Tutorials

Learn NVFLARE in the order that matches your goal. If you are new to NVFLARE, start with the current onboarding sequence. Experienced users can go directly to the focused feature tutorials, the searchable example catalog, or the structured training resources.

Current Onboarding

Start with the Quick Start and examples for the NVFLARE version you are using.

Quick Start — run your first federated learning job in the simulator, inspect the result, and choose your next step.
Hello World examples — introductory federated applications across frameworks (PyTorch, PyTorch Lightning, TensorFlow, JAX, NumPy, Hugging Face, Flower). Follow each example’s README for dependencies, data or model preparation, and the exact run command; some examples require arguments to job.py.

Feature Tutorials

Focused Jupyter notebooks for learning one NVFLARE capability, run locally from the tutorials directory. The README in that directory is the authoritative index: it records each notebook’s outcome, prerequisites, entry point, expected completion signal, tested environment, and last verification.

Build and test applications

Intro to the FL Simulator — run jobs on a single system with the simulator-managed server and simulated clients.
Job Recipe — build and run federated jobs with the high-level Recipe API instead of hand-written job configurations.

Operate and deploy systems

NVFLARE in POC Mode — configure and operate a full local FL system with separate server, client, and admin processes.
NVFlare CLI — manage setup, recipes, jobs, systems, studies, and deployment from the nvflare command groups.
Hello FLARE API — connect an admin session and drive a running FL system programmatically with the FLARE API.

Observe and troubleshoot runs

FLARE Logging — configure console and file logging for different use cases and run modes.

Example Catalog

Browse and filter the complete set of runnable examples by experience level, framework, algorithm, application, industry, API, and privacy algorithm in the NVFLARE Catalog.
See Example Applications for the major example applications and algorithms in this repository.
Video tutorials give quick introductions to FLARE for deep learning and traditional machine learning, including CIFAR10 classification and Kaplan-Meier survival analysis.

Structured Training

Note

The self-paced course was developed with NVFLARE 2.6, and the hosted DLI courses are versioned independently of the current NVFLARE release and may use earlier versions. Both remain useful structured paths, but not all content reflects the latest APIs; some examples may need adjustments for newer versions.

Self-Paced-Training Tutorials — a five-part, 12-chapter course with over 100 notebooks and 80 videos covering federated learning with NVFLARE, from fundamentals to system architecture, privacy and security, advanced topics, and industry applications.
Hosted NVIDIA DLI courses for guided, hands-on FLARE training (independently versioned; may use earlier NVFLARE releases): Introduction to Federated Learning with NVIDIA FLARE and Decentralized AI at Scale with NVIDIA FLARE.

Visit the NVIDIA FLARE developer portal for product information, tutorials, documentation, and developer resources.