An easy, silly, DIY Federated Learning framework with many baselines for individual researchers.
Project description
FedBase
An easy, modularized, DIY Federated Learning framework with many baselines for individual researchers.
Three steps to achieve FedAvg!
- Data partition
- Nodes and server simulation
- Train and test
Design philosophy
- Dataset
- Dataset
- MNIST
- CIFAR-10
- Fashion-MNIST
- ...
- Dataset partition
- IID
- Non-IID
- Dirichlet distribution
- N classes
- ...
- Fake data
- ...
- Dataset
- Node
- Local dataset
- Model
- Objective
- Optimizer
- Local update
- Test
- Server
- Model
- Aggregate
- Distribute
- Server & Node
- Topology
- Client sampling
- Exchange file
- Baselines
- Global
- Local
- FedAvg
- Visualization
How to develop your own FL with fedbase?
Baselines
- Centralized train
- Local train
- FedAvg
- Ditto
- Clustered FL
To be continued...
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
fedbase-0.2.7.tar.gz
(16.3 kB
view hashes)
Built Distribution
fedbase-0.2.7-py3-none-any.whl
(31.7 kB
view hashes)