davidilag's picture
Upload tokenizer
ac32e6b verified
raw
history blame
462 Bytes
{
"'": 20,
"-": 28,
"[PAD]": 40,
"[UNK]": 39,
"a": 33,
"b": 22,
"c": 19,
"d": 29,
"e": 4,
"f": 21,
"g": 26,
"h": 36,
"i": 16,
"j": 31,
"k": 9,
"l": 5,
"m": 37,
"n": 13,
"o": 24,
"p": 32,
"q": 23,
"r": 10,
"s": 17,
"t": 6,
"u": 25,
"v": 14,
"w": 7,
"x": 0,
"y": 15,
"z": 18,
"|": 1,
"á": 34,
"æ": 35,
"é": 30,
"í": 8,
"ð": 38,
"ó": 3,
"ø": 11,
"ú": 12,
"ü": 27,
"ý": 2
}