Practical overview
When it fits
Use FedLab when developing or studying federated-learning experiments in a PyTorch-based research framework, especially when modular simulation components and example workflows are useful.
Who it is for
- Researchers implementing federated-learning algorithms and experiments
- Developers building customizable federated-learning simulation workflows
- Learners exploring federated-learning pipelines through notebooks and examples
Documented uses
- Assemble a federated-learning simulation from modular components.
- Customize data partitioning, communication, compression and model-optimization elements.
- Explore baseline algorithm reproductions and pipeline tutorials.
- Run the documented standalone MNIST example with configurable experiment parameters.
Requirements & limits
The project is publicly available on GitHub under the Apache-2.0 license. The README documents installation from source and identifies fedlab 1.3.0 as an older stable pip version.
- The framework is based on PyTorch.
- Its listed simulation modules include communication, compression, model optimization and data partitioning.
- The repository includes baseline reproductions, notebook tutorials, examples and an API reference.
- The documented quick-start scenario uses partitioned MNIST.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Install FedLab from the source repository
Set up the repository’s latest source version with its listed Python requirements.
Before you start
- Git access to the FedLab repository
- Python package installer pip
Steps
Clone the SMILELab-FL/FedLab repository.
Change into the FedLab directory.
Install the listed dependencies with pip install -r requirements.txt.
Expected outcome
The repository and its listed requirements are installed locally.
Guide limitations
- These steps cover installation only; they do not establish successful execution in a particular environment.
- The README also lists fedlab 1.3.0 as an older stable package version.
Run the documented standalone MNIST example
Launch the repository’s standalone example using its partitioned MNIST dataset.
Before you start
- FedLab repository and its listed requirements installed
- A command-line environment with the documented Python example available
Steps
Change into ./examples/standalone/.
Run standalone.py with the documented total_clients, com_round, sample_ratio, batch_size, epochs and lr arguments.
Expected outcome
The documented command starts the standalone MNIST scenario with the supplied parameter values.
Guide limitations
- The README documents the command but does not establish results for a particular machine or environment.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 6 November 2026
- Original website title
- SMILELab-FL/FedLab · Official homepage title · retained in its source language