mattbonnell commited on
Commit
ef2cdb5
·
verified ·
1 Parent(s): ad82afb

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