sridhar1ga commited on
Commit
4555471
·
verified ·
1 Parent(s): d7cbc42

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +68 -68
vocab.json CHANGED
@@ -1,72 +1,72 @@
1
  {
2
- "!": 33,
3
- "'": 19,
4
- ",": 32,
5
- "-": 49,
6
- ".": 23,
7
- "0": 1,
8
- "1": 20,
9
- "2": 2,
10
- "4": 52,
11
- "5": 34,
12
- "6": 31,
13
- "7": 10,
14
- "8": 12,
15
- "9": 50,
16
- "?": 42,
17
- "A": 7,
18
- "B": 63,
19
- "C": 45,
20
- "D": 53,
21
- "E": 62,
22
- "F": 44,
23
- "G": 40,
24
- "H": 41,
25
- "I": 17,
26
- "J": 51,
27
- "K": 3,
28
- "L": 27,
29
- "M": 8,
30
- "N": 58,
31
- "O": 39,
32
- "P": 5,
33
- "Q": 16,
34
- "R": 64,
35
- "S": 6,
36
- "T": 9,
37
- "U": 47,
38
- "V": 61,
39
- "W": 67,
40
- "X": 43,
41
- "Y": 36,
42
- "Z": 18,
43
  "[PAD]": 69,
44
  "[UNK]": 68,
45
- "a": 26,
46
- "b": 0,
47
- "c": 57,
48
- "d": 30,
49
- "e": 13,
50
- "f": 60,
51
- "g": 48,
52
- "h": 35,
53
- "i": 14,
54
- "j": 66,
55
- "k": 54,
56
- "l": 22,
57
- "m": 55,
58
- "n": 4,
59
- "o": 15,
60
- "p": 37,
61
- "q": 46,
62
- "r": 25,
63
- "s": 29,
64
- "t": 24,
65
- "u": 21,
66
- "v": 56,
67
- "w": 11,
68
- "x": 38,
69
- "y": 28,
70
- "z": 59,
71
- "|": 65
72
  }
 
1
  {
2
+ "!": 15,
3
+ "'": 65,
4
+ ",": 0,
5
+ "-": 41,
6
+ ".": 16,
7
+ "0": 34,
8
+ "1": 1,
9
+ "2": 19,
10
+ "4": 56,
11
+ "5": 2,
12
+ "6": 59,
13
+ "7": 54,
14
+ "8": 49,
15
+ "9": 33,
16
+ "?": 37,
17
+ "A": 14,
18
+ "B": 18,
19
+ "C": 36,
20
+ "D": 35,
21
+ "E": 6,
22
+ "F": 11,
23
+ "G": 48,
24
+ "H": 44,
25
+ "I": 47,
26
+ "J": 12,
27
+ "K": 61,
28
+ "L": 45,
29
+ "M": 29,
30
+ "N": 7,
31
+ "O": 31,
32
+ "P": 62,
33
+ "Q": 43,
34
+ "R": 13,
35
+ "S": 8,
36
+ "T": 39,
37
+ "U": 25,
38
+ "V": 66,
39
+ "W": 46,
40
+ "X": 3,
41
+ "Y": 20,
42
+ "Z": 50,
43
  "[PAD]": 69,
44
  "[UNK]": 68,
45
+ "a": 30,
46
+ "b": 51,
47
+ "c": 52,
48
+ "d": 38,
49
+ "e": 40,
50
+ "f": 28,
51
+ "g": 26,
52
+ "h": 5,
53
+ "i": 22,
54
+ "j": 24,
55
+ "k": 63,
56
+ "l": 17,
57
+ "m": 9,
58
+ "n": 67,
59
+ "o": 10,
60
+ "p": 42,
61
+ "q": 32,
62
+ "r": 57,
63
+ "s": 55,
64
+ "t": 23,
65
+ "u": 58,
66
+ "v": 64,
67
+ "w": 60,
68
+ "x": 27,
69
+ "y": 53,
70
+ "z": 4,
71
+ "|": 21
72
  }