Upload tokenizer
385e4db
verified
|
|
{ |
|
|
"'": 3, |
|
|
"-": 43, |
|
|
"F": 50, |
|
|
"L": 19, |
|
|
"N": 11, |
|
|
"Y": 23, |
|
|
"[": 24, |
|
|
"[FY-NL]": 56, |
|
|
"[NL]": 57, |
|
|
"[PAD]": 55, |
|
|
"[UNK]": 54, |
|
|
"]": 12, |
|
|
"a": 0, |
|
|
"b": 48, |
|
|
"c": 53, |
|
|
"d": 34, |
|
|
"e": 27, |
|
|
"f": 46, |
|
|
"g": 2, |
|
|
"h": 22, |
|
|
"i": 31, |
|
|
"j": 26, |
|
|
"k": 7, |
|
|
"l": 6, |
|
|
"m": 44, |
|
|
"n": 5, |
|
|
"o": 39, |
|
|
"p": 51, |
|
|
"q": 18, |
|
|
"r": 28, |
|
|
"s": 35, |
|
|
"t": 42, |
|
|
"u": 32, |
|
|
"v": 8, |
|
|
"w": 9, |
|
|
"x": 33, |
|
|
"y": 30, |
|
|
"z": 49, |
|
|
"|": 13, |
|
|
"à": 4, |
|
|
"á": 14, |
|
|
"â": 36, |
|
|
"ä": 17, |
|
|
"è": 15, |
|
|
"é": 47, |
|
|
"ê": 38, |
|
|
"ë": 37, |
|
|
"ï": 41, |
|
|
"ó": 21, |
|
|
"ô": 1, |
|
|
"ö": 25, |
|
|
"ú": 20, |
|
|
"û": 29, |
|
|
"ü": 40, |
|
|
"–": 52, |
|
|
"’": 16, |
|
|
"“": 10, |
|
|
"”": 45 |
|
|
} |
|
|
|