This repository contains the benchmark suite and replication package for our paper "How well do LLMs reason over tabular data, really?", presented at the 4th Table Representation Learning Workshop at ACL 2025. It lets you reproduce the reasoning tests from the paper and explore how different models perform on table reasoning challenges.

Democratizing Insight Retrieval from (Semi-)Structured Data
opensource.org/license/MIT
Database Architectures

Wolff, C., & Hulsebos, M. (2025). How well do LLMs reason over tabular data, really?.