search

ray-project / ray

Ray is an AI compute engine. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.
https://ray.io
Apache License 2.0
33.72k stars 5.73k forks source link

[Docs] Create a tutorial showing how to use Ray with Feast. #23171

Open dynamicwebpaige opened 2 years ago

dynamicwebpaige commented 2 years ago

Search before asking

Description

Motivation

Feast is an operational data system for managing and serving machine learning features to models in production. It can serve features from a low-latency offline store (for real-time prediction), or from an off-line store (for scale-out batch scoring or training models).

Feature stores have emerged as an important component in the modern machine learning stack, as more data scientists and engineers work together to operationalize ML. One of the toughest challenges for operationalizing ML is data: how to compute and select features, store, validate serve, discover and share them.

Given the importance of feature stores for modern machine learning systems, it might make sense to include a tutorial for using Feast with Ray in the [ray.io]() docs.

Example

The third module of this Feast Workshop uses XGBoost and Feast with Ray, and would be useful to reference as inspiration for such a tutorial.

Use case

No response

Related issues

No response

Are you willing to submit a PR?

maxpumperla commented 2 years ago

@dynamicwebpaige great idea!

In the current structure "Ray ML" is split by libraries, and to a certain extent it has to be, as we have so much stuff. In your view, would it make sense to have Ray ML Examples section or even Examples for all of Ray? (I have an opinion on this, but love to hear yours).

dynamicwebpaige commented 2 years ago

@maxpumperla

I would love to see a ray-project/examples repo of vignettes that show how to use individual libraries (ex: Tune) with other components of the machine learning ecosystem; and Ray libraries composed together in end-to-end projects. We have few such examples of long-form documentation today.

dynamicwebpaige commented 2 years ago

Related to @edoakes' earlier issue: #13913

tianlinzx commented 2 years ago

Will this issue be picked up ? We have the same requirement to integrate Feast and Ray.

Spartee commented 2 years ago

@dynamicwebpaige I adapted the above example, put together by @dmatrix, to use Redis for my lightning talk at Ray Summit this wednesday. We are going to put more into this over time. Was hoping to get the HA Ray Cluster implementation running with RedisE, but only recently found the KubeRay docs (thanks to @simon-mo). We will get that going after the Summit however.

Codebase is here: https://github.com/RedisVentures/ray-feast