mattbonnell commited on
Commit
67c2dbc
·
verified ·
1 Parent(s): 7fbf32c

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