针对掩码扩散语言模型的机器去学习
Machine Unlearning for Masked Diffusion Language Models
AI总结：
本文提出了一种针对掩码扩散语言模型的去学习框架MDU，通过重新审视扩散过程中的知识学习，实现了高效的去学习性能。
摘要：
最近的掩码扩散语言模型（MDLMs），如LLaDA和Dream，已经达到了与自回归大语言模型相当的性能。与自回归模型不同，MDLMs通过并行迭代去噪掩码位置来生成文本。在微调过程中，MDLMs学习从掩码响应状态中恢复响应，从而将预测从提示-掩码无条件分布转向提示-条件分布。尽管生成和微调机制不同，针对MDLMs的机器去学习仍鲜有研究。在本文中，我们提出Masked Diffusion Unlearning（MDU），通过重新审视扩散过程中的知识学习，首次提出了针对MDLMs的去学习框架。具体而言，MDU在每个掩码响应位置上最小化从提示-条件预测到提示-掩码无条件锚点的正向KL散度，并通过温度缩放参数控制隐私-效用权衡。在标准基准和MDLM骨干网络上的实验证明，MDU在与现有LLM去学习方法相比时实现了较高的去学习性能。代码可在https://github.com/leegeoru/MDU上获得。
英文摘要：
Recent masked diffusion language models (MDLMs), such as LLaDA and Dream, have achieved performance comparable to autoregressive large language models. Unlike autoregressive models, which generate text sequentially, MDLMs generate text by iteratively denoising masked positions in parallel. During fine-tuning, MDLMs learn to recover responses from masked response states conditioned on a prompt, thereby shifting their predictions from a prompt-masked unconditional distribution toward a prompt-conditional distribution. Despite this distinct generative and fine-tuning mechanism, machine unlearning for MDLMs remains largely unexplored. In this paper, we propose Masked Diffusion Unlearning (MDU), the first unlearning framework for MDLMs, by revisiting the process of learning specific knowledge in terms of diffusion. Specifically, MDU minimizes a forward KL divergence from the prompt-conditional prediction to a prompt-masked unconditional anchor at every masked response position, with a temperature scaling parameter to control the privacy-utility trade-off. Our empirical results on standard benchmarks and MDLM backbones show that MDU achieves high unlearning performance compared to existing LLM unlearning methods. Code is available at https://github.com/leegeoru/MDU.
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。
