Upload tokenizer
Browse files- vocab.json +39 -59
vocab.json
CHANGED
@@ -1,62 +1,42 @@
|
|
1 |
{
|
2 |
-
"
|
3 |
-
"
|
4 |
-
"
|
5 |
-
"
|
6 |
-
"
|
7 |
-
"
|
8 |
-
"
|
9 |
-
"
|
10 |
-
"
|
11 |
-
"
|
12 |
-
"c": 43,
|
13 |
-
"e": 6,
|
14 |
-
"f": 51,
|
15 |
-
"j": 40,
|
16 |
-
"l": 12,
|
17 |
-
"m": 48,
|
18 |
-
"o": 33,
|
19 |
-
"q": 0,
|
20 |
-
"r": 32,
|
21 |
-
"t": 14,
|
22 |
-
"u": 30,
|
23 |
-
"w": 47,
|
24 |
-
"|": 44,
|
25 |
-
"ء": 31,
|
26 |
-
"آ": 28,
|
27 |
-
"أ": 45,
|
28 |
-
"إ": 27,
|
29 |
-
"ئ": 23,
|
30 |
-
"ا": 35,
|
31 |
-
"ب": 46,
|
32 |
"ة": 9,
|
33 |
-
"ت":
|
34 |
-
"ث":
|
35 |
-
"ج":
|
36 |
-
"ح":
|
37 |
-
"خ":
|
38 |
-
"د":
|
39 |
-
"ذ":
|
40 |
-
"ر":
|
41 |
-
"ز":
|
42 |
-
"س":
|
43 |
-
"ش":
|
44 |
-
"ص":
|
45 |
-
"ض":
|
46 |
-
"ط":
|
47 |
-
"ظ":
|
48 |
-
"ع":
|
49 |
-
"غ":
|
50 |
-
"ف":
|
51 |
-
"ق":
|
52 |
-
"ك":
|
53 |
-
"ل":
|
54 |
-
"م":
|
55 |
-
"ن":
|
56 |
-
"ه":
|
57 |
-
"و":
|
58 |
-
"ى":
|
59 |
-
"ي":
|
60 |
-
"ٱ":
|
61 |
-
"ڤ":
|
62 |
}
|
|
|
1 |
{
|
2 |
+
"[PAD]": 39,
|
3 |
+
"[UNK]": 38,
|
4 |
+
"|": 27,
|
5 |
+
"ء": 0,
|
6 |
+
"آ": 14,
|
7 |
+
"أ": 19,
|
8 |
+
"إ": 8,
|
9 |
+
"ئ": 2,
|
10 |
+
"ا": 3,
|
11 |
+
"ب": 37,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
"ة": 9,
|
13 |
+
"ت": 23,
|
14 |
+
"ث": 22,
|
15 |
+
"ج": 31,
|
16 |
+
"ح": 21,
|
17 |
+
"خ": 11,
|
18 |
+
"د": 13,
|
19 |
+
"ذ": 28,
|
20 |
+
"ر": 5,
|
21 |
+
"ز": 12,
|
22 |
+
"س": 16,
|
23 |
+
"ش": 15,
|
24 |
+
"ص": 10,
|
25 |
+
"ض": 30,
|
26 |
+
"ط": 6,
|
27 |
+
"ظ": 4,
|
28 |
+
"ع": 35,
|
29 |
+
"غ": 20,
|
30 |
+
"ف": 17,
|
31 |
+
"ق": 1,
|
32 |
+
"ك": 33,
|
33 |
+
"ل": 34,
|
34 |
+
"م": 24,
|
35 |
+
"ن": 36,
|
36 |
+
"ه": 18,
|
37 |
+
"و": 7,
|
38 |
+
"ى": 32,
|
39 |
+
"ي": 26,
|
40 |
+
"ٱ": 25,
|
41 |
+
"ڤ": 29
|
42 |
}
|