zoha's picture
add tokenizer
038226f
raw
history blame contribute delete
331 Bytes
{
"'": 10,
"[PAD]": 29,
"[UNK]": 28,
"a": 26,
"b": 27,
"c": 15,
"d": 12,
"e": 17,
"f": 20,
"g": 21,
"h": 7,
"i": 24,
"j": 2,
"k": 18,
"l": 25,
"m": 1,
"n": 0,
"o": 5,
"p": 16,
"q": 3,
"r": 13,
"s": 4,
"t": 19,
"u": 14,
"v": 11,
"w": 6,
"x": 23,
"y": 22,
"z": 9,
"|": 8
}