思考，但不要过度思考：重现递归语言模型
Think, But Don't Overthink: Reproducing Recursive Language Models
AI总结：
本研究通过评估不同递归深度对RLM性能的影响，发现更深层次递归会导致模型过度思考，从而降低性能并增加计算成本。
摘要：
本项目重现并扩展了Zhang等人（2026）最近提出的"递归语言模型"（RLMs）框架。该框架使大型语言模型（LLMs）能够通过将提示卸载到外部REPL环境中处理接近无限的上下文。虽然原始论文依赖于默认递归深度为1，并建议更深层次的递归作为未来方向，本研究特别研究了递归深度扩展的影响。使用最先进的开源代理模型（DeepSeek v3.2和Kimi K2），我在S-NIAH和OOLONG基准上评估了纯LLM、RLM（深度=1）和RLM（深度=2）。研究发现一种引人注目的现象：更深层次的递归导致模型"过度思考"。虽然深度为1的RLM在复杂推理任务上有效提升准确性，但应用更深层次的递归（深度=2）或在简单检索任务上使用RLM反而会降级性能，并使执行时间呈指数级增长（例如，从3.6秒增加到344.5秒）和token成本。代码和数据可在：https://github.com/drbillwang/rlm-reproduction获取。
英文摘要：
This project reproduces and extends the recently proposed ``Recursive Language Models'' (RLMs) framework by Zhang et al. (2026). This framework enables Large Language Models (LLMs) to process near-infinite contexts by offloading the prompt into an external REPL environment. While the original paper relies on a default recursion depth of 1 and suggests deeper recursion as a future direction, this study specifically investigates the impact of scaling the recursion depth. Using state-of-the-art open-source agentic models (DeepSeek v3.2 and Kimi K2), I evaluated pure LLM, RLM (depth=1), and RLM (depth=2) on the S-NIAH and OOLONG benchmarks. The findings reveal a compelling phenomenon: Deeper recursion causes models to ``overthink''. While depth-1 RLMs effectively boost accuracy on complex reasoning tasks, applying deeper recursion (depth=2) or using RLMs on simple retrieval tasks paradoxically degrades performance and exponentially inflates execution time (e.g., from 3.6s to 344.5s) and token costs. Code and data are available at: https://github.com/drbillwang/rlm-reproduction
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。
