---
title: README
emoji: 🐨
colorFrom: pink
colorTo: gray
sdk: gradio
pinned: false
---

<p align="center"><img src="https://relbench.stanford.edu/img/logo.png" alt="logo" width="600px" /></p>

----

[**Website**](https://relbench.stanford.edu) | [**Vision Paper**](https://relbench.stanford.edu/paper.pdf) | [**Mailing List**](https://groups.google.com/forum/#!forum/relbench/join) | [![Twitter](https://img.shields.io/twitter/url/https/twitter.com/cloudposse.svg?style=social&label=Follow%20%40RelBench)](https://twitter.com/RelBench)

Relational Deep Learning is a new approach for end-to-end representation learning on data spread across multiple tables, such as in a _relational database_ (see our [vision paper](https://relbench.stanford.edu/paper.pdf)). RelBench is the accompanying benchmark which seeks to facilitate efficient, robust and reproducible research in this direction. It comprises of a collection of realistic, large-scale, and diverse datasets structured as relational tables, along with machine learning tasks defined on them. It provides full support for data downloading, task specification and standardized evaluation in an ML-framework-agnostic manner. Additionally, there is seamless integration with [PyTorch Geometric](https://github.com/pyg-team/pytorch_geometric) to load the data as a graph and train GNN models, and with [PyTorch Frame](https://github.com/pyg-team/pytorch-frame) to encode the various types of table columns. Finally, there is a leaderboard for tracking progress.