# HaluNet: Learning Hallucination Risk from Internal Signals in LLM Question Answering

arXiv: https://arxiv.org/abs/2512.24562 · HTML: https://mlsanity.com/paper/2512.24562/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 32 references that are in the map.
- First posted on arXiv 31 Dec 2025.

## Builds on (most cited first, 18 of 18)

- [Axiomatic Attribution for Deep Networks](https://mlsanity.com/paper/1703.01365/) (2017)
- [TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension](https://mlsanity.com/paper/1705.03551/) (2017)
- [Know What You Don't Know: Unanswerable Questions for SQuAD](https://mlsanity.com/paper/1806.03822/) (2018)
- [Aleatoric and Epistemic Uncertainty in Machine Learning: An Introduction to Concepts and Methods](https://mlsanity.com/paper/1910.09457/) (2019)
- [Selective Classification for Deep Neural Networks](https://mlsanity.com/paper/1705.08500/) (2017)
- [How Can We Know When Language Models Know? On the Calibration of Language Models for Question Answering](https://mlsanity.com/paper/2012.00955/) (2020)
- [Semantic Entropy Probes: Robust and Cheap Hallucination Detection in LLMs](https://mlsanity.com/paper/2406.15927/) (2024)
- [A Token-level Reference-free Hallucination Detection Benchmark for Free-form Text Generation](https://mlsanity.com/paper/2104.08704/) (2021)
- [Enhancing Uncertainty-Based Hallucination Detection with Stronger Focus](https://mlsanity.com/paper/2311.13230/) (2023)
- [Knowledge-Driven CoT: Exploring Faithful Reasoning in LLMs for Knowledge-intensive Question Answering](https://mlsanity.com/paper/2308.13259/) (2023)
- [Uncertainty Estimation and Quantification for LLMs: A Simple Supervised Approach](https://mlsanity.com/paper/2404.15993/) (2024)
- [Can ChatGPT Replace Traditional KBQA Models? An In-depth Analysis of the Question Answering Performance of the GPT LLM Family](https://mlsanity.com/paper/2303.07992/) (2023)
- [Uncertainty in Language Models: Assessment through Rank-Calibration](https://mlsanity.com/paper/2404.03163/) (2024)
- [FactLLaMA: Optimizing Instruction-Following Language Models with External Knowledge for Automated Fact-Checking](https://mlsanity.com/paper/2309.00240/) (2023)
- [HALoGEN: Fantastic LLM Hallucinations and Where to Find Them](https://mlsanity.com/paper/2501.08292/) (2025)
- [Improving Uncertainty Quantification in Large Language Models via Semantic Embeddings](https://mlsanity.com/paper/2410.22685/) (2024)
- [Do Not Design, Learn: A Trainable Scoring Function for Uncertainty Estimation in Generative LLMs](https://mlsanity.com/paper/2406.11278/) (2024)
- [Resolving Conflicting Evidence in Automated Fact-Checking: A Study on Retrieval-Augmented LLMs](https://mlsanity.com/paper/2505.17762/) (2025)

## Built on by (most cited first, 2 of 2)

- [FaithSCAN: Model-Driven Single-Pass Hallucination Detection for Faithful Visual Question Answering](https://mlsanity.com/paper/2601.00269/) (2026)
- [From Tokens to Semantics: Leveraging Complementary Signals for Hallucination Detection in Black-Box LLMs](https://mlsanity.com/paper/2609.02679/) (2026)

## Abstract (arXiv)

Large language models (LLMs) achieve strong question answering (QA) performance but can produce fluent answers unsupported by available evidence. Existing hallucination detectors often rely on external verification, repeated sampling, or test-time judge calls, which can be costly for real-time QA. We propose HaluNet, a lightweight hallucination risk estimator that uses internal signals from one model generation. HaluNet jointly models token likelihood, predictive entropy, and hidden-state information, allowing probabilistic, distributional, and semantic evidence to inform an answer-level risk score. It is trained with LLM-as-a-Judge labels as scalable weak supervision and evaluated with independent human and multi-judge assessments. Experiments on SQuAD, TriviaQA, and Natural Questions show that HaluNet improves answer-level risk ranking across in-domain and out-of-domain settings. On a 300-example human evaluation, HaluNet achieves 0.874 AUROC and 0.869 AUPRC; its top 20% highest-risk answers contain 96.5% errors, yielding a 2.06× lift over the base error rate.