lparisi's picture
Upload tokenizer
a072212
raw
history blame
499 Bytes
{
"'": 13,
"[PAD]": 43,
"[UNK]": 42,
"a": 17,
"b": 24,
"c": 29,
"d": 5,
"e": 18,
"f": 40,
"g": 26,
"h": 11,
"i": 0,
"j": 2,
"k": 12,
"l": 21,
"m": 14,
"n": 22,
"o": 34,
"p": 38,
"q": 23,
"r": 41,
"s": 33,
"t": 8,
"u": 9,
"v": 36,
"w": 7,
"x": 19,
"y": 28,
"z": 10,
"|": 15,
"à": 35,
"á": 30,
"â": 39,
"ã": 20,
"ç": 1,
"é": 3,
"ê": 4,
"í": 37,
"ñ": 31,
"ó": 32,
"ô": 16,
"õ": 27,
"ú": 6,
"ü": 25
}