揭示机器遗忘中的潜在风险：多语言大语言模型中仅用英语进行遗忘的危险
Uncovering the Potential Risks in Unlearning: Danger of English-only Unlearning in Multilingual LLMs
AI总结：
本研究聚焦多语言LLMs仅用英语遗忘的盲点，发现全量微调后遗忘会出现语言混淆问题，提出N-Mix分数量化该现象，证明参考类指标会失效，呼吁开发基于语义的新型遗忘评估指标。
摘要：
已有多项研究表明，仅使用英语数据尝试擦除多语言知识，对多语言大语言模型（LLMs）而言是不够的，但这些研究的分析高度以性能为导向。本文将视角转向评估，指出了一个额外的盲点：当多语言LLM在进行遗忘操作前，已用平行多语言数据集完成全量微调时，该盲点便会显现。此时会出现语言混淆现象，即模型用与输入提示不同的语言作出回应。语言混淆是遗忘过程中的一个问题现象，会导致标准的基于参考的指标失效。我们分三步研究这一现象：（1）提出基于N元语法的语言混合（N-Mix）分数，以量化证明语言混淆在多语言LLMs中普遍存在且具有一致性；（2）证明当N-Mix分数较高时，基于参考的指标会产生假阴性结果；（3）提出需要一类新型的遗忘评估指标，可直接评估生成句子的内容，我们将这类指标称为基于语义的指标。
英文摘要：
There have been a couple of studies showing that attempting to erase multilingual knowledge using only English data is insufficient for multilingual LLMs. However, their analyses remain highly performance-oriented. In this paper, we switch the point of view to evaluation, and address an additional blind spot which reveals itself when the multilingual LLM is fully finetuned with parallel multilingual dataset before unlearning. Here, language confusion occurs whereby a model responds in language different from that of the input prompt. Language confusion is a problematic phenomenon in unlearning, causing the standard reference-based metrics to fail. We tackle this phenomenon in three steps: (1) introduce N-gram-based Language-Mix (N-Mix) score to quantitatively show the language confusion is pervasive and consistent in multilingual LLMs, (2) demonstrate that reference-based metrics result in false negatives when N-Mix score is high, and(3) suggest the need of new type of unlearning evaluation that can directly assess the content of the generated sentences. We call this type of metrics as semantic-based metric.
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。
