mattbonnell commited on
Commit
7bdb941
·
verified ·
1 Parent(s): 2db983f

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