paruwka's picture
Upload tokenizer
d519b28 verified
raw
history blame
750 Bytes
{
"&": 42,
"'": 4,
"(": 32,
")": 40,
"=": 19,
"[PAD]": 64,
"[UNK]": 63,
"a": 28,
"b": 18,
"c": 49,
"d": 54,
"e": 60,
"f": 36,
"g": 58,
"h": 11,
"i": 31,
"j": 2,
"k": 6,
"l": 14,
"m": 41,
"n": 22,
"o": 30,
"p": 35,
"q": 56,
"r": 53,
"s": 39,
"t": 43,
"u": 24,
"v": 20,
"w": 0,
"x": 50,
"y": 46,
"z": 55,
"|": 23,
"«": 33,
"»": 37,
"ß": 12,
"à": 44,
"á": 27,
"â": 61,
"ç": 62,
"è": 5,
"é": 17,
"ê": 26,
"ë": 34,
"î": 59,
"ï": 25,
"ó": 45,
"ô": 13,
"ö": 1,
"ù": 52,
"ú": 57,
"û": 48,
"ü": 9,
"ć": 47,
"ō": 15,
"œ": 21,
"ū": 16,
"̀": 8,
"́": 10,
"̂": 7,
"̧": 51,
"—": 3,
"’": 29,
"…": 38
}