Commit
•
3a51e23
1
Parent(s):
4be1ee9
Upload tokenizer
Browse files- special_tokens_map.json +4 -0
- tokenizer_config.json +12 -0
- vocab.json +59 -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": "mya",
|
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 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
" ": 56,
|
3 |
+
"'": 51,
|
4 |
+
"က": 7,
|
5 |
+
"ခ": 21,
|
6 |
+
"ဂ": 41,
|
7 |
+
"င": 8,
|
8 |
+
"စ": 20,
|
9 |
+
"ဆ": 30,
|
10 |
+
"ဇ": 39,
|
11 |
+
"ဈ": 55,
|
12 |
+
"ဉ": 37,
|
13 |
+
"ည": 13,
|
14 |
+
"ဋ": 50,
|
15 |
+
"ဌ": 49,
|
16 |
+
"ဍ": 53,
|
17 |
+
"ဏ": 43,
|
18 |
+
"တ": 9,
|
19 |
+
"ထ": 26,
|
20 |
+
"ဒ": 40,
|
21 |
+
"ဓ": 45,
|
22 |
+
"န": 16,
|
23 |
+
"ပ": 19,
|
24 |
+
"ဖ": 27,
|
25 |
+
"ဗ": 44,
|
26 |
+
"ဘ": 34,
|
27 |
+
"မ": 11,
|
28 |
+
"ယ": 29,
|
29 |
+
"ရ": 14,
|
30 |
+
"လ": 17,
|
31 |
+
"ဝ": 35,
|
32 |
+
"သ": 6,
|
33 |
+
"ဟ": 33,
|
34 |
+
"အ": 15,
|
35 |
+
"ဣ": 52,
|
36 |
+
"ဤ": 38,
|
37 |
+
"ဥ": 47,
|
38 |
+
"ဦ": 42,
|
39 |
+
"ဧ": 46,
|
40 |
+
"ဩ": 48,
|
41 |
+
"ါ": 25,
|
42 |
+
"ာ": 1,
|
43 |
+
"ိ": 3,
|
44 |
+
"ီ": 31,
|
45 |
+
"ု": 2,
|
46 |
+
"ူ": 23,
|
47 |
+
"ေ": 5,
|
48 |
+
"ဲ": 32,
|
49 |
+
"ံ": 28,
|
50 |
+
"့": 10,
|
51 |
+
"း": 4,
|
52 |
+
"္": 36,
|
53 |
+
"်": 0,
|
54 |
+
"ျ": 22,
|
55 |
+
"ြ": 12,
|
56 |
+
"ွ": 24,
|
57 |
+
"ှ": 18,
|
58 |
+
"ဿ": 54
|
59 |
+
}
|