#include "tensorstore/util/bfloat16.h" BFloat16 tensorstore::fmax(BFloat16 a, BFloat16 b); Computes the maximum of two values.