Tag: 32-bit floating point representation