hindi_turkish / vocab.json
Chahat7874's picture
Upload tokenizer
53e1390 verified
raw
history blame
840 Bytes
{
"[PAD]": 64,
"[UNK]": 63,
"|": 0,
"ं": 1,
"अ": 2,
"आ": 3,
"ई": 4,
"ओ": 5,
"क": 6,
"ग": 7,
"च": 8,
"ज": 9,
"ड": 10,
"त": 11,
"थ": 12,
"द": 13,
"ध": 14,
"न": 15,
"प": 16,
"फ": 17,
"ब": 18,
"भ": 19,
"म": 20,
"य": 21,
"र": 22,
"ल": 23,
"व": 24,
"श": 25,
"ष": 26,
"स": 27,
"ह": 28,
"़": 29,
"ा": 30,
"ि": 31,
"ी": 32,
"ु": 33,
"ू": 34,
"े": 35,
"ै": 36,
"ॉ": 37,
"ो": 38,
"्": 39,
"ం": 40,
"ఎ": 41,
"క": 42,
"చ": 43,
"ట": 44,
"డ": 45,
"త": 46,
"ద": 47,
"న": 48,
"ప": 49,
"మ": 50,
"ర": 51,
"ల": 52,
"వ": 53,
"శ": 54,
"ష": 55,
"ా": 56,
"ి": 57,
"ు": 58,
"ే": 59,
"ొ": 60,
"ో": 61,
"్": 62
}