trying-mms-finetuning / vocab.json
bishaltwr's picture
Upload tokenizer
cc1a5a3 verified
{
"npi": {
"!": 1,
"\"": 2,
"%": 3,
"'": 4,
"(": 5,
")": 6,
",": 7,
"-": 8,
".": 9,
"/": 10,
":": 11,
";": 12,
"?": 13,
"[PAD]": 85,
"[UNK]": 84,
"|": 0,
"ँ": 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,
"ऱ": 55,
"ल": 56,
"व": 57,
"श": 58,
"ष": 59,
"स": 60,
"ह": 61,
"ा": 62,
"ि": 63,
"ी": 64,
"ु": 65,
"ू": 66,
"ृ": 67,
"े": 68,
"ै": 69,
"ो": 70,
"ौ": 71,
"्": 72,
"ॠ": 73,
"।": 74,
"८": 75,
"‌": 76,
"‍": 77,
"–": 78,
"—": 79,
"‘": 80,
"’": 81,
"“": 82,
"”": 83
}
}