⚡ Ray Flashcards
🚀 What is Ray?
Ray is a distributed framework for building and running Python applications at scale, especially ML and AI workloads.
🧠 What makes Ray ideal for ML workloads?
Ray simplifies distributed training, hyperparameter tuning, and model serving using unified APIs.
📦 What are Ray Core, Tune, Serve, and Train?
These are Ray’s key libraries: Core (distributed execution), Tune (HP tuning), Train (distributed training), Serve (deployment).
🌍 How does Ray scale Python code?
Ray abstracts distributed computing through Python decorators and actors, scaling functions across machines.
🔄 What is Ray Actor?
Ray Actor is a stateful process that can execute methods remotely and retain its state across method calls.
🔧 What is Ray Tune?
Ray Tune is a scalable library for distributed hyperparameter search using various optimization algorithms.
📡 What is Ray Serve?
Ray Serve is a scalable and flexible model serving library for deploying ML models as microservices.
📊 What is Ray Dataset?
Ray Dataset provides a unified API for loading, transforming, and consuming large-scale tabular data.
🧱 Is Ray framework agnostic?
Yes, Ray supports TensorFlow, PyTorch, XGBoost, LightGBM, and many other libraries for distributed ML tasks.
💡 What is Ray’s ecosystem?
Ray integrates with tools like MLflow, Kubernetes, Airflow, Dask, and Hugging Face for end-to-end workflows.