mattbonnell commited on
Commit
7d35a3d
·
verified ·
1 Parent(s): 1ad5a79

Upload tokenizer

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