tamasheq-98-2 / vocab.json
ad019el's picture
add tokenizer
813926d
raw
history blame
746 Bytes
{
"/": 28,
"0": 12,
"1": 26,
"2": 39,
"7": 6,
"9": 34,
"[PAD]": 63,
"[UNK]": 62,
"_": 52,
"a": 36,
"c": 20,
"e": 8,
"f": 56,
"j": 45,
"l": 42,
"m": 48,
"o": 23,
"q": 49,
"r": 1,
"t": 47,
"u": 33,
"w": 54,
"|": 10,
"ء": 38,
"آ": 29,
"أ": 22,
"ؤ": 44,
"إ": 24,
"ئ": 9,
"ا": 7,
"ب": 30,
"ة": 27,
"ت": 37,
"ث": 50,
"ج": 35,
"ح": 46,
"خ": 19,
"د": 5,
"ذ": 14,
"ر": 41,
"ز": 15,
"س": 13,
"ش": 0,
"ص": 55,
"ض": 16,
"ط": 11,
"ظ": 53,
"ع": 58,
"غ": 59,
"ف": 21,
"ق": 25,
"ك": 60,
"ل": 4,
"م": 51,
"ن": 32,
"ه": 31,
"و": 61,
"ى": 3,
"ي": 43,
"ٱ": 18,
"چ": 40,
"ڤ": 2,
"ک": 57,
"ی": 17
}