google-deepmind / google-deepmind/optax
Timeline for JaxOpt migration
- Dominant language
- Python
- Stars
- 2.3k
- Forks
- 369
- Avg merge
- 10h 15m
- Merged PRs (30d)
- 7
Description
Hello, what's the current roadmap for jaxopt migration into optax? Will the scope of jaxopt be maintained, or will a trimming/expansion of features happen?
EDIT 27 October, 2024:
Dear all, based on optax's PR's, I suspect optax is adopting a ML-first/only roadmap. This is fine, as it's in their interest. However, the larger science community requires an orthogonal set of optimisation algorithms. I propose that, if there is enough interest among you all, to open a separate community-maintained repository covering the missing space of tools. It would focus on performance, robustness, and ease of use, and welcome ANY category of algorithm, predominantly focusing on the science community. If you're in favour leave a +1 or <3 on this comment, and I'll tag you when I open the repository.
EDIT 7 March, 2025:
Apologies on the delay. Becoming a father soon, and been busy preparing the nest. But, alas I have started a centralising repo. Please share feedback community. https://github.com/Joshuaalbert/curated_jax_benchmarks
Contributor guide
Assessment
This issue has not been assessed yet.