Upload tokenizer
Browse files- vocab.json +70 -0
vocab.json
ADDED
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"&": 58,
|
3 |
+
"'": 51,
|
4 |
+
"(": 17,
|
5 |
+
")": 36,
|
6 |
+
"-": 6,
|
7 |
+
"E": 16,
|
8 |
+
"F": 4,
|
9 |
+
"L": 50,
|
10 |
+
"N": 44,
|
11 |
+
"Y": 39,
|
12 |
+
"[": 60,
|
13 |
+
"[EN]": 67,
|
14 |
+
"[FY-NL]": 66,
|
15 |
+
"[PAD]": 65,
|
16 |
+
"[UNK]": 64,
|
17 |
+
"]": 10,
|
18 |
+
"a": 26,
|
19 |
+
"b": 29,
|
20 |
+
"c": 55,
|
21 |
+
"d": 8,
|
22 |
+
"e": 42,
|
23 |
+
"f": 22,
|
24 |
+
"g": 62,
|
25 |
+
"h": 23,
|
26 |
+
"i": 59,
|
27 |
+
"j": 19,
|
28 |
+
"k": 40,
|
29 |
+
"l": 57,
|
30 |
+
"m": 11,
|
31 |
+
"n": 9,
|
32 |
+
"o": 34,
|
33 |
+
"p": 12,
|
34 |
+
"q": 25,
|
35 |
+
"r": 15,
|
36 |
+
"s": 48,
|
37 |
+
"t": 0,
|
38 |
+
"u": 2,
|
39 |
+
"v": 37,
|
40 |
+
"w": 31,
|
41 |
+
"x": 20,
|
42 |
+
"y": 53,
|
43 |
+
"z": 61,
|
44 |
+
"|": 28,
|
45 |
+
"à": 7,
|
46 |
+
"á": 32,
|
47 |
+
"â": 24,
|
48 |
+
"ã": 1,
|
49 |
+
"ä": 30,
|
50 |
+
"è": 52,
|
51 |
+
"é": 49,
|
52 |
+
"ê": 54,
|
53 |
+
"ë": 45,
|
54 |
+
"í": 47,
|
55 |
+
"ï": 38,
|
56 |
+
"ó": 3,
|
57 |
+
"ô": 33,
|
58 |
+
"ú": 41,
|
59 |
+
"û": 13,
|
60 |
+
"ü": 14,
|
61 |
+
"ō": 5,
|
62 |
+
"š": 18,
|
63 |
+
"–": 46,
|
64 |
+
"—": 63,
|
65 |
+
"‘": 21,
|
66 |
+
"’": 43,
|
67 |
+
"“": 56,
|
68 |
+
"”": 27,
|
69 |
+
"…": 35
|
70 |
+
}
|