Upload tokenizer
ca78015
verified
|
{ |
|
"fry": { |
|
"'": 23, |
|
"(": 32, |
|
")": 17, |
|
"-": 53, |
|
"F": 57, |
|
"L": 25, |
|
"N": 62, |
|
"Y": 28, |
|
"[": 43, |
|
"[EN]": 65, |
|
"[FY-NL]": 67, |
|
"[NL]": 66, |
|
"[PAD]": 64, |
|
"[UNK]": 63, |
|
"]": 56, |
|
"a": 5, |
|
"b": 31, |
|
"c": 16, |
|
"d": 3, |
|
"e": 2, |
|
"f": 19, |
|
"g": 4, |
|
"h": 39, |
|
"i": 21, |
|
"j": 1, |
|
"k": 41, |
|
"l": 11, |
|
"m": 47, |
|
"n": 38, |
|
"o": 60, |
|
"p": 24, |
|
"q": 54, |
|
"r": 26, |
|
"s": 61, |
|
"t": 18, |
|
"u": 20, |
|
"v": 46, |
|
"w": 55, |
|
"x": 13, |
|
"y": 48, |
|
"z": 51, |
|
"|": 58, |
|
"´": 49, |
|
"à": 7, |
|
"á": 37, |
|
"â": 0, |
|
"ä": 10, |
|
"è": 29, |
|
"é": 14, |
|
"ê": 34, |
|
"ë": 45, |
|
"í": 35, |
|
"ï": 36, |
|
"ó": 52, |
|
"ô": 27, |
|
"ö": 33, |
|
"ú": 12, |
|
"û": 40, |
|
"ü": 22, |
|
"ă": 59, |
|
"š": 15, |
|
"ţ": 50, |
|
"–": 8, |
|
"‘": 42, |
|
"’": 44, |
|
"“": 30, |
|
"”": 6, |
|
"…": 9 |
|
} |
|
} |
|
|