Shehryar718
commited on
Commit
•
ab0f156
1
Parent(s):
3f5a4c5
Upload tokenizer
Browse files- vocab.json +84 -84
vocab.json
CHANGED
@@ -1,88 +1,88 @@
|
|
1 |
{
|
2 |
-
"'":
|
3 |
"[PAD]": 85,
|
4 |
"[UNK]": 84,
|
5 |
-
"`":
|
6 |
-
"|":
|
7 |
-
"،":
|
8 |
-
"ؑ":
|
9 |
-
"ؓ":
|
10 |
-
"ؔ":
|
11 |
-
"؛":
|
12 |
-
"؟":
|
13 |
-
"ء":
|
14 |
-
"آ":
|
15 |
-
"أ":
|
16 |
-
"ؤ":
|
17 |
-
"ئ":
|
18 |
-
"ا":
|
19 |
-
"ب":
|
20 |
-
"ت":
|
21 |
-
"ث":
|
22 |
-
"ج":
|
23 |
-
"ح":
|
24 |
-
"خ":
|
25 |
-
"د":
|
26 |
-
"ذ":
|
27 |
-
"ر":
|
28 |
-
"ز":
|
29 |
-
"س":
|
30 |
-
"ش":
|
31 |
-
"ص":
|
32 |
-
"ض":
|
33 |
-
"ط":
|
34 |
-
"ظ":
|
35 |
-
"ع":
|
36 |
-
"غ":
|
37 |
-
"ف":
|
38 |
-
"ق":
|
39 |
-
"ك":
|
40 |
-
"ل":
|
41 |
-
"م":
|
42 |
-
"ن":
|
43 |
-
"ه":
|
44 |
-
"و":
|
45 |
-
"ى":
|
46 |
-
"ي":
|
47 |
-
"ً":
|
48 |
-
"َ":
|
49 |
-
"ُ":
|
50 |
-
"ِ":
|
51 |
-
"ّ":
|
52 |
-
"ْ":
|
53 |
-
"ٓ":
|
54 |
-
"ٔ":
|
55 |
-
"ٗ":
|
56 |
-
"ٰ":
|
57 |
-
"ٹ":
|
58 |
-
"پ":
|
59 |
-
"چ":
|
60 |
-
"ڈ":
|
61 |
-
"ڑ":
|
62 |
-
"ژ":
|
63 |
-
"ک":
|
64 |
-
"گ":
|
65 |
-
"ں":
|
66 |
-
"ھ":
|
67 |
-
"ہ":
|
68 |
-
"ۂ":
|
69 |
-
"ۃ":
|
70 |
-
"ی":
|
71 |
-
"ے":
|
72 |
-
"ۓ":
|
73 |
-
"۔":
|
74 |
-
"’":
|
75 |
-
"…":
|
76 |
-
"ﭨ":
|
77 |
-
"ﮭ":
|
78 |
-
"ﮯ":
|
79 |
-
"ﯾ":
|
80 |
-
"ﷲ":
|
81 |
-
"ﷺ":
|
82 |
-
"ﺗ":
|
83 |
-
"ﺘ":
|
84 |
-
"ﺩ":
|
85 |
-
"ﺲ":
|
86 |
-
"ﻧ":
|
87 |
-
"ﻮ":
|
88 |
}
|
|
|
1 |
{
|
2 |
+
"'": 28,
|
3 |
"[PAD]": 85,
|
4 |
"[UNK]": 84,
|
5 |
+
"`": 49,
|
6 |
+
"|": 19,
|
7 |
+
"،": 36,
|
8 |
+
"ؑ": 76,
|
9 |
+
"ؓ": 45,
|
10 |
+
"ؔ": 57,
|
11 |
+
"؛": 7,
|
12 |
+
"؟": 25,
|
13 |
+
"ء": 18,
|
14 |
+
"آ": 59,
|
15 |
+
"أ": 35,
|
16 |
+
"ؤ": 12,
|
17 |
+
"ئ": 72,
|
18 |
+
"ا": 47,
|
19 |
+
"ب": 22,
|
20 |
+
"ت": 34,
|
21 |
+
"ث": 15,
|
22 |
+
"ج": 32,
|
23 |
+
"ح": 33,
|
24 |
+
"خ": 23,
|
25 |
+
"د": 82,
|
26 |
+
"ذ": 68,
|
27 |
+
"ر": 2,
|
28 |
+
"ز": 3,
|
29 |
+
"س": 75,
|
30 |
+
"ش": 14,
|
31 |
+
"ص": 30,
|
32 |
+
"ض": 17,
|
33 |
+
"ط": 69,
|
34 |
+
"ظ": 11,
|
35 |
+
"ع": 41,
|
36 |
+
"غ": 67,
|
37 |
+
"ف": 26,
|
38 |
+
"ق": 66,
|
39 |
+
"ك": 70,
|
40 |
+
"ل": 4,
|
41 |
+
"م": 62,
|
42 |
+
"ن": 5,
|
43 |
+
"ه": 16,
|
44 |
+
"و": 71,
|
45 |
+
"ى": 55,
|
46 |
+
"ي": 10,
|
47 |
+
"ً": 21,
|
48 |
+
"َ": 80,
|
49 |
+
"ُ": 64,
|
50 |
+
"ِ": 73,
|
51 |
+
"ّ": 53,
|
52 |
+
"ْ": 31,
|
53 |
+
"ٓ": 0,
|
54 |
+
"ٔ": 43,
|
55 |
+
"ٗ": 77,
|
56 |
+
"ٰ": 81,
|
57 |
+
"ٹ": 20,
|
58 |
+
"پ": 24,
|
59 |
+
"چ": 83,
|
60 |
+
"ڈ": 54,
|
61 |
+
"ڑ": 74,
|
62 |
+
"ژ": 44,
|
63 |
+
"ک": 78,
|
64 |
+
"گ": 29,
|
65 |
+
"ں": 1,
|
66 |
+
"ھ": 60,
|
67 |
+
"ہ": 61,
|
68 |
+
"ۂ": 27,
|
69 |
+
"ۃ": 9,
|
70 |
+
"ی": 50,
|
71 |
+
"ے": 38,
|
72 |
+
"ۓ": 48,
|
73 |
+
"۔": 52,
|
74 |
+
"’": 79,
|
75 |
+
"…": 51,
|
76 |
+
"ﭨ": 40,
|
77 |
+
"ﮭ": 42,
|
78 |
+
"ﮯ": 37,
|
79 |
+
"ﯾ": 6,
|
80 |
+
"ﷲ": 58,
|
81 |
+
"ﷺ": 65,
|
82 |
+
"ﺗ": 8,
|
83 |
+
"ﺘ": 46,
|
84 |
+
"ﺩ": 39,
|
85 |
+
"ﺲ": 63,
|
86 |
+
"ﻧ": 56,
|
87 |
+
"ﻮ": 13
|
88 |
}
|