catalanModel / vocab.json
Trimux's picture
Upload tokenizer
eb03239 verified
raw
history blame
434 Bytes
{
" ": 34,
"'": 18,
"-": 21,
"_": 20,
"a": 23,
"b": 29,
"c": 35,
"d": 13,
"e": 22,
"f": 1,
"g": 3,
"h": 8,
"i": 2,
"j": 36,
"l": 9,
"m": 4,
"n": 37,
"o": 5,
"p": 16,
"q": 27,
"r": 6,
"s": 17,
"t": 28,
"u": 12,
"v": 10,
"x": 25,
"y": 38,
"z": 0,
"à": 11,
"ç": 15,
"è": 7,
"é": 19,
"í": 30,
"ï": 32,
"ò": 33,
"ó": 31,
"ú": 14,
"ü": 26,
"—": 24
}