cdactvm's picture
Upload tokenizer
0350600 verified
raw
history blame
1.02 kB
{
"/": 1,
"0": 2,
"6": 3,
"8": 4,
"9": 5,
"I": 6,
"[PAD]": 79,
"[UNK]": 78,
"_": 7,
"|": 0,
"ಂ": 8,
"ಃ": 9,
"ಅ": 10,
"ಆ": 11,
"ಇ": 12,
"ಈ": 13,
"ಉ": 14,
"ಊ": 15,
"ಋ": 16,
"ಎ": 17,
"ಏ": 18,
"ಐ": 19,
"ಒ": 20,
"ಓ": 21,
"ಔ": 22,
"ಕ": 23,
"ಖ": 24,
"ಗ": 25,
"ಘ": 26,
"ಙ": 27,
"ಚ": 28,
"ಛ": 29,
"ಜ": 30,
"ಝ": 31,
"ಞ": 32,
"ಟ": 33,
"ಠ": 34,
"ಡ": 35,
"ಢ": 36,
"ಣ": 37,
"ತ": 38,
"ಥ": 39,
"ದ": 40,
"ಧ": 41,
"ನ": 42,
"ಪ": 43,
"ಫ": 44,
"ಬ": 45,
"ಭ": 46,
"ಮ": 47,
"ಯ": 48,
"ರ": 49,
"ಱ": 50,
"ಲ": 51,
"ಳ": 52,
"ವ": 53,
"ಶ": 54,
"ಷ": 55,
"ಸ": 56,
"ಹ": 57,
"ಾ": 58,
"ಿ": 59,
"ೀ": 60,
"ು": 61,
"ೂ": 62,
"ೃ": 63,
"ೆ": 64,
"ೇ": 65,
"ೈ": 66,
"ೊ": 67,
"ೋ": 68,
"ೌ": 69,
"್": 70,
"ೕ": 71,
"ೖ": 72,
"೦": 73,
"೨": 74,
"೪": 75,
"​": 76,
"‍": 77
}