aaniket commited on
Commit
52de8da
1 Parent(s): 0571ffa

Upload tokenizer

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