LLM在代码生成任务中能解人类无法理解的混乱题目，但可能依赖记忆而非推理。
LLM Performance for Code Generation on Noisy Tasks
- 通过噪声和删减手段测试模型对混乱代码任务的应对能力
- 所有模型在人类无法阅读的题目下仍能正确解题，但表现随数据污染程度下降
- 发现模型行为异常，提示训练数据污染与安全风险
本文研究大语言模型（LLMs）在被严重混淆的任务中识别与求解的能力。以编程竞赛和基准任务（LeetCode 和 MATH）为例，对比多种模型在噪声和红化等混淆方法下的表现。结果表明，所有评估的LLM均能在文本对人类完全不可读、关键指令或上下文缺失的情况下解题。我们提出‘急切模式匹配’概念描述该现象，其未出现在模型知识截止日期后发布的任务中，表明模型更依赖训练数据的记忆或过拟合，而非真实推理。报告了污染数据集与未见数据集间性能衰减模式的显著差异。讨论了对评测设计的影响，提醒在使用标准数据集时需谨慎。建议将性能在混淆下的衰减作为检测数据污染的指标，揭示自动化软件系统的潜在安全风险与可解释性问题。
原文摘要 · Abstract (English)
This paper investigates the ability of large language models (LLMs) to recognise and solve tasks which have been obfuscated beyond recognition. Focusing on competitive programming and benchmark tasks (LeetCode and MATH), we compare performance across multiple models and obfuscation methods, such as noise and redaction. We demonstrate that all evaluated LLMs can solve tasks obfuscated to a level where the text would be unintelligible to human readers, and does not contain key pieces of instruction or context. We introduce the concept of eager pattern matching to describe this behaviour, which is not observed in tasks published after the models' knowledge cutoff date, indicating strong memorisation or overfitting to training data, rather than legitimate reasoning about the presented problem. We report empirical evidence of distinct performance decay patterns between contaminated and unseen datasets. We discuss the implications for benchmarking and evaluations of model behaviour, arguing for caution when designing experiments using standard datasets. We also propose measuring the decay of performance under obfuscation as a possible strategy for detecting dataset contamination and highlighting potential safety risks and interpretability issues for automated software systems.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。