越流行，越难遗忘：面向大语言模型遗忘的自适应流行度

摘要

热门事实在预训练期间会被记忆得更深，并且比罕见事实更难被移除，然而现有的大语言模型遗忘方法无论训练数据频率如何，都会施加均匀的梯度压力。我们提出了 AdaPop（自适应流行度）方法，该方法将局部 token 置信度与由外部代理（例如 Wikidata 站点链接、LLM 作为裁判）导出的逐事实流行度相关指数相结合，并通过一个对偶上升控制器在每个 epoch 调整保留惩罚，从而自动平衡遗忘与保留。在三个模型家族和两个基准上，AdaPop 在释义查询下泄露的遗忘内容约为竞争方法的 1/5，在对抗性改写下约为竞争方法的 1/1.6。我们使用内部指标来支持这一分析：在我们的方法下，遗忘集的隐藏状态相比其他方法更远离遗忘前模型的状态，而保留集的表示仍与遗忘前模型的表示保持接近。

English

Popular facts are memorised more deeply during pretraining and resist removal longer than rare ones, yet existing LLM unlearning methods apply uniform gradient pressure regardless of training-data frequency. We propose the AdaPop (Adaptive Popularity) method, which combines local token confidence with a per-fact popularity-dependent exponent derived from an external proxy (e.g., Wikidata sitelinks, LLM-as-Judge), and automates the forget-retain balance via a dual-ascent controller that adjusts the retain penalty each epoch. Across three model families and two benchmarks, AdaPop leaks ~5x less forgotten content than competing methods under paraphrased queries and ~1.6x less under adversarial reformulations. We support our analysis with internal metrics: under our method, forget-set hidden states move further from the pre-unlearning model's states than under other methods, while retain-set representations remain close.