hts98 commited on
Commit
72c00b3
·
1 Parent(s): e4cfa88

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +104 -104
vocab.json CHANGED
@@ -1,110 +1,110 @@
1
  {
2
- ",": 69,
3
- ".": 22,
4
- "0": 24,
5
- "1": 50,
6
- "2": 48,
7
- "3": 25,
8
- "4": 99,
9
- "5": 42,
10
- "6": 87,
11
- "7": 105,
12
- "8": 18,
13
- "9": 64,
14
  "[PAD]": 107,
15
  "[UNK]": 106,
16
- "a": 93,
17
- "b": 73,
18
- "c": 35,
19
- "d": 43,
20
- "e": 68,
21
- "f": 75,
22
- "g": 84,
23
- "h": 77,
24
- "i": 101,
25
- "j": 19,
26
- "k": 92,
27
- "l": 5,
28
- "m": 81,
29
- "n": 103,
30
- "o": 7,
31
- "p": 45,
32
- "q": 95,
33
- "r": 102,
34
- "s": 1,
35
- "t": 12,
36
- "u": 104,
37
- "v": 16,
38
- "w": 20,
39
- "x": 3,
40
- "y": 23,
41
- "z": 11,
42
- "|": 2,
43
- "à": 83,
44
- "á": 57,
45
- "â": 80,
46
- "ã": 54,
47
- "è": 30,
48
- "é": 78,
49
- "ê": 37,
50
- "ì": 14,
51
- "í": 62,
52
- "ò": 28,
53
- "ó": 33,
54
- "ô": 53,
55
- "õ": 9,
56
- "ù": 85,
57
- "ú": 86,
58
- "ý": 15,
59
- "ă": 72,
60
- "đ": 52,
61
- "ĩ": 59,
62
- "ũ": 49,
63
- "ơ": 21,
64
  "ư": 29,
65
- "ạ": 97,
66
- "ả": 51,
67
- "ấ": 61,
68
- "ầ": 63,
69
- "ẩ": 55,
70
- "ẫ": 79,
71
- "ậ": 66,
72
- "ắ": 41,
73
- "ằ": 10,
74
- "ẳ": 26,
75
- "ẵ": 13,
76
- "ặ": 4,
77
- "ẹ": 40,
78
- "ẻ": 17,
79
- "ẽ": 88,
80
- "ế": 65,
81
  "ề": 58,
82
- "ể": 90,
83
- "ễ": 67,
84
- "ệ": 44,
85
- "ỉ": 96,
86
- "ị": 98,
87
- "ọ": 34,
88
- "ỏ": 46,
89
- "ố": 71,
90
- "ồ": 0,
91
- "ổ": 100,
92
- "ỗ": 8,
93
- "ộ": 70,
94
- "ớ": 94,
95
- "ờ": 47,
96
- "ở": 89,
97
- "ỡ": 31,
98
- "ợ": 38,
99
- "ụ": 27,
100
- "ủ": 36,
101
- "ứ": 32,
102
- "ừ": 76,
103
- "ử": 60,
104
- "ữ": 39,
105
- "ự": 6,
106
- "ỳ": 91,
107
- "ỷ": 74,
108
- "ỹ": 56,
109
- "₫": 82
110
  }
 
1
  {
2
+ ",": 90,
3
+ ".": 9,
4
+ "0": 19,
5
+ "1": 94,
6
+ "2": 23,
7
+ "3": 85,
8
+ "4": 89,
9
+ "5": 20,
10
+ "6": 40,
11
+ "7": 96,
12
+ "8": 33,
13
+ "9": 48,
14
  "[PAD]": 107,
15
  "[UNK]": 106,
16
+ "a": 37,
17
+ "b": 92,
18
+ "c": 17,
19
+ "d": 88,
20
+ "e": 71,
21
+ "f": 80,
22
+ "g": 76,
23
+ "h": 91,
24
+ "i": 28,
25
+ "j": 103,
26
+ "k": 95,
27
+ "l": 44,
28
+ "m": 41,
29
+ "n": 68,
30
+ "o": 18,
31
+ "p": 59,
32
+ "q": 67,
33
+ "r": 83,
34
+ "s": 26,
35
+ "t": 35,
36
+ "u": 93,
37
+ "v": 87,
38
+ "w": 82,
39
+ "x": 60,
40
+ "y": 52,
41
+ "z": 70,
42
+ "|": 6,
43
+ "à": 105,
44
+ "á": 99,
45
+ "â": 1,
46
+ "ã": 25,
47
+ "è": 24,
48
+ "é": 84,
49
+ "ê": 47,
50
+ "ì": 46,
51
+ "í": 34,
52
+ "ò": 51,
53
+ "ó": 39,
54
+ "ô": 32,
55
+ "õ": 63,
56
+ "ù": 65,
57
+ "ú": 72,
58
+ "ý": 3,
59
+ "ă": 74,
60
+ "đ": 8,
61
+ "ĩ": 79,
62
+ "ũ": 36,
63
+ "ơ": 69,
64
  "ư": 29,
65
+ "ạ": 16,
66
+ "ả": 64,
67
+ "ấ": 5,
68
+ "ầ": 54,
69
+ "ẩ": 31,
70
+ "ẫ": 7,
71
+ "ậ": 57,
72
+ "ắ": 101,
73
+ "ằ": 53,
74
+ "ẳ": 98,
75
+ "ẵ": 11,
76
+ "ặ": 61,
77
+ "ẹ": 78,
78
+ "ẻ": 0,
79
+ "ẽ": 21,
80
+ "ế": 49,
81
  "ề": 58,
82
+ "ể": 27,
83
+ "ễ": 77,
84
+ "ệ": 50,
85
+ "ỉ": 38,
86
+ "ị": 13,
87
+ "ọ": 42,
88
+ "ỏ": 56,
89
+ "ố": 15,
90
+ "ồ": 10,
91
+ "ổ": 86,
92
+ "ỗ": 62,
93
+ "ộ": 55,
94
+ "ớ": 14,
95
+ "ờ": 43,
96
+ "ở": 45,
97
+ "ỡ": 104,
98
+ "ợ": 4,
99
+ "ụ": 102,
100
+ "ủ": 30,
101
+ "ứ": 22,
102
+ "ừ": 66,
103
+ "ử": 12,
104
+ "ữ": 81,
105
+ "ự": 75,
106
+ "ỳ": 73,
107
+ "ỷ": 2,
108
+ "ỹ": 97,
109
+ "₫": 100
110
  }