riri-np-small / vocab.json
roviso's picture
Upload tokenizer
16242d5
raw
history blame
No virus
1.54 kB
{
"npi": {
"!": 1,
"\"": 2,
"'": 3,
"(": 4,
")": 5,
",": 6,
"-": 7,
".": 8,
"/": 9,
":": 10,
";": 11,
"?": 12,
"M": 13,
"[PAD]": 103,
"[UNK]": 102,
"e": 14,
"o": 15,
"r": 16,
"|": 0,
"¥": 17,
"·": 18,
"÷": 19,
"ँ": 20,
"ं": 21,
"ः": 22,
"अ": 23,
"आ": 24,
"इ": 25,
"ई": 26,
"उ": 27,
"ऊ": 28,
"ऋ": 29,
"ए": 30,
"ऐ": 31,
"ओ": 32,
"औ": 33,
"क": 34,
"ख": 35,
"ग": 36,
"घ": 37,
"ङ": 38,
"च": 39,
"छ": 40,
"ज": 41,
"झ": 42,
"ञ": 43,
"ट": 44,
"ठ": 45,
"ड": 46,
"ढ": 47,
"ण": 48,
"त": 49,
"थ": 50,
"द": 51,
"ध": 52,
"न": 53,
"प": 54,
"फ": 55,
"ब": 56,
"भ": 57,
"म": 58,
"य": 59,
"र": 60,
"ल": 61,
"व": 62,
"श": 63,
"ष": 64,
"स": 65,
"ह": 66,
"़": 67,
"ा": 68,
"ि": 69,
"ी": 70,
"ु": 71,
"ू": 72,
"ृ": 73,
"े": 74,
"ै": 75,
"ो": 76,
"ौ": 77,
"्": 78,
"ॐ": 79,
"।": 80,
"०": 81,
"१": 82,
"२": 83,
"३": 84,
"४": 85,
"५": 86,
"६": 87,
"७": 88,
"८": 89,
"९": 90,
"​": 91,
"‌": 92,
"‍": 93,
"–": 94,
"—": 95,
"‘": 96,
"’": 97,
"“": 98,
"”": 99,
"…": 100,
" ": 101
}
}