ad019el
/

Tamasheq-XLSR-53-mixed-english

Automatic Speech Recognition

Inference Endpoints

Model card Files Files and versions Metrics Training metrics Community

ad019el commited on Aug 28, 2023

Commit

b42ca29

•

1 Parent(s): 322b957

add tokenizer

Files changed (1) hide show

vocab.json +36 -54

vocab.json CHANGED Viewed

@@ -1,68 +1,50 @@
 {
   "&": 5,
-  "/": 29,
-  "0": 34,
-  "1": 38,
   "2": 0,
-  "7": 43,
-  "9": 56,
-  "[PAD]": 65,
-  "[UNK]": 64,
-  "_": 13,
-  "a": 35,
-  "c": 58,
-  "e": 31,
-  "f": 26,
-  "j": 49,
-  "l": 40,
-  "m": 27,
-  "o": 41,
-  "q": 46,
-  "r": 12,
-  "t": 50,
-  "u": 15,
-  "w": 42,
-  "|": 14,
-  "ء": 23,
-  "آ": 28,
   "أ": 4,
-  "ؤ": 52,
   "إ": 10,
-  "ئ": 21,
-  "ا": 19,
-  "ب": 47,
   "ة": 3,
-  "ت": 62,
-  "ث": 48,
-  "ج": 37,
   "ح": 9,
-  "خ": 45,
-  "د": 32,
-  "ذ": 18,
   "ر": 11,
-  "ز": 25,
   "س": 1,
-  "ش": 39,
   "ص": 2,
   "ض": 7,
-  "ط": 54,
-  "ظ": 55,
-  "ع": 16,
-  "غ": 51,
-  "ف": 53,
-  "ق": 60,
-  "ك": 63,
   "ل": 6,
-  "م": 36,
-  "ن": 33,
-  "ه": 30,
-  "و": 61,
-  "ى": 24,
-  "ي": 20,
-  "ٱ": 17,
-  "چ": 57,
-  "ڤ": 44,
   "ک": 8,
-  "ی": 22,
-  "’": 59
 }

 {
   "&": 5,
+  "0": 28,
   "2": 0,
+  "[PAD]": 47,
+  "[UNK]": 46,
+  "e": 25,
+  "|": 12,
+  "ء": 20,
+  "آ": 23,
   "أ": 4,
+  "ؤ": 37,
   "إ": 10,
+  "ئ": 18,
+  "ا": 16,
+  "ب": 34,
   "ة": 3,
+  "ت": 44,
+  "ث": 35,
+  "ج": 30,
   "ح": 9,
+  "خ": 33,
+  "د": 26,
+  "ذ": 15,
   "ر": 11,
+  "ز": 22,
   "س": 1,
+  "ش": 31,
   "ص": 2,
   "ض": 7,
+  "ط": 39,
+  "ظ": 40,
+  "ع": 13,
+  "غ": 36,
+  "ف": 38,
+  "ق": 42,
+  "ك": 45,
   "ل": 6,
+  "م": 29,
+  "ن": 27,
+  "ه": 24,
+  "و": 43,
+  "ى": 21,
+  "ي": 17,
+  "ٱ": 14,
+  "چ": 41,
+  "ڤ": 32,
   "ک": 8,
+  "ی": 19
 }