GeoGramBench: Benchmarking the Geometric Program Reasoning in Modern LLMs

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Luo, Shixian, Zhu, Zezhou, Yuan, Yu, Yang, Yuncheng, Shan, Lianlei, Wu, Yong
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908823843766272
author Luo, Shixian
Zhu, Zezhou
Yuan, Yu
Yang, Yuncheng
Shan, Lianlei
Wu, Yong
author_facet Luo, Shixian
Zhu, Zezhou
Yuan, Yu
Yang, Yuncheng
Shan, Lianlei
Wu, Yong
contents Geometric spatial reasoning forms the foundation of many applications in artificial intelligence, yet the ability of large language models (LLMs) to operate over geometric spatial information expressed in procedural code remains underexplored. In this paper, we address this gap by formalizing the Program-to-Geometry task, which challenges models to translate programmatic drawing code into accurate and abstract geometric reasoning. To evaluate this capability, we present GeoGramBench, a benchmark of 500 carefully refined problems organized by a tailored three-level taxonomy that considers geometric complexity rather than traditional mathematical reasoning complexity. Our comprehensive evaluation of 17 frontier LLMs reveals consistent and pronounced deficiencies: even the most advanced models achieve less than 50% accuracy at the highest abstraction level. These results highlight the unique challenges posed by program-driven spatial reasoning and establish GeoGramBench as a valuable resource for advancing research in symbolic-to-spatial geometric reasoning. Project page: https://github.com/LiAuto-DSR/GeoGramBench.
format Preprint
id arxiv_https___arxiv_org_abs_2505_17653
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle GeoGramBench: Benchmarking the Geometric Program Reasoning in Modern LLMs
Luo, Shixian
Zhu, Zezhou
Yuan, Yu
Yang, Yuncheng
Shan, Lianlei
Wu, Yong
Artificial Intelligence
Geometric spatial reasoning forms the foundation of many applications in artificial intelligence, yet the ability of large language models (LLMs) to operate over geometric spatial information expressed in procedural code remains underexplored. In this paper, we address this gap by formalizing the Program-to-Geometry task, which challenges models to translate programmatic drawing code into accurate and abstract geometric reasoning. To evaluate this capability, we present GeoGramBench, a benchmark of 500 carefully refined problems organized by a tailored three-level taxonomy that considers geometric complexity rather than traditional mathematical reasoning complexity. Our comprehensive evaluation of 17 frontier LLMs reveals consistent and pronounced deficiencies: even the most advanced models achieve less than 50% accuracy at the highest abstraction level. These results highlight the unique challenges posed by program-driven spatial reasoning and establish GeoGramBench as a valuable resource for advancing research in symbolic-to-spatial geometric reasoning. Project page: https://github.com/LiAuto-DSR/GeoGramBench.
title GeoGramBench: Benchmarking the Geometric Program Reasoning in Modern LLMs
topic Artificial Intelligence
url https://arxiv.org/abs/2505.17653