aaniket commited on
Commit
96dd8dc
1 Parent(s): 6c7c5e1

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