hadiqa123
/

xls-r-ur-large

Automatic Speech Recognition

Generated from Trainer

Inference Endpoints

Model card Files Files and versions Metrics Training metrics Community

hadiqa123 commited on Sep 20, 2022

Commit

a964faa

•

1 Parent(s): b8580dd

add tokenizer

Files changed (2) hide show

added_tokens.json +2 -2
vocab.json +53 -47

added_tokens.json CHANGED Viewed

@@ -1,4 +1,4 @@
 {
-  "</s>": 52,
-  "<s>": 51
 }

 {
+  "</s>": 58,
+  "<s>": 57
 }

vocab.json CHANGED Viewed

@@ -1,53 +1,59 @@
 {
-  "[PAD]": 50,
-  "[UNK]": 49,
   "|": 0,
   "،": 1,
   "ء": 2,
   "آ": 3,
-  "ؤ": 4,
-  "ئ": 5,
-  "ا": 6,
-  "ب": 7,
-  "ت": 8,
-  "ث": 9,
-  "ج": 10,
-  "ح": 11,
-  "خ": 12,
-  "د": 13,
-  "ذ": 14,
-  "ر": 15,
-  "ز": 16,
-  "س": 17,
-  "ش": 18,
-  "ص": 19,
-  "ض": 20,
-  "ط": 21,
-  "ظ": 22,
-  "ع": 23,
-  "غ": 24,
-  "ف": 25,
-  "ق": 26,
-  "ل": 27,
-  "م": 28,
-  "ن": 29,
-  "و": 30,
-  "ى": 31,
-  "ي": 32,
-  "ّ": 33,
-  "ٹ": 34,
-  "پ": 35,
-  "چ": 36,
-  "ڈ": 37,
-  "ڑ": 38,
-  "ژ": 39,
-  "ک": 40,
-  "گ": 41,
-  "ں": 42,
-  "ھ": 43,
-  "ہ": 44,
-  "ۂ": 45,
-  "ی": 46,
-  "ے": 47,
-  "۔": 48
 }

 {
+  "[PAD]": 56,
+  "[UNK]": 55,
   "|": 0,
   "،": 1,
   "ء": 2,
   "آ": 3,
+  "أ": 4,
+  "ؤ": 5,
+  "ئ": 6,
+  "ا": 7,
+  "ب": 8,
+  "ت": 9,
+  "ث": 10,
+  "ج": 11,
+  "ح": 12,
+  "خ": 13,
+  "د": 14,
+  "ذ": 15,
+  "ر": 16,
+  "ز": 17,
+  "س": 18,
+  "ش": 19,
+  "ص": 20,
+  "ض": 21,
+  "ط": 22,
+  "ظ": 23,
+  "ع": 24,
+  "غ": 25,
+  "ف": 26,
+  "ق": 27,
+  "ل": 28,
+  "م": 29,
+  "ن": 30,
+  "و": 31,
+  "ى": 32,
+  "ي": 33,
+  "ّ": 34,
+  "ٹ": 35,
+  "پ": 36,
+  "چ": 37,
+  "ڈ": 38,
+  "ڑ": 39,
+  "ژ": 40,
+  "ک": 41,
+  "گ": 42,
+  "ں": 43,
+  "ھ": 44,
+  "ہ": 45,
+  "ۂ": 46,
+  "ۃ": 47,
+  "ی": 48,
+  "ے": 49,
+  "ۓ": 50,
+  "۔": 51,
+  "’": 52,
+  "ﷲ": 53,
+  "ﷺ": 54
 }