tachiwin_tutunaku / vocab.json
ljcamargo's picture
Upload tokenizer
3d049ae
raw
history blame
415 Bytes
{
"'": 23,
"[PAD]": 36,
"[UNK]": 35,
"a": 30,
"b": 34,
"c": 25,
"d": 5,
"e": 26,
"f": 19,
"g": 13,
"h": 12,
"i": 22,
"j": 17,
"k": 32,
"l": 27,
"m": 33,
"n": 21,
"o": 10,
"p": 0,
"q": 7,
"r": 24,
"s": 14,
"t": 11,
"u": 20,
"v": 18,
"w": 4,
"x": 3,
"y": 15,
"z": 8,
"|": 29,
"á": 9,
"é": 1,
"í": 2,
"ñ": 6,
"ó": 28,
"ú": 31,
"ü": 16
}