FedVit is a transformer based pipeline for federated learning research!
We provide an end-to-end example for users to start running a medical vit!.
First of all, users need to clone the source code and install the required packages (we suggest python version >= 3.9).
git clone https://github.com/erfandarzi/Vit_FL.git
cd Vit_FL
We recommend using a new virtual environment to install Medscale:
conda create -n Vit_FL python=3.9
conda activate Vit_FL
If your backend is torch, please install torch in advance (torch-get-started). For example, if your cuda version is 11.3 please execute the following command:
conda install -y pytorch=1.10.1 torchvision=0.11.2 torchaudio=0.10.1 torchtext=0.11.1 cudatoolkit=11.3 -c pytorch -c conda-forge
For users with Apple M1 chips💻:
conda install pytorch torchvision torchaudio -c pytorch
Finally, after the backend is installed, you can install our repository from source
:
pip install .
Now, you have successfully installed the minimal version of our repository. (Optinal)
You can download lung cancer dataset from kaggle and put it in data/lung
directory.
To run a federated learning example with FedOPT
algorithm, for unbalanced dataset run:
python medscale/main.py --cfg scripts/my_configs/FedOPT_Lung_cancer.yaml
All adjustable parameters can be found in the second cell. Careful with the BATCH_SIZE
, as we are using it to parallelize head and time chunk calculations.
Our work is under MIT license. Credits to Federated Scope for their amazing pipeline.