mms_por_finetuning / vocab.json
lemorim's picture
Upload tokenizer
f75e6b4 verified
raw
history blame
480 Bytes
{
" ": 32,
"'": 11,
"-": 19,
"1": 2,
"2": 5,
"4": 16,
"_": 13,
"a": 38,
"b": 31,
"c": 22,
"d": 40,
"e": 27,
"f": 30,
"g": 20,
"h": 6,
"i": 29,
"j": 39,
"l": 4,
"m": 18,
"n": 35,
"o": 28,
"p": 8,
"q": 21,
"r": 33,
"s": 14,
"t": 26,
"u": 3,
"v": 17,
"x": 10,
"z": 23,
"à": 0,
"á": 42,
"â": 24,
"ã": 9,
"ç": 15,
"é": 7,
"ê": 12,
"í": 25,
"ó": 37,
"ô": 34,
"õ": 41,
"ú": 1,
"—": 36
}