CalVerT: Augmenting Agents with Calibrated Verifier Telemetry Improves Action and Learning in Knowledge-Intensive Tasks

Hugging Face Daily Papers Papers

Summary

CalVerT introduces calibrated verifier telemetry for LLM agents, providing confidence and grounding scores to reduce over-retrieval and unsupported answers, improving performance on knowledge-intensive QA tasks.

LLM agents in knowledge intensive question answering take retrieval and reasoning actions with incomplete knowledge about whether their current answer is uncertain, unsupported, or already complete. This produces two failure modes: committing to confident but unsupported answers, which hurts accuracy, and over-retrieving when the evidence in hand already suffices, resulting in wasted compute. To give agents a more complete picture of the state space they are operating in, we introduce calibrated verifier telemetry (CalVerT), which augments the agent's state with additional telemetry: a calibrated self-confidence score and a grounding verifier score. We show that CalVerT can improve agents in both training-free and training-based settings. On four QA benchmarks, we find that CalVerT raises F1 by triggering retrieval in cases where agents over-rely on parametric knowledge, while cutting redundant retrieval in cases where agents have sufficient context to answer. We show that CalVerT can augment existing QA frameworks without training. Moreover, CalVerT also improves trained systems: by simply augmenting an agent's state with telemetry, we observe improvements after reinforcement learning, as compared to an agent with identical training but no CalVerT telemetry.
Original Article
View Cached Full Text

Cached at: 06/23/26, 05:41 AM

Paper page - CalVerT: Augmenting Agents with Calibrated Verifier Telemetry Improves Action and Learning in Knowledge-Intensive Tasks

Source: https://huggingface.co/papers/2606.21777

Abstract

Calibrated verifier telemetry enhances LLM agents in knowledge-intensive question answering by providing confidence scores and grounding verification, reducing both over-retrieval and unsupported answers.

LLM agentsinknowledge intensive question answeringtakeretrievalandreasoningactions with incomplete knowledge about whether their current answer is uncertain, unsupported, or already complete. This produces two failure modes: committing to confident but unsupported answers, which hurts accuracy, and over-retrieving when the evidence in hand already suffices, resulting in wasted compute. To give agents a more complete picture of the state space they are operating in, we introduce calibrated verifier telemetry (CalVerT), which augments the agent’s state with additional telemetry: acalibrated self-confidence scoreand agrounding verifier score. We show that CalVerT can improve agents in both training-free and training-based settings. On four QA benchmarks, we find that CalVerT raises F1 by triggeringretrievalin cases where agents over-rely on parametric knowledge, while cutting redundantretrievalin cases where agents have sufficient context to answer. We show that CalVerT can augment existing QA frameworks without training. Moreover, CalVerT also improves trained systems: by simply augmenting an agent’s state with telemetry, we observe improvements afterreinforcement learning, as compared to an agent with identical training but no CalVerT telemetry.

View arXiv pageView PDFGitHub0Add to collection

Get this paper in your agent:

hf papers read 2606\.21777

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/2606.21777 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2606.21777 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2606.21777 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.