Upload tokenizer
Browse files- vocab.json +48 -48
vocab.json
CHANGED
@@ -1,53 +1,53 @@
|
|
1 |
{
|
2 |
"[PAD]": 50,
|
3 |
"[UNK]": 49,
|
4 |
-
"a":
|
5 |
-
"b":
|
6 |
-
"c":
|
7 |
-
"e":
|
8 |
-
"g":
|
9 |
-
"h":
|
10 |
-
"i":
|
11 |
-
"k":
|
12 |
-
"l":
|
13 |
-
"m":
|
14 |
-
"n":
|
15 |
-
"o":
|
16 |
-
"p":
|
17 |
-
"q":
|
18 |
"r": 19,
|
19 |
-
"s":
|
20 |
-
"t":
|
21 |
-
"u":
|
22 |
-
"v":
|
23 |
-
"x":
|
24 |
-
"y":
|
25 |
-
"|":
|
26 |
-
"à":
|
27 |
-
"á":
|
28 |
-
"ã":
|
29 |
-
"ê":
|
30 |
-
"ó":
|
31 |
-
"ô":
|
32 |
-
"ú":
|
33 |
-
"đ":
|
34 |
-
"ũ":
|
35 |
-
"ơ":
|
36 |
-
"ư":
|
37 |
-
"ạ":
|
38 |
-
"ả":
|
39 |
-
"ầ":
|
40 |
-
"ậ":
|
41 |
-
"ắ":
|
42 |
-
"ẹ":
|
43 |
-
"ẽ":
|
44 |
-
"ế":
|
45 |
-
"ề":
|
46 |
-
"ể":
|
47 |
-
"ọ":
|
48 |
-
"ố":
|
49 |
-
"ỗ":
|
50 |
-
"ớ":
|
51 |
-
"ờ":
|
52 |
-
"ợ":
|
53 |
}
|
|
|
1 |
{
|
2 |
"[PAD]": 50,
|
3 |
"[UNK]": 49,
|
4 |
+
"a": 14,
|
5 |
+
"b": 11,
|
6 |
+
"c": 46,
|
7 |
+
"e": 37,
|
8 |
+
"g": 15,
|
9 |
+
"h": 26,
|
10 |
+
"i": 29,
|
11 |
+
"k": 24,
|
12 |
+
"l": 0,
|
13 |
+
"m": 31,
|
14 |
+
"n": 41,
|
15 |
+
"o": 5,
|
16 |
+
"p": 28,
|
17 |
+
"q": 35,
|
18 |
"r": 19,
|
19 |
+
"s": 36,
|
20 |
+
"t": 25,
|
21 |
+
"u": 47,
|
22 |
+
"v": 30,
|
23 |
+
"x": 43,
|
24 |
+
"y": 20,
|
25 |
+
"|": 16,
|
26 |
+
"à": 42,
|
27 |
+
"á": 45,
|
28 |
+
"ã": 7,
|
29 |
+
"ê": 13,
|
30 |
+
"ó": 6,
|
31 |
+
"ô": 21,
|
32 |
+
"ú": 8,
|
33 |
+
"đ": 9,
|
34 |
+
"ũ": 44,
|
35 |
+
"ơ": 27,
|
36 |
+
"ư": 18,
|
37 |
+
"ạ": 12,
|
38 |
+
"ả": 4,
|
39 |
+
"ầ": 10,
|
40 |
+
"ậ": 34,
|
41 |
+
"ắ": 38,
|
42 |
+
"ẹ": 40,
|
43 |
+
"ẽ": 3,
|
44 |
+
"ế": 33,
|
45 |
+
"ề": 23,
|
46 |
+
"ể": 1,
|
47 |
+
"ọ": 2,
|
48 |
+
"ố": 17,
|
49 |
+
"ỗ": 22,
|
50 |
+
"ớ": 32,
|
51 |
+
"ờ": 48,
|
52 |
+
"ợ": 39
|
53 |
}
|