Implement multi-tensor parameter save/load facilities
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 314
- Forks
- 77
- Avg merge
- 4d 6m
- Merged PRs (30d)
- 10
Description
To fully support model saving (#687), we need functions to save parameters, probably in the form of a dictionary mapping names to tensors.
We can't just use safetensors or zarr out of the box, because LensKit supports multiple different tensor backends, and each component needs its tensors in the correct format. We can use one (or both) of these libraries, but with additional metadata on the form a tensor should take when deserialized.
Currently leaning to towards zarr, either directly or via Google TensorStore.
Contributor guide
No contributing guide indexed for this repository
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
No files or tests are named. Start by reviewing the model-saving work in issue #687, then compare how zarr and Google TensorStore could represent tensors and backend-specific metadata. Done means parameter save/load functions support named tensors and restore each tensor in the correct backend format.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100