nrshoudi's picture
Upload tokenizer
a3a9178
raw
history blame
454 Bytes
{
"<PAD>": 37,
"<UNK>": 36,
"|": 0,
"آ": 1,
"أ": 2,
"إ": 3,
"ا": 4,
"ب": 5,
"ة": 6,
"ت": 7,
"ج": 8,
"ح": 9,
"خ": 10,
"د": 11,
"ذ": 12,
"ر": 13,
"س": 14,
"ش": 15,
"ص": 16,
"ض": 17,
"ط": 18,
"ظ": 19,
"ع": 20,
"غ": 21,
"ف": 22,
"ق": 23,
"ك": 24,
"ل": 25,
"م": 26,
"ن": 27,
"ه": 28,
"و": 29,
"ى": 30,
"ي": 31,
"ً": 32,
"ُ": 33,
"ِ": 34,
"ّ": 35
}