samuelchiji's picture
Upload tokenizer
55f9c1f verified
raw
history blame
357 Bytes
{
"2": 20,
"[PAD]": 31,
"[UNK]": 30,
"a": 21,
"b": 11,
"c": 15,
"d": 18,
"e": 12,
"f": 3,
"g": 4,
"h": 26,
"i": 23,
"j": 5,
"k": 19,
"l": 29,
"m": 13,
"n": 14,
"o": 17,
"p": 25,
"q": 0,
"r": 10,
"s": 9,
"t": 1,
"u": 16,
"v": 8,
"w": 6,
"x": 27,
"y": 2,
"z": 28,
"|": 7,
"ẹ": 24,
"’": 22
}