Anujgr8's picture
Upload tokenizer
2e69138 verified
raw
history blame contribute delete
989 Bytes
{
"[PAD]": 76,
"[UNK]": 75,
"a": 1,
"h": 2,
"t": 3,
"|": 0,
" ": 4,
"ँ": 5,
"ं": 6,
"ः": 7,
"अ": 8,
"आ": 9,
"इ": 10,
"ई": 11,
"उ": 12,
"ऊ": 13,
"ऋ": 14,
"ऍ": 15,
"ए": 16,
"ऐ": 17,
"ऑ": 18,
"ओ": 19,
"औ": 20,
"क": 21,
"ख": 22,
"ग": 23,
"घ": 24,
"च": 25,
"छ": 26,
"ज": 27,
"झ": 28,
"ञ": 29,
"ट": 30,
"ठ": 31,
"ड": 32,
"ढ": 33,
"ण": 34,
"त": 35,
"थ": 36,
"द": 37,
"ध": 38,
"न": 39,
"प": 40,
"फ": 41,
"ब": 42,
"भ": 43,
"म": 44,
"य": 45,
"र": 46,
"ऱ": 47,
"ल": 48,
"ळ": 49,
"व": 50,
"श": 51,
"ष": 52,
"स": 53,
"ह": 54,
"़": 55,
"ा": 56,
"ि": 57,
"ी": 58,
"ु": 59,
"ू": 60,
"ृ": 61,
"े": 62,
"ै": 63,
"ॉ": 64,
"ो": 65,
"ौ": 66,
"्": 67,
"ख़": 68,
"ज़": 69,
"ड़": 70,
"ढ़": 71,
"फ़": 72,
"२": 73,
"‌": 74
}