End of training

all_results.json

@@ -1,12 +1,12 @@
 {
-    "epoch": 0.83,
-    "eval_loss": 0.20626504719257355,
-    "eval_runtime": 171.3898,
-    "eval_samples_per_second": 9.866,
-    "eval_steps_per_second": 0.309,
-    "eval_wer": 10.759335581366305,
-    "train_loss": 0.24075475883483888,
-    "train_runtime": 9843.0142,
-    "train_samples_per_second": 32.51,
-    "train_steps_per_second": 0.508
+    "epoch": 1.65,
+    "eval_loss": 0.19057515263557434,
+    "eval_runtime": 170.5616,
+    "eval_samples_per_second": 9.914,
+    "eval_steps_per_second": 0.311,
+    "eval_wer": 9.54165105532659,
+    "train_loss": 0.20601170344352723,
+    "train_runtime": 18822.9568,
+    "train_samples_per_second": 34.001,
+    "train_steps_per_second": 0.266
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 0.83,
-    "eval_loss": 0.20626504719257355,
-    "eval_runtime": 171.3898,
-    "eval_samples_per_second": 9.866,
-    "eval_steps_per_second": 0.309,
-    "eval_wer": 10.759335581366305
+    "epoch": 1.65,
+    "eval_loss": 0.19057515263557434,
+    "eval_runtime": 170.5616,
+    "eval_samples_per_second": 9.914,
+    "eval_steps_per_second": 0.311,
+    "eval_wer": 9.54165105532659
 }

train_results.json

@@ -1,7 +1,7 @@
 {
-    "epoch": 0.83,
-    "train_loss": 0.24075475883483888,
-    "train_runtime": 9843.0142,
-    "train_samples_per_second": 32.51,
-    "train_steps_per_second": 0.508
+    "epoch": 1.65,
+    "train_loss": 0.20601170344352723,
+    "train_runtime": 18822.9568,
+    "train_samples_per_second": 34.001,
+    "train_steps_per_second": 0.266
 }

trainer_state.json

@@ -1,1270 +1,1270 @@
 {
-  "best_metric": 10.759335581366305,
+  "best_metric": 9.54165105532659,
   "best_model_checkpoint": "/scratch/andoni.sudupe/checkpoints/whisper-small-basque/checkpoint-5000",
-  "epoch": 0.8259002312520648,
+  "epoch": 1.6518004625041294,
   "global_step": 5000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.0,
+      "epoch": 0.01,
       "learning_rate": 4.0000000000000003e-07,
-      "loss": 6.364,
+      "loss": 6.4054,
       "step": 25
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 9.000000000000001e-07,
-      "loss": 4.4115,
+      "loss": 4.2878,
       "step": 50
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.02,
       "learning_rate": 1.4000000000000001e-06,
-      "loss": 2.5368,
+      "loss": 2.4134,
       "step": 75
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.03,
       "learning_rate": 1.9000000000000002e-06,
-      "loss": 1.4237,
+      "loss": 1.331,
       "step": 100
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.04,
       "learning_rate": 2.4000000000000003e-06,
-      "loss": 1.0001,
+      "loss": 0.9224,
       "step": 125
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.05,
       "learning_rate": 2.9e-06,
-      "loss": 0.8307,
+      "loss": 0.7595,
       "step": 150
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.06,
       "learning_rate": 3.4000000000000005e-06,
-      "loss": 0.7202,
+      "loss": 0.6367,
       "step": 175
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.07,
       "learning_rate": 3.900000000000001e-06,
-      "loss": 0.6401,
+      "loss": 0.5795,
       "step": 200
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.07,
       "learning_rate": 4.4e-06,
-      "loss": 0.5868,
+      "loss": 0.5178,
       "step": 225
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.08,
       "learning_rate": 4.9000000000000005e-06,
-      "loss": 0.544,
+      "loss": 0.4734,
       "step": 250
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.09,
       "learning_rate": 5.400000000000001e-06,
-      "loss": 0.5086,
+      "loss": 0.434,
       "step": 275
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.1,
       "learning_rate": 5.9e-06,
-      "loss": 0.4674,
+      "loss": 0.4041,
       "step": 300
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.11,
       "learning_rate": 6.4000000000000006e-06,
-      "loss": 0.4273,
+      "loss": 0.3833,
       "step": 325
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 6.9e-06,
-      "loss": 0.4104,
+      "loss": 0.3556,
       "step": 350
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.12,
       "learning_rate": 7.4e-06,
-      "loss": 0.4043,
+      "loss": 0.3407,
       "step": 375
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.13,
       "learning_rate": 7.9e-06,
-      "loss": 0.3698,
+      "loss": 0.3233,
       "step": 400
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.14,
       "learning_rate": 8.400000000000001e-06,
-      "loss": 0.3599,
+      "loss": 0.3117,
       "step": 425
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.15,
       "learning_rate": 8.900000000000001e-06,
-      "loss": 0.3395,
+      "loss": 0.2857,
       "step": 450
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.16,
       "learning_rate": 9.4e-06,
-      "loss": 0.3311,
+      "loss": 0.283,
       "step": 475
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.17,
       "learning_rate": 9.9e-06,
-      "loss": 0.3123,
+      "loss": 0.2684,
       "step": 500
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.17,
       "learning_rate": 9.955555555555556e-06,
-      "loss": 0.3002,
+      "loss": 0.2553,
       "step": 525
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.18,
       "learning_rate": 9.9e-06,
-      "loss": 0.2907,
+      "loss": 0.2472,
       "step": 550
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.19,
       "learning_rate": 9.844444444444446e-06,
-      "loss": 0.285,
+      "loss": 0.2314,
       "step": 575
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.2,
       "learning_rate": 9.78888888888889e-06,
-      "loss": 0.2686,
+      "loss": 0.2284,
       "step": 600
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.21,
       "learning_rate": 9.733333333333334e-06,
-      "loss": 0.2659,
+      "loss": 0.2232,
       "step": 625
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.21,
       "learning_rate": 9.677777777777778e-06,
-      "loss": 0.2649,
+      "loss": 0.2169,
       "step": 650
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.22,
       "learning_rate": 9.622222222222222e-06,
-      "loss": 0.2534,
+      "loss": 0.2147,
       "step": 675
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.23,
       "learning_rate": 9.566666666666668e-06,
-      "loss": 0.2473,
+      "loss": 0.2054,
       "step": 700
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.24,
       "learning_rate": 9.511111111111112e-06,
-      "loss": 0.2512,
+      "loss": 0.2025,
       "step": 725
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.25,
       "learning_rate": 9.455555555555557e-06,
-      "loss": 0.2329,
+      "loss": 0.1952,
       "step": 750
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.26,
       "learning_rate": 9.4e-06,
-      "loss": 0.235,
+      "loss": 0.1969,
       "step": 775
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.26,
       "learning_rate": 9.344444444444446e-06,
-      "loss": 0.2283,
+      "loss": 0.1913,
       "step": 800
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.27,
       "learning_rate": 9.28888888888889e-06,
-      "loss": 0.2255,
+      "loss": 0.1874,
       "step": 825
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.28,
       "learning_rate": 9.233333333333334e-06,
-      "loss": 0.2274,
+      "loss": 0.1859,
       "step": 850
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.29,
       "learning_rate": 9.17777777777778e-06,
-      "loss": 0.2097,
+      "loss": 0.1753,
       "step": 875
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.3,
       "learning_rate": 9.122222222222223e-06,
-      "loss": 0.2065,
+      "loss": 0.1799,
       "step": 900
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.31,
       "learning_rate": 9.066666666666667e-06,
-      "loss": 0.2134,
+      "loss": 0.1779,
       "step": 925
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.31,
       "learning_rate": 9.011111111111111e-06,
-      "loss": 0.2038,
+      "loss": 0.1645,
       "step": 950
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.32,
       "learning_rate": 8.955555555555555e-06,
-      "loss": 0.1964,
+      "loss": 0.1673,
       "step": 975
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.33,
       "learning_rate": 8.900000000000001e-06,
-      "loss": 0.2059,
+      "loss": 0.17,
       "step": 1000
     },
     {
-      "epoch": 0.17,
-      "eval_loss": 0.30938294529914856,
-      "eval_runtime": 173.774,
-      "eval_samples_per_second": 9.731,
-      "eval_steps_per_second": 0.305,
-      "eval_wer": 17.659547895591356,
+      "epoch": 0.33,
+      "eval_loss": 0.2697364389896393,
+      "eval_runtime": 186.0082,
+      "eval_samples_per_second": 9.091,
+      "eval_steps_per_second": 0.285,
+      "eval_wer": 15.66129636568003,
       "step": 1000
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.34,
       "learning_rate": 8.844444444444445e-06,
-      "loss": 0.1904,
+      "loss": 0.1604,
       "step": 1025
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.35,
       "learning_rate": 8.788888888888891e-06,
-      "loss": 0.1968,
+      "loss": 0.16,
       "step": 1050
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 8.733333333333333e-06,
-      "loss": 0.1882,
+      "loss": 0.1554,
       "step": 1075
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.36,
       "learning_rate": 8.677777777777779e-06,
-      "loss": 0.1866,
+      "loss": 0.1584,
       "step": 1100
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.37,
       "learning_rate": 8.622222222222223e-06,
-      "loss": 0.1838,
+      "loss": 0.1519,
       "step": 1125
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.38,
       "learning_rate": 8.566666666666667e-06,
-      "loss": 0.1766,
+      "loss": 0.1504,
       "step": 1150
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.39,
       "learning_rate": 8.511111111111113e-06,
-      "loss": 0.1825,
+      "loss": 0.154,
       "step": 1175
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 8.455555555555555e-06,
-      "loss": 0.1772,
+      "loss": 0.144,
       "step": 1200
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.4,
       "learning_rate": 8.400000000000001e-06,
-      "loss": 0.1793,
+      "loss": 0.1513,
       "step": 1225
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.41,
       "learning_rate": 8.344444444444445e-06,
-      "loss": 0.1749,
+      "loss": 0.1479,
       "step": 1250
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.42,
       "learning_rate": 8.288888888888889e-06,
-      "loss": 0.177,
+      "loss": 0.152,
       "step": 1275
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.43,
       "learning_rate": 8.233333333333335e-06,
-      "loss": 0.1703,
+      "loss": 0.146,
       "step": 1300
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.44,
       "learning_rate": 8.177777777777779e-06,
-      "loss": 0.1678,
+      "loss": 0.1376,
       "step": 1325
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.45,
       "learning_rate": 8.122222222222223e-06,
-      "loss": 0.1808,
+      "loss": 0.1394,
       "step": 1350
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.45,
       "learning_rate": 8.066666666666667e-06,
-      "loss": 0.1676,
+      "loss": 0.1374,
       "step": 1375
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.46,
       "learning_rate": 8.011111111111113e-06,
-      "loss": 0.1664,
+      "loss": 0.1379,
       "step": 1400
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.47,
       "learning_rate": 7.955555555555557e-06,
-      "loss": 0.1691,
+      "loss": 0.1384,
       "step": 1425
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.48,
       "learning_rate": 7.9e-06,
-      "loss": 0.1598,
+      "loss": 0.1397,
       "step": 1450
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.49,
       "learning_rate": 7.844444444444446e-06,
-      "loss": 0.1636,
+      "loss": 0.1331,
       "step": 1475
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 7.788888888888889e-06,
-      "loss": 0.159,
+      "loss": 0.1322,
       "step": 1500
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.5,
       "learning_rate": 7.733333333333334e-06,
-      "loss": 0.1683,
+      "loss": 0.132,
       "step": 1525
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.51,
       "learning_rate": 7.677777777777778e-06,
-      "loss": 0.158,
+      "loss": 0.1334,
       "step": 1550
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.52,
       "learning_rate": 7.622222222222223e-06,
-      "loss": 0.1622,
+      "loss": 0.1296,
       "step": 1575
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.53,
       "learning_rate": 7.566666666666667e-06,
-      "loss": 0.154,
+      "loss": 0.1291,
       "step": 1600
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.54,
       "learning_rate": 7.511111111111111e-06,
-      "loss": 0.1562,
+      "loss": 0.1318,
       "step": 1625
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.55,
       "learning_rate": 7.455555555555556e-06,
-      "loss": 0.1577,
+      "loss": 0.1249,
       "step": 1650
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.55,
       "learning_rate": 7.4e-06,
-      "loss": 0.1531,
+      "loss": 0.1264,
       "step": 1675
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.56,
       "learning_rate": 7.344444444444445e-06,
-      "loss": 0.1564,
+      "loss": 0.1271,
       "step": 1700
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.57,
       "learning_rate": 7.28888888888889e-06,
-      "loss": 0.1435,
+      "loss": 0.1216,
       "step": 1725
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.58,
       "learning_rate": 7.233333333333334e-06,
-      "loss": 0.1507,
+      "loss": 0.126,
       "step": 1750
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.59,
       "learning_rate": 7.177777777777778e-06,
-      "loss": 0.1522,
+      "loss": 0.123,
       "step": 1775
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.59,
       "learning_rate": 7.122222222222222e-06,
-      "loss": 0.1523,
+      "loss": 0.1233,
       "step": 1800
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.6,
       "learning_rate": 7.066666666666667e-06,
-      "loss": 0.1505,
+      "loss": 0.1234,
       "step": 1825
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.61,
       "learning_rate": 7.011111111111112e-06,
-      "loss": 0.1501,
+      "loss": 0.1195,
       "step": 1850
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.62,
       "learning_rate": 6.955555555555557e-06,
-      "loss": 0.1373,
+      "loss": 0.1192,
       "step": 1875
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.63,
       "learning_rate": 6.9e-06,
-      "loss": 0.1418,
+      "loss": 0.1195,
       "step": 1900
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 6.844444444444445e-06,
-      "loss": 0.1356,
+      "loss": 0.1183,
       "step": 1925
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.64,
       "learning_rate": 6.788888888888889e-06,
-      "loss": 0.1491,
+      "loss": 0.1228,
       "step": 1950
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.65,
       "learning_rate": 6.733333333333334e-06,
-      "loss": 0.145,
+      "loss": 0.1179,
       "step": 1975
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.66,
       "learning_rate": 6.677777777777779e-06,
-      "loss": 0.1468,
+      "loss": 0.1163,
       "step": 2000
     },
     {
-      "epoch": 0.33,
-      "eval_loss": 0.24528370797634125,
-      "eval_runtime": 171.4379,
-      "eval_samples_per_second": 9.864,
-      "eval_steps_per_second": 0.309,
-      "eval_wer": 13.126014737105033,
+      "epoch": 0.66,
+      "eval_loss": 0.21983104944229126,
+      "eval_runtime": 170.2191,
+      "eval_samples_per_second": 9.934,
+      "eval_steps_per_second": 0.311,
+      "eval_wer": 11.727238666167104,
       "step": 2000
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.67,
       "learning_rate": 6.6222222222222236e-06,
-      "loss": 0.1422,
+      "loss": 0.1165,
       "step": 2025
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.68,
       "learning_rate": 6.566666666666667e-06,
-      "loss": 0.1339,
+      "loss": 0.1156,
       "step": 2050
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.69,
       "learning_rate": 6.511111111111112e-06,
-      "loss": 0.1362,
+      "loss": 0.1118,
       "step": 2075
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.69,
       "learning_rate": 6.455555555555556e-06,
-      "loss": 0.1395,
+      "loss": 0.1202,
       "step": 2100
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.7,
       "learning_rate": 6.4000000000000006e-06,
-      "loss": 0.1318,
+      "loss": 0.1181,
       "step": 2125
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.71,
       "learning_rate": 6.3444444444444454e-06,
-      "loss": 0.1348,
+      "loss": 0.1148,
       "step": 2150
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.72,
       "learning_rate": 6.28888888888889e-06,
-      "loss": 0.132,
+      "loss": 0.1099,
       "step": 2175
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.73,
       "learning_rate": 6.2333333333333335e-06,
-      "loss": 0.1439,
+      "loss": 0.1056,
       "step": 2200
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 6.177777777777778e-06,
-      "loss": 0.1342,
+      "loss": 0.1092,
       "step": 2225
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.74,
       "learning_rate": 6.1222222222222224e-06,
-      "loss": 0.1277,
+      "loss": 0.1082,
       "step": 2250
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.75,
       "learning_rate": 6.066666666666667e-06,
-      "loss": 0.1333,
+      "loss": 0.1143,
       "step": 2275
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.76,
       "learning_rate": 6.011111111111112e-06,
-      "loss": 0.1268,
+      "loss": 0.1083,
       "step": 2300
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.77,
       "learning_rate": 5.955555555555555e-06,
-      "loss": 0.137,
+      "loss": 0.1115,
       "step": 2325
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 5.9e-06,
-      "loss": 0.1311,
+      "loss": 0.108,
       "step": 2350
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.78,
       "learning_rate": 5.844444444444445e-06,
-      "loss": 0.1286,
+      "loss": 0.111,
       "step": 2375
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.79,
       "learning_rate": 5.788888888888889e-06,
-      "loss": 0.1221,
+      "loss": 0.1075,
       "step": 2400
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.8,
       "learning_rate": 5.733333333333334e-06,
-      "loss": 0.1342,
+      "loss": 0.1096,
       "step": 2425
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.81,
       "learning_rate": 5.677777777777779e-06,
-      "loss": 0.1326,
+      "loss": 0.1049,
       "step": 2450
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.82,
       "learning_rate": 5.622222222222222e-06,
-      "loss": 0.1261,
+      "loss": 0.1074,
       "step": 2475
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.83,
       "learning_rate": 5.566666666666667e-06,
-      "loss": 0.1329,
+      "loss": 0.1054,
       "step": 2500
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.83,
       "learning_rate": 5.511111111111112e-06,
-      "loss": 0.134,
+      "loss": 0.1118,
       "step": 2525
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.84,
       "learning_rate": 5.455555555555556e-06,
-      "loss": 0.1335,
+      "loss": 0.107,
       "step": 2550
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.85,
       "learning_rate": 5.400000000000001e-06,
-      "loss": 0.1279,
+      "loss": 0.1077,
       "step": 2575
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.86,
       "learning_rate": 5.344444444444446e-06,
-      "loss": 0.1295,
+      "loss": 0.1071,
       "step": 2600
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.87,
       "learning_rate": 5.288888888888889e-06,
-      "loss": 0.1236,
+      "loss": 0.1043,
       "step": 2625
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 5.233333333333334e-06,
-      "loss": 0.1196,
+      "loss": 0.1051,
       "step": 2650
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.88,
       "learning_rate": 5.177777777777779e-06,
-      "loss": 0.1253,
+      "loss": 0.1056,
       "step": 2675
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.89,
       "learning_rate": 5.122222222222223e-06,
-      "loss": 0.1199,
+      "loss": 0.1007,
       "step": 2700
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.9,
       "learning_rate": 5.0666666666666676e-06,
-      "loss": 0.1255,
+      "loss": 0.1021,
       "step": 2725
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.91,
       "learning_rate": 5.011111111111111e-06,
-      "loss": 0.1186,
+      "loss": 0.1052,
       "step": 2750
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.92,
       "learning_rate": 4.9555555555555565e-06,
-      "loss": 0.1245,
+      "loss": 0.0989,
       "step": 2775
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.93,
       "learning_rate": 4.9000000000000005e-06,
-      "loss": 0.1195,
+      "loss": 0.098,
       "step": 2800
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.93,
       "learning_rate": 4.8444444444444446e-06,
-      "loss": 0.1211,
+      "loss": 0.102,
       "step": 2825
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.94,
       "learning_rate": 4.7888888888888894e-06,
-      "loss": 0.1237,
+      "loss": 0.0981,
       "step": 2850
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.95,
       "learning_rate": 4.7333333333333335e-06,
-      "loss": 0.1277,
+      "loss": 0.0991,
       "step": 2875
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.96,
       "learning_rate": 4.677777777777778e-06,
-      "loss": 0.121,
+      "loss": 0.0994,
       "step": 2900
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.97,
       "learning_rate": 4.622222222222222e-06,
-      "loss": 0.1159,
+      "loss": 0.0986,
       "step": 2925
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.97,
       "learning_rate": 4.566666666666667e-06,
-      "loss": 0.1208,
+      "loss": 0.0967,
       "step": 2950
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.98,
       "learning_rate": 4.511111111111111e-06,
-      "loss": 0.1169,
+      "loss": 0.098,
       "step": 2975
     },
     {
-      "epoch": 0.5,
+      "epoch": 0.99,
       "learning_rate": 4.455555555555555e-06,
-      "loss": 0.1177,
+      "loss": 0.0966,
       "step": 3000
     },
     {
-      "epoch": 0.5,
-      "eval_loss": 0.22596605122089386,
-      "eval_runtime": 171.349,
-      "eval_samples_per_second": 9.869,
-      "eval_steps_per_second": 0.309,
-      "eval_wer": 11.952041963282129,
+      "epoch": 0.99,
+      "eval_loss": 0.20093278586864471,
+      "eval_runtime": 170.4727,
+      "eval_samples_per_second": 9.919,
+      "eval_steps_per_second": 0.311,
+      "eval_wer": 10.278506306981392,
       "step": 3000
     },
     {
-      "epoch": 0.5,
+      "epoch": 1.0,
       "learning_rate": 4.4e-06,
-      "loss": 0.1156,
+      "loss": 0.1016,
       "step": 3025
     },
     {
-      "epoch": 0.5,
+      "epoch": 1.01,
       "learning_rate": 4.344444444444445e-06,
-      "loss": 0.1195,
+      "loss": 0.0722,
       "step": 3050
     },
     {
-      "epoch": 0.51,
+      "epoch": 1.02,
       "learning_rate": 4.288888888888889e-06,
-      "loss": 0.1204,
+      "loss": 0.0746,
       "step": 3075
     },
     {
-      "epoch": 0.51,
+      "epoch": 1.02,
       "learning_rate": 4.233333333333334e-06,
-      "loss": 0.1187,
+      "loss": 0.0764,
       "step": 3100
     },
     {
-      "epoch": 0.52,
+      "epoch": 1.03,
       "learning_rate": 4.177777777777778e-06,
-      "loss": 0.114,
+      "loss": 0.0786,
       "step": 3125
     },
     {
-      "epoch": 0.52,
+      "epoch": 1.04,
       "learning_rate": 4.122222222222222e-06,
-      "loss": 0.119,
+      "loss": 0.0735,
       "step": 3150
     },
     {
-      "epoch": 0.52,
+      "epoch": 1.05,
       "learning_rate": 4.066666666666667e-06,
-      "loss": 0.1177,
+      "loss": 0.0784,
       "step": 3175
     },
     {
-      "epoch": 0.53,
+      "epoch": 1.06,
       "learning_rate": 4.011111111111111e-06,
-      "loss": 0.1138,
+      "loss": 0.0777,
       "step": 3200
     },
     {
-      "epoch": 0.53,
+      "epoch": 1.07,
       "learning_rate": 3.955555555555556e-06,
-      "loss": 0.1212,
+      "loss": 0.0734,
       "step": 3225
     },
     {
-      "epoch": 0.54,
+      "epoch": 1.07,
       "learning_rate": 3.900000000000001e-06,
-      "loss": 0.1151,
+      "loss": 0.0747,
       "step": 3250
     },
     {
-      "epoch": 0.54,
+      "epoch": 1.08,
       "learning_rate": 3.844444444444445e-06,
-      "loss": 0.1154,
+      "loss": 0.0774,
       "step": 3275
     },
     {
-      "epoch": 0.55,
+      "epoch": 1.09,
       "learning_rate": 3.7888888888888893e-06,
-      "loss": 0.1101,
+      "loss": 0.077,
       "step": 3300
     },
     {
-      "epoch": 0.55,
+      "epoch": 1.1,
       "learning_rate": 3.7333333333333337e-06,
-      "loss": 0.1233,
+      "loss": 0.0745,
       "step": 3325
     },
     {
-      "epoch": 0.55,
+      "epoch": 1.11,
       "learning_rate": 3.6777777777777778e-06,
-      "loss": 0.1052,
+      "loss": 0.0762,
       "step": 3350
     },
     {
-      "epoch": 0.56,
+      "epoch": 1.11,
       "learning_rate": 3.6222222222222226e-06,
-      "loss": 0.1138,
+      "loss": 0.0828,
       "step": 3375
     },
     {
-      "epoch": 0.56,
+      "epoch": 1.12,
       "learning_rate": 3.566666666666667e-06,
-      "loss": 0.1155,
+      "loss": 0.0739,
       "step": 3400
     },
     {
-      "epoch": 0.57,
+      "epoch": 1.13,
       "learning_rate": 3.511111111111111e-06,
-      "loss": 0.1102,
+      "loss": 0.0735,
       "step": 3425
     },
     {
-      "epoch": 0.57,
+      "epoch": 1.14,
       "learning_rate": 3.455555555555556e-06,
-      "loss": 0.1095,
+      "loss": 0.0725,
       "step": 3450
     },
     {
-      "epoch": 0.57,
+      "epoch": 1.15,
       "learning_rate": 3.4000000000000005e-06,
-      "loss": 0.1166,
+      "loss": 0.0743,
       "step": 3475
     },
     {
-      "epoch": 0.58,
+      "epoch": 1.16,
       "learning_rate": 3.3444444444444445e-06,
-      "loss": 0.1119,
+      "loss": 0.0725,
       "step": 3500
     },
     {
-      "epoch": 0.58,
+      "epoch": 1.16,
       "learning_rate": 3.2888888888888894e-06,
-      "loss": 0.1138,
+      "loss": 0.075,
       "step": 3525
     },
     {
-      "epoch": 0.59,
+      "epoch": 1.17,
       "learning_rate": 3.2333333333333334e-06,
-      "loss": 0.1107,
+      "loss": 0.0762,
       "step": 3550
     },
     {
-      "epoch": 0.59,
+      "epoch": 1.18,
       "learning_rate": 3.177777777777778e-06,
-      "loss": 0.1165,
+      "loss": 0.0723,
       "step": 3575
     },
     {
-      "epoch": 0.59,
+      "epoch": 1.19,
       "learning_rate": 3.1222222222222228e-06,
-      "loss": 0.1075,
+      "loss": 0.0798,
       "step": 3600
     },
     {
-      "epoch": 0.6,
+      "epoch": 1.2,
       "learning_rate": 3.066666666666667e-06,
-      "loss": 0.1115,
+      "loss": 0.073,
       "step": 3625
     },
     {
-      "epoch": 0.6,
+      "epoch": 1.21,
       "learning_rate": 3.0111111111111113e-06,
-      "loss": 0.113,
+      "loss": 0.0748,
       "step": 3650
     },
     {
-      "epoch": 0.61,
+      "epoch": 1.21,
       "learning_rate": 2.955555555555556e-06,
-      "loss": 0.1046,
+      "loss": 0.0726,
       "step": 3675
     },
     {
-      "epoch": 0.61,
+      "epoch": 1.22,
       "learning_rate": 2.9e-06,
-      "loss": 0.1118,
+      "loss": 0.0737,
       "step": 3700
     },
     {
-      "epoch": 0.62,
+      "epoch": 1.23,
       "learning_rate": 2.8444444444444446e-06,
-      "loss": 0.1084,
+      "loss": 0.0727,
       "step": 3725
     },
     {
-      "epoch": 0.62,
+      "epoch": 1.24,
       "learning_rate": 2.788888888888889e-06,
-      "loss": 0.1085,
+      "loss": 0.0716,
       "step": 3750
     },
     {
-      "epoch": 0.62,
+      "epoch": 1.25,
       "learning_rate": 2.7333333333333336e-06,
-      "loss": 0.1084,
+      "loss": 0.0723,
       "step": 3775
     },
     {
-      "epoch": 0.63,
+      "epoch": 1.26,
       "learning_rate": 2.677777777777778e-06,
-      "loss": 0.1105,
+      "loss": 0.0778,
       "step": 3800
     },
     {
-      "epoch": 0.63,
+      "epoch": 1.26,
       "learning_rate": 2.6222222222222225e-06,
-      "loss": 0.1143,
+      "loss": 0.0711,
       "step": 3825
     },
     {
-      "epoch": 0.64,
+      "epoch": 1.27,
       "learning_rate": 2.566666666666667e-06,
-      "loss": 0.1034,
+      "loss": 0.0694,
       "step": 3850
     },
     {
-      "epoch": 0.64,
+      "epoch": 1.28,
       "learning_rate": 2.5111111111111114e-06,
-      "loss": 0.1097,
+      "loss": 0.0781,
       "step": 3875
     },
     {
-      "epoch": 0.64,
+      "epoch": 1.29,
       "learning_rate": 2.455555555555556e-06,
-      "loss": 0.1162,
+      "loss": 0.0721,
       "step": 3900
     },
     {
-      "epoch": 0.65,
+      "epoch": 1.3,
       "learning_rate": 2.4000000000000003e-06,
-      "loss": 0.1048,
+      "loss": 0.0709,
       "step": 3925
     },
     {
-      "epoch": 0.65,
+      "epoch": 1.3,
       "learning_rate": 2.3444444444444448e-06,
-      "loss": 0.1125,
+      "loss": 0.0754,
       "step": 3950
     },
     {
-      "epoch": 0.66,
+      "epoch": 1.31,
       "learning_rate": 2.2888888888888892e-06,
-      "loss": 0.1003,
+      "loss": 0.0715,
       "step": 3975
     },
     {
-      "epoch": 0.66,
+      "epoch": 1.32,
       "learning_rate": 2.2333333333333333e-06,
-      "loss": 0.114,
+      "loss": 0.073,
       "step": 4000
     },
     {
-      "epoch": 0.66,
-      "eval_loss": 0.21108412742614746,
-      "eval_runtime": 171.078,
-      "eval_samples_per_second": 9.884,
-      "eval_steps_per_second": 0.31,
-      "eval_wer": 10.909204446109653,
+      "epoch": 1.32,
+      "eval_loss": 0.1944551318883896,
+      "eval_runtime": 170.2938,
+      "eval_samples_per_second": 9.93,
+      "eval_steps_per_second": 0.311,
+      "eval_wer": 9.847633320844261,
       "step": 4000
     },
     {
-      "epoch": 0.66,
+      "epoch": 1.33,
       "learning_rate": 2.1777777777777777e-06,
-      "loss": 0.1054,
+      "loss": 0.0714,
       "step": 4025
     },
     {
-      "epoch": 0.67,
+      "epoch": 1.34,
       "learning_rate": 2.1222222222222226e-06,
-      "loss": 0.109,
+      "loss": 0.0762,
       "step": 4050
     },
     {
-      "epoch": 0.67,
+      "epoch": 1.35,
       "learning_rate": 2.0666666666666666e-06,
-      "loss": 0.1035,
+      "loss": 0.0689,
       "step": 4075
     },
     {
-      "epoch": 0.68,
+      "epoch": 1.35,
       "learning_rate": 2.011111111111111e-06,
-      "loss": 0.11,
+      "loss": 0.0686,
       "step": 4100
     },
     {
-      "epoch": 0.68,
+      "epoch": 1.36,
       "learning_rate": 1.955555555555556e-06,
-      "loss": 0.0994,
+      "loss": 0.0684,
       "step": 4125
     },
     {
-      "epoch": 0.69,
+      "epoch": 1.37,
       "learning_rate": 1.9000000000000002e-06,
-      "loss": 0.1078,
+      "loss": 0.068,
       "step": 4150
     },
     {
-      "epoch": 0.69,
+      "epoch": 1.38,
       "learning_rate": 1.8444444444444445e-06,
-      "loss": 0.1106,
+      "loss": 0.0735,
       "step": 4175
     },
     {
-      "epoch": 0.69,
+      "epoch": 1.39,
       "learning_rate": 1.788888888888889e-06,
-      "loss": 0.1123,
+      "loss": 0.0768,
       "step": 4200
     },
     {
-      "epoch": 0.7,
+      "epoch": 1.4,
       "learning_rate": 1.7333333333333336e-06,
-      "loss": 0.1099,
+      "loss": 0.0681,
       "step": 4225
     },
     {
-      "epoch": 0.7,
+      "epoch": 1.4,
       "learning_rate": 1.6777777777777779e-06,
-      "loss": 0.1108,
+      "loss": 0.0712,
       "step": 4250
     },
     {
-      "epoch": 0.71,
+      "epoch": 1.41,
       "learning_rate": 1.6222222222222223e-06,
-      "loss": 0.1068,
+      "loss": 0.0682,
       "step": 4275
     },
     {
-      "epoch": 0.71,
+      "epoch": 1.42,
       "learning_rate": 1.566666666666667e-06,
-      "loss": 0.1069,
+      "loss": 0.0713,
       "step": 4300
     },
     {
-      "epoch": 0.71,
+      "epoch": 1.43,
       "learning_rate": 1.5111111111111112e-06,
-      "loss": 0.104,
+      "loss": 0.0672,
       "step": 4325
     },
     {
-      "epoch": 0.72,
+      "epoch": 1.44,
       "learning_rate": 1.4555555555555557e-06,
-      "loss": 0.1016,
+      "loss": 0.0745,
       "step": 4350
     },
     {
-      "epoch": 0.72,
+      "epoch": 1.45,
       "learning_rate": 1.4000000000000001e-06,
-      "loss": 0.101,
+      "loss": 0.071,
       "step": 4375
     },
     {
-      "epoch": 0.73,
+      "epoch": 1.45,
       "learning_rate": 1.3444444444444446e-06,
-      "loss": 0.0985,
+      "loss": 0.0754,
       "step": 4400
     },
     {
-      "epoch": 0.73,
+      "epoch": 1.46,
       "learning_rate": 1.288888888888889e-06,
-      "loss": 0.0998,
+      "loss": 0.0703,
       "step": 4425
     },
     {
-      "epoch": 0.74,
+      "epoch": 1.47,
       "learning_rate": 1.2333333333333335e-06,
-      "loss": 0.1042,
+      "loss": 0.071,
       "step": 4450
     },
     {
-      "epoch": 0.74,
+      "epoch": 1.48,
       "learning_rate": 1.1777777777777778e-06,
-      "loss": 0.1032,
+      "loss": 0.0713,
       "step": 4475
     },
     {
-      "epoch": 0.74,
+      "epoch": 1.49,
       "learning_rate": 1.1222222222222222e-06,
-      "loss": 0.0991,
+      "loss": 0.0659,
       "step": 4500
     },
     {
-      "epoch": 0.75,
+      "epoch": 1.49,
       "learning_rate": 1.066666666666667e-06,
-      "loss": 0.1089,
+      "loss": 0.0704,
       "step": 4525
     },
     {
-      "epoch": 0.75,
+      "epoch": 1.5,
       "learning_rate": 1.0111111111111111e-06,
-      "loss": 0.1066,
+      "loss": 0.0692,
       "step": 4550
     },
     {
-      "epoch": 0.76,
+      "epoch": 1.51,
       "learning_rate": 9.555555555555556e-07,
-      "loss": 0.1075,
+      "loss": 0.0655,
       "step": 4575
     },
     {
-      "epoch": 0.76,
+      "epoch": 1.52,
       "learning_rate": 9.000000000000001e-07,
-      "loss": 0.098,
+      "loss": 0.0674,
       "step": 4600
     },
     {
-      "epoch": 0.76,
+      "epoch": 1.53,
       "learning_rate": 8.444444444444445e-07,
-      "loss": 0.1008,
+      "loss": 0.0666,
       "step": 4625
     },
     {
-      "epoch": 0.77,
+      "epoch": 1.54,
       "learning_rate": 7.888888888888889e-07,
-      "loss": 0.1099,
+      "loss": 0.0718,
       "step": 4650
     },
     {
-      "epoch": 0.77,
+      "epoch": 1.54,
       "learning_rate": 7.333333333333334e-07,
-      "loss": 0.1039,
+      "loss": 0.0733,
       "step": 4675
     },
     {
-      "epoch": 0.78,
+      "epoch": 1.55,
       "learning_rate": 6.777777777777779e-07,
-      "loss": 0.1005,
+      "loss": 0.0718,
       "step": 4700
     },
     {
-      "epoch": 0.78,
+      "epoch": 1.56,
       "learning_rate": 6.222222222222223e-07,
-      "loss": 0.1029,
+      "loss": 0.0682,
       "step": 4725
     },
     {
-      "epoch": 0.78,
+      "epoch": 1.57,
       "learning_rate": 5.666666666666667e-07,
-      "loss": 0.1087,
+      "loss": 0.0703,
       "step": 4750
     },
     {
-      "epoch": 0.79,
+      "epoch": 1.58,
       "learning_rate": 5.111111111111112e-07,
-      "loss": 0.1031,
+      "loss": 0.0696,
       "step": 4775
     },
     {
-      "epoch": 0.79,
+      "epoch": 1.59,
       "learning_rate": 4.5555555555555563e-07,
-      "loss": 0.1014,
+      "loss": 0.0703,
       "step": 4800
     },
     {
-      "epoch": 0.8,
+      "epoch": 1.59,
       "learning_rate": 4.0000000000000003e-07,
-      "loss": 0.1109,
+      "loss": 0.0699,
       "step": 4825
     },
     {
-      "epoch": 0.8,
+      "epoch": 1.6,
       "learning_rate": 3.444444444444445e-07,
-      "loss": 0.0992,
+      "loss": 0.0658,
       "step": 4850
     },
     {
-      "epoch": 0.81,
+      "epoch": 1.61,
       "learning_rate": 2.888888888888889e-07,
-      "loss": 0.1012,
+      "loss": 0.0661,
       "step": 4875
     },
     {
-      "epoch": 0.81,
+      "epoch": 1.62,
       "learning_rate": 2.3333333333333336e-07,
-      "loss": 0.1006,
+      "loss": 0.0682,
       "step": 4900
     },
     {
-      "epoch": 0.81,
+      "epoch": 1.63,
       "learning_rate": 1.777777777777778e-07,
-      "loss": 0.1059,
+      "loss": 0.066,
       "step": 4925
     },
     {
-      "epoch": 0.82,
+      "epoch": 1.64,
       "learning_rate": 1.2222222222222225e-07,
-      "loss": 0.1011,
+      "loss": 0.0687,
       "step": 4950
     },
     {
-      "epoch": 0.82,
+      "epoch": 1.64,
       "learning_rate": 6.666666666666668e-08,
-      "loss": 0.1032,
+      "loss": 0.0656,
       "step": 4975
     },
     {
-      "epoch": 0.83,
+      "epoch": 1.65,
       "learning_rate": 1.1111111111111112e-08,
-      "loss": 0.0998,
+      "loss": 0.0666,
       "step": 5000
     },
     {
-      "epoch": 0.83,
-      "eval_loss": 0.20626504719257355,
-      "eval_runtime": 171.5268,
-      "eval_samples_per_second": 9.859,
-      "eval_steps_per_second": 0.309,
-      "eval_wer": 10.759335581366305,
+      "epoch": 1.65,
+      "eval_loss": 0.19057515263557434,
+      "eval_runtime": 173.3279,
+      "eval_samples_per_second": 9.756,
+      "eval_steps_per_second": 0.306,
+      "eval_wer": 9.54165105532659,
       "step": 5000
     },
     {
-      "epoch": 0.83,
+      "epoch": 1.65,
       "step": 5000,
-      "total_flos": 9.23473281024e+19,
-      "train_loss": 0.24075475883483888,
-      "train_runtime": 9843.0142,
-      "train_samples_per_second": 32.51,
-      "train_steps_per_second": 0.508
+      "total_flos": 1.846859986427904e+20,
+      "train_loss": 0.20601170344352723,
+      "train_runtime": 18822.9568,
+      "train_samples_per_second": 34.001,
+      "train_steps_per_second": 0.266
     }
   ],
   "max_steps": 5000,
-  "num_train_epochs": 1,
-  "total_flos": 9.23473281024e+19,
+  "num_train_epochs": 2,
+  "total_flos": 1.846859986427904e+20,
   "trial_name": null,
   "trial_params": null
 }