# Hallucination Detection via Activations of Open-Weight Proxy Analyzers

arXiv: https://arxiv.org/abs/2605.07209 · HTML: https://mlsanity.com/paper/2605.07209/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 15 of its 21 references that are in the map.
- First posted on arXiv 8 May 2026.

## Builds on (most cited first, 15 of 15)

- [Locating and Editing Factual Associations in GPT](https://mlsanity.com/paper/2202.05262/) (2022)
- [Survey of Hallucination in Natural Language Generation](https://mlsanity.com/paper/2202.03629/) (2022)
- [Transformer Feed-Forward Layers Are Key-Value Memories](https://mlsanity.com/paper/2012.14913/) (2020)
- [Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small](https://mlsanity.com/paper/2211.00593/) (2022)
- [SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models](https://mlsanity.com/paper/2303.08896/) (2023)
- [Adversarial NLI: A New Benchmark for Natural Language Understanding](https://mlsanity.com/paper/1910.14599/) (2019)
- [The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets](https://mlsanity.com/paper/2310.06824/) (2023)
- [The Internal State of an LLM Knows When It's Lying](https://mlsanity.com/paper/2304.13734/) (2023)
- [Ragas: Automated Evaluation of Retrieval Augmented Generation](https://mlsanity.com/paper/2309.15217/) (2023)
- [HaluEval: A Large-Scale Hallucination Evaluation Benchmark for Large Language Models](https://mlsanity.com/paper/2305.11747/) (2023)
- [INSIDE: LLMs' Internal States Retain the Power of Hallucination Detection](https://mlsanity.com/paper/2402.03744/) (2024)
- [MiniCheck: Efficient Fact-Checking of LLMs on Grounding Documents](https://mlsanity.com/paper/2404.10774/) (2024)
- [Lookback Lens: Detecting and Mitigating Contextual Hallucinations in Large Language Models Using Only Attention Maps](https://mlsanity.com/paper/2407.07071/) (2024)
- [Mechanistic Understanding and Mitigation of Language Model Non-Factual Hallucinations](https://mlsanity.com/paper/2403.18167/) (2024)
- [MedHallu: A Comprehensive Benchmark for Detecting Medical Hallucinations in Large Language Models](https://mlsanity.com/paper/2502.14302/) (2025)

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

- [Temporal Multi-Signal Fusion for Token-Level Hallucination Detection](https://mlsanity.com/paper/2608.18115/) (2026)
- [RAIM: Robust Aggregation of Inexpensive Models for Hallucination Detection](https://mlsanity.com/paper/2609.39229/) (2026)

## Abstract (arXiv)

We introduce a proxy-analyzer framework for detecting hallucinations in large language models. Instead of looking inside the generating model, our system reads already-generated text through a small locally hosted open-weight model and spots hallucinations using the reader's own internal activations. This works just as well when the generator is a closed API like GPT-4 as when it is any open-weight model. We built eighteen features grounded in how transformers process text, covering residual stream norms, per-head source-document attention, entropy, MLP activations, logit-lens trajectories, and three new token-level grounding statistics. We trained a stacking ensemble on 72,135 samples from five hallucination datasets. We tested across seven analyzer architectures from 0.5 billion to 9 billion parameters: Qwen2.5 at 0.5B and 7B, Gemma-2 at 2B and 9B, Pythia at 1.4B, and LLaMA-3 at both 3B and 8B. Across all seven, we consistently beat ReDeEP's token-level AUC of 0.73 on RAGTruth by 7.4 to 10.3 percentage points. Qwen2.5-7B reached an F1 of 0.717, just above ReDeEP's 0.713, while Qwen2.5-0.5B hit 0.706. The most striking finding is how tightly all seven models cluster: AUC spans only 2.3 percentage points across an eighteen-fold difference in model size. Even more surprising, our 3B LLaMA outperforms our 8B LLaMA on RAGTruth, showing that bigger is not always better even within the same model family. Both RAGTruth and LLM-AggreFact include outputs from multiple LLM families, so our results are not skewed toward any particular generator.