aaniket commited on
Commit
21076d7
1 Parent(s): 9ad7eeb

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +60 -60
vocab.json CHANGED
@@ -1,65 +1,65 @@
1
  {
2
  "[PAD]": 62,
3
  "[UNK]": 61,
4
- "aa": 50,
5
- "ae": 38,
6
- "ah": 7,
7
- "ao": 30,
8
- "aw": 43,
9
- "ax": 46,
10
- "ax-h": 2,
11
- "axr": 23,
12
- "ay": 53,
13
- "b": 32,
14
- "bcl": 14,
15
- "ch": 17,
16
- "d": 41,
17
- "dcl": 22,
18
- "dh": 1,
19
- "dx": 4,
20
- "eh": 58,
21
- "el": 24,
22
- "em": 10,
23
- "en": 20,
24
- "eng": 52,
25
- "epi": 15,
26
- "er": 29,
27
- "ey": 27,
28
- "f": 45,
29
- "g": 16,
30
- "gcl": 49,
31
- "h#": 6,
32
- "hh": 60,
33
- "hv": 5,
34
- "ih": 56,
35
- "ix": 59,
36
- "iy": 8,
37
- "jh": 9,
38
- "k": 48,
39
- "kcl": 31,
40
- "l": 25,
41
- "m": 44,
42
- "n": 12,
43
- "ng": 26,
44
- "nx": 21,
45
- "ow": 35,
46
- "oy": 47,
47
- "p": 3,
48
- "pau": 39,
49
- "pcl": 42,
50
  "q": 0,
51
- "r": 51,
52
- "s": 37,
53
- "sh": 54,
54
- "t": 18,
55
- "tcl": 40,
56
- "th": 33,
57
- "uh": 28,
58
- "uw": 34,
59
- "ux": 55,
60
- "v": 57,
61
- "w": 19,
62
- "y": 13,
63
- "z": 11,
64
- "zh": 36
65
  }
 
1
  {
2
  "[PAD]": 62,
3
  "[UNK]": 61,
4
+ "aa": 9,
5
+ "ae": 54,
6
+ "ah": 17,
7
+ "ao": 18,
8
+ "aw": 46,
9
+ "ax": 5,
10
+ "ax-h": 3,
11
+ "axr": 1,
12
+ "ay": 48,
13
+ "b": 19,
14
+ "bcl": 32,
15
+ "ch": 28,
16
+ "d": 34,
17
+ "dcl": 37,
18
+ "dh": 47,
19
+ "dx": 31,
20
+ "eh": 22,
21
+ "el": 33,
22
+ "em": 2,
23
+ "en": 36,
24
+ "eng": 57,
25
+ "epi": 16,
26
+ "er": 38,
27
+ "ey": 24,
28
+ "f": 60,
29
+ "g": 30,
30
+ "gcl": 23,
31
+ "h#": 53,
32
+ "hh": 11,
33
+ "hv": 58,
34
+ "ih": 4,
35
+ "ix": 49,
36
+ "iy": 42,
37
+ "jh": 14,
38
+ "k": 10,
39
+ "kcl": 41,
40
+ "l": 56,
41
+ "m": 27,
42
+ "n": 44,
43
+ "ng": 39,
44
+ "nx": 45,
45
+ "ow": 15,
46
+ "oy": 51,
47
+ "p": 6,
48
+ "pau": 59,
49
+ "pcl": 50,
50
  "q": 0,
51
+ "r": 13,
52
+ "s": 35,
53
+ "sh": 55,
54
+ "t": 43,
55
+ "tcl": 8,
56
+ "th": 52,
57
+ "uh": 21,
58
+ "uw": 40,
59
+ "ux": 25,
60
+ "v": 12,
61
+ "w": 29,
62
+ "y": 26,
63
+ "z": 20,
64
+ "zh": 7
65
  }