MO-IKE

New RL algorithm MO-IKE enhances LLM knowledge editing

Researchers have developed a new multi-objective reinforcement learning algorithm called MO-IKE to improve in-context knowledge editing for large language models. This method addresses limitations in previous approaches…