Beyond Accuracy: Evaluating the Reasoning Behavior of Large Language Models -- A Survey
This survey highlights a critical gap in evaluating LLMs for researchers and practitioners, emphasizing the need for deeper analysis beyond task performance, though it is incremental as it synthesizes existing work.
The paper addresses the uncertainty around large language models' reasoning abilities by reviewing studies that go beyond accuracy metrics to analyze their reasoning behavior, finding that LLMs often rely on surface-level patterns rather than sophisticated reasoning.
Large language models (LLMs) have recently shown impressive performance on tasks involving reasoning, leading to a lively debate on whether these models possess reasoning capabilities similar to humans. However, despite these successes, the depth of LLMs' reasoning abilities remains uncertain. This uncertainty partly stems from the predominant focus on task performance, measured through shallow accuracy metrics, rather than a thorough investigation of the models' reasoning behavior. This paper seeks to address this gap by providing a comprehensive review of studies that go beyond task accuracy, offering deeper insights into the models' reasoning processes. Furthermore, we survey prevalent methodologies to evaluate the reasoning behavior of LLMs, emphasizing current trends and efforts towards more nuanced reasoning analyses. Our review suggests that LLMs tend to rely on surface-level patterns and correlations in their training data, rather than on sophisticated reasoning abilities. Additionally, we identify the need for further research that delineates the key differences between human and LLM-based reasoning. Through this survey, we aim to shed light on the complex reasoning processes within LLMs.