samuelchiji's picture
Upload tokenizer
60e316f verified
raw
history blame
357 Bytes
{
"2": 29,
"[PAD]": 31,
"[UNK]": 30,
"a": 0,
"b": 14,
"c": 12,
"d": 7,
"e": 27,
"f": 5,
"g": 19,
"h": 10,
"i": 9,
"j": 16,
"k": 21,
"l": 20,
"m": 6,
"n": 13,
"o": 25,
"p": 4,
"q": 24,
"r": 8,
"s": 28,
"t": 17,
"u": 23,
"v": 3,
"w": 15,
"x": 11,
"y": 2,
"z": 22,
"|": 18,
"ẹ": 26,
"’": 1
}