sridhar1ga commited on
Commit
e57250a
1 Parent(s): 9c396a7

Upload tokenizer

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