Confidence Comes from Experience: Experiential Confidence Estimation from Reasoning to Agents

Hugging Face Daily Papers Papers
confidence-estimation language-models trustworthy-ai reasoning agents benchmarking

Summary

The paper introduces XConf, an experiential confidence estimation method that leverages a model's past experiences to enhance confidence calibration in language models across reasoning, coding, and agent tasks.

Reliable confidence estimation is increasingly central to the trustworthy deployment of language models: a calibrated estimate of the probability that an output is correct decides what to ship, what to escalate, and what to retry. Existing confidence estimators, however, share one design premise: they only read the current inference process, either by introspecting on it, scoring its token probabilities, or resampling it. We argue that the current inference is not a sufficient basis for confidence. We propose XConf (eXperiential Confidence): estimating confidence together with the model's accumulated experience. The experience is stored as a record of the model's own graded past episodes, each holding the task, the model's reflection, its stated confidence, the outcome, and a lesson written once the grade arrived. Given a new task, XConf's Recall stage retrieves past episodes on similar tasks met with a similar stated confidence, and reads off their historical success rate; its Reflect stage shows the model this record, has it name its recurring failure mode, and restate a confidence now informed by its own track records. Our estimator is format-general, requiring no logit access or weight updates, and costs only one answer generation. Across nine benchmarks spanning reasoning, coding, multimodal QA, and interactive agents, and four models from three families, XConf beats or matches ten-sample self-consistency in discrimination (AUROC) on 23 of 24 comparisons, with much lower calibration error (ECE), at a tenth of the generation cost. Used for selective prediction, abstaining on the 10% least-confident episodes raises the delivered success rate by up to 8.7 points on agent tasks. We therefore see experiential confidence estimation as a new paradigm for future general-purpose confidence estimation.
Original Article
View Cached Full Text

Cached at: 09/17/26, 06:52 AM

Paper page - Confidence Comes from Experience: Experiential Confidence Estimation from Reasoning to Agents

Source: https://huggingface.co/papers/2609.17708

Abstract

Reliableconfidenceestimationisincreasinglycentraltothetrustworthydeploymentoflanguagemodels:acalibratedestimateoftheprobabilitythatanoutputiscorrectdecideswhattoship,whattoescalate,andwhattoretry.Existingconfidenceestimators,however,shareonedesignpremise:theyonlyreadthecurrentinferenceprocess,eitherbyintrospectingonit,scoringitstokenprobabilities,orresamplingit.Wearguethatthecurrentinferenceisnotasufficientbasisforconfidence.WeproposeXConf(eXperientialConfidence):estimatingconfidencetogetherwiththemodel’saccumulatedexperience.Theexperienceisstoredasarecordofthemodel’sowngradedpastepisodes,eachholdingthetask,themodel’sreflection,itsstatedconfidence,theoutcome,andalessonwrittenoncethegradearrived.Givenanewtask,XConf’sRecallstageretrievespastepisodesonsimilartasksmetwithasimilarstatedconfidence,andreadsofftheirhistoricalsuccessrate;itsReflectstageshowsthemodelthisrecord,hasitnameitsrecurringfailuremode,andrestateaconfidencenowinformedbyitsowntrackrecords.Ourestimatorisformat-general,requiringnologitaccessorweightupdates,andcostsonlyoneanswergeneration.Acrossninebenchmarksspanningreasoning,coding,multimodalQA,andinteractiveagents,andfourmodelsfromthreefamilies,XConfbeatsormatchesten-sampleself-consistencyindiscrimination(AUROC)on23of24comparisons,withmuchlowercalibrationerror(ECE),atatenthofthegenerationcost.Usedforselectiveprediction,abstainingonthe10%least-confidentepisodesraisesthedeliveredsuccessratebyupto8.7pointsonagenttasks.Wethereforeseeexperientialconfidenceestimationasanewparadigmforfuturegeneral-purposeconfidenceestimation.

View arXiv pageView PDFProject pageGitHubAdd to collection

Get this paper in your agent:

hf papers read 2609\.17708

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2609.17708 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2609.17708 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2609.17708 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

CALIBER: Calibrating Confidence Before and After Reasoning in Language Models

arXiv cs.CL

The paper introduces CALIBER, a method for calibrating confidence in reasoning language models by eliciting confidence estimates both before and after reasoning, with supervision targets matched to the information state. It achieves significant reductions in Expected Calibration Error (up to 52.5%) and strong Brier scores and AUROC across multiple benchmarks.