Bug#1070718: ITP: python-gfloat -- Python module of generic floating point encode/decode logic
Package: wnpp
Severity: wishlist
Owner: Scott Kitterman <debian@kitterman.com>
X-Debbugs-Cc: debian-devel@lists.debian.org
* Package name : python-gfloat
Version : 0.1
Upstream Contact: Andrew Fitzgibbon <awf@graphcore.ai>
* URL : https://github.com/graphcore-research/gfloat
* License : Expat
Programming Lang: Python
Description : Python module of generic floating point encode/decode logic
An implementation of generic floating point encode/decode logic, handling
various current and proposed floating point types:
.
- IEEE 754: Binary16, Binary32
- OCP Float8: E5M2, E4M3
- IEEE WG P3109: P{p} for p in 1..7
- OCP MX Formats: E2M1, M2M3, E3M2, E8M0, INT8, and the MX block formats.
.
The library favours readability and extensibility over speed - for fast
implementations of these datatypes see, for example, ml_dtypes, bitstring, MX
PyTorch Emulation Library.
I intend to maintain this as part of the Debian Python Team. It's
needed to run tests for the newest version of python-bitstring.
Scott K
Reply to: