freerec 0.8.7

PyTorch library for recommender systems

Dataset processing | Training and Tuning | RecBoard

FreeRec is a repository designed for easy (recommendation) data pre-processing and model training. You are free to specify your own framework based on FreeRec.

Requirements:

Python >= 3.9 | PyTorch >=2.0 | TorchData >=0.6.0 | PyG >=2.3

conda create --name=FreeRec python=3.9
conda activate FreeRec

Note: After PyTorch 2.0, TorchData seems to have stopped being updated, and you can install it with --no-deps to avoid installing dependencies.

pip install --no-deps torchdata

Installation

pip install freerec

or (for latest)

pip install git+https://github.com/MTandHJ/freerec.git

Data Pipeline

Refer to here for dataset processing and splitting.

Training Flow

Reference Code

• TorchRec: https://github.com/pytorch/torchrec
• DeepCTR-Torch: https://github.com/shenweichen/DeepCTR-Torch
• FuxiCTR: https://github.com/xue-pai/FuxiCTR
• BARS: https://github.com/openbenchmark/BARS
• RecBole: https://github.com/RUCAIBox/RecBole

Acknowledgements

Thanks to ChatGPT for the annotation of some code. For this reason, some of the comments may be illogical.