让我自由表达？格式限制对大型语言模型性能影响的研究
Let Me Speak Freely? A Study on the Impact of Format Restrictions on Performance of Large Language Models
AI总结：
本研究探讨了结构化格式限制对大型语言模型推理和领域知识理解能力的影响，发现与自由格式相比，格式限制会导致模型推理能力显著下降，且约束越严格性能退化越严重。
摘要：
结构化生成（以JSON和XML等标准化格式生成内容）在实际应用中被广泛用于从大型语言模型（LLMs）中提取关键输出信息。本研究探讨了这种对生成空间的约束是否会影响LLMs的能力，包括推理和领域知识理解。具体而言，我们评估了LLMs在受限遵循结构化格式与生成自由格式响应时，在各类常见任务上的表现。令人惊讶的是，我们观察到在格式限制下，LLMs的推理能力出现显著下降。此外，我们发现更严格的格式约束通常会导致推理任务中更严重的性能退化。
英文摘要：
Structured generation, the process of producing content in standardized formats like JSON and XML, is widely utilized in real-world applications to extract key output information from large language models (LLMs). This study investigates whether such constraints on generation space impact LLMs abilities, including reasoning and domain knowledge comprehension. Specifically, we evaluate LLMs performance when restricted to adhere to structured formats versus generating free-form responses across various common tasks. Surprisingly, we observe a significant decline in LLMs reasoning abilities under format restrictions. Furthermore, we find that stricter format constraints generally lead to greater performance degradation in reasoning tasks.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
