Yingfeng
Yingfeng
1. Yes 2. rrf does not have any relationship with the relevance scores of each recall, but weighted_sum has. The final result of weighted_sum is decided by match_fulltext and match_dense....
> > Yes, if you use `match_tensor` outside of `fusion`, it means to use tensor index to perform the tensor recall. The tensor index is a pure memory solution and...
Could you provide the PDF leading to this segmentation fault error? Or else it's difficult to reproduce such an error.
Then could you provide some data leading to such error? We've never seen it before, it's caused by some third-party dependencies and could not be fixed without reproducing. Thank you...
> How about supporting multi-modal RAG? Under planning, it requires Infinity to be production ready such that Tensor based reranker as well as multi vector retrieval could be available.
SSO/IAM belongs to features of enterprise edition and will not be delivered in the open source edition.
Choose Table as the parsing template, then I've got the above result. Forget about the Chinese characters got by LLMs. One issue is that, the embedding process is pretty long....
It does not relate to the LLM/embedding closely. The key issue is you should choose `Table` instead of default `General` as the file parser template
You could try to adjust the prompt. By default, demo adopts deepseek.