Installation¶

Home | Installation | Running MixFlow

MixFlow is currently in active development and not available as a python library. It can, however, be used through the scripts available in the github repository.

1. Clone the repository¶

git clone https://github.com/Lotfollahi-lab/MixFlow.git
cd MixFlow

2. Create and activate an environment (optional but recommended)¶

For example, using conda:

conda create -n MixFlow python=3.10 -y
conda activate MixFlow

You can also use venv or another environment manager if you prefer.

3. Install MixFlow in editable mode¶

From the repository root:

pip install -e .

This installs MixFlow and its Python dependencies while keeping the source editable for development.

4. Install Exact Environment¶

We provide the requirements.txt file that can be used to recreate exactly the environment used for our benchmarks.

pip install -r requirements.txt

Once installed, you can follow the instructions in the Running MixFlow page for an example training run on the Norman dataset.