HINDI_MMS_DATA / vocab.json
FariqF's picture
Upload tokenizer
e1ec8b7 verified
raw
history blame contribute delete
907 Bytes
{
" ": 7,
"'": 44,
"-": 3,
"0": 8,
"1": 2,
"2": 18,
"3": 25,
"4": 1,
"8": 6,
"_": 33,
"`": 45,
"ँ": 16,
"ं": 30,
"ः": 35,
"अ": 4,
"आ": 55,
"इ": 5,
"ई": 34,
"उ": 32,
"ऊ": 65,
"ए": 11,
"ऐ": 12,
"ऑ": 19,
"ओ": 67,
"औ": 60,
"क": 70,
"ख": 42,
"ग": 49,
"घ": 47,
"च": 17,
"छ": 9,
"ज": 57,
"झ": 58,
"ञ": 26,
"ट": 68,
"ठ": 29,
"ड": 56,
"ढ": 50,
"ण": 43,
"त": 69,
"थ": 20,
"द": 62,
"ध": 37,
"न": 10,
"प": 48,
"फ": 0,
"ब": 41,
"भ": 21,
"म": 71,
"य": 51,
"र": 39,
"ल": 38,
"व": 46,
"श": 59,
"ष": 14,
"स": 40,
"ह": 36,
"़": 13,
"ा": 54,
"ि": 15,
"ी": 22,
"ु": 28,
"ू": 66,
"ृ": 63,
"ॅ": 24,
"े": 52,
"ै": 27,
"ॉ": 31,
"ो": 61,
"ौ": 64,
"्": 53,
"‍": 23
}