Ray is a distributed execution framework for AI applications.
Ray Features:
- Simplicity: Distributed processing can be described intuitively
- Flexibility: ML tasks such as data preprocessing, model learning, hyperparameter tuning, reinforcement learning, and model serving can be distributed.
- Scalability: Scalable from local machines to multiple clusters with almost no code changes.
The Ray library initially provided Ray Tune and Ray RLlib.
In recent years, the functionality has been rapidly extended to support other ML tasks.
