- Rating
- 2
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
This paper develops a sample-wise early stopping method, aiming to improve the performance and efficiency of LLM unlearning. The proposed method is a directly plug-in for existing unlearning algorithms. Both performance and efficiency are observed in the experiment results.
Critical strengths
- The presentation is good, and the paper is well written. The paper is clear even for people without LLM unlearning background.
- In the experiments, multiple LLM unlearning algorithms are included. Overall, the empirical validations are quite strong. The proposed strategy both improves performance and training efficiency.
- Studying unlearning/safety training from a data pruning perspective is well motivated.
Critical weaknesses
- The proposed method relies on training the entire forget dataset, which is somewhat constrained. This is confined to LLM unlearning's application to removing data memorization (e.g., TOFU and MUSE), where you know the specific texts you want the model to forget. However, if we look at the WMDP benchmark, the forget set is merely a 'proxy' of the harmful knowledge (e.g., how to make Bio weapons) that we want to remove. In this case, I do not think the proposed method can work.
- Lack of large-scale and direct comparison with other early stopping methods. While this is validated in Figure 5, this is only a partial evaluation involving a single algorithm.
- The method requires selecting a hard threshold for each task and each algorithm, which involves heavy hyperparameter tuning.
- While the performance gain is clear, I do not think efficiency improvement for LLM unlearning is a strong motivation. After all, LLM unlearning is actually cheap in compute, where you usually optimize for fewer than 1000 steps.
Questions for authors
- In table 1, why not experiment with GD+SWES? Is this because GD and WGA perform similarly? Can SWES also be integrated with representation-level unlearning methods such as RMU? It seems that SWES is naturally more compatible with unlearning algorithms using a reference model?
- I do not see why this paper is specific to unlearning. Can it be extended to other tasks, like safety optimization, which also involves constrained optimization.