2025
From Misleading Queries to Accurate Answers: A Three-Stage Fine-Tuning Method for LLMs
ACL 2025finding
Large language models (LLMs) exhibit excellent performance in natural language processing (NLP), but remain highly sensitive to the quality of input queries, especially when these queries contain misleading or inaccurate information. Existing methods focus on correcting the output, but they often ov…