Ethos：在正交参数空间中矫正语言模型
Ethos: Rectifying Language Models in Orthogonal Parameter Space
AI总结：
本文提出 Ethos 方法，通过在正交参数空间中利用奇异值分解区分一般知识与不良知识，并仅对不良知识任务向量进行否定，从而在去偏、去毒和记忆消除任务中更有效地移除不良知识并保持模型整体性能。
摘要：
语言模型（LMs）极大地推动了自然语言处理的研究。然而，语言模型也引发了关于生成有偏见或有毒内容以及可能泄露训练数据集中私人信息的担忧。在这项工作中，我们提出了一种新的高效方法 Ethos，用于矫正语言模型，以减轻输出中的毒性和偏见并避免隐私泄露。Ethos 建立在任务算术（task arithmetic）之上。然而，与当前的任务算术算法不同，Ethos 在重构任务向量时区分一般有益知识和不良知识。具体而言，Ethos 首先使用奇异值分解从预训练模型中获得一组主成分。然后，通过将任务向量投影到主成分上，Ethos 识别出编码一般知识或不良知识的主成分。Ethos 仅使用包含不良知识的任务向量进行否定操作，从而最大限度地减少对模型整体效用的附带损害。我们在三个不同的任务上展示了该方法的有效性：去偏、去毒和记忆消除。评估表明，与当前的任务算术方法相比，Ethos 在移除不良知识和保持整体模型性能方面更为有效。
英文摘要：
Language models (LMs) have greatly propelled the research on natural language processing. However, LMs also raise concerns regarding the generation of biased or toxic content and the potential disclosure of private information from the training dataset. In this work, we present a new efficient approach, Ethos, that rectifies LMs to mitigate toxicity and bias in outputs and avoid privacy leakage. Ethos is built on task arithmetic. However, unlike current task arithmetic algorithms, Ethos distinguishes general beneficial and undesired knowledge when reconstructing task vectors. Specifically, Ethos first obtains a set of principal components from the pre-trained models using singular value decomposition. Then, by projecting the task vector onto principal components, Ethos identifies the principal components that encode general or undesired knowledge. Ethos performs negating using the task vector with undesired knowledge only, thereby minimizing collateral damage on general model utility. We demonstrate the efficacy of our approach on three different tasks: debiasing, detoxification, and memorization unlearning. Evaluations show Ethos is more effective in removing undesired knowledge and maintaining the overall model performance compared to current task arithmetic methods.
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。
