Upload tokenizer
Browse files- vocab.json +95 -95
vocab.json
CHANGED
@@ -1,99 +1,99 @@
|
|
1 |
{
|
2 |
-
"'":
|
3 |
"[PAD]": 96,
|
4 |
"[UNK]": 95,
|
5 |
-
"a":
|
6 |
-
"b":
|
7 |
-
"c":
|
8 |
-
"d":
|
9 |
-
"e":
|
10 |
-
"f":
|
11 |
-
"g":
|
12 |
-
"h":
|
13 |
-
"i":
|
14 |
-
"j":
|
15 |
-
"k":
|
16 |
-
"l":
|
17 |
-
"m":
|
18 |
-
"n":
|
19 |
-
"o":
|
20 |
-
"p":
|
21 |
-
"q":
|
22 |
-
"r":
|
23 |
-
"s":
|
24 |
-
"t":
|
25 |
-
"u":
|
26 |
-
"v":
|
27 |
-
"w":
|
28 |
-
"x":
|
29 |
-
"y":
|
30 |
-
"z":
|
31 |
-
"|":
|
32 |
-
"à":
|
33 |
-
"á":
|
34 |
-
"â":
|
35 |
-
"ã":
|
36 |
-
"è":
|
37 |
-
"é":
|
38 |
-
"ê":
|
39 |
-
"ì":
|
40 |
-
"í":
|
41 |
-
"ò":
|
42 |
-
"ó":
|
43 |
-
"ô":
|
44 |
-
"õ":
|
45 |
-
"ù":
|
46 |
-
"ú":
|
47 |
-
"ý":
|
48 |
-
"ă":
|
49 |
-
"đ":
|
50 |
-
"ĩ":
|
51 |
-
"ũ":
|
52 |
-
"ơ":
|
53 |
-
"ư":
|
54 |
-
"ạ":
|
55 |
-
"ả":
|
56 |
-
"ấ":
|
57 |
-
"ầ":
|
58 |
-
"ẩ":
|
59 |
-
"ẫ":
|
60 |
-
"ậ":
|
61 |
-
"ắ":
|
62 |
-
"ằ":
|
63 |
-
"ẳ":
|
64 |
-
"ẵ":
|
65 |
-
"ặ":
|
66 |
-
"ẹ":
|
67 |
-
"ẻ":
|
68 |
-
"ẽ":
|
69 |
-
"ế":
|
70 |
-
"ề":
|
71 |
-
"ể":
|
72 |
-
"ễ":
|
73 |
-
"ệ":
|
74 |
-
"ỉ":
|
75 |
-
"ị":
|
76 |
-
"ọ":
|
77 |
-
"ỏ":
|
78 |
-
"ố":
|
79 |
-
"ồ":
|
80 |
-
"ổ":
|
81 |
-
"ỗ":
|
82 |
-
"ộ":
|
83 |
-
"ớ":
|
84 |
-
"ờ":
|
85 |
-
"ở":
|
86 |
-
"ỡ":
|
87 |
-
"ợ":
|
88 |
-
"ụ":
|
89 |
-
"ủ":
|
90 |
-
"ứ":
|
91 |
-
"ừ":
|
92 |
-
"ử":
|
93 |
-
"ữ":
|
94 |
-
"ự":
|
95 |
-
"ỳ":
|
96 |
-
"ỵ":
|
97 |
-
"ỷ":
|
98 |
-
"ỹ":
|
99 |
}
|
|
|
1 |
{
|
2 |
+
"'": 69,
|
3 |
"[PAD]": 96,
|
4 |
"[UNK]": 95,
|
5 |
+
"a": 29,
|
6 |
+
"b": 77,
|
7 |
+
"c": 78,
|
8 |
+
"d": 38,
|
9 |
+
"e": 3,
|
10 |
+
"f": 84,
|
11 |
+
"g": 9,
|
12 |
+
"h": 22,
|
13 |
+
"i": 51,
|
14 |
+
"j": 26,
|
15 |
+
"k": 19,
|
16 |
+
"l": 28,
|
17 |
+
"m": 46,
|
18 |
+
"n": 43,
|
19 |
+
"o": 48,
|
20 |
+
"p": 87,
|
21 |
+
"q": 65,
|
22 |
+
"r": 52,
|
23 |
+
"s": 83,
|
24 |
+
"t": 70,
|
25 |
+
"u": 42,
|
26 |
+
"v": 44,
|
27 |
+
"w": 1,
|
28 |
+
"x": 32,
|
29 |
+
"y": 92,
|
30 |
+
"z": 73,
|
31 |
+
"|": 49,
|
32 |
+
"à": 91,
|
33 |
+
"á": 0,
|
34 |
+
"â": 80,
|
35 |
+
"ã": 60,
|
36 |
+
"è": 89,
|
37 |
+
"é": 86,
|
38 |
+
"ê": 36,
|
39 |
+
"ì": 16,
|
40 |
+
"í": 64,
|
41 |
+
"ò": 61,
|
42 |
+
"ó": 8,
|
43 |
+
"ô": 54,
|
44 |
+
"õ": 56,
|
45 |
+
"ù": 53,
|
46 |
+
"ú": 7,
|
47 |
+
"ý": 31,
|
48 |
+
"ă": 66,
|
49 |
+
"đ": 24,
|
50 |
+
"ĩ": 81,
|
51 |
+
"ũ": 21,
|
52 |
+
"ơ": 85,
|
53 |
+
"ư": 17,
|
54 |
+
"ạ": 40,
|
55 |
+
"ả": 10,
|
56 |
+
"ấ": 39,
|
57 |
+
"ầ": 33,
|
58 |
+
"ẩ": 62,
|
59 |
+
"ẫ": 47,
|
60 |
+
"ậ": 55,
|
61 |
+
"ắ": 50,
|
62 |
+
"ằ": 75,
|
63 |
+
"ẳ": 82,
|
64 |
+
"ẵ": 23,
|
65 |
+
"ặ": 34,
|
66 |
+
"ẹ": 5,
|
67 |
+
"ẻ": 11,
|
68 |
+
"ẽ": 27,
|
69 |
+
"ế": 4,
|
70 |
+
"ề": 35,
|
71 |
+
"ể": 93,
|
72 |
+
"ễ": 76,
|
73 |
+
"ệ": 25,
|
74 |
+
"ỉ": 41,
|
75 |
+
"ị": 74,
|
76 |
+
"ọ": 13,
|
77 |
+
"ỏ": 18,
|
78 |
+
"ố": 37,
|
79 |
+
"ồ": 6,
|
80 |
+
"ổ": 12,
|
81 |
+
"ỗ": 59,
|
82 |
+
"ộ": 14,
|
83 |
+
"ớ": 79,
|
84 |
+
"ờ": 20,
|
85 |
+
"ở": 88,
|
86 |
+
"ỡ": 63,
|
87 |
+
"ợ": 68,
|
88 |
+
"ụ": 72,
|
89 |
+
"ủ": 71,
|
90 |
+
"ứ": 57,
|
91 |
+
"ừ": 2,
|
92 |
+
"ử": 67,
|
93 |
+
"ữ": 30,
|
94 |
+
"ự": 45,
|
95 |
+
"ỳ": 90,
|
96 |
+
"ỵ": 58,
|
97 |
+
"ỷ": 15,
|
98 |
+
"ỹ": 94
|
99 |
}
|