End of training

all_results.json

@@ -1,14 +1,14 @@
 {
     "epoch": 100.0,
-    "eval_loss": 0.6831369996070862,
-    "eval_runtime": 55.7549,
+    "eval_loss": 0.659227728843689,
+    "eval_runtime": 51.9908,
     "eval_samples": 1703,
-    "eval_samples_per_second": 30.544,
-    "eval_steps_per_second": 0.969,
-    "eval_wer": 0.4286635944700461,
-    "train_loss": 0.9041624231254105,
-    "train_runtime": 18573.3759,
+    "eval_samples_per_second": 32.756,
+    "eval_steps_per_second": 1.039,
+    "eval_wer": 0.45640552995391703,
+    "train_loss": 0.9130245208740234,
+    "train_runtime": 17767.5681,
     "train_samples": 3601,
-    "train_samples_per_second": 19.388,
-    "train_steps_per_second": 0.608
+    "train_samples_per_second": 20.267,
+    "train_steps_per_second": 0.636
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 100.0,
-    "eval_loss": 0.6831369996070862,
-    "eval_runtime": 55.7549,
+    "eval_loss": 0.659227728843689,
+    "eval_runtime": 51.9908,
     "eval_samples": 1703,
-    "eval_samples_per_second": 30.544,
-    "eval_steps_per_second": 0.969,
-    "eval_wer": 0.4286635944700461
+    "eval_samples_per_second": 32.756,
+    "eval_steps_per_second": 1.039,
+    "eval_wer": 0.45640552995391703
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5301fb5708bea0c9a55fac6bf9ab02b82cb49965ed803e96ab4ddae579b8b706
+oid sha256:04dc2d5e12b690a836e89132ee1906bc07361f91b36f64bf93b25dc84b5b8a23
 size 1262136881

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 100.0,
-    "train_loss": 0.9041624231254105,
-    "train_runtime": 18573.3759,
+    "train_loss": 0.9130245208740234,
+    "train_runtime": 17767.5681,
     "train_samples": 3601,
-    "train_samples_per_second": 19.388,
-    "train_steps_per_second": 0.608
+    "train_samples_per_second": 20.267,
+    "train_steps_per_second": 0.636
 }

trainer_state.json

@@ -10,892 +10,892 @@
     {
       "epoch": 0.88,
       "learning_rate": 5.82e-05,
-      "loss": 10.446,
+      "loss": 10.5082,
       "step": 100
     },
     {
       "epoch": 1.77,
       "learning_rate": 0.0001182,
-      "loss": 3.652,
+      "loss": 3.6378,
       "step": 200
     },
     {
       "epoch": 2.65,
       "learning_rate": 0.00017819999999999997,
-      "loss": 3.2102,
+      "loss": 3.2141,
       "step": 300
     },
     {
       "epoch": 3.54,
       "learning_rate": 0.0002382,
-      "loss": 3.1322,
+      "loss": 3.1644,
       "step": 400
     },
     {
       "epoch": 4.42,
       "learning_rate": 0.0002982,
-      "loss": 3.0798,
+      "loss": 3.0928,
       "step": 500
     },
     {
       "epoch": 4.42,
-      "eval_loss": 3.001023054122925,
-      "eval_runtime": 56.1867,
-      "eval_samples_per_second": 30.31,
-      "eval_steps_per_second": 0.961,
-      "eval_wer": 1.0011981566820276,
+      "eval_loss": 3.0804343223571777,
+      "eval_runtime": 51.9529,
+      "eval_samples_per_second": 32.78,
+      "eval_steps_per_second": 1.039,
+      "eval_wer": 1.0072811059907834,
       "step": 500
     },
     {
       "epoch": 5.31,
       "learning_rate": 0.00029730555555555554,
-      "loss": 2.5544,
+      "loss": 2.7805,
       "step": 600
     },
     {
       "epoch": 6.19,
       "learning_rate": 0.0002945277777777777,
-      "loss": 1.8568,
+      "loss": 1.9414,
       "step": 700
     },
     {
       "epoch": 7.08,
       "learning_rate": 0.00029174999999999996,
-      "loss": 1.6575,
+      "loss": 1.6707,
       "step": 800
     },
     {
       "epoch": 7.96,
       "learning_rate": 0.0002889722222222222,
-      "loss": 1.5125,
+      "loss": 1.5232,
       "step": 900
     },
     {
       "epoch": 8.85,
       "learning_rate": 0.0002861944444444444,
-      "loss": 1.4336,
+      "loss": 1.4505,
       "step": 1000
     },
     {
       "epoch": 8.85,
-      "eval_loss": 0.8481376767158508,
-      "eval_runtime": 54.9915,
-      "eval_samples_per_second": 30.968,
-      "eval_steps_per_second": 0.982,
-      "eval_wer": 0.6910599078341014,
+      "eval_loss": 0.9038452506065369,
+      "eval_runtime": 51.3716,
+      "eval_samples_per_second": 33.151,
+      "eval_steps_per_second": 1.051,
+      "eval_wer": 0.7329953917050691,
       "step": 1000
     },
     {
       "epoch": 9.73,
       "learning_rate": 0.0002834166666666666,
-      "loss": 1.3709,
+      "loss": 1.3908,
       "step": 1100
     },
     {
       "epoch": 10.62,
       "learning_rate": 0.00028063888888888886,
-      "loss": 1.3058,
+      "loss": 1.3335,
       "step": 1200
     },
     {
       "epoch": 11.5,
       "learning_rate": 0.0002778611111111111,
-      "loss": 1.2604,
+      "loss": 1.2725,
       "step": 1300
     },
     {
       "epoch": 12.39,
       "learning_rate": 0.0002750833333333333,
-      "loss": 1.2065,
+      "loss": 1.2043,
       "step": 1400
     },
     {
       "epoch": 13.27,
       "learning_rate": 0.00027230555555555553,
-      "loss": 1.2062,
+      "loss": 1.2207,
       "step": 1500
     },
     {
       "epoch": 13.27,
-      "eval_loss": 0.7312180995941162,
-      "eval_runtime": 55.5977,
-      "eval_samples_per_second": 30.631,
-      "eval_steps_per_second": 0.971,
-      "eval_wer": 0.6332718894009216,
+      "eval_loss": 0.7374930381774902,
+      "eval_runtime": 51.4675,
+      "eval_samples_per_second": 33.089,
+      "eval_steps_per_second": 1.049,
+      "eval_wer": 0.604516129032258,
       "step": 1500
     },
     {
       "epoch": 14.16,
       "learning_rate": 0.00026952777777777777,
-      "loss": 1.1712,
+      "loss": 1.1794,
       "step": 1600
     },
     {
       "epoch": 15.04,
       "learning_rate": 0.00026674999999999995,
-      "loss": 1.1348,
+      "loss": 1.1502,
       "step": 1700
     },
     {
       "epoch": 15.93,
       "learning_rate": 0.0002639722222222222,
-      "loss": 1.1077,
+      "loss": 1.1261,
       "step": 1800
     },
     {
       "epoch": 16.81,
       "learning_rate": 0.00026119444444444443,
-      "loss": 1.0821,
+      "loss": 1.1082,
       "step": 1900
     },
     {
       "epoch": 17.7,
       "learning_rate": 0.0002584166666666666,
-      "loss": 1.0481,
+      "loss": 1.0695,
       "step": 2000
     },
     {
       "epoch": 17.7,
-      "eval_loss": 0.6849815845489502,
-      "eval_runtime": 54.2403,
-      "eval_samples_per_second": 31.397,
-      "eval_steps_per_second": 0.996,
-      "eval_wer": 0.5358525345622119,
+      "eval_loss": 0.7119461894035339,
+      "eval_runtime": 51.1988,
+      "eval_samples_per_second": 33.263,
+      "eval_steps_per_second": 1.055,
+      "eval_wer": 0.544147465437788,
       "step": 2000
     },
     {
       "epoch": 18.58,
       "learning_rate": 0.0002556388888888889,
-      "loss": 1.0495,
+      "loss": 1.064,
       "step": 2100
     },
     {
       "epoch": 19.47,
       "learning_rate": 0.0002528611111111111,
-      "loss": 1.0119,
+      "loss": 1.0327,
       "step": 2200
     },
     {
       "epoch": 20.35,
       "learning_rate": 0.0002500833333333333,
-      "loss": 1.0156,
+      "loss": 1.0353,
       "step": 2300
     },
     {
       "epoch": 21.24,
       "learning_rate": 0.00024730555555555557,
-      "loss": 0.9916,
+      "loss": 1.0108,
       "step": 2400
     },
     {
       "epoch": 22.12,
       "learning_rate": 0.00024452777777777776,
-      "loss": 0.9837,
+      "loss": 1.0104,
       "step": 2500
     },
     {
       "epoch": 22.12,
-      "eval_loss": 0.6336787343025208,
-      "eval_runtime": 55.0498,
-      "eval_samples_per_second": 30.936,
-      "eval_steps_per_second": 0.981,
-      "eval_wer": 0.5316129032258065,
+      "eval_loss": 0.6069046258926392,
+      "eval_runtime": 51.5986,
+      "eval_samples_per_second": 33.005,
+      "eval_steps_per_second": 1.047,
+      "eval_wer": 0.5295852534562212,
       "step": 2500
     },
     {
       "epoch": 23.01,
       "learning_rate": 0.00024174999999999997,
-      "loss": 0.9558,
+      "loss": 0.9806,
       "step": 2600
     },
     {
       "epoch": 23.89,
       "learning_rate": 0.0002389722222222222,
-      "loss": 0.9523,
+      "loss": 0.9712,
       "step": 2700
     },
     {
       "epoch": 24.78,
       "learning_rate": 0.00023619444444444442,
-      "loss": 0.946,
+      "loss": 0.9703,
       "step": 2800
     },
     {
       "epoch": 25.66,
       "learning_rate": 0.00023341666666666663,
-      "loss": 0.909,
+      "loss": 0.9298,
       "step": 2900
     },
     {
       "epoch": 26.55,
       "learning_rate": 0.00023063888888888887,
-      "loss": 0.9108,
+      "loss": 0.9299,
       "step": 3000
     },
     {
       "epoch": 26.55,
-      "eval_loss": 0.6257887482643127,
-      "eval_runtime": 55.6823,
-      "eval_samples_per_second": 30.584,
-      "eval_steps_per_second": 0.97,
-      "eval_wer": 0.507926267281106,
+      "eval_loss": 0.6168299913406372,
+      "eval_runtime": 51.1346,
+      "eval_samples_per_second": 33.304,
+      "eval_steps_per_second": 1.056,
+      "eval_wer": 0.5206451612903226,
       "step": 3000
     },
     {
       "epoch": 27.43,
       "learning_rate": 0.00022786111111111108,
-      "loss": 0.9005,
+      "loss": 0.9159,
       "step": 3100
     },
     {
       "epoch": 28.32,
       "learning_rate": 0.00022508333333333332,
-      "loss": 0.9069,
+      "loss": 0.922,
       "step": 3200
     },
     {
       "epoch": 29.2,
       "learning_rate": 0.00022230555555555553,
-      "loss": 0.8605,
+      "loss": 0.8775,
       "step": 3300
     },
     {
       "epoch": 30.09,
       "learning_rate": 0.00021952777777777774,
-      "loss": 0.8815,
+      "loss": 0.889,
       "step": 3400
     },
     {
       "epoch": 30.97,
       "learning_rate": 0.00021674999999999998,
-      "loss": 0.8439,
+      "loss": 0.8588,
       "step": 3500
     },
     {
       "epoch": 30.97,
-      "eval_loss": 0.6301265954971313,
-      "eval_runtime": 54.5193,
-      "eval_samples_per_second": 31.237,
-      "eval_steps_per_second": 0.99,
-      "eval_wer": 0.48884792626728113,
+      "eval_loss": 0.6382256150245667,
+      "eval_runtime": 50.9107,
+      "eval_samples_per_second": 33.451,
+      "eval_steps_per_second": 1.061,
+      "eval_wer": 0.5170506912442396,
       "step": 3500
     },
     {
       "epoch": 31.86,
       "learning_rate": 0.0002139722222222222,
-      "loss": 0.8522,
+      "loss": 0.865,
       "step": 3600
     },
     {
       "epoch": 32.74,
       "learning_rate": 0.0002111944444444444,
-      "loss": 0.8477,
+      "loss": 0.8679,
       "step": 3700
     },
     {
       "epoch": 33.63,
       "learning_rate": 0.00020841666666666665,
-      "loss": 0.7978,
+      "loss": 0.8034,
       "step": 3800
     },
     {
       "epoch": 34.51,
       "learning_rate": 0.00020563888888888886,
-      "loss": 0.8127,
+      "loss": 0.7997,
       "step": 3900
     },
     {
       "epoch": 35.4,
       "learning_rate": 0.0002028611111111111,
-      "loss": 0.7901,
+      "loss": 0.7942,
       "step": 4000
     },
     {
       "epoch": 35.4,
-      "eval_loss": 0.6244927048683167,
-      "eval_runtime": 53.6992,
-      "eval_samples_per_second": 31.714,
-      "eval_steps_per_second": 1.006,
-      "eval_wer": 0.4976958525345622,
+      "eval_loss": 0.6048025488853455,
+      "eval_runtime": 50.6633,
+      "eval_samples_per_second": 33.614,
+      "eval_steps_per_second": 1.066,
+      "eval_wer": 0.49880184331797234,
       "step": 4000
     },
     {
       "epoch": 36.28,
-      "learning_rate": 0.0002000833333333333,
-      "loss": 0.7978,
+      "learning_rate": 0.00020011111111111108,
+      "loss": 0.796,
       "step": 4100
     },
     {
       "epoch": 37.17,
-      "learning_rate": 0.00019730555555555552,
-      "loss": 0.8046,
+      "learning_rate": 0.00019733333333333332,
+      "loss": 0.8013,
       "step": 4200
     },
     {
       "epoch": 38.05,
-      "learning_rate": 0.00019452777777777776,
-      "loss": 0.7892,
+      "learning_rate": 0.00019455555555555554,
+      "loss": 0.7807,
       "step": 4300
     },
     {
       "epoch": 38.94,
-      "learning_rate": 0.00019174999999999997,
-      "loss": 0.7657,
+      "learning_rate": 0.00019177777777777777,
+      "loss": 0.7579,
       "step": 4400
     },
     {
       "epoch": 39.82,
       "learning_rate": 0.00018899999999999999,
-      "loss": 0.7669,
+      "loss": 0.7808,
       "step": 4500
     },
     {
       "epoch": 39.82,
-      "eval_loss": 0.6164370775222778,
-      "eval_runtime": 54.5713,
-      "eval_samples_per_second": 31.207,
-      "eval_steps_per_second": 0.99,
-      "eval_wer": 0.4671889400921659,
+      "eval_loss": 0.6730400919914246,
+      "eval_runtime": 50.5688,
+      "eval_samples_per_second": 33.677,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.5083870967741936,
       "step": 4500
     },
     {
       "epoch": 40.71,
       "learning_rate": 0.0001862222222222222,
-      "loss": 0.7389,
+      "loss": 0.7556,
       "step": 4600
     },
     {
       "epoch": 41.59,
       "learning_rate": 0.00018344444444444444,
-      "loss": 0.7595,
+      "loss": 0.767,
       "step": 4700
     },
     {
       "epoch": 42.48,
       "learning_rate": 0.00018066666666666665,
-      "loss": 0.7264,
+      "loss": 0.7401,
       "step": 4800
     },
     {
       "epoch": 43.36,
       "learning_rate": 0.00017788888888888886,
-      "loss": 0.7251,
+      "loss": 0.7413,
       "step": 4900
     },
     {
       "epoch": 44.25,
       "learning_rate": 0.0001751111111111111,
-      "loss": 0.7196,
+      "loss": 0.743,
       "step": 5000
     },
     {
       "epoch": 44.25,
-      "eval_loss": 0.6039230227470398,
-      "eval_runtime": 55.8782,
-      "eval_samples_per_second": 30.477,
-      "eval_steps_per_second": 0.966,
-      "eval_wer": 0.4688479262672811,
+      "eval_loss": 0.6749109625816345,
+      "eval_runtime": 50.7919,
+      "eval_samples_per_second": 33.529,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.5011981566820276,
       "step": 5000
     },
     {
       "epoch": 45.13,
       "learning_rate": 0.0001723333333333333,
-      "loss": 0.6932,
+      "loss": 0.6921,
       "step": 5100
     },
     {
       "epoch": 46.02,
       "learning_rate": 0.00016955555555555555,
-      "loss": 0.6983,
+      "loss": 0.7032,
       "step": 5200
     },
     {
       "epoch": 46.9,
       "learning_rate": 0.00016677777777777776,
-      "loss": 0.6769,
+      "loss": 0.688,
       "step": 5300
     },
     {
       "epoch": 47.79,
       "learning_rate": 0.00016399999999999997,
-      "loss": 0.6928,
+      "loss": 0.6842,
       "step": 5400
     },
     {
       "epoch": 48.67,
       "learning_rate": 0.00016122222222222221,
-      "loss": 0.6715,
+      "loss": 0.6652,
       "step": 5500
     },
     {
       "epoch": 48.67,
-      "eval_loss": 0.5900057554244995,
-      "eval_runtime": 53.576,
-      "eval_samples_per_second": 31.787,
-      "eval_steps_per_second": 1.008,
-      "eval_wer": 0.45732718894009217,
+      "eval_loss": 0.6491430401802063,
+      "eval_runtime": 50.5445,
+      "eval_samples_per_second": 33.693,
+      "eval_steps_per_second": 1.068,
+      "eval_wer": 0.4735483870967742,
       "step": 5500
     },
     {
       "epoch": 49.56,
-      "learning_rate": 0.00015849999999999998,
-      "loss": 0.6833,
+      "learning_rate": 0.00015844444444444443,
+      "loss": 0.6798,
       "step": 5600
     },
     {
       "epoch": 50.44,
-      "learning_rate": 0.0001557222222222222,
-      "loss": 0.6673,
+      "learning_rate": 0.00015566666666666664,
+      "loss": 0.6639,
       "step": 5700
     },
     {
       "epoch": 51.33,
-      "learning_rate": 0.00015294444444444443,
-      "loss": 0.6791,
+      "learning_rate": 0.00015288888888888888,
+      "loss": 0.6691,
       "step": 5800
     },
     {
       "epoch": 52.21,
-      "learning_rate": 0.00015016666666666664,
-      "loss": 0.6292,
+      "learning_rate": 0.0001501111111111111,
+      "loss": 0.6275,
       "step": 5900
     },
     {
       "epoch": 53.1,
-      "learning_rate": 0.00014738888888888888,
-      "loss": 0.6441,
+      "learning_rate": 0.00014733333333333333,
+      "loss": 0.6386,
       "step": 6000
     },
     {
       "epoch": 53.1,
-      "eval_loss": 0.7002069354057312,
-      "eval_runtime": 55.3769,
-      "eval_samples_per_second": 30.753,
-      "eval_steps_per_second": 0.975,
-      "eval_wer": 0.479815668202765,
+      "eval_loss": 0.6927830576896667,
+      "eval_runtime": 50.5958,
+      "eval_samples_per_second": 33.659,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.49539170506912444,
       "step": 6000
     },
     {
       "epoch": 53.98,
-      "learning_rate": 0.0001446111111111111,
-      "loss": 0.6217,
+      "learning_rate": 0.00014455555555555554,
+      "loss": 0.6288,
       "step": 6100
     },
     {
       "epoch": 54.87,
-      "learning_rate": 0.00014183333333333333,
-      "loss": 0.6131,
+      "learning_rate": 0.00014177777777777778,
+      "loss": 0.6247,
       "step": 6200
     },
     {
       "epoch": 55.75,
-      "learning_rate": 0.00013905555555555554,
-      "loss": 0.6211,
+      "learning_rate": 0.00013902777777777777,
+      "loss": 0.6245,
       "step": 6300
     },
     {
       "epoch": 56.64,
-      "learning_rate": 0.00013627777777777775,
-      "loss": 0.6104,
+      "learning_rate": 0.00013624999999999998,
+      "loss": 0.606,
       "step": 6400
     },
     {
       "epoch": 57.52,
-      "learning_rate": 0.0001335,
-      "loss": 0.5938,
+      "learning_rate": 0.00013347222222222222,
+      "loss": 0.5945,
       "step": 6500
     },
     {
       "epoch": 57.52,
-      "eval_loss": 0.6249451637268066,
-      "eval_runtime": 55.668,
-      "eval_samples_per_second": 30.592,
-      "eval_steps_per_second": 0.97,
-      "eval_wer": 0.4578801843317972,
+      "eval_loss": 0.6358979344367981,
+      "eval_runtime": 51.0247,
+      "eval_samples_per_second": 33.376,
+      "eval_steps_per_second": 1.058,
+      "eval_wer": 0.479815668202765,
       "step": 6500
     },
     {
       "epoch": 58.41,
       "learning_rate": 0.0001307222222222222,
-      "loss": 0.6015,
+      "loss": 0.6067,
       "step": 6600
     },
     {
       "epoch": 59.29,
       "learning_rate": 0.00012794444444444442,
-      "loss": 0.591,
+      "loss": 0.5917,
       "step": 6700
     },
     {
       "epoch": 60.18,
       "learning_rate": 0.00012516666666666666,
-      "loss": 0.5734,
+      "loss": 0.5729,
       "step": 6800
     },
     {
       "epoch": 61.06,
       "learning_rate": 0.0001223888888888889,
-      "loss": 0.5685,
+      "loss": 0.5759,
       "step": 6900
     },
     {
       "epoch": 61.95,
       "learning_rate": 0.0001196111111111111,
-      "loss": 0.5541,
+      "loss": 0.5561,
       "step": 7000
     },
     {
       "epoch": 61.95,
-      "eval_loss": 0.6184473633766174,
-      "eval_runtime": 55.2876,
-      "eval_samples_per_second": 30.803,
-      "eval_steps_per_second": 0.977,
-      "eval_wer": 0.4424884792626728,
+      "eval_loss": 0.6409346461296082,
+      "eval_runtime": 51.2819,
+      "eval_samples_per_second": 33.209,
+      "eval_steps_per_second": 1.053,
+      "eval_wer": 0.4799078341013825,
       "step": 7000
     },
     {
       "epoch": 62.83,
       "learning_rate": 0.00011683333333333332,
-      "loss": 0.5546,
+      "loss": 0.5496,
       "step": 7100
     },
     {
       "epoch": 63.72,
       "learning_rate": 0.00011405555555555554,
-      "loss": 0.5473,
+      "loss": 0.5541,
       "step": 7200
     },
     {
       "epoch": 64.6,
       "learning_rate": 0.00011127777777777777,
-      "loss": 0.5592,
+      "loss": 0.555,
       "step": 7300
     },
     {
       "epoch": 65.49,
       "learning_rate": 0.0001085,
-      "loss": 0.5349,
+      "loss": 0.5393,
       "step": 7400
     },
     {
       "epoch": 66.37,
       "learning_rate": 0.0001057222222222222,
-      "loss": 0.5506,
+      "loss": 0.5464,
       "step": 7500
     },
     {
       "epoch": 66.37,
-      "eval_loss": 0.6962713003158569,
-      "eval_runtime": 55.4374,
-      "eval_samples_per_second": 30.719,
-      "eval_steps_per_second": 0.974,
-      "eval_wer": 0.45852534562211983,
+      "eval_loss": 0.6451661586761475,
+      "eval_runtime": 50.5882,
+      "eval_samples_per_second": 33.664,
+      "eval_steps_per_second": 1.067,
+      "eval_wer": 0.46912442396313364,
       "step": 7500
     },
     {
       "epoch": 67.26,
       "learning_rate": 0.00010294444444444443,
-      "loss": 0.5313,
+      "loss": 0.5308,
       "step": 7600
     },
     {
       "epoch": 68.14,
       "learning_rate": 0.00010016666666666666,
-      "loss": 0.5267,
+      "loss": 0.5304,
       "step": 7700
     },
     {
       "epoch": 69.03,
       "learning_rate": 9.738888888888888e-05,
-      "loss": 0.5222,
+      "loss": 0.5278,
       "step": 7800
     },
     {
       "epoch": 69.91,
       "learning_rate": 9.46111111111111e-05,
-      "loss": 0.5101,
+      "loss": 0.5112,
       "step": 7900
     },
     {
       "epoch": 70.8,
       "learning_rate": 9.183333333333332e-05,
-      "loss": 0.4998,
+      "loss": 0.5119,
       "step": 8000
     },
     {
       "epoch": 70.8,
-      "eval_loss": 0.6778160333633423,
-      "eval_runtime": 56.1738,
-      "eval_samples_per_second": 30.317,
-      "eval_steps_per_second": 0.961,
-      "eval_wer": 0.44682027649769585,
+      "eval_loss": 0.6376333832740784,
+      "eval_runtime": 50.3564,
+      "eval_samples_per_second": 33.819,
+      "eval_steps_per_second": 1.072,
+      "eval_wer": 0.4657142857142857,
       "step": 8000
     },
     {
       "epoch": 71.68,
       "learning_rate": 8.905555555555555e-05,
-      "loss": 0.4941,
+      "loss": 0.4974,
       "step": 8100
     },
     {
       "epoch": 72.57,
       "learning_rate": 8.627777777777776e-05,
-      "loss": 0.492,
+      "loss": 0.4986,
       "step": 8200
     },
     {
       "epoch": 73.45,
       "learning_rate": 8.349999999999998e-05,
-      "loss": 0.4741,
+      "loss": 0.4752,
       "step": 8300
     },
     {
       "epoch": 74.34,
       "learning_rate": 8.072222222222222e-05,
-      "loss": 0.495,
+      "loss": 0.4947,
       "step": 8400
     },
     {
       "epoch": 75.22,
-      "learning_rate": 7.794444444444445e-05,
-      "loss": 0.4729,
+      "learning_rate": 7.797222222222222e-05,
+      "loss": 0.474,
       "step": 8500
     },
     {
       "epoch": 75.22,
-      "eval_loss": 0.6383044719696045,
-      "eval_runtime": 53.9538,
-      "eval_samples_per_second": 31.564,
-      "eval_steps_per_second": 1.001,
-      "eval_wer": 0.4392626728110599,
+      "eval_loss": 0.6540603041648865,
+      "eval_runtime": 50.3966,
+      "eval_samples_per_second": 33.792,
+      "eval_steps_per_second": 1.072,
+      "eval_wer": 0.46995391705069123,
       "step": 8500
     },
     {
       "epoch": 76.11,
-      "learning_rate": 7.516666666666665e-05,
-      "loss": 0.4696,
+      "learning_rate": 7.519444444444445e-05,
+      "loss": 0.4724,
       "step": 8600
     },
     {
       "epoch": 76.99,
-      "learning_rate": 7.238888888888889e-05,
-      "loss": 0.4581,
+      "learning_rate": 7.241666666666666e-05,
+      "loss": 0.4668,
       "step": 8700
     },
     {
       "epoch": 77.88,
-      "learning_rate": 6.961111111111111e-05,
-      "loss": 0.4583,
+      "learning_rate": 6.963888888888889e-05,
+      "loss": 0.46,
       "step": 8800
     },
     {
       "epoch": 78.76,
-      "learning_rate": 6.683333333333332e-05,
-      "loss": 0.4451,
+      "learning_rate": 6.68611111111111e-05,
+      "loss": 0.45,
       "step": 8900
     },
     {
       "epoch": 79.65,
-      "learning_rate": 6.405555555555555e-05,
-      "loss": 0.4535,
+      "learning_rate": 6.408333333333332e-05,
+      "loss": 0.45,
       "step": 9000
     },
     {
       "epoch": 79.65,
-      "eval_loss": 0.6592639684677124,
-      "eval_runtime": 53.7527,
-      "eval_samples_per_second": 31.682,
-      "eval_steps_per_second": 1.005,
-      "eval_wer": 0.4368663594470046,
+      "eval_loss": 0.637355625629425,
+      "eval_runtime": 50.4224,
+      "eval_samples_per_second": 33.775,
+      "eval_steps_per_second": 1.071,
+      "eval_wer": 0.45714285714285713,
       "step": 9000
     },
     {
       "epoch": 80.53,
       "learning_rate": 6.130555555555555e-05,
-      "loss": 0.4324,
+      "loss": 0.4282,
       "step": 9100
     },
     {
       "epoch": 81.42,
       "learning_rate": 5.8527777777777774e-05,
-      "loss": 0.4546,
+      "loss": 0.4455,
       "step": 9200
     },
     {
       "epoch": 82.3,
       "learning_rate": 5.574999999999999e-05,
-      "loss": 0.4391,
+      "loss": 0.4423,
       "step": 9300
     },
     {
       "epoch": 83.19,
       "learning_rate": 5.297222222222222e-05,
-      "loss": 0.4306,
+      "loss": 0.4262,
       "step": 9400
     },
     {
       "epoch": 84.07,
       "learning_rate": 5.019444444444444e-05,
-      "loss": 0.4358,
+      "loss": 0.4315,
       "step": 9500
     },
     {
       "epoch": 84.07,
-      "eval_loss": 0.6913911700248718,
-      "eval_runtime": 55.4673,
-      "eval_samples_per_second": 30.703,
-      "eval_steps_per_second": 0.974,
-      "eval_wer": 0.4422119815668203,
+      "eval_loss": 0.6568067073822021,
+      "eval_runtime": 50.7772,
+      "eval_samples_per_second": 33.539,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.46248847926267284,
       "step": 9500
     },
     {
       "epoch": 84.96,
       "learning_rate": 4.741666666666666e-05,
-      "loss": 0.4095,
+      "loss": 0.4069,
       "step": 9600
     },
     {
       "epoch": 85.84,
       "learning_rate": 4.463888888888888e-05,
-      "loss": 0.4148,
+      "loss": 0.416,
       "step": 9700
     },
     {
       "epoch": 86.73,
       "learning_rate": 4.186111111111111e-05,
-      "loss": 0.4113,
+      "loss": 0.4106,
       "step": 9800
     },
     {
       "epoch": 87.61,
       "learning_rate": 3.9083333333333326e-05,
-      "loss": 0.405,
+      "loss": 0.4025,
       "step": 9900
     },
     {
       "epoch": 88.5,
       "learning_rate": 3.630555555555555e-05,
-      "loss": 0.402,
+      "loss": 0.3967,
       "step": 10000
     },
     {
       "epoch": 88.5,
-      "eval_loss": 0.6743763089179993,
-      "eval_runtime": 53.7113,
-      "eval_samples_per_second": 31.707,
-      "eval_steps_per_second": 1.005,
-      "eval_wer": 0.4269124423963134,
+      "eval_loss": 0.6636261343955994,
+      "eval_runtime": 50.7873,
+      "eval_samples_per_second": 33.532,
+      "eval_steps_per_second": 1.063,
+      "eval_wer": 0.46046082949308753,
       "step": 10000
     },
     {
       "epoch": 89.38,
       "learning_rate": 3.352777777777777e-05,
-      "loss": 0.3915,
+      "loss": 0.3857,
       "step": 10100
     },
     {
       "epoch": 90.27,
       "learning_rate": 3.0749999999999995e-05,
-      "loss": 0.394,
+      "loss": 0.3991,
       "step": 10200
     },
     {
       "epoch": 91.15,
       "learning_rate": 2.7972222222222217e-05,
-      "loss": 0.392,
+      "loss": 0.3899,
       "step": 10300
     },
     {
       "epoch": 92.04,
       "learning_rate": 2.519444444444444e-05,
-      "loss": 0.3937,
+      "loss": 0.3965,
       "step": 10400
     },
     {
       "epoch": 92.92,
       "learning_rate": 2.2416666666666665e-05,
-      "loss": 0.3946,
+      "loss": 0.3937,
       "step": 10500
     },
     {
       "epoch": 92.92,
-      "eval_loss": 0.6894700527191162,
-      "eval_runtime": 55.0145,
-      "eval_samples_per_second": 30.955,
-      "eval_steps_per_second": 0.982,
-      "eval_wer": 0.42746543778801843,
+      "eval_loss": 0.6537252068519592,
+      "eval_runtime": 49.8207,
+      "eval_samples_per_second": 34.183,
+      "eval_steps_per_second": 1.084,
+      "eval_wer": 0.4597235023041475,
       "step": 10500
     },
     {
       "epoch": 93.81,
       "learning_rate": 1.9638888888888887e-05,
-      "loss": 0.3881,
+      "loss": 0.3808,
       "step": 10600
     },
     {
       "epoch": 94.69,
       "learning_rate": 1.686111111111111e-05,
-      "loss": 0.3778,
+      "loss": 0.3701,
       "step": 10700
     },
     {
       "epoch": 95.58,
       "learning_rate": 1.4083333333333331e-05,
-      "loss": 0.3783,
+      "loss": 0.3746,
       "step": 10800
     },
     {
       "epoch": 96.46,
       "learning_rate": 1.1305555555555553e-05,
-      "loss": 0.3612,
+      "loss": 0.3647,
       "step": 10900
     },
     {
       "epoch": 97.35,
       "learning_rate": 8.527777777777777e-06,
-      "loss": 0.3734,
+      "loss": 0.3788,
       "step": 11000
     },
     {
       "epoch": 97.35,
-      "eval_loss": 0.6888979077339172,
-      "eval_runtime": 55.289,
-      "eval_samples_per_second": 30.802,
-      "eval_steps_per_second": 0.977,
-      "eval_wer": 0.4319815668202765,
+      "eval_loss": 0.6614137887954712,
+      "eval_runtime": 50.2299,
+      "eval_samples_per_second": 33.904,
+      "eval_steps_per_second": 1.075,
+      "eval_wer": 0.45889400921658985,
       "step": 11000
     },
     {
       "epoch": 98.23,
       "learning_rate": 5.749999999999999e-06,
-      "loss": 0.3635,
+      "loss": 0.3666,
       "step": 11100
     },
     {
       "epoch": 99.12,
-      "learning_rate": 2.9999999999999997e-06,
-      "loss": 0.3775,
+      "learning_rate": 2.972222222222222e-06,
+      "loss": 0.3701,
       "step": 11200
     },
     {
       "epoch": 100.0,
-      "learning_rate": 2.222222222222222e-07,
-      "loss": 0.3737,
+      "learning_rate": 1.9444444444444442e-07,
+      "loss": 0.3755,
       "step": 11300
     },
     {
       "epoch": 100.0,
       "step": 11300,
-      "total_flos": 4.438667922365353e+19,
-      "train_loss": 0.9041624231254105,
-      "train_runtime": 18573.3759,
-      "train_samples_per_second": 19.388,
-      "train_steps_per_second": 0.608
+      "total_flos": 4.438624660507792e+19,
+      "train_loss": 0.9130245208740234,
+      "train_runtime": 17767.5681,
+      "train_samples_per_second": 20.267,
+      "train_steps_per_second": 0.636
     }
   ],
   "max_steps": 11300,
   "num_train_epochs": 100,
-  "total_flos": 4.438667922365353e+19,
+  "total_flos": 4.438624660507792e+19,
   "trial_name": null,
   "trial_params": null
 }