cdactvm's picture
Upload tokenizer
c221eeb verified
raw
history blame contribute delete
944 Bytes
{
"[PAD]": 72,
"[UNK]": 71,
"|": 0,
"ँ": 1,
"ं": 2,
"ः": 3,
"अ": 4,
"आ": 5,
"इ": 6,
"ई": 7,
"उ": 8,
"ऊ": 9,
"ऋ": 10,
"ऍ": 11,
"ए": 12,
"ऐ": 13,
"ऑ": 14,
"ओ": 15,
"औ": 16,
"क": 17,
"ख": 18,
"ग": 19,
"घ": 20,
"च": 21,
"छ": 22,
"ज": 23,
"झ": 24,
"ञ": 25,
"ट": 26,
"ठ": 27,
"ड": 28,
"ढ": 29,
"ण": 30,
"त": 31,
"थ": 32,
"द": 33,
"ध": 34,
"न": 35,
"प": 36,
"फ": 37,
"ब": 38,
"भ": 39,
"म": 40,
"य": 41,
"र": 42,
"ल": 43,
"व": 44,
"श": 45,
"ष": 46,
"स": 47,
"ह": 48,
"़": 49,
"ा": 50,
"ि": 51,
"ी": 52,
"ु": 53,
"ू": 54,
"ृ": 55,
"ॅ": 56,
"े": 57,
"ै": 58,
"ॉ": 59,
"ो": 60,
"ौ": 61,
"्": 62,
"क़": 63,
"ख़": 64,
"ग़": 65,
"ज़": 66,
"ड़": 67,
"ढ़": 68,
"फ़": 69,
"ॠ": 70
}