lynguyenminh commited on
Commit
0577fdf
·
1 Parent(s): 05f5b55

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +95 -95
vocab.json CHANGED
@@ -1,99 +1,99 @@
1
  {
2
- "%": 48,
3
- "0": 32,
4
- "1": 20,
5
- "2": 10,
6
- "3": 24,
7
- "4": 37,
8
- "5": 21,
9
- "6": 1,
10
- "7": 76,
11
- "8": 6,
12
- "9": 47,
13
  "[PAD]": 96,
14
  "[UNK]": 95,
15
- "a": 86,
16
- "b": 69,
17
- "c": 87,
18
- "d": 61,
19
- "e": 16,
20
- "f": 5,
21
- "g": 74,
22
- "h": 19,
23
- "i": 44,
24
- "k": 63,
25
- "l": 89,
26
- "m": 49,
27
- "n": 75,
28
- "o": 4,
29
- "p": 29,
30
- "q": 14,
31
- "r": 15,
32
- "s": 42,
33
- "t": 52,
34
- "u": 46,
35
- "v": 2,
36
- "w": 78,
37
- "x": 71,
38
- "y": 80,
39
- "|": 3,
40
- "à": 60,
41
- "á": 79,
42
- "â": 55,
43
- "ã": 38,
44
- "è": 35,
45
- "é": 25,
46
- "ê": 66,
47
- "ì": 85,
48
- "í": 51,
49
- "ò": 94,
50
- "ó": 93,
51
- "ô": 39,
52
- "ù": 34,
53
- "ú": 17,
54
- "ý": 45,
55
- "ă": 62,
56
- "đ": 82,
57
- "ơ": 8,
58
- "ư": 58,
59
- "": 59,
60
- "": 92,
61
- "": 73,
62
- "": 31,
63
- "": 70,
64
- "": 40,
65
- "": 88,
66
- "": 0,
67
- "": 28,
68
- "": 54,
69
- "ặ": 68,
70
- "ẹ": 27,
71
- "ẻ": 7,
72
- "ẽ": 83,
73
- "ế": 65,
74
- "ề": 22,
75
- "ể": 56,
76
- "ệ": 41,
77
- "ỉ": 64,
78
- "ị": 13,
79
- "ọ": 50,
80
- "ỏ": 12,
81
- "ố": 11,
82
- "ồ": 23,
83
- "ổ": 91,
84
- "ỗ": 81,
85
- "ộ": 84,
86
- "ớ": 67,
87
- "ờ": 9,
88
- "ở": 53,
89
- "ỡ": 26,
90
- "ợ": 36,
91
- "ụ": 33,
92
- "ủ": 18,
93
- "ứ": 90,
94
- "ừ": 30,
95
- "ử": 43,
96
- "ữ": 57,
97
- "ự": 72,
98
- "ỳ": 77
99
  }
 
1
  {
2
+ "%": 36,
3
+ "0": 35,
4
+ "1": 22,
5
+ "2": 32,
6
+ "3": 71,
7
+ "4": 92,
8
+ "5": 67,
9
+ "6": 41,
10
+ "7": 26,
11
+ "8": 85,
12
+ "9": 1,
13
  "[PAD]": 96,
14
  "[UNK]": 95,
15
+ "a": 52,
16
+ "b": 20,
17
+ "c": 23,
18
+ "d": 70,
19
+ "e": 50,
20
+ "f": 19,
21
+ "g": 38,
22
+ "h": 7,
23
+ "i": 51,
24
+ "k": 64,
25
+ "l": 82,
26
+ "m": 60,
27
+ "n": 27,
28
+ "o": 43,
29
+ "p": 86,
30
+ "q": 11,
31
+ "r": 45,
32
+ "s": 13,
33
+ "t": 47,
34
+ "u": 44,
35
+ "v": 78,
36
+ "w": 66,
37
+ "x": 21,
38
+ "y": 63,
39
+ "|": 54,
40
+ "à": 28,
41
+ "á": 58,
42
+ "â": 56,
43
+ "ã": 87,
44
+ "è": 29,
45
+ "é": 49,
46
+ "ê": 65,
47
+ "ì": 2,
48
+ "í": 81,
49
+ "ò": 83,
50
+ "ó": 8,
51
+ "ô": 24,
52
+ "ù": 4,
53
+ "ú": 10,
54
+ "ý": 34,
55
+ "ă": 72,
56
+ "đ": 14,
57
+ "ơ": 88,
58
+ "ư": 94,
59
+ "̀": 77,
60
+ "̣": 89,
61
+ "": 76,
62
+ "": 57,
63
+ "": 37,
64
+ "": 39,
65
+ "": 53,
66
+ "": 80,
67
+ "": 25,
68
+ "": 68,
69
+ "ặ": 46,
70
+ "ẹ": 18,
71
+ "ẻ": 59,
72
+ "ẽ": 62,
73
+ "ế": 40,
74
+ "ề": 61,
75
+ "ể": 91,
76
+ "ệ": 33,
77
+ "ỉ": 48,
78
+ "ị": 30,
79
+ "ọ": 0,
80
+ "ỏ": 73,
81
+ "ố": 3,
82
+ "ồ": 42,
83
+ "ổ": 93,
84
+ "ỗ": 79,
85
+ "ộ": 55,
86
+ "ớ": 12,
87
+ "ờ": 84,
88
+ "ở": 15,
89
+ "ỡ": 31,
90
+ "ợ": 17,
91
+ "ụ": 16,
92
+ "ủ": 9,
93
+ "ứ": 5,
94
+ "ừ": 69,
95
+ "ử": 6,
96
+ "ữ": 75,
97
+ "ự": 90,
98
+ "ỳ": 74
99
  }