[docs] add lightgbm_ray to docs

[jameslamb]

I recently saw this blog post, "Introducing Distributed LightGBM Training with Ray".

ray is an open-source, Python-based framework for distributed computing. It's similar to Dask in a lot of ways, but was originally designed for reinforcement learning use cases (before growing into this general-purpose framework).

https://github.com/ray-project/lightgbm_ray is now officially a part of the Ray project, so I think it should be added to the "Distributed Learning Guide" here.

Notes for Reviewers

Tagging the lightgbm_ray contributors, @Yard1 and @richardliaw. Please let us know if you see any issues here (for example, I wasn't sure whether to use the name lightgbm_ray or LightGBM-Ray, since both are used in https://github.com/ray-project/lightgbm_ray).

[richardliaw]

awesome! Would you be willing to take contributions to the docs about how to use LightGBM-Ray?

[jameslamb]

Would you be willing to take contributions to the docs about how to use LightGBM-Ray?

I'd prefer not to have documentation here that has to be kept up to date with LightGBM-Ray. For example, I wouldn't support having code snippets and advice at the level of depth we provide for the Dask interface (https://lightgbm.readthedocs.io/en/latest/Parallel-Learning-Guide.html#dask).

However, we'd welcome an additional link or two to tutorials. Similar to how we handle SynapseML (formerly mmlspark)

https://github.com/microsoft/LightGBM/blame/dc09d1b41f95f413a4dcb648478fdebf862c49ea/docs/Parallel-Learning-Guide.rst#L52

https://github.com/microsoft/LightGBM/blame/dc09d1b41f95f413a4dcb648478fdebf862c49ea/docs/Parallel-Learning-Guide.rst#L458

and kubeflow

https://github.com/microsoft/LightGBM/blame/dc09d1b41f95f413a4dcb648478fdebf862c49ea/docs/Parallel-Learning-Guide.rst#L350-L352

https://github.com/microsoft/LightGBM/blame/dc09d1b41f95f413a4dcb648478fdebf862c49ea/docs/Parallel-Learning-Guide.rst#L474-L480

[Yard1]

We can perhaps link to our docs then? https://docs.ray.io/en/latest/lightgbm-ray.html

Thanks for the PR!

[jameslamb]

We can perhaps link to our docs then? https://docs.ray.io/en/latest/lightgbm-ray.html

oh sure, thanks for that! I'll add that right now.

While I'm looking there...did you run into some issues doing distributed learning-to-rank training? Noticed that there isn't a RayLGBMRanker at https://docs.ray.io/en/latest/lightgbm-ray.html#scikit-learn-api.

If you ran into any problems we'd welcome issue reports at https://github.com/microsoft/LightGBM/issues.

[Yard1]

We just hadn't gotten around to adding ranking support just yet - it's on our todo list. Will definitely let you folks know if we run into any issues!

[StrikerRUS]

LGTM, very nice addition to our docs!

[github-actions]

This pull request has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this.