shahukareem's picture
Upload tokenizer
c81137c
raw
history blame contribute delete
648 Bytes
{
"[PAD]": 53,
"[UNK]": 52,
"|": 0,
"ّ": 1,
"ހ": 2,
"ށ": 3,
"ނ": 4,
"ރ": 5,
"ބ": 6,
"ޅ": 7,
"ކ": 8,
"އ": 9,
"ވ": 10,
"މ": 11,
"ފ": 12,
"ދ": 13,
"ތ": 14,
"ލ": 15,
"ގ": 16,
"ޏ": 17,
"ސ": 18,
"ޑ": 19,
"ޒ": 20,
"ޓ": 21,
"ޔ": 22,
"ޕ": 23,
"ޖ": 24,
"ޗ": 25,
"ޘ": 26,
"ޙ": 27,
"ޚ": 28,
"ޛ": 29,
"ޝ": 30,
"ޞ": 31,
"ޟ": 32,
"ޠ": 33,
"ޡ": 34,
"ޢ": 35,
"ޣ": 36,
"ޤ": 37,
"ޥ": 38,
"ަ": 39,
"ާ": 40,
"ި": 41,
"ީ": 42,
"ު": 43,
"ޫ": 44,
"ެ": 45,
"ޭ": 46,
"ޮ": 47,
"ޯ": 48,
"ް": 49,
"ﷲ": 50,
"ﷺ": 51
}