SakshiRathi77
commited on
Commit
•
e0b2353
1
Parent(s):
e500924
Upload tokenizer
Browse files- added_tokens.json +2 -2
- vocab.json +68 -69
added_tokens.json
CHANGED
@@ -1,4 +1,4 @@
|
|
1 |
{
|
2 |
-
"</s>":
|
3 |
-
"<s>":
|
4 |
}
|
|
|
1 |
{
|
2 |
+
"</s>": 69,
|
3 |
+
"<s>": 68
|
4 |
}
|
vocab.json
CHANGED
@@ -1,72 +1,71 @@
|
|
1 |
{
|
2 |
-
"
|
3 |
-
"[
|
4 |
-
"[UNK]": 68,
|
5 |
"|": 0,
|
6 |
-
"ँ":
|
7 |
-
"ं":
|
8 |
-
"ः":
|
9 |
-
"अ":
|
10 |
-
"आ":
|
11 |
-
"इ":
|
12 |
-
"ई":
|
13 |
-
"उ":
|
14 |
-
"ऊ":
|
15 |
-
"ऋ":
|
16 |
-
"ए":
|
17 |
-
"ऐ":
|
18 |
-
"ऑ":
|
19 |
-
"ओ":
|
20 |
-
"औ":
|
21 |
-
"क":
|
22 |
-
"ख":
|
23 |
-
"ग":
|
24 |
-
"घ":
|
25 |
-
"च":
|
26 |
-
"छ":
|
27 |
-
"ज":
|
28 |
-
"झ":
|
29 |
-
"ञ":
|
30 |
-
"ट":
|
31 |
-
"ठ":
|
32 |
-
"ड":
|
33 |
-
"ढ":
|
34 |
-
"ण":
|
35 |
-
"त":
|
36 |
-
"थ":
|
37 |
-
"द":
|
38 |
-
"ध":
|
39 |
-
"न":
|
40 |
-
"प":
|
41 |
-
"फ":
|
42 |
-
"ब":
|
43 |
-
"भ":
|
44 |
-
"म":
|
45 |
-
"य":
|
46 |
-
"र":
|
47 |
-
"ल":
|
48 |
-
"व":
|
49 |
-
"श":
|
50 |
-
"ष":
|
51 |
-
"स":
|
52 |
-
"ह":
|
53 |
-
"़":
|
54 |
-
"ा":
|
55 |
-
"ि":
|
56 |
-
"ी":
|
57 |
-
"ु":
|
58 |
-
"ू":
|
59 |
-
"ृ":
|
60 |
-
"ॅ":
|
61 |
-
"े":
|
62 |
-
"ै":
|
63 |
-
"ॉ":
|
64 |
-
"ो":
|
65 |
-
"ौ":
|
66 |
-
"्":
|
67 |
-
"क़":
|
68 |
-
"ज़":
|
69 |
-
"ड़":
|
70 |
-
"ढ़":
|
71 |
-
"फ़":
|
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 |
}
|