aaniket commited on
Commit
d5a1162
1 Parent(s): 3f2ce83

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