Upload tokenizer
Browse files- vocab.json +70 -70
vocab.json
CHANGED
@@ -1,76 +1,76 @@
|
|
1 |
{
|
2 |
-
"'":
|
3 |
"[PAD]": 73,
|
4 |
"[UNK]": 72,
|
5 |
-
"|":
|
6 |
-
"ം":
|
7 |
-
"ഃ":
|
8 |
-
"അ":
|
9 |
-
"ആ":
|
10 |
-
"ഇ":
|
11 |
-
"ഈ":
|
12 |
-
"ഉ":
|
13 |
-
"ഊ":
|
14 |
-
"എ":
|
15 |
-
"ഏ":
|
16 |
-
"ഐ":
|
17 |
-
"ഒ":
|
18 |
-
"ഓ":
|
19 |
-
"ഔ":
|
20 |
-
"ക":
|
21 |
-
"ഖ":
|
22 |
-
"ഗ":
|
23 |
-
"ഘ":
|
24 |
-
"ങ":
|
25 |
-
"ച":
|
26 |
-
"ഛ":
|
27 |
-
"ജ":
|
28 |
-
"ഞ":
|
29 |
-
"ട":
|
30 |
-
"ഠ":
|
31 |
-
"ഡ":
|
32 |
-
"ഢ":
|
33 |
-
"ണ":
|
34 |
-
"ത":
|
35 |
-
"ഥ":
|
36 |
-
"ദ":
|
37 |
-
"ധ":
|
38 |
-
"ന":
|
39 |
-
"പ":
|
40 |
-
"ഫ":
|
41 |
-
"ബ":
|
42 |
-
"ഭ":
|
43 |
-
"മ":
|
44 |
-
"യ":
|
45 |
-
"ര":
|
46 |
-
"റ":
|
47 |
-
"ല":
|
48 |
-
"ള":
|
49 |
-
"ഴ":
|
50 |
-
"വ":
|
51 |
-
"ശ":
|
52 |
"ഷ": 63,
|
53 |
-
"സ":
|
54 |
-
"ഹ":
|
55 |
-
"ാ":
|
56 |
-
"ി":
|
57 |
-
"ീ":
|
58 |
-
"ു":
|
59 |
-
"ൂ":
|
60 |
-
"ൃ":
|
61 |
-
"െ":
|
62 |
-
"േ":
|
63 |
-
"ൈ":
|
64 |
-
"ൊ":
|
65 |
-
"ോ":
|
66 |
"ൌ": 56,
|
67 |
-
"്":
|
68 |
-
"ൗ":
|
69 |
-
"ൺ":
|
70 |
-
"ൻ":
|
71 |
-
"ർ":
|
72 |
-
"ൽ":
|
73 |
-
"ൾ":
|
74 |
-
"ൿ":
|
75 |
-
"’":
|
76 |
}
|
|
|
1 |
{
|
2 |
+
"'": 69,
|
3 |
"[PAD]": 73,
|
4 |
"[UNK]": 72,
|
5 |
+
"|": 31,
|
6 |
+
"ം": 25,
|
7 |
+
"ഃ": 6,
|
8 |
+
"അ": 55,
|
9 |
+
"ആ": 11,
|
10 |
+
"ഇ": 27,
|
11 |
+
"ഈ": 60,
|
12 |
+
"ഉ": 68,
|
13 |
+
"ഊ": 43,
|
14 |
+
"എ": 38,
|
15 |
+
"ഏ": 71,
|
16 |
+
"ഐ": 46,
|
17 |
+
"ഒ": 18,
|
18 |
+
"ഓ": 47,
|
19 |
+
"ഔ": 51,
|
20 |
+
"ക": 64,
|
21 |
+
"ഖ": 44,
|
22 |
+
"ഗ": 9,
|
23 |
+
"ഘ": 52,
|
24 |
+
"ങ": 61,
|
25 |
+
"ച": 19,
|
26 |
+
"ഛ": 7,
|
27 |
+
"ജ": 32,
|
28 |
+
"ഞ": 62,
|
29 |
+
"ട": 29,
|
30 |
+
"ഠ": 53,
|
31 |
+
"ഡ": 42,
|
32 |
+
"ഢ": 23,
|
33 |
+
"ണ": 67,
|
34 |
+
"ത": 8,
|
35 |
+
"ഥ": 24,
|
36 |
+
"ദ": 16,
|
37 |
+
"ധ": 48,
|
38 |
+
"ന": 36,
|
39 |
+
"പ": 28,
|
40 |
+
"ഫ": 5,
|
41 |
+
"ബ": 35,
|
42 |
+
"ഭ": 4,
|
43 |
+
"മ": 65,
|
44 |
+
"യ": 26,
|
45 |
+
"ര": 30,
|
46 |
+
"റ": 13,
|
47 |
+
"ല": 22,
|
48 |
+
"ള": 17,
|
49 |
+
"ഴ": 3,
|
50 |
+
"വ": 21,
|
51 |
+
"ശ": 10,
|
52 |
"ഷ": 63,
|
53 |
+
"സ": 54,
|
54 |
+
"ഹ": 15,
|
55 |
+
"ാ": 40,
|
56 |
+
"ി": 0,
|
57 |
+
"ീ": 34,
|
58 |
+
"ു": 59,
|
59 |
+
"ൂ": 57,
|
60 |
+
"ൃ": 70,
|
61 |
+
"െ": 1,
|
62 |
+
"േ": 33,
|
63 |
+
"ൈ": 41,
|
64 |
+
"ൊ": 58,
|
65 |
+
"ോ": 12,
|
66 |
"ൌ": 56,
|
67 |
+
"്": 14,
|
68 |
+
"ൗ": 49,
|
69 |
+
"ൺ": 39,
|
70 |
+
"ൻ": 45,
|
71 |
+
"ർ": 50,
|
72 |
+
"ൽ": 2,
|
73 |
+
"ൾ": 66,
|
74 |
+
"ൿ": 37,
|
75 |
+
"’": 20
|
76 |
}
|