End of training

all_results.json

@@ -1,12 +1,12 @@
 {
-    "epoch": 3.97,
-    "eval_loss": 0.21749083697795868,
-    "eval_runtime": 4779.1257,
-    "eval_samples_per_second": 1.055,
-    "eval_steps_per_second": 0.528,
-    "eval_wer": 9.261960952585282,
-    "train_loss": 0.3629964433670044,
-    "train_runtime": 41790.9887,
-    "train_samples_per_second": 0.479,
-    "train_steps_per_second": 0.12
+    "epoch": 2.42,
+    "eval_loss": 0.26386478543281555,
+    "eval_runtime": 855.6,
+    "eval_samples_per_second": 4.233,
+    "eval_steps_per_second": 2.117,
+    "eval_wer": 14.566367609845871,
+    "train_loss": 0.37738304929733274,
+    "train_runtime": 7350.5739,
+    "train_samples_per_second": 2.721,
+    "train_steps_per_second": 0.68
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 3.97,
-    "eval_loss": 0.21749083697795868,
-    "eval_runtime": 4779.1257,
-    "eval_samples_per_second": 1.055,
-    "eval_steps_per_second": 0.528,
-    "eval_wer": 9.261960952585282
+    "epoch": 2.42,
+    "eval_loss": 0.26386478543281555,
+    "eval_runtime": 855.6,
+    "eval_samples_per_second": 4.233,
+    "eval_steps_per_second": 2.117,
+    "eval_wer": 14.566367609845871
 }

runs/Aug10_10-44-35_dsai-server/events.out.tfevents.1691647999.dsai-server

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:adcd09e9b365a6b2f2e66f6a41f53ee1d915f7c63d20c29b85fbc147e21348a7
+size 40

train_results.json

@@ -1,7 +1,7 @@
 {
-    "epoch": 3.97,
-    "train_loss": 0.3629964433670044,
-    "train_runtime": 41790.9887,
-    "train_samples_per_second": 0.479,
-    "train_steps_per_second": 0.12
+    "epoch": 2.42,
+    "train_loss": 0.37738304929733274,
+    "train_runtime": 7350.5739,
+    "train_samples_per_second": 2.721,
+    "train_steps_per_second": 0.68
 }

trainer_state.json

@@ -1,1270 +1,1270 @@
 {
-  "best_metric": 9.261960952585282,
-  "best_model_checkpoint": "./output/small/cv9-special-batch4-lr6-small/checkpoint-5000",
-  "epoch": 3.965107057890563,
+  "best_metric": 14.566367609845871,
+  "best_model_checkpoint": "./output/small/cv9-special-batch4-lr6-small/checkpoint-4000",
+  "epoch": 2.420135527589545,
   "global_step": 5000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 4.4e-08,
-      "loss": 1.4649,
+      "loss": 1.3439,
       "step": 25
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.02,
       "learning_rate": 9.4e-08,
-      "loss": 1.399,
+      "loss": 1.2478,
       "step": 50
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.04,
       "learning_rate": 1.44e-07,
-      "loss": 1.3759,
+      "loss": 1.3734,
       "step": 75
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.05,
       "learning_rate": 1.94e-07,
-      "loss": 1.3914,
+      "loss": 1.3701,
       "step": 100
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.06,
       "learning_rate": 2.4399999999999996e-07,
-      "loss": 1.3497,
+      "loss": 1.2682,
       "step": 125
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.07,
       "learning_rate": 2.9399999999999996e-07,
-      "loss": 1.2231,
+      "loss": 1.2577,
       "step": 150
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.08,
       "learning_rate": 3.4399999999999996e-07,
-      "loss": 1.2017,
+      "loss": 1.2112,
       "step": 175
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.1,
       "learning_rate": 3.94e-07,
-      "loss": 1.1736,
+      "loss": 1.121,
       "step": 200
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.11,
       "learning_rate": 4.44e-07,
-      "loss": 1.0267,
+      "loss": 0.9726,
       "step": 225
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.12,
       "learning_rate": 4.94e-07,
-      "loss": 0.9892,
+      "loss": 0.9854,
       "step": 250
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.13,
       "learning_rate": 5.44e-07,
-      "loss": 0.8075,
+      "loss": 0.9003,
       "step": 275
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.15,
       "learning_rate": 5.939999999999999e-07,
-      "loss": 0.7824,
+      "loss": 0.764,
       "step": 300
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.16,
       "learning_rate": 6.44e-07,
-      "loss": 0.6551,
+      "loss": 0.7079,
       "step": 325
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.17,
       "learning_rate": 6.939999999999999e-07,
-      "loss": 0.5733,
+      "loss": 0.6626,
       "step": 350
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.18,
       "learning_rate": 7.44e-07,
-      "loss": 0.5097,
+      "loss": 0.6055,
       "step": 375
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.19,
       "learning_rate": 7.94e-07,
-      "loss": 0.5427,
+      "loss": 0.5399,
       "step": 400
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.21,
       "learning_rate": 8.439999999999999e-07,
-      "loss": 0.5368,
+      "loss": 0.4842,
       "step": 425
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.22,
       "learning_rate": 8.939999999999999e-07,
-      "loss": 0.549,
+      "loss": 0.4666,
       "step": 450
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.23,
       "learning_rate": 9.439999999999999e-07,
-      "loss": 0.5531,
+      "loss": 0.4912,
       "step": 475
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.24,
       "learning_rate": 9.94e-07,
-      "loss": 0.5898,
+      "loss": 0.4533,
       "step": 500
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.25,
       "learning_rate": 9.95111111111111e-07,
-      "loss": 0.4356,
+      "loss": 0.5009,
       "step": 525
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.27,
       "learning_rate": 9.895555555555554e-07,
-      "loss": 0.4242,
+      "loss": 0.4936,
       "step": 550
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.28,
       "learning_rate": 9.84e-07,
-      "loss": 0.4773,
+      "loss": 0.5088,
       "step": 575
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.29,
       "learning_rate": 9.784444444444444e-07,
-      "loss": 0.4798,
+      "loss": 0.4588,
       "step": 600
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.3,
       "learning_rate": 9.728888888888888e-07,
-      "loss": 0.4548,
+      "loss": 0.461,
       "step": 625
     },
     {
-      "epoch": 0.52,
+      "epoch": 0.31,
       "learning_rate": 9.673333333333332e-07,
-      "loss": 0.4685,
+      "loss": 0.389,
       "step": 650
     },
     {
-      "epoch": 0.54,
+      "epoch": 0.33,
       "learning_rate": 9.617777777777776e-07,
-      "loss": 0.4425,
+      "loss": 0.3996,
       "step": 675
     },
     {
-      "epoch": 0.56,
+      "epoch": 0.34,
       "learning_rate": 9.562222222222223e-07,
-      "loss": 0.491,
+      "loss": 0.4729,
       "step": 700
     },
     {
-      "epoch": 0.57,
+      "epoch": 0.35,
       "learning_rate": 9.506666666666667e-07,
-      "loss": 0.4578,
+      "loss": 0.4833,
       "step": 725
     },
     {
-      "epoch": 0.59,
+      "epoch": 0.36,
       "learning_rate": 9.451111111111111e-07,
-      "loss": 0.3686,
+      "loss": 0.4696,
       "step": 750
     },
     {
-      "epoch": 0.61,
+      "epoch": 0.38,
       "learning_rate": 9.395555555555556e-07,
-      "loss": 0.4792,
+      "loss": 0.4405,
       "step": 775
     },
     {
-      "epoch": 0.63,
+      "epoch": 0.39,
       "learning_rate": 9.34e-07,
-      "loss": 0.4405,
+      "loss": 0.4392,
       "step": 800
     },
     {
-      "epoch": 0.65,
+      "epoch": 0.4,
       "learning_rate": 9.284444444444444e-07,
-      "loss": 0.4577,
+      "loss": 0.4157,
       "step": 825
     },
     {
-      "epoch": 0.67,
+      "epoch": 0.41,
       "learning_rate": 9.228888888888888e-07,
-      "loss": 0.437,
+      "loss": 0.4139,
       "step": 850
     },
     {
-      "epoch": 0.69,
+      "epoch": 0.42,
       "learning_rate": 9.173333333333333e-07,
-      "loss": 0.4542,
+      "loss": 0.3918,
       "step": 875
     },
     {
-      "epoch": 0.71,
+      "epoch": 0.44,
       "learning_rate": 9.117777777777778e-07,
-      "loss": 0.4309,
+      "loss": 0.4873,
       "step": 900
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.45,
       "learning_rate": 9.062222222222222e-07,
-      "loss": 0.4273,
+      "loss": 0.3795,
       "step": 925
     },
     {
-      "epoch": 0.75,
+      "epoch": 0.46,
       "learning_rate": 9.006666666666666e-07,
-      "loss": 0.4513,
+      "loss": 0.3392,
       "step": 950
     },
     {
-      "epoch": 0.77,
+      "epoch": 0.47,
       "learning_rate": 8.95111111111111e-07,
-      "loss": 0.388,
+      "loss": 0.3928,
       "step": 975
     },
     {
-      "epoch": 0.79,
+      "epoch": 0.48,
       "learning_rate": 8.895555555555555e-07,
-      "loss": 0.4072,
+      "loss": 0.3934,
       "step": 1000
     },
     {
-      "epoch": 0.79,
-      "eval_loss": 0.36864736676216125,
-      "eval_runtime": 6765.5386,
-      "eval_samples_per_second": 0.745,
-      "eval_steps_per_second": 0.373,
-      "eval_wer": 16.28835013945505,
+      "epoch": 0.48,
+      "eval_loss": 0.3102651834487915,
+      "eval_runtime": 863.1504,
+      "eval_samples_per_second": 4.196,
+      "eval_steps_per_second": 2.098,
+      "eval_wer": 16.977225672877847,
       "step": 1000
     },
     {
-      "epoch": 0.81,
+      "epoch": 0.5,
       "learning_rate": 8.839999999999999e-07,
-      "loss": 0.4244,
+      "loss": 0.4245,
       "step": 1025
     },
     {
-      "epoch": 0.83,
+      "epoch": 0.51,
       "learning_rate": 8.784444444444444e-07,
-      "loss": 0.426,
+      "loss": 0.4051,
       "step": 1050
     },
     {
-      "epoch": 0.85,
+      "epoch": 0.52,
       "learning_rate": 8.728888888888889e-07,
-      "loss": 0.4758,
+      "loss": 0.3125,
       "step": 1075
     },
     {
-      "epoch": 0.87,
+      "epoch": 0.53,
       "learning_rate": 8.673333333333332e-07,
-      "loss": 0.3984,
+      "loss": 0.3947,
       "step": 1100
     },
     {
-      "epoch": 0.89,
+      "epoch": 0.54,
       "learning_rate": 8.617777777777777e-07,
-      "loss": 0.4527,
+      "loss": 0.4293,
       "step": 1125
     },
     {
-      "epoch": 0.91,
+      "epoch": 0.56,
       "learning_rate": 8.562222222222222e-07,
-      "loss": 0.3662,
+      "loss": 0.3565,
       "step": 1150
     },
     {
-      "epoch": 0.93,
+      "epoch": 0.57,
       "learning_rate": 8.506666666666667e-07,
-      "loss": 0.4354,
+      "loss": 0.4308,
       "step": 1175
     },
     {
-      "epoch": 0.95,
+      "epoch": 0.58,
       "learning_rate": 8.451111111111111e-07,
-      "loss": 0.3535,
+      "loss": 0.4256,
       "step": 1200
     },
     {
-      "epoch": 0.97,
+      "epoch": 0.59,
       "learning_rate": 8.395555555555556e-07,
-      "loss": 0.4261,
+      "loss": 0.3582,
       "step": 1225
     },
     {
-      "epoch": 0.99,
+      "epoch": 0.61,
       "learning_rate": 8.34e-07,
-      "loss": 0.4224,
+      "loss": 0.3952,
       "step": 1250
     },
     {
-      "epoch": 1.01,
+      "epoch": 0.62,
       "learning_rate": 8.284444444444444e-07,
-      "loss": 0.3804,
+      "loss": 0.3465,
       "step": 1275
     },
     {
-      "epoch": 1.03,
+      "epoch": 0.63,
       "learning_rate": 8.228888888888889e-07,
-      "loss": 0.3198,
+      "loss": 0.3402,
       "step": 1300
     },
     {
-      "epoch": 1.05,
+      "epoch": 0.64,
       "learning_rate": 8.173333333333333e-07,
-      "loss": 0.3635,
+      "loss": 0.3821,
       "step": 1325
     },
     {
-      "epoch": 1.07,
+      "epoch": 0.65,
       "learning_rate": 8.117777777777778e-07,
-      "loss": 0.3188,
+      "loss": 0.3882,
       "step": 1350
     },
     {
-      "epoch": 1.09,
+      "epoch": 0.67,
       "learning_rate": 8.062222222222221e-07,
-      "loss": 0.327,
+      "loss": 0.4284,
       "step": 1375
     },
     {
-      "epoch": 1.11,
+      "epoch": 0.68,
       "learning_rate": 8.006666666666666e-07,
-      "loss": 0.3135,
+      "loss": 0.3802,
       "step": 1400
     },
     {
-      "epoch": 1.13,
+      "epoch": 0.69,
       "learning_rate": 7.95111111111111e-07,
-      "loss": 0.308,
+      "loss": 0.4234,
       "step": 1425
     },
     {
-      "epoch": 1.15,
+      "epoch": 0.7,
       "learning_rate": 7.895555555555555e-07,
-      "loss": 0.3342,
+      "loss": 0.4355,
       "step": 1450
     },
     {
-      "epoch": 1.17,
+      "epoch": 0.71,
       "learning_rate": 7.84e-07,
-      "loss": 0.2946,
+      "loss": 0.3317,
       "step": 1475
     },
     {
-      "epoch": 1.19,
+      "epoch": 0.73,
       "learning_rate": 7.784444444444444e-07,
-      "loss": 0.338,
+      "loss": 0.3886,
       "step": 1500
     },
     {
-      "epoch": 1.21,
+      "epoch": 0.74,
       "learning_rate": 7.728888888888888e-07,
-      "loss": 0.2992,
+      "loss": 0.3849,
       "step": 1525
     },
     {
-      "epoch": 1.23,
+      "epoch": 0.75,
       "learning_rate": 7.673333333333332e-07,
-      "loss": 0.3421,
+      "loss": 0.4225,
       "step": 1550
     },
     {
-      "epoch": 1.25,
+      "epoch": 0.76,
       "learning_rate": 7.617777777777778e-07,
-      "loss": 0.3422,
+      "loss": 0.3237,
       "step": 1575
     },
     {
-      "epoch": 1.27,
+      "epoch": 0.77,
       "learning_rate": 7.562222222222222e-07,
-      "loss": 0.2927,
+      "loss": 0.3369,
       "step": 1600
     },
     {
-      "epoch": 1.29,
+      "epoch": 0.79,
       "learning_rate": 7.506666666666667e-07,
-      "loss": 0.329,
+      "loss": 0.3628,
       "step": 1625
     },
     {
-      "epoch": 1.31,
+      "epoch": 0.8,
       "learning_rate": 7.451111111111111e-07,
-      "loss": 0.3486,
+      "loss": 0.3315,
       "step": 1650
     },
     {
-      "epoch": 1.33,
+      "epoch": 0.81,
       "learning_rate": 7.395555555555555e-07,
-      "loss": 0.2915,
+      "loss": 0.3516,
       "step": 1675
     },
     {
-      "epoch": 1.35,
+      "epoch": 0.82,
       "learning_rate": 7.34e-07,
-      "loss": 0.3172,
+      "loss": 0.3805,
       "step": 1700
     },
     {
-      "epoch": 1.37,
+      "epoch": 0.83,
       "learning_rate": 7.284444444444444e-07,
-      "loss": 0.3376,
+      "loss": 0.3462,
       "step": 1725
     },
     {
-      "epoch": 1.39,
+      "epoch": 0.85,
       "learning_rate": 7.228888888888889e-07,
-      "loss": 0.317,
+      "loss": 0.3353,
       "step": 1750
     },
     {
-      "epoch": 1.41,
+      "epoch": 0.86,
       "learning_rate": 7.173333333333333e-07,
-      "loss": 0.3063,
+      "loss": 0.3835,
       "step": 1775
     },
     {
-      "epoch": 1.43,
+      "epoch": 0.87,
       "learning_rate": 7.117777777777777e-07,
-      "loss": 0.3421,
+      "loss": 0.3451,
       "step": 1800
     },
     {
-      "epoch": 1.45,
+      "epoch": 0.88,
       "learning_rate": 7.062222222222222e-07,
-      "loss": 0.3617,
+      "loss": 0.3736,
       "step": 1825
     },
     {
-      "epoch": 1.47,
+      "epoch": 0.9,
       "learning_rate": 7.006666666666666e-07,
-      "loss": 0.3695,
+      "loss": 0.3693,
       "step": 1850
     },
     {
-      "epoch": 1.49,
+      "epoch": 0.91,
       "learning_rate": 6.951111111111111e-07,
-      "loss": 0.3331,
+      "loss": 0.3199,
       "step": 1875
     },
     {
-      "epoch": 1.51,
+      "epoch": 0.92,
       "learning_rate": 6.895555555555555e-07,
-      "loss": 0.333,
+      "loss": 0.3639,
       "step": 1900
     },
     {
-      "epoch": 1.53,
+      "epoch": 0.93,
       "learning_rate": 6.84e-07,
-      "loss": 0.3854,
+      "loss": 0.3522,
       "step": 1925
     },
     {
-      "epoch": 1.55,
+      "epoch": 0.94,
       "learning_rate": 6.784444444444443e-07,
-      "loss": 0.3319,
+      "loss": 0.3967,
       "step": 1950
     },
     {
-      "epoch": 1.57,
+      "epoch": 0.96,
       "learning_rate": 6.728888888888888e-07,
-      "loss": 0.3786,
+      "loss": 0.3709,
       "step": 1975
     },
     {
-      "epoch": 1.59,
+      "epoch": 0.97,
       "learning_rate": 6.673333333333334e-07,
-      "loss": 0.3297,
+      "loss": 0.3279,
       "step": 2000
     },
     {
-      "epoch": 1.59,
-      "eval_loss": 0.2878722548484802,
-      "eval_runtime": 4872.6903,
-      "eval_samples_per_second": 1.035,
-      "eval_steps_per_second": 0.518,
-      "eval_wer": 12.381463205320745,
+      "epoch": 0.97,
+      "eval_loss": 0.2786530554294586,
+      "eval_runtime": 849.0799,
+      "eval_samples_per_second": 4.266,
+      "eval_steps_per_second": 2.133,
+      "eval_wer": 15.256498734759605,
       "step": 2000
     },
     {
-      "epoch": 1.61,
+      "epoch": 0.98,
       "learning_rate": 6.617777777777778e-07,
-      "loss": 0.32,
+      "loss": 0.4348,
       "step": 2025
     },
     {
-      "epoch": 1.63,
+      "epoch": 0.99,
       "learning_rate": 6.562222222222223e-07,
-      "loss": 0.3007,
+      "loss": 0.3908,
       "step": 2050
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 6.506666666666666e-07,
-      "loss": 0.3291,
+      "epoch": 1.0,
+      "learning_rate": 6.508888888888888e-07,
+      "loss": 0.3411,
       "step": 2075
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 6.451111111111111e-07,
-      "loss": 0.2643,
+      "epoch": 1.02,
+      "learning_rate": 6.453333333333333e-07,
+      "loss": 0.3441,
       "step": 2100
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 6.395555555555555e-07,
-      "loss": 0.3003,
+      "epoch": 1.03,
+      "learning_rate": 6.397777777777778e-07,
+      "loss": 0.2902,
       "step": 2125
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 6.34e-07,
-      "loss": 0.2525,
+      "epoch": 1.04,
+      "learning_rate": 6.342222222222222e-07,
+      "loss": 0.2429,
       "step": 2150
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 6.284444444444445e-07,
-      "loss": 0.3454,
+      "epoch": 1.05,
+      "learning_rate": 6.286666666666667e-07,
+      "loss": 0.248,
       "step": 2175
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 6.228888888888889e-07,
-      "loss": 0.2977,
+      "epoch": 1.06,
+      "learning_rate": 6.23111111111111e-07,
+      "loss": 0.3353,
       "step": 2200
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 6.173333333333333e-07,
-      "loss": 0.3601,
+      "epoch": 1.08,
+      "learning_rate": 6.175555555555555e-07,
+      "loss": 0.3036,
       "step": 2225
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.117777777777777e-07,
-      "loss": 0.3078,
+      "epoch": 1.09,
+      "learning_rate": 6.119999999999999e-07,
+      "loss": 0.3169,
       "step": 2250
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 6.062222222222222e-07,
-      "loss": 0.3807,
+      "epoch": 1.1,
+      "learning_rate": 6.064444444444444e-07,
+      "loss": 0.2947,
       "step": 2275
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 6.006666666666666e-07,
-      "loss": 0.2901,
+      "epoch": 1.11,
+      "learning_rate": 6.008888888888889e-07,
+      "loss": 0.2916,
       "step": 2300
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 5.951111111111111e-07,
-      "loss": 0.3622,
+      "epoch": 1.13,
+      "learning_rate": 5.953333333333333e-07,
+      "loss": 0.3078,
       "step": 2325
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 5.895555555555555e-07,
-      "loss": 0.3095,
+      "epoch": 1.14,
+      "learning_rate": 5.897777777777777e-07,
+      "loss": 0.2964,
       "step": 2350
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 5.839999999999999e-07,
-      "loss": 0.317,
+      "epoch": 1.15,
+      "learning_rate": 5.842222222222222e-07,
+      "loss": 0.2638,
       "step": 2375
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 5.784444444444444e-07,
-      "loss": 0.3335,
+      "epoch": 1.16,
+      "learning_rate": 5.786666666666667e-07,
+      "loss": 0.3004,
       "step": 2400
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 5.728888888888888e-07,
-      "loss": 0.3043,
+      "epoch": 1.17,
+      "learning_rate": 5.731111111111111e-07,
+      "loss": 0.3184,
       "step": 2425
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 5.673333333333334e-07,
-      "loss": 0.3073,
+      "epoch": 1.19,
+      "learning_rate": 5.675555555555556e-07,
+      "loss": 0.2896,
       "step": 2450
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 5.617777777777778e-07,
-      "loss": 0.3359,
+      "epoch": 1.2,
+      "learning_rate": 5.620000000000001e-07,
+      "loss": 0.2947,
       "step": 2475
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 5.562222222222222e-07,
-      "loss": 0.3164,
+      "epoch": 1.21,
+      "learning_rate": 5.564444444444444e-07,
+      "loss": 0.2729,
       "step": 2500
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 5.506666666666666e-07,
-      "loss": 0.3357,
+      "epoch": 1.22,
+      "learning_rate": 5.508888888888889e-07,
+      "loss": 0.2822,
       "step": 2525
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 5.451111111111111e-07,
-      "loss": 0.2738,
+      "epoch": 1.23,
+      "learning_rate": 5.453333333333333e-07,
+      "loss": 0.2689,
       "step": 2550
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 5.395555555555556e-07,
-      "loss": 0.2775,
+      "epoch": 1.25,
+      "learning_rate": 5.397777777777778e-07,
+      "loss": 0.3401,
       "step": 2575
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 5.34e-07,
-      "loss": 0.2357,
+      "epoch": 1.26,
+      "learning_rate": 5.342222222222222e-07,
+      "loss": 0.3384,
       "step": 2600
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 5.284444444444445e-07,
-      "loss": 0.2566,
+      "epoch": 1.27,
+      "learning_rate": 5.286666666666666e-07,
+      "loss": 0.313,
       "step": 2625
     },
     {
-      "epoch": 2.1,
-      "learning_rate": 5.228888888888888e-07,
-      "loss": 0.2582,
+      "epoch": 1.28,
+      "learning_rate": 5.23111111111111e-07,
+      "loss": 0.2596,
       "step": 2650
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 5.173333333333333e-07,
-      "loss": 0.2418,
+      "epoch": 1.29,
+      "learning_rate": 5.175555555555555e-07,
+      "loss": 0.3083,
       "step": 2675
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 5.117777777777777e-07,
-      "loss": 0.2724,
+      "epoch": 1.31,
+      "learning_rate": 5.12e-07,
+      "loss": 0.3538,
       "step": 2700
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 5.062222222222222e-07,
-      "loss": 0.2981,
+      "epoch": 1.32,
+      "learning_rate": 5.064444444444444e-07,
+      "loss": 0.2759,
       "step": 2725
     },
     {
-      "epoch": 2.18,
-      "learning_rate": 5.006666666666667e-07,
-      "loss": 0.2699,
+      "epoch": 1.33,
+      "learning_rate": 5.008888888888889e-07,
+      "loss": 0.2495,
       "step": 2750
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 4.951111111111111e-07,
-      "loss": 0.2552,
+      "epoch": 1.34,
+      "learning_rate": 4.953333333333333e-07,
+      "loss": 0.3429,
       "step": 2775
     },
     {
-      "epoch": 2.22,
-      "learning_rate": 4.895555555555555e-07,
-      "loss": 0.2797,
+      "epoch": 1.36,
+      "learning_rate": 4.897777777777778e-07,
+      "loss": 0.2638,
       "step": 2800
     },
     {
-      "epoch": 2.24,
-      "learning_rate": 4.839999999999999e-07,
-      "loss": 0.2609,
+      "epoch": 1.37,
+      "learning_rate": 4.842222222222222e-07,
+      "loss": 0.3351,
       "step": 2825
     },
     {
-      "epoch": 2.26,
-      "learning_rate": 4.784444444444444e-07,
-      "loss": 0.2689,
+      "epoch": 1.38,
+      "learning_rate": 4.786666666666667e-07,
+      "loss": 0.3105,
       "step": 2850
     },
     {
-      "epoch": 2.28,
-      "learning_rate": 4.728888888888889e-07,
-      "loss": 0.2731,
+      "epoch": 1.39,
+      "learning_rate": 4.7311111111111107e-07,
+      "loss": 0.2635,
       "step": 2875
     },
     {
-      "epoch": 2.3,
-      "learning_rate": 4.673333333333333e-07,
-      "loss": 0.2764,
+      "epoch": 1.4,
+      "learning_rate": 4.675555555555555e-07,
+      "loss": 0.321,
       "step": 2900
     },
     {
-      "epoch": 2.32,
-      "learning_rate": 4.6177777777777777e-07,
-      "loss": 0.2539,
+      "epoch": 1.42,
+      "learning_rate": 4.62e-07,
+      "loss": 0.2564,
       "step": 2925
     },
     {
-      "epoch": 2.34,
-      "learning_rate": 4.5622222222222217e-07,
-      "loss": 0.2972,
+      "epoch": 1.43,
+      "learning_rate": 4.564444444444444e-07,
+      "loss": 0.2625,
       "step": 2950
     },
     {
-      "epoch": 2.36,
-      "learning_rate": 4.506666666666666e-07,
-      "loss": 0.2594,
+      "epoch": 1.44,
+      "learning_rate": 4.508888888888889e-07,
+      "loss": 0.2598,
       "step": 2975
     },
     {
-      "epoch": 2.38,
-      "learning_rate": 4.451111111111111e-07,
-      "loss": 0.2717,
+      "epoch": 1.45,
+      "learning_rate": 4.4533333333333335e-07,
+      "loss": 0.28,
       "step": 3000
     },
     {
-      "epoch": 2.38,
-      "eval_loss": 0.24675099551677704,
-      "eval_runtime": 4884.1957,
-      "eval_samples_per_second": 1.033,
-      "eval_steps_per_second": 0.516,
-      "eval_wer": 10.519201888006865,
+      "epoch": 1.45,
+      "eval_loss": 0.271098792552948,
+      "eval_runtime": 857.8316,
+      "eval_samples_per_second": 4.222,
+      "eval_steps_per_second": 2.111,
+      "eval_wer": 14.994248907292384,
       "step": 3000
     },
     {
-      "epoch": 2.4,
-      "learning_rate": 4.3955555555555554e-07,
-      "loss": 0.2455,
+      "epoch": 1.46,
+      "learning_rate": 4.3977777777777775e-07,
+      "loss": 0.2818,
       "step": 3025
     },
     {
-      "epoch": 2.42,
-      "learning_rate": 4.34e-07,
-      "loss": 0.2464,
+      "epoch": 1.48,
+      "learning_rate": 4.342222222222222e-07,
+      "loss": 0.2364,
       "step": 3050
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 4.2844444444444445e-07,
-      "loss": 0.287,
+      "epoch": 1.49,
+      "learning_rate": 4.286666666666666e-07,
+      "loss": 0.2781,
       "step": 3075
     },
     {
-      "epoch": 2.46,
-      "learning_rate": 4.2288888888888886e-07,
-      "loss": 0.2777,
+      "epoch": 1.5,
+      "learning_rate": 4.2311111111111107e-07,
+      "loss": 0.2658,
       "step": 3100
     },
     {
-      "epoch": 2.48,
-      "learning_rate": 4.173333333333333e-07,
-      "loss": 0.2867,
+      "epoch": 1.51,
+      "learning_rate": 4.1755555555555553e-07,
+      "loss": 0.3219,
       "step": 3125
     },
     {
-      "epoch": 2.5,
-      "learning_rate": 4.1177777777777777e-07,
-      "loss": 0.252,
+      "epoch": 1.52,
+      "learning_rate": 4.12e-07,
+      "loss": 0.3786,
       "step": 3150
     },
     {
-      "epoch": 2.52,
-      "learning_rate": 4.0622222222222217e-07,
-      "loss": 0.2573,
+      "epoch": 1.54,
+      "learning_rate": 4.0644444444444444e-07,
+      "loss": 0.3083,
       "step": 3175
     },
     {
-      "epoch": 2.54,
-      "learning_rate": 4.0066666666666663e-07,
-      "loss": 0.2696,
+      "epoch": 1.55,
+      "learning_rate": 4.008888888888889e-07,
+      "loss": 0.2473,
       "step": 3200
     },
     {
-      "epoch": 2.56,
-      "learning_rate": 3.9511111111111114e-07,
-      "loss": 0.287,
+      "epoch": 1.56,
+      "learning_rate": 3.953333333333333e-07,
+      "loss": 0.2413,
       "step": 3225
     },
     {
-      "epoch": 2.58,
-      "learning_rate": 3.8955555555555554e-07,
-      "loss": 0.2637,
+      "epoch": 1.57,
+      "learning_rate": 3.8977777777777776e-07,
+      "loss": 0.3109,
       "step": 3250
     },
     {
-      "epoch": 2.6,
-      "learning_rate": 3.84e-07,
-      "loss": 0.2673,
+      "epoch": 1.59,
+      "learning_rate": 3.842222222222222e-07,
+      "loss": 0.3019,
       "step": 3275
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 3.7844444444444445e-07,
-      "loss": 0.232,
+      "epoch": 1.6,
+      "learning_rate": 3.786666666666666e-07,
+      "loss": 0.2829,
       "step": 3300
     },
     {
-      "epoch": 2.64,
-      "learning_rate": 3.7288888888888886e-07,
-      "loss": 0.2671,
+      "epoch": 1.61,
+      "learning_rate": 3.7311111111111107e-07,
+      "loss": 0.2777,
       "step": 3325
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 3.673333333333333e-07,
-      "loss": 0.2436,
+      "epoch": 1.62,
+      "learning_rate": 3.675555555555556e-07,
+      "loss": 0.2883,
       "step": 3350
     },
     {
-      "epoch": 2.68,
-      "learning_rate": 3.617777777777777e-07,
-      "loss": 0.2869,
+      "epoch": 1.63,
+      "learning_rate": 3.62e-07,
+      "loss": 0.2505,
       "step": 3375
     },
     {
-      "epoch": 2.7,
-      "learning_rate": 3.5622222222222223e-07,
-      "loss": 0.2166,
+      "epoch": 1.65,
+      "learning_rate": 3.5644444444444444e-07,
+      "loss": 0.2852,
       "step": 3400
     },
     {
-      "epoch": 2.72,
-      "learning_rate": 3.506666666666667e-07,
-      "loss": 0.2362,
+      "epoch": 1.66,
+      "learning_rate": 3.5088888888888885e-07,
+      "loss": 0.2943,
       "step": 3425
     },
     {
-      "epoch": 2.74,
-      "learning_rate": 3.451111111111111e-07,
-      "loss": 0.2947,
+      "epoch": 1.67,
+      "learning_rate": 3.453333333333333e-07,
+      "loss": 0.2918,
       "step": 3450
     },
     {
-      "epoch": 2.76,
-      "learning_rate": 3.3955555555555554e-07,
-      "loss": 0.2615,
+      "epoch": 1.68,
+      "learning_rate": 3.3977777777777776e-07,
+      "loss": 0.2685,
       "step": 3475
     },
     {
-      "epoch": 2.78,
-      "learning_rate": 3.34e-07,
-      "loss": 0.247,
+      "epoch": 1.69,
+      "learning_rate": 3.3422222222222216e-07,
+      "loss": 0.2846,
       "step": 3500
     },
     {
-      "epoch": 2.8,
-      "learning_rate": 3.284444444444444e-07,
-      "loss": 0.2134,
+      "epoch": 1.71,
+      "learning_rate": 3.2866666666666667e-07,
+      "loss": 0.2841,
       "step": 3525
     },
     {
-      "epoch": 2.82,
-      "learning_rate": 3.2288888888888886e-07,
-      "loss": 0.2627,
+      "epoch": 1.72,
+      "learning_rate": 3.2311111111111113e-07,
+      "loss": 0.305,
       "step": 3550
     },
     {
-      "epoch": 2.84,
-      "learning_rate": 3.173333333333333e-07,
-      "loss": 0.2419,
+      "epoch": 1.73,
+      "learning_rate": 3.1755555555555553e-07,
+      "loss": 0.2545,
       "step": 3575
     },
     {
-      "epoch": 2.85,
-      "learning_rate": 3.1177777777777777e-07,
-      "loss": 0.256,
+      "epoch": 1.74,
+      "learning_rate": 3.12e-07,
+      "loss": 0.2947,
       "step": 3600
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 3.0622222222222223e-07,
-      "loss": 0.25,
+      "epoch": 1.75,
+      "learning_rate": 3.0644444444444444e-07,
+      "loss": 0.2705,
       "step": 3625
     },
     {
-      "epoch": 2.89,
-      "learning_rate": 3.006666666666667e-07,
-      "loss": 0.2508,
+      "epoch": 1.77,
+      "learning_rate": 3.0088888888888885e-07,
+      "loss": 0.2912,
       "step": 3650
     },
     {
-      "epoch": 2.91,
-      "learning_rate": 2.951111111111111e-07,
-      "loss": 0.2479,
+      "epoch": 1.78,
+      "learning_rate": 2.953333333333333e-07,
+      "loss": 0.2586,
       "step": 3675
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 2.8955555555555555e-07,
-      "loss": 0.2689,
+      "epoch": 1.79,
+      "learning_rate": 2.897777777777778e-07,
+      "loss": 0.2673,
       "step": 3700
     },
     {
-      "epoch": 2.95,
-      "learning_rate": 2.8399999999999995e-07,
-      "loss": 0.25,
+      "epoch": 1.8,
+      "learning_rate": 2.842222222222222e-07,
+      "loss": 0.2871,
       "step": 3725
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 2.784444444444444e-07,
-      "loss": 0.2079,
+      "epoch": 1.82,
+      "learning_rate": 2.786666666666667e-07,
+      "loss": 0.2651,
       "step": 3750
     },
     {
-      "epoch": 2.99,
-      "learning_rate": 2.728888888888889e-07,
-      "loss": 0.2574,
+      "epoch": 1.83,
+      "learning_rate": 2.7311111111111113e-07,
+      "loss": 0.3039,
       "step": 3775
     },
     {
-      "epoch": 3.01,
-      "learning_rate": 2.673333333333333e-07,
-      "loss": 0.2398,
+      "epoch": 1.84,
+      "learning_rate": 2.6755555555555553e-07,
+      "loss": 0.2801,
       "step": 3800
     },
     {
-      "epoch": 3.03,
-      "learning_rate": 2.617777777777778e-07,
-      "loss": 0.2356,
+      "epoch": 1.85,
+      "learning_rate": 2.62e-07,
+      "loss": 0.2974,
       "step": 3825
     },
     {
-      "epoch": 3.05,
-      "learning_rate": 2.5622222222222223e-07,
-      "loss": 0.2483,
+      "epoch": 1.86,
+      "learning_rate": 2.564444444444444e-07,
+      "loss": 0.2564,
       "step": 3850
     },
     {
-      "epoch": 3.07,
-      "learning_rate": 2.5066666666666663e-07,
-      "loss": 0.1985,
+      "epoch": 1.88,
+      "learning_rate": 2.5088888888888885e-07,
+      "loss": 0.3262,
       "step": 3875
     },
     {
-      "epoch": 3.09,
-      "learning_rate": 2.451111111111111e-07,
-      "loss": 0.265,
+      "epoch": 1.89,
+      "learning_rate": 2.453333333333333e-07,
+      "loss": 0.2595,
       "step": 3900
     },
     {
-      "epoch": 3.11,
-      "learning_rate": 2.3955555555555555e-07,
-      "loss": 0.185,
+      "epoch": 1.9,
+      "learning_rate": 2.3977777777777776e-07,
+      "loss": 0.248,
       "step": 3925
     },
     {
-      "epoch": 3.13,
-      "learning_rate": 2.34e-07,
-      "loss": 0.1968,
+      "epoch": 1.91,
+      "learning_rate": 2.3422222222222222e-07,
+      "loss": 0.3078,
       "step": 3950
     },
     {
-      "epoch": 3.15,
-      "learning_rate": 2.2844444444444443e-07,
-      "loss": 0.2289,
+      "epoch": 1.92,
+      "learning_rate": 2.2866666666666665e-07,
+      "loss": 0.2846,
       "step": 3975
     },
     {
-      "epoch": 3.17,
-      "learning_rate": 2.2288888888888886e-07,
-      "loss": 0.2534,
+      "epoch": 1.94,
+      "learning_rate": 2.231111111111111e-07,
+      "loss": 0.3505,
       "step": 4000
     },
     {
-      "epoch": 3.17,
-      "eval_loss": 0.22480489313602448,
-      "eval_runtime": 4873.3769,
-      "eval_samples_per_second": 1.035,
-      "eval_steps_per_second": 0.518,
-      "eval_wer": 9.500107273117358,
+      "epoch": 1.94,
+      "eval_loss": 0.26386478543281555,
+      "eval_runtime": 852.2076,
+      "eval_samples_per_second": 4.25,
+      "eval_steps_per_second": 2.125,
+      "eval_wer": 14.566367609845871,
       "step": 4000
     },
     {
-      "epoch": 3.19,
+      "epoch": 1.95,
       "learning_rate": 2.1755555555555554e-07,
-      "loss": 0.2666,
+      "loss": 0.2606,
       "step": 4025
     },
     {
-      "epoch": 3.21,
+      "epoch": 1.96,
       "learning_rate": 2.12e-07,
-      "loss": 0.2444,
+      "loss": 0.2459,
       "step": 4050
     },
     {
-      "epoch": 3.23,
+      "epoch": 1.97,
       "learning_rate": 2.0644444444444445e-07,
-      "loss": 0.2116,
+      "loss": 0.3116,
       "step": 4075
     },
     {
-      "epoch": 3.25,
-      "learning_rate": 2.0088888888888888e-07,
-      "loss": 0.2535,
+      "epoch": 1.98,
+      "learning_rate": 2.011111111111111e-07,
+      "loss": 0.302,
       "step": 4100
     },
     {
-      "epoch": 3.27,
-      "learning_rate": 1.953333333333333e-07,
-      "loss": 0.2033,
+      "epoch": 2.0,
+      "learning_rate": 1.9555555555555555e-07,
+      "loss": 0.292,
       "step": 4125
     },
     {
-      "epoch": 3.29,
-      "learning_rate": 1.8977777777777777e-07,
-      "loss": 0.2423,
+      "epoch": 2.01,
+      "learning_rate": 1.8999999999999998e-07,
+      "loss": 0.3034,
       "step": 4150
     },
     {
-      "epoch": 3.31,
-      "learning_rate": 1.8422222222222222e-07,
-      "loss": 0.2163,
+      "epoch": 2.02,
+      "learning_rate": 1.8444444444444444e-07,
+      "loss": 0.262,
       "step": 4175
     },
     {
-      "epoch": 3.33,
-      "learning_rate": 1.7866666666666665e-07,
-      "loss": 0.2409,
+      "epoch": 2.03,
+      "learning_rate": 1.7888888888888887e-07,
+      "loss": 0.2639,
       "step": 4200
     },
     {
-      "epoch": 3.35,
-      "learning_rate": 1.731111111111111e-07,
-      "loss": 0.2406,
+      "epoch": 2.05,
+      "learning_rate": 1.7333333333333332e-07,
+      "loss": 0.264,
       "step": 4225
     },
     {
-      "epoch": 3.37,
-      "learning_rate": 1.6755555555555556e-07,
-      "loss": 0.2297,
+      "epoch": 2.06,
+      "learning_rate": 1.6777777777777778e-07,
+      "loss": 0.224,
       "step": 4250
     },
     {
-      "epoch": 3.39,
-      "learning_rate": 1.62e-07,
-      "loss": 0.2496,
+      "epoch": 2.07,
+      "learning_rate": 1.622222222222222e-07,
+      "loss": 0.2332,
       "step": 4275
     },
     {
-      "epoch": 3.41,
-      "learning_rate": 1.5644444444444442e-07,
-      "loss": 0.1907,
+      "epoch": 2.08,
+      "learning_rate": 1.5666666666666667e-07,
+      "loss": 0.2242,
       "step": 4300
     },
     {
-      "epoch": 3.43,
-      "learning_rate": 1.5088888888888888e-07,
-      "loss": 0.2242,
+      "epoch": 2.09,
+      "learning_rate": 1.511111111111111e-07,
+      "loss": 0.1942,
       "step": 4325
     },
     {
-      "epoch": 3.45,
-      "learning_rate": 1.4533333333333334e-07,
-      "loss": 0.213,
+      "epoch": 2.11,
+      "learning_rate": 1.4555555555555555e-07,
+      "loss": 0.2162,
       "step": 4350
     },
     {
-      "epoch": 3.47,
-      "learning_rate": 1.3977777777777777e-07,
-      "loss": 0.257,
+      "epoch": 2.12,
+      "learning_rate": 1.4e-07,
+      "loss": 0.2474,
       "step": 4375
     },
     {
-      "epoch": 3.49,
-      "learning_rate": 1.342222222222222e-07,
-      "loss": 0.2563,
+      "epoch": 2.13,
+      "learning_rate": 1.3444444444444444e-07,
+      "loss": 0.2182,
       "step": 4400
     },
     {
-      "epoch": 3.51,
-      "learning_rate": 1.2866666666666668e-07,
-      "loss": 0.2258,
+      "epoch": 2.14,
+      "learning_rate": 1.2888888888888887e-07,
+      "loss": 0.2821,
       "step": 4425
     },
     {
-      "epoch": 3.53,
-      "learning_rate": 1.231111111111111e-07,
-      "loss": 0.2249,
+      "epoch": 2.15,
+      "learning_rate": 1.2333333333333333e-07,
+      "loss": 0.3,
       "step": 4450
     },
     {
-      "epoch": 3.55,
-      "learning_rate": 1.1755555555555554e-07,
-      "loss": 0.2147,
+      "epoch": 2.17,
+      "learning_rate": 1.1777777777777777e-07,
+      "loss": 0.2229,
       "step": 4475
     },
     {
-      "epoch": 3.57,
-      "learning_rate": 1.12e-07,
-      "loss": 0.2242,
+      "epoch": 2.18,
+      "learning_rate": 1.1222222222222221e-07,
+      "loss": 0.2092,
       "step": 4500
     },
     {
-      "epoch": 3.59,
-      "learning_rate": 1.0644444444444444e-07,
-      "loss": 0.2095,
+      "epoch": 2.19,
+      "learning_rate": 1.0666666666666667e-07,
+      "loss": 0.2451,
       "step": 4525
     },
     {
-      "epoch": 3.61,
-      "learning_rate": 1.0088888888888888e-07,
-      "loss": 0.2454,
+      "epoch": 2.2,
+      "learning_rate": 1.011111111111111e-07,
+      "loss": 0.2575,
       "step": 4550
     },
     {
-      "epoch": 3.63,
-      "learning_rate": 9.533333333333334e-08,
-      "loss": 0.202,
+      "epoch": 2.21,
+      "learning_rate": 9.555555555555556e-08,
+      "loss": 0.2143,
       "step": 4575
     },
     {
-      "epoch": 3.65,
-      "learning_rate": 8.977777777777777e-08,
-      "loss": 0.1916,
+      "epoch": 2.23,
+      "learning_rate": 9e-08,
+      "loss": 0.2353,
       "step": 4600
     },
     {
-      "epoch": 3.67,
-      "learning_rate": 8.422222222222223e-08,
-      "loss": 0.259,
+      "epoch": 2.24,
+      "learning_rate": 8.444444444444444e-08,
+      "loss": 0.2568,
       "step": 4625
     },
     {
-      "epoch": 3.69,
-      "learning_rate": 7.866666666666666e-08,
-      "loss": 0.2247,
+      "epoch": 2.25,
+      "learning_rate": 7.888888888888889e-08,
+      "loss": 0.2388,
       "step": 4650
     },
     {
-      "epoch": 3.71,
-      "learning_rate": 7.311111111111111e-08,
-      "loss": 0.2259,
+      "epoch": 2.26,
+      "learning_rate": 7.333333333333333e-08,
+      "loss": 0.2883,
       "step": 4675
     },
     {
-      "epoch": 3.73,
-      "learning_rate": 6.755555555555554e-08,
-      "loss": 0.2357,
+      "epoch": 2.27,
+      "learning_rate": 6.777777777777778e-08,
+      "loss": 0.2509,
       "step": 4700
     },
     {
-      "epoch": 3.75,
-      "learning_rate": 6.2e-08,
-      "loss": 0.2094,
+      "epoch": 2.29,
+      "learning_rate": 6.222222222222221e-08,
+      "loss": 0.2651,
       "step": 4725
     },
     {
-      "epoch": 3.77,
-      "learning_rate": 5.644444444444444e-08,
-      "loss": 0.2214,
+      "epoch": 2.3,
+      "learning_rate": 5.666666666666666e-08,
+      "loss": 0.3152,
       "step": 4750
     },
     {
-      "epoch": 3.79,
-      "learning_rate": 5.0888888888888886e-08,
-      "loss": 0.2127,
+      "epoch": 2.31,
+      "learning_rate": 5.1111111111111114e-08,
+      "loss": 0.2032,
       "step": 4775
     },
     {
-      "epoch": 3.81,
-      "learning_rate": 4.5333333333333336e-08,
-      "loss": 0.252,
+      "epoch": 2.32,
+      "learning_rate": 4.555555555555556e-08,
+      "loss": 0.2442,
       "step": 4800
     },
     {
-      "epoch": 3.83,
-      "learning_rate": 3.977777777777778e-08,
-      "loss": 0.2356,
+      "epoch": 2.34,
+      "learning_rate": 4e-08,
+      "loss": 0.2139,
       "step": 4825
     },
     {
-      "epoch": 3.85,
-      "learning_rate": 3.422222222222222e-08,
-      "loss": 0.2828,
+      "epoch": 2.35,
+      "learning_rate": 3.4444444444444444e-08,
+      "loss": 0.2782,
       "step": 4850
     },
     {
-      "epoch": 3.87,
-      "learning_rate": 2.8666666666666665e-08,
-      "loss": 0.2111,
+      "epoch": 2.36,
+      "learning_rate": 2.8888888888888887e-08,
+      "loss": 0.2401,
       "step": 4875
     },
     {
-      "epoch": 3.89,
-      "learning_rate": 2.311111111111111e-08,
-      "loss": 0.2316,
+      "epoch": 2.37,
+      "learning_rate": 2.3333333333333334e-08,
+      "loss": 0.2247,
       "step": 4900
     },
     {
-      "epoch": 3.91,
-      "learning_rate": 1.7555555555555555e-08,
-      "loss": 0.2319,
+      "epoch": 2.38,
+      "learning_rate": 1.7777777777777777e-08,
+      "loss": 0.2341,
       "step": 4925
     },
     {
-      "epoch": 3.93,
-      "learning_rate": 1.2e-08,
-      "loss": 0.2153,
+      "epoch": 2.4,
+      "learning_rate": 1.2222222222222222e-08,
+      "loss": 0.2994,
       "step": 4950
     },
     {
-      "epoch": 3.95,
-      "learning_rate": 6.444444444444444e-09,
-      "loss": 0.1792,
+      "epoch": 2.41,
+      "learning_rate": 6.666666666666667e-09,
+      "loss": 0.288,
       "step": 4975
     },
     {
-      "epoch": 3.97,
-      "learning_rate": 8.888888888888889e-10,
-      "loss": 0.2226,
+      "epoch": 2.42,
+      "learning_rate": 1.111111111111111e-09,
+      "loss": 0.2412,
       "step": 5000
     },
     {
-      "epoch": 3.97,
-      "eval_loss": 0.21749083697795868,
-      "eval_runtime": 4878.0654,
-      "eval_samples_per_second": 1.034,
-      "eval_steps_per_second": 0.517,
-      "eval_wer": 9.261960952585282,
+      "epoch": 2.42,
+      "eval_loss": 0.2629786729812622,
+      "eval_runtime": 844.6741,
+      "eval_samples_per_second": 4.288,
+      "eval_steps_per_second": 2.144,
+      "eval_wer": 14.570968484011962,
       "step": 5000
     },
     {
-      "epoch": 3.97,
+      "epoch": 2.42,
       "step": 5000,
-      "total_flos": 5.77084225019904e+18,
-      "train_loss": 0.3629964433670044,
-      "train_runtime": 41790.9887,
-      "train_samples_per_second": 0.479,
-      "train_steps_per_second": 0.12
+      "total_flos": 5.76997649399808e+18,
+      "train_loss": 0.37738304929733274,
+      "train_runtime": 7350.5739,
+      "train_samples_per_second": 2.721,
+      "train_steps_per_second": 0.68
     }
   ],
   "max_steps": 5000,
-  "num_train_epochs": 4,
-  "total_flos": 5.77084225019904e+18,
+  "num_train_epochs": 3,
+  "total_flos": 5.76997649399808e+18,
   "trial_name": null,
   "trial_params": null
 }