Upload tokenizer
Browse files- vocab.json +62 -62
vocab.json
CHANGED
@@ -1,66 +1,66 @@
|
|
1 |
{
|
2 |
"[PAD]": 63,
|
3 |
"[UNK]": 62,
|
4 |
-
"|":
|
5 |
-
"ਂ":
|
6 |
-
"ਅ":
|
7 |
-
"ਆ":
|
8 |
-
"ਇ":
|
9 |
-
"ਈ":
|
10 |
-
"ਉ":
|
11 |
-
"ਊ":
|
12 |
-
"ਏ":
|
13 |
-
"ਐ":
|
14 |
-
"ਓ":
|
15 |
-
"ਔ":
|
16 |
-
"ਕ":
|
17 |
-
"ਖ":
|
18 |
-
"ਗ":
|
19 |
-
"ਘ":
|
20 |
-
"ਚ":
|
21 |
-
"ਛ":
|
22 |
-
"ਜ":
|
23 |
-
"ਝ":
|
24 |
-
"ਟ":
|
25 |
-
"ਠ":
|
26 |
-
"ਡ":
|
27 |
-
"ਢ":
|
28 |
-
"ਣ":
|
29 |
-
"ਤ":
|
30 |
-
"ਥ":
|
31 |
-
"ਦ":
|
32 |
-
"ਧ":
|
33 |
-
"ਨ":
|
34 |
-
"ਪ":
|
35 |
-
"ਫ":
|
36 |
-
"ਬ":
|
37 |
-
"ਭ":
|
38 |
-
"ਮ":
|
39 |
-
"ਯ":
|
40 |
-
"ਰ":
|
41 |
-
"ਲ":
|
42 |
-
"ਲ਼":
|
43 |
-
"ਵ":
|
44 |
-
"ਸ਼":
|
45 |
-
"ਸ":
|
46 |
-
"ਹ":
|
47 |
-
"਼":
|
48 |
-
"ਾ":
|
49 |
-
"ਿ":
|
50 |
-
"ੀ":
|
51 |
-
"ੁ":
|
52 |
-
"ੂ":
|
53 |
-
"ੇ":
|
54 |
-
"ੈ":
|
55 |
-
"ੋ":
|
56 |
-
"ੌ":
|
57 |
-
"੍":
|
58 |
-
"ਖ਼":
|
59 |
-
"ਗ਼":
|
60 |
-
"ਜ਼":
|
61 |
-
"ੜ":
|
62 |
-
"ਫ਼":
|
63 |
-
"ੰ":
|
64 |
-
"ੱ":
|
65 |
-
"…":
|
66 |
}
|
|
|
1 |
{
|
2 |
"[PAD]": 63,
|
3 |
"[UNK]": 62,
|
4 |
+
"|": 43,
|
5 |
+
"ਂ": 32,
|
6 |
+
"ਅ": 57,
|
7 |
+
"ਆ": 1,
|
8 |
+
"ਇ": 55,
|
9 |
+
"ਈ": 44,
|
10 |
+
"ਉ": 24,
|
11 |
+
"ਊ": 56,
|
12 |
+
"ਏ": 51,
|
13 |
+
"ਐ": 35,
|
14 |
+
"ਓ": 27,
|
15 |
+
"ਔ": 16,
|
16 |
+
"ਕ": 50,
|
17 |
+
"ਖ": 25,
|
18 |
+
"ਗ": 17,
|
19 |
+
"ਘ": 3,
|
20 |
+
"ਚ": 20,
|
21 |
+
"ਛ": 15,
|
22 |
+
"ਜ": 7,
|
23 |
+
"ਝ": 52,
|
24 |
+
"ਟ": 8,
|
25 |
+
"ਠ": 12,
|
26 |
+
"ਡ": 38,
|
27 |
+
"ਢ": 26,
|
28 |
+
"ਣ": 31,
|
29 |
+
"ਤ": 46,
|
30 |
+
"ਥ": 4,
|
31 |
+
"ਦ": 18,
|
32 |
+
"ਧ": 34,
|
33 |
+
"ਨ": 40,
|
34 |
+
"ਪ": 45,
|
35 |
+
"ਫ": 58,
|
36 |
+
"ਬ": 14,
|
37 |
+
"ਭ": 21,
|
38 |
+
"ਮ": 42,
|
39 |
+
"ਯ": 30,
|
40 |
+
"ਰ": 54,
|
41 |
+
"ਲ": 11,
|
42 |
+
"ਲ਼": 37,
|
43 |
+
"ਵ": 9,
|
44 |
+
"ਸ਼": 28,
|
45 |
+
"ਸ": 49,
|
46 |
+
"ਹ": 19,
|
47 |
+
"਼": 61,
|
48 |
+
"ਾ": 22,
|
49 |
+
"ਿ": 33,
|
50 |
+
"ੀ": 10,
|
51 |
+
"ੁ": 47,
|
52 |
+
"ੂ": 13,
|
53 |
+
"ੇ": 39,
|
54 |
+
"ੈ": 23,
|
55 |
+
"ੋ": 29,
|
56 |
+
"ੌ": 48,
|
57 |
+
"੍": 0,
|
58 |
+
"ਖ਼": 6,
|
59 |
+
"ਗ਼": 53,
|
60 |
+
"ਜ਼": 36,
|
61 |
+
"ੜ": 2,
|
62 |
+
"ਫ਼": 5,
|
63 |
+
"ੰ": 41,
|
64 |
+
"ੱ": 59,
|
65 |
+
"…": 60
|
66 |
}
|