End of training

all_results.json

@@ -0,0 +1,14 @@
+{
+    "epoch": 199.9,
+    "eval_loss": 2.3263261318206787,
+    "eval_runtime": 7.517,
+    "eval_samples": 63,
+    "eval_samples_per_second": 8.381,
+    "eval_steps_per_second": 1.064,
+    "eval_wer": 0.7803052978150254,
+    "train_loss": 2.0372400057315825,
+    "train_runtime": 18220.1543,
+    "train_samples": 563,
+    "train_samples_per_second": 6.18,
+    "train_steps_per_second": 0.088
+}

eval_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 199.9,
+    "eval_loss": 2.3263261318206787,
+    "eval_runtime": 7.517,
+    "eval_samples": 63,
+    "eval_samples_per_second": 8.381,
+    "eval_steps_per_second": 1.064,
+    "eval_wer": 0.7803052978150254
+}

runs/Nov07_10-39-49_70ac8591dae5/events.out.tfevents.1667836705.70ac8591dae5.3645.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53cf66fc2423d43ea314b2128dbebc313271a1cee195d72950a76ff1fa58f192
+size 358

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 199.9,
+    "train_loss": 2.0372400057315825,
+    "train_runtime": 18220.1543,
+    "train_samples": 563,
+    "train_samples_per_second": 6.18,
+    "train_steps_per_second": 0.088
+}

trainer_state.json

@@ -0,0 +1,2785 @@
+{
+  "best_metric": 2.3263261318206787,
+  "best_model_checkpoint": "ai-light-dance_drums_ft_pretrain_wav2vec2-base/checkpoint-1168",
+  "epoch": 199.90140845070422,
+  "global_step": 1600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.9,
+      "eval_loss": 129.7303924560547,
+      "eval_runtime": 7.5345,
+      "eval_samples_per_second": 8.362,
+      "eval_steps_per_second": 1.062,
+      "eval_wer": 1.082609997006884,
+      "step": 8
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00014,
+      "loss": 58.4765,
+      "step": 10
+    },
+    {
+      "epoch": 1.9,
+      "eval_loss": 92.10203552246094,
+      "eval_runtime": 7.564,
+      "eval_samples_per_second": 8.329,
+      "eval_steps_per_second": 1.058,
+      "eval_wer": 1.3475007482789585,
+      "step": 16
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00032,
+      "loss": 37.6409,
+      "step": 20
+    },
+    {
+      "epoch": 2.9,
+      "eval_loss": 23.880123138427734,
+      "eval_runtime": 7.5165,
+      "eval_samples_per_second": 8.382,
+      "eval_steps_per_second": 1.064,
+      "eval_wer": 1.0,
+      "step": 24
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003984810126582279,
+      "loss": 6.3535,
+      "step": 30
+    },
+    {
+      "epoch": 3.9,
+      "eval_loss": 13.232953071594238,
+      "eval_runtime": 7.5025,
+      "eval_samples_per_second": 8.397,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 1.0,
+      "step": 32
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0003959493670886076,
+      "loss": 5.0319,
+      "step": 40
+    },
+    {
+      "epoch": 4.9,
+      "eval_loss": 8.361285209655762,
+      "eval_runtime": 7.5177,
+      "eval_samples_per_second": 8.38,
+      "eval_steps_per_second": 1.064,
+      "eval_wer": 1.0,
+      "step": 40
+    },
+    {
+      "epoch": 5.9,
+      "eval_loss": 5.518415927886963,
+      "eval_runtime": 7.547,
+      "eval_samples_per_second": 8.348,
+      "eval_steps_per_second": 1.06,
+      "eval_wer": 1.0,
+      "step": 48
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00039341772151898737,
+      "loss": 4.7733,
+      "step": 50
+    },
+    {
+      "epoch": 6.9,
+      "eval_loss": 4.171965599060059,
+      "eval_runtime": 7.494,
+      "eval_samples_per_second": 8.407,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 1.0,
+      "step": 56
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.0003908860759493671,
+      "loss": 3.8911,
+      "step": 60
+    },
+    {
+      "epoch": 7.9,
+      "eval_loss": 3.9397051334381104,
+      "eval_runtime": 7.5681,
+      "eval_samples_per_second": 8.324,
+      "eval_steps_per_second": 1.057,
+      "eval_wer": 0.9994013768332835,
+      "step": 64
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 0.00038835443037974686,
+      "loss": 3.6,
+      "step": 70
+    },
+    {
+      "epoch": 8.9,
+      "eval_loss": 4.074159622192383,
+      "eval_runtime": 7.4884,
+      "eval_samples_per_second": 8.413,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.9994013768332835,
+      "step": 72
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 0.00038582278481012663,
+      "loss": 3.1586,
+      "step": 80
+    },
+    {
+      "epoch": 9.9,
+      "eval_loss": 3.8395957946777344,
+      "eval_runtime": 7.5346,
+      "eval_samples_per_second": 8.361,
+      "eval_steps_per_second": 1.062,
+      "eval_wer": 0.9997006884166417,
+      "step": 80
+    },
+    {
+      "epoch": 10.9,
+      "eval_loss": 3.8820667266845703,
+      "eval_runtime": 7.5684,
+      "eval_samples_per_second": 8.324,
+      "eval_steps_per_second": 1.057,
+      "eval_wer": 0.9994013768332835,
+      "step": 88
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 0.00038329113924050635,
+      "loss": 3.3812,
+      "step": 90
+    },
+    {
+      "epoch": 11.9,
+      "eval_loss": 3.9755468368530273,
+      "eval_runtime": 7.5139,
+      "eval_samples_per_second": 8.384,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.9979048189164921,
+      "step": 96
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 0.00038075949367088606,
+      "loss": 3.0088,
+      "step": 100
+    },
+    {
+      "epoch": 12.9,
+      "eval_loss": 4.529850006103516,
+      "eval_runtime": 7.5521,
+      "eval_samples_per_second": 8.342,
+      "eval_steps_per_second": 1.059,
+      "eval_wer": 0.9970068841664173,
+      "step": 104
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 0.00037822784810126583,
+      "loss": 2.7633,
+      "step": 110
+    },
+    {
+      "epoch": 13.9,
+      "eval_loss": 4.139248371124268,
+      "eval_runtime": 7.5513,
+      "eval_samples_per_second": 8.343,
+      "eval_steps_per_second": 1.059,
+      "eval_wer": 0.9964082609997007,
+      "step": 112
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 0.0003756962025316456,
+      "loss": 2.6359,
+      "step": 120
+    },
+    {
+      "epoch": 14.9,
+      "eval_loss": 3.8528494834899902,
+      "eval_runtime": 7.555,
+      "eval_samples_per_second": 8.339,
+      "eval_steps_per_second": 1.059,
+      "eval_wer": 0.9985034420832086,
+      "step": 120
+    },
+    {
+      "epoch": 15.9,
+      "eval_loss": 3.4155988693237305,
+      "eval_runtime": 7.5157,
+      "eval_samples_per_second": 8.382,
+      "eval_steps_per_second": 1.064,
+      "eval_wer": 0.9973061957497755,
+      "step": 128
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 0.0003731645569620253,
+      "loss": 3.1964,
+      "step": 130
+    },
+    {
+      "epoch": 16.9,
+      "eval_loss": 3.374109983444214,
+      "eval_runtime": 7.5568,
+      "eval_samples_per_second": 8.337,
+      "eval_steps_per_second": 1.059,
+      "eval_wer": 0.996707572583059,
+      "step": 136
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 0.0003706329113924051,
+      "loss": 2.4962,
+      "step": 140
+    },
+    {
+      "epoch": 17.9,
+      "eval_loss": 3.6131513118743896,
+      "eval_runtime": 7.535,
+      "eval_samples_per_second": 8.361,
+      "eval_steps_per_second": 1.062,
+      "eval_wer": 0.9922178988326849,
+      "step": 144
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 0.0003681012658227848,
+      "loss": 2.4907,
+      "step": 150
+    },
+    {
+      "epoch": 18.9,
+      "eval_loss": 3.5363380908966064,
+      "eval_runtime": 7.513,
+      "eval_samples_per_second": 8.385,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.9958096378329842,
+      "step": 152
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 0.0003655696202531646,
+      "loss": 2.413,
+      "step": 160
+    },
+    {
+      "epoch": 19.9,
+      "eval_loss": 3.6351335048675537,
+      "eval_runtime": 7.5038,
+      "eval_samples_per_second": 8.396,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.9928165219994014,
+      "step": 160
+    },
+    {
+      "epoch": 20.9,
+      "eval_loss": 3.6112582683563232,
+      "eval_runtime": 7.5044,
+      "eval_samples_per_second": 8.395,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.9925172104160431,
+      "step": 168
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 0.0003630379746835443,
+      "loss": 2.5555,
+      "step": 170
+    },
+    {
+      "epoch": 21.9,
+      "eval_loss": 3.7119691371917725,
+      "eval_runtime": 7.5276,
+      "eval_samples_per_second": 8.369,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.9904220293325352,
+      "step": 176
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 0.00036050632911392407,
+      "loss": 2.2909,
+      "step": 180
+    },
+    {
+      "epoch": 22.9,
+      "eval_loss": 3.43344783782959,
+      "eval_runtime": 7.4907,
+      "eval_samples_per_second": 8.41,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.99131996408261,
+      "step": 184
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 0.00035797468354430384,
+      "loss": 2.298,
+      "step": 190
+    },
+    {
+      "epoch": 23.9,
+      "eval_loss": 3.5457568168640137,
+      "eval_runtime": 7.4274,
+      "eval_samples_per_second": 8.482,
+      "eval_steps_per_second": 1.077,
+      "eval_wer": 0.99131996408261,
+      "step": 192
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 0.00035544303797468356,
+      "loss": 2.2366,
+      "step": 200
+    },
+    {
+      "epoch": 24.9,
+      "eval_loss": 3.7104668617248535,
+      "eval_runtime": 7.5469,
+      "eval_samples_per_second": 8.348,
+      "eval_steps_per_second": 1.06,
+      "eval_wer": 0.9898234061658187,
+      "step": 200
+    },
+    {
+      "epoch": 25.9,
+      "eval_loss": 4.229844093322754,
+      "eval_runtime": 7.5144,
+      "eval_samples_per_second": 8.384,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.9865309787488776,
+      "step": 208
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 0.00035291139240506333,
+      "loss": 2.3428,
+      "step": 210
+    },
+    {
+      "epoch": 26.9,
+      "eval_loss": 3.3207037448883057,
+      "eval_runtime": 7.5005,
+      "eval_samples_per_second": 8.399,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.9889254714157438,
+      "step": 216
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 0.00035037974683544304,
+      "loss": 2.2519,
+      "step": 220
+    },
+    {
+      "epoch": 27.9,
+      "eval_loss": 3.154818058013916,
+      "eval_runtime": 7.5146,
+      "eval_samples_per_second": 8.384,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.9895240945824604,
+      "step": 224
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 0.0003478481012658228,
+      "loss": 2.0429,
+      "step": 230
+    },
+    {
+      "epoch": 28.9,
+      "eval_loss": 3.7694132328033447,
+      "eval_runtime": 7.5309,
+      "eval_samples_per_second": 8.366,
+      "eval_steps_per_second": 1.062,
+      "eval_wer": 0.9877282250823107,
+      "step": 232
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 0.0003453164556962026,
+      "loss": 2.1886,
+      "step": 240
+    },
+    {
+      "epoch": 29.9,
+      "eval_loss": 3.5341339111328125,
+      "eval_runtime": 7.4895,
+      "eval_samples_per_second": 8.412,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.9877282250823107,
+      "step": 240
+    },
+    {
+      "epoch": 30.9,
+      "eval_loss": 3.138820171356201,
+      "eval_runtime": 7.4982,
+      "eval_samples_per_second": 8.402,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.9850344208320863,
+      "step": 248
+    },
+    {
+      "epoch": 31.23,
+      "learning_rate": 0.0003427848101265823,
+      "loss": 2.2182,
+      "step": 250
+    },
+    {
+      "epoch": 31.9,
+      "eval_loss": 2.938950538635254,
+      "eval_runtime": 7.535,
+      "eval_samples_per_second": 8.361,
+      "eval_steps_per_second": 1.062,
+      "eval_wer": 0.9817419934151451,
+      "step": 256
+    },
+    {
+      "epoch": 32.45,
+      "learning_rate": 0.000340253164556962,
+      "loss": 1.9479,
+      "step": 260
+    },
+    {
+      "epoch": 32.9,
+      "eval_loss": 3.0059807300567627,
+      "eval_runtime": 7.5148,
+      "eval_samples_per_second": 8.383,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.98263992816522,
+      "step": 264
+    },
+    {
+      "epoch": 33.68,
+      "learning_rate": 0.0003377215189873418,
+      "loss": 1.9703,
+      "step": 270
+    },
+    {
+      "epoch": 33.9,
+      "eval_loss": 3.257068395614624,
+      "eval_runtime": 7.5032,
+      "eval_samples_per_second": 8.396,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.9787488775815624,
+      "step": 272
+    },
+    {
+      "epoch": 34.9,
+      "learning_rate": 0.00033518987341772156,
+      "loss": 1.9385,
+      "step": 280
+    },
+    {
+      "epoch": 34.9,
+      "eval_loss": 3.1085846424102783,
+      "eval_runtime": 7.5414,
+      "eval_samples_per_second": 8.354,
+      "eval_steps_per_second": 1.061,
+      "eval_wer": 0.9775516312481293,
+      "step": 280
+    },
+    {
+      "epoch": 35.9,
+      "eval_loss": 2.8231420516967773,
+      "eval_runtime": 7.5276,
+      "eval_samples_per_second": 8.369,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.9655791679137983,
+      "step": 288
+    },
+    {
+      "epoch": 36.23,
+      "learning_rate": 0.0003326582278481013,
+      "loss": 2.0297,
+      "step": 290
+    },
+    {
+      "epoch": 36.9,
+      "eval_loss": 2.8960585594177246,
+      "eval_runtime": 7.4607,
+      "eval_samples_per_second": 8.444,
+      "eval_steps_per_second": 1.072,
+      "eval_wer": 0.9667764142472314,
+      "step": 296
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 0.000330126582278481,
+      "loss": 1.8406,
+      "step": 300
+    },
+    {
+      "epoch": 37.9,
+      "eval_loss": 2.882859706878662,
+      "eval_runtime": 7.4625,
+      "eval_samples_per_second": 8.442,
+      "eval_steps_per_second": 1.072,
+      "eval_wer": 0.9670757258305896,
+      "step": 304
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 0.00032759493670886077,
+      "loss": 1.8707,
+      "step": 310
+    },
+    {
+      "epoch": 38.9,
+      "eval_loss": 3.1193575859069824,
+      "eval_runtime": 7.5322,
+      "eval_samples_per_second": 8.364,
+      "eval_steps_per_second": 1.062,
+      "eval_wer": 0.9583956899131997,
+      "step": 312
+    },
+    {
+      "epoch": 39.9,
+      "learning_rate": 0.00032506329113924054,
+      "loss": 1.7798,
+      "step": 320
+    },
+    {
+      "epoch": 39.9,
+      "eval_loss": 3.0685505867004395,
+      "eval_runtime": 7.5634,
+      "eval_samples_per_second": 8.33,
+      "eval_steps_per_second": 1.058,
+      "eval_wer": 0.959892247829991,
+      "step": 320
+    },
+    {
+      "epoch": 40.9,
+      "eval_loss": 2.727964162826538,
+      "eval_runtime": 7.4937,
+      "eval_samples_per_second": 8.407,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.9527087698293923,
+      "step": 328
+    },
+    {
+      "epoch": 41.23,
+      "learning_rate": 0.00032253164556962026,
+      "loss": 1.9163,
+      "step": 330
+    },
+    {
+      "epoch": 41.9,
+      "eval_loss": 2.6320998668670654,
+      "eval_runtime": 7.5136,
+      "eval_samples_per_second": 8.385,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.9476204729123017,
+      "step": 336
+    },
+    {
+      "epoch": 42.45,
+      "learning_rate": 0.00032,
+      "loss": 1.7248,
+      "step": 340
+    },
+    {
+      "epoch": 42.9,
+      "eval_loss": 2.6813271045684814,
+      "eval_runtime": 7.5252,
+      "eval_samples_per_second": 8.372,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.9413349296617779,
+      "step": 344
+    },
+    {
+      "epoch": 43.68,
+      "learning_rate": 0.00031746835443037974,
+      "loss": 1.7602,
+      "step": 350
+    },
+    {
+      "epoch": 43.9,
+      "eval_loss": 2.725241184234619,
+      "eval_runtime": 7.544,
+      "eval_samples_per_second": 8.351,
+      "eval_steps_per_second": 1.06,
+      "eval_wer": 0.9419335528284944,
+      "step": 352
+    },
+    {
+      "epoch": 44.9,
+      "learning_rate": 0.0003149367088607595,
+      "loss": 1.7357,
+      "step": 360
+    },
+    {
+      "epoch": 44.9,
+      "eval_loss": 3.03348708152771,
+      "eval_runtime": 7.4787,
+      "eval_samples_per_second": 8.424,
+      "eval_steps_per_second": 1.07,
+      "eval_wer": 0.9398383717449865,
+      "step": 360
+    },
+    {
+      "epoch": 45.9,
+      "eval_loss": 2.8731963634490967,
+      "eval_runtime": 7.5025,
+      "eval_samples_per_second": 8.397,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.9344507632445376,
+      "step": 368
+    },
+    {
+      "epoch": 46.23,
+      "learning_rate": 0.0003124050632911393,
+      "loss": 1.7997,
+      "step": 370
+    },
+    {
+      "epoch": 46.9,
+      "eval_loss": 2.7708871364593506,
+      "eval_runtime": 7.5081,
+      "eval_samples_per_second": 8.391,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.9353486979946124,
+      "step": 376
+    },
+    {
+      "epoch": 47.45,
+      "learning_rate": 0.000309873417721519,
+      "loss": 1.6268,
+      "step": 380
+    },
+    {
+      "epoch": 47.9,
+      "eval_loss": 2.768134832382202,
+      "eval_runtime": 7.5125,
+      "eval_samples_per_second": 8.386,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.9278659084106555,
+      "step": 384
+    },
+    {
+      "epoch": 48.68,
+      "learning_rate": 0.0003073417721518987,
+      "loss": 1.6527,
+      "step": 390
+    },
+    {
+      "epoch": 48.9,
+      "eval_loss": 2.825923204421997,
+      "eval_runtime": 7.512,
+      "eval_samples_per_second": 8.387,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.9302604010775217,
+      "step": 392
+    },
+    {
+      "epoch": 49.9,
+      "learning_rate": 0.0003048101265822785,
+      "loss": 1.5715,
+      "step": 400
+    },
+    {
+      "epoch": 49.9,
+      "eval_loss": 2.8840715885162354,
+      "eval_runtime": 7.525,
+      "eval_samples_per_second": 8.372,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.9341514516611793,
+      "step": 400
+    },
+    {
+      "epoch": 50.9,
+      "eval_loss": 2.794407367706299,
+      "eval_runtime": 7.4742,
+      "eval_samples_per_second": 8.429,
+      "eval_steps_per_second": 1.07,
+      "eval_wer": 0.9221789883268483,
+      "step": 408
+    },
+    {
+      "epoch": 51.23,
+      "learning_rate": 0.00030227848101265826,
+      "loss": 1.6903,
+      "step": 410
+    },
+    {
+      "epoch": 51.9,
+      "eval_loss": 3.1597204208374023,
+      "eval_runtime": 7.5,
+      "eval_samples_per_second": 8.4,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.9203831188266985,
+      "step": 416
+    },
+    {
+      "epoch": 52.45,
+      "learning_rate": 0.000299746835443038,
+      "loss": 1.5722,
+      "step": 420
+    },
+    {
+      "epoch": 52.9,
+      "eval_loss": 2.559480667114258,
+      "eval_runtime": 7.4456,
+      "eval_samples_per_second": 8.461,
+      "eval_steps_per_second": 1.074,
+      "eval_wer": 0.9212810535767734,
+      "step": 424
+    },
+    {
+      "epoch": 53.68,
+      "learning_rate": 0.0002972151898734177,
+      "loss": 1.539,
+      "step": 430
+    },
+    {
+      "epoch": 53.9,
+      "eval_loss": 2.816045045852661,
+      "eval_runtime": 7.4737,
+      "eval_samples_per_second": 8.43,
+      "eval_steps_per_second": 1.07,
+      "eval_wer": 0.9117030829093086,
+      "step": 432
+    },
+    {
+      "epoch": 54.9,
+      "learning_rate": 0.0002946835443037975,
+      "loss": 1.538,
+      "step": 440
+    },
+    {
+      "epoch": 54.9,
+      "eval_loss": 2.565622091293335,
+      "eval_runtime": 7.4709,
+      "eval_samples_per_second": 8.433,
+      "eval_steps_per_second": 1.071,
+      "eval_wer": 0.9167913798263992,
+      "step": 440
+    },
+    {
+      "epoch": 55.9,
+      "eval_loss": 2.9076685905456543,
+      "eval_runtime": 7.5072,
+      "eval_samples_per_second": 8.392,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.9075127207422927,
+      "step": 448
+    },
+    {
+      "epoch": 56.23,
+      "learning_rate": 0.00029215189873417724,
+      "loss": 1.624,
+      "step": 450
+    },
+    {
+      "epoch": 56.9,
+      "eval_loss": 2.872545003890991,
+      "eval_runtime": 7.4843,
+      "eval_samples_per_second": 8.418,
+      "eval_steps_per_second": 1.069,
+      "eval_wer": 0.8985333732415445,
+      "step": 456
+    },
+    {
+      "epoch": 57.45,
+      "learning_rate": 0.00028962025316455695,
+      "loss": 1.5052,
+      "step": 460
+    },
+    {
+      "epoch": 57.9,
+      "eval_loss": 2.621701240539551,
+      "eval_runtime": 7.5267,
+      "eval_samples_per_second": 8.37,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.8901526489075127,
+      "step": 464
+    },
+    {
+      "epoch": 58.68,
+      "learning_rate": 0.0002870886075949367,
+      "loss": 1.4367,
+      "step": 470
+    },
+    {
+      "epoch": 58.9,
+      "eval_loss": 2.5042622089385986,
+      "eval_runtime": 7.5036,
+      "eval_samples_per_second": 8.396,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.8976354384914697,
+      "step": 472
+    },
+    {
+      "epoch": 59.9,
+      "learning_rate": 0.0002845569620253165,
+      "loss": 1.4814,
+      "step": 480
+    },
+    {
+      "epoch": 59.9,
+      "eval_loss": 3.090513229370117,
+      "eval_runtime": 7.5001,
+      "eval_samples_per_second": 8.4,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.8949416342412452,
+      "step": 480
+    },
+    {
+      "epoch": 60.9,
+      "eval_loss": 2.7212915420532227,
+      "eval_runtime": 7.5374,
+      "eval_samples_per_second": 8.358,
+      "eval_steps_per_second": 1.061,
+      "eval_wer": 0.899131996408261,
+      "step": 488
+    },
+    {
+      "epoch": 61.23,
+      "learning_rate": 0.0002820253164556962,
+      "loss": 1.5696,
+      "step": 490
+    },
+    {
+      "epoch": 61.9,
+      "eval_loss": 2.6882176399230957,
+      "eval_runtime": 7.5273,
+      "eval_samples_per_second": 8.37,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.8913498952409459,
+      "step": 496
+    },
+    {
+      "epoch": 62.45,
+      "learning_rate": 0.00027949367088607593,
+      "loss": 1.4408,
+      "step": 500
+    },
+    {
+      "epoch": 62.9,
+      "eval_loss": 2.732247829437256,
+      "eval_runtime": 7.5125,
+      "eval_samples_per_second": 8.386,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.8736905118228075,
+      "step": 504
+    },
+    {
+      "epoch": 63.68,
+      "learning_rate": 0.0002769620253164557,
+      "loss": 1.4065,
+      "step": 510
+    },
+    {
+      "epoch": 63.9,
+      "eval_loss": 2.7024385929107666,
+      "eval_runtime": 7.5194,
+      "eval_samples_per_second": 8.378,
+      "eval_steps_per_second": 1.064,
+      "eval_wer": 0.8826698593235558,
+      "step": 512
+    },
+    {
+      "epoch": 64.9,
+      "learning_rate": 0.00027443037974683547,
+      "loss": 1.3989,
+      "step": 520
+    },
+    {
+      "epoch": 64.9,
+      "eval_loss": 2.6808624267578125,
+      "eval_runtime": 7.5114,
+      "eval_samples_per_second": 8.387,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.8805746782400479,
+      "step": 520
+    },
+    {
+      "epoch": 65.9,
+      "eval_loss": 2.6340062618255615,
+      "eval_runtime": 7.4638,
+      "eval_samples_per_second": 8.441,
+      "eval_steps_per_second": 1.072,
+      "eval_wer": 0.8865609099072134,
+      "step": 528
+    },
+    {
+      "epoch": 66.23,
+      "learning_rate": 0.0002718987341772152,
+      "loss": 1.5102,
+      "step": 530
+    },
+    {
+      "epoch": 66.9,
+      "eval_loss": 2.865128993988037,
+      "eval_runtime": 7.4899,
+      "eval_samples_per_second": 8.411,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.885064351990422,
+      "step": 536
+    },
+    {
+      "epoch": 67.45,
+      "learning_rate": 0.00026936708860759496,
+      "loss": 1.4158,
+      "step": 540
+    },
+    {
+      "epoch": 67.9,
+      "eval_loss": 2.7928247451782227,
+      "eval_runtime": 7.4996,
+      "eval_samples_per_second": 8.401,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.8757856929063155,
+      "step": 544
+    },
+    {
+      "epoch": 68.68,
+      "learning_rate": 0.0002668354430379747,
+      "loss": 1.3322,
+      "step": 550
+    },
+    {
+      "epoch": 68.9,
+      "eval_loss": 2.754878282546997,
+      "eval_runtime": 7.529,
+      "eval_samples_per_second": 8.368,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.8799760550733313,
+      "step": 552
+    },
+    {
+      "epoch": 69.9,
+      "learning_rate": 0.00026430379746835445,
+      "loss": 1.4226,
+      "step": 560
+    },
+    {
+      "epoch": 69.9,
+      "eval_loss": 2.6617932319641113,
+      "eval_runtime": 7.6013,
+      "eval_samples_per_second": 8.288,
+      "eval_steps_per_second": 1.052,
+      "eval_wer": 0.8775815624064651,
+      "step": 560
+    },
+    {
+      "epoch": 70.9,
+      "eval_loss": 2.6936562061309814,
+      "eval_runtime": 7.5105,
+      "eval_samples_per_second": 8.388,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.8650104759054176,
+      "step": 568
+    },
+    {
+      "epoch": 71.23,
+      "learning_rate": 0.0002617721518987342,
+      "loss": 1.4735,
+      "step": 570
+    },
+    {
+      "epoch": 71.9,
+      "eval_loss": 2.628467321395874,
+      "eval_runtime": 7.5847,
+      "eval_samples_per_second": 8.306,
+      "eval_steps_per_second": 1.055,
+      "eval_wer": 0.8668063454055672,
+      "step": 576
+    },
+    {
+      "epoch": 72.45,
+      "learning_rate": 0.00025924050632911394,
+      "loss": 1.338,
+      "step": 580
+    },
+    {
+      "epoch": 72.9,
+      "eval_loss": 2.5627574920654297,
+      "eval_runtime": 7.5097,
+      "eval_samples_per_second": 8.389,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.8668063454055672,
+      "step": 584
+    },
+    {
+      "epoch": 73.68,
+      "learning_rate": 0.00025670886075949365,
+      "loss": 1.335,
+      "step": 590
+    },
+    {
+      "epoch": 73.9,
+      "eval_loss": 2.4783389568328857,
+      "eval_runtime": 7.4904,
+      "eval_samples_per_second": 8.411,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.8608201137384017,
+      "step": 592
+    },
+    {
+      "epoch": 74.9,
+      "learning_rate": 0.0002541772151898734,
+      "loss": 1.3433,
+      "step": 600
+    },
+    {
+      "epoch": 74.9,
+      "eval_loss": 2.654869318008423,
+      "eval_runtime": 7.5287,
+      "eval_samples_per_second": 8.368,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.8605208021550435,
+      "step": 600
+    },
+    {
+      "epoch": 75.9,
+      "eval_loss": 2.3850927352905273,
+      "eval_runtime": 7.5228,
+      "eval_samples_per_second": 8.375,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.853636635737803,
+      "step": 608
+    },
+    {
+      "epoch": 76.23,
+      "learning_rate": 0.0002516455696202532,
+      "loss": 1.4341,
+      "step": 610
+    },
+    {
+      "epoch": 76.9,
+      "eval_loss": 2.6057424545288086,
+      "eval_runtime": 7.5058,
+      "eval_samples_per_second": 8.394,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.8641125411553428,
+      "step": 616
+    },
+    {
+      "epoch": 77.45,
+      "learning_rate": 0.0002491139240506329,
+      "loss": 1.3036,
+      "step": 620
+    },
+    {
+      "epoch": 77.9,
+      "eval_loss": 2.414400100708008,
+      "eval_runtime": 7.5426,
+      "eval_samples_per_second": 8.353,
+      "eval_steps_per_second": 1.061,
+      "eval_wer": 0.8614187369051183,
+      "step": 624
+    },
+    {
+      "epoch": 78.68,
+      "learning_rate": 0.00024658227848101263,
+      "loss": 1.2617,
+      "step": 630
+    },
+    {
+      "epoch": 78.9,
+      "eval_loss": 2.500195264816284,
+      "eval_runtime": 7.4699,
+      "eval_samples_per_second": 8.434,
+      "eval_steps_per_second": 1.071,
+      "eval_wer": 0.8596228674049686,
+      "step": 632
+    },
+    {
+      "epoch": 79.9,
+      "learning_rate": 0.00024405063291139243,
+      "loss": 1.308,
+      "step": 640
+    },
+    {
+      "epoch": 79.9,
+      "eval_loss": 2.492858648300171,
+      "eval_runtime": 7.51,
+      "eval_samples_per_second": 8.389,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.8575276863214606,
+      "step": 640
+    },
+    {
+      "epoch": 80.9,
+      "eval_loss": 2.6992645263671875,
+      "eval_runtime": 7.5434,
+      "eval_samples_per_second": 8.352,
+      "eval_steps_per_second": 1.061,
+      "eval_wer": 0.8461538461538461,
+      "step": 648
+    },
+    {
+      "epoch": 81.23,
+      "learning_rate": 0.00024151898734177217,
+      "loss": 1.3877,
+      "step": 650
+    },
+    {
+      "epoch": 81.9,
+      "eval_loss": 2.587432861328125,
+      "eval_runtime": 7.5224,
+      "eval_samples_per_second": 8.375,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.8509428314875785,
+      "step": 656
+    },
+    {
+      "epoch": 82.45,
+      "learning_rate": 0.0002389873417721519,
+      "loss": 1.2553,
+      "step": 660
+    },
+    {
+      "epoch": 82.9,
+      "eval_loss": 2.6430294513702393,
+      "eval_runtime": 7.5162,
+      "eval_samples_per_second": 8.382,
+      "eval_steps_per_second": 1.064,
+      "eval_wer": 0.853636635737803,
+      "step": 664
+    },
+    {
+      "epoch": 83.68,
+      "learning_rate": 0.00023645569620253169,
+      "loss": 1.211,
+      "step": 670
+    },
+    {
+      "epoch": 83.9,
+      "eval_loss": 3.0369396209716797,
+      "eval_runtime": 7.4884,
+      "eval_samples_per_second": 8.413,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.8497455851541454,
+      "step": 672
+    },
+    {
+      "epoch": 84.9,
+      "learning_rate": 0.0002339240506329114,
+      "loss": 1.2647,
+      "step": 680
+    },
+    {
+      "epoch": 84.9,
+      "eval_loss": 2.701183795928955,
+      "eval_runtime": 7.5489,
+      "eval_samples_per_second": 8.346,
+      "eval_steps_per_second": 1.06,
+      "eval_wer": 0.84375935348698,
+      "step": 680
+    },
+    {
+      "epoch": 85.9,
+      "eval_loss": 2.5128839015960693,
+      "eval_runtime": 7.5032,
+      "eval_samples_per_second": 8.396,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.8581263094881771,
+      "step": 688
+    },
+    {
+      "epoch": 86.23,
+      "learning_rate": 0.00023139240506329115,
+      "loss": 1.3168,
+      "step": 690
+    },
+    {
+      "epoch": 86.9,
+      "eval_loss": 2.512349843978882,
+      "eval_runtime": 7.5078,
+      "eval_samples_per_second": 8.391,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.8452559114037713,
+      "step": 696
+    },
+    {
+      "epoch": 87.45,
+      "learning_rate": 0.00022886075949367092,
+      "loss": 1.1997,
+      "step": 700
+    },
+    {
+      "epoch": 87.9,
+      "eval_loss": 2.459209442138672,
+      "eval_runtime": 7.507,
+      "eval_samples_per_second": 8.392,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.850344208320862,
+      "step": 704
+    },
+    {
+      "epoch": 88.68,
+      "learning_rate": 0.00022632911392405066,
+      "loss": 1.1866,
+      "step": 710
+    },
+    {
+      "epoch": 88.9,
+      "eval_loss": 2.6305787563323975,
+      "eval_runtime": 7.5109,
+      "eval_samples_per_second": 8.388,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.8419634839868303,
+      "step": 712
+    },
+    {
+      "epoch": 89.9,
+      "learning_rate": 0.00022379746835443038,
+      "loss": 1.2396,
+      "step": 720
+    },
+    {
+      "epoch": 89.9,
+      "eval_loss": 2.4730277061462402,
+      "eval_runtime": 7.5381,
+      "eval_samples_per_second": 8.358,
+      "eval_steps_per_second": 1.061,
+      "eval_wer": 0.8458545345704879,
+      "step": 720
+    },
+    {
+      "epoch": 90.9,
+      "eval_loss": 2.6145691871643066,
+      "eval_runtime": 7.4985,
+      "eval_samples_per_second": 8.402,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.8488476504040706,
+      "step": 728
+    },
+    {
+      "epoch": 91.23,
+      "learning_rate": 0.00022126582278481012,
+      "loss": 1.3184,
+      "step": 730
+    },
+    {
+      "epoch": 91.9,
+      "eval_loss": 2.620443105697632,
+      "eval_runtime": 7.5157,
+      "eval_samples_per_second": 8.383,
+      "eval_steps_per_second": 1.064,
+      "eval_wer": 0.8422627955701886,
+      "step": 736
+    },
+    {
+      "epoch": 92.45,
+      "learning_rate": 0.0002187341772151899,
+      "loss": 1.1704,
+      "step": 740
+    },
+    {
+      "epoch": 92.9,
+      "eval_loss": 2.8895633220672607,
+      "eval_runtime": 7.493,
+      "eval_samples_per_second": 8.408,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.8440586650703382,
+      "step": 744
+    },
+    {
+      "epoch": 93.68,
+      "learning_rate": 0.00021620253164556964,
+      "loss": 1.1436,
+      "step": 750
+    },
+    {
+      "epoch": 93.9,
+      "eval_loss": 2.997091770172119,
+      "eval_runtime": 7.5434,
+      "eval_samples_per_second": 8.352,
+      "eval_steps_per_second": 1.061,
+      "eval_wer": 0.8389703681532475,
+      "step": 752
+    },
+    {
+      "epoch": 94.9,
+      "learning_rate": 0.00021367088607594935,
+      "loss": 1.1716,
+      "step": 760
+    },
+    {
+      "epoch": 94.9,
+      "eval_loss": 2.729346513748169,
+      "eval_runtime": 7.4914,
+      "eval_samples_per_second": 8.41,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.8479497156539958,
+      "step": 760
+    },
+    {
+      "epoch": 95.9,
+      "eval_loss": 2.9620306491851807,
+      "eval_runtime": 7.4885,
+      "eval_samples_per_second": 8.413,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.8425621071535468,
+      "step": 768
+    },
+    {
+      "epoch": 96.23,
+      "learning_rate": 0.00021113924050632915,
+      "loss": 1.2487,
+      "step": 770
+    },
+    {
+      "epoch": 96.9,
+      "eval_loss": 2.6880440711975098,
+      "eval_runtime": 7.5581,
+      "eval_samples_per_second": 8.335,
+      "eval_steps_per_second": 1.058,
+      "eval_wer": 0.8332834480694403,
+      "step": 776
+    },
+    {
+      "epoch": 97.45,
+      "learning_rate": 0.00020860759493670887,
+      "loss": 1.118,
+      "step": 780
+    },
+    {
+      "epoch": 97.9,
+      "eval_loss": 2.6754331588745117,
+      "eval_runtime": 7.4991,
+      "eval_samples_per_second": 8.401,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.8320862017360072,
+      "step": 784
+    },
+    {
+      "epoch": 98.68,
+      "learning_rate": 0.0002060759493670886,
+      "loss": 1.186,
+      "step": 790
+    },
+    {
+      "epoch": 98.9,
+      "eval_loss": 2.6924636363983154,
+      "eval_runtime": 7.4702,
+      "eval_samples_per_second": 8.433,
+      "eval_steps_per_second": 1.071,
+      "eval_wer": 0.8341813828195151,
+      "step": 792
+    },
+    {
+      "epoch": 99.9,
+      "learning_rate": 0.00020354430379746838,
+      "loss": 1.1373,
+      "step": 800
+    },
+    {
+      "epoch": 99.9,
+      "eval_loss": 2.9206714630126953,
+      "eval_runtime": 7.4921,
+      "eval_samples_per_second": 8.409,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.8338820712361569,
+      "step": 800
+    },
+    {
+      "epoch": 100.9,
+      "eval_loss": 2.8559393882751465,
+      "eval_runtime": 7.4976,
+      "eval_samples_per_second": 8.403,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.8353786291529482,
+      "step": 808
+    },
+    {
+      "epoch": 101.23,
+      "learning_rate": 0.00020101265822784813,
+      "loss": 1.2086,
+      "step": 810
+    },
+    {
+      "epoch": 101.9,
+      "eval_loss": 2.977409601211548,
+      "eval_runtime": 7.6531,
+      "eval_samples_per_second": 8.232,
+      "eval_steps_per_second": 1.045,
+      "eval_wer": 0.8335827596527986,
+      "step": 816
+    },
+    {
+      "epoch": 102.45,
+      "learning_rate": 0.00019848101265822785,
+      "loss": 1.1227,
+      "step": 820
+    },
+    {
+      "epoch": 102.9,
+      "eval_loss": 3.0108087062835693,
+      "eval_runtime": 7.4953,
+      "eval_samples_per_second": 8.405,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.8192158036516013,
+      "step": 824
+    },
+    {
+      "epoch": 103.68,
+      "learning_rate": 0.00019594936708860762,
+      "loss": 1.1446,
+      "step": 830
+    },
+    {
+      "epoch": 103.9,
+      "eval_loss": 2.8996899127960205,
+      "eval_runtime": 7.5008,
+      "eval_samples_per_second": 8.399,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.8269979048189164,
+      "step": 832
+    },
+    {
+      "epoch": 104.9,
+      "learning_rate": 0.00019341772151898736,
+      "loss": 1.1142,
+      "step": 840
+    },
+    {
+      "epoch": 104.9,
+      "eval_loss": 2.662632703781128,
+      "eval_runtime": 7.5291,
+      "eval_samples_per_second": 8.368,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.8305896438192159,
+      "step": 840
+    },
+    {
+      "epoch": 105.9,
+      "eval_loss": 2.7736780643463135,
+      "eval_runtime": 7.5386,
+      "eval_samples_per_second": 8.357,
+      "eval_steps_per_second": 1.061,
+      "eval_wer": 0.8195151152349596,
+      "step": 848
+    },
+    {
+      "epoch": 106.23,
+      "learning_rate": 0.0001908860759493671,
+      "loss": 1.1665,
+      "step": 850
+    },
+    {
+      "epoch": 106.9,
+      "eval_loss": 2.544724225997925,
+      "eval_runtime": 7.4986,
+      "eval_samples_per_second": 8.402,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.8186171804848847,
+      "step": 856
+    },
+    {
+      "epoch": 107.45,
+      "learning_rate": 0.00018835443037974685,
+      "loss": 1.1,
+      "step": 860
+    },
+    {
+      "epoch": 107.9,
+      "eval_loss": 2.4471564292907715,
+      "eval_runtime": 7.4869,
+      "eval_samples_per_second": 8.415,
+      "eval_steps_per_second": 1.069,
+      "eval_wer": 0.8311882669859324,
+      "step": 864
+    },
+    {
+      "epoch": 108.68,
+      "learning_rate": 0.0001858227848101266,
+      "loss": 1.0674,
+      "step": 870
+    },
+    {
+      "epoch": 108.9,
+      "eval_loss": 2.406235933303833,
+      "eval_runtime": 7.5313,
+      "eval_samples_per_second": 8.365,
+      "eval_steps_per_second": 1.062,
+      "eval_wer": 0.8225082310685423,
+      "step": 872
+    },
+    {
+      "epoch": 109.9,
+      "learning_rate": 0.00018329113924050634,
+      "loss": 1.0556,
+      "step": 880
+    },
+    {
+      "epoch": 109.9,
+      "eval_loss": 2.4098057746887207,
+      "eval_runtime": 7.4614,
+      "eval_samples_per_second": 8.444,
+      "eval_steps_per_second": 1.072,
+      "eval_wer": 0.8246034121520502,
+      "step": 880
+    },
+    {
+      "epoch": 110.9,
+      "eval_loss": 2.344670534133911,
+      "eval_runtime": 7.5006,
+      "eval_samples_per_second": 8.399,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.8255013469021251,
+      "step": 888
+    },
+    {
+      "epoch": 111.23,
+      "learning_rate": 0.00018075949367088608,
+      "loss": 1.1834,
+      "step": 890
+    },
+    {
+      "epoch": 111.9,
+      "eval_loss": 2.557053804397583,
+      "eval_runtime": 7.5289,
+      "eval_samples_per_second": 8.368,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.8066447171505537,
+      "step": 896
+    },
+    {
+      "epoch": 112.45,
+      "learning_rate": 0.00017822784810126582,
+      "loss": 1.0533,
+      "step": 900
+    },
+    {
+      "epoch": 112.9,
+      "eval_loss": 2.5983164310455322,
+      "eval_runtime": 7.5273,
+      "eval_samples_per_second": 8.369,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.8150254414845854,
+      "step": 904
+    },
+    {
+      "epoch": 113.68,
+      "learning_rate": 0.0001756962025316456,
+      "loss": 1.101,
+      "step": 910
+    },
+    {
+      "epoch": 113.9,
+      "eval_loss": 2.6911237239837646,
+      "eval_runtime": 7.6253,
+      "eval_samples_per_second": 8.262,
+      "eval_steps_per_second": 1.049,
+      "eval_wer": 0.7949715653995809,
+      "step": 912
+    },
+    {
+      "epoch": 114.9,
+      "learning_rate": 0.0001731645569620253,
+      "loss": 1.0633,
+      "step": 920
+    },
+    {
+      "epoch": 114.9,
+      "eval_loss": 2.573255777359009,
+      "eval_runtime": 7.4999,
+      "eval_samples_per_second": 8.4,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.8078419634839868,
+      "step": 920
+    },
+    {
+      "epoch": 115.9,
+      "eval_loss": 2.5812604427337646,
+      "eval_runtime": 7.5652,
+      "eval_samples_per_second": 8.328,
+      "eval_steps_per_second": 1.057,
+      "eval_wer": 0.8192158036516013,
+      "step": 928
+    },
+    {
+      "epoch": 116.23,
+      "learning_rate": 0.00017063291139240508,
+      "loss": 1.1512,
+      "step": 930
+    },
+    {
+      "epoch": 116.9,
+      "eval_loss": 2.623731851577759,
+      "eval_runtime": 7.5059,
+      "eval_samples_per_second": 8.393,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.8135288835677941,
+      "step": 936
+    },
+    {
+      "epoch": 117.45,
+      "learning_rate": 0.00016810126582278483,
+      "loss": 1.0317,
+      "step": 940
+    },
+    {
+      "epoch": 117.9,
+      "eval_loss": 2.5565974712371826,
+      "eval_runtime": 7.5083,
+      "eval_samples_per_second": 8.391,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.8030529781502544,
+      "step": 944
+    },
+    {
+      "epoch": 118.68,
+      "learning_rate": 0.00016556962025316457,
+      "loss": 1.0117,
+      "step": 950
+    },
+    {
+      "epoch": 118.9,
+      "eval_loss": 2.5485475063323975,
+      "eval_runtime": 7.4871,
+      "eval_samples_per_second": 8.414,
+      "eval_steps_per_second": 1.069,
+      "eval_wer": 0.8015564202334631,
+      "step": 952
+    },
+    {
+      "epoch": 119.9,
+      "learning_rate": 0.00016303797468354431,
+      "loss": 1.0556,
+      "step": 960
+    },
+    {
+      "epoch": 119.9,
+      "eval_loss": 2.4270551204681396,
+      "eval_runtime": 7.5607,
+      "eval_samples_per_second": 8.333,
+      "eval_steps_per_second": 1.058,
+      "eval_wer": 0.7979646812331638,
+      "step": 960
+    },
+    {
+      "epoch": 120.9,
+      "eval_loss": 2.6578969955444336,
+      "eval_runtime": 7.5132,
+      "eval_samples_per_second": 8.385,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.7940736306495061,
+      "step": 968
+    },
+    {
+      "epoch": 121.23,
+      "learning_rate": 0.00016050632911392406,
+      "loss": 1.1204,
+      "step": 970
+    },
+    {
+      "epoch": 121.9,
+      "eval_loss": 2.6150407791137695,
+      "eval_runtime": 7.467,
+      "eval_samples_per_second": 8.437,
+      "eval_steps_per_second": 1.071,
+      "eval_wer": 0.7943729422328644,
+      "step": 976
+    },
+    {
+      "epoch": 122.45,
+      "learning_rate": 0.0001579746835443038,
+      "loss": 1.0378,
+      "step": 980
+    },
+    {
+      "epoch": 122.9,
+      "eval_loss": 2.549945116043091,
+      "eval_runtime": 7.4938,
+      "eval_samples_per_second": 8.407,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.8024543549835379,
+      "step": 984
+    },
+    {
+      "epoch": 123.68,
+      "learning_rate": 0.00015544303797468355,
+      "loss": 1.0213,
+      "step": 990
+    },
+    {
+      "epoch": 123.9,
+      "eval_loss": 2.72731876373291,
+      "eval_runtime": 7.5219,
+      "eval_samples_per_second": 8.376,
+      "eval_steps_per_second": 1.064,
+      "eval_wer": 0.7937743190661478,
+      "step": 992
+    },
+    {
+      "epoch": 124.9,
+      "learning_rate": 0.0001529113924050633,
+      "loss": 1.0247,
+      "step": 1000
+    },
+    {
+      "epoch": 124.9,
+      "eval_loss": 2.752246618270874,
+      "eval_runtime": 7.5067,
+      "eval_samples_per_second": 8.392,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.7862915294821909,
+      "step": 1000
+    },
+    {
+      "epoch": 125.9,
+      "eval_loss": 2.916325569152832,
+      "eval_runtime": 7.5407,
+      "eval_samples_per_second": 8.355,
+      "eval_steps_per_second": 1.061,
+      "eval_wer": 0.797066746483089,
+      "step": 1008
+    },
+    {
+      "epoch": 126.23,
+      "learning_rate": 0.00015037974683544303,
+      "loss": 1.0939,
+      "step": 1010
+    },
+    {
+      "epoch": 126.9,
+      "eval_loss": 2.622096300125122,
+      "eval_runtime": 7.5442,
+      "eval_samples_per_second": 8.351,
+      "eval_steps_per_second": 1.06,
+      "eval_wer": 0.789583956899132,
+      "step": 1016
+    },
+    {
+      "epoch": 127.45,
+      "learning_rate": 0.0001478481012658228,
+      "loss": 1.0399,
+      "step": 1020
+    },
+    {
+      "epoch": 127.9,
+      "eval_loss": 2.9418468475341797,
+      "eval_runtime": 7.4804,
+      "eval_samples_per_second": 8.422,
+      "eval_steps_per_second": 1.069,
+      "eval_wer": 0.80035917390003,
+      "step": 1024
+    },
+    {
+      "epoch": 128.68,
+      "learning_rate": 0.00014531645569620252,
+      "loss": 1.0233,
+      "step": 1030
+    },
+    {
+      "epoch": 128.9,
+      "eval_loss": 2.755779266357422,
+      "eval_runtime": 7.4931,
+      "eval_samples_per_second": 8.408,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.7856929063154744,
+      "step": 1032
+    },
+    {
+      "epoch": 129.9,
+      "learning_rate": 0.0001427848101265823,
+      "loss": 0.9702,
+      "step": 1040
+    },
+    {
+      "epoch": 129.9,
+      "eval_loss": 2.574506998062134,
+      "eval_runtime": 7.5545,
+      "eval_samples_per_second": 8.339,
+      "eval_steps_per_second": 1.059,
+      "eval_wer": 0.7904818916492068,
+      "step": 1040
+    },
+    {
+      "epoch": 130.9,
+      "eval_loss": 2.6720142364501953,
+      "eval_runtime": 7.4821,
+      "eval_samples_per_second": 8.42,
+      "eval_steps_per_second": 1.069,
+      "eval_wer": 0.7898832684824902,
+      "step": 1048
+    },
+    {
+      "epoch": 131.23,
+      "learning_rate": 0.00014025316455696204,
+      "loss": 1.0676,
+      "step": 1050
+    },
+    {
+      "epoch": 131.9,
+      "eval_loss": 2.690053701400757,
+      "eval_runtime": 7.499,
+      "eval_samples_per_second": 8.401,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.8021550434001796,
+      "step": 1056
+    },
+    {
+      "epoch": 132.45,
+      "learning_rate": 0.00013772151898734178,
+      "loss": 1.0044,
+      "step": 1060
+    },
+    {
+      "epoch": 132.9,
+      "eval_loss": 2.659367561340332,
+      "eval_runtime": 7.5012,
+      "eval_samples_per_second": 8.399,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.7773121819814427,
+      "step": 1064
+    },
+    {
+      "epoch": 133.68,
+      "learning_rate": 0.00013518987341772153,
+      "loss": 1.0276,
+      "step": 1070
+    },
+    {
+      "epoch": 133.9,
+      "eval_loss": 2.4738552570343018,
+      "eval_runtime": 7.4676,
+      "eval_samples_per_second": 8.436,
+      "eval_steps_per_second": 1.071,
+      "eval_wer": 0.7931756958994313,
+      "step": 1072
+    },
+    {
+      "epoch": 134.9,
+      "learning_rate": 0.00013265822784810127,
+      "loss": 0.949,
+      "step": 1080
+    },
+    {
+      "epoch": 134.9,
+      "eval_loss": 2.539825677871704,
+      "eval_runtime": 7.5085,
+      "eval_samples_per_second": 8.391,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.775516312481293,
+      "step": 1080
+    },
+    {
+      "epoch": 135.9,
+      "eval_loss": 2.6266517639160156,
+      "eval_runtime": 7.4936,
+      "eval_samples_per_second": 8.407,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.7797066746483089,
+      "step": 1088
+    },
+    {
+      "epoch": 136.23,
+      "learning_rate": 0.00013012658227848101,
+      "loss": 1.0508,
+      "step": 1090
+    },
+    {
+      "epoch": 136.9,
+      "eval_loss": 2.482872247695923,
+      "eval_runtime": 7.5101,
+      "eval_samples_per_second": 8.389,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.772223885064352,
+      "step": 1096
+    },
+    {
+      "epoch": 137.45,
+      "learning_rate": 0.00012759493670886076,
+      "loss": 0.9937,
+      "step": 1100
+    },
+    {
+      "epoch": 137.9,
+      "eval_loss": 2.4289324283599854,
+      "eval_runtime": 7.4931,
+      "eval_samples_per_second": 8.408,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.777611493564801,
+      "step": 1104
+    },
+    {
+      "epoch": 138.68,
+      "learning_rate": 0.0001250632911392405,
+      "loss": 0.9677,
+      "step": 1110
+    },
+    {
+      "epoch": 138.9,
+      "eval_loss": 2.584481954574585,
+      "eval_runtime": 7.4633,
+      "eval_samples_per_second": 8.441,
+      "eval_steps_per_second": 1.072,
+      "eval_wer": 0.7815025441484585,
+      "step": 1112
+    },
+    {
+      "epoch": 139.9,
+      "learning_rate": 0.00012253164556962027,
+      "loss": 1.0115,
+      "step": 1120
+    },
+    {
+      "epoch": 139.9,
+      "eval_loss": 2.713249921798706,
+      "eval_runtime": 7.4798,
+      "eval_samples_per_second": 8.423,
+      "eval_steps_per_second": 1.07,
+      "eval_wer": 0.7704280155642024,
+      "step": 1120
+    },
+    {
+      "epoch": 140.9,
+      "eval_loss": 2.429744005203247,
+      "eval_runtime": 7.5782,
+      "eval_samples_per_second": 8.313,
+      "eval_steps_per_second": 1.056,
+      "eval_wer": 0.7835977252319665,
+      "step": 1128
+    },
+    {
+      "epoch": 141.23,
+      "learning_rate": 0.00012,
+      "loss": 1.049,
+      "step": 1130
+    },
+    {
+      "epoch": 141.9,
+      "eval_loss": 2.342973232269287,
+      "eval_runtime": 7.5821,
+      "eval_samples_per_second": 8.309,
+      "eval_steps_per_second": 1.055,
+      "eval_wer": 0.7829991020652499,
+      "step": 1136
+    },
+    {
+      "epoch": 142.45,
+      "learning_rate": 0.00011746835443037976,
+      "loss": 0.9412,
+      "step": 1140
+    },
+    {
+      "epoch": 142.9,
+      "eval_loss": 2.6201610565185547,
+      "eval_runtime": 7.5078,
+      "eval_samples_per_second": 8.391,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.7698293923974858,
+      "step": 1144
+    },
+    {
+      "epoch": 143.68,
+      "learning_rate": 0.00011493670886075949,
+      "loss": 0.9647,
+      "step": 1150
+    },
+    {
+      "epoch": 143.9,
+      "eval_loss": 2.507197141647339,
+      "eval_runtime": 7.5411,
+      "eval_samples_per_second": 8.354,
+      "eval_steps_per_second": 1.061,
+      "eval_wer": 0.7710266387309189,
+      "step": 1152
+    },
+    {
+      "epoch": 144.9,
+      "learning_rate": 0.00011240506329113925,
+      "loss": 0.9839,
+      "step": 1160
+    },
+    {
+      "epoch": 144.9,
+      "eval_loss": 2.364020824432373,
+      "eval_runtime": 7.5081,
+      "eval_samples_per_second": 8.391,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.7752170008979348,
+      "step": 1160
+    },
+    {
+      "epoch": 145.9,
+      "eval_loss": 2.3263261318206787,
+      "eval_runtime": 7.4816,
+      "eval_samples_per_second": 8.421,
+      "eval_steps_per_second": 1.069,
+      "eval_wer": 0.7803052978150254,
+      "step": 1168
+    },
+    {
+      "epoch": 146.23,
+      "learning_rate": 0.000109873417721519,
+      "loss": 1.0245,
+      "step": 1170
+    },
+    {
+      "epoch": 146.9,
+      "eval_loss": 2.4204721450805664,
+      "eval_runtime": 7.4826,
+      "eval_samples_per_second": 8.42,
+      "eval_steps_per_second": 1.069,
+      "eval_wer": 0.7683328344806944,
+      "step": 1176
+    },
+    {
+      "epoch": 147.45,
+      "learning_rate": 0.00010734177215189874,
+      "loss": 0.9537,
+      "step": 1180
+    },
+    {
+      "epoch": 147.9,
+      "eval_loss": 2.3592872619628906,
+      "eval_runtime": 7.5011,
+      "eval_samples_per_second": 8.399,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.7832984136486082,
+      "step": 1184
+    },
+    {
+      "epoch": 148.68,
+      "learning_rate": 0.0001048101265822785,
+      "loss": 0.9787,
+      "step": 1190
+    },
+    {
+      "epoch": 148.9,
+      "eval_loss": 2.5318942070007324,
+      "eval_runtime": 7.597,
+      "eval_samples_per_second": 8.293,
+      "eval_steps_per_second": 1.053,
+      "eval_wer": 0.7740197545645017,
+      "step": 1192
+    },
+    {
+      "epoch": 149.9,
+      "learning_rate": 0.00010227848101265822,
+      "loss": 0.9443,
+      "step": 1200
+    },
+    {
+      "epoch": 149.9,
+      "eval_loss": 2.492283582687378,
+      "eval_runtime": 7.4715,
+      "eval_samples_per_second": 8.432,
+      "eval_steps_per_second": 1.071,
+      "eval_wer": 0.7734211313977851,
+      "step": 1200
+    },
+    {
+      "epoch": 150.9,
+      "eval_loss": 2.393569231033325,
+      "eval_runtime": 7.4907,
+      "eval_samples_per_second": 8.41,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.7725231966477103,
+      "step": 1208
+    },
+    {
+      "epoch": 151.23,
+      "learning_rate": 9.974683544303798e-05,
+      "loss": 1.0125,
+      "step": 1210
+    },
+    {
+      "epoch": 151.9,
+      "eval_loss": 2.4754488468170166,
+      "eval_runtime": 7.5067,
+      "eval_samples_per_second": 8.393,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.761448668063454,
+      "step": 1216
+    },
+    {
+      "epoch": 152.45,
+      "learning_rate": 9.721518987341773e-05,
+      "loss": 0.943,
+      "step": 1220
+    },
+    {
+      "epoch": 152.9,
+      "eval_loss": 2.434126615524292,
+      "eval_runtime": 7.5336,
+      "eval_samples_per_second": 8.362,
+      "eval_steps_per_second": 1.062,
+      "eval_wer": 0.7701287039808441,
+      "step": 1224
+    },
+    {
+      "epoch": 153.68,
+      "learning_rate": 9.468354430379748e-05,
+      "loss": 0.9482,
+      "step": 1230
+    },
+    {
+      "epoch": 153.9,
+      "eval_loss": 2.423170328140259,
+      "eval_runtime": 7.517,
+      "eval_samples_per_second": 8.381,
+      "eval_steps_per_second": 1.064,
+      "eval_wer": 0.7698293923974858,
+      "step": 1232
+    },
+    {
+      "epoch": 154.9,
+      "learning_rate": 9.215189873417723e-05,
+      "loss": 0.8958,
+      "step": 1240
+    },
+    {
+      "epoch": 154.9,
+      "eval_loss": 2.4942498207092285,
+      "eval_runtime": 7.5276,
+      "eval_samples_per_second": 8.369,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.7515713858126309,
+      "step": 1240
+    },
+    {
+      "epoch": 155.9,
+      "eval_loss": 2.516117572784424,
+      "eval_runtime": 7.574,
+      "eval_samples_per_second": 8.318,
+      "eval_steps_per_second": 1.056,
+      "eval_wer": 0.7680335228973362,
+      "step": 1248
+    },
+    {
+      "epoch": 156.23,
+      "learning_rate": 8.962025316455697e-05,
+      "loss": 1.0073,
+      "step": 1250
+    },
+    {
+      "epoch": 156.9,
+      "eval_loss": 2.533907890319824,
+      "eval_runtime": 7.5082,
+      "eval_samples_per_second": 8.391,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.7698293923974858,
+      "step": 1256
+    },
+    {
+      "epoch": 157.45,
+      "learning_rate": 8.708860759493672e-05,
+      "loss": 0.9784,
+      "step": 1260
+    },
+    {
+      "epoch": 157.9,
+      "eval_loss": 2.4986894130706787,
+      "eval_runtime": 7.5302,
+      "eval_samples_per_second": 8.366,
+      "eval_steps_per_second": 1.062,
+      "eval_wer": 0.7560610595630051,
+      "step": 1264
+    },
+    {
+      "epoch": 158.68,
+      "learning_rate": 8.455696202531646e-05,
+      "loss": 0.904,
+      "step": 1270
+    },
+    {
+      "epoch": 158.9,
+      "eval_loss": 2.4728736877441406,
+      "eval_runtime": 7.5286,
+      "eval_samples_per_second": 8.368,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.7560610595630051,
+      "step": 1272
+    },
+    {
+      "epoch": 159.9,
+      "learning_rate": 8.202531645569622e-05,
+      "loss": 0.9352,
+      "step": 1280
+    },
+    {
+      "epoch": 159.9,
+      "eval_loss": 2.4667537212371826,
+      "eval_runtime": 7.4806,
+      "eval_samples_per_second": 8.422,
+      "eval_steps_per_second": 1.069,
+      "eval_wer": 0.7590541753965878,
+      "step": 1280
+    },
+    {
+      "epoch": 160.9,
+      "eval_loss": 2.4546968936920166,
+      "eval_runtime": 7.4781,
+      "eval_samples_per_second": 8.425,
+      "eval_steps_per_second": 1.07,
+      "eval_wer": 0.7557617479796468,
+      "step": 1288
+    },
+    {
+      "epoch": 161.23,
+      "learning_rate": 7.949367088607596e-05,
+      "loss": 1.0036,
+      "step": 1290
+    },
+    {
+      "epoch": 161.9,
+      "eval_loss": 2.6064584255218506,
+      "eval_runtime": 7.5036,
+      "eval_samples_per_second": 8.396,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.764741095480395,
+      "step": 1296
+    },
+    {
+      "epoch": 162.45,
+      "learning_rate": 7.69620253164557e-05,
+      "loss": 0.9437,
+      "step": 1300
+    },
+    {
+      "epoch": 162.9,
+      "eval_loss": 2.5465590953826904,
+      "eval_runtime": 7.6278,
+      "eval_samples_per_second": 8.259,
+      "eval_steps_per_second": 1.049,
+      "eval_wer": 0.7641424723136785,
+      "step": 1304
+    },
+    {
+      "epoch": 163.68,
+      "learning_rate": 7.443037974683545e-05,
+      "loss": 0.8998,
+      "step": 1310
+    },
+    {
+      "epoch": 163.9,
+      "eval_loss": 2.5044281482696533,
+      "eval_runtime": 7.5244,
+      "eval_samples_per_second": 8.373,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.7629452259802454,
+      "step": 1312
+    },
+    {
+      "epoch": 164.9,
+      "learning_rate": 7.189873417721519e-05,
+      "loss": 0.9195,
+      "step": 1320
+    },
+    {
+      "epoch": 164.9,
+      "eval_loss": 2.4214060306549072,
+      "eval_runtime": 7.5004,
+      "eval_samples_per_second": 8.4,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.7632445375636037,
+      "step": 1320
+    },
+    {
+      "epoch": 165.9,
+      "eval_loss": 2.3591132164001465,
+      "eval_runtime": 7.4748,
+      "eval_samples_per_second": 8.428,
+      "eval_steps_per_second": 1.07,
+      "eval_wer": 0.7608500448967375,
+      "step": 1328
+    },
+    {
+      "epoch": 166.23,
+      "learning_rate": 6.936708860759494e-05,
+      "loss": 0.9795,
+      "step": 1330
+    },
+    {
+      "epoch": 166.9,
+      "eval_loss": 2.4736063480377197,
+      "eval_runtime": 7.4961,
+      "eval_samples_per_second": 8.404,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.7542651900628554,
+      "step": 1336
+    },
+    {
+      "epoch": 167.45,
+      "learning_rate": 6.68354430379747e-05,
+      "loss": 0.9041,
+      "step": 1340
+    },
+    {
+      "epoch": 167.9,
+      "eval_loss": 2.4042837619781494,
+      "eval_runtime": 7.5141,
+      "eval_samples_per_second": 8.384,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.7596527985633044,
+      "step": 1344
+    },
+    {
+      "epoch": 168.68,
+      "learning_rate": 6.430379746835444e-05,
+      "loss": 0.9111,
+      "step": 1350
+    },
+    {
+      "epoch": 168.9,
+      "eval_loss": 2.4593567848205566,
+      "eval_runtime": 7.5475,
+      "eval_samples_per_second": 8.347,
+      "eval_steps_per_second": 1.06,
+      "eval_wer": 0.7416941035618079,
+      "step": 1352
+    },
+    {
+      "epoch": 169.9,
+      "learning_rate": 6.177215189873418e-05,
+      "loss": 0.8902,
+      "step": 1360
+    },
+    {
+      "epoch": 169.9,
+      "eval_loss": 2.425161361694336,
+      "eval_runtime": 7.4767,
+      "eval_samples_per_second": 8.426,
+      "eval_steps_per_second": 1.07,
+      "eval_wer": 0.7491768931457647,
+      "step": 1360
+    },
+    {
+      "epoch": 170.9,
+      "eval_loss": 2.4007468223571777,
+      "eval_runtime": 7.5101,
+      "eval_samples_per_second": 8.389,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.7542651900628554,
+      "step": 1368
+    },
+    {
+      "epoch": 171.23,
+      "learning_rate": 5.924050632911393e-05,
+      "loss": 0.9956,
+      "step": 1370
+    },
+    {
+      "epoch": 171.9,
+      "eval_loss": 2.3502612113952637,
+      "eval_runtime": 7.4634,
+      "eval_samples_per_second": 8.441,
+      "eval_steps_per_second": 1.072,
+      "eval_wer": 0.7605507333133792,
+      "step": 1376
+    },
+    {
+      "epoch": 172.45,
+      "learning_rate": 5.670886075949368e-05,
+      "loss": 0.8645,
+      "step": 1380
+    },
+    {
+      "epoch": 172.9,
+      "eval_loss": 2.3732781410217285,
+      "eval_runtime": 7.525,
+      "eval_samples_per_second": 8.372,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.7470817120622568,
+      "step": 1384
+    },
+    {
+      "epoch": 173.68,
+      "learning_rate": 5.417721518987342e-05,
+      "loss": 0.8989,
+      "step": 1390
+    },
+    {
+      "epoch": 173.9,
+      "eval_loss": 2.3426337242126465,
+      "eval_runtime": 7.5601,
+      "eval_samples_per_second": 8.333,
+      "eval_steps_per_second": 1.058,
+      "eval_wer": 0.7545645016462137,
+      "step": 1392
+    },
+    {
+      "epoch": 174.9,
+      "learning_rate": 5.1645569620253165e-05,
+      "loss": 0.8961,
+      "step": 1400
+    },
+    {
+      "epoch": 174.9,
+      "eval_loss": 2.4074254035949707,
+      "eval_runtime": 7.5242,
+      "eval_samples_per_second": 8.373,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.7521700089793475,
+      "step": 1400
+    },
+    {
+      "epoch": 175.9,
+      "eval_loss": 2.4200074672698975,
+      "eval_runtime": 7.512,
+      "eval_samples_per_second": 8.387,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.7545645016462137,
+      "step": 1408
+    },
+    {
+      "epoch": 176.23,
+      "learning_rate": 4.9113924050632915e-05,
+      "loss": 0.9481,
+      "step": 1410
+    },
+    {
+      "epoch": 176.9,
+      "eval_loss": 2.361619234085083,
+      "eval_runtime": 7.5221,
+      "eval_samples_per_second": 8.375,
+      "eval_steps_per_second": 1.064,
+      "eval_wer": 0.7503741394791978,
+      "step": 1416
+    },
+    {
+      "epoch": 177.45,
+      "learning_rate": 4.658227848101266e-05,
+      "loss": 0.9347,
+      "step": 1420
+    },
+    {
+      "epoch": 177.9,
+      "eval_loss": 2.3545103073120117,
+      "eval_runtime": 7.5422,
+      "eval_samples_per_second": 8.353,
+      "eval_steps_per_second": 1.061,
+      "eval_wer": 0.7539658784794971,
+      "step": 1424
+    },
+    {
+      "epoch": 178.68,
+      "learning_rate": 4.405063291139241e-05,
+      "loss": 0.9105,
+      "step": 1430
+    },
+    {
+      "epoch": 178.9,
+      "eval_loss": 2.3841302394866943,
+      "eval_runtime": 7.516,
+      "eval_samples_per_second": 8.382,
+      "eval_steps_per_second": 1.064,
+      "eval_wer": 0.7539658784794971,
+      "step": 1432
+    },
+    {
+      "epoch": 179.9,
+      "learning_rate": 4.1518987341772154e-05,
+      "loss": 0.8936,
+      "step": 1440
+    },
+    {
+      "epoch": 179.9,
+      "eval_loss": 2.4328415393829346,
+      "eval_runtime": 7.5588,
+      "eval_samples_per_second": 8.335,
+      "eval_steps_per_second": 1.058,
+      "eval_wer": 0.7530679437294223,
+      "step": 1440
+    },
+    {
+      "epoch": 180.9,
+      "eval_loss": 2.4595539569854736,
+      "eval_runtime": 7.5083,
+      "eval_samples_per_second": 8.391,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.7473810236456151,
+      "step": 1448
+    },
+    {
+      "epoch": 181.23,
+      "learning_rate": 3.8987341772151905e-05,
+      "loss": 0.9511,
+      "step": 1450
+    },
+    {
+      "epoch": 181.9,
+      "eval_loss": 2.4178130626678467,
+      "eval_runtime": 7.5338,
+      "eval_samples_per_second": 8.362,
+      "eval_steps_per_second": 1.062,
+      "eval_wer": 0.7509727626459144,
+      "step": 1456
+    },
+    {
+      "epoch": 182.45,
+      "learning_rate": 3.645569620253165e-05,
+      "loss": 0.8743,
+      "step": 1460
+    },
+    {
+      "epoch": 182.9,
+      "eval_loss": 2.4075419902801514,
+      "eval_runtime": 7.4937,
+      "eval_samples_per_second": 8.407,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.7512720742292727,
+      "step": 1464
+    },
+    {
+      "epoch": 183.68,
+      "learning_rate": 3.392405063291139e-05,
+      "loss": 0.8905,
+      "step": 1470
+    },
+    {
+      "epoch": 183.9,
+      "eval_loss": 2.3900365829467773,
+      "eval_runtime": 7.5106,
+      "eval_samples_per_second": 8.388,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.7524693205627058,
+      "step": 1472
+    },
+    {
+      "epoch": 184.9,
+      "learning_rate": 3.1392405063291144e-05,
+      "loss": 0.8968,
+      "step": 1480
+    },
+    {
+      "epoch": 184.9,
+      "eval_loss": 2.4383199214935303,
+      "eval_runtime": 7.5418,
+      "eval_samples_per_second": 8.353,
+      "eval_steps_per_second": 1.061,
+      "eval_wer": 0.7542651900628554,
+      "step": 1480
+    },
+    {
+      "epoch": 185.9,
+      "eval_loss": 2.4400503635406494,
+      "eval_runtime": 7.5025,
+      "eval_samples_per_second": 8.397,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.7518706973959892,
+      "step": 1488
+    },
+    {
+      "epoch": 186.23,
+      "learning_rate": 2.8860759493670884e-05,
+      "loss": 0.9459,
+      "step": 1490
+    },
+    {
+      "epoch": 186.9,
+      "eval_loss": 2.4344472885131836,
+      "eval_runtime": 7.4831,
+      "eval_samples_per_second": 8.419,
+      "eval_steps_per_second": 1.069,
+      "eval_wer": 0.7497755163124813,
+      "step": 1496
+    },
+    {
+      "epoch": 187.45,
+      "learning_rate": 2.6329113924050635e-05,
+      "loss": 0.9273,
+      "step": 1500
+    },
+    {
+      "epoch": 187.9,
+      "eval_loss": 2.417949914932251,
+      "eval_runtime": 7.5261,
+      "eval_samples_per_second": 8.371,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.7449865309787489,
+      "step": 1504
+    },
+    {
+      "epoch": 188.68,
+      "learning_rate": 2.379746835443038e-05,
+      "loss": 0.9523,
+      "step": 1510
+    },
+    {
+      "epoch": 188.9,
+      "eval_loss": 2.4077460765838623,
+      "eval_runtime": 7.5161,
+      "eval_samples_per_second": 8.382,
+      "eval_steps_per_second": 1.064,
+      "eval_wer": 0.7524693205627058,
+      "step": 1512
+    },
+    {
+      "epoch": 189.9,
+      "learning_rate": 2.1265822784810126e-05,
+      "loss": 0.8903,
+      "step": 1520
+    },
+    {
+      "epoch": 189.9,
+      "eval_loss": 2.4481801986694336,
+      "eval_runtime": 7.5019,
+      "eval_samples_per_second": 8.398,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.7521700089793475,
+      "step": 1520
+    },
+    {
+      "epoch": 190.9,
+      "eval_loss": 2.4506967067718506,
+      "eval_runtime": 7.5305,
+      "eval_samples_per_second": 8.366,
+      "eval_steps_per_second": 1.062,
+      "eval_wer": 0.7572583058964382,
+      "step": 1528
+    },
+    {
+      "epoch": 191.23,
+      "learning_rate": 1.8734177215189874e-05,
+      "loss": 0.9759,
+      "step": 1530
+    },
+    {
+      "epoch": 191.9,
+      "eval_loss": 2.4391028881073,
+      "eval_runtime": 7.5878,
+      "eval_samples_per_second": 8.303,
+      "eval_steps_per_second": 1.054,
+      "eval_wer": 0.7563603711463633,
+      "step": 1536
+    },
+    {
+      "epoch": 192.45,
+      "learning_rate": 1.620253164556962e-05,
+      "loss": 0.887,
+      "step": 1540
+    },
+    {
+      "epoch": 192.9,
+      "eval_loss": 2.4516143798828125,
+      "eval_runtime": 7.5568,
+      "eval_samples_per_second": 8.337,
+      "eval_steps_per_second": 1.059,
+      "eval_wer": 0.7521700089793475,
+      "step": 1544
+    },
+    {
+      "epoch": 193.68,
+      "learning_rate": 1.3670886075949368e-05,
+      "loss": 0.8796,
+      "step": 1550
+    },
+    {
+      "epoch": 193.9,
+      "eval_loss": 2.440375328063965,
+      "eval_runtime": 7.5815,
+      "eval_samples_per_second": 8.31,
+      "eval_steps_per_second": 1.055,
+      "eval_wer": 0.7542651900628554,
+      "step": 1552
+    },
+    {
+      "epoch": 194.9,
+      "learning_rate": 1.1139240506329114e-05,
+      "loss": 0.861,
+      "step": 1560
+    },
+    {
+      "epoch": 194.9,
+      "eval_loss": 2.4268441200256348,
+      "eval_runtime": 7.5115,
+      "eval_samples_per_second": 8.387,
+      "eval_steps_per_second": 1.065,
+      "eval_wer": 0.7518706973959892,
+      "step": 1560
+    },
+    {
+      "epoch": 195.9,
+      "eval_loss": 2.4319472312927246,
+      "eval_runtime": 7.4835,
+      "eval_samples_per_second": 8.418,
+      "eval_steps_per_second": 1.069,
+      "eval_wer": 0.7506734510625561,
+      "step": 1568
+    },
+    {
+      "epoch": 196.23,
+      "learning_rate": 8.607594936708861e-06,
+      "loss": 0.9349,
+      "step": 1570
+    },
+    {
+      "epoch": 196.9,
+      "eval_loss": 2.422971487045288,
+      "eval_runtime": 7.5236,
+      "eval_samples_per_second": 8.374,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.7539658784794971,
+      "step": 1576
+    },
+    {
+      "epoch": 197.45,
+      "learning_rate": 6.075949367088608e-06,
+      "loss": 0.9154,
+      "step": 1580
+    },
+    {
+      "epoch": 197.9,
+      "eval_loss": 2.4296085834503174,
+      "eval_runtime": 7.5409,
+      "eval_samples_per_second": 8.354,
+      "eval_steps_per_second": 1.061,
+      "eval_wer": 0.7557617479796468,
+      "step": 1584
+    },
+    {
+      "epoch": 198.68,
+      "learning_rate": 3.544303797468355e-06,
+      "loss": 0.8695,
+      "step": 1590
+    },
+    {
+      "epoch": 198.9,
+      "eval_loss": 2.430819511413574,
+      "eval_runtime": 7.5825,
+      "eval_samples_per_second": 8.309,
+      "eval_steps_per_second": 1.055,
+      "eval_wer": 0.7557617479796468,
+      "step": 1592
+    },
+    {
+      "epoch": 199.9,
+      "learning_rate": 1.0126582278481013e-06,
+      "loss": 0.8754,
+      "step": 1600
+    },
+    {
+      "epoch": 199.9,
+      "eval_loss": 2.428027868270874,
+      "eval_runtime": 7.4856,
+      "eval_samples_per_second": 8.416,
+      "eval_steps_per_second": 1.069,
+      "eval_wer": 0.7545645016462137,
+      "step": 1600
+    },
+    {
+      "epoch": 199.9,
+      "step": 1600,
+      "total_flos": 1.1848602213269387e+19,
+      "train_loss": 2.0372400057315825,
+      "train_runtime": 18220.1543,
+      "train_samples_per_second": 6.18,
+      "train_steps_per_second": 0.088
+    }
+  ],
+  "max_steps": 1600,
+  "num_train_epochs": 200,
+  "total_flos": 1.1848602213269387e+19,
+  "trial_name": null,
+  "trial_params": null
+}