End of training

all_results.json

@@ -1,12 +1,12 @@
 {
     "epoch": 2.17,
-    "eval_loss": 0.5219861268997192,
-    "eval_runtime": 361.899,
-    "eval_samples_per_second": 3.562,
-    "eval_steps_per_second": 1.782,
-    "eval_wer": 43.8592127355264,
-    "train_loss": 0.7047004199822744,
-    "train_runtime": 3798.8957,
-    "train_samples_per_second": 3.159,
-    "train_steps_per_second": 1.579
+    "eval_loss": 0.33273452520370483,
+    "eval_runtime": 2214.5995,
+    "eval_samples_per_second": 0.582,
+    "eval_steps_per_second": 0.291,
+    "eval_wer": 26.8453454387165,
+    "train_loss": 0.38768368325630825,
+    "train_runtime": 19861.2799,
+    "train_samples_per_second": 0.604,
+    "train_steps_per_second": 0.302
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 2.17,
-    "eval_loss": 0.5219861268997192,
-    "eval_runtime": 361.899,
-    "eval_samples_per_second": 3.562,
-    "eval_steps_per_second": 1.782,
-    "eval_wer": 43.8592127355264
+    "eval_loss": 0.33273452520370483,
+    "eval_runtime": 2214.5995,
+    "eval_samples_per_second": 0.582,
+    "eval_steps_per_second": 0.291,
+    "eval_wer": 26.8453454387165
 }

train_results.json

@@ -1,7 +1,7 @@
 {
     "epoch": 2.17,
-    "train_loss": 0.7047004199822744,
-    "train_runtime": 3798.8957,
-    "train_samples_per_second": 3.159,
-    "train_steps_per_second": 1.579
+    "train_loss": 0.38768368325630825,
+    "train_runtime": 19861.2799,
+    "train_samples_per_second": 0.604,
+    "train_steps_per_second": 0.302
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 43.8592127355264,
-  "best_model_checkpoint": "/home/rcj2772/data_sets/Seneca/output_dir/checkpoint-6000",
+  "best_metric": 26.8453454387165,
+  "best_model_checkpoint": "/home/rcj2772/data_sets/Seneca/output_dir/checkpoint-5000",
   "epoch": 2.1746666666666665,
   "global_step": 6000,
   "is_hyper_param_search": false,
@@ -9,1511 +9,1511 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 4.2000000000000006e-07,
-      "loss": 5.6013,
+      "learning_rate": 3.8e-07,
+      "loss": 4.7275,
       "step": 25
     },
     {
       "epoch": 0.01,
-      "learning_rate": 9.200000000000001e-07,
-      "loss": 5.3689,
+      "learning_rate": 8.8e-07,
+      "loss": 4.0879,
       "step": 50
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.42e-06,
-      "loss": 4.6359,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 3.1622,
       "step": 75
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.9200000000000003e-06,
-      "loss": 4.1888,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 2.6329,
       "step": 100
     },
     {
       "epoch": 0.02,
-      "learning_rate": 2.42e-06,
-      "loss": 3.5889,
+      "learning_rate": 2.38e-06,
+      "loss": 1.8973,
       "step": 125
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.92e-06,
-      "loss": 3.0842,
+      "learning_rate": 2.88e-06,
+      "loss": 1.5389,
       "step": 150
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.4200000000000007e-06,
-      "loss": 2.8286,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 1.3676,
       "step": 175
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.920000000000001e-06,
-      "loss": 2.4452,
+      "learning_rate": 3.88e-06,
+      "loss": 1.207,
       "step": 200
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.42e-06,
-      "loss": 2.1272,
+      "learning_rate": 4.38e-06,
+      "loss": 1.0281,
       "step": 225
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.92e-06,
-      "loss": 1.9312,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 0.9425,
       "step": 250
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.420000000000001e-06,
-      "loss": 1.8329,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.9098,
       "step": 275
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.92e-06,
-      "loss": 1.6862,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 0.7999,
       "step": 300
     },
     {
       "epoch": 0.05,
-      "learning_rate": 6.42e-06,
-      "loss": 1.6483,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.9003,
       "step": 325
     },
     {
       "epoch": 0.06,
-      "learning_rate": 6.92e-06,
-      "loss": 1.4521,
+      "learning_rate": 6.88e-06,
+      "loss": 0.7452,
       "step": 350
     },
     {
       "epoch": 0.06,
-      "learning_rate": 7.420000000000001e-06,
-      "loss": 1.4175,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7833,
       "step": 375
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.92e-06,
-      "loss": 1.4054,
+      "learning_rate": 7.88e-06,
+      "loss": 0.7824,
       "step": 400
     },
     {
       "epoch": 0.07,
-      "learning_rate": 8.42e-06,
-      "loss": 1.3339,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7376,
       "step": 425
     },
     {
       "epoch": 0.07,
-      "learning_rate": 8.920000000000001e-06,
-      "loss": 1.2991,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 0.6737,
       "step": 450
     },
     {
       "epoch": 0.08,
-      "learning_rate": 9.42e-06,
-      "loss": 1.2699,
+      "learning_rate": 9.38e-06,
+      "loss": 0.7435,
       "step": 475
     },
     {
       "epoch": 0.08,
-      "learning_rate": 9.920000000000002e-06,
-      "loss": 1.1429,
+      "learning_rate": 9.88e-06,
+      "loss": 0.6097,
       "step": 500
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.961818181818183e-06,
-      "loss": 1.2096,
+      "learning_rate": 9.965454545454546e-06,
+      "loss": 0.7438,
       "step": 525
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.916363636363637e-06,
-      "loss": 1.1677,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 0.7065,
       "step": 550
     },
     {
       "epoch": 0.1,
-      "learning_rate": 9.870909090909092e-06,
-      "loss": 1.035,
+      "learning_rate": 9.874545454545455e-06,
+      "loss": 0.6087,
       "step": 575
     },
     {
       "epoch": 0.1,
-      "learning_rate": 9.825454545454546e-06,
-      "loss": 1.0198,
+      "learning_rate": 9.82909090909091e-06,
+      "loss": 0.637,
       "step": 600
     },
     {
       "epoch": 0.1,
-      "learning_rate": 9.780000000000001e-06,
-      "loss": 0.989,
+      "learning_rate": 9.783636363636365e-06,
+      "loss": 0.5986,
       "step": 625
     },
     {
       "epoch": 0.11,
-      "learning_rate": 9.734545454545455e-06,
-      "loss": 0.9738,
+      "learning_rate": 9.738181818181818e-06,
+      "loss": 0.5865,
       "step": 650
     },
     {
       "epoch": 0.11,
-      "learning_rate": 9.68909090909091e-06,
-      "loss": 0.9576,
+      "learning_rate": 9.692727272727274e-06,
+      "loss": 0.5588,
       "step": 675
     },
     {
       "epoch": 0.12,
-      "learning_rate": 9.643636363636364e-06,
-      "loss": 0.9663,
+      "learning_rate": 9.64727272727273e-06,
+      "loss": 0.6164,
       "step": 700
     },
     {
       "epoch": 0.12,
-      "learning_rate": 9.598181818181818e-06,
-      "loss": 0.9403,
+      "learning_rate": 9.601818181818183e-06,
+      "loss": 0.6035,
       "step": 725
     },
     {
       "epoch": 0.12,
-      "learning_rate": 9.552727272727273e-06,
-      "loss": 0.8925,
+      "learning_rate": 9.556363636363637e-06,
+      "loss": 0.5321,
       "step": 750
     },
     {
       "epoch": 0.13,
-      "learning_rate": 9.507272727272729e-06,
-      "loss": 0.8861,
+      "learning_rate": 9.510909090909092e-06,
+      "loss": 0.5492,
       "step": 775
     },
     {
       "epoch": 0.13,
-      "learning_rate": 9.461818181818183e-06,
-      "loss": 0.9019,
+      "learning_rate": 9.465454545454546e-06,
+      "loss": 0.5241,
       "step": 800
     },
     {
       "epoch": 0.14,
-      "learning_rate": 9.416363636363636e-06,
-      "loss": 0.831,
+      "learning_rate": 9.42e-06,
+      "loss": 0.4835,
       "step": 825
     },
     {
       "epoch": 0.14,
-      "learning_rate": 9.370909090909092e-06,
-      "loss": 1.0007,
+      "learning_rate": 9.374545454545455e-06,
+      "loss": 0.6401,
       "step": 850
     },
     {
       "epoch": 0.15,
-      "learning_rate": 9.325454545454547e-06,
-      "loss": 0.9652,
+      "learning_rate": 9.32909090909091e-06,
+      "loss": 0.6224,
       "step": 875
     },
     {
       "epoch": 0.15,
-      "learning_rate": 9.280000000000001e-06,
-      "loss": 0.8738,
+      "learning_rate": 9.283636363636364e-06,
+      "loss": 0.5077,
       "step": 900
     },
     {
       "epoch": 0.15,
-      "learning_rate": 9.234545454545455e-06,
-      "loss": 0.8946,
+      "learning_rate": 9.23818181818182e-06,
+      "loss": 0.5315,
       "step": 925
     },
     {
       "epoch": 0.16,
-      "learning_rate": 9.18909090909091e-06,
-      "loss": 0.773,
+      "learning_rate": 9.192727272727273e-06,
+      "loss": 0.4291,
       "step": 950
     },
     {
       "epoch": 0.16,
-      "learning_rate": 9.143636363636364e-06,
-      "loss": 0.8357,
+      "learning_rate": 9.147272727272727e-06,
+      "loss": 0.5389,
       "step": 975
     },
     {
       "epoch": 0.17,
-      "learning_rate": 9.09818181818182e-06,
-      "loss": 0.7706,
+      "learning_rate": 9.101818181818183e-06,
+      "loss": 0.4698,
       "step": 1000
     },
     {
       "epoch": 0.17,
-      "eval_loss": 0.818512499332428,
-      "eval_runtime": 412.4709,
-      "eval_samples_per_second": 3.125,
-      "eval_steps_per_second": 1.564,
-      "eval_wer": 69.79665443691313,
+      "eval_loss": 0.516851007938385,
+      "eval_runtime": 2299.3338,
+      "eval_samples_per_second": 0.561,
+      "eval_steps_per_second": 0.281,
+      "eval_wer": 42.61550898575959,
       "step": 1000
     },
     {
       "epoch": 0.17,
-      "learning_rate": 9.052727272727273e-06,
-      "loss": 0.8804,
+      "learning_rate": 9.056363636363638e-06,
+      "loss": 0.5734,
       "step": 1025
     },
     {
       "epoch": 0.17,
-      "learning_rate": 9.007272727272729e-06,
-      "loss": 0.8028,
+      "learning_rate": 9.010909090909092e-06,
+      "loss": 0.4825,
       "step": 1050
     },
     {
       "epoch": 0.18,
-      "learning_rate": 8.961818181818182e-06,
-      "loss": 0.8313,
+      "learning_rate": 8.965454545454546e-06,
+      "loss": 0.4937,
       "step": 1075
     },
     {
       "epoch": 0.18,
-      "learning_rate": 8.916363636363638e-06,
-      "loss": 0.8308,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 0.502,
       "step": 1100
     },
     {
       "epoch": 0.19,
-      "learning_rate": 8.870909090909091e-06,
-      "loss": 0.7963,
+      "learning_rate": 8.874545454545456e-06,
+      "loss": 0.5304,
       "step": 1125
     },
     {
       "epoch": 0.19,
-      "learning_rate": 8.825454545454545e-06,
-      "loss": 0.8294,
+      "learning_rate": 8.82909090909091e-06,
+      "loss": 0.585,
       "step": 1150
     },
     {
       "epoch": 0.2,
-      "learning_rate": 8.78e-06,
-      "loss": 0.7245,
+      "learning_rate": 8.783636363636364e-06,
+      "loss": 0.4542,
       "step": 1175
     },
     {
       "epoch": 0.2,
-      "learning_rate": 8.734545454545456e-06,
-      "loss": 0.7297,
+      "learning_rate": 8.73818181818182e-06,
+      "loss": 0.461,
       "step": 1200
     },
     {
       "epoch": 0.2,
-      "learning_rate": 8.68909090909091e-06,
-      "loss": 0.7972,
+      "learning_rate": 8.692727272727273e-06,
+      "loss": 0.5081,
       "step": 1225
     },
     {
       "epoch": 0.21,
-      "learning_rate": 8.643636363636364e-06,
-      "loss": 0.7423,
+      "learning_rate": 8.647272727272727e-06,
+      "loss": 0.4471,
       "step": 1250
     },
     {
       "epoch": 0.21,
-      "learning_rate": 8.598181818181819e-06,
-      "loss": 0.7589,
+      "learning_rate": 8.601818181818182e-06,
+      "loss": 0.4815,
       "step": 1275
     },
     {
       "epoch": 0.22,
-      "learning_rate": 8.552727272727274e-06,
-      "loss": 0.7391,
+      "learning_rate": 8.556363636363638e-06,
+      "loss": 0.4804,
       "step": 1300
     },
     {
       "epoch": 0.22,
-      "learning_rate": 8.507272727272728e-06,
-      "loss": 0.7139,
+      "learning_rate": 8.510909090909092e-06,
+      "loss": 0.4413,
       "step": 1325
     },
     {
       "epoch": 0.23,
-      "learning_rate": 8.461818181818182e-06,
-      "loss": 0.7484,
+      "learning_rate": 8.465454545454547e-06,
+      "loss": 0.4752,
       "step": 1350
     },
     {
       "epoch": 0.23,
-      "learning_rate": 8.416363636363637e-06,
-      "loss": 0.8506,
+      "learning_rate": 8.42e-06,
+      "loss": 0.5209,
       "step": 1375
     },
     {
       "epoch": 0.23,
-      "learning_rate": 8.370909090909091e-06,
-      "loss": 0.7147,
+      "learning_rate": 8.374545454545454e-06,
+      "loss": 0.4457,
       "step": 1400
     },
     {
       "epoch": 0.24,
-      "learning_rate": 8.325454545454547e-06,
-      "loss": 0.795,
+      "learning_rate": 8.32909090909091e-06,
+      "loss": 0.5209,
       "step": 1425
     },
     {
       "epoch": 0.24,
-      "learning_rate": 8.28e-06,
-      "loss": 0.6987,
+      "learning_rate": 8.283636363636365e-06,
+      "loss": 0.4721,
       "step": 1450
     },
     {
       "epoch": 0.25,
-      "learning_rate": 8.234545454545456e-06,
-      "loss": 0.6036,
+      "learning_rate": 8.238181818181819e-06,
+      "loss": 0.3808,
       "step": 1475
     },
     {
       "epoch": 0.25,
-      "learning_rate": 8.18909090909091e-06,
-      "loss": 0.6356,
+      "learning_rate": 8.192727272727273e-06,
+      "loss": 0.3892,
       "step": 1500
     },
     {
       "epoch": 0.25,
-      "learning_rate": 8.143636363636365e-06,
-      "loss": 0.7819,
+      "learning_rate": 8.147272727272728e-06,
+      "loss": 0.4843,
       "step": 1525
     },
     {
       "epoch": 0.26,
-      "learning_rate": 8.098181818181819e-06,
-      "loss": 0.7489,
+      "learning_rate": 8.101818181818184e-06,
+      "loss": 0.4976,
       "step": 1550
     },
     {
       "epoch": 0.26,
-      "learning_rate": 8.052727272727272e-06,
-      "loss": 0.7443,
+      "learning_rate": 8.056363636363636e-06,
+      "loss": 0.5162,
       "step": 1575
     },
     {
       "epoch": 0.27,
-      "learning_rate": 8.007272727272728e-06,
-      "loss": 0.6522,
+      "learning_rate": 8.010909090909091e-06,
+      "loss": 0.4288,
       "step": 1600
     },
     {
       "epoch": 0.27,
-      "learning_rate": 7.961818181818183e-06,
-      "loss": 0.6936,
+      "learning_rate": 7.965454545454547e-06,
+      "loss": 0.437,
       "step": 1625
     },
     {
       "epoch": 0.28,
-      "learning_rate": 7.916363636363637e-06,
-      "loss": 0.7222,
+      "learning_rate": 7.92e-06,
+      "loss": 0.4668,
       "step": 1650
     },
     {
       "epoch": 0.28,
-      "learning_rate": 7.870909090909091e-06,
-      "loss": 0.6555,
+      "learning_rate": 7.874545454545454e-06,
+      "loss": 0.4039,
       "step": 1675
     },
     {
       "epoch": 0.28,
-      "learning_rate": 7.825454545454546e-06,
-      "loss": 0.6729,
+      "learning_rate": 7.82909090909091e-06,
+      "loss": 0.4512,
       "step": 1700
     },
     {
       "epoch": 0.29,
-      "learning_rate": 7.78e-06,
-      "loss": 0.6733,
+      "learning_rate": 7.783636363636365e-06,
+      "loss": 0.4357,
       "step": 1725
     },
     {
       "epoch": 0.29,
-      "learning_rate": 7.734545454545455e-06,
-      "loss": 0.6301,
+      "learning_rate": 7.738181818181819e-06,
+      "loss": 0.3964,
       "step": 1750
     },
     {
       "epoch": 0.3,
-      "learning_rate": 7.68909090909091e-06,
-      "loss": 0.702,
+      "learning_rate": 7.692727272727273e-06,
+      "loss": 0.4589,
       "step": 1775
     },
     {
       "epoch": 0.3,
-      "learning_rate": 7.643636363636365e-06,
-      "loss": 0.7571,
+      "learning_rate": 7.647272727272728e-06,
+      "loss": 0.4641,
       "step": 1800
     },
     {
       "epoch": 0.3,
-      "learning_rate": 7.598181818181819e-06,
-      "loss": 0.5723,
+      "learning_rate": 7.6018181818181826e-06,
+      "loss": 0.3623,
       "step": 1825
     },
     {
       "epoch": 0.31,
-      "learning_rate": 7.552727272727274e-06,
-      "loss": 0.6493,
+      "learning_rate": 7.556363636363637e-06,
+      "loss": 0.3959,
       "step": 1850
     },
     {
       "epoch": 0.31,
-      "learning_rate": 7.507272727272728e-06,
-      "loss": 0.6668,
+      "learning_rate": 7.510909090909092e-06,
+      "loss": 0.3921,
       "step": 1875
     },
     {
       "epoch": 0.32,
-      "learning_rate": 7.461818181818182e-06,
-      "loss": 0.6291,
+      "learning_rate": 7.4654545454545455e-06,
+      "loss": 0.3909,
       "step": 1900
     },
     {
       "epoch": 0.32,
-      "learning_rate": 7.416363636363637e-06,
-      "loss": 0.7087,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 0.4608,
       "step": 1925
     },
     {
       "epoch": 0.33,
-      "learning_rate": 7.370909090909092e-06,
-      "loss": 0.6432,
+      "learning_rate": 7.3745454545454556e-06,
+      "loss": 0.408,
       "step": 1950
     },
     {
       "epoch": 0.33,
-      "learning_rate": 7.325454545454546e-06,
-      "loss": 0.5941,
+      "learning_rate": 7.32909090909091e-06,
+      "loss": 0.3655,
       "step": 1975
     },
     {
       "epoch": 0.33,
-      "learning_rate": 7.280000000000001e-06,
-      "loss": 0.6779,
+      "learning_rate": 7.283636363636364e-06,
+      "loss": 0.4237,
       "step": 2000
     },
     {
       "epoch": 0.33,
-      "eval_loss": 0.6389220356941223,
-      "eval_runtime": 383.9898,
-      "eval_samples_per_second": 3.357,
-      "eval_steps_per_second": 1.68,
-      "eval_wer": 55.07120203967415,
+      "eval_loss": 0.39532899856567383,
+      "eval_runtime": 2277.9452,
+      "eval_samples_per_second": 0.566,
+      "eval_steps_per_second": 0.283,
+      "eval_wer": 35.03513463093091,
       "step": 2000
     },
     {
       "epoch": 0.34,
-      "learning_rate": 7.234545454545455e-06,
-      "loss": 0.6888,
+      "learning_rate": 7.2381818181818185e-06,
+      "loss": 0.4668,
       "step": 2025
     },
     {
       "epoch": 0.34,
-      "learning_rate": 7.18909090909091e-06,
-      "loss": 0.644,
+      "learning_rate": 7.192727272727273e-06,
+      "loss": 0.3957,
       "step": 2050
     },
     {
       "epoch": 0.35,
-      "learning_rate": 7.145454545454547e-06,
-      "loss": 0.6554,
+      "learning_rate": 7.1472727272727285e-06,
+      "loss": 0.4053,
       "step": 2075
     },
     {
       "epoch": 0.35,
-      "learning_rate": 7.100000000000001e-06,
-      "loss": 0.6161,
+      "learning_rate": 7.101818181818182e-06,
+      "loss": 0.3659,
       "step": 2100
     },
     {
       "epoch": 0.35,
-      "learning_rate": 7.054545454545455e-06,
-      "loss": 0.6484,
+      "learning_rate": 7.056363636363637e-06,
+      "loss": 0.3957,
       "step": 2125
     },
     {
       "epoch": 0.36,
-      "learning_rate": 7.00909090909091e-06,
-      "loss": 0.6378,
+      "learning_rate": 7.0109090909090915e-06,
+      "loss": 0.4341,
       "step": 2150
     },
     {
       "epoch": 0.36,
-      "learning_rate": 6.963636363636364e-06,
-      "loss": 0.6438,
+      "learning_rate": 6.965454545454546e-06,
+      "loss": 0.3944,
       "step": 2175
     },
     {
       "epoch": 0.37,
-      "learning_rate": 6.918181818181818e-06,
-      "loss": 0.6121,
+      "learning_rate": 6.92e-06,
+      "loss": 0.3924,
       "step": 2200
     },
     {
       "epoch": 0.37,
-      "learning_rate": 6.872727272727273e-06,
-      "loss": 0.5537,
+      "learning_rate": 6.874545454545455e-06,
+      "loss": 0.3448,
       "step": 2225
     },
     {
       "epoch": 0.38,
-      "learning_rate": 6.827272727272728e-06,
-      "loss": 0.6023,
+      "learning_rate": 6.82909090909091e-06,
+      "loss": 0.3856,
       "step": 2250
     },
     {
       "epoch": 0.38,
-      "learning_rate": 6.781818181818183e-06,
-      "loss": 0.6361,
+      "learning_rate": 6.7836363636363644e-06,
+      "loss": 0.4088,
       "step": 2275
     },
     {
       "epoch": 0.38,
-      "learning_rate": 6.7363636363636365e-06,
-      "loss": 0.663,
+      "learning_rate": 6.738181818181819e-06,
+      "loss": 0.4379,
       "step": 2300
     },
     {
       "epoch": 0.39,
-      "learning_rate": 6.690909090909091e-06,
-      "loss": 0.5898,
+      "learning_rate": 6.692727272727273e-06,
+      "loss": 0.3856,
       "step": 2325
     },
     {
       "epoch": 0.39,
-      "learning_rate": 6.645454545454546e-06,
-      "loss": 0.6508,
+      "learning_rate": 6.647272727272727e-06,
+      "loss": 0.4111,
       "step": 2350
     },
     {
       "epoch": 0.4,
-      "learning_rate": 6.600000000000001e-06,
-      "loss": 0.5475,
+      "learning_rate": 6.601818181818183e-06,
+      "loss": 0.3716,
       "step": 2375
     },
     {
       "epoch": 0.4,
-      "learning_rate": 6.554545454545455e-06,
-      "loss": 0.5629,
+      "learning_rate": 6.5563636363636374e-06,
+      "loss": 0.3523,
       "step": 2400
     },
     {
       "epoch": 0.4,
-      "learning_rate": 6.5090909090909095e-06,
-      "loss": 0.5163,
+      "learning_rate": 6.510909090909091e-06,
+      "loss": 0.3096,
       "step": 2425
     },
     {
       "epoch": 0.41,
-      "learning_rate": 6.463636363636364e-06,
-      "loss": 0.5804,
+      "learning_rate": 6.465454545454546e-06,
+      "loss": 0.3672,
       "step": 2450
     },
     {
       "epoch": 0.41,
-      "learning_rate": 6.418181818181819e-06,
-      "loss": 0.5706,
+      "learning_rate": 6.42e-06,
+      "loss": 0.3415,
       "step": 2475
     },
     {
       "epoch": 1.0,
-      "learning_rate": 6.372727272727274e-06,
-      "loss": 0.4578,
+      "learning_rate": 6.374545454545456e-06,
+      "loss": 0.2064,
       "step": 2500
     },
     {
       "epoch": 1.01,
-      "learning_rate": 6.327272727272727e-06,
-      "loss": 0.4942,
+      "learning_rate": 6.3290909090909096e-06,
+      "loss": 0.2218,
       "step": 2525
     },
     {
       "epoch": 1.01,
-      "learning_rate": 6.2818181818181825e-06,
-      "loss": 0.4727,
+      "learning_rate": 6.283636363636364e-06,
+      "loss": 0.2255,
       "step": 2550
     },
     {
       "epoch": 1.02,
-      "learning_rate": 6.236363636363637e-06,
-      "loss": 0.3782,
+      "learning_rate": 6.238181818181819e-06,
+      "loss": 0.1819,
       "step": 2575
     },
     {
       "epoch": 1.02,
-      "learning_rate": 6.190909090909092e-06,
-      "loss": 0.4767,
+      "learning_rate": 6.192727272727273e-06,
+      "loss": 0.221,
       "step": 2600
     },
     {
       "epoch": 1.02,
-      "learning_rate": 6.1454545454545454e-06,
-      "loss": 0.5233,
+      "learning_rate": 6.147272727272727e-06,
+      "loss": 0.2702,
       "step": 2625
     },
     {
       "epoch": 1.03,
-      "learning_rate": 6.1e-06,
-      "loss": 0.4596,
+      "learning_rate": 6.101818181818182e-06,
+      "loss": 0.2232,
       "step": 2650
     },
     {
       "epoch": 1.03,
-      "learning_rate": 6.0545454545454555e-06,
-      "loss": 0.4481,
+      "learning_rate": 6.056363636363637e-06,
+      "loss": 0.2465,
       "step": 2675
     },
     {
       "epoch": 1.04,
-      "learning_rate": 6.00909090909091e-06,
-      "loss": 0.4935,
+      "learning_rate": 6.010909090909092e-06,
+      "loss": 0.2437,
       "step": 2700
     },
     {
       "epoch": 1.04,
-      "learning_rate": 5.963636363636364e-06,
-      "loss": 0.3752,
+      "learning_rate": 5.965454545454546e-06,
+      "loss": 0.1712,
       "step": 2725
     },
     {
       "epoch": 1.05,
-      "learning_rate": 5.9181818181818184e-06,
-      "loss": 0.5005,
+      "learning_rate": 5.92e-06,
+      "loss": 0.2478,
       "step": 2750
     },
     {
       "epoch": 1.05,
-      "learning_rate": 5.872727272727273e-06,
-      "loss": 0.5152,
+      "learning_rate": 5.874545454545455e-06,
+      "loss": 0.2263,
       "step": 2775
     },
     {
       "epoch": 1.05,
-      "learning_rate": 5.8272727272727285e-06,
-      "loss": 0.4325,
+      "learning_rate": 5.82909090909091e-06,
+      "loss": 0.2031,
       "step": 2800
     },
     {
       "epoch": 1.06,
-      "learning_rate": 5.781818181818181e-06,
-      "loss": 0.5238,
+      "learning_rate": 5.783636363636365e-06,
+      "loss": 0.2435,
       "step": 2825
     },
     {
       "epoch": 1.06,
-      "learning_rate": 5.736363636363637e-06,
-      "loss": 0.4427,
+      "learning_rate": 5.7381818181818185e-06,
+      "loss": 0.2155,
       "step": 2850
     },
     {
       "epoch": 1.07,
-      "learning_rate": 5.690909090909091e-06,
-      "loss": 0.4761,
+      "learning_rate": 5.692727272727273e-06,
+      "loss": 0.2485,
       "step": 2875
     },
     {
       "epoch": 1.07,
-      "learning_rate": 5.645454545454546e-06,
-      "loss": 0.4914,
+      "learning_rate": 5.647272727272728e-06,
+      "loss": 0.252,
       "step": 2900
     },
     {
       "epoch": 1.07,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 0.435,
+      "learning_rate": 5.601818181818183e-06,
+      "loss": 0.222,
       "step": 2925
     },
     {
       "epoch": 1.08,
-      "learning_rate": 5.554545454545454e-06,
-      "loss": 0.5136,
+      "learning_rate": 5.556363636363636e-06,
+      "loss": 0.2737,
       "step": 2950
     },
     {
       "epoch": 1.08,
-      "learning_rate": 5.50909090909091e-06,
-      "loss": 0.453,
+      "learning_rate": 5.5109090909090914e-06,
+      "loss": 0.2088,
       "step": 2975
     },
     {
       "epoch": 1.09,
-      "learning_rate": 5.463636363636364e-06,
-      "loss": 0.5047,
+      "learning_rate": 5.465454545454546e-06,
+      "loss": 0.2547,
       "step": 3000
     },
     {
       "epoch": 1.09,
-      "eval_loss": 0.5776195526123047,
-      "eval_runtime": 363.6882,
-      "eval_samples_per_second": 3.544,
-      "eval_steps_per_second": 1.773,
-      "eval_wer": 48.64125365337976,
+      "eval_loss": 0.36878591775894165,
+      "eval_runtime": 2273.9498,
+      "eval_samples_per_second": 0.567,
+      "eval_steps_per_second": 0.284,
+      "eval_wer": 33.685716062433926,
       "step": 3000
     },
     {
       "epoch": 1.09,
-      "learning_rate": 5.418181818181819e-06,
-      "loss": 0.4433,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 0.2017,
       "step": 3025
     },
     {
       "epoch": 1.1,
-      "learning_rate": 5.372727272727273e-06,
-      "loss": 0.4821,
+      "learning_rate": 5.374545454545454e-06,
+      "loss": 0.2424,
       "step": 3050
     },
     {
       "epoch": 1.1,
-      "learning_rate": 5.327272727272727e-06,
-      "loss": 0.5219,
+      "learning_rate": 5.329090909090909e-06,
+      "loss": 0.2506,
       "step": 3075
     },
     {
       "epoch": 1.1,
-      "learning_rate": 5.281818181818183e-06,
-      "loss": 0.4334,
+      "learning_rate": 5.283636363636364e-06,
+      "loss": 0.2602,
       "step": 3100
     },
     {
       "epoch": 1.11,
-      "learning_rate": 5.236363636363637e-06,
-      "loss": 0.5332,
+      "learning_rate": 5.238181818181819e-06,
+      "loss": 0.293,
       "step": 3125
     },
     {
       "epoch": 1.11,
-      "learning_rate": 5.190909090909091e-06,
-      "loss": 0.4962,
+      "learning_rate": 5.192727272727273e-06,
+      "loss": 0.2234,
       "step": 3150
     },
     {
       "epoch": 1.12,
-      "learning_rate": 5.145454545454546e-06,
-      "loss": 0.4089,
+      "learning_rate": 5.147272727272727e-06,
+      "loss": 0.2034,
       "step": 3175
     },
     {
       "epoch": 1.12,
-      "learning_rate": 5.1e-06,
-      "loss": 0.4765,
+      "learning_rate": 5.101818181818182e-06,
+      "loss": 0.2769,
       "step": 3200
     },
     {
       "epoch": 1.12,
-      "learning_rate": 5.054545454545455e-06,
-      "loss": 0.4863,
+      "learning_rate": 5.056363636363637e-06,
+      "loss": 0.25,
       "step": 3225
     },
     {
       "epoch": 1.13,
-      "learning_rate": 5.009090909090909e-06,
-      "loss": 0.5127,
+      "learning_rate": 5.010909090909092e-06,
+      "loss": 0.2425,
       "step": 3250
     },
     {
       "epoch": 1.13,
-      "learning_rate": 4.963636363636364e-06,
-      "loss": 0.476,
+      "learning_rate": 4.965454545454546e-06,
+      "loss": 0.22,
       "step": 3275
     },
     {
       "epoch": 1.14,
-      "learning_rate": 4.918181818181819e-06,
-      "loss": 0.4638,
+      "learning_rate": 4.92e-06,
+      "loss": 0.2266,
       "step": 3300
     },
     {
       "epoch": 1.14,
-      "learning_rate": 4.872727272727273e-06,
-      "loss": 0.4698,
+      "learning_rate": 4.874545454545455e-06,
+      "loss": 0.2221,
       "step": 3325
     },
     {
       "epoch": 1.15,
-      "learning_rate": 4.827272727272728e-06,
-      "loss": 0.3705,
+      "learning_rate": 4.8290909090909095e-06,
+      "loss": 0.1786,
       "step": 3350
     },
     {
       "epoch": 1.15,
-      "learning_rate": 4.7818181818181825e-06,
-      "loss": 0.4775,
+      "learning_rate": 4.783636363636364e-06,
+      "loss": 0.2556,
       "step": 3375
     },
     {
       "epoch": 1.15,
-      "learning_rate": 4.736363636363637e-06,
-      "loss": 0.4774,
+      "learning_rate": 4.738181818181819e-06,
+      "loss": 0.2448,
       "step": 3400
     },
     {
       "epoch": 1.16,
-      "learning_rate": 4.690909090909092e-06,
-      "loss": 0.4395,
+      "learning_rate": 4.692727272727273e-06,
+      "loss": 0.19,
       "step": 3425
     },
     {
       "epoch": 1.16,
-      "learning_rate": 4.645454545454545e-06,
-      "loss": 0.5421,
+      "learning_rate": 4.647272727272728e-06,
+      "loss": 0.2793,
       "step": 3450
     },
     {
       "epoch": 1.17,
-      "learning_rate": 4.600000000000001e-06,
-      "loss": 0.4346,
+      "learning_rate": 4.6018181818181825e-06,
+      "loss": 0.2135,
       "step": 3475
     },
     {
       "epoch": 1.17,
-      "learning_rate": 4.554545454545455e-06,
-      "loss": 0.4454,
+      "learning_rate": 4.556363636363636e-06,
+      "loss": 0.1994,
       "step": 3500
     },
     {
       "epoch": 1.17,
-      "learning_rate": 4.50909090909091e-06,
-      "loss": 0.441,
+      "learning_rate": 4.510909090909092e-06,
+      "loss": 0.215,
       "step": 3525
     },
     {
       "epoch": 1.18,
-      "learning_rate": 4.463636363636364e-06,
-      "loss": 0.5132,
+      "learning_rate": 4.4654545454545454e-06,
+      "loss": 0.2709,
       "step": 3550
     },
     {
       "epoch": 1.18,
-      "learning_rate": 4.418181818181818e-06,
-      "loss": 0.4546,
+      "learning_rate": 4.42e-06,
+      "loss": 0.2189,
       "step": 3575
     },
     {
       "epoch": 1.19,
-      "learning_rate": 4.372727272727273e-06,
-      "loss": 0.4235,
+      "learning_rate": 4.374545454545455e-06,
+      "loss": 0.2234,
       "step": 3600
     },
     {
       "epoch": 1.19,
-      "learning_rate": 4.327272727272728e-06,
-      "loss": 0.4955,
+      "learning_rate": 4.329090909090909e-06,
+      "loss": 0.2716,
       "step": 3625
     },
     {
       "epoch": 1.2,
-      "learning_rate": 4.281818181818182e-06,
-      "loss": 0.4967,
+      "learning_rate": 4.283636363636365e-06,
+      "loss": 0.2459,
       "step": 3650
     },
     {
       "epoch": 1.2,
-      "learning_rate": 4.236363636363637e-06,
-      "loss": 0.4617,
+      "learning_rate": 4.238181818181818e-06,
+      "loss": 0.2378,
       "step": 3675
     },
     {
       "epoch": 1.2,
-      "learning_rate": 4.190909090909091e-06,
-      "loss": 0.4491,
+      "learning_rate": 4.192727272727273e-06,
+      "loss": 0.2116,
       "step": 3700
     },
     {
       "epoch": 1.21,
-      "learning_rate": 4.145454545454546e-06,
-      "loss": 0.4087,
+      "learning_rate": 4.147272727272728e-06,
+      "loss": 0.1569,
       "step": 3725
     },
     {
       "epoch": 1.21,
-      "learning_rate": 4.1e-06,
-      "loss": 0.4776,
+      "learning_rate": 4.101818181818182e-06,
+      "loss": 0.2136,
       "step": 3750
     },
     {
       "epoch": 1.22,
-      "learning_rate": 4.054545454545455e-06,
-      "loss": 0.3885,
+      "learning_rate": 4.056363636363637e-06,
+      "loss": 0.1828,
       "step": 3775
     },
     {
       "epoch": 1.22,
-      "learning_rate": 4.009090909090909e-06,
-      "loss": 0.4575,
+      "learning_rate": 4.010909090909091e-06,
+      "loss": 0.213,
       "step": 3800
     },
     {
       "epoch": 1.22,
-      "learning_rate": 3.963636363636364e-06,
-      "loss": 0.5708,
+      "learning_rate": 3.965454545454546e-06,
+      "loss": 0.2872,
       "step": 3825
     },
     {
       "epoch": 1.23,
-      "learning_rate": 3.918181818181819e-06,
-      "loss": 0.4305,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 0.1848,
       "step": 3850
     },
     {
       "epoch": 1.23,
-      "learning_rate": 3.872727272727273e-06,
-      "loss": 0.4109,
+      "learning_rate": 3.874545454545454e-06,
+      "loss": 0.1915,
       "step": 3875
     },
     {
       "epoch": 1.24,
-      "learning_rate": 3.827272727272728e-06,
-      "loss": 0.4232,
+      "learning_rate": 3.82909090909091e-06,
+      "loss": 0.2201,
       "step": 3900
     },
     {
       "epoch": 1.24,
-      "learning_rate": 3.781818181818182e-06,
-      "loss": 0.4658,
+      "learning_rate": 3.783636363636364e-06,
+      "loss": 0.2255,
       "step": 3925
     },
     {
       "epoch": 1.25,
-      "learning_rate": 3.736363636363637e-06,
-      "loss": 0.3914,
+      "learning_rate": 3.7381818181818185e-06,
+      "loss": 0.2062,
       "step": 3950
     },
     {
       "epoch": 1.25,
-      "learning_rate": 3.690909090909091e-06,
-      "loss": 0.4737,
+      "learning_rate": 3.6927272727272727e-06,
+      "loss": 0.2402,
       "step": 3975
     },
     {
       "epoch": 1.25,
-      "learning_rate": 3.645454545454546e-06,
-      "loss": 0.4432,
+      "learning_rate": 3.6472727272727277e-06,
+      "loss": 0.1923,
       "step": 4000
     },
     {
       "epoch": 1.25,
-      "eval_loss": 0.5460641384124756,
-      "eval_runtime": 352.1387,
-      "eval_samples_per_second": 3.66,
-      "eval_steps_per_second": 1.832,
-      "eval_wer": 45.32056464150239,
+      "eval_loss": 0.34995561838150024,
+      "eval_runtime": 2323.4377,
+      "eval_samples_per_second": 0.555,
+      "eval_steps_per_second": 0.278,
+      "eval_wer": 30.632423356756423,
       "step": 4000
     },
     {
       "epoch": 1.26,
-      "learning_rate": 3.6000000000000003e-06,
-      "loss": 0.4855,
+      "learning_rate": 3.601818181818182e-06,
+      "loss": 0.2281,
       "step": 4025
     },
     {
       "epoch": 1.26,
-      "learning_rate": 3.554545454545455e-06,
-      "loss": 0.4771,
+      "learning_rate": 3.5563636363636365e-06,
+      "loss": 0.2396,
       "step": 4050
     },
     {
       "epoch": 1.27,
-      "learning_rate": 3.509090909090909e-06,
-      "loss": 0.5081,
+      "learning_rate": 3.510909090909091e-06,
+      "loss": 0.2664,
       "step": 4075
     },
     {
       "epoch": 1.27,
-      "learning_rate": 3.463636363636364e-06,
-      "loss": 0.3598,
+      "learning_rate": 3.4654545454545457e-06,
+      "loss": 0.1738,
       "step": 4100
     },
     {
       "epoch": 1.27,
-      "learning_rate": 3.4181818181818182e-06,
-      "loss": 0.4573,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 0.2027,
       "step": 4125
     },
     {
       "epoch": 1.28,
-      "learning_rate": 3.3727272727272732e-06,
-      "loss": 0.4805,
+      "learning_rate": 3.374545454545455e-06,
+      "loss": 0.2423,
       "step": 4150
     },
     {
       "epoch": 1.28,
-      "learning_rate": 3.3272727272727274e-06,
-      "loss": 0.4409,
+      "learning_rate": 3.3290909090909095e-06,
+      "loss": 0.2195,
       "step": 4175
     },
     {
       "epoch": 1.29,
-      "learning_rate": 3.281818181818182e-06,
-      "loss": 0.4803,
+      "learning_rate": 3.2836363636363637e-06,
+      "loss": 0.2307,
       "step": 4200
     },
     {
       "epoch": 1.29,
-      "learning_rate": 3.236363636363636e-06,
-      "loss": 0.4092,
+      "learning_rate": 3.2381818181818187e-06,
+      "loss": 0.1837,
       "step": 4225
     },
     {
       "epoch": 1.3,
-      "learning_rate": 3.190909090909091e-06,
-      "loss": 0.4431,
+      "learning_rate": 3.192727272727273e-06,
+      "loss": 0.2315,
       "step": 4250
     },
     {
       "epoch": 1.3,
       "learning_rate": 3.147272727272728e-06,
-      "loss": 0.4414,
+      "loss": 0.2101,
       "step": 4275
     },
     {
       "epoch": 1.3,
       "learning_rate": 3.101818181818182e-06,
-      "loss": 0.443,
+      "loss": 0.189,
       "step": 4300
     },
     {
       "epoch": 1.31,
       "learning_rate": 3.0563636363636366e-06,
-      "loss": 0.4769,
+      "loss": 0.225,
       "step": 4325
     },
     {
       "epoch": 1.31,
       "learning_rate": 3.010909090909091e-06,
-      "loss": 0.3794,
+      "loss": 0.1643,
       "step": 4350
     },
     {
       "epoch": 1.32,
       "learning_rate": 2.965454545454546e-06,
-      "loss": 0.3931,
+      "loss": 0.2064,
       "step": 4375
     },
     {
       "epoch": 1.32,
       "learning_rate": 2.92e-06,
-      "loss": 0.4102,
+      "loss": 0.1916,
       "step": 4400
     },
     {
       "epoch": 1.32,
       "learning_rate": 2.874545454545455e-06,
-      "loss": 0.5113,
+      "loss": 0.2625,
       "step": 4425
     },
     {
       "epoch": 1.33,
       "learning_rate": 2.829090909090909e-06,
-      "loss": 0.4647,
+      "loss": 0.2485,
       "step": 4450
     },
     {
       "epoch": 1.33,
       "learning_rate": 2.7836363636363638e-06,
-      "loss": 0.5214,
+      "loss": 0.2437,
       "step": 4475
     },
     {
       "epoch": 1.34,
       "learning_rate": 2.738181818181818e-06,
-      "loss": 0.4365,
+      "loss": 0.1991,
       "step": 4500
     },
     {
       "epoch": 1.34,
       "learning_rate": 2.692727272727273e-06,
-      "loss": 0.488,
+      "loss": 0.266,
       "step": 4525
     },
     {
       "epoch": 1.35,
       "learning_rate": 2.647272727272727e-06,
-      "loss": 0.4531,
+      "loss": 0.241,
       "step": 4550
     },
     {
       "epoch": 1.35,
       "learning_rate": 2.601818181818182e-06,
-      "loss": 0.4889,
+      "loss": 0.2263,
       "step": 4575
     },
     {
       "epoch": 1.35,
       "learning_rate": 2.5563636363636368e-06,
-      "loss": 0.4286,
+      "loss": 0.1896,
       "step": 4600
     },
     {
       "epoch": 1.36,
       "learning_rate": 2.510909090909091e-06,
-      "loss": 0.3778,
+      "loss": 0.1949,
       "step": 4625
     },
     {
       "epoch": 1.36,
       "learning_rate": 2.4654545454545455e-06,
-      "loss": 0.3764,
+      "loss": 0.168,
       "step": 4650
     },
     {
       "epoch": 1.37,
       "learning_rate": 2.42e-06,
-      "loss": 0.4754,
+      "loss": 0.2271,
       "step": 4675
     },
     {
       "epoch": 1.37,
       "learning_rate": 2.3745454545454547e-06,
-      "loss": 0.4664,
+      "loss": 0.2126,
       "step": 4700
     },
     {
       "epoch": 1.37,
       "learning_rate": 2.3290909090909093e-06,
-      "loss": 0.4955,
+      "loss": 0.2661,
       "step": 4725
     },
     {
       "epoch": 1.38,
       "learning_rate": 2.283636363636364e-06,
-      "loss": 0.4556,
+      "loss": 0.2146,
       "step": 4750
     },
     {
       "epoch": 1.38,
       "learning_rate": 2.238181818181818e-06,
-      "loss": 0.4609,
+      "loss": 0.2034,
       "step": 4775
     },
     {
       "epoch": 1.39,
       "learning_rate": 2.192727272727273e-06,
-      "loss": 0.3585,
+      "loss": 0.1588,
       "step": 4800
     },
     {
       "epoch": 1.39,
       "learning_rate": 2.1472727272727277e-06,
-      "loss": 0.3905,
+      "loss": 0.154,
       "step": 4825
     },
     {
       "epoch": 1.4,
       "learning_rate": 2.1018181818181823e-06,
-      "loss": 0.3687,
+      "loss": 0.1563,
       "step": 4850
     },
     {
       "epoch": 1.4,
       "learning_rate": 2.0563636363636365e-06,
-      "loss": 0.3655,
+      "loss": 0.1709,
       "step": 4875
     },
     {
       "epoch": 1.4,
       "learning_rate": 2.010909090909091e-06,
-      "loss": 0.3919,
+      "loss": 0.1699,
       "step": 4900
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.9654545454545457e-06,
-      "loss": 0.4405,
+      "loss": 0.2218,
       "step": 4925
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.9200000000000003e-06,
-      "loss": 0.4722,
+      "loss": 0.2127,
       "step": 4950
     },
     {
       "epoch": 2.0,
       "learning_rate": 1.8745454545454546e-06,
-      "loss": 0.3391,
+      "loss": 0.0894,
       "step": 4975
     },
     {
       "epoch": 2.01,
       "learning_rate": 1.8290909090909092e-06,
-      "loss": 0.3717,
+      "loss": 0.1281,
       "step": 5000
     },
     {
       "epoch": 2.01,
-      "eval_loss": 0.5240125060081482,
-      "eval_runtime": 360.6936,
-      "eval_samples_per_second": 3.574,
-      "eval_steps_per_second": 1.788,
-      "eval_wer": 45.619053541446426,
+      "eval_loss": 0.33273452520370483,
+      "eval_runtime": 2209.286,
+      "eval_samples_per_second": 0.583,
+      "eval_steps_per_second": 0.292,
+      "eval_wer": 26.8453454387165,
       "step": 5000
     },
     {
       "epoch": 2.01,
       "learning_rate": 1.7836363636363638e-06,
-      "loss": 0.2858,
+      "loss": 0.0663,
       "step": 5025
     },
     {
       "epoch": 2.02,
       "learning_rate": 1.7381818181818182e-06,
-      "loss": 0.3426,
+      "loss": 0.0889,
       "step": 5050
     },
     {
       "epoch": 2.02,
       "learning_rate": 1.6927272727272728e-06,
-      "loss": 0.3141,
+      "loss": 0.084,
       "step": 5075
     },
     {
       "epoch": 2.02,
       "learning_rate": 1.6472727272727274e-06,
-      "loss": 0.4064,
+      "loss": 0.1086,
       "step": 5100
     },
     {
       "epoch": 2.03,
       "learning_rate": 1.6018181818181818e-06,
-      "loss": 0.3487,
+      "loss": 0.0903,
       "step": 5125
     },
     {
       "epoch": 2.03,
       "learning_rate": 1.5563636363636364e-06,
-      "loss": 0.3834,
+      "loss": 0.0924,
       "step": 5150
     },
     {
       "epoch": 2.04,
       "learning_rate": 1.510909090909091e-06,
-      "loss": 0.3244,
+      "loss": 0.0874,
       "step": 5175
     },
     {
       "epoch": 2.04,
       "learning_rate": 1.4654545454545454e-06,
-      "loss": 0.3632,
+      "loss": 0.0851,
       "step": 5200
     },
     {
       "epoch": 2.05,
       "learning_rate": 1.42e-06,
-      "loss": 0.3758,
+      "loss": 0.0997,
       "step": 5225
     },
     {
       "epoch": 2.05,
       "learning_rate": 1.3745454545454548e-06,
-      "loss": 0.3333,
+      "loss": 0.0864,
       "step": 5250
     },
     {
       "epoch": 2.05,
       "learning_rate": 1.3290909090909094e-06,
-      "loss": 0.3499,
+      "loss": 0.1078,
       "step": 5275
     },
     {
       "epoch": 2.06,
       "learning_rate": 1.2836363636363637e-06,
-      "loss": 0.3138,
+      "loss": 0.0734,
       "step": 5300
     },
     {
       "epoch": 2.06,
       "learning_rate": 1.2381818181818183e-06,
-      "loss": 0.3901,
+      "loss": 0.1124,
       "step": 5325
     },
     {
       "epoch": 2.07,
       "learning_rate": 1.1927272727272727e-06,
-      "loss": 0.3502,
+      "loss": 0.0891,
       "step": 5350
     },
     {
       "epoch": 2.07,
       "learning_rate": 1.1472727272727275e-06,
-      "loss": 0.3431,
+      "loss": 0.0933,
       "step": 5375
     },
     {
       "epoch": 2.07,
       "learning_rate": 1.101818181818182e-06,
-      "loss": 0.3167,
+      "loss": 0.0791,
       "step": 5400
     },
     {
       "epoch": 2.08,
       "learning_rate": 1.0563636363636365e-06,
-      "loss": 0.2958,
+      "loss": 0.073,
       "step": 5425
     },
     {
       "epoch": 2.08,
       "learning_rate": 1.0109090909090911e-06,
-      "loss": 0.3056,
+      "loss": 0.0724,
       "step": 5450
     },
     {
       "epoch": 2.09,
       "learning_rate": 9.654545454545455e-07,
-      "loss": 0.3011,
+      "loss": 0.0618,
       "step": 5475
     },
     {
       "epoch": 2.09,
       "learning_rate": 9.200000000000001e-07,
-      "loss": 0.2788,
+      "loss": 0.0672,
       "step": 5500
     },
     {
       "epoch": 2.1,
       "learning_rate": 8.745454545454546e-07,
-      "loss": 0.3341,
+      "loss": 0.0978,
       "step": 5525
     },
     {
       "epoch": 2.1,
       "learning_rate": 8.290909090909091e-07,
-      "loss": 0.4218,
+      "loss": 0.0849,
       "step": 5550
     },
     {
       "epoch": 2.1,
       "learning_rate": 7.836363636363637e-07,
-      "loss": 0.3287,
+      "loss": 0.0858,
       "step": 5575
     },
     {
       "epoch": 2.11,
       "learning_rate": 7.381818181818183e-07,
-      "loss": 0.3101,
+      "loss": 0.0819,
       "step": 5600
     },
     {
       "epoch": 2.11,
       "learning_rate": 6.927272727272729e-07,
-      "loss": 0.3661,
+      "loss": 0.0885,
       "step": 5625
     },
     {
       "epoch": 2.12,
       "learning_rate": 6.472727272727273e-07,
-      "loss": 0.3343,
+      "loss": 0.0774,
       "step": 5650
     },
     {
       "epoch": 2.12,
       "learning_rate": 6.018181818181819e-07,
-      "loss": 0.4131,
+      "loss": 0.1247,
       "step": 5675
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.563636363636364e-07,
-      "loss": 0.2915,
+      "loss": 0.0896,
       "step": 5700
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.109090909090909e-07,
-      "loss": 0.3419,
+      "loss": 0.0797,
       "step": 5725
     },
     {
       "epoch": 2.13,
       "learning_rate": 4.6545454545454546e-07,
-      "loss": 0.3723,
+      "loss": 0.1102,
       "step": 5750
     },
     {
       "epoch": 2.14,
       "learning_rate": 4.2000000000000006e-07,
-      "loss": 0.3391,
+      "loss": 0.0942,
       "step": 5775
     },
     {
       "epoch": 2.14,
       "learning_rate": 3.745454545454546e-07,
-      "loss": 0.3551,
+      "loss": 0.0735,
       "step": 5800
     },
     {
       "epoch": 2.15,
       "learning_rate": 3.290909090909091e-07,
-      "loss": 0.3701,
+      "loss": 0.1001,
       "step": 5825
     },
     {
       "epoch": 2.15,
       "learning_rate": 2.836363636363637e-07,
-      "loss": 0.3364,
+      "loss": 0.072,
       "step": 5850
     },
     {
       "epoch": 2.15,
       "learning_rate": 2.381818181818182e-07,
-      "loss": 0.3804,
+      "loss": 0.0961,
       "step": 5875
     },
     {
       "epoch": 2.16,
       "learning_rate": 1.9272727272727272e-07,
-      "loss": 0.3342,
+      "loss": 0.0892,
       "step": 5900
     },
     {
       "epoch": 2.16,
       "learning_rate": 1.4727272727272728e-07,
-      "loss": 0.3559,
+      "loss": 0.0795,
       "step": 5925
     },
     {
       "epoch": 2.17,
       "learning_rate": 1.0181818181818183e-07,
-      "loss": 0.3329,
+      "loss": 0.092,
       "step": 5950
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.636363636363637e-08,
-      "loss": 0.351,
+      "loss": 0.1005,
       "step": 5975
     },
     {
       "epoch": 2.17,
       "learning_rate": 1.090909090909091e-08,
-      "loss": 0.3244,
+      "loss": 0.0701,
       "step": 6000
     },
     {
       "epoch": 2.17,
-      "eval_loss": 0.5219861268997192,
-      "eval_runtime": 356.3084,
-      "eval_samples_per_second": 3.618,
-      "eval_steps_per_second": 1.81,
-      "eval_wer": 43.8592127355264,
+      "eval_loss": 0.3435710072517395,
+      "eval_runtime": 2232.4891,
+      "eval_samples_per_second": 0.577,
+      "eval_steps_per_second": 0.289,
+      "eval_wer": 27.27442323238605,
       "step": 6000
     },
     {
       "epoch": 2.17,
       "step": 6000,
-      "total_flos": 2.954266088885453e+17,
-      "train_loss": 0.7047004199822744,
-      "train_runtime": 3798.8957,
-      "train_samples_per_second": 3.159,
-      "train_steps_per_second": 1.579
+      "total_flos": 1.2247259017327411e+19,
+      "train_loss": 0.38768368325630825,
+      "train_runtime": 19861.2799,
+      "train_samples_per_second": 0.604,
+      "train_steps_per_second": 0.302
     }
   ],
   "max_steps": 6000,
   "num_train_epochs": 9223372036854775807,
-  "total_flos": 2.954266088885453e+17,
+  "total_flos": 1.2247259017327411e+19,
   "trial_name": null,
   "trial_params": null
 }