mattbonnell commited on
Commit
ea574d9
·
verified ·
1 Parent(s): 7bdb941

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