aaniket commited on
Commit
6ceadf0
1 Parent(s): 43acdc6

Upload tokenizer

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