Upload tokenizer
Browse files- vocab.json +14 -15
vocab.json
CHANGED
@@ -1,21 +1,20 @@
|
|
1 |
{
|
2 |
"!": 0,
|
3 |
"\"": 26,
|
4 |
-
",":
|
5 |
".": 28,
|
6 |
":": 41,
|
7 |
-
"[PAD]":
|
8 |
-
"[UNK]":
|
9 |
-
"p": 47,
|
10 |
"|": 4,
|
11 |
" ": 21,
|
12 |
"،": 16,
|
13 |
-
"؟":
|
14 |
"ء": 30,
|
15 |
"آ": 37,
|
16 |
"أ": 38,
|
17 |
"ؤ": 5,
|
18 |
-
"إ":
|
19 |
"ئ": 40,
|
20 |
"ا": 22,
|
21 |
"ب": 34,
|
@@ -26,26 +25,26 @@
|
|
26 |
"ح": 27,
|
27 |
"خ": 44,
|
28 |
"د": 9,
|
29 |
-
"ذ":
|
30 |
"ر": 14,
|
31 |
"ز": 43,
|
32 |
-
"س":
|
33 |
"ش": 29,
|
34 |
"ص": 2,
|
35 |
-
"ض":
|
36 |
"ط": 31,
|
37 |
-
"ظ":
|
38 |
"ع": 46,
|
39 |
-
"غ":
|
40 |
"ـ": 25,
|
41 |
-
"ف":
|
42 |
"ق": 39,
|
43 |
"ك": 1,
|
44 |
"ل": 11,
|
45 |
-
"م":
|
46 |
"ن": 8,
|
47 |
"ه": 3,
|
48 |
-
"و":
|
49 |
"ى": 35,
|
50 |
"ي": 42,
|
51 |
"ً": 12,
|
@@ -53,7 +52,7 @@
|
|
53 |
"ٍ": 45,
|
54 |
"َ": 10,
|
55 |
"ُ": 20,
|
56 |
-
"ِ":
|
57 |
"ّ": 23,
|
58 |
"ْ": 15,
|
59 |
"ٱ": 36,
|
|
|
1 |
{
|
2 |
"!": 0,
|
3 |
"\"": 26,
|
4 |
+
",": 47,
|
5 |
".": 28,
|
6 |
":": 41,
|
7 |
+
"[PAD]": 60,
|
8 |
+
"[UNK]": 59,
|
|
|
9 |
"|": 4,
|
10 |
" ": 21,
|
11 |
"،": 16,
|
12 |
+
"؟": 56,
|
13 |
"ء": 30,
|
14 |
"آ": 37,
|
15 |
"أ": 38,
|
16 |
"ؤ": 5,
|
17 |
+
"إ": 49,
|
18 |
"ئ": 40,
|
19 |
"ا": 22,
|
20 |
"ب": 34,
|
|
|
25 |
"ح": 27,
|
26 |
"خ": 44,
|
27 |
"د": 9,
|
28 |
+
"ذ": 54,
|
29 |
"ر": 14,
|
30 |
"ز": 43,
|
31 |
+
"س": 55,
|
32 |
"ش": 29,
|
33 |
"ص": 2,
|
34 |
+
"ض": 50,
|
35 |
"ط": 31,
|
36 |
+
"ظ": 57,
|
37 |
"ع": 46,
|
38 |
+
"غ": 53,
|
39 |
"ـ": 25,
|
40 |
+
"ف": 48,
|
41 |
"ق": 39,
|
42 |
"ك": 1,
|
43 |
"ل": 11,
|
44 |
+
"م": 52,
|
45 |
"ن": 8,
|
46 |
"ه": 3,
|
47 |
+
"و": 51,
|
48 |
"ى": 35,
|
49 |
"ي": 42,
|
50 |
"ً": 12,
|
|
|
52 |
"ٍ": 45,
|
53 |
"َ": 10,
|
54 |
"ُ": 20,
|
55 |
+
"ِ": 58,
|
56 |
"ّ": 23,
|
57 |
"ْ": 15,
|
58 |
"ٱ": 36,
|