[ACL'25 Main] SelfElicit: Your Language Model Secretly Knows Where is the Relevant Evidence! | 让你的LLM更好地利用上下文文档：一个基于注意力的简单方案 - View it on GitHub