Upload tokenizer
Browse files- special_tokens_map.json +14 -2
- vocab.json +11 -12
special_tokens_map.json
CHANGED
@@ -1,4 +1,16 @@
|
|
1 |
{
|
2 |
-
"pad_token":
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4 |
}
|
|
|
1 |
{
|
2 |
+
"pad_token": {
|
3 |
+
"content": "a",
|
4 |
+
"lstrip": false,
|
5 |
+
"normalized": false,
|
6 |
+
"rstrip": false,
|
7 |
+
"single_word": false
|
8 |
+
},
|
9 |
+
"unk_token": {
|
10 |
+
"content": "<unk>",
|
11 |
+
"lstrip": false,
|
12 |
+
"normalized": false,
|
13 |
+
"rstrip": false,
|
14 |
+
"single_word": false
|
15 |
+
}
|
16 |
}
|
vocab.json
CHANGED
@@ -9,44 +9,43 @@
|
|
9 |
"c": 7,
|
10 |
"d": 25,
|
11 |
"e": 23,
|
|
|
12 |
"f": 31,
|
13 |
"g": 29,
|
14 |
"h": 8,
|
15 |
"i": 3,
|
|
|
16 |
"j": 20,
|
17 |
"k": 12,
|
18 |
"l": 9,
|
19 |
"m": 11,
|
20 |
"n": 5,
|
21 |
"o": 17,
|
|
|
22 |
"p": 13,
|
23 |
"q": 18,
|
24 |
"r": 26,
|
25 |
"s": 15,
|
26 |
"t": 6,
|
27 |
"u": 4,
|
|
|
28 |
"v": 32,
|
|
|
29 |
"x": 10,
|
30 |
"y": 16,
|
31 |
"z": 22,
|
32 |
"|": 2,
|
33 |
"á": 14,
|
|
|
34 |
"é": 30,
|
35 |
"ì": 37,
|
36 |
-
"í":
|
|
|
37 |
"ï": 36,
|
38 |
"ñ": 34,
|
39 |
"ó": 27,
|
|
|
40 |
"ú": 21,
|
41 |
-
"
|
42 |
-
"
|
43 |
-
"e:": 41,
|
44 |
-
"á:": 42,
|
45 |
-
"i:": 43,
|
46 |
-
"u:": 44,
|
47 |
-
"í:": 45,
|
48 |
-
"í": 46,
|
49 |
-
"ú:": 47,
|
50 |
-
"ó:": 48,
|
51 |
-
"o:": 49
|
52 |
}
|
|
|
9 |
"c": 7,
|
10 |
"d": 25,
|
11 |
"e": 23,
|
12 |
+
"e:": 41,
|
13 |
"f": 31,
|
14 |
"g": 29,
|
15 |
"h": 8,
|
16 |
"i": 3,
|
17 |
+
"i:": 43,
|
18 |
"j": 20,
|
19 |
"k": 12,
|
20 |
"l": 9,
|
21 |
"m": 11,
|
22 |
"n": 5,
|
23 |
"o": 17,
|
24 |
+
"o:": 49,
|
25 |
"p": 13,
|
26 |
"q": 18,
|
27 |
"r": 26,
|
28 |
"s": 15,
|
29 |
"t": 6,
|
30 |
"u": 4,
|
31 |
+
"u:": 44,
|
32 |
"v": 32,
|
33 |
+
"w": 40,
|
34 |
"x": 10,
|
35 |
"y": 16,
|
36 |
"z": 22,
|
37 |
"|": 2,
|
38 |
"á": 14,
|
39 |
+
"á:": 42,
|
40 |
"é": 30,
|
41 |
"ì": 37,
|
42 |
+
"í": 46,
|
43 |
+
"í:": 45,
|
44 |
"ï": 36,
|
45 |
"ñ": 34,
|
46 |
"ó": 27,
|
47 |
+
"ó:": 48,
|
48 |
"ú": 21,
|
49 |
+
"ú:": 47,
|
50 |
+
"̱": 1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
51 |
}
|