# LoVeC: Reinforcement Learning for Better Verbalized Confidence in Long-Form Generations Source: [LoVeC: Reinforcement Learning for Better Verbalized Confidence in Long-Form Generations]() Collection index: [llms.txt]() Data updated: 2026-05-16T05:00:05.655Z Authors: Caiqi Zhang, Xiaochen Zhu, Chengzu Li, Nigel Collier, Andreas Vlachos Published: 2025-05-29T18:05:20.000Z ## Abstract Hallucination remains a major challenge for the safe and trustworthy deployment of large language models (LLMs) in factual content generation. Prior work has explored confidence estimation as an effective approach to hallucination detection, but often relies on post-hoc self-consistency methods that require computationally expensive sampling. Verbalized confidence offers a more efficient alternative, but existing approaches are largely limited to short-form question answering (QA) tasks and do not generalize well to open-ended generation. In this paper, we propose LoVeC (Long-form Verbalized Confidence), a novel reinforcement learning based method that trains LLMs to append an on-the-fly numerical confidence score to each generated statement during long-form generation. The confidence score serves as a direct and interpretable signal of the factuality of generation. We introduce two evaluation settings, free-form tagging and iterative tagging, to assess different verbalized confidence estimation methods. Experiments on three long-form QA datasets show that our RL-trained models achieve better calibration and generalize robustly across domains. Also, our method is highly efficient, being 20 times faster than traditional self-consistency methods while achieving better calibration. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Confidence Estimation (Iterative Tagging) on Bios](): Confidence Estimation (Iterative Tagging); BIOS; Brier Score (BS): 7.5. - [Confidence Estimation (Iterative Tagging) on PopQA](): Confidence Estimation (Iterative Tagging); PopQA; Brier Score: 9.6. - [Confidence Estimation (Freeform Tagging) on WildHallu](): Confidence Estimation (Freeform Tagging); WildHallu; Brier Score (BS): 4.1. - [Confidence Estimation (Iterative Tagging) on WildHallu](): Confidence Estimation (Iterative Tagging); WildHallu; Brier Score (BS): 5.7. - [Confidence Estimation (Freeform Tagging) on Bios](): Confidence Estimation (Freeform Tagging); BIOS; Brier Score (BS): 9.2. - [Confidence Estimation (Freeform Tagging) on PopQA](): Confidence Estimation (Freeform Tagging); PopQA; Brier Score (BS): 10.1. - [Short-form Question Answering on Trivia QA](): Short-form Question Answering; Trivia QA; AUROC: 71.2. ## Original sources and implementation [Original paper]() [PDF]()