kingabzpro commited on
Commit
50b6465
1 Parent(s): d143682

Upload tokenizer

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