How to obtain the half precision floating point representation of a number?

Asked Apr 18 '19 at 17:23

Active Apr 18 '19 at 17:23

Viewed 676 times

I want to obtain the binary representation of a variable x in half precision floating point representation. x can be anything (like -1.25 or 10 etc). I have tried quite a few things but can't really get this right.
I have tried making my own function but it doesn't work for all numbers. It's filled with bugs.
I have tried using struct but I don't really know much on how to use it.

var = pack('e', 0.25)
t = unpack('H',var)
bin(t[0] & 0xffff)

This doesn't seem to give correct representation.

asked Apr 18 '19 at 17:23

Black Jack 21

1

[Related](https://stackoverflow.com/questions/33451800/decimal-to-binary-half-precision-ieee-754-in-python): use `np.float16(var).tostring()`. – Trevor Keller Apr 18 '19 at 17:29
This seems helpful: https://stackoverflow.com/a/16444778/3377926 – Bendik Apr 18 '19 at 17:30
@TrevorKeller `bin(np.float16(117.0).view('H'))[2:].zfill(16)` was just what I needed! Thanks – Black Jack 21 Apr 18 '19 at 18:20

How to obtain the half precision floating point representation of a number?

0 Answers0