AOTI Cuda Partitioner Should Not Take Method Name as an arg
Open
@JacobSzwejbka is already working on this.
Since Oct 8, 2025.
- Dominant language
- Python
- Stars
- 5k
- Forks
- 1.2k
- Avg merge
- 2d 10h
- Merged PRs (30d)
- 581
Description
🚀 The feature, motivation and pitch
This doesnt seem like it should be the users responsibility to manage. Ideally the delegate would name its blobs on its own and know which one its looking for, this is what we expect of other backends when putting blobs in the named data map. Because it takes it as an arg it also creates a weird pattern where we have to create a separate partitioner instance for each entry point when we should really be able to use the same one across all of them.
example usage https://github.com/huggingface/optimum-executorch/commit/dfa2d7ab0bce75d865e923717072b645ed0432b1
Alternatives
No response
Additional context
No response
RFC (Optional)
No response
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Assessment
This issue has not been assessed yet.