IEEE 754-2008 contains a half precision that is only 16 bits wide. The leftmost bit is still the sign bit, the exponent is 5 bits wide and has a bias of 15, and the mantissa is 10 bits long. A hidden 1 is assumed.

Can I get some explanation and work show?

convert 2.6875*10^1, and 1.45751953125*10^-1 then add those two using16 bit representation. assuming 1 guard bit, 1 round bit and sticky bit.

I got 0 10011 1010111000 for 2.6875*10^1