Hi @blah . Thanks for your work. I recently started playing around with n8n and am trying to build a decent assistant with RAG, including reranking. I found an info about n8n-nodes-query-retriever-rerank community node and decided to use it. For bigger volume vector stores it is a real game changer! I think this node should become an official n8n node!
One comment about its reaction to Timeout on LLM node. I am running an extremely slow setup on local laptop with CPUs only and my LLM node takes more than 60 seconds to answer. I got “Request timed out” every time reranker node used LLM and did not know how to solve it until I realized about this Timeout option. If it is somehow possible to inform better the n8n user about it, then maybe this could be an improvement.
Debugging and LLM Debug analysis options are amazing too! Incredibly helpful in further optimizations.