xacl's picture
Upload tokenizer
fc49a2b
raw
history blame
331 Bytes
{
"'": 10,
"[PAD]": 29,
"[UNK]": 28,
"a": 12,
"b": 7,
"c": 9,
"d": 21,
"e": 11,
"f": 22,
"g": 15,
"h": 5,
"i": 0,
"j": 6,
"k": 8,
"l": 27,
"m": 25,
"n": 17,
"o": 20,
"p": 1,
"q": 2,
"r": 26,
"s": 23,
"t": 24,
"u": 14,
"v": 16,
"w": 13,
"x": 3,
"y": 18,
"z": 19,
"|": 4
}