让微调模型自述隐藏行为，更可靠地发现其潜在问题。
Revealing Hidden Model Behaviors with Task-Specific Self-Reports
- 用轻量级LoRA适配器让模型用自然语言描述自身隐藏行为。
- 在7种植入行为中全部检测到，且误报率比基线低一半。
- 适合需要审计模型微调后真实学习内容的开发者和研究者。
微调可能使语言模型产生隐藏行为——在特定条件下给出错误答案，或仅在触及某话题时提供有害建议。我们提出稳定适配器自报告（SAR），一种轻量级LoRA适配器，仅使用模型和训练数据即可让微调后的模型用自然语言描述自身隐藏行为。在七种植入行为中，SAR全部成功检测出隐藏行为，即使模型已泛化为广泛偏差，训练数据也无法预测。与最接近的基线——内省适配器（IA）相比，后者漏检部分行为且会持续编造错误行为，而SAR在所有失败场景中仍保留有效信号，误报率约降低一半。这为从业者提供了更可靠的工具，用于审计微调模型并回答‘它实际学到了什么’等问题。
原文摘要 · Abstract (English)
Fine-tuning can give a language model a hidden behavior--it may give false answers under a narrow condition, or give harmful advice only when a prompt touches a particular topic. We introduce the Stabilized Adapter for self-Report (SAR), a lightweight LoRA adapter that makes a fine-tuned model describe its own hidden behavior in plain language, using only the model and the dataset it was trained on. Across seven implanted behaviors, SAR detects the hidden behavior in every one--even when the model has generalized into broad misalignment that the training data alone does not predict. Introspection Adapters (IA), the closest existing baseline, detects some behaviors from our suite but misses others entirely--and where it misses, it hallucinates, consistently reporting wrong behaviors. SAR retains positive signal on every setting where IA fails and roughly halves the rate of hallucinations. This gives practitioners a more reliable tool to audit a fine-tuned model and answer ``what did it actually learn?'' type of questions.
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。