End of training

all_results.json

@@ -1,12 +1,12 @@
 {
-    "epoch": 12.01,
-    "eval_loss": 0.3812163174152374,
-    "eval_runtime": 951.9575,
-    "eval_samples_per_second": 6.924,
-    "eval_steps_per_second": 0.433,
-    "eval_wer": 18.775568066750374,
-    "train_loss": 0.106446673027426,
-    "train_runtime": 27653.1068,
-    "train_samples_per_second": 5.786,
-    "train_steps_per_second": 0.181
+    "epoch": 1.34,
+    "eval_loss": 0.2256375402212143,
+    "eval_runtime": 1793.4047,
+    "eval_samples_per_second": 3.675,
+    "eval_steps_per_second": 0.459,
+    "eval_wer": 13.996111628660538,
+    "train_loss": 0.2607154120922089,
+    "train_runtime": 15779.8251,
+    "train_samples_per_second": 1.267,
+    "train_steps_per_second": 0.317
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 12.01,
-    "eval_loss": 0.3812163174152374,
-    "eval_runtime": 951.9575,
-    "eval_samples_per_second": 6.924,
-    "eval_steps_per_second": 0.433,
-    "eval_wer": 18.775568066750374
+    "epoch": 1.34,
+    "eval_loss": 0.2256375402212143,
+    "eval_runtime": 1793.4047,
+    "eval_samples_per_second": 3.675,
+    "eval_steps_per_second": 0.459,
+    "eval_wer": 13.996111628660538
 }

nohup.out

@@ -201089,3 +201089,25 @@ Training completed. Do not forget to share your model on huggingface.co/models =
    ca66a5d..029efe6  main -> main
 
 
+To https://huggingface.co/xezpeleta/whisper-medium-eu
+   029efe6..257c7d5  main -> main
+
+07/23/2023 03:05:47 - WARNING - huggingface_hub.repository - To https://huggingface.co/xezpeleta/whisper-medium-eu
+   029efe6..257c7d5  main -> main
+
+***** train metrics *****
+  epoch                    =       1.34
+  train_loss               =     0.2607
+  train_runtime            = 4:22:59.82
+  train_samples_per_second =      1.267
+  train_steps_per_second   =      0.317
+07/23/2023 03:05:50 - INFO - __main__ - *** Evaluate ***
+[INFO|trainer.py:2955] 2023-07-23 03:05:50,448 >> ***** Running Evaluation *****
+[INFO|trainer.py:2959] 2023-07-23 03:05:50,448 >>   Num examples: Unknown
+[INFO|trainer.py:2960] 2023-07-23 03:05:50,448 >>   Batch size = 8
+
+[INFO|trainer_utils.py:689] 2023-07-23 03:05:52,742 >> The following columns in the evaluation set don't have a corresponding argument in `WhisperForConditionalGeneration.forward` and have been ignored: input_length. If input_length are not expected by `WhisperForConditionalGeneration.forward`,  you can safely ignore this message.
+[INFO|trainer.py:2700] 2023-07-23 03:35:43,854 >> Saving model checkpoint to ./
+[INFO|configuration_utils.py:447] 2023-07-23 03:35:43,854 >> Configuration saved in ./config.json
+[INFO|modeling_utils.py:1693] 2023-07-23 03:35:47,924 >> Model weights saved in ./pytorch_model.bin
+[INFO|feature_extraction_utils.py:368] 2023-07-23 03:35:47,934 >> Feature extractor saved in ./preprocessor_config.json

runs/Jul22_22-41-53_tknadmin-System-Product-Name/events.out.tfevents.1690076143.tknadmin-System-Product-Name.3969872.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd2634054e3fd7f0c8665ce7a06b958944c2ac37b607eaaa0c041c13f11b262c
+size 358

train_results.json

@@ -1,7 +1,7 @@
 {
-    "epoch": 12.01,
-    "train_loss": 0.106446673027426,
-    "train_runtime": 27653.1068,
-    "train_samples_per_second": 5.786,
-    "train_steps_per_second": 0.181
+    "epoch": 1.34,
+    "train_loss": 0.2607154120922089,
+    "train_runtime": 15779.8251,
+    "train_samples_per_second": 1.267,
+    "train_steps_per_second": 0.317
 }

trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 18.775568066750374,
-  "best_model_checkpoint": "./checkpoint-4000",
-  "epoch": 12.0112,
+  "best_metric": 13.996111628660538,
+  "best_model_checkpoint": "./checkpoint-5000",
+  "epoch": 1.3408,
   "global_step": 5000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
@@ -9,1262 +9,1262 @@
   "log_history": [
     {
       "epoch": 0.01,
-      "learning_rate": 4.4e-07,
-      "loss": 1.6639,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 1.4182,
       "step": 25
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.400000000000001e-07,
-      "loss": 1.4728,
+      "loss": 1.292,
       "step": 50
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.44e-06,
-      "loss": 1.1563,
+      "loss": 1.0018,
       "step": 75
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.94e-06,
-      "loss": 0.9368,
+      "loss": 0.7765,
       "step": 100
     },
     {
       "epoch": 0.03,
       "learning_rate": 2.4400000000000004e-06,
-      "loss": 0.8889,
+      "loss": 0.7103,
       "step": 125
     },
     {
       "epoch": 0.03,
       "learning_rate": 2.9400000000000002e-06,
-      "loss": 0.8114,
+      "loss": 0.6597,
       "step": 150
     },
     {
       "epoch": 0.04,
       "learning_rate": 3.44e-06,
-      "loss": 0.6924,
+      "loss": 0.6657,
       "step": 175
     },
     {
       "epoch": 0.04,
       "learning_rate": 3.94e-06,
-      "loss": 0.661,
+      "loss": 0.5853,
       "step": 200
     },
     {
       "epoch": 0.04,
       "learning_rate": 4.440000000000001e-06,
-      "loss": 0.6427,
+      "loss": 0.5273,
       "step": 225
     },
     {
       "epoch": 0.05,
       "learning_rate": 4.94e-06,
-      "loss": 0.5217,
+      "loss": 0.5979,
       "step": 250
     },
     {
       "epoch": 0.06,
       "learning_rate": 5.4400000000000004e-06,
-      "loss": 0.4991,
+      "loss": 0.5861,
       "step": 275
     },
     {
       "epoch": 0.06,
       "learning_rate": 5.94e-06,
-      "loss": 0.4768,
+      "loss": 0.5085,
       "step": 300
     },
     {
       "epoch": 0.07,
       "learning_rate": 6.440000000000001e-06,
-      "loss": 0.4632,
+      "loss": 0.4827,
       "step": 325
     },
     {
       "epoch": 0.07,
       "learning_rate": 6.9400000000000005e-06,
-      "loss": 0.4289,
+      "loss": 0.4909,
       "step": 350
     },
     {
       "epoch": 0.07,
       "learning_rate": 7.440000000000001e-06,
-      "loss": 0.404,
+      "loss": 0.4651,
       "step": 375
     },
     {
       "epoch": 0.08,
       "learning_rate": 7.94e-06,
-      "loss": 0.3833,
+      "loss": 0.494,
       "step": 400
     },
     {
-      "epoch": 1.0,
+      "epoch": 0.09,
       "learning_rate": 8.44e-06,
-      "loss": 0.4201,
+      "loss": 0.4188,
       "step": 425
     },
     {
-      "epoch": 1.01,
+      "epoch": 0.09,
       "learning_rate": 8.94e-06,
-      "loss": 0.3989,
+      "loss": 0.3849,
       "step": 450
     },
     {
-      "epoch": 1.01,
+      "epoch": 0.1,
       "learning_rate": 9.440000000000001e-06,
-      "loss": 0.3453,
+      "loss": 0.4577,
       "step": 475
     },
     {
-      "epoch": 1.02,
+      "epoch": 0.1,
       "learning_rate": 9.940000000000001e-06,
-      "loss": 0.3559,
+      "loss": 0.4415,
       "step": 500
     },
     {
-      "epoch": 1.02,
+      "epoch": 0.1,
       "learning_rate": 9.951111111111111e-06,
-      "loss": 0.3429,
+      "loss": 0.4615,
       "step": 525
     },
     {
-      "epoch": 1.03,
+      "epoch": 0.11,
       "learning_rate": 9.895555555555557e-06,
-      "loss": 0.3701,
+      "loss": 0.4282,
       "step": 550
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 9.84e-06,
-      "loss": 0.3084,
+      "epoch": 0.12,
+      "learning_rate": 9.842222222222223e-06,
+      "loss": 0.4481,
       "step": 575
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.784444444444445e-06,
-      "loss": 0.2956,
+      "epoch": 0.12,
+      "learning_rate": 9.786666666666667e-06,
+      "loss": 0.4441,
       "step": 600
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.72888888888889e-06,
-      "loss": 0.2923,
+      "epoch": 0.12,
+      "learning_rate": 9.731111111111113e-06,
+      "loss": 0.4238,
       "step": 625
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.673333333333334e-06,
-      "loss": 0.2786,
+      "epoch": 0.13,
+      "learning_rate": 9.675555555555555e-06,
+      "loss": 0.4245,
       "step": 650
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.617777777777778e-06,
-      "loss": 0.2427,
+      "epoch": 0.14,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 0.4118,
       "step": 675
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.562222222222223e-06,
-      "loss": 0.2387,
+      "epoch": 0.14,
+      "learning_rate": 9.564444444444445e-06,
+      "loss": 0.4111,
       "step": 700
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.506666666666667e-06,
-      "loss": 0.2371,
+      "epoch": 0.14,
+      "learning_rate": 9.508888888888889e-06,
+      "loss": 0.3642,
       "step": 725
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.451111111111112e-06,
-      "loss": 0.2114,
+      "epoch": 0.15,
+      "learning_rate": 9.453333333333335e-06,
+      "loss": 0.401,
       "step": 750
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.395555555555556e-06,
-      "loss": 0.1894,
+      "epoch": 0.15,
+      "learning_rate": 9.397777777777779e-06,
+      "loss": 0.3855,
       "step": 775
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.340000000000002e-06,
-      "loss": 0.1867,
+      "epoch": 0.16,
+      "learning_rate": 9.342222222222223e-06,
+      "loss": 0.3668,
       "step": 800
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 9.284444444444444e-06,
-      "loss": 0.1927,
+      "epoch": 0.17,
+      "learning_rate": 9.286666666666667e-06,
+      "loss": 0.3794,
       "step": 825
     },
     {
-      "epoch": 2.01,
-      "learning_rate": 9.22888888888889e-06,
-      "loss": 0.1951,
+      "epoch": 0.17,
+      "learning_rate": 9.231111111111111e-06,
+      "loss": 0.4296,
       "step": 850
     },
     {
-      "epoch": 2.01,
-      "learning_rate": 9.173333333333334e-06,
-      "loss": 0.1775,
+      "epoch": 0.17,
+      "learning_rate": 9.175555555555557e-06,
+      "loss": 0.4003,
       "step": 875
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 9.117777777777778e-06,
-      "loss": 0.1672,
+      "epoch": 0.18,
+      "learning_rate": 9.12e-06,
+      "loss": 0.374,
       "step": 900
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 9.062222222222224e-06,
-      "loss": 0.1591,
+      "epoch": 0.18,
+      "learning_rate": 9.064444444444447e-06,
+      "loss": 0.4051,
       "step": 925
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 9.006666666666666e-06,
-      "loss": 0.1703,
+      "epoch": 0.19,
+      "learning_rate": 9.008888888888889e-06,
+      "loss": 0.3806,
       "step": 950
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 8.951111111111112e-06,
-      "loss": 0.1636,
+      "epoch": 0.2,
+      "learning_rate": 8.953333333333335e-06,
+      "loss": 0.4161,
       "step": 975
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 8.895555555555556e-06,
-      "loss": 0.1413,
+      "epoch": 0.2,
+      "learning_rate": 8.897777777777779e-06,
+      "loss": 0.4198,
       "step": 1000
     },
     {
-      "epoch": 2.04,
-      "eval_loss": 0.31775403022766113,
-      "eval_runtime": 963.5028,
-      "eval_samples_per_second": 6.841,
-      "eval_steps_per_second": 0.428,
-      "eval_wer": 22.013852322896838,
+      "epoch": 0.2,
+      "eval_loss": 0.41016528010368347,
+      "eval_runtime": 1814.0427,
+      "eval_samples_per_second": 3.633,
+      "eval_steps_per_second": 0.454,
+      "eval_wer": 28.34865729677184,
       "step": 1000
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 8.84e-06,
-      "loss": 0.1371,
+      "epoch": 0.2,
+      "learning_rate": 8.842222222222223e-06,
+      "loss": 0.409,
       "step": 1025
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 8.784444444444446e-06,
-      "loss": 0.1494,
+      "epoch": 0.21,
+      "learning_rate": 8.786666666666668e-06,
+      "loss": 0.3674,
       "step": 1050
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 8.72888888888889e-06,
-      "loss": 0.1196,
+      "epoch": 0.21,
+      "learning_rate": 8.73111111111111e-06,
+      "loss": 0.3591,
       "step": 1075
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 8.673333333333334e-06,
-      "loss": 0.1152,
+      "epoch": 0.22,
+      "learning_rate": 8.675555555555556e-06,
+      "loss": 0.3892,
       "step": 1100
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 8.617777777777778e-06,
-      "loss": 0.1158,
+      "epoch": 0.23,
+      "learning_rate": 8.62e-06,
+      "loss": 0.3843,
       "step": 1125
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 8.562222222222224e-06,
-      "loss": 0.1054,
+      "epoch": 0.23,
+      "learning_rate": 8.564444444444445e-06,
+      "loss": 0.3605,
       "step": 1150
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 8.506666666666668e-06,
-      "loss": 0.0912,
+      "epoch": 0.23,
+      "learning_rate": 8.50888888888889e-06,
+      "loss": 0.326,
       "step": 1175
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 8.451111111111112e-06,
-      "loss": 0.0915,
+      "epoch": 0.24,
+      "learning_rate": 8.453333333333334e-06,
+      "loss": 0.3103,
       "step": 1200
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 8.395555555555557e-06,
-      "loss": 0.0901,
+      "epoch": 0.24,
+      "learning_rate": 8.397777777777778e-06,
+      "loss": 0.2766,
       "step": 1225
     },
     {
-      "epoch": 3.0,
-      "learning_rate": 8.34e-06,
-      "loss": 0.0953,
+      "epoch": 0.25,
+      "learning_rate": 8.342222222222222e-06,
+      "loss": 0.3204,
       "step": 1250
     },
     {
-      "epoch": 3.01,
-      "learning_rate": 8.284444444444446e-06,
-      "loss": 0.0923,
+      "epoch": 0.26,
+      "learning_rate": 8.286666666666668e-06,
+      "loss": 0.3426,
       "step": 1275
     },
     {
-      "epoch": 3.01,
-      "learning_rate": 8.22888888888889e-06,
-      "loss": 0.0749,
+      "epoch": 0.26,
+      "learning_rate": 8.231111111111112e-06,
+      "loss": 0.3417,
       "step": 1300
     },
     {
-      "epoch": 3.02,
-      "learning_rate": 8.173333333333334e-06,
-      "loss": 0.0796,
+      "epoch": 0.27,
+      "learning_rate": 8.175555555555556e-06,
+      "loss": 0.3179,
       "step": 1325
     },
     {
-      "epoch": 3.02,
-      "learning_rate": 8.11777777777778e-06,
-      "loss": 0.0786,
+      "epoch": 0.27,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 0.2598,
       "step": 1350
     },
     {
-      "epoch": 3.03,
-      "learning_rate": 8.062222222222222e-06,
-      "loss": 0.0819,
+      "epoch": 0.28,
+      "learning_rate": 8.064444444444444e-06,
+      "loss": 0.3453,
       "step": 1375
     },
     {
-      "epoch": 3.03,
-      "learning_rate": 8.006666666666667e-06,
-      "loss": 0.0747,
+      "epoch": 0.28,
+      "learning_rate": 8.00888888888889e-06,
+      "loss": 0.2752,
       "step": 1400
     },
     {
-      "epoch": 3.04,
-      "learning_rate": 7.951111111111111e-06,
-      "loss": 0.0646,
+      "epoch": 0.28,
+      "learning_rate": 7.953333333333334e-06,
+      "loss": 0.2927,
       "step": 1425
     },
     {
-      "epoch": 3.04,
-      "learning_rate": 7.895555555555557e-06,
-      "loss": 0.0707,
+      "epoch": 0.29,
+      "learning_rate": 7.897777777777778e-06,
+      "loss": 0.3859,
       "step": 1450
     },
     {
-      "epoch": 3.05,
-      "learning_rate": 7.840000000000001e-06,
-      "loss": 0.0639,
+      "epoch": 0.29,
+      "learning_rate": 7.842222222222224e-06,
+      "loss": 0.3137,
       "step": 1475
     },
     {
-      "epoch": 3.05,
-      "learning_rate": 7.784444444444445e-06,
-      "loss": 0.0544,
+      "epoch": 0.3,
+      "learning_rate": 7.786666666666666e-06,
+      "loss": 0.2678,
       "step": 1500
     },
     {
-      "epoch": 3.06,
-      "learning_rate": 7.72888888888889e-06,
-      "loss": 0.0531,
+      "epoch": 0.3,
+      "learning_rate": 7.731111111111112e-06,
+      "loss": 0.2803,
       "step": 1525
     },
     {
-      "epoch": 3.06,
-      "learning_rate": 7.673333333333333e-06,
-      "loss": 0.0518,
+      "epoch": 0.31,
+      "learning_rate": 7.675555555555556e-06,
+      "loss": 0.2828,
       "step": 1550
     },
     {
-      "epoch": 3.07,
-      "learning_rate": 7.617777777777778e-06,
-      "loss": 0.0461,
+      "epoch": 0.32,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 0.3655,
       "step": 1575
     },
     {
-      "epoch": 3.07,
-      "learning_rate": 7.562222222222223e-06,
-      "loss": 0.0424,
+      "epoch": 0.32,
+      "learning_rate": 7.564444444444446e-06,
+      "loss": 0.3321,
       "step": 1600
     },
     {
-      "epoch": 3.08,
-      "learning_rate": 7.506666666666668e-06,
-      "loss": 0.0392,
+      "epoch": 0.33,
+      "learning_rate": 7.50888888888889e-06,
+      "loss": 0.3649,
       "step": 1625
     },
     {
-      "epoch": 4.0,
-      "learning_rate": 7.451111111111111e-06,
-      "loss": 0.0376,
+      "epoch": 0.33,
+      "learning_rate": 7.453333333333334e-06,
+      "loss": 0.3229,
       "step": 1650
     },
     {
-      "epoch": 4.01,
-      "learning_rate": 7.395555555555556e-06,
-      "loss": 0.0447,
+      "epoch": 0.34,
+      "learning_rate": 7.3977777777777786e-06,
+      "loss": 0.3115,
       "step": 1675
     },
     {
-      "epoch": 4.01,
-      "learning_rate": 7.340000000000001e-06,
-      "loss": 0.0362,
+      "epoch": 0.34,
+      "learning_rate": 7.342222222222223e-06,
+      "loss": 0.2925,
       "step": 1700
     },
     {
-      "epoch": 4.02,
-      "learning_rate": 7.284444444444445e-06,
-      "loss": 0.0368,
+      "epoch": 0.34,
+      "learning_rate": 7.2866666666666675e-06,
+      "loss": 0.3014,
       "step": 1725
     },
     {
-      "epoch": 4.02,
-      "learning_rate": 7.22888888888889e-06,
-      "loss": 0.0367,
+      "epoch": 0.35,
+      "learning_rate": 7.231111111111112e-06,
+      "loss": 0.3303,
       "step": 1750
     },
     {
-      "epoch": 4.03,
-      "learning_rate": 7.173333333333335e-06,
-      "loss": 0.0391,
+      "epoch": 0.35,
+      "learning_rate": 7.1755555555555556e-06,
+      "loss": 0.3174,
       "step": 1775
     },
     {
-      "epoch": 4.03,
-      "learning_rate": 7.117777777777778e-06,
-      "loss": 0.0331,
+      "epoch": 0.36,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 0.3249,
       "step": 1800
     },
     {
-      "epoch": 4.04,
-      "learning_rate": 7.062222222222223e-06,
-      "loss": 0.028,
+      "epoch": 0.36,
+      "learning_rate": 7.0644444444444445e-06,
+      "loss": 0.2678,
       "step": 1825
     },
     {
-      "epoch": 4.04,
-      "learning_rate": 7.006666666666667e-06,
-      "loss": 0.0315,
+      "epoch": 0.37,
+      "learning_rate": 7.008888888888889e-06,
+      "loss": 0.3088,
       "step": 1850
     },
     {
-      "epoch": 4.05,
-      "learning_rate": 6.951111111111112e-06,
-      "loss": 0.0308,
+      "epoch": 0.38,
+      "learning_rate": 6.953333333333334e-06,
+      "loss": 0.2515,
       "step": 1875
     },
     {
-      "epoch": 4.05,
-      "learning_rate": 6.8955555555555565e-06,
-      "loss": 0.0248,
+      "epoch": 0.38,
+      "learning_rate": 6.897777777777779e-06,
+      "loss": 0.2838,
       "step": 1900
     },
     {
-      "epoch": 4.06,
-      "learning_rate": 6.8400000000000014e-06,
-      "loss": 0.023,
+      "epoch": 0.39,
+      "learning_rate": 6.842222222222222e-06,
+      "loss": 0.2494,
       "step": 1925
     },
     {
-      "epoch": 4.06,
-      "learning_rate": 6.784444444444445e-06,
-      "loss": 0.0233,
+      "epoch": 0.39,
+      "learning_rate": 6.786666666666667e-06,
+      "loss": 0.205,
       "step": 1950
     },
     {
-      "epoch": 4.07,
-      "learning_rate": 6.7288888888888895e-06,
-      "loss": 0.0236,
+      "epoch": 0.4,
+      "learning_rate": 6.731111111111111e-06,
+      "loss": 0.2439,
       "step": 1975
     },
     {
-      "epoch": 4.07,
-      "learning_rate": 6.6733333333333335e-06,
-      "loss": 0.0181,
+      "epoch": 0.4,
+      "learning_rate": 6.675555555555556e-06,
+      "loss": 0.2547,
       "step": 2000
     },
     {
-      "epoch": 4.07,
-      "eval_loss": 0.33755382895469666,
-      "eval_runtime": 952.5181,
-      "eval_samples_per_second": 6.92,
-      "eval_steps_per_second": 0.433,
-      "eval_wer": 20.286362347604197,
+      "epoch": 0.4,
+      "eval_loss": 0.31417879462242126,
+      "eval_runtime": 1807.9177,
+      "eval_samples_per_second": 3.646,
+      "eval_steps_per_second": 0.456,
+      "eval_wer": 21.643241929604276,
       "step": 2000
     },
     {
-      "epoch": 4.08,
-      "learning_rate": 6.617777777777778e-06,
-      "loss": 0.0176,
+      "epoch": 0.41,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 0.212,
       "step": 2025
     },
     {
-      "epoch": 4.08,
-      "learning_rate": 6.562222222222223e-06,
-      "loss": 0.0177,
+      "epoch": 0.41,
+      "learning_rate": 6.564444444444446e-06,
+      "loss": 0.2386,
       "step": 2050
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 6.5066666666666665e-06,
-      "loss": 0.0198,
+      "epoch": 0.41,
+      "learning_rate": 6.508888888888889e-06,
+      "loss": 0.2429,
       "step": 2075
     },
     {
-      "epoch": 5.01,
-      "learning_rate": 6.451111111111111e-06,
-      "loss": 0.0188,
+      "epoch": 0.42,
+      "learning_rate": 6.453333333333334e-06,
+      "loss": 0.3079,
       "step": 2100
     },
     {
-      "epoch": 5.01,
-      "learning_rate": 6.395555555555556e-06,
-      "loss": 0.0153,
+      "epoch": 0.42,
+      "learning_rate": 6.397777777777778e-06,
+      "loss": 0.2576,
       "step": 2125
     },
     {
-      "epoch": 5.02,
-      "learning_rate": 6.34e-06,
-      "loss": 0.0157,
+      "epoch": 0.43,
+      "learning_rate": 6.342222222222223e-06,
+      "loss": 0.2558,
       "step": 2150
     },
     {
-      "epoch": 5.02,
-      "learning_rate": 6.284444444444445e-06,
-      "loss": 0.0196,
+      "epoch": 0.43,
+      "learning_rate": 6.286666666666668e-06,
+      "loss": 0.2904,
       "step": 2175
     },
     {
-      "epoch": 5.03,
-      "learning_rate": 6.22888888888889e-06,
-      "loss": 0.0153,
+      "epoch": 0.44,
+      "learning_rate": 6.231111111111111e-06,
+      "loss": 0.2423,
       "step": 2200
     },
     {
-      "epoch": 5.03,
-      "learning_rate": 6.173333333333333e-06,
-      "loss": 0.0132,
+      "epoch": 0.45,
+      "learning_rate": 6.175555555555556e-06,
+      "loss": 0.255,
       "step": 2225
     },
     {
-      "epoch": 5.04,
-      "learning_rate": 6.117777777777778e-06,
-      "loss": 0.0144,
+      "epoch": 0.45,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 0.2142,
       "step": 2250
     },
     {
-      "epoch": 5.04,
-      "learning_rate": 6.062222222222223e-06,
-      "loss": 0.0145,
+      "epoch": 0.46,
+      "learning_rate": 6.064444444444445e-06,
+      "loss": 0.2687,
       "step": 2275
     },
     {
-      "epoch": 5.05,
-      "learning_rate": 6.006666666666667e-06,
-      "loss": 0.0131,
+      "epoch": 0.46,
+      "learning_rate": 6.00888888888889e-06,
+      "loss": 0.2617,
       "step": 2300
     },
     {
-      "epoch": 5.05,
-      "learning_rate": 5.951111111111112e-06,
-      "loss": 0.0107,
+      "epoch": 0.47,
+      "learning_rate": 5.9533333333333345e-06,
+      "loss": 0.2414,
       "step": 2325
     },
     {
-      "epoch": 5.06,
-      "learning_rate": 5.895555555555557e-06,
-      "loss": 0.0122,
+      "epoch": 0.47,
+      "learning_rate": 5.897777777777778e-06,
+      "loss": 0.2048,
       "step": 2350
     },
     {
-      "epoch": 5.06,
-      "learning_rate": 5.84e-06,
-      "loss": 0.0121,
+      "epoch": 0.47,
+      "learning_rate": 5.8422222222222226e-06,
+      "loss": 0.222,
       "step": 2375
     },
     {
-      "epoch": 5.07,
-      "learning_rate": 5.784444444444445e-06,
-      "loss": 0.0103,
+      "epoch": 0.48,
+      "learning_rate": 5.7866666666666674e-06,
+      "loss": 0.2453,
       "step": 2400
     },
     {
-      "epoch": 5.07,
-      "learning_rate": 5.72888888888889e-06,
-      "loss": 0.0089,
+      "epoch": 0.48,
+      "learning_rate": 5.7311111111111115e-06,
+      "loss": 0.2099,
       "step": 2425
     },
     {
-      "epoch": 5.08,
-      "learning_rate": 5.673333333333334e-06,
-      "loss": 0.0087,
+      "epoch": 0.49,
+      "learning_rate": 5.675555555555556e-06,
+      "loss": 0.2515,
       "step": 2450
     },
     {
-      "epoch": 6.0,
-      "learning_rate": 5.617777777777779e-06,
-      "loss": 0.0087,
+      "epoch": 0.49,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 0.2232,
       "step": 2475
     },
     {
-      "epoch": 6.01,
-      "learning_rate": 5.562222222222222e-06,
-      "loss": 0.0106,
+      "epoch": 0.5,
+      "learning_rate": 5.5644444444444444e-06,
+      "loss": 0.1946,
       "step": 2500
     },
     {
-      "epoch": 6.01,
-      "learning_rate": 5.506666666666667e-06,
-      "loss": 0.0094,
+      "epoch": 0.51,
+      "learning_rate": 5.508888888888889e-06,
+      "loss": 0.2176,
       "step": 2525
     },
     {
-      "epoch": 6.02,
-      "learning_rate": 5.451111111111112e-06,
-      "loss": 0.0082,
+      "epoch": 0.51,
+      "learning_rate": 5.453333333333334e-06,
+      "loss": 0.2565,
       "step": 2550
     },
     {
-      "epoch": 6.02,
-      "learning_rate": 5.3955555555555565e-06,
-      "loss": 0.0079,
+      "epoch": 0.52,
+      "learning_rate": 5.397777777777778e-06,
+      "loss": 0.2452,
       "step": 2575
     },
     {
-      "epoch": 6.03,
-      "learning_rate": 5.3400000000000005e-06,
-      "loss": 0.0084,
+      "epoch": 0.52,
+      "learning_rate": 5.342222222222223e-06,
+      "loss": 0.2851,
       "step": 2600
     },
     {
-      "epoch": 6.03,
-      "learning_rate": 5.2844444444444454e-06,
-      "loss": 0.0076,
+      "epoch": 0.53,
+      "learning_rate": 5.286666666666666e-06,
+      "loss": 0.1891,
       "step": 2625
     },
     {
-      "epoch": 6.04,
-      "learning_rate": 5.228888888888889e-06,
-      "loss": 0.0069,
+      "epoch": 0.53,
+      "learning_rate": 5.231111111111111e-06,
+      "loss": 0.2404,
       "step": 2650
     },
     {
-      "epoch": 6.04,
-      "learning_rate": 5.1733333333333335e-06,
-      "loss": 0.0067,
+      "epoch": 0.54,
+      "learning_rate": 5.175555555555556e-06,
+      "loss": 0.2037,
       "step": 2675
     },
     {
-      "epoch": 6.05,
-      "learning_rate": 5.117777777777778e-06,
-      "loss": 0.0073,
+      "epoch": 0.54,
+      "learning_rate": 5.12e-06,
+      "loss": 0.215,
       "step": 2700
     },
     {
-      "epoch": 6.05,
-      "learning_rate": 5.062222222222222e-06,
-      "loss": 0.0062,
+      "epoch": 0.55,
+      "learning_rate": 5.064444444444445e-06,
+      "loss": 0.2115,
       "step": 2725
     },
     {
-      "epoch": 6.06,
-      "learning_rate": 5.006666666666667e-06,
-      "loss": 0.0057,
+      "epoch": 0.55,
+      "learning_rate": 5.00888888888889e-06,
+      "loss": 0.2491,
       "step": 2750
     },
     {
-      "epoch": 6.06,
-      "learning_rate": 4.951111111111111e-06,
-      "loss": 0.0064,
+      "epoch": 0.56,
+      "learning_rate": 4.953333333333334e-06,
+      "loss": 0.1979,
       "step": 2775
     },
     {
-      "epoch": 6.07,
-      "learning_rate": 4.895555555555556e-06,
-      "loss": 0.0053,
+      "epoch": 0.56,
+      "learning_rate": 4.897777777777778e-06,
+      "loss": 0.224,
       "step": 2800
     },
     {
-      "epoch": 6.07,
-      "learning_rate": 4.84e-06,
-      "loss": 0.0047,
+      "epoch": 0.56,
+      "learning_rate": 4.842222222222223e-06,
+      "loss": 0.2065,
       "step": 2825
     },
     {
-      "epoch": 6.08,
-      "learning_rate": 4.784444444444445e-06,
-      "loss": 0.005,
+      "epoch": 0.57,
+      "learning_rate": 4.786666666666667e-06,
+      "loss": 0.2144,
       "step": 2850
     },
     {
-      "epoch": 6.08,
-      "learning_rate": 4.728888888888889e-06,
-      "loss": 0.0047,
+      "epoch": 0.57,
+      "learning_rate": 4.731111111111112e-06,
+      "loss": 0.2141,
       "step": 2875
     },
     {
-      "epoch": 7.0,
-      "learning_rate": 4.673333333333333e-06,
-      "loss": 0.0052,
+      "epoch": 0.58,
+      "learning_rate": 4.675555555555556e-06,
+      "loss": 0.1953,
       "step": 2900
     },
     {
-      "epoch": 7.01,
-      "learning_rate": 4.617777777777778e-06,
-      "loss": 0.005,
+      "epoch": 0.58,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 0.1907,
       "step": 2925
     },
     {
-      "epoch": 7.01,
-      "learning_rate": 4.562222222222222e-06,
-      "loss": 0.0041,
+      "epoch": 0.59,
+      "learning_rate": 4.564444444444445e-06,
+      "loss": 0.2292,
       "step": 2950
     },
     {
-      "epoch": 7.02,
-      "learning_rate": 4.506666666666667e-06,
-      "loss": 0.0045,
+      "epoch": 0.59,
+      "learning_rate": 4.50888888888889e-06,
+      "loss": 0.2164,
       "step": 2975
     },
     {
-      "epoch": 7.02,
-      "learning_rate": 4.451111111111112e-06,
-      "loss": 0.0044,
+      "epoch": 0.6,
+      "learning_rate": 4.453333333333334e-06,
+      "loss": 0.2145,
       "step": 3000
     },
     {
-      "epoch": 7.02,
-      "eval_loss": 0.3602650463581085,
-      "eval_runtime": 960.4855,
-      "eval_samples_per_second": 6.862,
-      "eval_steps_per_second": 0.429,
-      "eval_wer": 18.876827737048888,
+      "epoch": 0.6,
+      "eval_loss": 0.260960191488266,
+      "eval_runtime": 1802.8082,
+      "eval_samples_per_second": 3.656,
+      "eval_steps_per_second": 0.457,
+      "eval_wer": 17.515897768236865,
       "step": 3000
     },
     {
-      "epoch": 7.03,
-      "learning_rate": 4.395555555555556e-06,
-      "loss": 0.0043,
+      "epoch": 0.6,
+      "learning_rate": 4.397777777777778e-06,
+      "loss": 0.1934,
       "step": 3025
     },
     {
-      "epoch": 7.03,
-      "learning_rate": 4.34e-06,
-      "loss": 0.0041,
+      "epoch": 0.61,
+      "learning_rate": 4.3422222222222225e-06,
+      "loss": 0.1973,
       "step": 3050
     },
     {
-      "epoch": 7.04,
-      "learning_rate": 4.284444444444445e-06,
-      "loss": 0.0035,
+      "epoch": 0.61,
+      "learning_rate": 4.2866666666666666e-06,
+      "loss": 0.1767,
       "step": 3075
     },
     {
-      "epoch": 7.04,
-      "learning_rate": 4.228888888888889e-06,
-      "loss": 0.0044,
+      "epoch": 0.62,
+      "learning_rate": 4.2311111111111114e-06,
+      "loss": 0.1918,
       "step": 3100
     },
     {
-      "epoch": 7.05,
-      "learning_rate": 4.173333333333334e-06,
-      "loss": 0.0039,
+      "epoch": 0.62,
+      "learning_rate": 4.175555555555556e-06,
+      "loss": 0.1946,
       "step": 3125
     },
     {
-      "epoch": 7.05,
-      "learning_rate": 4.117777777777779e-06,
-      "loss": 0.0031,
+      "epoch": 0.63,
+      "learning_rate": 4.12e-06,
+      "loss": 0.1897,
       "step": 3150
     },
     {
-      "epoch": 7.06,
-      "learning_rate": 4.062222222222223e-06,
-      "loss": 0.0032,
+      "epoch": 0.64,
+      "learning_rate": 4.064444444444444e-06,
+      "loss": 0.2185,
       "step": 3175
     },
     {
-      "epoch": 7.06,
-      "learning_rate": 4.006666666666667e-06,
-      "loss": 0.0034,
+      "epoch": 0.64,
+      "learning_rate": 4.008888888888889e-06,
+      "loss": 0.1954,
       "step": 3200
     },
     {
-      "epoch": 7.07,
-      "learning_rate": 3.951111111111112e-06,
-      "loss": 0.0028,
+      "epoch": 0.65,
+      "learning_rate": 3.953333333333333e-06,
+      "loss": 0.2318,
       "step": 3225
     },
     {
-      "epoch": 7.07,
-      "learning_rate": 3.895555555555556e-06,
-      "loss": 0.0032,
+      "epoch": 0.65,
+      "learning_rate": 3.897777777777778e-06,
+      "loss": 0.2615,
       "step": 3250
     },
     {
-      "epoch": 7.08,
-      "learning_rate": 3.8400000000000005e-06,
-      "loss": 0.003,
+      "epoch": 0.66,
+      "learning_rate": 3.842222222222223e-06,
+      "loss": 0.1846,
       "step": 3275
     },
     {
-      "epoch": 8.0,
-      "learning_rate": 3.784444444444445e-06,
-      "loss": 0.0034,
+      "epoch": 1.0,
+      "learning_rate": 3.7866666666666667e-06,
+      "loss": 0.222,
       "step": 3300
     },
     {
-      "epoch": 8.01,
-      "learning_rate": 3.728888888888889e-06,
-      "loss": 0.003,
+      "epoch": 1.01,
+      "learning_rate": 3.7311111111111116e-06,
+      "loss": 0.2224,
       "step": 3325
     },
     {
-      "epoch": 8.01,
-      "learning_rate": 3.673333333333334e-06,
-      "loss": 0.0027,
+      "epoch": 1.01,
+      "learning_rate": 3.675555555555556e-06,
+      "loss": 0.2128,
       "step": 3350
     },
     {
-      "epoch": 8.02,
-      "learning_rate": 3.617777777777778e-06,
-      "loss": 0.0026,
+      "epoch": 1.02,
+      "learning_rate": 3.62e-06,
+      "loss": 0.2002,
       "step": 3375
     },
     {
-      "epoch": 8.02,
-      "learning_rate": 3.5622222222222224e-06,
-      "loss": 0.0028,
+      "epoch": 1.02,
+      "learning_rate": 3.564444444444445e-06,
+      "loss": 0.1861,
       "step": 3400
     },
     {
-      "epoch": 8.03,
-      "learning_rate": 3.5066666666666673e-06,
-      "loss": 0.0029,
+      "epoch": 1.03,
+      "learning_rate": 3.508888888888889e-06,
+      "loss": 0.176,
       "step": 3425
     },
     {
-      "epoch": 8.03,
-      "learning_rate": 3.4511111111111113e-06,
-      "loss": 0.0027,
+      "epoch": 1.03,
+      "learning_rate": 3.4533333333333334e-06,
+      "loss": 0.1659,
       "step": 3450
     },
     {
-      "epoch": 8.04,
-      "learning_rate": 3.3955555555555558e-06,
-      "loss": 0.0022,
+      "epoch": 1.04,
+      "learning_rate": 3.3977777777777783e-06,
+      "loss": 0.1545,
       "step": 3475
     },
     {
-      "epoch": 8.04,
-      "learning_rate": 3.3400000000000006e-06,
-      "loss": 0.0024,
+      "epoch": 1.04,
+      "learning_rate": 3.3422222222222224e-06,
+      "loss": 0.1314,
       "step": 3500
     },
     {
-      "epoch": 8.05,
-      "learning_rate": 3.2844444444444447e-06,
-      "loss": 0.0028,
+      "epoch": 1.05,
+      "learning_rate": 3.286666666666667e-06,
+      "loss": 0.1573,
       "step": 3525
     },
     {
-      "epoch": 8.05,
-      "learning_rate": 3.228888888888889e-06,
-      "loss": 0.0025,
+      "epoch": 1.05,
+      "learning_rate": 3.2311111111111117e-06,
+      "loss": 0.1696,
       "step": 3550
     },
     {
-      "epoch": 8.06,
-      "learning_rate": 3.173333333333334e-06,
-      "loss": 0.0022,
+      "epoch": 1.06,
+      "learning_rate": 3.1755555555555557e-06,
+      "loss": 0.1348,
       "step": 3575
     },
     {
-      "epoch": 8.06,
-      "learning_rate": 3.117777777777778e-06,
-      "loss": 0.0021,
+      "epoch": 1.06,
+      "learning_rate": 3.12e-06,
+      "loss": 0.1477,
       "step": 3600
     },
     {
-      "epoch": 8.07,
-      "learning_rate": 3.0622222222222225e-06,
-      "loss": 0.002,
+      "epoch": 1.07,
+      "learning_rate": 3.064444444444445e-06,
+      "loss": 0.1464,
       "step": 3625
     },
     {
-      "epoch": 8.07,
-      "learning_rate": 3.0066666666666674e-06,
-      "loss": 0.0019,
+      "epoch": 1.07,
+      "learning_rate": 3.008888888888889e-06,
+      "loss": 0.1027,
       "step": 3650
     },
     {
-      "epoch": 8.08,
-      "learning_rate": 2.9511111111111114e-06,
-      "loss": 0.0019,
+      "epoch": 1.08,
+      "learning_rate": 2.9533333333333336e-06,
+      "loss": 0.1032,
       "step": 3675
     },
     {
-      "epoch": 8.08,
-      "learning_rate": 2.895555555555556e-06,
-      "loss": 0.0019,
+      "epoch": 1.08,
+      "learning_rate": 2.8977777777777785e-06,
+      "loss": 0.0937,
       "step": 3700
     },
     {
-      "epoch": 9.0,
-      "learning_rate": 2.84e-06,
-      "loss": 0.0021,
+      "epoch": 1.09,
+      "learning_rate": 2.8422222222222225e-06,
+      "loss": 0.0975,
       "step": 3725
     },
     {
-      "epoch": 9.01,
-      "learning_rate": 2.784444444444445e-06,
-      "loss": 0.0021,
+      "epoch": 1.09,
+      "learning_rate": 2.786666666666667e-06,
+      "loss": 0.0922,
       "step": 3750
     },
     {
-      "epoch": 9.01,
-      "learning_rate": 2.7288888888888893e-06,
-      "loss": 0.0018,
+      "epoch": 1.1,
+      "learning_rate": 2.7311111111111114e-06,
+      "loss": 0.1045,
       "step": 3775
     },
     {
-      "epoch": 9.02,
-      "learning_rate": 2.6733333333333333e-06,
-      "loss": 0.0019,
+      "epoch": 1.1,
+      "learning_rate": 2.675555555555556e-06,
+      "loss": 0.0816,
       "step": 3800
     },
     {
-      "epoch": 9.02,
-      "learning_rate": 2.617777777777778e-06,
-      "loss": 0.0018,
+      "epoch": 1.11,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 0.1011,
       "step": 3825
     },
     {
-      "epoch": 9.03,
-      "learning_rate": 2.5622222222222226e-06,
-      "loss": 0.0021,
+      "epoch": 1.11,
+      "learning_rate": 2.5644444444444444e-06,
+      "loss": 0.0883,
       "step": 3850
     },
     {
-      "epoch": 9.03,
-      "learning_rate": 2.5066666666666667e-06,
-      "loss": 0.0017,
+      "epoch": 1.12,
+      "learning_rate": 2.5088888888888892e-06,
+      "loss": 0.0872,
       "step": 3875
     },
     {
-      "epoch": 9.04,
-      "learning_rate": 2.451111111111111e-06,
-      "loss": 0.0018,
+      "epoch": 1.12,
+      "learning_rate": 2.4533333333333333e-06,
+      "loss": 0.1131,
       "step": 3900
     },
     {
-      "epoch": 9.04,
-      "learning_rate": 2.3955555555555556e-06,
-      "loss": 0.0017,
+      "epoch": 1.13,
+      "learning_rate": 2.397777777777778e-06,
+      "loss": 0.1031,
       "step": 3925
     },
     {
-      "epoch": 9.05,
-      "learning_rate": 2.3400000000000005e-06,
-      "loss": 0.002,
+      "epoch": 1.13,
+      "learning_rate": 2.342222222222222e-06,
+      "loss": 0.0985,
       "step": 3950
     },
     {
-      "epoch": 9.05,
-      "learning_rate": 2.2844444444444445e-06,
-      "loss": 0.0017,
+      "epoch": 1.14,
+      "learning_rate": 2.2866666666666667e-06,
+      "loss": 0.1057,
       "step": 3975
     },
     {
-      "epoch": 9.06,
-      "learning_rate": 2.228888888888889e-06,
-      "loss": 0.0016,
+      "epoch": 1.14,
+      "learning_rate": 2.2311111111111115e-06,
+      "loss": 0.0828,
       "step": 4000
     },
     {
-      "epoch": 9.06,
-      "eval_loss": 0.3812163174152374,
-      "eval_runtime": 968.0397,
-      "eval_samples_per_second": 6.809,
-      "eval_steps_per_second": 0.426,
-      "eval_wer": 18.775568066750374,
+      "epoch": 1.14,
+      "eval_loss": 0.23880085349082947,
+      "eval_runtime": 1805.0898,
+      "eval_samples_per_second": 3.651,
+      "eval_steps_per_second": 0.456,
+      "eval_wer": 15.300336182105392,
       "step": 4000
     },
     {
-      "epoch": 9.06,
-      "learning_rate": 2.1733333333333334e-06,
-      "loss": 0.0016,
+      "epoch": 1.15,
+      "learning_rate": 2.1755555555555556e-06,
+      "loss": 0.0873,
       "step": 4025
     },
     {
-      "epoch": 9.07,
-      "learning_rate": 2.117777777777778e-06,
-      "loss": 0.0015,
+      "epoch": 1.15,
+      "learning_rate": 2.12e-06,
+      "loss": 0.0848,
       "step": 4050
     },
     {
-      "epoch": 9.07,
-      "learning_rate": 2.0622222222222223e-06,
-      "loss": 0.0015,
+      "epoch": 1.16,
+      "learning_rate": 2.064444444444445e-06,
+      "loss": 0.0936,
       "step": 4075
     },
     {
-      "epoch": 9.08,
-      "learning_rate": 2.006666666666667e-06,
-      "loss": 0.0015,
+      "epoch": 1.16,
+      "learning_rate": 2.008888888888889e-06,
+      "loss": 0.0965,
       "step": 4100
     },
     {
-      "epoch": 10.0,
-      "learning_rate": 1.9511111111111113e-06,
-      "loss": 0.0015,
+      "epoch": 1.17,
+      "learning_rate": 1.9533333333333334e-06,
+      "loss": 0.0923,
       "step": 4125
     },
     {
-      "epoch": 10.01,
-      "learning_rate": 1.8955555555555557e-06,
-      "loss": 0.0016,
+      "epoch": 1.17,
+      "learning_rate": 1.8977777777777779e-06,
+      "loss": 0.0793,
       "step": 4150
     },
     {
-      "epoch": 10.01,
-      "learning_rate": 1.8400000000000002e-06,
-      "loss": 0.0015,
+      "epoch": 1.18,
+      "learning_rate": 1.8422222222222225e-06,
+      "loss": 0.0848,
       "step": 4175
     },
     {
-      "epoch": 10.02,
-      "learning_rate": 1.7844444444444444e-06,
-      "loss": 0.0015,
+      "epoch": 1.18,
+      "learning_rate": 1.7866666666666668e-06,
+      "loss": 0.0956,
       "step": 4200
     },
     {
-      "epoch": 10.02,
-      "learning_rate": 1.728888888888889e-06,
-      "loss": 0.0014,
+      "epoch": 1.19,
+      "learning_rate": 1.7311111111111112e-06,
+      "loss": 0.0814,
       "step": 4225
     },
     {
-      "epoch": 10.03,
-      "learning_rate": 1.6733333333333335e-06,
-      "loss": 0.0016,
+      "epoch": 1.19,
+      "learning_rate": 1.675555555555556e-06,
+      "loss": 0.1086,
       "step": 4250
     },
     {
-      "epoch": 10.03,
-      "learning_rate": 1.6177777777777778e-06,
-      "loss": 0.0015,
+      "epoch": 1.2,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 0.1057,
       "step": 4275
     },
     {
-      "epoch": 10.04,
-      "learning_rate": 1.5622222222222225e-06,
-      "loss": 0.0015,
+      "epoch": 1.2,
+      "learning_rate": 1.5644444444444446e-06,
+      "loss": 0.091,
       "step": 4300
     },
     {
-      "epoch": 10.04,
-      "learning_rate": 1.506666666666667e-06,
-      "loss": 0.0016,
+      "epoch": 1.21,
+      "learning_rate": 1.5088888888888889e-06,
+      "loss": 0.0857,
       "step": 4325
     },
     {
-      "epoch": 10.05,
-      "learning_rate": 1.4511111111111112e-06,
-      "loss": 0.0015,
+      "epoch": 1.21,
+      "learning_rate": 1.4533333333333335e-06,
+      "loss": 0.0904,
       "step": 4350
     },
     {
-      "epoch": 10.05,
-      "learning_rate": 1.3955555555555556e-06,
-      "loss": 0.0014,
+      "epoch": 1.22,
+      "learning_rate": 1.397777777777778e-06,
+      "loss": 0.0714,
       "step": 4375
     },
     {
-      "epoch": 10.06,
-      "learning_rate": 1.34e-06,
-      "loss": 0.0014,
+      "epoch": 1.22,
+      "learning_rate": 1.3422222222222222e-06,
+      "loss": 0.071,
       "step": 4400
     },
     {
-      "epoch": 10.06,
-      "learning_rate": 1.2844444444444445e-06,
-      "loss": 0.0013,
+      "epoch": 1.23,
+      "learning_rate": 1.286666666666667e-06,
+      "loss": 0.0689,
       "step": 4425
     },
     {
-      "epoch": 10.07,
-      "learning_rate": 1.228888888888889e-06,
-      "loss": 0.0013,
+      "epoch": 1.23,
+      "learning_rate": 1.2311111111111112e-06,
+      "loss": 0.0597,
       "step": 4450
     },
     {
-      "epoch": 10.07,
-      "learning_rate": 1.1733333333333335e-06,
-      "loss": 0.0013,
+      "epoch": 1.24,
+      "learning_rate": 1.1755555555555556e-06,
+      "loss": 0.0737,
       "step": 4475
     },
     {
-      "epoch": 10.08,
-      "learning_rate": 1.117777777777778e-06,
-      "loss": 0.0013,
+      "epoch": 1.24,
+      "learning_rate": 1.12e-06,
+      "loss": 0.0712,
       "step": 4500
     },
     {
-      "epoch": 10.08,
-      "learning_rate": 1.0622222222222222e-06,
-      "loss": 0.0013,
+      "epoch": 1.25,
+      "learning_rate": 1.0644444444444445e-06,
+      "loss": 0.0641,
       "step": 4525
     },
     {
-      "epoch": 11.0,
-      "learning_rate": 1.0066666666666668e-06,
-      "loss": 0.0014,
+      "epoch": 1.25,
+      "learning_rate": 1.008888888888889e-06,
+      "loss": 0.0605,
       "step": 4550
     },
     {
-      "epoch": 11.01,
-      "learning_rate": 9.511111111111111e-07,
-      "loss": 0.0014,
+      "epoch": 1.26,
+      "learning_rate": 9.533333333333335e-07,
+      "loss": 0.0981,
       "step": 4575
     },
     {
-      "epoch": 11.01,
-      "learning_rate": 8.955555555555557e-07,
-      "loss": 0.0013,
+      "epoch": 1.26,
+      "learning_rate": 8.977777777777778e-07,
+      "loss": 0.0751,
       "step": 4600
     },
     {
-      "epoch": 11.02,
-      "learning_rate": 8.400000000000001e-07,
-      "loss": 0.0013,
+      "epoch": 1.27,
+      "learning_rate": 8.422222222222224e-07,
+      "loss": 0.0645,
       "step": 4625
     },
     {
-      "epoch": 11.02,
-      "learning_rate": 7.844444444444445e-07,
-      "loss": 0.0014,
+      "epoch": 1.27,
+      "learning_rate": 7.866666666666667e-07,
+      "loss": 0.0573,
       "step": 4650
     },
     {
-      "epoch": 11.03,
-      "learning_rate": 7.28888888888889e-07,
-      "loss": 0.0015,
+      "epoch": 1.28,
+      "learning_rate": 7.311111111111112e-07,
+      "loss": 0.0672,
       "step": 4675
     },
     {
-      "epoch": 11.03,
-      "learning_rate": 6.733333333333334e-07,
-      "loss": 0.0013,
+      "epoch": 1.28,
+      "learning_rate": 6.755555555555555e-07,
+      "loss": 0.0891,
       "step": 4700
     },
     {
-      "epoch": 11.04,
-      "learning_rate": 6.177777777777778e-07,
-      "loss": 0.0013,
+      "epoch": 1.29,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 0.0605,
       "step": 4725
     },
     {
-      "epoch": 11.04,
-      "learning_rate": 5.622222222222223e-07,
-      "loss": 0.0013,
+      "epoch": 1.29,
+      "learning_rate": 5.644444444444445e-07,
+      "loss": 0.0686,
       "step": 4750
     },
     {
-      "epoch": 11.05,
-      "learning_rate": 5.066666666666667e-07,
-      "loss": 0.0013,
+      "epoch": 1.3,
+      "learning_rate": 5.088888888888889e-07,
+      "loss": 0.0708,
       "step": 4775
     },
     {
-      "epoch": 11.05,
-      "learning_rate": 4.511111111111111e-07,
-      "loss": 0.0013,
+      "epoch": 1.3,
+      "learning_rate": 4.533333333333334e-07,
+      "loss": 0.0594,
       "step": 4800
     },
     {
-      "epoch": 11.06,
-      "learning_rate": 3.9555555555555557e-07,
-      "loss": 0.0013,
+      "epoch": 1.31,
+      "learning_rate": 3.9777777777777783e-07,
+      "loss": 0.0615,
       "step": 4825
     },
     {
-      "epoch": 11.06,
-      "learning_rate": 3.4000000000000003e-07,
-      "loss": 0.0013,
+      "epoch": 1.31,
+      "learning_rate": 3.422222222222223e-07,
+      "loss": 0.0574,
       "step": 4850
     },
     {
-      "epoch": 11.07,
-      "learning_rate": 2.844444444444445e-07,
-      "loss": 0.0013,
+      "epoch": 1.32,
+      "learning_rate": 2.866666666666667e-07,
+      "loss": 0.0772,
       "step": 4875
     },
     {
-      "epoch": 11.07,
-      "learning_rate": 2.2888888888888892e-07,
-      "loss": 0.0012,
+      "epoch": 1.32,
+      "learning_rate": 2.3111111111111112e-07,
+      "loss": 0.0713,
       "step": 4900
     },
     {
-      "epoch": 11.08,
-      "learning_rate": 1.7333333333333335e-07,
-      "loss": 0.0012,
+      "epoch": 1.33,
+      "learning_rate": 1.7555555555555558e-07,
+      "loss": 0.0634,
       "step": 4925
     },
     {
-      "epoch": 12.0,
-      "learning_rate": 1.1777777777777778e-07,
-      "loss": 0.0013,
+      "epoch": 1.33,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 0.078,
       "step": 4950
     },
     {
-      "epoch": 12.01,
-      "learning_rate": 6.222222222222223e-08,
-      "loss": 0.0013,
+      "epoch": 1.34,
+      "learning_rate": 6.444444444444445e-08,
+      "loss": 0.0827,
       "step": 4975
     },
     {
-      "epoch": 12.01,
-      "learning_rate": 6.666666666666667e-09,
-      "loss": 0.0012,
+      "epoch": 1.34,
+      "learning_rate": 8.88888888888889e-09,
+      "loss": 0.0729,
       "step": 5000
     },
     {
-      "epoch": 12.01,
-      "eval_loss": 0.3913772702217102,
-      "eval_runtime": 973.2149,
-      "eval_samples_per_second": 6.772,
-      "eval_steps_per_second": 0.423,
-      "eval_wer": 18.83024828871157,
+      "epoch": 1.34,
+      "eval_loss": 0.2256375402212143,
+      "eval_runtime": 1809.5302,
+      "eval_samples_per_second": 3.642,
+      "eval_steps_per_second": 0.455,
+      "eval_wer": 13.996111628660538,
       "step": 5000
     },
     {
-      "epoch": 12.01,
+      "epoch": 1.34,
       "step": 5000,
-      "total_flos": 4.616673800159232e+19,
-      "train_loss": 0.106446673027426,
-      "train_runtime": 27653.1068,
-      "train_samples_per_second": 5.786,
-      "train_steps_per_second": 0.181
+      "total_flos": 2.041005694058496e+19,
+      "train_loss": 0.2607154120922089,
+      "train_runtime": 15779.8251,
+      "train_samples_per_second": 1.267,
+      "train_steps_per_second": 0.317
     }
   ],
   "max_steps": 5000,
   "num_train_epochs": 9223372036854775807,
-  "total_flos": 4.616673800159232e+19,
+  "total_flos": 2.041005694058496e+19,
   "trial_name": null,
   "trial_params": null
 }