MathHay: An Automated Benchmark for Long-Context Mathematical Reasoning in LLMs

TL;DR

MathHay is an automated benchmark designed to evaluate the long-context mathematical reasoning abilities of large language models, revealing that even top models struggle significantly, thus highlighting the need for further advancements.

Contribution

The paper introduces MathHay, a novel benchmark specifically targeting long-context mathematical reasoning in LLMs, filling a gap in existing evaluation tools.

Findings

Top models achieve only around 51% accuracy on MathHay.

Even the best model struggles with long-context mathematical reasoning.

MathHay reveals significant room for improvement in LLM capabilities.

Abstract

Recent large language models (LLMs) have demonstrated versatile capabilities in long-context scenarios. Although some recent benchmarks have been developed to evaluate the long-context capabilities of LLMs, there is a lack of benchmarks evaluating the mathematical reasoning abilities of LLMs over long contexts, which is crucial for LLMs' application in real-world scenarios. In this paper, we introduce MathHay, an automated benchmark designed to assess the long-context mathematical reasoning capabilities of LLMs. Unlike previous benchmarks like Needle in a Haystack, which focus primarily on information retrieval within long texts, MathHay demands models with both information-seeking and complex mathematical reasoning abilities. We conduct extensive experiments on MathHay to assess the long-context mathematical reasoning abilities of eight top-performing LLMs. Even the best-performing model, Gemini-1.5-Pro-002, still struggles with mathematical reasoning over long contexts, achieving only 51.26% accuracy at 128K tokens. This highlights the significant room for improvement on the MathHay benchmark.