Source: Notebook experiment 24: Self-RAG
Runnable scaffold of decide-to-retrieve, answer, grounding check, retry.
Contents
- What: runnable experiment in the owner’s public Notebook repo
- Author: Loic (coolloic)
- Date: undated (read 2026-11-14)
- Original: experiments/24-self-rag
Summary
A prompted approximation of Self-RAG: the model decides whether to retrieve, answers, checks whether the answer is grounded, and retrieves more and retries up to MAX_TRIES.
Key claims
- Reflection decisions (retrieve? grounded?) let the model gate its own retrieval and self-verify.
- Retries must be bounded.