infiniflow
[Question]: Why does Ragflow keep saying the knowledge base is empty?
Self Checks
- [x] I have searched for existing issues search for existing issues, including closed ones.
- [x] I confirm that I am using English to submit this report (Language Policy).
- [x] Non-english title submitions will be closed directly ( 非英文标题的提交将会被直接关闭 ) (Language Policy).
- [x] Please do not modify this template :) and fill in all the required fields.
Describe your problem
I have created a new knowledge base and added a document to it. All models are well configured. But when I create an assistant with the knowledge base selected, and ask something related to the document in the chat, it always says the knowledge base is empty. Does anyone have the same problem?
文档是否已成功解析/分块？
I have encountered the same problem. The file has been successfully parsed, the fragments have been divided, and the knowledge base has been set up in the "Assistant Settings" section. However, when asked about the large model, it still prompts that it cannot find the relevant content of the knowledge base. Sometimes it simply says that the knowledge base is empty.
文档是否已成功解析/分块？
I have encountered the same problem. The file has been successfully parsed, the fragments have been divided, and the knowledge base has been set up in the "Assistant Settings" section. However, when asked about the large model, it still prompts that it cannot find the relevant content of the knowledge base. Sometimes it simply says that the knowledge base is empty.
Exactly. Sometimes the feedback of Ragflow only cover very general terms or concepts of the knowledge base, without touching any specific content that is really related to my question.
It seems that all the problems start from version 0.17.2. When I was using 0.17.1, everything worked quite well.
Have the documents been successfully parsed/chunked?
Now I get something better, after reducing the chunk size from 256 to 128 or much lower to 50, it is improving now. It looks like something related to the embedding model. In my case, I set all the models from Ollama and the embedding is nomic-embed-text. What should be the Max Tokens for this model?
Have the documents been successfully parsed/chunked?
sometimes the parsing process gets stuck at the beginning, any way to resolve this problem?