Beyond Self-Knowledge: Propagating Uncertainty Across Reasoning and Retrieval in LLMs

• URL: http://arxiv.org/abs/2607.25600v2
• Date: Wed, 29 Jul 2026 07:55:33 GMT
• Title: Beyond Self-Knowledge: Propagating Uncertainty Across Reasoning and Retrieval in LLMs
• Abstract summary: Verbalized confidence from black-box language models can serve as an actionable signal for retrieval routing.<n>We evaluate 27,000 policy instances across six QA benchmarks, three model families, and three retrieval policies.
• Score: 4.476374205849672
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Retrieval-augmented generation improves knowledge-intensive question answering, but indiscriminate retrieval can introduce irrelevant evidence and unnecessary computation. We investigate whether verbalized confidence from black-box language models can serve as an actionable signal for retrieval routing. Our method, BeyondUncertainty, first elicits a structured provisional answer and confidence estimate, then applies a model-specific threshold selected on held-out validation data and frozen before test evaluation. Low-confidence questions receive top-5 TF-IDF retrieval followed by a second answer call, whereas high-confidence questions return the provisional answer directly. We evaluate 27,000 policy instances across six QA benchmarks, three model families, and three retrieval policies. BeyondUncertainty achieves 0.483 mean token-level F1, compared with 0.467 for always retrieval and 0.401 for no retrieval, while reducing retrieved passages by 20.4\% relative to always retrieval. When matched on the number of questions routed to retrieval within each dataset-model cell, it outperforms a post-hoc random allocation in 17 of 18 settings, with an average gain of 0.024 F1. Although poorly calibrated as an absolute probability, probe uncertainty modestly predicts question-level retrieval benefit (AUROC = 0.628). However, the additional probe increases total token usage by 28.2\%, revealing a trade-off between more selective evidence acquisition and end-to-end token efficiency.

Related papers

• Evaluating Confidence-Gated Retrieval with Matched Trajectory Replay [3.888537003098864]
  We propose matched trajectory replay, a controlled protocol for comparing confidence-to-action mappings.<n>We use it to compare raw verbalized confidence with post-hoc isotonic calibration in a multi-hop question-answering system.<n>Overall accuracy improves by up to 15 percentage points on HotpotQA but falls by up to 17 percentage points on MuSiQue.
  arXiv  Detail & Related papers  (2026-08-27T09:15:33Z)
• Confidently Wrong: Detecting Hallucinations in Financial Question Answering from LLM Internal States [0.0]
  Hedged, uncertain answers invite scrutiny, whereas confident errors silently degrade downstream decisions without warning.<n>We ask how reliably such confidently wrong answers, or confident hallucinations, can be detected from a model's internal activations.<n>We train linear probes on the residual stream and evaluate them on two established question-answering benchmarks.
  arXiv  Detail & Related papers  (2026-07-13T11:22:25Z)
• Conditional Inference Trees and Forests for Feature Selection [0.0]
  Conditional inference trees (CIT) and conditional inference forests (CIF) reduce split-selection bias by testing features before choosing split thresholds.<n>We study CIT and CIF as top-$k$ feature-ranking methods for downstream prediction using real-data benchmarks, runtime ablations, and synthetic feature-recovery experiments.
  arXiv  Detail & Related papers  (2026-07-01T19:23:07Z)
• When Does Learning to Stop Help? A Cost-Aware Study of Early Exits in Reasoning Models [4.8190992438931035]
  LearnStop is a hidden-state-free logistic stopper over prefix-observable features.<n>A trajectory decomposition predicts the regime: learning pays where answers oscillate and correctness evidence is spread.<n>Cost accounting sharpens the picture further -- the same policy that saves 32% of tokens under KV-cache forking costs 121% extra under black-box repeated prefilling.
  arXiv  Detail & Related papers  (2026-06-29T19:33:42Z)
• Know Before You Fetch: Calibrated Retrieval-Budget Allocation for Retrieval-Augmented Generation [10.60293484295824]
  We formulate adaptive RAG as calibrated retrieval-budget allocation.<n>We calibrate sequence log-probability and prefix-logit uncertainty signals into probabilities of correctness.<n>For sequence log-probability, ECE drops from 0.275 to 0.062 on TriviaQA, 0.643 to 0.009 on NQ, and 0.711 to 0.031 on MS MARCO.
  arXiv  Detail & Related papers  (2026-06-29T08:36:33Z)
• AB-RAG: Adaptive Budgeted Retrieval-Augmented Generation for Reliable Question Answering [0.0]
  AB-RAG is a training-free and backbone-agnostic framework that generates an answer.<n>It estimates its confidence from a combination of three signals.<n>It then decides whether to stop or to retrieve more evidence.
  arXiv  Detail & Related papers  (2026-06-27T21:08:14Z)
• Towards Dependable Retrieval-Augmented Generation Using Factual Confidence Prediction [0.28932261919131014]
  It is critical to associate meaningful confidence measures about the factuality of the retrieval process with the generated answers.<n>We present a new, two-staged approach to predict fact faithfulness of the output of retrieval-augmented generations.<n>Our work helps to establish a novel type of certified RAG systems for a broad range of natural language industry applications.
  arXiv  Detail & Related papers  (2026-05-04T11:28:19Z)
• Confidence Before Answering: A Paradigm Shift for Efficient LLM Uncertainty Estimation [47.91529693614168]
  Existing methods are predominantly answer-first, producing confidence only after generating an answer.<n>We study a confidence-first paradigm, where the model outputs its confidence before answering, interpreting this score as the model's probability of answering the question correctly.<n>We propose CoCA, a reinforcement learning framework that jointly optimize confidence calibration and answer accuracy via segmented credit assignment.
  arXiv  Detail & Related papers  (2026-03-06T04:03:13Z)
• Know When You're Wrong: Aligning Confidence with Correctness for LLM Error Detection [0.0]
  Large language models (LLMs) are increasingly deployed in critical decision-making systems.<n>We introduce a normalized confidence score based on output anchor token probabilities.<n>This enables direct detection of errors and hallucinations with minimal overhead.
  arXiv  Detail & Related papers  (2026-02-18T07:05:12Z)
• Reference-Free Rating of LLM Responses via Latent Information [53.463883683503106]
  We study the common practice of asking a judge model to assign Likert-scale scores to free-text responses.<n>We then propose and evaluate Latent Judges, which derive scalar ratings from internal model signals.<n>Across a broad suite of pairwise and single-rating benchmarks, latent methods match or surpass standard prompting.
  arXiv  Detail & Related papers  (2025-09-29T12:15:52Z)
• Eigen-1: Adaptive Multi-Agent Refinement with Monitor-Based RAG for Scientific Reasoning [53.45095336430027]
  We develop a unified framework that combines implicit retrieval and structured collaboration.<n>On Humanity's Last Exam (HLE) Bio/Chem Gold, our framework achieves 48.3% accuracy.<n>Results on SuperGPQA and TRQA confirm robustness across domains.
  arXiv  Detail & Related papers  (2025-09-25T14:05:55Z)
• Mitigating LLM Hallucinations via Conformal Abstention [70.83870602967625]
  We develop a principled procedure for determining when a large language model should abstain from responding in a general domain. We leverage conformal prediction techniques to develop an abstention procedure that benefits from rigorous theoretical guarantees on the hallucination rate (error rate) Experimentally, our resulting conformal abstention method reliably bounds the hallucination rate on various closed-book, open-domain generative question answering datasets.
  arXiv  Detail & Related papers  (2024-04-04T11:32:03Z)
• Llamas Know What GPTs Don't Show: Surrogate Models for Confidence Estimation [70.27452774899189]
  Large language models (LLMs) should signal low confidence on examples where they are incorrect, instead of misleading the user. As of November 2023, state-of-the-art LLMs do not provide access to these probabilities. Our best method composing linguistic confidences and surrogate model probabilities gives state-of-the-art confidence estimates on all 12 datasets.
  arXiv  Detail & Related papers  (2023-11-15T11:27:44Z)
• Conservative Prediction via Data-Driven Confidence Minimization [70.93946578046003]
  In safety-critical applications of machine learning, it is often desirable for a model to be conservative. We propose the Data-Driven Confidence Minimization framework, which minimizes confidence on an uncertainty dataset.
  arXiv  Detail & Related papers  (2023-06-08T07:05:36Z)
• Improving Passage Retrieval with Zero-Shot Question Generation [109.11542468380331]
  We propose a simple and effective re-ranking method for improving passage retrieval in open question answering. The re-ranker re-scores retrieved passages with a zero-shot question generation model, which uses a pre-trained language model to compute the probability of the input question conditioned on a retrieved passage.
  arXiv  Detail & Related papers  (2022-04-15T14:51:41Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.