Wheels with Python and libtorch stable ABI
- Dominant language
- Python
- Stars
- 16
- Forks
- 8
- PR merge metrics
- No merged PRs in 30d
Description
Hi, do we consider publishing wheels with Python and libtorch stable ABI? Basically this means we no longer need to build a different wheel for every Python and PyTorch version. We only need to build different wheels for Windows/Linux and CUDA 12/CUDA 13/ROCm 7.
I've also started discussion in [nunchaku](https://github.com/nunchaku-ai/nunchaku/issues/659) and [causal-conv1d](https://github.com/Dao-AILab/causal-conv1d/pull/102). I'm willing to help convert more wheels into this format if you need.
Contributor guide
Research direction
Start by reading the linked nunchaku issue and causal-conv1d pull request to understand how stable-ABI wheels are being approached. No files, tests, or entry points are named in this issue; completion would require an agreed wheel compatibility matrix and a defined plan for converting the relevant wheels.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- build-system
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100