SineProject：面向稳定视觉语言对齐的机器遗忘方法
SineProject: Machine Unlearning for Stable Vision-Language Alignment
对做多模态大模型安全对齐的研究者，这篇论文指出遗忘训练会破坏投影器的雅可比条件数，并给出用正弦调制可训练参数稳定跨模态嵌入的轻量方案，可直接用于视觉语言模型的知识擦除与安全微调。
原文依据 augments the frozen projector with sinusoidally modulated trainable parameters that improve the Jacobian's spectral conditioning
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论多模态大语言模型（MLLMs）日益需要在不进行完整重训练的情况下遗忘特定知识，例如不安全或隐私信息。然而，现有的遗忘方法往往会破坏视觉语言对齐，导致模型同时拒绝有害查询和良性查询。我们将这一失败归因于投影器网络：在遗忘过程中，其雅可比矩阵变得严重病态，导致优化不稳定以及跨模态嵌入漂移。我们提出 SineProject，一种简单的方法，通过在冻结的投影器上增加正弦调制的可训练参数，改善雅可比矩阵的谱条件，并在整个遗忘过程中稳定对齐。在 LLaVA-v1.5-7B 和 13B 上，于标准安全与隐私遗忘基准进行评估，SineProject 在实现对目标信息完全遗忘的同时减少了良性查询的拒绝，以可忽略的计算开销实现了最优的遗忘-保留权衡。
英文原文摘要 1003 字符
Multimodal Large Language Models (MLLMs) increasingly need to forget specific knowledge, such as unsafe or private information, without full retraining. However, existing unlearning methods often disrupt vision-language alignment, causing models to reject both harmful and benign queries simultaneously. We trace this failure to the projector network: during unlearning, its Jacobian becomes severely ill-conditioned, leading to unstable optimization and drift in cross-modal embeddings. We introduce SineProject, a simple approach that augments the frozen projector with sinusoidally modulated trainable parameters that improve the Jacobian's spectral conditioning and stabilize alignment throughout unlearning. Evaluated across standard safety and privacy unlearning benchmarks using LLaVA-v1.5-7B and 13B, SineProject reduces benign-query refusals while achieving complete forgetting of targeted information, delivering state-of-the-art forget-retain trade-offs with negligible computational overhead