The CodeFlare Stack is a set of tooling and best of breed code and models to help you be productive at leveraging Cloud GPU resources for ML tasks.
The CodeFlare Stack is set up to accommodate your choices. So, first, choose what you would like to accomplish:
-
Introductory Demo I want to see the experience in action.
Goals: Learning
You Provide: nothing, it just works!
CodeFlare Stack Provides: Ray cluster | Kubernetes management | Distributed training job -
Train a Masked Language Model (Demo)
Goals: Learning
You Provide: nothing, it just works!
CodeFlare Stack Provides: S3 data | Ray cluster | Kubernetes management | Distributed training job | Pop-up Dashboards -
Bring Your Own Code I have my own training code, and want to run it at scale.
Goals: Productive Use
You Provide: Python source code | S3 data | Command-line options to tweak the run parameters
CodeFlare Stack Provides: Ray cluster | Kubernetes management | Distributed training job | Link S3 credentials | Pop-up Dashboards