trust-align. Codes and datasets for the paper Measuring and Enhancing Trustworthiness of LLMs in RAG through Grounded Attributions and Learning to Refuse