End of training

all_results.json

@@ -1,12 +1,12 @@
 {
-    "epoch": 7.92,
-    "eval_loss": 0.15959425270557404,
-    "eval_runtime": 1037.8061,
-    "eval_samples_per_second": 4.859,
-    "eval_steps_per_second": 1.215,
-    "eval_wer": 6.472859901308732,
-    "train_loss": 0.30382301959991453,
-    "train_runtime": 10304.0521,
-    "train_samples_per_second": 3.882,
-    "train_steps_per_second": 0.485
+    "epoch": 4.84,
+    "eval_loss": 0.25939562916755676,
+    "eval_runtime": 627.5878,
+    "eval_samples_per_second": 5.771,
+    "eval_steps_per_second": 1.444,
+    "eval_wer": 14.409937888198757,
+    "train_loss": 0.330156898021698,
+    "train_runtime": 8255.5438,
+    "train_samples_per_second": 4.845,
+    "train_steps_per_second": 0.606
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 7.92,
-    "eval_loss": 0.15959425270557404,
-    "eval_runtime": 1037.8061,
-    "eval_samples_per_second": 4.859,
-    "eval_steps_per_second": 1.215,
-    "eval_wer": 6.472859901308732
+    "epoch": 4.84,
+    "eval_loss": 0.25939562916755676,
+    "eval_runtime": 627.5878,
+    "eval_samples_per_second": 5.771,
+    "eval_steps_per_second": 1.444,
+    "eval_wer": 14.409937888198757
 }

runs/Aug10_14-26-15_dsai-server/events.out.tfevents.1691661685.dsai-server

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d505dcb3d89eddf00be0189aac58ebeca131d57fad16cd693b12518bc20b2a2a
+size 40

train_results.json

@@ -1,7 +1,7 @@
 {
-    "epoch": 7.92,
-    "train_loss": 0.30382301959991453,
-    "train_runtime": 10304.0521,
-    "train_samples_per_second": 3.882,
-    "train_steps_per_second": 0.485
+    "epoch": 4.84,
+    "train_loss": 0.330156898021698,
+    "train_runtime": 8255.5438,
+    "train_samples_per_second": 4.845,
+    "train_steps_per_second": 0.606
 }

trainer_state.json

@@ -1,1270 +1,1270 @@
 {
-  "best_metric": 6.472859901308732,
+  "best_metric": 14.409937888198757,
   "best_model_checkpoint": "./output/small/cv9-special-batch8-lr6-small/checkpoint-5000",
-  "epoch": 7.9239302694136295,
+  "epoch": 4.84027105517909,
   "global_step": 5000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.04,
+      "epoch": 0.02,
       "learning_rate": 4.4e-08,
-      "loss": 1.4328,
+      "loss": 1.2956,
       "step": 25
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.05,
       "learning_rate": 9.4e-08,
-      "loss": 1.4148,
+      "loss": 1.4083,
       "step": 50
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.07,
       "learning_rate": 1.44e-07,
-      "loss": 1.4051,
+      "loss": 1.3825,
       "step": 75
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.1,
       "learning_rate": 1.94e-07,
-      "loss": 1.4043,
+      "loss": 1.394,
       "step": 100
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.12,
       "learning_rate": 2.4399999999999996e-07,
-      "loss": 1.2898,
+      "loss": 1.2318,
       "step": 125
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.15,
       "learning_rate": 2.9399999999999996e-07,
-      "loss": 1.1444,
+      "loss": 1.2293,
       "step": 150
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.17,
       "learning_rate": 3.4399999999999996e-07,
-      "loss": 1.0854,
+      "loss": 1.1937,
       "step": 175
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.19,
       "learning_rate": 3.94e-07,
-      "loss": 0.9595,
+      "loss": 1.0569,
       "step": 200
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.22,
       "learning_rate": 4.44e-07,
-      "loss": 0.9625,
+      "loss": 0.9308,
       "step": 225
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.24,
       "learning_rate": 4.94e-07,
-      "loss": 0.9066,
+      "loss": 0.8598,
       "step": 250
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.27,
       "learning_rate": 5.44e-07,
-      "loss": 0.6918,
+      "loss": 0.7757,
       "step": 275
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.29,
       "learning_rate": 5.939999999999999e-07,
-      "loss": 0.6384,
+      "loss": 0.6895,
       "step": 300
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.31,
       "learning_rate": 6.44e-07,
-      "loss": 0.589,
+      "loss": 0.5567,
       "step": 325
     },
     {
-      "epoch": 0.55,
+      "epoch": 0.34,
       "learning_rate": 6.939999999999999e-07,
-      "loss": 0.572,
+      "loss": 0.5519,
       "step": 350
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.36,
       "learning_rate": 7.44e-07,
-      "loss": 0.509,
+      "loss": 0.5751,
       "step": 375
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.39,
       "learning_rate": 7.94e-07,
-      "loss": 0.5418,
+      "loss": 0.5369,
       "step": 400
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.41,
       "learning_rate": 8.439999999999999e-07,
-      "loss": 0.5221,
+      "loss": 0.4853,
       "step": 425
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.44,
       "learning_rate": 8.939999999999999e-07,
-      "loss": 0.5158,
+      "loss": 0.4981,
       "step": 450
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.46,
       "learning_rate": 9.439999999999999e-07,
-      "loss": 0.5084,
+      "loss": 0.4249,
       "step": 475
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.48,
       "learning_rate": 9.94e-07,
-      "loss": 0.4492,
+      "loss": 0.4384,
       "step": 500
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.51,
       "learning_rate": 9.95111111111111e-07,
-      "loss": 0.465,
+      "loss": 0.4681,
       "step": 525
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.53,
       "learning_rate": 9.895555555555554e-07,
-      "loss": 0.4877,
+      "loss": 0.3946,
       "step": 550
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.56,
       "learning_rate": 9.84e-07,
-      "loss": 0.4555,
+      "loss": 0.4322,
       "step": 575
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.58,
       "learning_rate": 9.784444444444444e-07,
-      "loss": 0.4367,
+      "loss": 0.467,
       "step": 600
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.61,
       "learning_rate": 9.728888888888888e-07,
-      "loss": 0.4637,
+      "loss": 0.4168,
       "step": 625
     },
     {
-      "epoch": 1.03,
+      "epoch": 0.63,
       "learning_rate": 9.673333333333332e-07,
-      "loss": 0.4346,
+      "loss": 0.3774,
       "step": 650
     },
     {
-      "epoch": 1.07,
+      "epoch": 0.65,
       "learning_rate": 9.617777777777776e-07,
-      "loss": 0.413,
+      "loss": 0.4176,
       "step": 675
     },
     {
-      "epoch": 1.11,
+      "epoch": 0.68,
       "learning_rate": 9.562222222222223e-07,
-      "loss": 0.3905,
+      "loss": 0.4306,
       "step": 700
     },
     {
-      "epoch": 1.15,
+      "epoch": 0.7,
       "learning_rate": 9.506666666666667e-07,
-      "loss": 0.3906,
+      "loss": 0.457,
       "step": 725
     },
     {
-      "epoch": 1.19,
+      "epoch": 0.73,
       "learning_rate": 9.451111111111111e-07,
-      "loss": 0.3564,
+      "loss": 0.383,
       "step": 750
     },
     {
-      "epoch": 1.23,
+      "epoch": 0.75,
       "learning_rate": 9.395555555555556e-07,
-      "loss": 0.3979,
+      "loss": 0.4315,
       "step": 775
     },
     {
-      "epoch": 1.27,
+      "epoch": 0.77,
       "learning_rate": 9.34e-07,
-      "loss": 0.3028,
+      "loss": 0.3539,
       "step": 800
     },
     {
-      "epoch": 1.31,
+      "epoch": 0.8,
       "learning_rate": 9.284444444444444e-07,
-      "loss": 0.3383,
+      "loss": 0.3698,
       "step": 825
     },
     {
-      "epoch": 1.35,
+      "epoch": 0.82,
       "learning_rate": 9.228888888888888e-07,
-      "loss": 0.3794,
+      "loss": 0.3899,
       "step": 850
     },
     {
-      "epoch": 1.39,
+      "epoch": 0.85,
       "learning_rate": 9.173333333333333e-07,
-      "loss": 0.364,
+      "loss": 0.3652,
       "step": 875
     },
     {
-      "epoch": 1.43,
+      "epoch": 0.87,
       "learning_rate": 9.117777777777778e-07,
-      "loss": 0.3377,
+      "loss": 0.3892,
       "step": 900
     },
     {
-      "epoch": 1.47,
+      "epoch": 0.9,
       "learning_rate": 9.062222222222222e-07,
-      "loss": 0.36,
+      "loss": 0.3941,
       "step": 925
     },
     {
-      "epoch": 1.51,
+      "epoch": 0.92,
       "learning_rate": 9.006666666666666e-07,
-      "loss": 0.3508,
+      "loss": 0.3609,
       "step": 950
     },
     {
-      "epoch": 1.55,
+      "epoch": 0.94,
       "learning_rate": 8.95111111111111e-07,
-      "loss": 0.3646,
+      "loss": 0.3917,
       "step": 975
     },
     {
-      "epoch": 1.58,
+      "epoch": 0.97,
       "learning_rate": 8.895555555555555e-07,
-      "loss": 0.3812,
+      "loss": 0.3678,
       "step": 1000
     },
     {
-      "epoch": 1.58,
-      "eval_loss": 0.3284018039703369,
-      "eval_runtime": 1137.8075,
-      "eval_samples_per_second": 4.432,
-      "eval_steps_per_second": 1.108,
-      "eval_wer": 14.348852177644284,
+      "epoch": 0.97,
+      "eval_loss": 0.2947046160697937,
+      "eval_runtime": 601.1133,
+      "eval_samples_per_second": 6.025,
+      "eval_steps_per_second": 1.507,
+      "eval_wer": 16.33770416379112,
       "step": 1000
     },
     {
-      "epoch": 1.62,
+      "epoch": 0.99,
       "learning_rate": 8.839999999999999e-07,
-      "loss": 0.3607,
+      "loss": 0.4326,
       "step": 1025
     },
     {
-      "epoch": 1.66,
+      "epoch": 1.02,
       "learning_rate": 8.784444444444444e-07,
-      "loss": 0.3375,
+      "loss": 0.3505,
       "step": 1050
     },
     {
-      "epoch": 1.7,
+      "epoch": 1.04,
       "learning_rate": 8.728888888888889e-07,
-      "loss": 0.3931,
+      "loss": 0.3147,
       "step": 1075
     },
     {
-      "epoch": 1.74,
+      "epoch": 1.06,
       "learning_rate": 8.673333333333332e-07,
-      "loss": 0.3583,
+      "loss": 0.334,
       "step": 1100
     },
     {
-      "epoch": 1.78,
+      "epoch": 1.09,
       "learning_rate": 8.617777777777777e-07,
-      "loss": 0.3617,
+      "loss": 0.3459,
       "step": 1125
     },
     {
-      "epoch": 1.82,
+      "epoch": 1.11,
       "learning_rate": 8.562222222222222e-07,
-      "loss": 0.3749,
+      "loss": 0.3134,
       "step": 1150
     },
     {
-      "epoch": 1.86,
+      "epoch": 1.14,
       "learning_rate": 8.506666666666667e-07,
-      "loss": 0.373,
+      "loss": 0.336,
       "step": 1175
     },
     {
-      "epoch": 1.9,
+      "epoch": 1.16,
       "learning_rate": 8.451111111111111e-07,
-      "loss": 0.3757,
+      "loss": 0.3134,
       "step": 1200
     },
     {
-      "epoch": 1.94,
+      "epoch": 1.19,
       "learning_rate": 8.395555555555556e-07,
-      "loss": 0.387,
+      "loss": 0.3092,
       "step": 1225
     },
     {
-      "epoch": 1.98,
+      "epoch": 1.21,
       "learning_rate": 8.34e-07,
-      "loss": 0.3225,
+      "loss": 0.3159,
       "step": 1250
     },
     {
-      "epoch": 2.02,
+      "epoch": 1.23,
       "learning_rate": 8.284444444444444e-07,
-      "loss": 0.3071,
+      "loss": 0.3306,
       "step": 1275
     },
     {
-      "epoch": 2.06,
+      "epoch": 1.26,
       "learning_rate": 8.228888888888889e-07,
-      "loss": 0.3026,
+      "loss": 0.3221,
       "step": 1300
     },
     {
-      "epoch": 2.1,
+      "epoch": 1.28,
       "learning_rate": 8.173333333333333e-07,
-      "loss": 0.3326,
+      "loss": 0.3118,
       "step": 1325
     },
     {
-      "epoch": 2.14,
+      "epoch": 1.31,
       "learning_rate": 8.117777777777778e-07,
-      "loss": 0.3171,
+      "loss": 0.3057,
       "step": 1350
     },
     {
-      "epoch": 2.18,
+      "epoch": 1.33,
       "learning_rate": 8.062222222222221e-07,
-      "loss": 0.2751,
+      "loss": 0.3335,
       "step": 1375
     },
     {
-      "epoch": 2.22,
+      "epoch": 1.36,
       "learning_rate": 8.006666666666666e-07,
-      "loss": 0.3204,
+      "loss": 0.3211,
       "step": 1400
     },
     {
-      "epoch": 2.26,
+      "epoch": 1.38,
       "learning_rate": 7.95111111111111e-07,
-      "loss": 0.2981,
+      "loss": 0.3073,
       "step": 1425
     },
     {
-      "epoch": 2.3,
+      "epoch": 1.4,
       "learning_rate": 7.895555555555555e-07,
-      "loss": 0.3096,
+      "loss": 0.3228,
       "step": 1450
     },
     {
-      "epoch": 2.34,
+      "epoch": 1.43,
       "learning_rate": 7.84e-07,
-      "loss": 0.2902,
+      "loss": 0.3155,
       "step": 1475
     },
     {
-      "epoch": 2.38,
+      "epoch": 1.45,
       "learning_rate": 7.784444444444444e-07,
-      "loss": 0.2813,
+      "loss": 0.3426,
       "step": 1500
     },
     {
-      "epoch": 2.42,
+      "epoch": 1.48,
       "learning_rate": 7.728888888888888e-07,
-      "loss": 0.3082,
+      "loss": 0.2791,
       "step": 1525
     },
     {
-      "epoch": 2.46,
+      "epoch": 1.5,
       "learning_rate": 7.673333333333332e-07,
-      "loss": 0.3051,
+      "loss": 0.3174,
       "step": 1550
     },
     {
-      "epoch": 2.5,
+      "epoch": 1.52,
       "learning_rate": 7.617777777777778e-07,
-      "loss": 0.2862,
+      "loss": 0.3169,
       "step": 1575
     },
     {
-      "epoch": 2.54,
+      "epoch": 1.55,
       "learning_rate": 7.562222222222222e-07,
-      "loss": 0.2478,
+      "loss": 0.2963,
       "step": 1600
     },
     {
-      "epoch": 2.58,
+      "epoch": 1.57,
       "learning_rate": 7.506666666666667e-07,
-      "loss": 0.2556,
+      "loss": 0.28,
       "step": 1625
     },
     {
-      "epoch": 2.61,
+      "epoch": 1.6,
       "learning_rate": 7.451111111111111e-07,
-      "loss": 0.2488,
+      "loss": 0.3186,
       "step": 1650
     },
     {
-      "epoch": 2.65,
+      "epoch": 1.62,
       "learning_rate": 7.395555555555555e-07,
-      "loss": 0.2963,
+      "loss": 0.3624,
       "step": 1675
     },
     {
-      "epoch": 2.69,
+      "epoch": 1.65,
       "learning_rate": 7.34e-07,
-      "loss": 0.2866,
+      "loss": 0.3469,
       "step": 1700
     },
     {
-      "epoch": 2.73,
+      "epoch": 1.67,
       "learning_rate": 7.284444444444444e-07,
-      "loss": 0.2856,
+      "loss": 0.3254,
       "step": 1725
     },
     {
-      "epoch": 2.77,
+      "epoch": 1.69,
       "learning_rate": 7.228888888888889e-07,
-      "loss": 0.2732,
+      "loss": 0.2798,
       "step": 1750
     },
     {
-      "epoch": 2.81,
+      "epoch": 1.72,
       "learning_rate": 7.173333333333333e-07,
-      "loss": 0.306,
+      "loss": 0.3179,
       "step": 1775
     },
     {
-      "epoch": 2.85,
+      "epoch": 1.74,
       "learning_rate": 7.117777777777777e-07,
-      "loss": 0.2954,
+      "loss": 0.3337,
       "step": 1800
     },
     {
-      "epoch": 2.89,
+      "epoch": 1.77,
       "learning_rate": 7.062222222222222e-07,
-      "loss": 0.264,
+      "loss": 0.3224,
       "step": 1825
     },
     {
-      "epoch": 2.93,
+      "epoch": 1.79,
       "learning_rate": 7.006666666666666e-07,
-      "loss": 0.2844,
+      "loss": 0.2911,
       "step": 1850
     },
     {
-      "epoch": 2.97,
+      "epoch": 1.82,
       "learning_rate": 6.951111111111111e-07,
-      "loss": 0.2856,
+      "loss": 0.3199,
       "step": 1875
     },
     {
-      "epoch": 3.01,
+      "epoch": 1.84,
       "learning_rate": 6.895555555555555e-07,
-      "loss": 0.2715,
+      "loss": 0.302,
       "step": 1900
     },
     {
-      "epoch": 3.05,
+      "epoch": 1.86,
       "learning_rate": 6.84e-07,
-      "loss": 0.2429,
+      "loss": 0.2863,
       "step": 1925
     },
     {
-      "epoch": 3.09,
+      "epoch": 1.89,
       "learning_rate": 6.784444444444443e-07,
-      "loss": 0.2367,
+      "loss": 0.3445,
       "step": 1950
     },
     {
-      "epoch": 3.13,
+      "epoch": 1.91,
       "learning_rate": 6.728888888888888e-07,
-      "loss": 0.2324,
+      "loss": 0.2905,
       "step": 1975
     },
     {
-      "epoch": 3.17,
+      "epoch": 1.94,
       "learning_rate": 6.673333333333334e-07,
-      "loss": 0.243,
+      "loss": 0.2802,
       "step": 2000
     },
     {
-      "epoch": 3.17,
-      "eval_loss": 0.2384428232908249,
-      "eval_runtime": 1063.3104,
-      "eval_samples_per_second": 4.743,
-      "eval_steps_per_second": 1.186,
-      "eval_wer": 10.152327826646642,
+      "epoch": 1.94,
+      "eval_loss": 0.268283873796463,
+      "eval_runtime": 598.3597,
+      "eval_samples_per_second": 6.053,
+      "eval_steps_per_second": 1.514,
+      "eval_wer": 14.810213940648723,
       "step": 2000
     },
     {
-      "epoch": 3.21,
+      "epoch": 1.96,
       "learning_rate": 6.617777777777778e-07,
-      "loss": 0.2703,
+      "loss": 0.306,
       "step": 2025
     },
     {
-      "epoch": 3.25,
+      "epoch": 1.98,
       "learning_rate": 6.562222222222223e-07,
-      "loss": 0.295,
+      "loss": 0.3184,
       "step": 2050
     },
     {
-      "epoch": 3.29,
+      "epoch": 2.01,
       "learning_rate": 6.506666666666666e-07,
-      "loss": 0.2683,
+      "loss": 0.3043,
       "step": 2075
     },
     {
-      "epoch": 3.33,
+      "epoch": 2.03,
       "learning_rate": 6.451111111111111e-07,
-      "loss": 0.2618,
+      "loss": 0.2558,
       "step": 2100
     },
     {
-      "epoch": 3.37,
+      "epoch": 2.06,
       "learning_rate": 6.395555555555555e-07,
-      "loss": 0.2385,
+      "loss": 0.2471,
       "step": 2125
     },
     {
-      "epoch": 3.41,
+      "epoch": 2.08,
       "learning_rate": 6.34e-07,
-      "loss": 0.2422,
+      "loss": 0.2577,
       "step": 2150
     },
     {
-      "epoch": 3.45,
+      "epoch": 2.11,
       "learning_rate": 6.284444444444445e-07,
-      "loss": 0.2486,
+      "loss": 0.2516,
       "step": 2175
     },
     {
-      "epoch": 3.49,
+      "epoch": 2.13,
       "learning_rate": 6.228888888888889e-07,
-      "loss": 0.2198,
+      "loss": 0.2571,
       "step": 2200
     },
     {
-      "epoch": 3.53,
+      "epoch": 2.15,
       "learning_rate": 6.173333333333333e-07,
-      "loss": 0.23,
+      "loss": 0.261,
       "step": 2225
     },
     {
-      "epoch": 3.57,
+      "epoch": 2.18,
       "learning_rate": 6.117777777777777e-07,
-      "loss": 0.2443,
+      "loss": 0.2884,
       "step": 2250
     },
     {
-      "epoch": 3.61,
+      "epoch": 2.2,
       "learning_rate": 6.062222222222222e-07,
-      "loss": 0.2439,
+      "loss": 0.2695,
       "step": 2275
     },
     {
-      "epoch": 3.65,
+      "epoch": 2.23,
       "learning_rate": 6.006666666666666e-07,
-      "loss": 0.2579,
+      "loss": 0.2538,
       "step": 2300
     },
     {
-      "epoch": 3.68,
+      "epoch": 2.25,
       "learning_rate": 5.951111111111111e-07,
-      "loss": 0.2318,
+      "loss": 0.2577,
       "step": 2325
     },
     {
-      "epoch": 3.72,
+      "epoch": 2.27,
       "learning_rate": 5.895555555555555e-07,
-      "loss": 0.2471,
+      "loss": 0.2299,
       "step": 2350
     },
     {
-      "epoch": 3.76,
+      "epoch": 2.3,
       "learning_rate": 5.839999999999999e-07,
-      "loss": 0.2224,
+      "loss": 0.2452,
       "step": 2375
     },
     {
-      "epoch": 3.8,
+      "epoch": 2.32,
       "learning_rate": 5.784444444444444e-07,
-      "loss": 0.2422,
+      "loss": 0.2798,
       "step": 2400
     },
     {
-      "epoch": 3.84,
+      "epoch": 2.35,
       "learning_rate": 5.728888888888888e-07,
-      "loss": 0.2285,
+      "loss": 0.2839,
       "step": 2425
     },
     {
-      "epoch": 3.88,
+      "epoch": 2.37,
       "learning_rate": 5.673333333333334e-07,
-      "loss": 0.2241,
+      "loss": 0.2453,
       "step": 2450
     },
     {
-      "epoch": 3.92,
+      "epoch": 2.4,
       "learning_rate": 5.617777777777778e-07,
-      "loss": 0.2263,
+      "loss": 0.2626,
       "step": 2475
     },
     {
-      "epoch": 3.96,
+      "epoch": 2.42,
       "learning_rate": 5.562222222222222e-07,
-      "loss": 0.2113,
+      "loss": 0.2432,
       "step": 2500
     },
     {
-      "epoch": 4.0,
+      "epoch": 2.44,
       "learning_rate": 5.506666666666666e-07,
-      "loss": 0.2461,
+      "loss": 0.2598,
       "step": 2525
     },
     {
-      "epoch": 4.04,
+      "epoch": 2.47,
       "learning_rate": 5.451111111111111e-07,
-      "loss": 0.2179,
+      "loss": 0.2737,
       "step": 2550
     },
     {
-      "epoch": 4.08,
+      "epoch": 2.49,
       "learning_rate": 5.395555555555556e-07,
-      "loss": 0.1978,
+      "loss": 0.2467,
       "step": 2575
     },
     {
-      "epoch": 4.12,
+      "epoch": 2.52,
       "learning_rate": 5.34e-07,
-      "loss": 0.2208,
+      "loss": 0.242,
       "step": 2600
     },
     {
-      "epoch": 4.16,
+      "epoch": 2.54,
       "learning_rate": 5.284444444444445e-07,
-      "loss": 0.1959,
+      "loss": 0.2424,
       "step": 2625
     },
     {
-      "epoch": 4.2,
+      "epoch": 2.57,
       "learning_rate": 5.228888888888888e-07,
-      "loss": 0.2211,
+      "loss": 0.2666,
       "step": 2650
     },
     {
-      "epoch": 4.24,
+      "epoch": 2.59,
       "learning_rate": 5.173333333333333e-07,
-      "loss": 0.1886,
+      "loss": 0.2473,
       "step": 2675
     },
     {
-      "epoch": 4.28,
+      "epoch": 2.61,
       "learning_rate": 5.117777777777777e-07,
-      "loss": 0.2161,
+      "loss": 0.2862,
       "step": 2700
     },
     {
-      "epoch": 4.32,
+      "epoch": 2.64,
       "learning_rate": 5.062222222222222e-07,
-      "loss": 0.1965,
+      "loss": 0.2658,
       "step": 2725
     },
     {
-      "epoch": 4.36,
+      "epoch": 2.66,
       "learning_rate": 5.006666666666667e-07,
-      "loss": 0.2322,
+      "loss": 0.2696,
       "step": 2750
     },
     {
-      "epoch": 4.4,
+      "epoch": 2.69,
       "learning_rate": 4.951111111111111e-07,
-      "loss": 0.221,
+      "loss": 0.2564,
       "step": 2775
     },
     {
-      "epoch": 4.44,
+      "epoch": 2.71,
       "learning_rate": 4.895555555555555e-07,
-      "loss": 0.1939,
+      "loss": 0.2573,
       "step": 2800
     },
     {
-      "epoch": 4.48,
+      "epoch": 2.73,
       "learning_rate": 4.839999999999999e-07,
-      "loss": 0.2081,
+      "loss": 0.2479,
       "step": 2825
     },
     {
-      "epoch": 4.52,
+      "epoch": 2.76,
       "learning_rate": 4.784444444444444e-07,
-      "loss": 0.2016,
+      "loss": 0.2618,
       "step": 2850
     },
     {
-      "epoch": 4.56,
+      "epoch": 2.78,
       "learning_rate": 4.728888888888889e-07,
-      "loss": 0.2085,
+      "loss": 0.2359,
       "step": 2875
     },
     {
-      "epoch": 4.6,
+      "epoch": 2.81,
       "learning_rate": 4.673333333333333e-07,
-      "loss": 0.2066,
+      "loss": 0.2513,
       "step": 2900
     },
     {
-      "epoch": 4.64,
+      "epoch": 2.83,
       "learning_rate": 4.6177777777777777e-07,
-      "loss": 0.2011,
+      "loss": 0.2417,
       "step": 2925
     },
     {
-      "epoch": 4.68,
+      "epoch": 2.86,
       "learning_rate": 4.5622222222222217e-07,
-      "loss": 0.2275,
+      "loss": 0.2306,
       "step": 2950
     },
     {
-      "epoch": 4.71,
+      "epoch": 2.88,
       "learning_rate": 4.506666666666666e-07,
-      "loss": 0.2265,
+      "loss": 0.2593,
       "step": 2975
     },
     {
-      "epoch": 4.75,
+      "epoch": 2.9,
       "learning_rate": 4.451111111111111e-07,
-      "loss": 0.2088,
+      "loss": 0.2488,
       "step": 3000
     },
     {
-      "epoch": 4.75,
-      "eval_loss": 0.19160950183868408,
-      "eval_runtime": 1003.8357,
-      "eval_samples_per_second": 5.024,
-      "eval_steps_per_second": 1.256,
-      "eval_wer": 7.944647071443896,
+      "epoch": 2.9,
+      "eval_loss": 0.26232555508613586,
+      "eval_runtime": 611.1177,
+      "eval_samples_per_second": 5.927,
+      "eval_steps_per_second": 1.483,
+      "eval_wer": 14.939038417299285,
       "step": 3000
     },
     {
-      "epoch": 4.79,
+      "epoch": 2.93,
       "learning_rate": 4.3955555555555554e-07,
-      "loss": 0.2188,
+      "loss": 0.2372,
       "step": 3025
     },
     {
-      "epoch": 4.83,
+      "epoch": 2.95,
       "learning_rate": 4.34e-07,
-      "loss": 0.2171,
+      "loss": 0.2681,
       "step": 3050
     },
     {
-      "epoch": 4.87,
+      "epoch": 2.98,
       "learning_rate": 4.2844444444444445e-07,
-      "loss": 0.2175,
+      "loss": 0.2543,
       "step": 3075
     },
     {
-      "epoch": 4.91,
+      "epoch": 3.0,
       "learning_rate": 4.2288888888888886e-07,
-      "loss": 0.1954,
+      "loss": 0.2389,
       "step": 3100
     },
     {
-      "epoch": 4.95,
+      "epoch": 3.03,
       "learning_rate": 4.173333333333333e-07,
-      "loss": 0.204,
+      "loss": 0.2287,
       "step": 3125
     },
     {
-      "epoch": 4.99,
+      "epoch": 3.05,
       "learning_rate": 4.1177777777777777e-07,
-      "loss": 0.1982,
+      "loss": 0.2168,
       "step": 3150
     },
     {
-      "epoch": 5.03,
+      "epoch": 3.07,
       "learning_rate": 4.0622222222222217e-07,
-      "loss": 0.178,
+      "loss": 0.2341,
       "step": 3175
     },
     {
-      "epoch": 5.07,
+      "epoch": 3.1,
       "learning_rate": 4.0066666666666663e-07,
-      "loss": 0.1816,
+      "loss": 0.2393,
       "step": 3200
     },
     {
-      "epoch": 5.11,
+      "epoch": 3.12,
       "learning_rate": 3.9511111111111114e-07,
-      "loss": 0.1868,
+      "loss": 0.2227,
       "step": 3225
     },
     {
-      "epoch": 5.15,
+      "epoch": 3.15,
       "learning_rate": 3.8955555555555554e-07,
-      "loss": 0.1831,
+      "loss": 0.213,
       "step": 3250
     },
     {
-      "epoch": 5.19,
+      "epoch": 3.17,
       "learning_rate": 3.84e-07,
-      "loss": 0.2125,
+      "loss": 0.1968,
       "step": 3275
     },
     {
-      "epoch": 5.23,
+      "epoch": 3.19,
       "learning_rate": 3.7844444444444445e-07,
-      "loss": 0.1779,
+      "loss": 0.2131,
       "step": 3300
     },
     {
-      "epoch": 5.27,
+      "epoch": 3.22,
       "learning_rate": 3.7288888888888886e-07,
-      "loss": 0.1762,
+      "loss": 0.2086,
       "step": 3325
     },
     {
-      "epoch": 5.31,
+      "epoch": 3.24,
       "learning_rate": 3.673333333333333e-07,
-      "loss": 0.1805,
+      "loss": 0.1916,
       "step": 3350
     },
     {
-      "epoch": 5.35,
+      "epoch": 3.27,
       "learning_rate": 3.617777777777777e-07,
-      "loss": 0.2095,
+      "loss": 0.2665,
       "step": 3375
     },
     {
-      "epoch": 5.39,
+      "epoch": 3.29,
       "learning_rate": 3.5622222222222223e-07,
-      "loss": 0.1855,
+      "loss": 0.2082,
       "step": 3400
     },
     {
-      "epoch": 5.43,
+      "epoch": 3.32,
       "learning_rate": 3.506666666666667e-07,
-      "loss": 0.184,
+      "loss": 0.2042,
       "step": 3425
     },
     {
-      "epoch": 5.47,
+      "epoch": 3.34,
       "learning_rate": 3.451111111111111e-07,
-      "loss": 0.1778,
+      "loss": 0.2426,
       "step": 3450
     },
     {
-      "epoch": 5.51,
+      "epoch": 3.36,
       "learning_rate": 3.3955555555555554e-07,
-      "loss": 0.1979,
+      "loss": 0.2116,
       "step": 3475
     },
     {
-      "epoch": 5.55,
+      "epoch": 3.39,
       "learning_rate": 3.34e-07,
-      "loss": 0.1684,
+      "loss": 0.1946,
       "step": 3500
     },
     {
-      "epoch": 5.59,
+      "epoch": 3.41,
       "learning_rate": 3.284444444444444e-07,
-      "loss": 0.2078,
+      "loss": 0.2226,
       "step": 3525
     },
     {
-      "epoch": 5.63,
+      "epoch": 3.44,
       "learning_rate": 3.2288888888888886e-07,
-      "loss": 0.1852,
+      "loss": 0.2169,
       "step": 3550
     },
     {
-      "epoch": 5.67,
+      "epoch": 3.46,
       "learning_rate": 3.173333333333333e-07,
-      "loss": 0.188,
+      "loss": 0.2413,
       "step": 3575
     },
     {
-      "epoch": 5.71,
+      "epoch": 3.48,
       "learning_rate": 3.1177777777777777e-07,
-      "loss": 0.178,
+      "loss": 0.2535,
       "step": 3600
     },
     {
-      "epoch": 5.74,
+      "epoch": 3.51,
       "learning_rate": 3.0622222222222223e-07,
-      "loss": 0.1843,
+      "loss": 0.2209,
       "step": 3625
     },
     {
-      "epoch": 5.78,
+      "epoch": 3.53,
       "learning_rate": 3.006666666666667e-07,
-      "loss": 0.2204,
+      "loss": 0.2466,
       "step": 3650
     },
     {
-      "epoch": 5.82,
+      "epoch": 3.56,
       "learning_rate": 2.951111111111111e-07,
-      "loss": 0.2049,
+      "loss": 0.2282,
       "step": 3675
     },
     {
-      "epoch": 5.86,
+      "epoch": 3.58,
       "learning_rate": 2.8955555555555555e-07,
-      "loss": 0.1702,
+      "loss": 0.2395,
       "step": 3700
     },
     {
-      "epoch": 5.9,
+      "epoch": 3.61,
       "learning_rate": 2.8399999999999995e-07,
-      "loss": 0.1929,
+      "loss": 0.2301,
       "step": 3725
     },
     {
-      "epoch": 5.94,
+      "epoch": 3.63,
       "learning_rate": 2.784444444444444e-07,
-      "loss": 0.1808,
+      "loss": 0.2006,
       "step": 3750
     },
     {
-      "epoch": 5.98,
+      "epoch": 3.65,
       "learning_rate": 2.728888888888889e-07,
-      "loss": 0.1835,
+      "loss": 0.2152,
       "step": 3775
     },
     {
-      "epoch": 6.02,
+      "epoch": 3.68,
       "learning_rate": 2.673333333333333e-07,
-      "loss": 0.156,
+      "loss": 0.229,
       "step": 3800
     },
     {
-      "epoch": 6.06,
+      "epoch": 3.7,
       "learning_rate": 2.617777777777778e-07,
-      "loss": 0.1728,
+      "loss": 0.224,
       "step": 3825
     },
     {
-      "epoch": 6.1,
+      "epoch": 3.73,
       "learning_rate": 2.5622222222222223e-07,
-      "loss": 0.1886,
+      "loss": 0.2649,
       "step": 3850
     },
     {
-      "epoch": 6.14,
+      "epoch": 3.75,
       "learning_rate": 2.5066666666666663e-07,
-      "loss": 0.1515,
+      "loss": 0.2083,
       "step": 3875
     },
     {
-      "epoch": 6.18,
+      "epoch": 3.78,
       "learning_rate": 2.451111111111111e-07,
-      "loss": 0.1548,
+      "loss": 0.2142,
       "step": 3900
     },
     {
-      "epoch": 6.22,
+      "epoch": 3.8,
       "learning_rate": 2.3955555555555555e-07,
-      "loss": 0.1775,
+      "loss": 0.2195,
       "step": 3925
     },
     {
-      "epoch": 6.26,
+      "epoch": 3.82,
       "learning_rate": 2.34e-07,
-      "loss": 0.1601,
+      "loss": 0.2088,
       "step": 3950
     },
     {
-      "epoch": 6.3,
+      "epoch": 3.85,
       "learning_rate": 2.2844444444444443e-07,
-      "loss": 0.1879,
+      "loss": 0.2225,
       "step": 3975
     },
     {
-      "epoch": 6.34,
+      "epoch": 3.87,
       "learning_rate": 2.2288888888888886e-07,
-      "loss": 0.1798,
+      "loss": 0.2066,
       "step": 4000
     },
     {
-      "epoch": 6.34,
-      "eval_loss": 0.1675407886505127,
-      "eval_runtime": 1057.188,
-      "eval_samples_per_second": 4.77,
-      "eval_steps_per_second": 1.193,
-      "eval_wer": 6.848315812057498,
+      "epoch": 3.87,
+      "eval_loss": 0.25899189710617065,
+      "eval_runtime": 623.3342,
+      "eval_samples_per_second": 5.811,
+      "eval_steps_per_second": 1.453,
+      "eval_wer": 14.455946629859673,
       "step": 4000
     },
     {
-      "epoch": 6.38,
+      "epoch": 3.9,
       "learning_rate": 2.1733333333333332e-07,
-      "loss": 0.172,
+      "loss": 0.2293,
       "step": 4025
     },
     {
-      "epoch": 6.42,
+      "epoch": 3.92,
       "learning_rate": 2.1177777777777778e-07,
-      "loss": 0.1642,
+      "loss": 0.2206,
       "step": 4050
     },
     {
-      "epoch": 6.46,
+      "epoch": 3.94,
       "learning_rate": 2.062222222222222e-07,
-      "loss": 0.1672,
+      "loss": 0.2143,
       "step": 4075
     },
     {
-      "epoch": 6.5,
+      "epoch": 3.97,
       "learning_rate": 2.0066666666666666e-07,
-      "loss": 0.1864,
+      "loss": 0.2178,
       "step": 4100
     },
     {
-      "epoch": 6.54,
+      "epoch": 3.99,
       "learning_rate": 1.9511111111111112e-07,
-      "loss": 0.1633,
+      "loss": 0.2214,
       "step": 4125
     },
     {
-      "epoch": 6.58,
+      "epoch": 4.02,
       "learning_rate": 1.8955555555555555e-07,
-      "loss": 0.1621,
+      "loss": 0.1748,
       "step": 4150
     },
     {
-      "epoch": 6.62,
+      "epoch": 4.04,
       "learning_rate": 1.8399999999999998e-07,
-      "loss": 0.1781,
+      "loss": 0.2046,
       "step": 4175
     },
     {
-      "epoch": 6.66,
+      "epoch": 4.07,
       "learning_rate": 1.7844444444444444e-07,
-      "loss": 0.1659,
+      "loss": 0.2031,
       "step": 4200
     },
     {
-      "epoch": 6.7,
+      "epoch": 4.09,
       "learning_rate": 1.728888888888889e-07,
-      "loss": 0.1782,
+      "loss": 0.2178,
       "step": 4225
     },
     {
-      "epoch": 6.74,
+      "epoch": 4.11,
       "learning_rate": 1.6733333333333332e-07,
-      "loss": 0.1735,
+      "loss": 0.1946,
       "step": 4250
     },
     {
-      "epoch": 6.77,
+      "epoch": 4.14,
       "learning_rate": 1.6177777777777775e-07,
-      "loss": 0.1771,
+      "loss": 0.244,
       "step": 4275
     },
     {
-      "epoch": 6.81,
+      "epoch": 4.16,
       "learning_rate": 1.5622222222222224e-07,
-      "loss": 0.1822,
+      "loss": 0.2003,
       "step": 4300
     },
     {
-      "epoch": 6.85,
+      "epoch": 4.19,
       "learning_rate": 1.5066666666666667e-07,
-      "loss": 0.1537,
+      "loss": 0.2167,
       "step": 4325
     },
     {
-      "epoch": 6.89,
+      "epoch": 4.21,
       "learning_rate": 1.451111111111111e-07,
-      "loss": 0.1823,
+      "loss": 0.2167,
       "step": 4350
     },
     {
-      "epoch": 6.93,
+      "epoch": 4.24,
       "learning_rate": 1.3955555555555553e-07,
-      "loss": 0.1811,
+      "loss": 0.2025,
       "step": 4375
     },
     {
-      "epoch": 6.97,
+      "epoch": 4.26,
       "learning_rate": 1.34e-07,
-      "loss": 0.1792,
+      "loss": 0.2313,
       "step": 4400
     },
     {
-      "epoch": 7.01,
+      "epoch": 4.28,
       "learning_rate": 1.2844444444444444e-07,
-      "loss": 0.1369,
+      "loss": 0.1842,
       "step": 4425
     },
     {
-      "epoch": 7.05,
+      "epoch": 4.31,
       "learning_rate": 1.228888888888889e-07,
-      "loss": 0.197,
+      "loss": 0.2034,
       "step": 4450
     },
     {
-      "epoch": 7.09,
+      "epoch": 4.33,
       "learning_rate": 1.1733333333333333e-07,
-      "loss": 0.1807,
+      "loss": 0.1849,
       "step": 4475
     },
     {
-      "epoch": 7.13,
+      "epoch": 4.36,
       "learning_rate": 1.1177777777777778e-07,
-      "loss": 0.1656,
+      "loss": 0.2283,
       "step": 4500
     },
     {
-      "epoch": 7.17,
+      "epoch": 4.38,
       "learning_rate": 1.0622222222222221e-07,
-      "loss": 0.1539,
+      "loss": 0.1916,
       "step": 4525
     },
     {
-      "epoch": 7.21,
+      "epoch": 4.4,
       "learning_rate": 1.0066666666666667e-07,
-      "loss": 0.1563,
+      "loss": 0.2096,
       "step": 4550
     },
     {
-      "epoch": 7.25,
+      "epoch": 4.43,
       "learning_rate": 9.51111111111111e-08,
-      "loss": 0.1651,
+      "loss": 0.1966,
       "step": 4575
     },
     {
-      "epoch": 7.29,
-      "learning_rate": 8.955555555555555e-08,
-      "loss": 0.1486,
+      "epoch": 4.45,
+      "learning_rate": 8.977777777777777e-08,
+      "loss": 0.2004,
       "step": 4600
     },
     {
-      "epoch": 7.33,
-      "learning_rate": 8.4e-08,
-      "loss": 0.167,
+      "epoch": 4.48,
+      "learning_rate": 8.422222222222223e-08,
+      "loss": 0.1891,
       "step": 4625
     },
     {
-      "epoch": 7.37,
-      "learning_rate": 7.844444444444444e-08,
-      "loss": 0.1484,
+      "epoch": 4.5,
+      "learning_rate": 7.866666666666666e-08,
+      "loss": 0.2059,
       "step": 4650
     },
     {
-      "epoch": 7.41,
-      "learning_rate": 7.288888888888888e-08,
-      "loss": 0.1605,
+      "epoch": 4.53,
+      "learning_rate": 7.311111111111111e-08,
+      "loss": 0.2054,
       "step": 4675
     },
     {
-      "epoch": 7.45,
-      "learning_rate": 6.733333333333333e-08,
-      "loss": 0.1729,
+      "epoch": 4.55,
+      "learning_rate": 6.755555555555554e-08,
+      "loss": 0.2186,
       "step": 4700
     },
     {
-      "epoch": 7.49,
-      "learning_rate": 6.177777777777777e-08,
-      "loss": 0.1555,
+      "epoch": 4.57,
+      "learning_rate": 6.2e-08,
+      "loss": 0.1954,
       "step": 4725
     },
     {
-      "epoch": 7.53,
-      "learning_rate": 5.622222222222222e-08,
-      "loss": 0.1583,
+      "epoch": 4.6,
+      "learning_rate": 5.644444444444444e-08,
+      "loss": 0.2086,
       "step": 4750
     },
     {
-      "epoch": 7.57,
-      "learning_rate": 5.0666666666666664e-08,
-      "loss": 0.1577,
+      "epoch": 4.62,
+      "learning_rate": 5.0888888888888886e-08,
+      "loss": 0.2241,
       "step": 4775
     },
     {
-      "epoch": 7.61,
-      "learning_rate": 4.511111111111111e-08,
-      "loss": 0.1596,
+      "epoch": 4.65,
+      "learning_rate": 4.5333333333333336e-08,
+      "loss": 0.2256,
       "step": 4800
     },
     {
-      "epoch": 7.65,
-      "learning_rate": 3.955555555555555e-08,
-      "loss": 0.1739,
+      "epoch": 4.67,
+      "learning_rate": 3.977777777777778e-08,
+      "loss": 0.213,
       "step": 4825
     },
     {
-      "epoch": 7.69,
+      "epoch": 4.7,
       "learning_rate": 3.422222222222222e-08,
-      "loss": 0.1583,
+      "loss": 0.1952,
       "step": 4850
     },
     {
-      "epoch": 7.73,
+      "epoch": 4.72,
       "learning_rate": 2.8666666666666665e-08,
-      "loss": 0.1556,
+      "loss": 0.1955,
       "step": 4875
     },
     {
-      "epoch": 7.77,
+      "epoch": 4.74,
       "learning_rate": 2.311111111111111e-08,
-      "loss": 0.164,
+      "loss": 0.1706,
       "step": 4900
     },
     {
-      "epoch": 7.81,
+      "epoch": 4.77,
       "learning_rate": 1.7555555555555555e-08,
-      "loss": 0.1564,
+      "loss": 0.2055,
       "step": 4925
     },
     {
-      "epoch": 7.84,
+      "epoch": 4.79,
       "learning_rate": 1.2e-08,
-      "loss": 0.1689,
+      "loss": 0.1925,
       "step": 4950
     },
     {
-      "epoch": 7.88,
+      "epoch": 4.82,
       "learning_rate": 6.444444444444444e-09,
-      "loss": 0.1571,
+      "loss": 0.188,
       "step": 4975
     },
     {
-      "epoch": 7.92,
+      "epoch": 4.84,
       "learning_rate": 8.888888888888889e-10,
-      "loss": 0.1671,
+      "loss": 0.2122,
       "step": 5000
     },
     {
-      "epoch": 7.92,
-      "eval_loss": 0.15959425270557404,
-      "eval_runtime": 1033.6723,
-      "eval_samples_per_second": 4.879,
-      "eval_steps_per_second": 1.22,
-      "eval_wer": 6.472859901308732,
+      "epoch": 4.84,
+      "eval_loss": 0.25939562916755676,
+      "eval_runtime": 623.5586,
+      "eval_samples_per_second": 5.809,
+      "eval_steps_per_second": 1.453,
+      "eval_wer": 14.409937888198757,
       "step": 5000
     },
     {
-      "epoch": 7.92,
+      "epoch": 4.84,
       "step": 5000,
-      "total_flos": 1.15333155237888e+19,
-      "train_loss": 0.30382301959991453,
-      "train_runtime": 10304.0521,
-      "train_samples_per_second": 3.882,
-      "train_steps_per_second": 0.485
+      "total_flos": 1.153995298799616e+19,
+      "train_loss": 0.330156898021698,
+      "train_runtime": 8255.5438,
+      "train_samples_per_second": 4.845,
+      "train_steps_per_second": 0.606
     }
   ],
   "max_steps": 5000,
-  "num_train_epochs": 8,
-  "total_flos": 1.15333155237888e+19,
+  "num_train_epochs": 5,
+  "total_flos": 1.153995298799616e+19,
   "trial_name": null,
   "trial_params": null
 }