henilp105's picture
Upload tokenizer
4f8496e
raw
history blame
814 Bytes
{
"[PAD]": 62,
"[UNK]": 61,
"|": 0,
"ం": 1,
"ః": 2,
"అ": 3,
"ఆ": 4,
"ఇ": 5,
"ఈ": 6,
"ఉ": 7,
"ఊ": 8,
"ఎ": 9,
"ఏ": 10,
"ఐ": 11,
"ఒ": 12,
"ఓ": 13,
"క": 14,
"ఖ": 15,
"గ": 16,
"ఘ": 17,
"చ": 18,
"ఛ": 19,
"జ": 20,
"ఞ": 21,
"ట": 22,
"ఠ": 23,
"డ": 24,
"ఢ": 25,
"ణ": 26,
"త": 27,
"థ": 28,
"ద": 29,
"ధ": 30,
"న": 31,
"ప": 32,
"ఫ": 33,
"బ": 34,
"భ": 35,
"మ": 36,
"య": 37,
"ర": 38,
"ల": 39,
"ళ": 40,
"వ": 41,
"శ": 42,
"ష": 43,
"స": 44,
"హ": 45,
"ా": 46,
"ి": 47,
"ీ": 48,
"ు": 49,
"ూ": 50,
"ృ": 51,
"ె": 52,
"ే": 53,
"ై": 54,
"ొ": 55,
"ో": 56,
"ౌ": 57,
"్": 58,
"౩": 59,
"‌": 60
}