jrs-a's picture
Upload tokenizer
f355ea7
raw
history blame
287 Bytes
{
"[PAD]": 25,
"[UNK]": 24,
"a": 5,
"b": 0,
"c": 1,
"d": 12,
"e": 9,
"f": 14,
"g": 22,
"h": 17,
"i": 18,
"k": 10,
"l": 13,
"m": 6,
"n": 19,
"o": 7,
"p": 2,
"r": 11,
"s": 4,
"t": 16,
"u": 21,
"v": 3,
"w": 23,
"x": 20,
"y": 15,
"|": 8
}