hindi_turkish / vocab.json
Chahat7874's picture
Upload tokenizer
a18d6ea verified
raw
history blame
879 Bytes
{
"[PAD]": 67,
"[UNK]": 66,
"|": 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,
"़": 28,
"ा": 29,
"ि": 30,
"ी": 31,
"ु": 32,
"ू": 33,
"े": 34,
"ै": 35,
"ॉ": 36,
"ो": 37,
"्": 38,
"ం": 39,
"అ": 40,
"ఎ": 41,
"క": 42,
"గ": 43,
"చ": 44,
"ట": 45,
"డ": 46,
"త": 47,
"ద": 48,
"న": 49,
"ప": 50,
"బ": 51,
"మ": 52,
"ర": 53,
"ల": 54,
"వ": 55,
"శ": 56,
"ష": 57,
"స": 58,
"ా": 59,
"ి": 60,
"ు": 61,
"ే": 62,
"ొ": 63,
"ో": 64,
"్": 65
}