#include "tensorstore/util/bfloat16.h"
tensorstore::BFloat16::operator float() const;

Lossless conversion to float.