mattbonnell commited on
Commit
ae0b28e
·
verified ·
1 Parent(s): 5baa43e

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +44 -44
vocab.json CHANGED
@@ -1,48 +1,48 @@
1
  {
2
  "[PAD]": 45,
3
  "[UNK]": 44,
4
- "a": 22,
5
- "b": 31,
6
- "d": 3,
7
- "e": 20,
8
- "f": 43,
9
- "h": 6,
10
- "i": 34,
11
- "j": 32,
12
- "k": 24,
13
- "l": 10,
14
- "m": 25,
15
- "n": 16,
16
- "o": 15,
17
- "p": 18,
18
- "s": 8,
19
- "t": 26,
20
- "u": 0,
21
- "v": 21,
22
- "w": 5,
23
- "z": 37,
24
- "|": 19,
25
- "æ": 35,
26
- "ð": 17,
27
- "ŋ": 4,
28
- "ɐ": 14,
29
- "ɑ": 30,
30
- "ɔ": 9,
31
- "ə": 11,
32
- "ɚ": 7,
33
- "ɛ": 29,
34
- "ɜ": 39,
35
- "ɡ": 42,
36
- "ɪ": 2,
37
- "ɹ": 12,
38
- "ɾ": 40,
39
- "ʃ": 33,
40
- "ʊ": 38,
41
- "ʌ": 28,
42
- "ʒ": 41,
43
- "ʔ": 23,
44
- "ː": 27,
45
- "̩": 1,
46
- "θ": 13,
47
- "ᵻ": 36
48
  }
 
1
  {
2
  "[PAD]": 45,
3
  "[UNK]": 44,
4
+ "a": 15,
5
+ "b": 28,
6
+ "d": 34,
7
+ "e": 43,
8
+ "f": 42,
9
+ "h": 13,
10
+ "i": 39,
11
+ "j": 22,
12
+ "k": 8,
13
+ "l": 11,
14
+ "m": 20,
15
+ "n": 25,
16
+ "o": 19,
17
+ "p": 5,
18
+ "s": 35,
19
+ "t": 16,
20
+ "u": 2,
21
+ "v": 26,
22
+ "w": 4,
23
+ "z": 14,
24
+ "|": 6,
25
+ "æ": 40,
26
+ "ð": 36,
27
+ "ŋ": 30,
28
+ "ɐ": 38,
29
+ "ɑ": 37,
30
+ "ɔ": 32,
31
+ "ə": 12,
32
+ "ɚ": 1,
33
+ "ɛ": 18,
34
+ "ɜ": 24,
35
+ "ɡ": 0,
36
+ "ɪ": 7,
37
+ "ɹ": 33,
38
+ "ɾ": 17,
39
+ "ʃ": 23,
40
+ "ʊ": 3,
41
+ "ʌ": 9,
42
+ "ʒ": 21,
43
+ "ʔ": 27,
44
+ "ː": 31,
45
+ "̩": 29,
46
+ "θ": 10,
47
+ "ᵻ": 41
48
  }