SakshiRathi77 commited on
Commit
e0b2353
1 Parent(s): e500924

Upload tokenizer

Browse files
Files changed (2) hide show
  1. added_tokens.json +2 -2
  2. vocab.json +68 -69
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 70,
3
- "<s>": 69
4
  }
 
1
  {
2
+ "</s>": 69,
3
+ "<s>": 68
4
  }
vocab.json CHANGED
@@ -1,72 +1,71 @@
1
  {
2
- "&": 1,
3
- "[PAD]": 69,
4
- "[UNK]": 68,
5
  "|": 0,
6
- "ँ": 3,
7
- "ं": 4,
8
- "ः": 5,
9
- "अ": 6,
10
- "आ": 7,
11
- "इ": 8,
12
- "ई": 9,
13
- "उ": 10,
14
- "ऊ": 11,
15
- "ऋ": 12,
16
- "ए": 13,
17
- "ऐ": 14,
18
- "ऑ": 15,
19
- "ओ": 16,
20
- "औ": 17,
21
- "क": 18,
22
- "ख": 19,
23
- "ग": 20,
24
- "घ": 21,
25
- "च": 22,
26
- "छ": 23,
27
- "ज": 24,
28
- "झ": 25,
29
- "ञ": 26,
30
- "ट": 27,
31
- "ठ": 28,
32
- "ड": 29,
33
- "ढ": 30,
34
- "ण": 31,
35
- "त": 32,
36
- "थ": 33,
37
- "द": 34,
38
- "ध": 35,
39
- "न": 36,
40
- "प": 37,
41
- "फ": 38,
42
- "ब": 39,
43
- "भ": 40,
44
- "म": 41,
45
- "य": 42,
46
- "र": 43,
47
- "ल": 44,
48
- "व": 45,
49
- "श": 46,
50
- "ष": 47,
51
- "स": 48,
52
- "ह": 49,
53
- "़": 50,
54
- "ा": 51,
55
- "ि": 52,
56
- "ी": 53,
57
- "ु": 54,
58
- "ू": 55,
59
- "ृ": 56,
60
- "ॅ": 57,
61
- "े": 58,
62
- "ै": 59,
63
- "ॉ": 60,
64
- "ो": 61,
65
- "ौ": 62,
66
- "्": 63,
67
- "क़": 64,
68
- "ज़": 65,
69
- "ड़": 66,
70
- "ढ़": 67,
71
- "फ़": 68
72
  }
 
1
  {
2
+ "[PAD]": 68,
3
+ "[UNK]": 67,
 
4
  "|": 0,
5
+ "ँ": 2,
6
+ "ं": 3,
7
+ "ः": 4,
8
+ "अ": 5,
9
+ "आ": 6,
10
+ "इ": 7,
11
+ "ई": 8,
12
+ "उ": 9,
13
+ "ऊ": 10,
14
+ "ऋ": 11,
15
+ "ए": 12,
16
+ "ऐ": 13,
17
+ "ऑ": 14,
18
+ "ओ": 15,
19
+ "औ": 16,
20
+ "क": 17,
21
+ "ख": 18,
22
+ "ग": 19,
23
+ "घ": 20,
24
+ "च": 21,
25
+ "छ": 22,
26
+ "ज": 23,
27
+ "झ": 24,
28
+ "ञ": 25,
29
+ "ट": 26,
30
+ "ठ": 27,
31
+ "ड": 28,
32
+ "ढ": 29,
33
+ "ण": 30,
34
+ "त": 31,
35
+ "थ": 32,
36
+ "द": 33,
37
+ "ध": 34,
38
+ "न": 35,
39
+ "प": 36,
40
+ "फ": 37,
41
+ "ब": 38,
42
+ "भ": 39,
43
+ "म": 40,
44
+ "य": 41,
45
+ "र": 42,
46
+ "ल": 43,
47
+ "व": 44,
48
+ "श": 45,
49
+ "ष": 46,
50
+ "स": 47,
51
+ "ह": 48,
52
+ "़": 49,
53
+ "ा": 50,
54
+ "ि": 51,
55
+ "ी": 52,
56
+ "ु": 53,
57
+ "ू": 54,
58
+ "ृ": 55,
59
+ "ॅ": 56,
60
+ "े": 57,
61
+ "ै": 58,
62
+ "ॉ": 59,
63
+ "ो": 60,
64
+ "ौ": 61,
65
+ "्": 62,
66
+ "क़": 63,
67
+ "ज़": 64,
68
+ "ड़": 65,
69
+ "ढ़": 66,
70
+ "फ़": 67
71
  }