Cautious Context Steering for Efficient LLM Personalization

Cautious Context Steering (CCS) introduces a lightweight adapter to personalize LMs via in-context learning without the overhead of two forward passes per step. This method addresses data sparsity and generalization issues found in traditional per-user adapter training or fixed-coefficient steering.