henilp105's picture
Upload tokenizer
d63fc58
raw
history blame
801 Bytes
{
"[PAD]": 61,
"[UNK]": 60,
"|": 0,
"ం": 1,
"అ": 2,
"ఆ": 3,
"ఇ": 4,
"ఈ": 5,
"ఉ": 6,
"ఊ": 7,
"ఎ": 8,
"ఏ": 9,
"ఐ": 10,
"ఒ": 11,
"ఓ": 12,
"క": 13,
"ఖ": 14,
"గ": 15,
"ఘ": 16,
"చ": 17,
"ఛ": 18,
"జ": 19,
"ఞ": 20,
"ట": 21,
"ఠ": 22,
"డ": 23,
"ఢ": 24,
"ణ": 25,
"త": 26,
"థ": 27,
"ద": 28,
"ధ": 29,
"న": 30,
"ప": 31,
"ఫ": 32,
"బ": 33,
"భ": 34,
"మ": 35,
"య": 36,
"ర": 37,
"ల": 38,
"ళ": 39,
"వ": 40,
"శ": 41,
"ష": 42,
"స": 43,
"హ": 44,
"ా": 45,
"ి": 46,
"ీ": 47,
"ు": 48,
"ూ": 49,
"ృ": 50,
"ె": 51,
"ే": 52,
"ై": 53,
"ొ": 54,
"ో": 55,
"ౌ": 56,
"్": 57,
"౩": 58,
"‌": 59
}