Introduction to Ray for distributed and machine learning applications in Python
summary
Introductory, hands-on tour of Ray Core 2.0 walks code not concepts: it converts Python functions into remote functions and actors, demonstrates the shared object store, and shows the autoscaler spinning up nodes. Concrete results include a Fibonacci demo (serial ~1.9s vs parallel much faster) and a Monte Carlo π run (parallel ~42 ms), image-transform benchmarks, and a tree-of-actors hyperparameter example comparing XGBoost, linear regression, and decision tree training. Notes cite real-world use (Cohere, OpenAI, Hugging Face) and working notebooks on GitHub.
AI-written summary · about
chapters (12)
- 0:00Welcome
- 3:53Why Ray and ecosystem
- 8:48Ray architecture
- 15:26Ray design patterns
- 21:55Ray APIs and object store
- 33:50Use cases and integrations
- 42:02Ray for large language models
- 56:27Parallelizing Python
- 1:00:38Autoscaler demo
- 1:16:31Ray actors
- 1:27:13Actor tree pattern
- 1:30:22Summary
related talks
01h 30mDistributed Python with Ray-Hands on with the Ray 2.0 APIs for scaling Python Workloads
15mRay: A System for Distributed Applications
02h 51mDistributed Python with Ray Hands on with the Ray Core APIs
24mRay: A System for High-performance, Distributed ML Applications
26mRay: A System for High-performance, Distributed Python Applications
42mRay: A distributed system for emerging AI applications
11mKeynote Q&A
31mNext-Generation Data Science Workflows Using Ray