lynguyenminh commited on
Commit
1dcb092
1 Parent(s): 18a4b91

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +104 -104
vocab.json CHANGED
@@ -1,109 +1,109 @@
1
  {
2
- "/": 73,
3
- "0": 101,
4
- "1": 81,
5
- "2": 43,
6
- "3": 95,
7
- "4": 16,
8
- "5": 74,
9
- "6": 32,
10
- "7": 47,
11
- "8": 26,
12
- "9": 69,
13
  "[PAD]": 106,
14
  "[UNK]": 105,
15
- "a": 99,
16
- "b": 41,
17
- "c": 100,
18
- "d": 40,
19
- "e": 49,
20
- "f": 42,
21
- "g": 17,
22
- "h": 80,
23
- "i": 30,
24
- "k": 36,
25
- "l": 51,
26
- "m": 10,
27
- "n": 76,
28
- "o": 104,
29
- "p": 70,
30
- "q": 71,
31
- "r": 33,
32
- "s": 27,
33
- "t": 14,
34
- "u": 68,
35
- "v": 34,
36
- "w": 64,
37
- "x": 52,
38
- "y": 7,
39
- "z": 77,
40
- "|": 24,
41
- "à": 96,
42
- "á": 29,
43
- "â": 84,
44
- "ã": 21,
45
- "è": 28,
46
- "é": 37,
47
- "ê": 90,
48
- "ì": 35,
49
- "í": 72,
50
- "ò": 5,
51
- "ó": 85,
52
- "ô": 57,
53
- "õ": 0,
54
- "ù": 82,
55
- "ú": 89,
56
- "ý": 18,
57
- "ă": 58,
58
- "đ": 54,
59
- "ĩ": 13,
60
- "ũ": 12,
61
- "ơ": 48,
62
- "ư": 50,
63
- "̀": 83,
64
- "́": 56,
65
- "̣": 44,
66
- "ạ": 94,
67
- "ả": 86,
68
- "ấ": 23,
69
- "ầ": 78,
70
- "ẩ": 60,
71
- "ẫ": 46,
72
- "ậ": 65,
73
- "ắ": 88,
74
- "ằ": 87,
75
- "ẳ": 61,
76
- "ẵ": 53,
77
- "ặ": 79,
78
- "ẹ": 62,
79
- "ẻ": 4,
80
- "ẽ": 91,
81
- "ế": 25,
82
- "ề": 19,
83
- "ể": 6,
84
- "ễ": 9,
85
- "ệ": 75,
86
- "ỉ": 55,
87
- "ị": 66,
88
- "ọ": 1,
89
- "ỏ": 22,
90
- "ố": 2,
91
- "ồ": 63,
92
- "ổ": 45,
93
- "ỗ": 8,
94
- "ộ": 39,
95
- "ớ": 67,
96
- "ờ": 93,
97
- "ở": 92,
98
- "ỡ": 31,
99
- "ợ": 59,
100
- "ụ": 11,
101
- "ủ": 98,
102
- "ứ": 3,
103
  "ừ": 97,
104
- "ử": 20,
105
- "ữ": 102,
106
- "ự": 15,
107
- "ỳ": 38,
108
- "ỷ": 103
109
  }
 
1
  {
2
+ "/": 86,
3
+ "0": 2,
4
+ "1": 51,
5
+ "2": 68,
6
+ "3": 48,
7
+ "4": 63,
8
+ "5": 85,
9
+ "6": 92,
10
+ "7": 103,
11
+ "8": 0,
12
+ "9": 43,
13
  "[PAD]": 106,
14
  "[UNK]": 105,
15
+ "a": 45,
16
+ "b": 62,
17
+ "c": 25,
18
+ "d": 41,
19
+ "e": 76,
20
+ "f": 27,
21
+ "g": 19,
22
+ "h": 7,
23
+ "i": 17,
24
+ "k": 31,
25
+ "l": 81,
26
+ "m": 46,
27
+ "n": 12,
28
+ "o": 18,
29
+ "p": 1,
30
+ "q": 56,
31
+ "r": 55,
32
+ "s": 8,
33
+ "t": 53,
34
+ "u": 90,
35
+ "v": 59,
36
+ "w": 82,
37
+ "x": 75,
38
+ "y": 98,
39
+ "z": 70,
40
+ "|": 3,
41
+ "à": 38,
42
+ "á": 6,
43
+ "â": 104,
44
+ "ã": 22,
45
+ "è": 102,
46
+ "é": 11,
47
+ "ê": 44,
48
+ "ì": 69,
49
+ "í": 87,
50
+ "ò": 77,
51
+ "ó": 84,
52
+ "ô": 94,
53
+ "õ": 100,
54
+ "ù": 96,
55
+ "ú": 30,
56
+ "ý": 54,
57
+ "ă": 9,
58
+ "đ": 13,
59
+ "ĩ": 14,
60
+ "ũ": 101,
61
+ "ơ": 20,
62
+ "ư": 67,
63
+ "̀": 23,
64
+ "́": 24,
65
+ "̣": 74,
66
+ "ạ": 39,
67
+ "ả": 47,
68
+ "ấ": 15,
69
+ "ầ": 10,
70
+ "ẩ": 36,
71
+ "ẫ": 49,
72
+ "ậ": 58,
73
+ "ắ": 61,
74
+ "ằ": 34,
75
+ "ẳ": 28,
76
+ "ẵ": 37,
77
+ "ặ": 29,
78
+ "ẹ": 65,
79
+ "ẻ": 73,
80
+ "ẽ": 72,
81
+ "ế": 80,
82
+ "ề": 35,
83
+ "ể": 93,
84
+ "ễ": 16,
85
+ "ệ": 89,
86
+ "ỉ": 32,
87
+ "ị": 99,
88
+ "ọ": 66,
89
+ "ỏ": 42,
90
+ "ố": 83,
91
+ "ồ": 40,
92
+ "ổ": 88,
93
+ "ỗ": 52,
94
+ "ộ": 79,
95
+ "ớ": 78,
96
+ "ờ": 21,
97
+ "ở": 60,
98
+ "ỡ": 33,
99
+ "ợ": 5,
100
+ "ụ": 91,
101
+ "ủ": 4,
102
+ "ứ": 64,
103
  "ừ": 97,
104
+ "ử": 95,
105
+ "ữ": 26,
106
+ "ự": 57,
107
+ "ỳ": 71,
108
+ "ỷ": 50
109
  }