scratch / vocab.json
yip-i's picture
Upload tokenizer
c5310b3
raw
history blame contribute delete
331 Bytes
{
"'": 27,
"[PAD]": 29,
"[UNK]": 28,
"a": 18,
"b": 5,
"c": 10,
"d": 4,
"e": 12,
"f": 23,
"g": 0,
"h": 21,
"i": 24,
"j": 17,
"k": 16,
"l": 9,
"m": 3,
"n": 11,
"o": 25,
"p": 19,
"q": 22,
"r": 26,
"s": 8,
"t": 6,
"u": 7,
"v": 20,
"w": 14,
"x": 2,
"y": 13,
"z": 15,
"|": 1
}