Sajjo's picture
Upload tokenizer
c22c0d2 verified
raw
history blame
969 Bytes
{
"&": 1,
"[PAD]": 74,
"[UNK]": 73,
"|": 0,
"ँ": 3,
"ं": 4,
"ः": 5,
"अ": 6,
"आ": 7,
"इ": 8,
"ई": 9,
"उ": 10,
"ऊ": 11,
"ऋ": 12,
"ए": 13,
"ऐ": 14,
"ऑ": 15,
"ओ": 16,
"औ": 17,
"क": 18,
"ख": 19,
"ग": 20,
"घ": 21,
"च": 22,
"छ": 23,
"ज": 24,
"झ": 25,
"ञ": 26,
"ट": 27,
"ठ": 28,
"ड": 29,
"ढ": 30,
"ण": 31,
"त": 32,
"थ": 33,
"द": 34,
"ध": 35,
"न": 36,
"प": 37,
"फ": 38,
"ब": 39,
"भ": 40,
"म": 41,
"य": 42,
"र": 43,
"ल": 44,
"व": 45,
"श": 46,
"ष": 47,
"स": 48,
"ह": 49,
"़": 50,
"ा": 51,
"ि": 52,
"ी": 53,
"ु": 54,
"ू": 55,
"ृ": 56,
"ॅ": 57,
"े": 58,
"ै": 59,
"ॉ": 60,
"ो": 61,
"ौ": 62,
"्": 63,
"क़": 64,
"ख़": 65,
"ग़": 66,
"ज़": 67,
"ड़": 68,
"ढ़": 69,
"फ़": 70,
"।": 71,
"–": 72,
"’": 73
}