SpaRC and SpaRP: Spatial Reasoning Characterization and Path Generation for Understanding Spatial Reasoning Capability of Large Language Models

TL;DR

This paper evaluates the spatial reasoning abilities of large language models, introduces new datasets and frameworks for analysis, and finds that larger models and fine-tuning improve performance, though gaps remain especially between proprietary and open-source models.

Contribution

The paper introduces the SpaRC framework and SpaRP datasets for detailed spatial reasoning analysis of LLMs, revealing performance gaps and scaling effects.

Findings

Model size correlates with spatial reasoning performance.

Fine-tuning improves LLMs' spatial reasoning accuracy.

Proprietary LLMs outperform open-source models in spatial tasks.

Abstract

Spatial reasoning is a crucial component of both biological and artificial intelligence. In this work, we present a comprehensive study of the capability of current state-of-the-art large language models (LLMs) on spatial reasoning. To support our study, we created and contribute a novel Spatial Reasoning Characterization (SpaRC) framework and Spatial Reasoning Paths (SpaRP) datasets, to enable an in-depth understanding of the spatial relations and compositions as well as the usefulness of spatial reasoning chains. We found that all the state-of-the-art LLMs do not perform well on the datasets -- their performances are consistently low across different setups. The spatial reasoning capability improves substantially as model sizes scale up. Finetuning both large language models (e.g., Llama-2-70B) and smaller ones (e.g., Llama-2-13B) can significantly improve their F1-scores by 7--32 absolute points. We also found that the top proprietary LLMs still significantly outperform their open-source counterparts in topological spatial understanding and reasoning.