mattbonnell commited on
Commit
60bc83e
·
verified ·
1 Parent(s): 0057435

Upload tokenizer

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