RUPBench: Benchmarking Reasoning Under Perturbations for Robustness Evaluation in Large Language Models

TL;DR

This paper introduces RUPBench, a comprehensive benchmark for evaluating the robustness of large language models across diverse reasoning tasks and textual perturbations, revealing model strengths and weaknesses.

Contribution

RUPBench is the first benchmark to systematically assess LLM robustness across multiple reasoning types and perturbation levels, providing detailed analysis of model error patterns.

Findings

Larger models show greater robustness to perturbations

Common errors include logical inconsistencies and lexical misunderstandings

Performance drops significantly under certain perturbations

Abstract

With the increasing use of large language models (LLMs), ensuring reliable performance in diverse, real-world environments is essential. Despite their remarkable achievements, LLMs often struggle with adversarial inputs, significantly impacting their effectiveness in practical applications. To systematically understand the robustness of LLMs, we present RUPBench, a comprehensive benchmark designed to evaluate LLM robustness across diverse reasoning tasks. Our benchmark incorporates 15 reasoning datasets, categorized into commonsense, arithmetic, logical, and knowledge-intensive reasoning, and introduces nine types of textual perturbations at lexical, syntactic, and semantic levels. By examining the performance of state-of-the-art LLMs such as GPT-4o, Llama3, Phi-3, and Gemma on both original and perturbed datasets, we provide a detailed analysis of their robustness and error patterns. Our findings highlight that larger models tend to exhibit greater robustness to perturbations. Additionally, common error types are identified through manual inspection, revealing specific challenges faced by LLMs in different reasoning contexts. This work provides insights into areas where LLMs need further improvement to handle diverse and noisy inputs effectively.