hungadc commited on
Commit
a9aa948
1 Parent(s): ef10515

Upload tokenizer

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