add tokenizer
Browse files- vocab.json +2 -4
vocab.json
CHANGED
@@ -1,7 +1,5 @@
|
|
1 |
{
|
2 |
" ": 23,
|
3 |
-
"PAD": 43,
|
4 |
-
"[UNK]": 44,
|
5 |
"آ": 1,
|
6 |
"أ": 33,
|
7 |
"ؤ": 0,
|
@@ -12,7 +10,7 @@
|
|
12 |
"ث": 11,
|
13 |
"ج": 36,
|
14 |
"ح": 38,
|
15 |
-
"خ":
|
16 |
"د": 8,
|
17 |
"ذ": 37,
|
18 |
"ر": 14,
|
@@ -27,7 +25,7 @@
|
|
27 |
"غ": 39,
|
28 |
"ف": 28,
|
29 |
"ق": 16,
|
30 |
-
"ل":
|
31 |
"م": 13,
|
32 |
"ن": 32,
|
33 |
"و": 21,
|
|
|
1 |
{
|
2 |
" ": 23,
|
|
|
|
|
3 |
"آ": 1,
|
4 |
"أ": 33,
|
5 |
"ؤ": 0,
|
|
|
10 |
"ث": 11,
|
11 |
"ج": 36,
|
12 |
"ح": 38,
|
13 |
+
"خ": 20,
|
14 |
"د": 8,
|
15 |
"ذ": 37,
|
16 |
"ر": 14,
|
|
|
25 |
"غ": 39,
|
26 |
"ف": 28,
|
27 |
"ق": 16,
|
28 |
+
"ل": 22,
|
29 |
"م": 13,
|
30 |
"ن": 32,
|
31 |
"و": 21,
|