End of training

all_results.json

@@ -1,12 +1,12 @@
 {
-    "epoch": 11.88,
-    "eval_loss": 0.12232152372598648,
-    "eval_runtime": 1042.7201,
-    "eval_samples_per_second": 4.836,
-    "eval_steps_per_second": 0.807,
-    "eval_wer": 4.711435314310234,
-    "train_loss": 0.26830010027885437,
-    "train_runtime": 12204.8548,
-    "train_samples_per_second": 4.916,
-    "train_steps_per_second": 0.41
+    "epoch": 7.26,
+    "eval_loss": 0.25948023796081543,
+    "eval_runtime": 2105.9954,
+    "eval_samples_per_second": 1.72,
+    "eval_steps_per_second": 0.287,
+    "eval_wer": 14.345525649873474,
+    "train_loss": 0.30027096328735353,
+    "train_runtime": 47995.2614,
+    "train_samples_per_second": 1.25,
+    "train_steps_per_second": 0.104
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 11.88,
-    "eval_loss": 0.12232152372598648,
-    "eval_runtime": 1042.7201,
-    "eval_samples_per_second": 4.836,
-    "eval_steps_per_second": 0.807,
-    "eval_wer": 4.711435314310234
+    "epoch": 7.26,
+    "eval_loss": 0.25948023796081543,
+    "eval_runtime": 2105.9954,
+    "eval_samples_per_second": 1.72,
+    "eval_steps_per_second": 0.287,
+    "eval_wer": 14.345525649873474
 }

runs/Aug10_18-47-27_dsai-server/events.out.tfevents.1691719290.dsai-server

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e8338ec150a5d69ba8ca402bebb1b7e4ee03ab95dc910ed6c9309f8e2d560c8
+size 40

train_results.json

@@ -1,7 +1,7 @@
 {
-    "epoch": 11.88,
-    "train_loss": 0.26830010027885437,
-    "train_runtime": 12204.8548,
-    "train_samples_per_second": 4.916,
-    "train_steps_per_second": 0.41
+    "epoch": 7.26,
+    "train_loss": 0.30027096328735353,
+    "train_runtime": 47995.2614,
+    "train_samples_per_second": 1.25,
+    "train_steps_per_second": 0.104
 }

trainer_state.json

@@ -1,1270 +1,1270 @@
 {
-  "best_metric": 4.711435314310234,
+  "best_metric": 14.345525649873474,
   "best_model_checkpoint": "./output/small/cv9-special-batch12-lr6-small/checkpoint-5000",
-  "epoch": 11.876484560570072,
+  "epoch": 7.256894049346879,
   "global_step": 5000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.06,
+      "epoch": 0.04,
       "learning_rate": 4.4e-08,
-      "loss": 1.4193,
+      "loss": 1.33,
       "step": 25
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.07,
       "learning_rate": 9.4e-08,
-      "loss": 1.4255,
+      "loss": 1.4048,
       "step": 50
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.11,
       "learning_rate": 1.44e-07,
-      "loss": 1.43,
+      "loss": 1.3935,
       "step": 75
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.15,
       "learning_rate": 1.94e-07,
-      "loss": 1.2985,
+      "loss": 1.3644,
       "step": 100
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.18,
       "learning_rate": 2.4399999999999996e-07,
-      "loss": 1.1832,
+      "loss": 1.3174,
       "step": 125
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.22,
       "learning_rate": 2.9399999999999996e-07,
-      "loss": 1.155,
+      "loss": 1.1595,
       "step": 150
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.25,
       "learning_rate": 3.4399999999999996e-07,
-      "loss": 1.1072,
+      "loss": 1.1359,
       "step": 175
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.29,
       "learning_rate": 3.94e-07,
-      "loss": 0.9823,
+      "loss": 1.0684,
       "step": 200
     },
     {
-      "epoch": 0.53,
+      "epoch": 0.33,
       "learning_rate": 4.44e-07,
-      "loss": 0.8871,
+      "loss": 0.8967,
       "step": 225
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.36,
       "learning_rate": 4.94e-07,
-      "loss": 0.7735,
+      "loss": 0.8263,
       "step": 250
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.4,
       "learning_rate": 5.44e-07,
-      "loss": 0.6804,
+      "loss": 0.6897,
       "step": 275
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.44,
       "learning_rate": 5.939999999999999e-07,
-      "loss": 0.6176,
+      "loss": 0.598,
       "step": 300
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.47,
       "learning_rate": 6.44e-07,
-      "loss": 0.5623,
+      "loss": 0.5159,
       "step": 325
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.51,
       "learning_rate": 6.939999999999999e-07,
-      "loss": 0.5292,
+      "loss": 0.5284,
       "step": 350
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.54,
       "learning_rate": 7.44e-07,
-      "loss": 0.5682,
+      "loss": 0.4766,
       "step": 375
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.58,
       "learning_rate": 7.94e-07,
-      "loss": 0.4774,
+      "loss": 0.5041,
       "step": 400
     },
     {
-      "epoch": 1.01,
+      "epoch": 0.62,
       "learning_rate": 8.439999999999999e-07,
-      "loss": 0.491,
+      "loss": 0.4576,
       "step": 425
     },
     {
-      "epoch": 1.07,
+      "epoch": 0.65,
       "learning_rate": 8.939999999999999e-07,
-      "loss": 0.4744,
+      "loss": 0.4458,
       "step": 450
     },
     {
-      "epoch": 1.13,
+      "epoch": 0.69,
       "learning_rate": 9.439999999999999e-07,
-      "loss": 0.4452,
+      "loss": 0.4805,
       "step": 475
     },
     {
-      "epoch": 1.19,
+      "epoch": 0.73,
       "learning_rate": 9.94e-07,
-      "loss": 0.4444,
+      "loss": 0.4435,
       "step": 500
     },
     {
-      "epoch": 1.25,
+      "epoch": 0.76,
       "learning_rate": 9.95111111111111e-07,
-      "loss": 0.398,
+      "loss": 0.4378,
       "step": 525
     },
     {
-      "epoch": 1.31,
+      "epoch": 0.8,
       "learning_rate": 9.895555555555554e-07,
-      "loss": 0.4174,
+      "loss": 0.3972,
       "step": 550
     },
     {
-      "epoch": 1.37,
+      "epoch": 0.83,
       "learning_rate": 9.84e-07,
-      "loss": 0.4164,
+      "loss": 0.4128,
       "step": 575
     },
     {
-      "epoch": 1.43,
+      "epoch": 0.87,
       "learning_rate": 9.784444444444444e-07,
-      "loss": 0.366,
+      "loss": 0.404,
       "step": 600
     },
     {
-      "epoch": 1.48,
+      "epoch": 0.91,
       "learning_rate": 9.728888888888888e-07,
-      "loss": 0.37,
+      "loss": 0.3985,
       "step": 625
     },
     {
-      "epoch": 1.54,
+      "epoch": 0.94,
       "learning_rate": 9.673333333333332e-07,
-      "loss": 0.4138,
+      "loss": 0.4062,
       "step": 650
     },
     {
-      "epoch": 1.6,
+      "epoch": 0.98,
       "learning_rate": 9.617777777777776e-07,
-      "loss": 0.3824,
+      "loss": 0.4198,
       "step": 675
     },
     {
-      "epoch": 1.66,
+      "epoch": 1.02,
       "learning_rate": 9.562222222222223e-07,
-      "loss": 0.3827,
+      "loss": 0.4174,
       "step": 700
     },
     {
-      "epoch": 1.72,
+      "epoch": 1.05,
       "learning_rate": 9.506666666666667e-07,
-      "loss": 0.3889,
+      "loss": 0.3474,
       "step": 725
     },
     {
-      "epoch": 1.78,
+      "epoch": 1.09,
       "learning_rate": 9.451111111111111e-07,
-      "loss": 0.3985,
+      "loss": 0.3489,
       "step": 750
     },
     {
-      "epoch": 1.84,
+      "epoch": 1.12,
       "learning_rate": 9.395555555555556e-07,
-      "loss": 0.4033,
+      "loss": 0.363,
       "step": 775
     },
     {
-      "epoch": 1.9,
+      "epoch": 1.16,
       "learning_rate": 9.34e-07,
-      "loss": 0.4152,
+      "loss": 0.3143,
       "step": 800
     },
     {
-      "epoch": 1.96,
+      "epoch": 1.2,
       "learning_rate": 9.284444444444444e-07,
-      "loss": 0.4014,
+      "loss": 0.3435,
       "step": 825
     },
     {
-      "epoch": 2.02,
+      "epoch": 1.23,
       "learning_rate": 9.228888888888888e-07,
-      "loss": 0.343,
+      "loss": 0.3572,
       "step": 850
     },
     {
-      "epoch": 2.08,
+      "epoch": 1.27,
       "learning_rate": 9.173333333333333e-07,
-      "loss": 0.3204,
+      "loss": 0.3339,
       "step": 875
     },
     {
-      "epoch": 2.14,
+      "epoch": 1.31,
       "learning_rate": 9.117777777777778e-07,
-      "loss": 0.3402,
+      "loss": 0.3632,
       "step": 900
     },
     {
-      "epoch": 2.2,
+      "epoch": 1.34,
       "learning_rate": 9.062222222222222e-07,
-      "loss": 0.316,
+      "loss": 0.3272,
       "step": 925
     },
     {
-      "epoch": 2.26,
+      "epoch": 1.38,
       "learning_rate": 9.006666666666666e-07,
-      "loss": 0.3197,
+      "loss": 0.3494,
       "step": 950
     },
     {
-      "epoch": 2.32,
+      "epoch": 1.42,
       "learning_rate": 8.95111111111111e-07,
-      "loss": 0.3238,
+      "loss": 0.3234,
       "step": 975
     },
     {
-      "epoch": 2.38,
+      "epoch": 1.45,
       "learning_rate": 8.895555555555555e-07,
-      "loss": 0.3296,
+      "loss": 0.3456,
       "step": 1000
     },
     {
-      "epoch": 2.38,
-      "eval_loss": 0.30187273025512695,
-      "eval_runtime": 825.6555,
-      "eval_samples_per_second": 6.108,
-      "eval_steps_per_second": 1.019,
-      "eval_wer": 13.14739326324823,
+      "epoch": 1.45,
+      "eval_loss": 0.28774985671043396,
+      "eval_runtime": 2339.756,
+      "eval_samples_per_second": 1.548,
+      "eval_steps_per_second": 0.258,
+      "eval_wer": 16.057050839659535,
       "step": 1000
     },
     {
-      "epoch": 2.43,
+      "epoch": 1.49,
       "learning_rate": 8.839999999999999e-07,
-      "loss": 0.337,
+      "loss": 0.3324,
       "step": 1025
     },
     {
-      "epoch": 2.49,
+      "epoch": 1.52,
       "learning_rate": 8.784444444444444e-07,
-      "loss": 0.2967,
+      "loss": 0.3465,
       "step": 1050
     },
     {
-      "epoch": 2.55,
+      "epoch": 1.56,
       "learning_rate": 8.728888888888889e-07,
-      "loss": 0.3238,
+      "loss": 0.3212,
       "step": 1075
     },
     {
-      "epoch": 2.61,
+      "epoch": 1.6,
       "learning_rate": 8.673333333333332e-07,
-      "loss": 0.3102,
+      "loss": 0.3115,
       "step": 1100
     },
     {
-      "epoch": 2.67,
+      "epoch": 1.63,
       "learning_rate": 8.617777777777777e-07,
-      "loss": 0.3003,
+      "loss": 0.3248,
       "step": 1125
     },
     {
-      "epoch": 2.73,
+      "epoch": 1.67,
       "learning_rate": 8.562222222222222e-07,
-      "loss": 0.3089,
+      "loss": 0.3355,
       "step": 1150
     },
     {
-      "epoch": 2.79,
+      "epoch": 1.71,
       "learning_rate": 8.506666666666667e-07,
-      "loss": 0.3021,
+      "loss": 0.3348,
       "step": 1175
     },
     {
-      "epoch": 2.85,
+      "epoch": 1.74,
       "learning_rate": 8.451111111111111e-07,
-      "loss": 0.3172,
+      "loss": 0.3514,
       "step": 1200
     },
     {
-      "epoch": 2.91,
+      "epoch": 1.78,
       "learning_rate": 8.395555555555556e-07,
-      "loss": 0.3249,
+      "loss": 0.3329,
       "step": 1225
     },
     {
-      "epoch": 2.97,
+      "epoch": 1.81,
       "learning_rate": 8.34e-07,
-      "loss": 0.2983,
+      "loss": 0.3165,
       "step": 1250
     },
     {
-      "epoch": 3.03,
+      "epoch": 1.85,
       "learning_rate": 8.284444444444444e-07,
-      "loss": 0.3062,
+      "loss": 0.3505,
       "step": 1275
     },
     {
-      "epoch": 3.09,
+      "epoch": 1.89,
       "learning_rate": 8.228888888888889e-07,
-      "loss": 0.2794,
+      "loss": 0.336,
       "step": 1300
     },
     {
-      "epoch": 3.15,
+      "epoch": 1.92,
       "learning_rate": 8.173333333333333e-07,
-      "loss": 0.2703,
+      "loss": 0.2869,
       "step": 1325
     },
     {
-      "epoch": 3.21,
+      "epoch": 1.96,
       "learning_rate": 8.117777777777778e-07,
-      "loss": 0.2569,
+      "loss": 0.3571,
       "step": 1350
     },
     {
-      "epoch": 3.27,
+      "epoch": 2.0,
       "learning_rate": 8.062222222222221e-07,
-      "loss": 0.2508,
+      "loss": 0.3251,
       "step": 1375
     },
     {
-      "epoch": 3.33,
+      "epoch": 2.03,
       "learning_rate": 8.006666666666666e-07,
-      "loss": 0.2639,
+      "loss": 0.2643,
       "step": 1400
     },
     {
-      "epoch": 3.38,
+      "epoch": 2.07,
       "learning_rate": 7.95111111111111e-07,
-      "loss": 0.2583,
+      "loss": 0.2766,
       "step": 1425
     },
     {
-      "epoch": 3.44,
+      "epoch": 2.1,
       "learning_rate": 7.895555555555555e-07,
-      "loss": 0.2641,
+      "loss": 0.2818,
       "step": 1450
     },
     {
-      "epoch": 3.5,
+      "epoch": 2.14,
       "learning_rate": 7.84e-07,
-      "loss": 0.2433,
+      "loss": 0.2777,
       "step": 1475
     },
     {
-      "epoch": 3.56,
+      "epoch": 2.18,
       "learning_rate": 7.784444444444444e-07,
-      "loss": 0.2846,
+      "loss": 0.2959,
       "step": 1500
     },
     {
-      "epoch": 3.62,
+      "epoch": 2.21,
       "learning_rate": 7.728888888888888e-07,
-      "loss": 0.2616,
+      "loss": 0.2794,
       "step": 1525
     },
     {
-      "epoch": 3.68,
+      "epoch": 2.25,
       "learning_rate": 7.673333333333332e-07,
-      "loss": 0.2548,
+      "loss": 0.2803,
       "step": 1550
     },
     {
-      "epoch": 3.74,
+      "epoch": 2.29,
       "learning_rate": 7.617777777777778e-07,
-      "loss": 0.2702,
+      "loss": 0.2553,
       "step": 1575
     },
     {
-      "epoch": 3.8,
+      "epoch": 2.32,
       "learning_rate": 7.562222222222222e-07,
-      "loss": 0.2762,
+      "loss": 0.2529,
       "step": 1600
     },
     {
-      "epoch": 3.86,
+      "epoch": 2.36,
       "learning_rate": 7.506666666666667e-07,
-      "loss": 0.2907,
+      "loss": 0.2604,
       "step": 1625
     },
     {
-      "epoch": 3.92,
+      "epoch": 2.39,
       "learning_rate": 7.451111111111111e-07,
-      "loss": 0.2648,
+      "loss": 0.2903,
       "step": 1650
     },
     {
-      "epoch": 3.98,
+      "epoch": 2.43,
       "learning_rate": 7.395555555555555e-07,
-      "loss": 0.2408,
+      "loss": 0.2517,
       "step": 1675
     },
     {
-      "epoch": 4.04,
+      "epoch": 2.47,
       "learning_rate": 7.34e-07,
-      "loss": 0.2274,
+      "loss": 0.291,
       "step": 1700
     },
     {
-      "epoch": 4.1,
+      "epoch": 2.5,
       "learning_rate": 7.284444444444444e-07,
-      "loss": 0.2322,
+      "loss": 0.245,
       "step": 1725
     },
     {
-      "epoch": 4.16,
+      "epoch": 2.54,
       "learning_rate": 7.228888888888889e-07,
-      "loss": 0.2313,
+      "loss": 0.2662,
       "step": 1750
     },
     {
-      "epoch": 4.22,
+      "epoch": 2.58,
       "learning_rate": 7.173333333333333e-07,
-      "loss": 0.2296,
+      "loss": 0.2815,
       "step": 1775
     },
     {
-      "epoch": 4.28,
+      "epoch": 2.61,
       "learning_rate": 7.117777777777777e-07,
-      "loss": 0.2444,
+      "loss": 0.2606,
       "step": 1800
     },
     {
-      "epoch": 4.33,
+      "epoch": 2.65,
       "learning_rate": 7.062222222222222e-07,
-      "loss": 0.2252,
+      "loss": 0.2766,
       "step": 1825
     },
     {
-      "epoch": 4.39,
+      "epoch": 2.69,
       "learning_rate": 7.006666666666666e-07,
-      "loss": 0.2257,
+      "loss": 0.2815,
       "step": 1850
     },
     {
-      "epoch": 4.45,
+      "epoch": 2.72,
       "learning_rate": 6.951111111111111e-07,
-      "loss": 0.2302,
+      "loss": 0.2982,
       "step": 1875
     },
     {
-      "epoch": 4.51,
+      "epoch": 2.76,
       "learning_rate": 6.895555555555555e-07,
-      "loss": 0.2272,
+      "loss": 0.2849,
       "step": 1900
     },
     {
-      "epoch": 4.57,
+      "epoch": 2.79,
       "learning_rate": 6.84e-07,
-      "loss": 0.2377,
+      "loss": 0.2792,
       "step": 1925
     },
     {
-      "epoch": 4.63,
+      "epoch": 2.83,
       "learning_rate": 6.784444444444443e-07,
-      "loss": 0.2111,
+      "loss": 0.2775,
       "step": 1950
     },
     {
-      "epoch": 4.69,
+      "epoch": 2.87,
       "learning_rate": 6.728888888888888e-07,
-      "loss": 0.2443,
+      "loss": 0.2516,
       "step": 1975
     },
     {
-      "epoch": 4.75,
+      "epoch": 2.9,
       "learning_rate": 6.673333333333334e-07,
-      "loss": 0.2247,
+      "loss": 0.2388,
       "step": 2000
     },
     {
-      "epoch": 4.75,
-      "eval_loss": 0.20599021017551422,
-      "eval_runtime": 1037.1517,
-      "eval_samples_per_second": 4.862,
-      "eval_steps_per_second": 0.811,
-      "eval_wer": 8.704140742329972,
+      "epoch": 2.9,
+      "eval_loss": 0.26580867171287537,
+      "eval_runtime": 2332.4486,
+      "eval_samples_per_second": 1.553,
+      "eval_steps_per_second": 0.259,
+      "eval_wer": 15.044858523119393,
       "step": 2000
     },
     {
-      "epoch": 4.81,
+      "epoch": 2.94,
       "learning_rate": 6.617777777777778e-07,
-      "loss": 0.2064,
+      "loss": 0.2649,
       "step": 2025
     },
     {
-      "epoch": 4.87,
+      "epoch": 2.98,
       "learning_rate": 6.562222222222223e-07,
-      "loss": 0.2275,
+      "loss": 0.2612,
       "step": 2050
     },
     {
-      "epoch": 4.93,
+      "epoch": 3.01,
       "learning_rate": 6.506666666666666e-07,
-      "loss": 0.2277,
+      "loss": 0.2679,
       "step": 2075
     },
     {
-      "epoch": 4.99,
+      "epoch": 3.05,
       "learning_rate": 6.451111111111111e-07,
-      "loss": 0.2226,
+      "loss": 0.2426,
       "step": 2100
     },
     {
-      "epoch": 5.05,
+      "epoch": 3.08,
       "learning_rate": 6.395555555555555e-07,
-      "loss": 0.2078,
+      "loss": 0.236,
       "step": 2125
     },
     {
-      "epoch": 5.11,
+      "epoch": 3.12,
       "learning_rate": 6.34e-07,
-      "loss": 0.2107,
+      "loss": 0.2336,
       "step": 2150
     },
     {
-      "epoch": 5.17,
+      "epoch": 3.16,
       "learning_rate": 6.284444444444445e-07,
-      "loss": 0.1824,
+      "loss": 0.2178,
       "step": 2175
     },
     {
-      "epoch": 5.23,
+      "epoch": 3.19,
       "learning_rate": 6.228888888888889e-07,
-      "loss": 0.1923,
+      "loss": 0.2155,
       "step": 2200
     },
     {
-      "epoch": 5.29,
+      "epoch": 3.23,
       "learning_rate": 6.173333333333333e-07,
-      "loss": 0.2209,
+      "loss": 0.2377,
       "step": 2225
     },
     {
-      "epoch": 5.34,
+      "epoch": 3.27,
       "learning_rate": 6.117777777777777e-07,
-      "loss": 0.185,
+      "loss": 0.2223,
       "step": 2250
     },
     {
-      "epoch": 5.4,
+      "epoch": 3.3,
       "learning_rate": 6.062222222222222e-07,
-      "loss": 0.1951,
+      "loss": 0.2255,
       "step": 2275
     },
     {
-      "epoch": 5.46,
+      "epoch": 3.34,
       "learning_rate": 6.006666666666666e-07,
-      "loss": 0.1872,
+      "loss": 0.2168,
       "step": 2300
     },
     {
-      "epoch": 5.52,
+      "epoch": 3.37,
       "learning_rate": 5.951111111111111e-07,
-      "loss": 0.204,
+      "loss": 0.2243,
       "step": 2325
     },
     {
-      "epoch": 5.58,
+      "epoch": 3.41,
       "learning_rate": 5.895555555555555e-07,
-      "loss": 0.2137,
+      "loss": 0.2455,
       "step": 2350
     },
     {
-      "epoch": 5.64,
+      "epoch": 3.45,
       "learning_rate": 5.839999999999999e-07,
-      "loss": 0.2045,
+      "loss": 0.2472,
       "step": 2375
     },
     {
-      "epoch": 5.7,
+      "epoch": 3.48,
       "learning_rate": 5.784444444444444e-07,
-      "loss": 0.2049,
+      "loss": 0.2481,
       "step": 2400
     },
     {
-      "epoch": 5.76,
+      "epoch": 3.52,
       "learning_rate": 5.728888888888888e-07,
-      "loss": 0.2005,
+      "loss": 0.2264,
       "step": 2425
     },
     {
-      "epoch": 5.82,
+      "epoch": 3.56,
       "learning_rate": 5.673333333333334e-07,
-      "loss": 0.1907,
+      "loss": 0.2448,
       "step": 2450
     },
     {
-      "epoch": 5.88,
+      "epoch": 3.59,
       "learning_rate": 5.617777777777778e-07,
-      "loss": 0.193,
+      "loss": 0.2249,
       "step": 2475
     },
     {
-      "epoch": 5.94,
+      "epoch": 3.63,
       "learning_rate": 5.562222222222222e-07,
-      "loss": 0.2018,
+      "loss": 0.2489,
       "step": 2500
     },
     {
-      "epoch": 6.0,
+      "epoch": 3.66,
       "learning_rate": 5.506666666666666e-07,
-      "loss": 0.1988,
+      "loss": 0.2208,
       "step": 2525
     },
     {
-      "epoch": 6.06,
+      "epoch": 3.7,
       "learning_rate": 5.451111111111111e-07,
-      "loss": 0.1784,
+      "loss": 0.2492,
       "step": 2550
     },
     {
-      "epoch": 6.12,
+      "epoch": 3.74,
       "learning_rate": 5.395555555555556e-07,
-      "loss": 0.1668,
+      "loss": 0.2288,
       "step": 2575
     },
     {
-      "epoch": 6.18,
+      "epoch": 3.77,
       "learning_rate": 5.34e-07,
-      "loss": 0.189,
+      "loss": 0.222,
       "step": 2600
     },
     {
-      "epoch": 6.24,
+      "epoch": 3.81,
       "learning_rate": 5.284444444444445e-07,
-      "loss": 0.1727,
+      "loss": 0.2518,
       "step": 2625
     },
     {
-      "epoch": 6.29,
+      "epoch": 3.85,
       "learning_rate": 5.228888888888888e-07,
-      "loss": 0.1645,
+      "loss": 0.2361,
       "step": 2650
     },
     {
-      "epoch": 6.35,
+      "epoch": 3.88,
       "learning_rate": 5.173333333333333e-07,
-      "loss": 0.1644,
+      "loss": 0.2276,
       "step": 2675
     },
     {
-      "epoch": 6.41,
+      "epoch": 3.92,
       "learning_rate": 5.117777777777777e-07,
-      "loss": 0.1707,
+      "loss": 0.2282,
       "step": 2700
     },
     {
-      "epoch": 6.47,
+      "epoch": 3.96,
       "learning_rate": 5.062222222222222e-07,
-      "loss": 0.1783,
+      "loss": 0.2072,
       "step": 2725
     },
     {
-      "epoch": 6.53,
+      "epoch": 3.99,
       "learning_rate": 5.006666666666667e-07,
-      "loss": 0.1775,
+      "loss": 0.2265,
       "step": 2750
     },
     {
-      "epoch": 6.59,
+      "epoch": 4.03,
       "learning_rate": 4.951111111111111e-07,
-      "loss": 0.1714,
+      "loss": 0.1864,
       "step": 2775
     },
     {
-      "epoch": 6.65,
+      "epoch": 4.06,
       "learning_rate": 4.895555555555555e-07,
-      "loss": 0.173,
+      "loss": 0.2122,
       "step": 2800
     },
     {
-      "epoch": 6.71,
+      "epoch": 4.1,
       "learning_rate": 4.839999999999999e-07,
-      "loss": 0.1759,
+      "loss": 0.2117,
       "step": 2825
     },
     {
-      "epoch": 6.77,
+      "epoch": 4.14,
       "learning_rate": 4.784444444444444e-07,
-      "loss": 0.174,
+      "loss": 0.2124,
       "step": 2850
     },
     {
-      "epoch": 6.83,
+      "epoch": 4.17,
       "learning_rate": 4.728888888888889e-07,
-      "loss": 0.1837,
+      "loss": 0.2166,
       "step": 2875
     },
     {
-      "epoch": 6.89,
+      "epoch": 4.21,
       "learning_rate": 4.673333333333333e-07,
-      "loss": 0.1818,
+      "loss": 0.2116,
       "step": 2900
     },
     {
-      "epoch": 6.95,
+      "epoch": 4.25,
       "learning_rate": 4.6177777777777777e-07,
-      "loss": 0.2009,
+      "loss": 0.1993,
       "step": 2925
     },
     {
-      "epoch": 7.01,
+      "epoch": 4.28,
       "learning_rate": 4.5622222222222217e-07,
-      "loss": 0.1756,
+      "loss": 0.2089,
       "step": 2950
     },
     {
-      "epoch": 7.07,
+      "epoch": 4.32,
       "learning_rate": 4.506666666666666e-07,
-      "loss": 0.1515,
+      "loss": 0.1977,
       "step": 2975
     },
     {
-      "epoch": 7.13,
+      "epoch": 4.35,
       "learning_rate": 4.451111111111111e-07,
-      "loss": 0.1526,
+      "loss": 0.2224,
       "step": 3000
     },
     {
-      "epoch": 7.13,
-      "eval_loss": 0.15617816150188446,
-      "eval_runtime": 1028.0416,
-      "eval_samples_per_second": 4.905,
-      "eval_steps_per_second": 0.818,
-      "eval_wer": 6.2196953443467065,
+      "epoch": 4.35,
+      "eval_loss": 0.2609769105911255,
+      "eval_runtime": 2335.1545,
+      "eval_samples_per_second": 1.551,
+      "eval_steps_per_second": 0.259,
+      "eval_wer": 14.414538762364849,
       "step": 3000
     },
     {
-      "epoch": 7.19,
+      "epoch": 4.39,
       "learning_rate": 4.3955555555555554e-07,
-      "loss": 0.1682,
+      "loss": 0.1923,
       "step": 3025
     },
     {
-      "epoch": 7.24,
+      "epoch": 4.43,
       "learning_rate": 4.34e-07,
-      "loss": 0.1561,
+      "loss": 0.2334,
       "step": 3050
     },
     {
-      "epoch": 7.3,
+      "epoch": 4.46,
       "learning_rate": 4.2844444444444445e-07,
-      "loss": 0.1673,
+      "loss": 0.1788,
       "step": 3075
     },
     {
-      "epoch": 7.36,
+      "epoch": 4.5,
       "learning_rate": 4.2288888888888886e-07,
-      "loss": 0.173,
+      "loss": 0.186,
       "step": 3100
     },
     {
-      "epoch": 7.42,
+      "epoch": 4.54,
       "learning_rate": 4.173333333333333e-07,
-      "loss": 0.1634,
+      "loss": 0.1885,
       "step": 3125
     },
     {
-      "epoch": 7.48,
+      "epoch": 4.57,
       "learning_rate": 4.1177777777777777e-07,
-      "loss": 0.154,
+      "loss": 0.2059,
       "step": 3150
     },
     {
-      "epoch": 7.54,
+      "epoch": 4.61,
       "learning_rate": 4.0622222222222217e-07,
-      "loss": 0.15,
+      "loss": 0.1958,
       "step": 3175
     },
     {
-      "epoch": 7.6,
+      "epoch": 4.64,
       "learning_rate": 4.0066666666666663e-07,
-      "loss": 0.1621,
+      "loss": 0.2102,
       "step": 3200
     },
     {
-      "epoch": 7.66,
+      "epoch": 4.68,
       "learning_rate": 3.9511111111111114e-07,
-      "loss": 0.1502,
+      "loss": 0.1949,
       "step": 3225
     },
     {
-      "epoch": 7.72,
+      "epoch": 4.72,
       "learning_rate": 3.8955555555555554e-07,
-      "loss": 0.1747,
+      "loss": 0.204,
       "step": 3250
     },
     {
-      "epoch": 7.78,
+      "epoch": 4.75,
       "learning_rate": 3.84e-07,
-      "loss": 0.1662,
+      "loss": 0.2164,
       "step": 3275
     },
     {
-      "epoch": 7.84,
+      "epoch": 4.79,
       "learning_rate": 3.7844444444444445e-07,
-      "loss": 0.1667,
+      "loss": 0.2081,
       "step": 3300
     },
     {
-      "epoch": 7.9,
+      "epoch": 4.83,
       "learning_rate": 3.7288888888888886e-07,
-      "loss": 0.1527,
+      "loss": 0.2139,
       "step": 3325
     },
     {
-      "epoch": 7.96,
+      "epoch": 4.86,
       "learning_rate": 3.673333333333333e-07,
-      "loss": 0.1332,
+      "loss": 0.1991,
       "step": 3350
     },
     {
-      "epoch": 8.02,
+      "epoch": 4.9,
       "learning_rate": 3.617777777777777e-07,
-      "loss": 0.1455,
+      "loss": 0.2241,
       "step": 3375
     },
     {
-      "epoch": 8.08,
+      "epoch": 4.93,
       "learning_rate": 3.5622222222222223e-07,
-      "loss": 0.1437,
+      "loss": 0.1938,
       "step": 3400
     },
     {
-      "epoch": 8.14,
+      "epoch": 4.97,
       "learning_rate": 3.506666666666667e-07,
-      "loss": 0.1326,
+      "loss": 0.2083,
       "step": 3425
     },
     {
-      "epoch": 8.19,
+      "epoch": 5.01,
       "learning_rate": 3.451111111111111e-07,
-      "loss": 0.1668,
+      "loss": 0.1856,
       "step": 3450
     },
     {
-      "epoch": 8.25,
+      "epoch": 5.04,
       "learning_rate": 3.3955555555555554e-07,
-      "loss": 0.1561,
+      "loss": 0.1823,
       "step": 3475
     },
     {
-      "epoch": 8.31,
+      "epoch": 5.08,
       "learning_rate": 3.34e-07,
-      "loss": 0.1428,
+      "loss": 0.1829,
       "step": 3500
     },
     {
-      "epoch": 8.37,
+      "epoch": 5.12,
       "learning_rate": 3.284444444444444e-07,
-      "loss": 0.1512,
+      "loss": 0.2002,
       "step": 3525
     },
     {
-      "epoch": 8.43,
+      "epoch": 5.15,
       "learning_rate": 3.2288888888888886e-07,
-      "loss": 0.1399,
+      "loss": 0.1843,
       "step": 3550
     },
     {
-      "epoch": 8.49,
+      "epoch": 5.19,
       "learning_rate": 3.173333333333333e-07,
-      "loss": 0.1402,
+      "loss": 0.1815,
       "step": 3575
     },
     {
-      "epoch": 8.55,
+      "epoch": 5.22,
       "learning_rate": 3.1177777777777777e-07,
-      "loss": 0.143,
+      "loss": 0.1992,
       "step": 3600
     },
     {
-      "epoch": 8.61,
+      "epoch": 5.26,
       "learning_rate": 3.0622222222222223e-07,
-      "loss": 0.1489,
+      "loss": 0.1791,
       "step": 3625
     },
     {
-      "epoch": 8.67,
+      "epoch": 5.3,
       "learning_rate": 3.006666666666667e-07,
-      "loss": 0.1512,
+      "loss": 0.2037,
       "step": 3650
     },
     {
-      "epoch": 8.73,
+      "epoch": 5.33,
       "learning_rate": 2.951111111111111e-07,
-      "loss": 0.1306,
+      "loss": 0.2108,
       "step": 3675
     },
     {
-      "epoch": 8.79,
+      "epoch": 5.37,
       "learning_rate": 2.8955555555555555e-07,
-      "loss": 0.1402,
+      "loss": 0.1891,
       "step": 3700
     },
     {
-      "epoch": 8.85,
+      "epoch": 5.41,
       "learning_rate": 2.8399999999999995e-07,
-      "loss": 0.1474,
+      "loss": 0.1735,
       "step": 3725
     },
     {
-      "epoch": 8.91,
+      "epoch": 5.44,
       "learning_rate": 2.784444444444444e-07,
-      "loss": 0.1392,
+      "loss": 0.1955,
       "step": 3750
     },
     {
-      "epoch": 8.97,
+      "epoch": 5.48,
       "learning_rate": 2.728888888888889e-07,
-      "loss": 0.1451,
+      "loss": 0.1858,
       "step": 3775
     },
     {
-      "epoch": 9.03,
+      "epoch": 5.52,
       "learning_rate": 2.673333333333333e-07,
-      "loss": 0.1399,
+      "loss": 0.1715,
       "step": 3800
     },
     {
-      "epoch": 9.09,
+      "epoch": 5.55,
       "learning_rate": 2.617777777777778e-07,
-      "loss": 0.1352,
+      "loss": 0.1609,
       "step": 3825
     },
     {
-      "epoch": 9.14,
+      "epoch": 5.59,
       "learning_rate": 2.5622222222222223e-07,
-      "loss": 0.1348,
+      "loss": 0.1892,
       "step": 3850
     },
     {
-      "epoch": 9.2,
+      "epoch": 5.62,
       "learning_rate": 2.5066666666666663e-07,
-      "loss": 0.1272,
+      "loss": 0.2001,
       "step": 3875
     },
     {
-      "epoch": 9.26,
+      "epoch": 5.66,
       "learning_rate": 2.451111111111111e-07,
-      "loss": 0.1309,
+      "loss": 0.1571,
       "step": 3900
     },
     {
-      "epoch": 9.32,
+      "epoch": 5.7,
       "learning_rate": 2.3955555555555555e-07,
-      "loss": 0.1509,
+      "loss": 0.1655,
       "step": 3925
     },
     {
-      "epoch": 9.38,
+      "epoch": 5.73,
       "learning_rate": 2.34e-07,
-      "loss": 0.1321,
+      "loss": 0.1897,
       "step": 3950
     },
     {
-      "epoch": 9.44,
+      "epoch": 5.77,
       "learning_rate": 2.2844444444444443e-07,
-      "loss": 0.1295,
+      "loss": 0.2051,
       "step": 3975
     },
     {
-      "epoch": 9.5,
+      "epoch": 5.81,
       "learning_rate": 2.2288888888888886e-07,
-      "loss": 0.1476,
+      "loss": 0.1881,
       "step": 4000
     },
     {
-      "epoch": 9.5,
-      "eval_loss": 0.13044601678848267,
-      "eval_runtime": 1031.7698,
-      "eval_samples_per_second": 4.888,
-      "eval_steps_per_second": 0.815,
-      "eval_wer": 5.035400128727741,
+      "epoch": 5.81,
+      "eval_loss": 0.25893890857696533,
+      "eval_runtime": 2333.5911,
+      "eval_samples_per_second": 1.552,
+      "eval_steps_per_second": 0.259,
+      "eval_wer": 14.3869335173683,
       "step": 4000
     },
     {
-      "epoch": 9.56,
+      "epoch": 5.84,
       "learning_rate": 2.1733333333333332e-07,
-      "loss": 0.1407,
+      "loss": 0.1935,
       "step": 4025
     },
     {
-      "epoch": 9.62,
+      "epoch": 5.88,
       "learning_rate": 2.1177777777777778e-07,
-      "loss": 0.1288,
+      "loss": 0.183,
       "step": 4050
     },
     {
-      "epoch": 9.68,
+      "epoch": 5.91,
       "learning_rate": 2.062222222222222e-07,
-      "loss": 0.1405,
+      "loss": 0.2036,
       "step": 4075
     },
     {
-      "epoch": 9.74,
+      "epoch": 5.95,
       "learning_rate": 2.0066666666666666e-07,
-      "loss": 0.1344,
+      "loss": 0.1761,
       "step": 4100
     },
     {
-      "epoch": 9.8,
+      "epoch": 5.99,
       "learning_rate": 1.9511111111111112e-07,
-      "loss": 0.1394,
+      "loss": 0.1876,
       "step": 4125
     },
     {
-      "epoch": 9.86,
+      "epoch": 6.02,
       "learning_rate": 1.8955555555555555e-07,
-      "loss": 0.126,
+      "loss": 0.1727,
       "step": 4150
     },
     {
-      "epoch": 9.92,
+      "epoch": 6.06,
       "learning_rate": 1.8399999999999998e-07,
-      "loss": 0.127,
+      "loss": 0.176,
       "step": 4175
     },
     {
-      "epoch": 9.98,
+      "epoch": 6.1,
       "learning_rate": 1.7844444444444444e-07,
-      "loss": 0.1389,
+      "loss": 0.1598,
       "step": 4200
     },
     {
-      "epoch": 10.04,
+      "epoch": 6.13,
       "learning_rate": 1.728888888888889e-07,
-      "loss": 0.1353,
+      "loss": 0.1585,
       "step": 4225
     },
     {
-      "epoch": 10.1,
+      "epoch": 6.17,
       "learning_rate": 1.6733333333333332e-07,
-      "loss": 0.1323,
+      "loss": 0.1761,
       "step": 4250
     },
     {
-      "epoch": 10.15,
+      "epoch": 6.2,
       "learning_rate": 1.6177777777777775e-07,
-      "loss": 0.1217,
+      "loss": 0.1862,
       "step": 4275
     },
     {
-      "epoch": 10.21,
+      "epoch": 6.24,
       "learning_rate": 1.5622222222222224e-07,
-      "loss": 0.1517,
+      "loss": 0.1647,
       "step": 4300
     },
     {
-      "epoch": 10.27,
+      "epoch": 6.28,
       "learning_rate": 1.5066666666666667e-07,
-      "loss": 0.1206,
+      "loss": 0.1823,
       "step": 4325
     },
     {
-      "epoch": 10.33,
+      "epoch": 6.31,
       "learning_rate": 1.451111111111111e-07,
-      "loss": 0.1311,
+      "loss": 0.1878,
       "step": 4350
     },
     {
-      "epoch": 10.39,
+      "epoch": 6.35,
       "learning_rate": 1.3955555555555553e-07,
-      "loss": 0.1289,
+      "loss": 0.1761,
       "step": 4375
     },
     {
-      "epoch": 10.45,
+      "epoch": 6.39,
       "learning_rate": 1.34e-07,
-      "loss": 0.1241,
+      "loss": 0.1759,
       "step": 4400
     },
     {
-      "epoch": 10.51,
+      "epoch": 6.42,
       "learning_rate": 1.2844444444444444e-07,
-      "loss": 0.1287,
+      "loss": 0.1721,
       "step": 4425
     },
     {
-      "epoch": 10.57,
+      "epoch": 6.46,
       "learning_rate": 1.228888888888889e-07,
-      "loss": 0.1218,
+      "loss": 0.1856,
       "step": 4450
     },
     {
-      "epoch": 10.63,
+      "epoch": 6.49,
       "learning_rate": 1.1733333333333333e-07,
-      "loss": 0.1257,
+      "loss": 0.174,
       "step": 4475
     },
     {
-      "epoch": 10.69,
+      "epoch": 6.53,
       "learning_rate": 1.1177777777777778e-07,
-      "loss": 0.1369,
+      "loss": 0.1733,
       "step": 4500
     },
     {
-      "epoch": 10.75,
+      "epoch": 6.57,
       "learning_rate": 1.0622222222222221e-07,
-      "loss": 0.1204,
+      "loss": 0.1755,
       "step": 4525
     },
     {
-      "epoch": 10.81,
+      "epoch": 6.6,
       "learning_rate": 1.0066666666666667e-07,
-      "loss": 0.1181,
+      "loss": 0.1847,
       "step": 4550
     },
     {
-      "epoch": 10.87,
+      "epoch": 6.64,
       "learning_rate": 9.51111111111111e-08,
-      "loss": 0.1225,
+      "loss": 0.1563,
       "step": 4575
     },
     {
-      "epoch": 10.93,
+      "epoch": 6.68,
       "learning_rate": 8.955555555555555e-08,
-      "loss": 0.1326,
+      "loss": 0.1894,
       "step": 4600
     },
     {
-      "epoch": 10.99,
+      "epoch": 6.71,
       "learning_rate": 8.4e-08,
-      "loss": 0.1232,
+      "loss": 0.1782,
       "step": 4625
     },
     {
-      "epoch": 11.05,
+      "epoch": 6.75,
       "learning_rate": 7.844444444444444e-08,
-      "loss": 0.1226,
+      "loss": 0.167,
       "step": 4650
     },
     {
-      "epoch": 11.1,
+      "epoch": 6.79,
       "learning_rate": 7.288888888888888e-08,
-      "loss": 0.1188,
+      "loss": 0.1693,
       "step": 4675
     },
     {
-      "epoch": 11.16,
+      "epoch": 6.82,
       "learning_rate": 6.733333333333333e-08,
-      "loss": 0.1275,
+      "loss": 0.171,
       "step": 4700
     },
     {
-      "epoch": 11.22,
+      "epoch": 6.86,
       "learning_rate": 6.177777777777777e-08,
-      "loss": 0.1319,
+      "loss": 0.1732,
       "step": 4725
     },
     {
-      "epoch": 11.28,
+      "epoch": 6.89,
       "learning_rate": 5.622222222222222e-08,
-      "loss": 0.1224,
+      "loss": 0.1608,
       "step": 4750
     },
     {
-      "epoch": 11.34,
+      "epoch": 6.93,
       "learning_rate": 5.0666666666666664e-08,
-      "loss": 0.1191,
+      "loss": 0.1773,
       "step": 4775
     },
     {
-      "epoch": 11.4,
+      "epoch": 6.97,
       "learning_rate": 4.511111111111111e-08,
-      "loss": 0.1189,
+      "loss": 0.1669,
       "step": 4800
     },
     {
-      "epoch": 11.46,
+      "epoch": 7.0,
       "learning_rate": 3.955555555555555e-08,
-      "loss": 0.1194,
+      "loss": 0.1902,
       "step": 4825
     },
     {
-      "epoch": 11.52,
+      "epoch": 7.04,
       "learning_rate": 3.4e-08,
-      "loss": 0.1333,
+      "loss": 0.1679,
       "step": 4850
     },
     {
-      "epoch": 11.58,
+      "epoch": 7.08,
       "learning_rate": 2.8444444444444443e-08,
-      "loss": 0.1292,
+      "loss": 0.1598,
       "step": 4875
     },
     {
-      "epoch": 11.64,
+      "epoch": 7.11,
       "learning_rate": 2.2888888888888887e-08,
-      "loss": 0.1169,
+      "loss": 0.172,
       "step": 4900
     },
     {
-      "epoch": 11.7,
+      "epoch": 7.15,
       "learning_rate": 1.7333333333333333e-08,
-      "loss": 0.1154,
+      "loss": 0.1469,
       "step": 4925
     },
     {
-      "epoch": 11.76,
+      "epoch": 7.18,
       "learning_rate": 1.1777777777777778e-08,
-      "loss": 0.124,
+      "loss": 0.1546,
       "step": 4950
     },
     {
-      "epoch": 11.82,
+      "epoch": 7.22,
       "learning_rate": 6.222222222222222e-09,
-      "loss": 0.1362,
+      "loss": 0.1758,
       "step": 4975
     },
     {
-      "epoch": 11.88,
+      "epoch": 7.26,
       "learning_rate": 6.666666666666666e-10,
-      "loss": 0.1173,
+      "loss": 0.1783,
       "step": 5000
     },
     {
-      "epoch": 11.88,
-      "eval_loss": 0.12232152372598648,
-      "eval_runtime": 1046.7582,
-      "eval_samples_per_second": 4.818,
-      "eval_steps_per_second": 0.803,
-      "eval_wer": 4.711435314310234,
+      "epoch": 7.26,
+      "eval_loss": 0.25948023796081543,
+      "eval_runtime": 2325.2937,
+      "eval_samples_per_second": 1.558,
+      "eval_steps_per_second": 0.26,
+      "eval_wer": 14.345525649873474,
       "step": 5000
     },
     {
-      "epoch": 11.88,
+      "epoch": 7.26,
       "step": 5000,
-      "total_flos": 1.728655406456832e+19,
-      "train_loss": 0.26830010027885437,
-      "train_runtime": 12204.8548,
-      "train_samples_per_second": 4.916,
-      "train_steps_per_second": 0.41
+      "total_flos": 1.730098333458432e+19,
+      "train_loss": 0.30027096328735353,
+      "train_runtime": 47995.2614,
+      "train_samples_per_second": 1.25,
+      "train_steps_per_second": 0.104
     }
   ],
   "max_steps": 5000,
-  "num_train_epochs": 12,
-  "total_flos": 1.728655406456832e+19,
+  "num_train_epochs": 8,
+  "total_flos": 1.730098333458432e+19,
   "trial_name": null,
   "trial_params": null
 }