mattbonnell commited on
Commit
7dc2f11
·
verified ·
1 Parent(s): 450aca0

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