Installation

Prerequisites

• Python >= 3.10

• PyTorch (CPU or GPU)

Clone the repository first:

git clone https://github.com/pliang279/MultiBench.git
cd MultiBench

Windows users should check out the Windows Subsystem for Linux (WSL).

Virtual environment

It is advised to use a virtual environment to manage dependencies. You can create one using uv (recommended) or the built-in venv module.

Using uv (recommended):

uv venv
source .venv/bin/activate  # On Windows, use .venv\Scripts\activate

Using venv:

python -m venv .venv
source .venv/bin/activate  # On Windows, use .venv\Scripts\activate

If you are not using uv, simply drop the uv prefix from the installation commands below (e.g. pip install -r requirements.txt).

Installing dependencies

GPU support. Install PyTorch using the appropriate command from the PyTorch website. The following installs the latest stable PyTorch with CUDA support, then the remaining requirements:

uv pip install -r requirements.txt

CPU only. Install the CPU build of PyTorch directly:

uv pip install torch torchvision --index-url https://download.pytorch.org/whl/cpu
uv pip install memory-profiler scikit-learn scipy matplotlib h5py tqdm

The quickstart examples also use a few dataset-specific packages:

uv pip install gdown yfinance pandas pmdarima fannypack "numpy<2"

Conda (alternative). An environment.yml is also provided:

conda env create [-n ENVNAME] -f environment.yml

From there, you should be able to try out the example scripts and the rest of the code by running a Python kernel inside the repository folder. Head to the Quick Start guide for a runnable first experiment.