Commit
•
a730290
1
Parent(s):
d633e16
Upload tokenizer
Browse files- special_tokens_map.json +4 -0
- tokenizer_config.json +12 -0
- vocab.json +63 -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": "hoy",
|
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 +1,63 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
" ": 60,
|
3 |
+
"'": 59,
|
4 |
+
"-": 48,
|
5 |
+
"|": 0,
|
6 |
+
"ँ": 58,
|
7 |
+
"ं": 18,
|
8 |
+
"अ": 19,
|
9 |
+
"आ": 15,
|
10 |
+
"इ": 32,
|
11 |
+
"ई": 27,
|
12 |
+
"उ": 33,
|
13 |
+
"ऊ": 29,
|
14 |
+
"ए": 55,
|
15 |
+
"ओ": 51,
|
16 |
+
"औ": 53,
|
17 |
+
"क": 6,
|
18 |
+
"ख": 39,
|
19 |
+
"ग": 20,
|
20 |
+
"घ": 42,
|
21 |
+
"च": 31,
|
22 |
+
"छ": 52,
|
23 |
+
"ज": 30,
|
24 |
+
"झ": 54,
|
25 |
+
"ट": 25,
|
26 |
+
"ठ": 49,
|
27 |
+
"ड": 37,
|
28 |
+
"ढ": 40,
|
29 |
+
"ण": 43,
|
30 |
+
"त": 9,
|
31 |
+
"थ": 47,
|
32 |
+
"द": 2,
|
33 |
+
"ध": 38,
|
34 |
+
"न": 4,
|
35 |
+
"प": 24,
|
36 |
+
"फ": 45,
|
37 |
+
"ब": 26,
|
38 |
+
"भ": 36,
|
39 |
+
"म": 14,
|
40 |
+
"य": 11,
|
41 |
+
"र": 3,
|
42 |
+
"ल": 21,
|
43 |
+
"ळ": 13,
|
44 |
+
"व": 22,
|
45 |
+
"श": 28,
|
46 |
+
"ष": 50,
|
47 |
+
"स": 10,
|
48 |
+
"ह": 23,
|
49 |
+
"़": 56,
|
50 |
+
"ा": 1,
|
51 |
+
"ि": 12,
|
52 |
+
"ी": 7,
|
53 |
+
"ु": 5,
|
54 |
+
"ू": 35,
|
55 |
+
"ृ": 57,
|
56 |
+
"े": 17,
|
57 |
+
"ै": 44,
|
58 |
+
"ॉ": 41,
|
59 |
+
"ो": 16,
|
60 |
+
"ौ": 46,
|
61 |
+
"्": 8,
|
62 |
+
"": 34
|
63 |
+
}
|