aaniket's picture
Upload tokenizer
3a69fc1
raw
history blame
935 Bytes
{
"'": 11,
"[PAD]": 73,
"[UNK]": 72,
"a": 42,
"e": 41,
"f": 21,
"i": 35,
"l": 65,
"m": 23,
"p": 53,
"r": 39,
"u": 49,
"w": 10,
"|": 0,
"ँ": 69,
"ं": 5,
"ः": 45,
"अ": 48,
"आ": 18,
"इ": 34,
"ई": 68,
"उ": 9,
"ऊ": 22,
"ए": 24,
"ऐ": 58,
"ऑ": 15,
"ओ": 70,
"औ": 8,
"क": 26,
"ख": 28,
"ग": 12,
"घ": 66,
"च": 29,
"छ": 59,
"ज": 64,
"झ": 56,
"ट": 17,
"ठ": 3,
"ड": 16,
"ढ": 4,
"ण": 60,
"त": 6,
"थ": 13,
"द": 32,
"ध": 2,
"न": 71,
"प": 55,
"फ": 1,
"ब": 19,
"भ": 50,
"म": 20,
"य": 62,
"र": 47,
"ल": 51,
"व": 44,
"श": 54,
"ष": 36,
"स": 57,
"ह": 30,
"़": 27,
"ा": 52,
"ि": 38,
"ी": 67,
"ु": 43,
"ू": 63,
"ृ": 61,
"े": 37,
"ै": 14,
"ॉ": 25,
"ो": 7,
"ौ": 40,
"्": 31,
"ड़": 33,
"।": 46
}