On Effects of Steering Latent Representation for Large Language Model Unlearning
This addresses the need for more effective and robust unlearning methods in large language models, though it appears incremental as it builds on existing representation steering techniques.
The paper tackles the problem of understanding and improving representation steering for large language model unlearning, showing that steering forget representations reduces token confidence and proposing Adaptive RMU to make unlearning effective across most layers, with experiments demonstrating significant performance improvements over prior methods.
Representation Misdirection for Unlearning (RMU), which steers model representation in the intermediate layer to a target random representation, is an effective method for large language model (LLM) unlearning. Despite its high performance, the underlying cause and explanation remain underexplored. In this paper, we theoretically demonstrate that steering forget representations in the intermediate layer reduces token confidence, causing LLMs to generate wrong or nonsense responses. We investigate how the coefficient influences the alignment of forget-sample representations with the random direction and hint at the optimal coefficient values for effective unlearning across different network layers. We show that RMU unlearned models are robust against adversarial jailbreak attacks. Furthermore, our empirical analysis shows that RMU is less effective when applied to the middle and later layers in LLMs. To resolve this drawback, we propose Adaptive RMU--a simple yet effective alternative method that makes unlearning effective with most layers. Extensive experiments demonstrate that Adaptive RMU significantly improves the unlearning performance compared to prior art while incurring no additional computational cost.