aaniket commited on
Commit
e0a4274
1 Parent(s): 44d4763

Upload tokenizer

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