Referring to the IEEE Std $754$ binary$32$ and binary$64$ formats. We can postulate a similar albeit smaller binary$16$ floating$-$point number format. Let's suppose that binary$16$ has a sign bit, a $6-$bit biased exponent, where bias $=2^{6-1}-1=31,$ and a $9-$bit fraction with an implied leading $1$ bit.

$\backslash$table$[[s,e,f$