aminnaghavi's picture
add tokenizer
858c571
raw
history blame
331 Bytes
{
"'": 7,
"[PAD]": 30,
"[UNK]": 30,
"a": 12,
"b": 8,
"c": 2,
"d": 10,
"e": 18,
"f": 11,
"g": 27,
"h": 22,
"i": 17,
"j": 19,
"k": 24,
"l": 20,
"m": 0,
"n": 9,
"o": 16,
"p": 1,
"q": 4,
"r": 6,
"s": 25,
"t": 14,
"u": 5,
"v": 15,
"w": 26,
"x": 23,
"y": 21,
"z": 3,
"|": 13
}