mattbonnell commited on
Commit
1036534
·
verified ·
1 Parent(s): 35e211e

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +44 -44
vocab.json CHANGED
@@ -1,49 +1,49 @@
1
  {
2
  "[PAD]": 46,
3
  "[UNK]": 45,
4
- "a": 6,
5
- "b": 18,
6
  "d": 17,
7
- "e": 9,
8
- "f": 42,
9
- "h": 1,
10
- "i": 37,
11
- "j": 38,
12
- "k": 23,
13
- "l": 41,
14
- "m": 7,
15
- "n": 26,
16
- "o": 33,
17
- "p": 31,
18
- "s": 11,
19
- "t": 39,
20
- "u": 19,
21
- "v": 5,
22
- "w": 2,
23
- "z": 40,
24
- "|": 15,
25
- "æ": 8,
26
- "ð": 3,
27
- "ŋ": 10,
28
- "ɐ": 34,
29
- "ɑ": 30,
30
- "ɔ": 28,
31
- "ə": 12,
32
- "ɚ": 13,
33
- "ɛ": 16,
34
- "ɜ": 0,
35
- "ɡ": 22,
36
- "ɣ": 29,
37
- "ɪ": 24,
38
- "ɹ": 36,
39
- "ɾ": 21,
40
- "ʃ": 32,
41
- "ʊ": 43,
42
- "ʌ": 20,
43
- "ʒ": 35,
44
- "ʔ": 14,
45
- "ː": 44,
46
- "̩": 25,
47
- "θ": 27,
48
- "ᵻ": 4
49
  }
 
1
  {
2
  "[PAD]": 46,
3
  "[UNK]": 45,
4
+ "a": 4,
5
+ "b": 26,
6
  "d": 17,
7
+ "e": 7,
8
+ "f": 28,
9
+ "h": 30,
10
+ "i": 32,
11
+ "j": 42,
12
+ "k": 33,
13
+ "l": 8,
14
+ "m": 15,
15
+ "n": 25,
16
+ "o": 40,
17
+ "p": 29,
18
+ "s": 39,
19
+ "t": 24,
20
+ "u": 0,
21
+ "v": 22,
22
+ "w": 20,
23
+ "z": 38,
24
+ "|": 13,
25
+ "æ": 16,
26
+ "ð": 31,
27
+ "ŋ": 43,
28
+ "ɐ": 12,
29
+ "ɑ": 44,
30
+ "ɔ": 19,
31
+ "ə": 1,
32
+ "ɚ": 27,
33
+ "ɛ": 14,
34
+ "ɜ": 6,
35
+ "ɡ": 23,
36
+ "ɣ": 41,
37
+ "ɪ": 11,
38
+ "ɹ": 37,
39
+ "ɾ": 35,
40
+ "ʃ": 21,
41
+ "ʊ": 9,
42
+ "ʌ": 18,
43
+ "ʒ": 5,
44
+ "ʔ": 3,
45
+ "ː": 34,
46
+ "̩": 36,
47
+ "θ": 10,
48
+ "ᵻ": 2
49
  }