Add results

all_results.json

@@ -1,12 +1,12 @@
 {
-    "epoch": 12.01,
+    "epoch": 6.06,
     "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
+    "train_loss": 0.15837200704216958,
+    "train_runtime": 29005.3522,
+    "train_samples_per_second": 5.516,
+    "train_steps_per_second": 0.172
 }

ggml-small.eu.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2e35b59747aa9d0976001fe99c33ff3f422c3af984ee4440b5c1392c282b6d51
+oid sha256:939aa258c716f2fa50d1b18d3a649fcaac19fbfc865df87e09303b0e4383ef73
 size 487601984

nohup.out

@@ -70598,3 +70598,21 @@ The progress bars may be unreliable.
 
 
 
+To https://huggingface.co/xezpeleta/whisper-small-eu
+   df538ec..b382d42  main -> main
+
+12/29/2023 22:56:29 - WARNING - huggingface_hub.repository - To https://huggingface.co/xezpeleta/whisper-small-eu
+   df538ec..b382d42  main -> main
+
+***** train metrics *****
+  epoch                    =       6.06
+  train_loss               =     0.1584
+  train_runtime            = 8:03:25.35
+  train_samples_per_second =      5.516
+  train_steps_per_second   =      0.172
+12/29/2023 22:56:32 - INFO - __main__ - *** Evaluate ***
+[INFO|trainer.py:2955] 2023-12-29 22:56:32,446 >> ***** Running Evaluation *****
+[INFO|trainer.py:2959] 2023-12-29 22:56:32,446 >>   Num examples: Unknown
+[INFO|trainer.py:2960] 2023-12-29 22:56:32,446 >>   Batch size = 16
+
+[INFO|trainer_utils.py:689] 2023-12-29 22:56:36,997 >> 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.

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": 6.06,
+    "train_loss": 0.15837200704216958,
+    "train_runtime": 29005.3522,
+    "train_samples_per_second": 5.516,
+    "train_steps_per_second": 0.172
 }

trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 18.775568066750374,
-  "best_model_checkpoint": "./checkpoint-4000",
-  "epoch": 12.0112,
+  "best_metric": 12.012786552211754,
+  "best_model_checkpoint": "./checkpoint-5000",
+  "epoch": 6.0604,
   "global_step": 5000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
@@ -10,1261 +10,1261 @@
     {
       "epoch": 0.01,
       "learning_rate": 4.4e-07,
-      "loss": 1.6639,
+      "loss": 1.6183,
       "step": 25
     },
     {
       "epoch": 0.01,
       "learning_rate": 9.400000000000001e-07,
-      "loss": 1.4728,
+      "loss": 1.4856,
       "step": 50
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.44e-06,
-      "loss": 1.1563,
+      "loss": 1.1898,
       "step": 75
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.94e-06,
-      "loss": 0.9368,
+      "loss": 0.9354,
       "step": 100
     },
     {
       "epoch": 0.03,
       "learning_rate": 2.4400000000000004e-06,
-      "loss": 0.8889,
+      "loss": 0.8172,
       "step": 125
     },
     {
       "epoch": 0.03,
       "learning_rate": 2.9400000000000002e-06,
-      "loss": 0.8114,
+      "loss": 0.7386,
       "step": 150
     },
     {
       "epoch": 0.04,
       "learning_rate": 3.44e-06,
-      "loss": 0.6924,
+      "loss": 0.6768,
       "step": 175
     },
     {
       "epoch": 0.04,
       "learning_rate": 3.94e-06,
-      "loss": 0.661,
+      "loss": 0.6069,
       "step": 200
     },
     {
       "epoch": 0.04,
       "learning_rate": 4.440000000000001e-06,
-      "loss": 0.6427,
+      "loss": 0.5739,
       "step": 225
     },
     {
       "epoch": 0.05,
       "learning_rate": 4.94e-06,
-      "loss": 0.5217,
+      "loss": 0.5269,
       "step": 250
     },
     {
       "epoch": 0.06,
       "learning_rate": 5.4400000000000004e-06,
-      "loss": 0.4991,
+      "loss": 0.5011,
       "step": 275
     },
     {
       "epoch": 0.06,
       "learning_rate": 5.94e-06,
-      "loss": 0.4768,
+      "loss": 0.4842,
       "step": 300
     },
     {
       "epoch": 0.07,
       "learning_rate": 6.440000000000001e-06,
-      "loss": 0.4632,
+      "loss": 0.4698,
       "step": 325
     },
     {
       "epoch": 0.07,
       "learning_rate": 6.9400000000000005e-06,
-      "loss": 0.4289,
+      "loss": 0.446,
       "step": 350
     },
     {
       "epoch": 0.07,
       "learning_rate": 7.440000000000001e-06,
-      "loss": 0.404,
+      "loss": 0.4378,
       "step": 375
     },
     {
       "epoch": 0.08,
       "learning_rate": 7.94e-06,
-      "loss": 0.3833,
+      "loss": 0.3966,
       "step": 400
     },
     {
-      "epoch": 1.0,
+      "epoch": 0.09,
       "learning_rate": 8.44e-06,
-      "loss": 0.4201,
+      "loss": 0.3916,
       "step": 425
     },
     {
-      "epoch": 1.01,
+      "epoch": 0.09,
       "learning_rate": 8.94e-06,
-      "loss": 0.3989,
+      "loss": 0.3746,
       "step": 450
     },
     {
-      "epoch": 1.01,
+      "epoch": 0.1,
       "learning_rate": 9.440000000000001e-06,
-      "loss": 0.3453,
+      "loss": 0.3372,
       "step": 475
     },
     {
-      "epoch": 1.02,
+      "epoch": 0.1,
       "learning_rate": 9.940000000000001e-06,
-      "loss": 0.3559,
+      "loss": 0.329,
       "step": 500
     },
     {
-      "epoch": 1.02,
+      "epoch": 0.1,
       "learning_rate": 9.951111111111111e-06,
-      "loss": 0.3429,
+      "loss": 0.3364,
       "step": 525
     },
     {
-      "epoch": 1.03,
+      "epoch": 0.11,
       "learning_rate": 9.895555555555557e-06,
-      "loss": 0.3701,
+      "loss": 0.3074,
       "step": 550
     },
     {
-      "epoch": 1.03,
+      "epoch": 0.12,
       "learning_rate": 9.84e-06,
-      "loss": 0.3084,
+      "loss": 0.3134,
       "step": 575
     },
     {
-      "epoch": 1.04,
+      "epoch": 0.12,
       "learning_rate": 9.784444444444445e-06,
-      "loss": 0.2956,
+      "loss": 0.348,
       "step": 600
     },
     {
-      "epoch": 1.04,
+      "epoch": 0.12,
       "learning_rate": 9.72888888888889e-06,
-      "loss": 0.2923,
+      "loss": 0.3037,
       "step": 625
     },
     {
-      "epoch": 1.05,
+      "epoch": 0.13,
       "learning_rate": 9.673333333333334e-06,
-      "loss": 0.2786,
+      "loss": 0.2768,
       "step": 650
     },
     {
-      "epoch": 1.05,
+      "epoch": 0.14,
       "learning_rate": 9.617777777777778e-06,
-      "loss": 0.2427,
+      "loss": 0.3341,
       "step": 675
     },
     {
-      "epoch": 1.06,
+      "epoch": 0.14,
       "learning_rate": 9.562222222222223e-06,
-      "loss": 0.2387,
+      "loss": 0.3203,
       "step": 700
     },
     {
-      "epoch": 1.06,
+      "epoch": 0.14,
       "learning_rate": 9.506666666666667e-06,
-      "loss": 0.2371,
+      "loss": 0.2681,
       "step": 725
     },
     {
-      "epoch": 1.07,
+      "epoch": 0.15,
       "learning_rate": 9.451111111111112e-06,
-      "loss": 0.2114,
+      "loss": 0.2906,
       "step": 750
     },
     {
-      "epoch": 1.07,
+      "epoch": 0.15,
       "learning_rate": 9.395555555555556e-06,
-      "loss": 0.1894,
+      "loss": 0.2918,
       "step": 775
     },
     {
-      "epoch": 1.08,
+      "epoch": 1.0,
       "learning_rate": 9.340000000000002e-06,
-      "loss": 0.1867,
+      "loss": 0.2697,
       "step": 800
     },
     {
-      "epoch": 2.0,
+      "epoch": 1.01,
       "learning_rate": 9.284444444444444e-06,
-      "loss": 0.1927,
+      "loss": 0.2868,
       "step": 825
     },
     {
-      "epoch": 2.01,
+      "epoch": 1.01,
       "learning_rate": 9.22888888888889e-06,
-      "loss": 0.1951,
+      "loss": 0.2654,
       "step": 850
     },
     {
-      "epoch": 2.01,
+      "epoch": 1.02,
       "learning_rate": 9.173333333333334e-06,
-      "loss": 0.1775,
+      "loss": 0.2549,
       "step": 875
     },
     {
-      "epoch": 2.02,
+      "epoch": 1.02,
       "learning_rate": 9.117777777777778e-06,
-      "loss": 0.1672,
+      "loss": 0.2508,
       "step": 900
     },
     {
-      "epoch": 2.02,
+      "epoch": 1.03,
       "learning_rate": 9.062222222222224e-06,
-      "loss": 0.1591,
+      "loss": 0.2285,
       "step": 925
     },
     {
-      "epoch": 2.03,
+      "epoch": 1.03,
       "learning_rate": 9.006666666666666e-06,
-      "loss": 0.1703,
+      "loss": 0.2362,
       "step": 950
     },
     {
-      "epoch": 2.03,
+      "epoch": 1.04,
       "learning_rate": 8.951111111111112e-06,
-      "loss": 0.1636,
+      "loss": 0.2239,
       "step": 975
     },
     {
-      "epoch": 2.04,
+      "epoch": 1.04,
       "learning_rate": 8.895555555555556e-06,
-      "loss": 0.1413,
+      "loss": 0.2009,
       "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": 1.04,
+      "eval_loss": 0.24461345374584198,
+      "eval_runtime": 1705.6758,
+      "eval_samples_per_second": 7.337,
+      "eval_steps_per_second": 0.459,
+      "eval_wer": 17.688063892927563,
       "step": 1000
     },
     {
-      "epoch": 2.04,
+      "epoch": 1.05,
       "learning_rate": 8.84e-06,
-      "loss": 0.1371,
+      "loss": 0.196,
       "step": 1025
     },
     {
-      "epoch": 2.05,
+      "epoch": 1.05,
       "learning_rate": 8.784444444444446e-06,
-      "loss": 0.1494,
+      "loss": 0.2057,
       "step": 1050
     },
     {
-      "epoch": 2.05,
+      "epoch": 1.06,
       "learning_rate": 8.72888888888889e-06,
-      "loss": 0.1196,
+      "loss": 0.2115,
       "step": 1075
     },
     {
-      "epoch": 2.06,
+      "epoch": 1.06,
       "learning_rate": 8.673333333333334e-06,
-      "loss": 0.1152,
+      "loss": 0.1891,
       "step": 1100
     },
     {
-      "epoch": 2.06,
+      "epoch": 1.07,
       "learning_rate": 8.617777777777778e-06,
-      "loss": 0.1158,
+      "loss": 0.1985,
       "step": 1125
     },
     {
-      "epoch": 2.07,
+      "epoch": 1.07,
       "learning_rate": 8.562222222222224e-06,
-      "loss": 0.1054,
+      "loss": 0.184,
       "step": 1150
     },
     {
-      "epoch": 2.07,
+      "epoch": 1.08,
       "learning_rate": 8.506666666666668e-06,
-      "loss": 0.0912,
+      "loss": 0.1581,
       "step": 1175
     },
     {
-      "epoch": 2.08,
+      "epoch": 1.08,
       "learning_rate": 8.451111111111112e-06,
-      "loss": 0.0915,
+      "loss": 0.1609,
       "step": 1200
     },
     {
-      "epoch": 2.08,
+      "epoch": 1.09,
       "learning_rate": 8.395555555555557e-06,
-      "loss": 0.0901,
+      "loss": 0.1528,
       "step": 1225
     },
     {
-      "epoch": 3.0,
+      "epoch": 1.09,
       "learning_rate": 8.34e-06,
-      "loss": 0.0953,
+      "loss": 0.1387,
       "step": 1250
     },
     {
-      "epoch": 3.01,
+      "epoch": 1.1,
       "learning_rate": 8.284444444444446e-06,
-      "loss": 0.0923,
+      "loss": 0.1312,
       "step": 1275
     },
     {
-      "epoch": 3.01,
+      "epoch": 1.1,
       "learning_rate": 8.22888888888889e-06,
-      "loss": 0.0749,
+      "loss": 0.1471,
       "step": 1300
     },
     {
-      "epoch": 3.02,
+      "epoch": 1.11,
       "learning_rate": 8.173333333333334e-06,
-      "loss": 0.0796,
+      "loss": 0.1208,
       "step": 1325
     },
     {
-      "epoch": 3.02,
+      "epoch": 1.11,
       "learning_rate": 8.11777777777778e-06,
-      "loss": 0.0786,
+      "loss": 0.1387,
       "step": 1350
     },
     {
-      "epoch": 3.03,
+      "epoch": 1.12,
       "learning_rate": 8.062222222222222e-06,
-      "loss": 0.0819,
+      "loss": 0.1393,
       "step": 1375
     },
     {
-      "epoch": 3.03,
+      "epoch": 1.12,
       "learning_rate": 8.006666666666667e-06,
-      "loss": 0.0747,
+      "loss": 0.1347,
       "step": 1400
     },
     {
-      "epoch": 3.04,
+      "epoch": 1.13,
       "learning_rate": 7.951111111111111e-06,
-      "loss": 0.0646,
+      "loss": 0.1278,
       "step": 1425
     },
     {
-      "epoch": 3.04,
+      "epoch": 1.13,
       "learning_rate": 7.895555555555557e-06,
-      "loss": 0.0707,
+      "loss": 0.1528,
       "step": 1450
     },
     {
-      "epoch": 3.05,
+      "epoch": 1.14,
       "learning_rate": 7.840000000000001e-06,
-      "loss": 0.0639,
+      "loss": 0.1614,
       "step": 1475
     },
     {
-      "epoch": 3.05,
+      "epoch": 1.14,
       "learning_rate": 7.784444444444445e-06,
-      "loss": 0.0544,
+      "loss": 0.1255,
       "step": 1500
     },
     {
-      "epoch": 3.06,
+      "epoch": 1.15,
       "learning_rate": 7.72888888888889e-06,
-      "loss": 0.0531,
+      "loss": 0.138,
       "step": 1525
     },
     {
-      "epoch": 3.06,
+      "epoch": 1.15,
       "learning_rate": 7.673333333333333e-06,
-      "loss": 0.0518,
+      "loss": 0.1341,
       "step": 1550
     },
     {
-      "epoch": 3.07,
+      "epoch": 2.0,
       "learning_rate": 7.617777777777778e-06,
-      "loss": 0.0461,
+      "loss": 0.1223,
       "step": 1575
     },
     {
-      "epoch": 3.07,
+      "epoch": 2.01,
       "learning_rate": 7.562222222222223e-06,
-      "loss": 0.0424,
+      "loss": 0.1272,
       "step": 1600
     },
     {
-      "epoch": 3.08,
+      "epoch": 2.01,
       "learning_rate": 7.506666666666668e-06,
-      "loss": 0.0392,
+      "loss": 0.1324,
       "step": 1625
     },
     {
-      "epoch": 4.0,
+      "epoch": 2.02,
       "learning_rate": 7.451111111111111e-06,
-      "loss": 0.0376,
+      "loss": 0.1277,
       "step": 1650
     },
     {
-      "epoch": 4.01,
+      "epoch": 2.02,
       "learning_rate": 7.395555555555556e-06,
-      "loss": 0.0447,
+      "loss": 0.1184,
       "step": 1675
     },
     {
-      "epoch": 4.01,
+      "epoch": 2.03,
       "learning_rate": 7.340000000000001e-06,
-      "loss": 0.0362,
+      "loss": 0.1117,
       "step": 1700
     },
     {
-      "epoch": 4.02,
+      "epoch": 2.03,
       "learning_rate": 7.284444444444445e-06,
-      "loss": 0.0368,
+      "loss": 0.1092,
       "step": 1725
     },
     {
-      "epoch": 4.02,
+      "epoch": 2.04,
       "learning_rate": 7.22888888888889e-06,
-      "loss": 0.0367,
+      "loss": 0.1027,
       "step": 1750
     },
     {
-      "epoch": 4.03,
+      "epoch": 2.04,
       "learning_rate": 7.173333333333335e-06,
-      "loss": 0.0391,
+      "loss": 0.1018,
       "step": 1775
     },
     {
-      "epoch": 4.03,
+      "epoch": 2.05,
       "learning_rate": 7.117777777777778e-06,
-      "loss": 0.0331,
+      "loss": 0.0909,
       "step": 1800
     },
     {
-      "epoch": 4.04,
+      "epoch": 2.05,
       "learning_rate": 7.062222222222223e-06,
-      "loss": 0.028,
+      "loss": 0.104,
       "step": 1825
     },
     {
-      "epoch": 4.04,
+      "epoch": 2.06,
       "learning_rate": 7.006666666666667e-06,
-      "loss": 0.0315,
+      "loss": 0.1002,
       "step": 1850
     },
     {
-      "epoch": 4.05,
+      "epoch": 2.06,
       "learning_rate": 6.951111111111112e-06,
-      "loss": 0.0308,
+      "loss": 0.0964,
       "step": 1875
     },
     {
-      "epoch": 4.05,
+      "epoch": 2.07,
       "learning_rate": 6.8955555555555565e-06,
-      "loss": 0.0248,
+      "loss": 0.0874,
       "step": 1900
     },
     {
-      "epoch": 4.06,
+      "epoch": 2.07,
       "learning_rate": 6.8400000000000014e-06,
-      "loss": 0.023,
+      "loss": 0.0885,
       "step": 1925
     },
     {
-      "epoch": 4.06,
+      "epoch": 2.08,
       "learning_rate": 6.784444444444445e-06,
-      "loss": 0.0233,
+      "loss": 0.084,
       "step": 1950
     },
     {
-      "epoch": 4.07,
+      "epoch": 2.08,
       "learning_rate": 6.7288888888888895e-06,
-      "loss": 0.0236,
+      "loss": 0.0766,
       "step": 1975
     },
     {
-      "epoch": 4.07,
+      "epoch": 2.09,
       "learning_rate": 6.6733333333333335e-06,
-      "loss": 0.0181,
+      "loss": 0.0759,
       "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": 2.09,
+      "eval_loss": 0.21021738648414612,
+      "eval_runtime": 1689.1729,
+      "eval_samples_per_second": 7.409,
+      "eval_steps_per_second": 0.464,
+      "eval_wer": 14.258399888466212,
       "step": 2000
     },
     {
-      "epoch": 4.08,
+      "epoch": 2.09,
       "learning_rate": 6.617777777777778e-06,
-      "loss": 0.0176,
+      "loss": 0.0705,
       "step": 2025
     },
     {
-      "epoch": 4.08,
+      "epoch": 2.1,
       "learning_rate": 6.562222222222223e-06,
-      "loss": 0.0177,
+      "loss": 0.0634,
       "step": 2050
     },
     {
-      "epoch": 5.0,
+      "epoch": 2.1,
       "learning_rate": 6.5066666666666665e-06,
-      "loss": 0.0198,
+      "loss": 0.0671,
       "step": 2075
     },
     {
-      "epoch": 5.01,
+      "epoch": 2.11,
       "learning_rate": 6.451111111111111e-06,
-      "loss": 0.0188,
+      "loss": 0.0658,
       "step": 2100
     },
     {
-      "epoch": 5.01,
+      "epoch": 2.11,
       "learning_rate": 6.395555555555556e-06,
-      "loss": 0.0153,
+      "loss": 0.0601,
       "step": 2125
     },
     {
-      "epoch": 5.02,
+      "epoch": 2.12,
       "learning_rate": 6.34e-06,
-      "loss": 0.0157,
+      "loss": 0.0701,
       "step": 2150
     },
     {
-      "epoch": 5.02,
+      "epoch": 2.12,
       "learning_rate": 6.284444444444445e-06,
-      "loss": 0.0196,
+      "loss": 0.0642,
       "step": 2175
     },
     {
-      "epoch": 5.03,
+      "epoch": 2.13,
       "learning_rate": 6.22888888888889e-06,
-      "loss": 0.0153,
+      "loss": 0.0612,
       "step": 2200
     },
     {
-      "epoch": 5.03,
+      "epoch": 2.13,
       "learning_rate": 6.173333333333333e-06,
-      "loss": 0.0132,
+      "loss": 0.0571,
       "step": 2225
     },
     {
-      "epoch": 5.04,
+      "epoch": 2.14,
       "learning_rate": 6.117777777777778e-06,
-      "loss": 0.0144,
+      "loss": 0.1102,
       "step": 2250
     },
     {
-      "epoch": 5.04,
+      "epoch": 2.14,
       "learning_rate": 6.062222222222223e-06,
-      "loss": 0.0145,
+      "loss": 0.0744,
       "step": 2275
     },
     {
-      "epoch": 5.05,
+      "epoch": 2.15,
       "learning_rate": 6.006666666666667e-06,
-      "loss": 0.0131,
+      "loss": 0.0638,
       "step": 2300
     },
     {
-      "epoch": 5.05,
+      "epoch": 2.15,
       "learning_rate": 5.951111111111112e-06,
-      "loss": 0.0107,
+      "loss": 0.0638,
       "step": 2325
     },
     {
-      "epoch": 5.06,
+      "epoch": 3.0,
       "learning_rate": 5.895555555555557e-06,
-      "loss": 0.0122,
+      "loss": 0.0601,
       "step": 2350
     },
     {
-      "epoch": 5.06,
+      "epoch": 3.01,
       "learning_rate": 5.84e-06,
-      "loss": 0.0121,
+      "loss": 0.056,
       "step": 2375
     },
     {
-      "epoch": 5.07,
+      "epoch": 3.01,
       "learning_rate": 5.784444444444445e-06,
-      "loss": 0.0103,
+      "loss": 0.062,
       "step": 2400
     },
     {
-      "epoch": 5.07,
+      "epoch": 3.02,
       "learning_rate": 5.72888888888889e-06,
-      "loss": 0.0089,
+      "loss": 0.0645,
       "step": 2425
     },
     {
-      "epoch": 5.08,
+      "epoch": 3.02,
       "learning_rate": 5.673333333333334e-06,
-      "loss": 0.0087,
+      "loss": 0.0595,
       "step": 2450
     },
     {
-      "epoch": 6.0,
+      "epoch": 3.03,
       "learning_rate": 5.617777777777779e-06,
-      "loss": 0.0087,
+      "loss": 0.0543,
       "step": 2475
     },
     {
-      "epoch": 6.01,
+      "epoch": 3.03,
       "learning_rate": 5.562222222222222e-06,
-      "loss": 0.0106,
+      "loss": 0.0525,
       "step": 2500
     },
     {
-      "epoch": 6.01,
+      "epoch": 3.04,
       "learning_rate": 5.506666666666667e-06,
-      "loss": 0.0094,
+      "loss": 0.0494,
       "step": 2525
     },
     {
-      "epoch": 6.02,
+      "epoch": 3.04,
       "learning_rate": 5.451111111111112e-06,
-      "loss": 0.0082,
+      "loss": 0.0473,
       "step": 2550
     },
     {
-      "epoch": 6.02,
+      "epoch": 3.05,
       "learning_rate": 5.3955555555555565e-06,
-      "loss": 0.0079,
+      "loss": 0.0483,
       "step": 2575
     },
     {
-      "epoch": 6.03,
+      "epoch": 3.05,
       "learning_rate": 5.3400000000000005e-06,
-      "loss": 0.0084,
+      "loss": 0.0467,
       "step": 2600
     },
     {
-      "epoch": 6.03,
+      "epoch": 3.06,
       "learning_rate": 5.2844444444444454e-06,
-      "loss": 0.0076,
+      "loss": 0.0503,
       "step": 2625
     },
     {
-      "epoch": 6.04,
+      "epoch": 3.06,
       "learning_rate": 5.228888888888889e-06,
-      "loss": 0.0069,
+      "loss": 0.0428,
       "step": 2650
     },
     {
-      "epoch": 6.04,
+      "epoch": 3.07,
       "learning_rate": 5.1733333333333335e-06,
-      "loss": 0.0067,
+      "loss": 0.0418,
       "step": 2675
     },
     {
-      "epoch": 6.05,
+      "epoch": 3.07,
       "learning_rate": 5.117777777777778e-06,
-      "loss": 0.0073,
+      "loss": 0.0424,
       "step": 2700
     },
     {
-      "epoch": 6.05,
+      "epoch": 3.08,
       "learning_rate": 5.062222222222222e-06,
-      "loss": 0.0062,
+      "loss": 0.0406,
       "step": 2725
     },
     {
-      "epoch": 6.06,
+      "epoch": 3.08,
       "learning_rate": 5.006666666666667e-06,
-      "loss": 0.0057,
+      "loss": 0.0327,
       "step": 2750
     },
     {
-      "epoch": 6.06,
+      "epoch": 3.09,
       "learning_rate": 4.951111111111111e-06,
-      "loss": 0.0064,
+      "loss": 0.0371,
       "step": 2775
     },
     {
-      "epoch": 6.07,
+      "epoch": 3.09,
       "learning_rate": 4.895555555555556e-06,
-      "loss": 0.0053,
+      "loss": 0.0313,
       "step": 2800
     },
     {
-      "epoch": 6.07,
+      "epoch": 3.1,
       "learning_rate": 4.84e-06,
-      "loss": 0.0047,
+      "loss": 0.0295,
       "step": 2825
     },
     {
-      "epoch": 6.08,
+      "epoch": 3.1,
       "learning_rate": 4.784444444444445e-06,
-      "loss": 0.005,
+      "loss": 0.0284,
       "step": 2850
     },
     {
-      "epoch": 6.08,
+      "epoch": 3.11,
       "learning_rate": 4.728888888888889e-06,
-      "loss": 0.0047,
+      "loss": 0.0322,
       "step": 2875
     },
     {
-      "epoch": 7.0,
+      "epoch": 3.11,
       "learning_rate": 4.673333333333333e-06,
-      "loss": 0.0052,
+      "loss": 0.0307,
       "step": 2900
     },
     {
-      "epoch": 7.01,
+      "epoch": 3.12,
       "learning_rate": 4.617777777777778e-06,
-      "loss": 0.005,
+      "loss": 0.0316,
       "step": 2925
     },
     {
-      "epoch": 7.01,
+      "epoch": 3.12,
       "learning_rate": 4.562222222222222e-06,
-      "loss": 0.0041,
+      "loss": 0.0286,
       "step": 2950
     },
     {
-      "epoch": 7.02,
+      "epoch": 3.13,
       "learning_rate": 4.506666666666667e-06,
-      "loss": 0.0045,
+      "loss": 0.0289,
       "step": 2975
     },
     {
-      "epoch": 7.02,
+      "epoch": 3.13,
       "learning_rate": 4.451111111111112e-06,
-      "loss": 0.0044,
+      "loss": 0.0264,
       "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": 3.13,
+      "eval_loss": 0.2200043797492981,
+      "eval_runtime": 1693.8446,
+      "eval_samples_per_second": 7.389,
+      "eval_steps_per_second": 0.462,
+      "eval_wer": 13.689776733254993,
       "step": 3000
     },
     {
-      "epoch": 7.03,
+      "epoch": 3.14,
       "learning_rate": 4.395555555555556e-06,
-      "loss": 0.0043,
+      "loss": 0.0609,
       "step": 3025
     },
     {
-      "epoch": 7.03,
+      "epoch": 3.14,
       "learning_rate": 4.34e-06,
-      "loss": 0.0041,
+      "loss": 0.0455,
       "step": 3050
     },
     {
-      "epoch": 7.04,
+      "epoch": 3.15,
       "learning_rate": 4.284444444444445e-06,
-      "loss": 0.0035,
+      "loss": 0.0323,
       "step": 3075
     },
     {
-      "epoch": 7.04,
+      "epoch": 3.15,
       "learning_rate": 4.228888888888889e-06,
-      "loss": 0.0044,
+      "loss": 0.0298,
       "step": 3100
     },
     {
-      "epoch": 7.05,
+      "epoch": 3.16,
       "learning_rate": 4.173333333333334e-06,
-      "loss": 0.0039,
+      "loss": 0.0265,
       "step": 3125
     },
     {
-      "epoch": 7.05,
+      "epoch": 4.0,
       "learning_rate": 4.117777777777779e-06,
-      "loss": 0.0031,
+      "loss": 0.1155,
       "step": 3150
     },
     {
-      "epoch": 7.06,
+      "epoch": 4.01,
       "learning_rate": 4.062222222222223e-06,
-      "loss": 0.0032,
+      "loss": 0.1149,
       "step": 3175
     },
     {
-      "epoch": 7.06,
+      "epoch": 4.01,
       "learning_rate": 4.006666666666667e-06,
-      "loss": 0.0034,
+      "loss": 0.1179,
       "step": 3200
     },
     {
-      "epoch": 7.07,
+      "epoch": 4.02,
       "learning_rate": 3.951111111111112e-06,
-      "loss": 0.0028,
+      "loss": 0.1228,
       "step": 3225
     },
     {
-      "epoch": 7.07,
+      "epoch": 4.02,
       "learning_rate": 3.895555555555556e-06,
-      "loss": 0.0032,
+      "loss": 0.1806,
       "step": 3250
     },
     {
-      "epoch": 7.08,
+      "epoch": 4.03,
       "learning_rate": 3.8400000000000005e-06,
-      "loss": 0.003,
+      "loss": 0.2425,
       "step": 3275
     },
     {
-      "epoch": 8.0,
+      "epoch": 4.03,
       "learning_rate": 3.784444444444445e-06,
-      "loss": 0.0034,
+      "loss": 0.1122,
       "step": 3300
     },
     {
-      "epoch": 8.01,
+      "epoch": 4.04,
       "learning_rate": 3.728888888888889e-06,
-      "loss": 0.003,
+      "loss": 0.0986,
       "step": 3325
     },
     {
-      "epoch": 8.01,
+      "epoch": 4.04,
       "learning_rate": 3.673333333333334e-06,
-      "loss": 0.0027,
+      "loss": 0.0942,
       "step": 3350
     },
     {
-      "epoch": 8.02,
+      "epoch": 4.05,
       "learning_rate": 3.617777777777778e-06,
-      "loss": 0.0026,
+      "loss": 0.1099,
       "step": 3375
     },
     {
-      "epoch": 8.02,
+      "epoch": 4.05,
       "learning_rate": 3.5622222222222224e-06,
-      "loss": 0.0028,
+      "loss": 0.1332,
       "step": 3400
     },
     {
-      "epoch": 8.03,
+      "epoch": 4.06,
       "learning_rate": 3.5066666666666673e-06,
-      "loss": 0.0029,
+      "loss": 0.1368,
       "step": 3425
     },
     {
-      "epoch": 8.03,
+      "epoch": 4.06,
       "learning_rate": 3.4511111111111113e-06,
-      "loss": 0.0027,
+      "loss": 0.164,
       "step": 3450
     },
     {
-      "epoch": 8.04,
+      "epoch": 4.07,
       "learning_rate": 3.3955555555555558e-06,
-      "loss": 0.0022,
+      "loss": 0.174,
       "step": 3475
     },
     {
-      "epoch": 8.04,
+      "epoch": 4.07,
       "learning_rate": 3.3400000000000006e-06,
-      "loss": 0.0024,
+      "loss": 0.1675,
       "step": 3500
     },
     {
-      "epoch": 8.05,
+      "epoch": 4.08,
       "learning_rate": 3.2844444444444447e-06,
-      "loss": 0.0028,
+      "loss": 0.0996,
       "step": 3525
     },
     {
-      "epoch": 8.05,
+      "epoch": 4.08,
       "learning_rate": 3.228888888888889e-06,
-      "loss": 0.0025,
+      "loss": 0.0795,
       "step": 3550
     },
     {
-      "epoch": 8.06,
+      "epoch": 4.09,
       "learning_rate": 3.173333333333334e-06,
-      "loss": 0.0022,
+      "loss": 0.0683,
       "step": 3575
     },
     {
-      "epoch": 8.06,
+      "epoch": 4.09,
       "learning_rate": 3.117777777777778e-06,
-      "loss": 0.0021,
+      "loss": 0.0635,
       "step": 3600
     },
     {
-      "epoch": 8.07,
+      "epoch": 4.1,
       "learning_rate": 3.0622222222222225e-06,
-      "loss": 0.002,
+      "loss": 0.0751,
       "step": 3625
     },
     {
-      "epoch": 8.07,
+      "epoch": 4.1,
       "learning_rate": 3.0066666666666674e-06,
-      "loss": 0.0019,
+      "loss": 0.0712,
       "step": 3650
     },
     {
-      "epoch": 8.08,
+      "epoch": 4.11,
       "learning_rate": 2.9511111111111114e-06,
-      "loss": 0.0019,
+      "loss": 0.0703,
       "step": 3675
     },
     {
-      "epoch": 8.08,
+      "epoch": 4.11,
       "learning_rate": 2.895555555555556e-06,
-      "loss": 0.0019,
+      "loss": 0.1152,
       "step": 3700
     },
     {
-      "epoch": 9.0,
+      "epoch": 4.12,
       "learning_rate": 2.84e-06,
-      "loss": 0.0021,
+      "loss": 0.1039,
       "step": 3725
     },
     {
-      "epoch": 9.01,
+      "epoch": 4.12,
       "learning_rate": 2.784444444444445e-06,
-      "loss": 0.0021,
+      "loss": 0.0863,
       "step": 3750
     },
     {
-      "epoch": 9.01,
+      "epoch": 4.13,
       "learning_rate": 2.7288888888888893e-06,
-      "loss": 0.0018,
+      "loss": 0.0882,
       "step": 3775
     },
     {
-      "epoch": 9.02,
+      "epoch": 4.13,
       "learning_rate": 2.6733333333333333e-06,
-      "loss": 0.0019,
+      "loss": 0.0677,
       "step": 3800
     },
     {
-      "epoch": 9.02,
+      "epoch": 4.14,
       "learning_rate": 2.617777777777778e-06,
-      "loss": 0.0018,
+      "loss": 0.0511,
       "step": 3825
     },
     {
-      "epoch": 9.03,
+      "epoch": 4.14,
       "learning_rate": 2.5622222222222226e-06,
-      "loss": 0.0021,
+      "loss": 0.0283,
       "step": 3850
     },
     {
-      "epoch": 9.03,
+      "epoch": 4.15,
       "learning_rate": 2.5066666666666667e-06,
-      "loss": 0.0017,
+      "loss": 0.0246,
       "step": 3875
     },
     {
-      "epoch": 9.04,
+      "epoch": 4.15,
       "learning_rate": 2.451111111111111e-06,
-      "loss": 0.0018,
+      "loss": 0.0231,
       "step": 3900
     },
     {
-      "epoch": 9.04,
+      "epoch": 5.0,
       "learning_rate": 2.3955555555555556e-06,
-      "loss": 0.0017,
+      "loss": 0.0442,
       "step": 3925
     },
     {
-      "epoch": 9.05,
+      "epoch": 5.01,
       "learning_rate": 2.3400000000000005e-06,
-      "loss": 0.002,
+      "loss": 0.063,
       "step": 3950
     },
     {
-      "epoch": 9.05,
+      "epoch": 5.01,
       "learning_rate": 2.2844444444444445e-06,
-      "loss": 0.0017,
+      "loss": 0.0548,
       "step": 3975
     },
     {
-      "epoch": 9.06,
+      "epoch": 5.02,
       "learning_rate": 2.228888888888889e-06,
-      "loss": 0.0016,
+      "loss": 0.0633,
       "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": 5.02,
+      "eval_loss": 0.1954876184463501,
+      "eval_runtime": 1710.419,
+      "eval_samples_per_second": 7.317,
+      "eval_steps_per_second": 0.458,
+      "eval_wer": 12.553526260232228,
       "step": 4000
     },
     {
-      "epoch": 9.06,
+      "epoch": 5.02,
       "learning_rate": 2.1733333333333334e-06,
-      "loss": 0.0016,
+      "loss": 0.0801,
       "step": 4025
     },
     {
-      "epoch": 9.07,
+      "epoch": 5.03,
       "learning_rate": 2.117777777777778e-06,
-      "loss": 0.0015,
+      "loss": 0.1296,
       "step": 4050
     },
     {
-      "epoch": 9.07,
+      "epoch": 5.03,
       "learning_rate": 2.0622222222222223e-06,
-      "loss": 0.0015,
+      "loss": 0.0944,
       "step": 4075
     },
     {
-      "epoch": 9.08,
+      "epoch": 5.04,
       "learning_rate": 2.006666666666667e-06,
-      "loss": 0.0015,
+      "loss": 0.0541,
       "step": 4100
     },
     {
-      "epoch": 10.0,
+      "epoch": 5.04,
       "learning_rate": 1.9511111111111113e-06,
-      "loss": 0.0015,
+      "loss": 0.0511,
       "step": 4125
     },
     {
-      "epoch": 10.01,
+      "epoch": 5.05,
       "learning_rate": 1.8955555555555557e-06,
-      "loss": 0.0016,
+      "loss": 0.0524,
       "step": 4150
     },
     {
-      "epoch": 10.01,
+      "epoch": 5.05,
       "learning_rate": 1.8400000000000002e-06,
-      "loss": 0.0015,
+      "loss": 0.0656,
       "step": 4175
     },
     {
-      "epoch": 10.02,
+      "epoch": 5.06,
       "learning_rate": 1.7844444444444444e-06,
-      "loss": 0.0015,
+      "loss": 0.0675,
       "step": 4200
     },
     {
-      "epoch": 10.02,
+      "epoch": 5.06,
       "learning_rate": 1.728888888888889e-06,
-      "loss": 0.0014,
+      "loss": 0.0916,
       "step": 4225
     },
     {
-      "epoch": 10.03,
+      "epoch": 5.07,
       "learning_rate": 1.6733333333333335e-06,
-      "loss": 0.0016,
+      "loss": 0.0977,
       "step": 4250
     },
     {
-      "epoch": 10.03,
+      "epoch": 5.07,
       "learning_rate": 1.6177777777777778e-06,
-      "loss": 0.0015,
+      "loss": 0.108,
       "step": 4275
     },
     {
-      "epoch": 10.04,
+      "epoch": 5.08,
       "learning_rate": 1.5622222222222225e-06,
-      "loss": 0.0015,
+      "loss": 0.0672,
       "step": 4300
     },
     {
-      "epoch": 10.04,
+      "epoch": 5.08,
       "learning_rate": 1.506666666666667e-06,
-      "loss": 0.0016,
+      "loss": 0.0444,
       "step": 4325
     },
     {
-      "epoch": 10.05,
+      "epoch": 5.09,
       "learning_rate": 1.4511111111111112e-06,
-      "loss": 0.0015,
+      "loss": 0.0371,
       "step": 4350
     },
     {
-      "epoch": 10.05,
+      "epoch": 5.09,
       "learning_rate": 1.3955555555555556e-06,
-      "loss": 0.0014,
+      "loss": 0.0368,
       "step": 4375
     },
     {
-      "epoch": 10.06,
+      "epoch": 5.1,
       "learning_rate": 1.34e-06,
-      "loss": 0.0014,
+      "loss": 0.038,
       "step": 4400
     },
     {
-      "epoch": 10.06,
+      "epoch": 5.1,
       "learning_rate": 1.2844444444444445e-06,
-      "loss": 0.0013,
+      "loss": 0.0366,
       "step": 4425
     },
     {
-      "epoch": 10.07,
+      "epoch": 5.11,
       "learning_rate": 1.228888888888889e-06,
-      "loss": 0.0013,
+      "loss": 0.041,
       "step": 4450
     },
     {
-      "epoch": 10.07,
+      "epoch": 5.11,
       "learning_rate": 1.1733333333333335e-06,
-      "loss": 0.0013,
+      "loss": 0.0613,
       "step": 4475
     },
     {
-      "epoch": 10.08,
+      "epoch": 5.12,
       "learning_rate": 1.117777777777778e-06,
-      "loss": 0.0013,
+      "loss": 0.06,
       "step": 4500
     },
     {
-      "epoch": 10.08,
+      "epoch": 5.12,
       "learning_rate": 1.0622222222222222e-06,
-      "loss": 0.0013,
+      "loss": 0.0519,
       "step": 4525
     },
     {
-      "epoch": 11.0,
+      "epoch": 5.13,
       "learning_rate": 1.0066666666666668e-06,
-      "loss": 0.0014,
+      "loss": 0.0475,
       "step": 4550
     },
     {
-      "epoch": 11.01,
+      "epoch": 5.13,
       "learning_rate": 9.511111111111111e-07,
-      "loss": 0.0014,
+      "loss": 0.0328,
       "step": 4575
     },
     {
-      "epoch": 11.01,
+      "epoch": 5.14,
       "learning_rate": 8.955555555555557e-07,
-      "loss": 0.0013,
+      "loss": 0.046,
       "step": 4600
     },
     {
-      "epoch": 11.02,
+      "epoch": 5.14,
       "learning_rate": 8.400000000000001e-07,
-      "loss": 0.0013,
+      "loss": 0.0184,
       "step": 4625
     },
     {
-      "epoch": 11.02,
+      "epoch": 5.15,
       "learning_rate": 7.844444444444445e-07,
-      "loss": 0.0014,
+      "loss": 0.014,
       "step": 4650
     },
     {
-      "epoch": 11.03,
+      "epoch": 5.15,
       "learning_rate": 7.28888888888889e-07,
-      "loss": 0.0015,
+      "loss": 0.0138,
       "step": 4675
     },
     {
-      "epoch": 11.03,
+      "epoch": 6.0,
       "learning_rate": 6.733333333333334e-07,
-      "loss": 0.0013,
+      "loss": 0.0116,
       "step": 4700
     },
     {
-      "epoch": 11.04,
+      "epoch": 6.01,
       "learning_rate": 6.177777777777778e-07,
-      "loss": 0.0013,
+      "loss": 0.014,
       "step": 4725
     },
     {
-      "epoch": 11.04,
+      "epoch": 6.01,
       "learning_rate": 5.622222222222223e-07,
-      "loss": 0.0013,
+      "loss": 0.0156,
       "step": 4750
     },
     {
-      "epoch": 11.05,
+      "epoch": 6.02,
       "learning_rate": 5.066666666666667e-07,
-      "loss": 0.0013,
+      "loss": 0.0136,
       "step": 4775
     },
     {
-      "epoch": 11.05,
+      "epoch": 6.02,
       "learning_rate": 4.511111111111111e-07,
-      "loss": 0.0013,
+      "loss": 0.0144,
       "step": 4800
     },
     {
-      "epoch": 11.06,
+      "epoch": 6.03,
       "learning_rate": 3.9555555555555557e-07,
-      "loss": 0.0013,
+      "loss": 0.0155,
       "step": 4825
     },
     {
-      "epoch": 11.06,
+      "epoch": 6.03,
       "learning_rate": 3.4000000000000003e-07,
-      "loss": 0.0013,
+      "loss": 0.0187,
       "step": 4850
     },
     {
-      "epoch": 11.07,
+      "epoch": 6.04,
       "learning_rate": 2.844444444444445e-07,
-      "loss": 0.0013,
+      "loss": 0.0189,
       "step": 4875
     },
     {
-      "epoch": 11.07,
+      "epoch": 6.04,
       "learning_rate": 2.2888888888888892e-07,
-      "loss": 0.0012,
+      "loss": 0.0187,
       "step": 4900
     },
     {
-      "epoch": 11.08,
+      "epoch": 6.05,
       "learning_rate": 1.7333333333333335e-07,
-      "loss": 0.0012,
+      "loss": 0.0194,
       "step": 4925
     },
     {
-      "epoch": 12.0,
+      "epoch": 6.05,
       "learning_rate": 1.1777777777777778e-07,
-      "loss": 0.0013,
+      "loss": 0.019,
       "step": 4950
     },
     {
-      "epoch": 12.01,
+      "epoch": 6.06,
       "learning_rate": 6.222222222222223e-08,
-      "loss": 0.0013,
+      "loss": 0.0183,
       "step": 4975
     },
     {
-      "epoch": 12.01,
+      "epoch": 6.06,
       "learning_rate": 6.666666666666667e-09,
-      "loss": 0.0012,
+      "loss": 0.0199,
       "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": 6.06,
+      "eval_loss": 0.19964517652988434,
+      "eval_runtime": 1705.0556,
+      "eval_samples_per_second": 7.34,
+      "eval_steps_per_second": 0.459,
+      "eval_wer": 12.012786552211754,
       "step": 5000
     },
     {
-      "epoch": 12.01,
+      "epoch": 6.06,
       "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": 4.612864472875008e+19,
+      "train_loss": 0.15837200704216958,
+      "train_runtime": 29005.3522,
+      "train_samples_per_second": 5.516,
+      "train_steps_per_second": 0.172
     }
   ],
   "max_steps": 5000,
   "num_train_epochs": 9223372036854775807,
-  "total_flos": 4.616673800159232e+19,
+  "total_flos": 4.612864472875008e+19,
   "trial_name": null,
   "trial_params": null
 }