Anujgr8's picture
Upload tokenizer
3214e7d verified
raw
history blame
827 Bytes
{
"[PAD]": 63,
"[UNK]": 62,
"|": 0,
"ँ": 1,
"ं": 2,
"अ": 3,
"आ": 4,
"इ": 5,
"ई": 6,
"उ": 7,
"ऊ": 8,
"ऋ": 9,
"ए": 10,
"ऐ": 11,
"ऑ": 12,
"ओ": 13,
"औ": 14,
"क": 15,
"ख": 16,
"ग": 17,
"घ": 18,
"च": 19,
"छ": 20,
"ज": 21,
"झ": 22,
"ञ": 23,
"ट": 24,
"ठ": 25,
"ड": 26,
"ढ": 27,
"ण": 28,
"त": 29,
"थ": 30,
"द": 31,
"ध": 32,
"न": 33,
"प": 34,
"फ": 35,
"ब": 36,
"भ": 37,
"म": 38,
"य": 39,
"र": 40,
"ल": 41,
"ळ": 42,
"व": 43,
"श": 44,
"ष": 45,
"स": 46,
"ह": 47,
"़": 48,
"ा": 49,
"ि": 50,
"ी": 51,
"ु": 52,
"ू": 53,
"ृ": 54,
"ॅ": 55,
"े": 56,
"ै": 57,
"ॉ": 58,
"ो": 59,
"ौ": 60,
"्": 61
}