add tokenizer
Browse files- vocab.json +14 -14
vocab.json
CHANGED
|
@@ -1,19 +1,19 @@
|
|
| 1 |
{
|
| 2 |
-
" ":
|
| 3 |
"[PAD]": 16,
|
| 4 |
"[UNK]": 15,
|
| 5 |
"ا": 4,
|
| 6 |
-
"ب":
|
| 7 |
-
"ت":
|
| 8 |
-
"د":
|
| 9 |
-
"ز":
|
| 10 |
-
"س":
|
| 11 |
-
"ل":
|
| 12 |
-
"م":
|
| 13 |
-
"ڑ":
|
| 14 |
-
"ک":
|
| 15 |
-
"ھ":
|
| 16 |
-
"ہ":
|
| 17 |
-
"ی":
|
| 18 |
-
"ے":
|
| 19 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
" ": 3,
|
| 3 |
"[PAD]": 16,
|
| 4 |
"[UNK]": 15,
|
| 5 |
"ا": 4,
|
| 6 |
+
"ب": 8,
|
| 7 |
+
"ت": 1,
|
| 8 |
+
"د": 2,
|
| 9 |
+
"ز": 9,
|
| 10 |
+
"س": 10,
|
| 11 |
+
"ل": 6,
|
| 12 |
+
"م": 0,
|
| 13 |
+
"ڑ": 14,
|
| 14 |
+
"ک": 11,
|
| 15 |
+
"ھ": 5,
|
| 16 |
+
"ہ": 12,
|
| 17 |
+
"ی": 7,
|
| 18 |
+
"ے": 13
|
| 19 |
}
|