mattbonnell commited on
Commit
dfd5fbf
·
verified ·
1 Parent(s): f900425

Upload tokenizer

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