Installation

Before you start, you’ll need to setup your environment and install the appropriate packages. 🤗 SetFit is tested on Python 3.7+.

pip

The most straightforward way to install 🤗 Datasets is with pip:

pip install setfit

If you have a CUDA-capable graphics card, then it is recommended to install torch with CUDA support to train and performing inference much more quickly:

pip install torch --index-url https://download.pytorch.org/whl/cu118

Installing from source

Building 🤗 SetFit from source lets you make changes to the code base. To install from the source, clone the repository and install 🤗 SetFit in editable mode with the following commands:

git clone https://github.com/huggingface/setfit.git
cd setfit
pip install -e .

If you just want the bleeding-edge version without making any changes of your own, then install from source by running:

pip install git+https://github.com/huggingface/setfit.git

Conda

If conda is your package management system of choice, then you can install 🤗 SetFit like so:

conda install -c conda-forge setfit