Upload tokenizer
Browse files- added_tokens.json +2 -2
- vocab.json +60 -62
added_tokens.json
CHANGED
@@ -1,4 +1,4 @@
|
|
1 |
{
|
2 |
-
"</s>":
|
3 |
-
"<s>":
|
4 |
}
|
|
|
1 |
{
|
2 |
+
"</s>": 62,
|
3 |
+
"<s>": 61
|
4 |
}
|
vocab.json
CHANGED
@@ -1,65 +1,63 @@
|
|
1 |
{
|
2 |
-
"[PAD]":
|
3 |
-
"[UNK]":
|
4 |
-
"f": 1,
|
5 |
-
"j": 2,
|
6 |
"|": 0,
|
7 |
-
"ء":
|
8 |
-
"آ":
|
9 |
-
"ؤ":
|
10 |
-
"ئ":
|
11 |
-
"ا":
|
12 |
-
"ب":
|
13 |
-
"ت":
|
14 |
-
"ث":
|
15 |
-
"ج":
|
16 |
-
"ح":
|
17 |
-
"خ":
|
18 |
-
"د":
|
19 |
-
"ذ":
|
20 |
-
"ر":
|
21 |
-
"ز":
|
22 |
-
"س":
|
23 |
-
"ش":
|
24 |
-
"ص":
|
25 |
-
"ض":
|
26 |
-
"ط":
|
27 |
-
"ظ":
|
28 |
-
"ع":
|
29 |
-
"غ":
|
30 |
-
"ـ":
|
31 |
-
"ف":
|
32 |
-
"ق":
|
33 |
-
"ك":
|
34 |
-
"ل":
|
35 |
-
"م":
|
36 |
-
"ن":
|
37 |
-
"ه":
|
38 |
-
"و":
|
39 |
-
"ى":
|
40 |
-
"ي":
|
41 |
-
"ٰ":
|
42 |
-
"ټ":
|
43 |
-
"پ":
|
44 |
-
"ځ":
|
45 |
-
"څ":
|
46 |
-
"چ":
|
47 |
-
"ډ":
|
48 |
-
"ړ":
|
49 |
-
"ږ":
|
50 |
-
"ژ":
|
51 |
-
"ښ":
|
52 |
-
"ک":
|
53 |
-
"ګ":
|
54 |
-
"گ":
|
55 |
-
"ں":
|
56 |
-
"ڼ":
|
57 |
-
"ھ":
|
58 |
-
"ۀ":
|
59 |
-
"ہ":
|
60 |
-
"ی":
|
61 |
-
"ۍ":
|
62 |
-
"ې":
|
63 |
-
"ے":
|
64 |
-
"۔":
|
65 |
}
|
|
|
1 |
{
|
2 |
+
"[PAD]": 60,
|
3 |
+
"[UNK]": 59,
|
|
|
|
|
4 |
"|": 0,
|
5 |
+
"ء": 1,
|
6 |
+
"آ": 2,
|
7 |
+
"ؤ": 3,
|
8 |
+
"ئ": 4,
|
9 |
+
"ا": 5,
|
10 |
+
"ب": 6,
|
11 |
+
"ت": 7,
|
12 |
+
"ث": 8,
|
13 |
+
"ج": 9,
|
14 |
+
"ح": 10,
|
15 |
+
"خ": 11,
|
16 |
+
"د": 12,
|
17 |
+
"ذ": 13,
|
18 |
+
"ر": 14,
|
19 |
+
"ز": 15,
|
20 |
+
"س": 16,
|
21 |
+
"ش": 17,
|
22 |
+
"ص": 18,
|
23 |
+
"ض": 19,
|
24 |
+
"ط": 20,
|
25 |
+
"ظ": 21,
|
26 |
+
"ع": 22,
|
27 |
+
"غ": 23,
|
28 |
+
"ـ": 24,
|
29 |
+
"ف": 25,
|
30 |
+
"ق": 26,
|
31 |
+
"ك": 27,
|
32 |
+
"ل": 28,
|
33 |
+
"م": 29,
|
34 |
+
"ن": 30,
|
35 |
+
"ه": 31,
|
36 |
+
"و": 32,
|
37 |
+
"ى": 33,
|
38 |
+
"ي": 34,
|
39 |
+
"ٰ": 35,
|
40 |
+
"ټ": 36,
|
41 |
+
"پ": 37,
|
42 |
+
"ځ": 38,
|
43 |
+
"څ": 39,
|
44 |
+
"چ": 40,
|
45 |
+
"ډ": 41,
|
46 |
+
"ړ": 42,
|
47 |
+
"ږ": 43,
|
48 |
+
"ژ": 44,
|
49 |
+
"ښ": 45,
|
50 |
+
"ک": 46,
|
51 |
+
"ګ": 47,
|
52 |
+
"گ": 48,
|
53 |
+
"ں": 49,
|
54 |
+
"ڼ": 50,
|
55 |
+
"ھ": 51,
|
56 |
+
"ۀ": 52,
|
57 |
+
"ہ": 53,
|
58 |
+
"ی": 54,
|
59 |
+
"ۍ": 55,
|
60 |
+
"ې": 56,
|
61 |
+
"ے": 57,
|
62 |
+
"۔": 58
|
63 |
}
|