Mardiyyah's picture
updating base pidgin model with correct tokenizer
7e324a5 verified
raw
history blame
499 Bytes
{
"'": 4,
"/": 1,
"0": 3,
"1": 34,
"2": 22,
"3": 41,
"4": 28,
"5": 12,
"6": 14,
"7": 31,
"8": 26,
"9": 39,
"@": 17,
"[PAD]": 44,
"[UNK]": 43,
"a": 23,
"b": 24,
"c": 16,
"d": 13,
"e": 2,
"f": 15,
"g": 0,
"h": 25,
"i": 20,
"j": 33,
"k": 32,
"l": 40,
"m": 37,
"n": 42,
"o": 21,
"p": 27,
"q": 8,
"r": 29,
"s": 36,
"t": 18,
"u": 38,
"v": 9,
"w": 30,
"x": 5,
"y": 35,
"z": 10,
"|": 19,
"ç": 6,
"è": 11,
"é": 7
}