RAG accuracy jumped from 10% to 60% when I added outcome scoring

by roampal • 7 months ago

11 7 7 months ago

mistrial9• 7 months ago

What is this kind of blog post? It is like advertising only, with urgent "install this code now" talk at the end. Impolite at best.. not great front page material IMHO

roampal3• 7 months ago

Fair point, the install instructions at the end were meant as a "here's how to try it if interested" but I can see how it reads as pushy. The core of the post is about the outcome scoring approach itself. Should've led with more depth on the methodology. Thanks for the feedback.

udfalkso• 7 months ago

It’s not pushy at all imo

ramenlover• 7 months ago

How did you measure the 60% improvement rate?

roampal3• 7 months ago

Ran a 4-way comparison test across 200 query-memory pairs:

- Baseline RAG (embedding similarity only): 10%

- RAG + reranker: 20%

- Outcomes only (no reranker): 60%

- RAG + outcome scoring (mature memories with 20+ uses): 60%

"Accuracy" = correct memory ranked #1 for the query. The outcome scoring uses Wilson score lower bound - memories that consistently get positive feedback from the "user" get boosted, ones that fail get demoted.

Test methodology: https://github.com/roampal-ai/roampal/blob/main/dev/benchmar...

realaleris149• 7 months ago

When I say "thanks, that worked," that memory gets promoted. When I say "no, that's wrong," it gets demoted. … No manual tagging.

I think this is also a kind of tagging.

roampal3• 7 months ago

You're right, it is a form of tagging technically. The difference is you're already saying "thanks that worked" or "nah that's wrong" anyway. No extra step, it just listens.

RAG accuracy jumped from 10% to 60% when I added outcome scoring

Related Stories