Ray provides developers with granular control for how their ML applications should run in a distributed environment. However, it’s not always natural for developers to think about the world in terms of tasks, actors, object management, and placement groups.
This is where Ray libraries come in - offering a familiar level of abstraction to accelerate developer productivity in the Ray ecosystem, while abstracting away the Ray building blocks.
Ray’s libraries simplify common tasks like data processing, model training, online inference, hyperparameter tuning, and reinforcement learning. Built on Ray, Ray libraries deliver scalable performance without the complexity.
How Ray libraries simplify Ray application development
Common use cases across each Ray library
How to build an end-to-end LLM workflow for Named Entity Recognition (NER) with Ray and Anyscale
If you're exploring Ray, comfortable with Python, and generally interested in LLMs, this session is for you. Whether you're exploring Ray on your own or considering it for your team, you’ll learn how to use Ray’s libraries to scale real workloads without getting lost in the complexity of the lower-level Ray details.