https://promptfoo.dev/ logo
For testing large files in RAG how would you sugge...
# ❓|questions
s
Trying to test our systems document summarizer using promptfoo. Our documents are fairly large. It feels like to do a good test I'd need to have the chunks and use those as input, but I might also need to chunk up the known good summary I want to compare against due to token limitations? Any guidance around what best practices would be for this?
i
if I understand correctly, one option is to hook it up to the actual RAG chunker you're using in your application. This example doesn't do chunking specifically but it shows how you can implement custom retrieval: . Hope it's helpful
s
Gets me closer at least. Thanks