437866,
+      "learning_rate": 4.838276299112801e-05,
+      "loss": 0.0281,
+      "step": 3828
+    },
+    {
+      "epoch": 1.6180819602872836,
+      "grad_norm": 0.6071678400039673,
+      "learning_rate": 4.838191803971272e-05,
+      "loss": 0.0281,
+      "step": 3830
+    },
+    {
+      "epoch": 1.618926911702577,
+      "grad_norm": 1.6799170970916748,
+      "learning_rate": 4.838107308829742e-05,
+      "loss": 0.0694,
+      "step": 3832
+    },
+    {
+      "epoch": 1.6197718631178706,
+      "grad_norm": 1.0800015926361084,
+      "learning_rate": 4.8380228136882135e-05,
+      "loss": 0.0589,
+      "step": 3834
+    },
+    {
+      "epoch": 1.6206168145331643,
+      "grad_norm": 0.9297267198562622,
+      "learning_rate": 4.8379383185466835e-05,
+      "loss": 0.0276,
+      "step": 3836
+    },
+    {
+      "epoch": 1.621461765948458,
+      "grad_norm": 0.6874012351036072,
+      "learning_rate": 4.837853823405154e-05,
+      "loss": 0.0321,
+      "step": 3838
+    },
+    {
+      "epoch": 1.6223067173637515,
+      "grad_norm": 0.8664073944091797,
+      "learning_rate": 4.837769328263625e-05,
+      "loss": 0.0322,
+      "step": 3840
+    },
+    {
+      "epoch": 1.623151668779045,
+      "grad_norm": 1.3340883255004883,
+      "learning_rate": 4.837684833122096e-05,
+      "loss": 0.0442,
+      "step": 3842
+    },
+    {
+      "epoch": 1.623996620194339,
+      "grad_norm": 1.2647576332092285,
+      "learning_rate": 4.8376003379805664e-05,
+      "loss": 0.0795,
+      "step": 3844
+    },
+    {
+      "epoch": 1.6248415716096325,
+      "grad_norm": 1.0526044368743896,
+      "learning_rate": 4.837515842839037e-05,
+      "loss": 0.0379,
+      "step": 3846
+    },
+    {
+      "epoch": 1.625686523024926,
+      "grad_norm": 0.8544779419898987,
+      "learning_rate": 4.837431347697507e-05,
+      "loss": 0.0446,
+      "step": 3848
+    },
+    {
+      "epoch": 1.6265314744402197,
+      "grad_norm": 1.369625449180603,
+      "learning_rate": 4.8373468525559786e-05,
+      "loss": 0.0624,
+      "step": 3850
+    },
+    {
+      "epoch": 1.6273764258555135,
+      "grad_norm": 0.6502795219421387,
+      "learning_rate": 4.8372623574144486e-05,
+      "loss": 0.0625,
+      "step": 3852
+    },
+    {
+      "epoch": 1.628221377270807,
+      "grad_norm": 1.5579420328140259,
+      "learning_rate": 4.83717786227292e-05,
+      "loss": 0.0634,
+      "step": 3854
+    },
+    {
+      "epoch": 1.6290663286861005,
+      "grad_norm": 1.384822130203247,
+      "learning_rate": 4.83709336713139e-05,
+      "loss": 0.0428,
+      "step": 3856
+    },
+    {
+      "epoch": 1.6299112801013942,
+      "grad_norm": 0.95893394947052,
+      "learning_rate": 4.837008871989861e-05,
+      "loss": 0.0487,
+      "step": 3858
+    },
+    {
+      "epoch": 1.630756231516688,
+      "grad_norm": 1.7926594018936157,
+      "learning_rate": 4.8369243768483315e-05,
+      "loss": 0.056,
+      "step": 3860
+    },
+    {
+      "epoch": 1.6316011829319814,
+      "grad_norm": 2.259760856628418,
+      "learning_rate": 4.836839881706802e-05,
+      "loss": 0.0356,
+      "step": 3862
+    },
+    {
+      "epoch": 1.632446134347275,
+      "grad_norm": 0.7830055356025696,
+      "learning_rate": 4.836755386565273e-05,
+      "loss": 0.0371,
+      "step": 3864
+    },
+    {
+      "epoch": 1.6332910857625687,
+      "grad_norm": 0.8463422060012817,
+      "learning_rate": 4.8366708914237437e-05,
+      "loss": 0.0532,
+      "step": 3866
+    },
+    {
+      "epoch": 1.6341360371778624,
+      "grad_norm": 1.0259829759597778,
+      "learning_rate": 4.836586396282214e-05,
+      "loss": 0.0461,
+      "step": 3868
+    },
+    {
+      "epoch": 1.6349809885931559,
+      "grad_norm": 0.7621318697929382,
+      "learning_rate": 4.836501901140685e-05,
+      "loss": 0.0515,
+      "step": 3870
+    },
+    {
+      "epoch": 1.6358259400084494,
+      "grad_norm": 1.094347596168518,
+      "learning_rate": 4.836417405999155e-05,
+      "loss": 0.0504,
+      "step": 3872
+    },
+    {
+      "epoch": 1.6366708914237431,
+      "grad_norm": 1.4083515405654907,
+      "learning_rate": 4.836332910857626e-05,
+      "loss": 0.0464,
+      "step": 3874
+    },
+    {
+      "epoch": 1.6375158428390368,
+      "grad_norm": 1.221833348274231,
+      "learning_rate": 4.8362484157160966e-05,
+      "loss": 0.0393,
+      "step": 3876
+    },
+    {
+      "epoch": 1.6383607942543303,
+      "grad_norm": 1.5514699220657349,
+      "learning_rate": 4.836163920574567e-05,
+      "loss": 0.0591,
+      "step": 3878
+    },
+    {
+      "epoch": 1.6392057456696238,
+      "grad_norm": 1.4877634048461914,
+      "learning_rate": 4.836079425433038e-05,
+      "loss": 0.0546,
+      "step": 3880
+    },
+    {
+      "epoch": 1.6400506970849176,
+      "grad_norm": 1.098713994026184,
+      "learning_rate": 4.835994930291509e-05,
+      "loss": 0.0385,
+      "step": 3882
+    },
+    {
+      "epoch": 1.6408956485002113,
+      "grad_norm": 1.0514299869537354,
+      "learning_rate": 4.835910435149979e-05,
+      "loss": 0.0556,
+      "step": 3884
+    },
+    {
+      "epoch": 1.6417405999155048,
+      "grad_norm": 0.7681018710136414,
+      "learning_rate": 4.83582594000845e-05,
+      "loss": 0.0355,
+      "step": 3886
+    },
+    {
+      "epoch": 1.6425855513307985,
+      "grad_norm": 1.322800636291504,
+      "learning_rate": 4.83574144486692e-05,
+      "loss": 0.0453,
+      "step": 3888
+    },
+    {
+      "epoch": 1.6434305027460923,
+      "grad_norm": 1.2197022438049316,
+      "learning_rate": 4.835656949725391e-05,
+      "loss": 0.0542,
+      "step": 3890
+    },
+    {
+      "epoch": 1.6442754541613858,
+      "grad_norm": 1.9387176036834717,
+      "learning_rate": 4.8355724545838617e-05,
+      "loss": 0.06,
+      "step": 3892
+    },
+    {
+      "epoch": 1.6451204055766793,
+      "grad_norm": 0.9585774540901184,
+      "learning_rate": 4.8354879594423324e-05,
+      "loss": 0.0847,
+      "step": 3894
+    },
+    {
+      "epoch": 1.645965356991973,
+      "grad_norm": 0.7485412359237671,
+      "learning_rate": 4.835403464300803e-05,
+      "loss": 0.0278,
+      "step": 3896
+    },
+    {
+      "epoch": 1.6468103084072667,
+      "grad_norm": 0.7363729476928711,
+      "learning_rate": 4.835318969159274e-05,
+      "loss": 0.0548,
+      "step": 3898
+    },
+    {
+      "epoch": 1.6476552598225602,
+      "grad_norm": 2.1429085731506348,
+      "learning_rate": 4.835234474017744e-05,
+      "loss": 0.044,
+      "step": 3900
+    },
+    {
+      "epoch": 1.6476552598225602,
+      "eval_accuracy": 0.7290050008622176,
+      "eval_cer": 0.07787785491853172,
+      "eval_loss": 0.17509505152702332,
+      "eval_runtime": 862.0291,
+      "eval_samples_per_second": 13.454,
+      "eval_steps_per_second": 0.421,
+      "step": 3900
+    },
+    {
+      "epoch": 1.6485002112378537,
+      "grad_norm": 1.3191771507263184,
+      "learning_rate": 4.835149978876215e-05,
+      "loss": 0.05,
+      "step": 3902
+    },
+    {
+      "epoch": 1.6493451626531475,
+      "grad_norm": 1.1920831203460693,
+      "learning_rate": 4.835065483734685e-05,
+      "loss": 0.045,
+      "step": 3904
+    },
+    {
+      "epoch": 1.6501901140684412,
+      "grad_norm": 0.7942538261413574,
+      "learning_rate": 4.834980988593156e-05,
+      "loss": 0.0335,
+      "step": 3906
+    },
+    {
+      "epoch": 1.6510350654837347,
+      "grad_norm": 1.081873893737793,
+      "learning_rate": 4.834896493451627e-05,
+      "loss": 0.0959,
+      "step": 3908
+    },
+    {
+      "epoch": 1.6518800168990282,
+      "grad_norm": 0.9521547555923462,
+      "learning_rate": 4.8348119983100975e-05,
+      "loss": 0.0277,
+      "step": 3910
+    },
+    {
+      "epoch": 1.652724968314322,
+      "grad_norm": 1.2026457786560059,
+      "learning_rate": 4.834727503168568e-05,
+      "loss": 0.0523,
+      "step": 3912
+    },
+    {
+      "epoch": 1.6535699197296156,
+      "grad_norm": 0.7098683714866638,
+      "learning_rate": 4.834643008027039e-05,
+      "loss": 0.0256,
+      "step": 3914
+    },
+    {
+      "epoch": 1.6544148711449091,
+      "grad_norm": 1.0528465509414673,
+      "learning_rate": 4.834558512885509e-05,
+      "loss": 0.064,
+      "step": 3916
+    },
+    {
+      "epoch": 1.6552598225602027,
+      "grad_norm": 1.4617663621902466,
+      "learning_rate": 4.8344740177439803e-05,
+      "loss": 0.0489,
+      "step": 3918
+    },
+    {
+      "epoch": 1.6561047739754964,
+      "grad_norm": 1.6092969179153442,
+      "learning_rate": 4.8343895226024504e-05,
+      "loss": 0.0505,
+      "step": 3920
+    },
+    {
+      "epoch": 1.65694972539079,
+      "grad_norm": 1.4073854684829712,
+      "learning_rate": 4.834305027460921e-05,
+      "loss": 0.0411,
+      "step": 3922
+    },
+    {
+      "epoch": 1.6577946768060836,
+      "grad_norm": 1.2704623937606812,
+      "learning_rate": 4.834220532319392e-05,
+      "loss": 0.0713,
+      "step": 3924
+    },
+    {
+      "epoch": 1.6586396282213771,
+      "grad_norm": 1.0156190395355225,
+      "learning_rate": 4.8341360371778625e-05,
+      "loss": 0.0335,
+      "step": 3926
+    },
+    {
+      "epoch": 1.6594845796366708,
+      "grad_norm": 1.200533151626587,
+      "learning_rate": 4.834051542036333e-05,
+      "loss": 0.0465,
+      "step": 3928
+    },
+    {
+      "epoch": 1.6603295310519646,
+      "grad_norm": 0.7413454651832581,
+      "learning_rate": 4.833967046894804e-05,
+      "loss": 0.044,
+      "step": 3930
+    },
+    {
+      "epoch": 1.661174482467258,
+      "grad_norm": 1.4523422718048096,
+      "learning_rate": 4.833882551753274e-05,
+      "loss": 0.0612,
+      "step": 3932
+    },
+    {
+      "epoch": 1.6620194338825518,
+      "grad_norm": 1.259691596031189,
+      "learning_rate": 4.8337980566117454e-05,
+      "loss": 0.061,
+      "step": 3934
+    },
+    {
+      "epoch": 1.6628643852978455,
+      "grad_norm": 1.7233320474624634,
+      "learning_rate": 4.8337135614702155e-05,
+      "loss": 0.0653,
+      "step": 3936
+    },
+    {
+      "epoch": 1.663709336713139,
+      "grad_norm": 1.2285038232803345,
+      "learning_rate": 4.833629066328686e-05,
+      "loss": 0.0703,
+      "step": 3938
+    },
+    {
+      "epoch": 1.6645542881284325,
+      "grad_norm": 1.288918375968933,
+      "learning_rate": 4.833544571187157e-05,
+      "loss": 0.0618,
+      "step": 3940
+    },
+    {
+      "epoch": 1.6653992395437263,
+      "grad_norm": 0.7824044823646545,
+      "learning_rate": 4.8334600760456276e-05,
+      "loss": 0.048,
+      "step": 3942
+    },
+    {
+      "epoch": 1.66624419095902,
+      "grad_norm": 1.211120843887329,
+      "learning_rate": 4.8333755809040983e-05,
+      "loss": 0.0669,
+      "step": 3944
+    },
+    {
+      "epoch": 1.6670891423743135,
+      "grad_norm": 2.0919220447540283,
+      "learning_rate": 4.833291085762569e-05,
+      "loss": 0.0803,
+      "step": 3946
+    },
+    {
+      "epoch": 1.667934093789607,
+      "grad_norm": 0.8903474807739258,
+      "learning_rate": 4.833206590621039e-05,
+      "loss": 0.063,
+      "step": 3948
+    },
+    {
+      "epoch": 1.6687790452049007,
+      "grad_norm": 1.258203148841858,
+      "learning_rate": 4.8331220954795105e-05,
+      "loss": 0.0331,
+      "step": 3950
+    },
+    {
+      "epoch": 1.6696239966201945,
+      "grad_norm": 1.2711974382400513,
+      "learning_rate": 4.8330376003379805e-05,
+      "loss": 0.0414,
+      "step": 3952
+    },
+    {
+      "epoch": 1.670468948035488,
+      "grad_norm": 0.8380125164985657,
+      "learning_rate": 4.832953105196452e-05,
+      "loss": 0.0445,
+      "step": 3954
+    },
+    {
+      "epoch": 1.6713138994507815,
+      "grad_norm": 1.392994999885559,
+      "learning_rate": 4.832868610054922e-05,
+      "loss": 0.0605,
+      "step": 3956
+    },
+    {
+      "epoch": 1.6721588508660752,
+      "grad_norm": 1.1180285215377808,
+      "learning_rate": 4.832784114913393e-05,
+      "loss": 0.0493,
+      "step": 3958
+    },
+    {
+      "epoch": 1.673003802281369,
+      "grad_norm": 1.3766900300979614,
+      "learning_rate": 4.8326996197718634e-05,
+      "loss": 0.057,
+      "step": 3960
+    },
+    {
+      "epoch": 1.6738487536966624,
+      "grad_norm": 1.6987857818603516,
+      "learning_rate": 4.832615124630334e-05,
+      "loss": 0.0476,
+      "step": 3962
+    },
+    {
+      "epoch": 1.674693705111956,
+      "grad_norm": 1.4040523767471313,
+      "learning_rate": 4.832530629488805e-05,
+      "loss": 0.0397,
+      "step": 3964
+    },
+    {
+      "epoch": 1.6755386565272496,
+      "grad_norm": 0.45623424649238586,
+      "learning_rate": 4.8324461343472756e-05,
+      "loss": 0.0455,
+      "step": 3966
+    },
+    {
+      "epoch": 1.6763836079425434,
+      "grad_norm": 0.9928030967712402,
+      "learning_rate": 4.8323616392057456e-05,
+      "loss": 0.0448,
+      "step": 3968
+    },
+    {
+      "epoch": 1.6772285593578369,
+      "grad_norm": 0.9238046407699585,
+      "learning_rate": 4.832277144064217e-05,
+      "loss": 0.0295,
+      "step": 3970
+    },
+    {
+      "epoch": 1.6780735107731304,
+      "grad_norm": 0.2931238114833832,
+      "learning_rate": 4.832192648922687e-05,
+      "loss": 0.0247,
+      "step": 3972
+    },
+    {
+      "epoch": 1.6789184621884243,
+      "grad_norm": 0.8757814168930054,
+      "learning_rate": 4.832108153781158e-05,
+      "loss": 0.0358,
+      "step": 3974
+    },
+    {
+      "epoch": 1.6797634136037178,
+      "grad_norm": 0.8881898522377014,
+      "learning_rate": 4.8320236586396285e-05,
+      "loss": 0.0331,
+      "step": 3976
+    },
+    {
+      "epoch": 1.6806083650190113,
+      "grad_norm": 0.7834998965263367,
+      "learning_rate": 4.831939163498099e-05,
+      "loss": 0.0407,
+      "step": 3978
+    },
+    {
+      "epoch": 1.681453316434305,
+      "grad_norm": 1.1790554523468018,
+      "learning_rate": 4.83185466835657e-05,
+      "loss": 0.0513,
+      "step": 3980
+    },
+    {
+      "epoch": 1.6822982678495988,
+      "grad_norm": 0.5270421504974365,
+      "learning_rate": 4.8317701732150407e-05,
+      "loss": 0.0395,
+      "step": 3982
+    },
+    {
+      "epoch": 1.6831432192648923,
+      "grad_norm": 1.120709776878357,
+      "learning_rate": 4.831685678073511e-05,
+      "loss": 0.0394,
+      "step": 3984
+    },
+    {
+      "epoch": 1.6839881706801858,
+      "grad_norm": 0.8393281102180481,
+      "learning_rate": 4.831601182931982e-05,
+      "loss": 0.0342,
+      "step": 3986
+    },
+    {
+      "epoch": 1.6848331220954795,
+      "grad_norm": 0.6471826434135437,
+      "learning_rate": 4.831516687790452e-05,
+      "loss": 0.0257,
+      "step": 3988
+    },
+    {
+      "epoch": 1.6856780735107733,
+      "grad_norm": 1.7182902097702026,
+      "learning_rate": 4.831432192648923e-05,
+      "loss": 0.0506,
+      "step": 3990
+    },
+    {
+      "epoch": 1.6865230249260668,
+      "grad_norm": 0.7699361443519592,
+      "learning_rate": 4.8313476975073936e-05,
+      "loss": 0.0408,
+      "step": 3992
+    },
+    {
+      "epoch": 1.6873679763413603,
+      "grad_norm": 0.8868207335472107,
+      "learning_rate": 4.831263202365864e-05,
+      "loss": 0.0359,
+      "step": 3994
+    },
+    {
+      "epoch": 1.688212927756654,
+      "grad_norm": 1.3517881631851196,
+      "learning_rate": 4.831178707224335e-05,
+      "loss": 0.0439,
+      "step": 3996
+    },
+    {
+      "epoch": 1.6890578791719477,
+      "grad_norm": 1.8134877681732178,
+      "learning_rate": 4.831094212082806e-05,
+      "loss": 0.0508,
+      "step": 3998
+    },
+    {
+      "epoch": 1.6899028305872412,
+      "grad_norm": 0.8728620409965515,
+      "learning_rate": 4.831009716941276e-05,
+      "loss": 0.0638,
+      "step": 4000
+    },
+    {
+      "epoch": 1.6899028305872412,
+      "eval_accuracy": 0.7381445076737368,
+      "eval_cer": 0.07271716788552979,
+      "eval_loss": 0.17253856360912323,
+      "eval_runtime": 854.712,
+      "eval_samples_per_second": 13.569,
+      "eval_steps_per_second": 0.425,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 118350,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 50,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.6614257405400023e+19,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}