Description:
I was trying to build a RAG application where the retrieved context is HUGE. When I run it from the langserve playground, my browser crashes due to out of memory. I suspect it was because saving the stream log in the frontend was using to much memory (64 chunks with 2000 characters per chunk).
Steps to Reproduce:
Build a RAG application
Retrieve with a large context
Attempted Solution:
Reducing the number of chunks works, but I really need a huge number of chunks for my application.
Same problem here stream_logs seems to get a lot of events and then the browser tab crashes. I also retrieve with a huge retriever that is using a lot of llm calls and so on.
Description: I was trying to build a RAG application where the retrieved context is HUGE. When I run it from the langserve playground, my browser crashes due to out of memory. I suspect it was because saving the stream log in the frontend was using to much memory (64 chunks with 2000 characters per chunk).
Steps to Reproduce:
Attempted Solution: Reducing the number of chunks works, but I really need a huge number of chunks for my application.