ayertey01's picture
Upload tokenizer
6090b55
raw
history blame
394 Bytes
{
"[PAD]": 34,
"[UNK]": 33,
"a": 7,
"b": 8,
"c": 12,
"d": 18,
"e": 20,
"f": 26,
"g": 13,
"h": 14,
"i": 30,
"j": 15,
"k": 3,
"l": 16,
"m": 29,
"n": 2,
"o": 21,
"p": 25,
"r": 6,
"s": 1,
"t": 32,
"u": 0,
"v": 10,
"w": 17,
"y": 31,
"z": 9,
"|": 22,
"á": 28,
"ã": 27,
"õ": 11,
"ĩ": 5,
"ũ": 23,
"ɔ": 4,
"ɛ": 19,
"́": 24
}