I really enjoy the video, especially the update on the text-base RAG. A nice tip I got from that is to use LLMs to generate some questions so that we can better retrieve the documents. This helps to mitigate issues where query embeddings and document embeddings don't aligned or don't belong in the same embedding space.
I really enjoy the video, especially the update on the text-base RAG. A nice tip I got from that is to use LLMs to generate some questions so that we can better retrieve the documents. This helps to mitigate issues where query embeddings and document embeddings don't aligned or don't belong in the same embedding space.