awaelchli commented 2 years ago

🚀 Feature

Add Bagua to the supported strategy types in Lite.

Motivation

Users who want to try Bagua but are hesitant to convert fully to Lightning can take a shortcut and just use Lite to benchark their code against SOTA strategies with only a few code changes.

Pitch

In theory Lite is strategy agnostic and can support Bagua out of the box. However, we should verify that first with a simple test and then add it to the supported types.

Alternatives

No.

Additional context

12069

If you enjoy Lightning, check out our other projects! ⚡

Metrics: Machine learning metrics for distributed, scalable PyTorch applications.
Lite: enables pure PyTorch users to scale their existing code on any kind of device while retaining full control over their own loops and optimization logic.
Flash: The fastest way to get a Lightning baseline! A collection of tasks for fast prototyping, baselining, fine-tuning, and solving problems with deep learning.
Bolts: Pretrained SOTA Deep Learning models, callbacks, and more for research and production with PyTorch Lightning and PyTorch.
Lightning Transformers: Flexible interface for high-performance research using SOTA Transformers leveraging Pytorch Lightning, Transformers, and Hydra.

cc @borda @carmocca @justusschock @awaelchli @wangraying

carmocca commented 2 years ago

Considering that we are re-thinking Lite and that Bagua hasn't had a release since January, I suggest freezing this work