aaniket commited on
Commit
1474b58
1 Parent(s): 294e834

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