lmcinnes / lmcinnes/pynndescent
Backporting from UMAP 0.4dev
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 970
- Forks
- 107
- PR merge metrics
- No merged PRs in 30d
Description
In https://github.com/lmcinnes/pynndescent/pull/16, there was mention of the UMAP 0.4dev branch having diverged substantially from pynndescent for the sparse code.
The non-sparse code doesn't seem to have been changed as much, though. The main change is in the nearest neighbor descent code; apart from that there is some variance in numba decoration, but that's about it.
If there is interest in porting UMAP 0.4dev code from `utils.py`, `nn_descent.py` and `rp_trees.py` back into pynndescent separately from the sparse or threaded code, I am happy to create a PR, so at least there are some diffs which can be examined to work out which bits to use from each code base.
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.
Research direction
Compare UMAP 0.4dev with pynndescent in utils.py, nn_descent.py, and rp_trees.py, starting with the nearest neighbor descent code and the differences in numba decoration. Determine which non-sparse changes should be ported separately from sparse or threaded code; done means the relevant diffs have been examined and the selected changes are backported.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- search
- Issue type
- Refactor
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 15/100