aminnaghavi's picture
add tokenizer
97d3d0c
raw
history blame
331 Bytes
{
"'": 27,
"[PAD]": 29,
"[UNK]": 28,
"a": 0,
"b": 13,
"c": 5,
"d": 15,
"e": 25,
"f": 10,
"g": 2,
"h": 11,
"i": 4,
"j": 26,
"k": 1,
"l": 6,
"m": 7,
"n": 17,
"o": 14,
"p": 16,
"q": 18,
"r": 24,
"s": 21,
"t": 19,
"u": 20,
"v": 22,
"w": 3,
"x": 23,
"y": 12,
"z": 9,
"|": 8
}