Not Every Token Needs Forgetting: Selective Unlearning to Limit Change in Utility in Large Language Model Unlearning
This addresses the need to remove private or copyrighted content from LLMs with minimal utility loss, representing an incremental improvement over existing unlearning methods.
The paper tackles the problem of large language model unlearning by proposing Selective Unlearning, which identifies and forgets only critical tokens related to unwanted information, rather than all tokens. Experiments show it effectively unlearns target data while preserving model utility on retaining sets.
Large Language Model (LLM) unlearning has recently gained significant attention, driven by the need to remove unwanted information, such as private, sensitive, or copyrighted content, from LLMs. However, conventional unlearning approaches indiscriminately update model parameters to forget all tokens in a target document, including common tokens (e.g., pronouns, prepositions, general nouns) that carry general knowledge. In this paper, we highlight that not every token needs forgetting. We propose Selective Unlearning (SU), which identifies a critical subset of tokens within the forgetting set that is relevant to the unwanted information, and unlearns only those tokens. Experiments on two benchmarks and six baseline unlearning algorithms demonstrate that SU not only achieves effective unlearning on the targeted forget data, but also significantly preserves the model's utility in the retaining set.