Enhancing the fine-grained and coarse-grained knowledge awareness of large language models to improve their performance on complex knowledge-intensive question answering tasks.
