youtu-graphrag
youtu-graphrag copied to clipboard
修改提示词
如何修改提示词能让回答更准确?如果在知识问答阶段,无法在图谱中检索到相关内容,可以直接调用本地大模型进行回答吗?
In our paper and code, we have considered two response modes in Question-and-Answer: open& reject. In the open mode, the LLMs are allowed to answer questions based on the retrieved content. If no effective information is found in the retrieved content, LLMs are permitted to respond using their own knowledge.
这个项目是不是不建议使用比deepseek3低的,参数少的模型,我使用了其他低的模型,修改提示词之后还是得不到想要的回复