GSS: Gated Subspace Steering for Selective Memorization Mitigation in LLMs
This addresses privacy and generalization issues in LLMs for AI safety and deployment, though it is an incremental improvement over existing mitigation methods.
The paper tackles the problem of LLMs memorizing training sequences, which harms generalization and privacy, by proposing Gated Subspace Steering (GSS) for selective mitigation, achieving state-of-the-art memorization reduction with 100-1000x less compute than alternatives.
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.