Skip to content
Back to search

NVFlare

NVIDIA Federated Learning Application Runtime Environment

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/NVIDIA/NVFlare with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/NVIDIA/NVFlare'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/nvidia--nvflare/badge.svg)](/listings/nvidia--nvflare)

HTML

<a href="/listings/nvidia--nvflare"><img src="/api/listings/nvidia--nvflare/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

To install the [current release](https://pypi.org/project/nvflare/):

$ python -m pip install nvflare

For detailed installation please refer to [NVIDIA FLARE installation](https://nvflare.readthedocs.io/en/main/installation.html).

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 211 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. <img src="https://raw.githubusercontent.com/NVIDIA/NVFlare/main/docs/resources/nvidia_eye.wwPt122j.png" alt="NVIDIA Logo" width="200">
  2. # NVIDIA FLARE
  3. Build federated computing applications that collaborate across organizational and geographic boundaries while each
  4. site keeps control of its data and local execution.
  5. [Website](https://nvidia.github.io/NVFlare) |
  6. [NVIDIA FLARE Developer Site](https://developer.nvidia.com/flare) |
  7. [Documentation](https://nvflare.readthedocs.io/en/main/) |
  8. [Quick Start](https://nvflare.readthedocs.io/en/main/quickstart.html) |
  9. [Examples](https://nvidia.github.io/NVFlare/catalog/) |
  10. [Discussions](https://github.com/NVIDIA/NVFlare/discussions)
  11. [Paper](https://arxiv.org/abs/2210.13291) |
  12. [Blogs](https://developer.nvidia.com/blog/tag/federated-learning) |
  13. [Talks & Papers](https://nvflare.readthedocs.io/en/main/publications_and_talks.html) |
  14. [Webinars](https://nvidia.github.io/NVFlare/webinars) |
  15. [Research](./research/README.md)
  16. [![Blossom-CI](https://github.com/NVIDIA/nvflare/workflows/Blossom-CI/badge.svg?branch=main)](https://github.com/NVIDIA/nvflare/actions)
  17. [![documentation](https://readthedocs.org/projects/nvflare/badge/?version=main)](https://nvflare.readthedocs.io/en/main/?badge=main)
  18. [![license](https://img.shields.io/badge/License-Apache%202.0-brightgreen.svg)](./LICENSE)
  19. [![pypi](https://badge.fury.io/py/nvflare.svg)](https://badge.fury.io/py/nvflare)
  20. [![pyversion](https://img.shields.io/pypi/pyversions/nvflare.svg)](https://badge.fury.io/py/nvflare)
  21. [![downloads](https://static.pepy.tech/badge/nvflare)](https://pepy.tech/project/nvflare)
  22. [![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/NVIDIA/NVFlare)
  23. ## What is NVIDIA FLARE?
  24. **[NVIDIA FLARE™](https://developer.nvidia.com/flare)** (**NV**IDIA **F**ederated **L**earning **A**pplication **R**untime **E**nvironment) is a
  25. domain-agnostic, open-source, extensible Python SDK for federated learning and other federated-computing applications.
  26. It coordinates approved local computation and exchanges application-defined results across independently controlled
  27. sites, while each site retains its source datasets and controls what results may leave.
  28. Start with a small federated-learning example on your computer. Then adapt your training code or explore statistics,
  29. custom workflows, and deployment options.
  30. ## How it works
  31. <p align="center">
  32. <img src="https://raw.githubusercontent.com/NVIDIA/NVFlare/f0aac04afdd7f0fab7675e8f5259a5a40cc9592c/docs/resources/federated_learning_overview.png"
  33. alt="Two hospitals train models on private local data and send model updates to a federated server, which aggregates them into a global model."
  34. width="650">
  35. </p>
  36. In this federated-learning workflow, the server sends a model to participating clients. Each client trains the model
  37. on its local data and returns model updates, keeping its source dataset local. The server aggregates the updates into
  38. a global model and repeats the process across training rounds.
  39. The example below runs this workflow on your computer, with two simulated clients and a server.
  40. ## Try it locally
  41. Create and activate a fresh Python virtual environment as described in the
  42. [installation guide](https://nvflare.readthedocs.io/en/main/installation.html), then choose **one** installation:
  43. **Released package, once NVFLARE 2.10.0 or later is available on [PyPI](https://pypi.org/project/nvflare/):**
  44. ```bash
  45. python -m pip install "nvflare>=2.10.0"
  46. ```
  47. **Development version from `main`, including before the 2.10.0 release:**
  48. ```bash
  49. python -m pip install "git+https://github.com/NVIDIA/NVFlare.git@main"
  50. ```
  51. The development option requires Git; pip manages a temporary checkout. `main` is the unstable development branch.
  52. Use a separate virtual environment for each installation. If you use an earlier stable release, follow its
  53. [version of the Quick Start](https://nvflare.readthedocs.io/en/stable/quickstart.html); the `nvflare examples` command
  54. was introduced in 2.10.0.
  55. After either installation, retrieve the matching Hello PyTorch example, install its dependencies, and run it:
  56. ```bash
  57. nvflare examples get hello-pt
  58. cd hello-pt
  59. python -m pip install -r requirements.txt
  60. python job.py --log_config progress
  61. ```
  62. This runs a two-client simulation on CPU using synthetic data; no dataset download is required. For other examples,
  63. retrieve a revision-matched copy with `nvflare examples get`. Add the extras the example's README names, such as
  64. `HE`, to the same NVFLARE installation, install its `requirements.txt`, then follow its README. If an older
  65. `requirements.txt` still has a line that installs `nvflare` or `nvflare-nightly`, skip it so the installed NVFLARE
  66. version is preserved.
  67. The installed NVFLARE version selects the example version: `nvflare examples get` downloads the example from the
  68. source revision recorded in your installation. To reproduce a released version, pin that version when installing it.
  69. For download failures and GitHub rate limits, see the
  70. [examples CLI guide](https://nvflare.readthedocs.io/en/main/user_guide/nvflare_cli/examples_command.html#automation).
  71. Continue with the [Quick Start for main](https://nvflare.readthedocs.io/en/main/quickstart.html) to understand the run
  72. and inspect its results. If you use a released package, select its version in the documentation and follow the
  73. matching example README for dependencies, commands, and troubleshooting. The `nvflare examples` command requires
  74. NVFLARE 2.10.0 or later, or a development installation from `main`.
  75. ## Choose your path
  76. Start with the guide for the concept and supported workflow, then use a maintained example as the runnable reference.
  77. | Goal | Read first | Then run or adapt |
  78. |---|---|---|
  79. | Adapt existing training code | [Quick Start](https://nvflare.readthedocs.io/en/main/quickstart.html), then [Agent Skills](https://nvflare.readthedocs.io/en/main/user_guide/agent_skills/index.html) for supported projects or [API selection](https://nvflare.readthedocs.io/en/main/user_guide/data_scientist_guide/api_selection.html) for manual and custom integration | Try the [Agent Skills conversion examples](./examples/hello-world/agent-skills/README.md), or use [Hello PyTorch with the Client API and a Recipe](./examples/hello-world/hello-pt/README.md) |
  80. | Fine-tune large language models | [Federated LLM guide](https://nvflare.readthedocs.io/en/main/programming_guide/llm_fine_tuning.html) | [Hugging Face LoRA fine-tuning](./examples/hello-world/hello-huggingface/README.md) or [NeMo supervised fine-tuning](./integration/nemo/examples/supervised_fine_tuning/README.md) |
  81. | Federated analytics and statistics | [Agent Skills](https://nvflare.readthedocs.io/en/main/user_guide/agent_skills/index.html) for supported tabular and image datasets, or the [Federated Statistics guide](https://nvflare.readthedocs.io/en/main/examples/federated_statistics_overview.html) for manual and custom workflows | Agent-assisted [tabular](./examples/hello-world/agent-skills/fedstats-tabular/README.md) or [image statistics](./examples/hello-world/agent-skills/fedstats-image/README.md), or manual [DataFrame](./examples/advanced/federated-statistics/df_stats/README.md) or [image statistics](./examples/advanced/federated-statistics/image_stats/README.md) |
  82. | Research and design custom FL algorithms or workflows | [Collaboration API guide (Technical Preview)](https://nvflare.readthedocs.io/en/main/user_guide/data_scientist_guide/collab_api.html) | [Hello FedAvg with the Collab API](./examples/hello-world/hello-collab/README.md), then [split learning, asynchronous aggregation, and swarm examples](./examples/advanced/collab/README.md) |
  83. | Learn a specific NVIDIA FLARE feature interactively | [Feature Tutorials](https://nvflare.readthedocs.io/en/main/tutorials.html#feature-tutorials) | [Simulator](./examples/tutorials/flare_simulator.ipynb), [Recipe](./examples/tutorials/job_recipe.ipynb), or [logging](./examples/tutorials/logging.ipynb) notebook |
  84. | Follow structured NVIDIA FLARE training | [Tutorials and training](https://nvflare.readthedocs.io/en/main/tutorials.html) | [Self-paced curriculum](./examples/tutorials/self-paced-training/README.md) or [NVIDIA DLI introductory course](https://learn.nvidia.com/courses/course-detail?course_id=course-v1:DLI+S-FX-28+V1) |
  85. | Validate POC and production environments | [Deployment guide](https://nvflare.readthedocs.io/en/main/user_guide/admin_guide/deployment/overview.html) | [Hello PyTorch environment example](./examples/advanced/hello-pt-environments/README.md) |
  86. | Add privacy and governance controls | [Security overview](https://nvflare.readthedocs.io/en/main/system_architecture/security_overview.html) | [DP](./examples/hello-world/hello-dp/README.md), [HE](./examples/advanced/cifar10/pt/cifar10-real-world/cifar10_fedavg_he/README.md), [PSI](./examples/advanced/psi/user_email_match/README.md), [confidential computing](./examples/advanced/cvm_builder/README.md), or [site policies](./examples/advanced/federated-policies/README.rst) |
  87. Use the [NVIDIA FLARE website](https://nvidia.github.io/NVFlare) to discover tutorials, examples, research, webinars,
  88. and events. The [documentation](https://nvflare.readthedocs.io/en/main/) provides the authoritative guides. After
  89. installing NVFLARE 2.10.0 or later, or the development version from `main`, `nvflare examples list` shows the curated
  90. catalog and `nvflare examples get <name>` retrieves an example matched to that installation's source revision. Each
  91. example README then owns its exact setup and run steps.
  92. ## Federated learning and computing
  93. **Federated learning** lets sites collaboratively train or evaluate models on their local data. Training and evaluation
  94. happen where the data resides; the underlying training examples are not exchanged between sites or sent to a central
  95. server. This can enable collaboration where privacy, regulation, data sovereignty, intellectual property, data
  96. ownership, or the cost of moving data makes central collection impractical. Federated learning is one application of
  97. the broader federated computing paradigm.
  98. **Federated computing** brings approved computation to data distributed across multiple parties or locations. In
  99. addition to model training and evaluation, it supports analytics, statistics, site-local data processing, and custom
  100. multi-site computation. Participating sites run tasks locally and share only outputs permitted by the collaboration.
  101. NVIDIA FLARE builds its federated-learning capabilities on top of a general federated-computing core. That core
  102. provides workflow orchestration, task execution, communication, security, and lifecycle services shared by training,
  103. evaluation, analytics, statistics, and custom distributed applications.
  104. A **job** packages the application logic and configuration. A **workflow** coordinates tasks across participating
  105. sites. Each site executes its task against local resources, and policies determine what can run and what may leave the
  106. site. Depending on the application, the shared result can be an aggregate model, evaluation, statistic, intersection,
  107. or another collaboration artifact.
  108. NVIDIA FLARE also supports decentralized and client-controlled workflows, as well as federated computing applications
  109. whose shared result is a statistic, evaluation, intersection, or another permitted application result rather than a
  110. model.
  111. > **Keep data at its source.** NVIDIA FLARE moves approved computation to participating sites. In a properly designed
  112. > and governed federation, raw datasets remain at their source and only outputs permitted by the collaboration and
  113. > each site's policies leave the site. Application owners and site operators define and enforce those policies for
  114. > their data and threat model.
  115. ## Capabilities
  116. | Area | Capabilities |
  117. |---|---|
  118. | Federated applications | Training and fine-tuning, evaluation and cross-site validation, analytics and statistics, site-local data processing, and custom multi-site computation |
  119. | Models and frameworks | Any model type or ML/AI framework through extensible Python APIs, with maintained integrations and examples for PyTorch, TensorFlow, JAX, scikit-learn, XGBoost, Hugging Face, NeMo, and Flower |
  120. | Algorithms and workflows | FedAvg, FedProx, FedOpt, SCAFFOLD, Ditto, cyclic and swarm learning, horizontal and vertical FL, and custom server- or client-controlled workflows |
  121. | Privacy, security, and governance | Site-controlled authorization, audit logging, secure provisioning, differential privacy, homomorphic encryption, private set intersection, and confidential computing options |
  122. | Runtime and operations | Local simulation, proof-of-concept environments, production provisioning, cloud and on-premises deployment, resiliency, monitoring, and experiment tracking |
  123. FLARE's component and event architecture lets applications replace or extend controllers, aggregators, executors,
  124. filters, persistence, communication, and deployment behavior without rewriting the entire system.
  125. ## From local development to production
  126. The application stays consistent while its execution environment changes:
  127. | Stage | Purpose |
  128. |---|---|
  129. | [Simulator](https://nvflare.readthedocs.io/en/main/user_guide/nvflare_cli/fl_simulator.html) (`SimEnv`) | Run a local job directly to test application code and algorithms, without first starting and administering a FLARE deployment. |
  130. | [Proof&nbsp;of&nbsp;Concept&nbsp;(POC)](https://nvflare.readthedocs.io/en/main/user_guide/nvflare_cli/poc_command.html)&nbsp;(`PocEnv`) | Start a local FLARE deployment to practice job submission, administration, and deployment behavior before moving to provisioned sites. |
  131. | [Production](https://nvflare.readthedocs.io/en/main/user_guide/admin_guide/deployment/overview.html) (`ProdEnv`) | Run across provisioned sites with production identities, authorization, networking, policies, and operations. |
  132. See the [run modes](https://nvflare.readthedocs.io/en/main/run_mode.html) and
  133. [deployment guide](https://nvflare.readthedocs.io/en/main/user_guide/admin_guide/deployment/overview.html) for the
  134. differences and operational requirements.
  135. ## Industry use cases
  136. Federated computing is useful when organizations need a shared result but cannot freely pool the source data. NVIDIA
  137. FLARE has been applied to:
  138. | Industry | Representative problems and evidence |
  139. |---|---|
  140. | Healthcare and oncology | Multi-institution medical imaging, tumor segmentation, survival analysis, and collaborative cancer research. See the [industry use cases](https://nvflare.readthedocs.io/en/main/industry_use_cases.html) and [healthcare research implementations](./research/README.md). |
  141. | Life sciences and drug discovery | Collaborative protein-property prediction, molecular modeling, and model fine-tuning while organizations retain proprietary research data. See the [life-sciences case studies](https://nvflare.readthedocs.io/en/main/industry_use_cases.html#healthcare-life-sciences). |
  142. | Financial services | Cross-institution fraud detection, risk analysis, and federated statistics over transaction data. See the [federated fraud-detection research workflow](./research/fsi-fraud-detection/README.md). |
  143. | Automotive, edge, and industrial systems | Learning from distributed vehicles, devices, sensors, and facilities without centralizing every raw stream. See the [FLARE Day deployments and talks](https://developer.nvidia.com/flare-day-2025). |
  144. | Government and scientific computing | Collaboration across facilities where sovereignty, classification, scale, or data movement constrains centralization. See the [industry use cases](https://nvflare.readthedocs.io/en/main/industry_use_cases.html#government-national-security). |
  145. These references describe particular deployments and studies. The privacy, security, governance, and model-quality
  146. properties of a new application still depend on its design, policies, data, and validation.
  147. ### Solution ecosystem
  148. Independent software vendors (ISVs) and global systems integrators (GSIs) use NVIDIA FLARE to build federated
  149. platforms, governed data-collaboration solutions, and customer implementations. Public technical sessions include
  150. Rhino Health's federated computing platform, an Apheris and BioNeMo data-collaboration implementation, and Deloitte's
  151. FedRAG implementation. Explore these and other ecosystem work in the [NVIDIA FLARE webinar
  152. series](https://nvidia.github.io/NVFlare/webinars/) and [FLARE Day](https://nvidia.github.io/NVFlare/flareDay/).
  153. ## Project and community
  154. - Read [What's New](https://nvflare.readthedocs.io/en/main/whats_new.html) and the
  155. [talks and publications](https://nvflare.readthedocs.io/en/main/publications_and_talks.html).
  156. - Ask questions and share ideas in [GitHub Discussions](https://github.com/NVIDIA/NVFlare/discussions).
  157. - Review the [contributing guide](./CONTRIBUTING.md) and open
  158. [good first issues](https://github.com/NVIDIA/NVFlare/issues?q=is%3Aissue+is%3Aopen+label%3A%22good+first+issue%22).
  159. - Cite the [NVIDIA FLARE paper](https://arxiv.org/abs/2210.13291) when the project supports your work.
  160. NVIDIA FLARE is released under the [Apache 2.0 license](./LICENSE).

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.