This chapter explores other methods for Large Language Model (LLM) interaction, focusing on Active Learning, Imitation Learning, Model Surgery, and Interaction Message Fusion. Active Learning is discussed as a method for efficiently selecting and labeling data to enhance model performance, with a focus on query strategies and oracles. Imitation Learning is presented as an approach where models learn by mimicking expert behavior, with applications in both offline and online settings. Model Surgery is introduced as a suite of post-training targeted interventions that enable precise and efficient modification of an LLM’s knowledge or behavior–such as updating, correcting, or removing specific knowledge–without the need for full retraining. This includes techniques like knowledge editing and machine unlearning. The chapter also introduces Interaction Message Fusion, a comprehensive framework for integrating external information into language models, categorized along three dimensions: data, model, and training. Each method is examined for its potential to improve the adaptability and effectiveness of language models in interactive scenarios.