sanchit-gandhi HF staff commited on
Commit
99025de
โ€ข
1 Parent(s): 2f6ab17

Upload tokenizer

Browse files
Files changed (3) hide show
  1. special_tokens_map.json +4 -0
  2. tokenizer_config.json +12 -0
  3. vocab.json +62 -0
special_tokens_map.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "pad_token": "แ",
3
+ "unk_token": "<unk>"
4
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_blank": true,
3
+ "clean_up_tokenization_spaces": true,
4
+ "is_uroman": false,
5
+ "language": "mnw",
6
+ "model_max_length": 1000000000000000019884624838656,
7
+ "normalize": true,
8
+ "pad_token": "แ",
9
+ "phonemize": false,
10
+ "tokenizer_class": "VitsTokenizer",
11
+ "unk_token": "<unk>"
12
+ }
vocab.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ " ": 23,
3
+ "_": 38,
4
+ "แ€€": 29,
5
+ "แ€": 24,
6
+ "แ€‚": 39,
7
+ "แ€ƒ": 54,
8
+ "แ€…": 47,
9
+ "แ€†": 59,
10
+ "แ€‡": 40,
11
+ "แ€Š": 22,
12
+ "แ€": 6,
13
+ "แ€": 52,
14
+ "แ€": 21,
15
+ "แ€‘": 25,
16
+ "แ€’": 32,
17
+ "แ€“": 5,
18
+ "แ€”": 49,
19
+ "แ€•": 10,
20
+ "แ€–": 53,
21
+ "แ€—": 4,
22
+ "แ€˜": 19,
23
+ "แ€™": 9,
24
+ "แ€š": 17,
25
+ "แ€›": 11,
26
+ "แ€œ": 14,
27
+ "แ€": 50,
28
+ "แ€ž": 15,
29
+ "แ€Ÿ": 44,
30
+ "แ€ก": 20,
31
+ "แ€ฃ": 26,
32
+ "แ€ฅ": 3,
33
+ "แ€จ": 18,
34
+ "แ€ฉ": 37,
35
+ "แ€ซ": 1,
36
+ "แ€ฌ": 58,
37
+ "แ€ญ": 30,
38
+ "แ€ฎ": 33,
39
+ "แ€ฏ": 51,
40
+ "แ€ฐ": 28,
41
+ "แ€ฑ": 31,
42
+ "แ€ฒ": 41,
43
+ "แ€ณ": 27,
44
+ "แ€ด": 12,
45
+ "แ€ต": 35,
46
+ "แ€ถ": 48,
47
+ "แ€ธ": 56,
48
+ "แ€น": 43,
49
+ "แ€บ": 45,
50
+ "แ€ป": 34,
51
+ "แ€ผ": 46,
52
+ "แ€ฝ": 16,
53
+ "แ€พ": 42,
54
+ "แ€ฟ": 57,
55
+ "แ†": 2,
56
+ "แš": 8,
57
+ "แœ": 36,
58
+ "แ": 0,
59
+ "แž": 55,
60
+ "แŸ": 13,
61
+ "แ ": 7
62
+ }