RAG Eval Field Card — metrics cheat sheet (P@k / R@k / Hit@k)

You can chat with your docs and still have no idea whether retrieval improved.

This Field Card is a reusable reading layer for:

• Precision@k / Recall@k / Hit@k in plain English

• Common misreads (including why P@3 often sits near ~0.33 with one gold doc)

• How to compare chunk vs rerank without changing ten knobs at once

• A one-week practice plan + printable checklist

Honest limits:

• Educational / practice card

• Synthetic-style examples only

• Faithfulness stays N/A unless you run a real judge

• Not a production RAG platform and not an accuracy guarantee

Files:

• field-card.md

• checklist-one-pager.md

• LICENSE-NOTE.txt

Want a runnable stdlib Python harness too?

• Lite $2.99 — https://payhip.com/b/cmjJk

• Pro $11.99 — https://payhip.com/b/UdQ0M

Author: Dicardo9 / Moxuan Ding (I made this).