riri-small / vocab.json
roviso's picture
Upload tokenizer
da567a3
raw
history blame
1.39 kB
{
"npi": {
" ": 0,
"!": 1,
"\"": 2,
"'": 3,
"(": 4,
")": 5,
",": 6,
"-": 7,
".": 8,
"/": 9,
":": 10,
";": 11,
"?": 12,
"[PAD]": 93,
"[UNK]": 92,
"|": 0,
"¥": 13,
"ँ": 14,
"ं": 15,
"ः": 16,
"अ": 17,
"आ": 18,
"इ": 19,
"ई": 20,
"उ": 21,
"ऊ": 22,
"ऋ": 23,
"ए": 24,
"ऐ": 25,
"ओ": 26,
"औ": 27,
"क": 28,
"ख": 29,
"ग": 30,
"घ": 31,
"ङ": 32,
"च": 33,
"छ": 34,
"ज": 35,
"झ": 36,
"ञ": 37,
"ट": 38,
"ठ": 39,
"ड": 40,
"ढ": 41,
"ण": 42,
"त": 43,
"थ": 44,
"द": 45,
"ध": 46,
"न": 47,
"प": 48,
"फ": 49,
"ब": 50,
"भ": 51,
"म": 52,
"य": 53,
"र": 54,
"ल": 55,
"व": 56,
"श": 57,
"ष": 58,
"स": 59,
"ह": 60,
"ा": 61,
"ि": 62,
"ी": 63,
"ु": 64,
"ू": 65,
"ृ": 66,
"े": 67,
"ै": 68,
"ो": 69,
"ौ": 70,
"्": 71,
"।": 72,
"०": 73,
"१": 74,
"२": 75,
"३": 76,
"४": 77,
"५": 78,
"६": 79,
"७": 80,
"८": 81,
"९": 82,
"​": 83,
"‌": 84,
"‍": 85,
"–": 86,
"‘": 87,
"’": 88,
"“": 89,
"”": 90
}
}