A Closer Look at Logical Reasoning with LLMs: The Choice of Tool Matters
This work addresses a gap in understanding performance variations in LLM-based logical reasoning, which is important for researchers and practitioners in AI and NLP, though it is incremental as it focuses on comparing existing tools.
The study investigated how the choice of symbolic solver (e.g., Z3, Pyke, Prover9) affects the performance of LLMs in logical reasoning tasks, finding that tool-executable rates varied by nearly 50% across benchmarks, with a strong correlation between translation quality and outcome accuracy for Prover9.
The emergence of Large Language Models (LLMs) has demonstrated promising progress in solving logical reasoning tasks effectively. Several recent approaches have proposed to change the role of the LLM from the reasoner into a translator between natural language statements and symbolic representations which are then sent to external symbolic solvers to resolve. This paradigm has established the current state-of-the-art result in logical reasoning (i.e., deductive reasoning). However, it remains unclear whether the variance in performance of these approaches stems from the methodologies employed or the specific symbolic solvers utilized. There is a lack of consistent comparison between symbolic solvers and how they influence the overall reported performance. This is important, as each symbolic solver also has its own input symbolic language, presenting varying degrees of challenge in the translation process. To address this gap, we perform experiments on 3 deductive reasoning benchmarks with LLMs augmented with widely used symbolic solvers: Z3, Pyke, and Prover9. The tool-executable rates of symbolic translation generated by different LLMs exhibit a near 50% performance variation. This highlights a significant difference in performance rooted in very basic choices of tools. The almost linear correlation between the executable rate of translations and the accuracy of the outcomes from Prover9 highlight a strong alignment between LLMs ability to translate into Prover9 symbolic language, and the correctness of those translations.