aaniket commited on
Commit
c2e3075
1 Parent(s): 6b2be5a

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +61 -61
vocab.json CHANGED
@@ -1,66 +1,66 @@
1
  {
2
  "[PAD]": 63,
3
  "[UNK]": 62,
4
- "aa": 37,
5
- "ae": 38,
6
- "ah": 16,
7
- "ao": 9,
8
- "aw": 34,
9
- "ax": 49,
10
- "ax-h": 15,
11
- "axr": 5,
12
- "ay": 27,
13
- "b": 4,
14
- "bcl": 1,
15
- "ch": 45,
16
- "d": 31,
17
- "dcl": 18,
18
- "dh": 8,
19
- "dx": 19,
20
- "eh": 12,
21
- "el": 44,
22
- "em": 22,
23
- "en": 7,
24
- "eng": 33,
25
- "epi": 57,
26
- "er": 13,
27
- "ey": 20,
28
- "f": 29,
29
- "g": 59,
30
- "gcl": 0,
31
- "h#": 21,
32
- "hh": 25,
33
- "hv": 60,
34
- "ih": 42,
35
- "ix": 41,
36
- "iy": 53,
37
- "jh": 51,
38
- "k": 40,
39
- "kcl": 43,
40
- "l": 3,
41
- "m": 26,
42
- "n": 52,
43
- "ng": 39,
44
- "nx": 6,
45
- "ow": 35,
46
- "oy": 2,
47
- "p": 56,
48
- "pau": 14,
49
- "pcl": 48,
50
- "q": 58,
51
- "r": 30,
52
- "s": 55,
53
- "sh": 10,
54
- "t": 54,
55
- "tcl": 47,
56
- "th": 36,
57
- "uh": 24,
58
- "uw": 46,
59
- "ux": 11,
60
- "v": 17,
61
- "w": 23,
62
- "y": 32,
63
- "z": 50,
64
- "zh": 28,
65
  "|": 61
66
  }
 
1
  {
2
  "[PAD]": 63,
3
  "[UNK]": 62,
4
+ "aa": 24,
5
+ "ae": 6,
6
+ "ah": 12,
7
+ "ao": 40,
8
+ "aw": 35,
9
+ "ax": 15,
10
+ "ax-h": 28,
11
+ "axr": 47,
12
+ "ay": 55,
13
+ "b": 9,
14
+ "bcl": 54,
15
+ "ch": 36,
16
+ "d": 45,
17
+ "dcl": 52,
18
+ "dh": 21,
19
+ "dx": 37,
20
+ "eh": 0,
21
+ "el": 48,
22
+ "em": 59,
23
+ "en": 14,
24
+ "eng": 53,
25
+ "epi": 19,
26
+ "er": 25,
27
+ "ey": 16,
28
+ "f": 50,
29
+ "g": 46,
30
+ "gcl": 17,
31
+ "h#": 5,
32
+ "hh": 32,
33
+ "hv": 2,
34
+ "ih": 44,
35
+ "ix": 57,
36
+ "iy": 29,
37
+ "jh": 30,
38
+ "k": 3,
39
+ "kcl": 8,
40
+ "l": 1,
41
+ "m": 31,
42
+ "n": 42,
43
+ "ng": 43,
44
+ "nx": 60,
45
+ "ow": 27,
46
+ "oy": 49,
47
+ "p": 38,
48
+ "pau": 22,
49
+ "pcl": 10,
50
+ "q": 39,
51
+ "r": 33,
52
+ "s": 7,
53
+ "sh": 58,
54
+ "t": 41,
55
+ "tcl": 26,
56
+ "th": 13,
57
+ "uh": 4,
58
+ "uw": 56,
59
+ "ux": 23,
60
+ "v": 11,
61
+ "w": 18,
62
+ "y": 20,
63
+ "z": 34,
64
+ "zh": 51,
65
  "|": 61
66
  }