aminnaghavi's picture
add tokenizer
4c205ca
raw
history blame contribute delete
331 Bytes
{
"'": 24,
"[PAD]": 29,
"[UNK]": 28,
"a": 20,
"b": 1,
"c": 9,
"d": 26,
"e": 27,
"f": 0,
"g": 3,
"h": 4,
"i": 5,
"j": 15,
"k": 18,
"l": 6,
"m": 12,
"n": 21,
"o": 14,
"p": 10,
"q": 19,
"r": 8,
"s": 2,
"t": 17,
"u": 23,
"v": 16,
"w": 25,
"x": 7,
"y": 13,
"z": 22,
"|": 11
}