提出一种精准抑制大模型记忆的动态干预方法
GSS: Gated Subspace Steering for Selective Memorization Mitigation in LLMs
- 通过探测器+调节器分步干预，只在需要时修正记忆内容
- 在四个数据集上减少记忆率，计算量仅为现有方法的1/100至1/1000
- 适合关注模型隐私与泛化能力的研究者使用
大语言模型会原样复现训练数据中的序列，这损害了泛化能力和隐私安全。现有方法对所有词元统一干预，导致正常泛化性能下降。我们实证发现记忆行为稀疏、间断且依赖上下文，因此需要基于上下文的动态干预而非静态参数修改。为此，我们提出一种新型选择性记忆抑制方法——门控子空间引导（GSS），将干预分解为探测器（识别记忆相关激活）和调节器（仅当探测值超过阈值时施加修正）。最优探测-调节组合由基于最优子空间引导的原理性优化框架得出。在四个基准测试中，GSS在记忆抑制效果上达到或超越当前最优水平，同时计算开销仅为优化类方法的100到1000倍。此外，我们还揭示了神经表征中记忆几何结构的新理论见解。
原文摘要 · Abstract (English)
Large language models (LLMs) can memorize and reproduce training sequences verbatim -- a tendency that undermines both generalization and privacy. Existing mitigation methods apply interventions uniformly, degrading performance on the majority of tokens that generalize normally. We show empirically that memorization is sparse, intermittent, and token-conditioned, suggesting that effective mitigation requires context-aware intervention rather than static parameter modification. To this end, we propose a novel and effective selective memorization mitigation method -- Gated Subspace Steering (GSS), which decomposes intervention into a probe (detecting memorization-relevant activations) and a steer (applying targeted correction only when the probe exceeds a threshold). The optimal probe-steer pair emerges from a principled optimization framework based on optimal subspace steering. Experiments on four benchmarks show GSS matches or exceeds state-of-the-art memorization reduction while requiring $100-1000 \times$ less compute than optimization-based alternatives. Furthermore, we provide new theoretical insights into the geometry of memorization in neural representations.
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。