Distributed Python with Ray Hands on with the Ray Core APIs
summary
The session demonstrates concrete Ray Core use: writing remote functions and actors to build a working parameter server (3 workers, updates over ~20 epochs) and a tree of actors supervisor that launches workers and reports state on the Ray dashboard. It runs distributed hyperparameter tuning for XGBoost with four actors doing trials, shows a scikit‑learn drop‑in distributed adapter, and deploys a composite Hugging Face pipeline (sentiment → translate) via Ray Serve; all examples run end‑to‑end and visualize actor replicas, checkpoints, and results.
AI-written summary · about
chapters (12)
- 0:00Start
- 1:30:20Programming
- 1:30:54Parameter Server
- 1:36:55Tree of Actors
- 1:44:07Tree Actor Pattern
- 1:56:49Fault Tolerance Mechanisms
- 2:08:56Trio Pattern
- 2:14:08Trial Schedulers
- 2:14:54Scikit-Learn
- 2:27:45Create a Deployment
- 2:38:17Composition Model
- 2:41:29Composed Model
related talks
01h 30mDistributed Python with Ray-Hands on with the Ray 2.0 APIs for scaling Python Workloads
01h 31mIntroduction to Ray for distributed and machine learning applications in Python
57mIntroduction to Ray for distributed and machine learning applications in Python
26mRay: A System for High-performance, Distributed Python Applications
39mScaling Python for Machine Learning: Beyond Data Parallelism
31mScaling Python to thousands of nodes with Ray
28mHuggingFace + Ray AIR Integration: A Python developer’s guide to
29mLearning to Scale Data Science, Machine Learning, and Pandas with Ray and Modin