约束熵遗忘：大型语言模型的原偶框架
Constrained Entropic Unlearning: A Primal-Dual Framework for Large Language Models
AI总结：
本文提出一种基于约束优化和原偶算法的LLM遗忘框架，通过logit边际展平损失强制遗忘并硬约束保留，在TOFU和MUSE基准上达到或超越最先进水平。
摘要：
在真实环境中部署的大型语言模型（LLMs）日益面临遗忘敏感、过时或专有信息的需求。现有的遗忘方法通常将遗忘和保留表述为正则化的权衡，将两个目标合并为单一的标量化损失。这常常导致优化不稳定，并在保留数据上性能下降，尤其是在激进遗忘的情况下。我们提出了一种将LLM遗忘表述为约束优化问题的新方法：通过一种新颖的logit边际展平损失来强制遗忘，该损失明确地将指定遗忘集上的输出分布推向均匀分布，同时通过一个对独立保留集的硬约束来保持保留能力。与基于熵的目标相比，我们的损失函数无需softmax，数值稳定，并保持非消失梯度，从而实现更高效和稳健的优化。我们使用一种可扩展的原偶算法来解决约束问题，该算法通过偶变量的动态揭示了遗忘与保留之间的权衡，且无需任何额外的计算开销。在TOFU和MUSE基准上，跨多种LLM架构的评估表明，我们的方法始终达到或超过最先进的基线，有效移除目标信息，同时保留下游实用性。
英文摘要：
Large Language Models (LLMs) deployed in real-world settings increasingly face the need to unlearn sensitive, outdated, or proprietary information. Existing unlearning methods typically formulate forgetting and retention as a regularized trade-off, combining both objectives into a single scalarized loss. This often leads to unstable optimization and degraded performance on retained data, especially under aggressive forgetting. We propose a new formulation of LLM unlearning as a constrained optimization problem: forgetting is enforced via a novel logit-margin flattening loss that explicitly drives the output distribution toward uniformity on a designated forget set, while retention is preserved through a hard constraint on a separate retain set. Compared to entropy-based objectives, our loss is softmax-free, numerically stable, and maintains non-vanishing gradients, enabling more efficient and robust optimization. We solve the constrained problem using a scalable primal-dual algorithm that exposes the trade-off between forgetting and retention through the dynamics of the dual variable, all without any extra computational overhead. Evaluations on the TOFU and MUSE benchmarks across diverse LLM architectures demonstrate that our approach consistently matches or exceeds state-of-the-art baselines, effectively removing targeted information while preserving downstream utility.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
