henilp105's picture
Upload tokenizer
2432e00
raw
history blame
849 Bytes
{
"0": 1,
"2": 2,
"[PAD]": 65,
"[UNK]": 64,
"|": 0,
"ఁ": 3,
"ం": 4,
"ః": 5,
"అ": 6,
"ఆ": 7,
"ఇ": 8,
"ఈ": 9,
"ఉ": 10,
"ఊ": 11,
"ఎ": 12,
"ఏ": 13,
"ఐ": 14,
"ఒ": 15,
"ఓ": 16,
"క": 17,
"ఖ": 18,
"గ": 19,
"ఘ": 20,
"చ": 21,
"ఛ": 22,
"జ": 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,
"౩": 62,
"‌": 63
}