LLM rerankers for production RAG: tips and tricks

(fin.ai)
4 points mathcircler | 2 comments | 15 Sep 25 10:02 UTC | HN request time: 0.469s | source
1. alexpivnenko ◴[15 Sep 25 10:51 UTC] No.45248225[source]▶
Surprised that removing spaces actually had such a big effect on latency.

Also props for including the prompt and AB results