ayertey01's picture
Upload tokenizer
f2c520c
raw
history blame
394 Bytes
{
"[PAD]": 34,
"[UNK]": 33,
"a": 18,
"b": 25,
"c": 28,
"d": 17,
"e": 6,
"f": 16,
"g": 23,
"h": 14,
"i": 10,
"j": 13,
"k": 0,
"l": 1,
"m": 5,
"n": 9,
"o": 15,
"p": 2,
"r": 24,
"s": 12,
"t": 31,
"u": 7,
"v": 32,
"w": 4,
"y": 29,
"z": 22,
"|": 11,
"á": 19,
"ã": 21,
"õ": 8,
"ĩ": 26,
"ũ": 20,
"ɔ": 3,
"ɛ": 27,
"́": 30
}