# Your Retriever Already Knows: Distribution-Shape QPP for RAG Retrieval Sufficiency

arXiv: https://arxiv.org/abs/2609.11646 · HTML: https://mlsanity.com/paper/2609.11646/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in RAG robustness and adaptive retrieval, one of 15 directions in Retrieval-augmented generation.
- Builds on 5 of its 21 references that are in the map.
- First posted on arXiv 10 Sep 2026.

## Placed in

- Retrieval-augmented generation / [RAG robustness and adaptive retrieval](https://mlsanity.com/rag/rag-robustness-and-adaptive-retrieval/)

## Builds on (most cited first, 5 of 5)

- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection](https://mlsanity.com/paper/2310.11511/) (2023)
- [ColPali: Efficient Document Retrieval with Vision Language Models](https://mlsanity.com/paper/2407.01449/) (2024)
- [Query Performance Prediction for Neural IR: Are We There Yet?](https://mlsanity.com/paper/2302.09947/) (2023)
- [Uncovering the Limitations of Query Performance Prediction: Failures, Insights, and Implications for Selective Query Processing](https://mlsanity.com/paper/2504.01101/) (2025)

## Abstract (arXiv)

Standard Retrieval-Augmented Generation (RAG) pipelines often provide no reliable inference-time signal of whether retrieval succeeded; on ambiguous or out-of-scope queries, generation may then hallucinate. Motivated by a Czech nuclear-regulator deployment where data sensitivity precludes third-party LLM APIs, we compare three Query Performance Prediction (QPP) paradigms for retrieval sufficiency in RAG: score-based features, a content-based LLM judge, and a hybrid. On the eight ViDoRe vision domains (14,514 queries), our 24 non-lexical features (GeneralQPP; 15 distribution-shape, 5 query-surface, 4 global) reach a weighted-average AUROC of 0.856 at 2 ms per query, ahead of a classic-QPP literature pool (Classic Full, 0.835) and well above a local Qwen3.5 LLM judge (0.649, +0.207 gap; ~3000× faster and cheaper per query). Adding the LLM judgment as one feature (hybrid) matches S1 on ViDoRe (0.863) but gains a statistically significant edge on S'UJB (AUROC 0.911 at Hit@5, adversarial-detection 0.954; 1,510 queries, 500 synthetic adversarial), at LLM latency. Rankings agree across datasets (Spearman ρ = 0.90). Under Leave-One-Domain-Out, S1 drops to 0.706; a 13-feature LODO-stepwise subset (S1-Lean) recovers to 0.719 (+0.032 over the literature pool).