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