The bias of a half-precision floating-point number's exponent. - View it on GitHub