crossdelenna's picture
add tokenizer
aa39d51
raw
history blame
331 Bytes
{
"'": 27,
"[PAD]": 29,
"[UNK]": 28,
"a": 26,
"b": 12,
"c": 16,
"d": 2,
"e": 23,
"f": 6,
"g": 21,
"h": 17,
"i": 25,
"j": 15,
"k": 8,
"l": 3,
"m": 18,
"n": 10,
"o": 5,
"p": 24,
"q": 0,
"r": 22,
"s": 13,
"t": 1,
"u": 4,
"v": 9,
"w": 11,
"x": 14,
"y": 19,
"z": 20,
"|": 7
}