LLM Unlearning Without an Expert Curated Dataset
This addresses the need for scalable unlearning in AI to mitigate harmful content, though it is incremental by improving dataset generation rather than introducing a new unlearning paradigm.
The paper tackled the problem of removing sensitive or copyrighted knowledge from large language models without full retraining by automating the generation of forget sets using a structured prompting pipeline, showing that synthetic datasets outperform baseline alternatives and are comparable to expert-curated ones in unlearning domains like biosecurity and Harry Potter novels.
Modern large language models often encode sensitive, harmful, or copyrighted knowledge, raising the need for post-hoc unlearning-the ability to remove specific domains of knowledge from a model without full retraining. A major bottleneck in current unlearning pipelines is constructing effective forget sets-datasets that approximate the target domain and guide the model to forget it. In this work, we introduce a scalable, automated approach to generate high-quality forget sets using language models themselves. Our method synthesizes textbook-style data through a structured prompting pipeline, requiring only a domain name as input. Through experiments on unlearning biosecurity, cybersecurity, and Harry Potter novels, we show that our synthetic datasets consistently outperform the baseline synthetic alternatives and are comparable to the expert-curated ones. Additionally, ablation studies reveal that the multi-step generation pipeline significantly boosts data diversity, which in turn improves unlearning utility. Overall, our findings suggest that synthetic datasets offer a promising path toward practical, scalable unlearning for a wide range of emerging domains without the need for manual intervention. We release our code and dataset at https://github.com/xyzhu123/Synthetic_Textbook.