sanchit-gandhi's picture
add tokenizer
e5eaaa7
raw
history blame
380 Bytes
{
"'": 14,
"<": 27,
"</s>": 2,
"<pad>": 0,
"<s>": 1,
"<unk>": 3,
">": 20,
"a": 16,
"b": 22,
"c": 15,
"d": 6,
"e": 11,
"f": 25,
"g": 31,
"h": 13,
"i": 29,
"j": 18,
"k": 17,
"l": 26,
"m": 5,
"n": 23,
"o": 19,
"p": 33,
"q": 7,
"r": 8,
"s": 28,
"t": 21,
"u": 30,
"v": 9,
"w": 12,
"x": 4,
"y": 24,
"z": 32,
"|": 10
}