henilp105's picture
Upload tokenizer
ad5cf5b
raw
history blame
840 Bytes
{
"[PAD]": 64,
"[UNK]": 63,
"|": 0,
"ం": 1,
"ః": 2,
"అ": 3,
"ఆ": 4,
"ఇ": 5,
"ఈ": 6,
"ఉ": 7,
"ఊ": 8,
"ఋ": 9,
"ఎ": 10,
"ఏ": 11,
"ఐ": 12,
"ఒ": 13,
"ఓ": 14,
"ఔ": 15,
"క": 16,
"ఖ": 17,
"గ": 18,
"ఘ": 19,
"చ": 20,
"ఛ": 21,
"జ": 22,
"ఝ": 23,
"ఞ": 24,
"ట": 25,
"ఠ": 26,
"డ": 27,
"ఢ": 28,
"ణ": 29,
"త": 30,
"థ": 31,
"ద": 32,
"ధ": 33,
"న": 34,
"ప": 35,
"ఫ": 36,
"బ": 37,
"భ": 38,
"మ": 39,
"య": 40,
"ర": 41,
"ల": 42,
"ళ": 43,
"వ": 44,
"శ": 45,
"ష": 46,
"స": 47,
"హ": 48,
"ా": 49,
"ి": 50,
"ీ": 51,
"ు": 52,
"ూ": 53,
"ృ": 54,
"ె": 55,
"ే": 56,
"ై": 57,
"ొ": 58,
"ో": 59,
"ౌ": 60,
"్": 61,
"ౖ": 62
}