Commit
•
b8b5733
1
Parent(s):
cccf916
Upload tokenizer
Browse files- special_tokens_map.json +4 -0
- tokenizer_config.json +12 -0
- vocab.json +86 -0
special_tokens_map.json
ADDED
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"pad_token": "—",
|
3 |
+
"unk_token": "<unk>"
|
4 |
+
}
|
tokenizer_config.json
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_blank": true,
|
3 |
+
"clean_up_tokenization_spaces": true,
|
4 |
+
"is_uroman": false,
|
5 |
+
"language": "mal",
|
6 |
+
"model_max_length": 1000000000000000019884624838656,
|
7 |
+
"normalize": true,
|
8 |
+
"pad_token": "—",
|
9 |
+
"phonemize": false,
|
10 |
+
"tokenizer_class": "VitsTokenizer",
|
11 |
+
"unk_token": "<unk>"
|
12 |
+
}
|
vocab.json
ADDED
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
" ": 61,
|
3 |
+
"'": 81,
|
4 |
+
"-": 33,
|
5 |
+
"0": 6,
|
6 |
+
"1": 57,
|
7 |
+
"2": 54,
|
8 |
+
"3": 66,
|
9 |
+
"4": 23,
|
10 |
+
"5": 79,
|
11 |
+
"6": 46,
|
12 |
+
"_": 43,
|
13 |
+
"c": 75,
|
14 |
+
"i": 18,
|
15 |
+
"m": 30,
|
16 |
+
"o": 73,
|
17 |
+
"q": 82,
|
18 |
+
"ം": 29,
|
19 |
+
"ഃ": 38,
|
20 |
+
"അ": 63,
|
21 |
+
"ആ": 26,
|
22 |
+
"ഇ": 10,
|
23 |
+
"ഈ": 67,
|
24 |
+
"ഉ": 19,
|
25 |
+
"ഊ": 13,
|
26 |
+
"എ": 47,
|
27 |
+
"ഏ": 45,
|
28 |
+
"ഐ": 80,
|
29 |
+
"ഒ": 5,
|
30 |
+
"ഓ": 51,
|
31 |
+
"ഔ": 68,
|
32 |
+
"ക": 56,
|
33 |
+
"ഖ": 8,
|
34 |
+
"ഗ": 39,
|
35 |
+
"ഘ": 50,
|
36 |
+
"ങ": 59,
|
37 |
+
"ച": 14,
|
38 |
+
"ഛ": 35,
|
39 |
+
"ജ": 31,
|
40 |
+
"ഞ": 9,
|
41 |
+
"ട": 37,
|
42 |
+
"ഠ": 22,
|
43 |
+
"ഡ": 11,
|
44 |
+
"ഢ": 62,
|
45 |
+
"ണ": 40,
|
46 |
+
"ത": 1,
|
47 |
+
"ഥ": 49,
|
48 |
+
"ദ": 78,
|
49 |
+
"ധ": 32,
|
50 |
+
"ന": 71,
|
51 |
+
"പ": 69,
|
52 |
+
"ഫ": 20,
|
53 |
+
"ബ": 65,
|
54 |
+
"ഭ": 60,
|
55 |
+
"മ": 64,
|
56 |
+
"യ": 16,
|
57 |
+
"ര": 2,
|
58 |
+
"റ": 44,
|
59 |
+
"ല": 21,
|
60 |
+
"ള": 3,
|
61 |
+
"ഴ": 4,
|
62 |
+
"വ": 76,
|
63 |
+
"ശ": 83,
|
64 |
+
"ഷ": 17,
|
65 |
+
"സ": 27,
|
66 |
+
"ഹ": 34,
|
67 |
+
"ാ": 15,
|
68 |
+
"ി": 12,
|
69 |
+
"ീ": 52,
|
70 |
+
"ു": 28,
|
71 |
+
"ൂ": 7,
|
72 |
+
"ൃ": 74,
|
73 |
+
"െ": 24,
|
74 |
+
"േ": 55,
|
75 |
+
"ൈ": 53,
|
76 |
+
"ൊ": 72,
|
77 |
+
"ോ": 42,
|
78 |
+
"്": 36,
|
79 |
+
"ൗ": 25,
|
80 |
+
"ൺ": 70,
|
81 |
+
"ൻ": 77,
|
82 |
+
"ർ": 48,
|
83 |
+
"ൽ": 41,
|
84 |
+
"ൾ": 58,
|
85 |
+
"—": 0
|
86 |
+
}
|