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