paper-with-me

홈 › Papers

Detection Without Correction: A Robust Asymmetry in Activation-Based Hallucination Probing

2026-03-20 · Dip Roy, Rajiv Misra, Sanjay Kumar Singh, Anisha Roy arxiv

Activation-based linear probing is widely proposed as a method for both detecting and correcting hallucinations in autoregressive language models. We present an empirical study across seven models spanning 117M to 7B parameters and three architecture families (GPT-2, Pythia, Qwen-2.5) that documents a robust asymmetry: linear probes can detect hallucination signals with above-chance accuracy in larger models, but activation steering along the probe-derived direction fails to correct hallucinations in 7 of 7 models tested. We further find that output-confidence baselines outperform activation probes on raw detection AUC at every model above 410M parameters, with the gap reaching 0.157 AUC for Pythia-6.9B. The probe's distinguishing value is therefore not detection accuracy but temporal positioning: probe signals are accessible at position zero (before any output tokens are produced), enabling pre-generation flagging that output-based methods structurally cannot provide. The temporal signal is statistically significant in two of seven models (Pythia-1.4B, p = 0.012; Qwen2.5-7B, p = 0.038) and absent in models below 400M parameters and in the base-only Pythia-6.9B. We position these findings as a clean negative result for the dominant probing-as-detection-and-control research direction and as initial evidence that probe-based methods occupy a complementary deployment niche, namely pre-generation flagging, rather than competing with output-based detectors on raw accuracy.

📄 PDF Abstract BibTeX arXiv:2604.13068

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Hallucination as Trajectory Commitment: Causal Evidence for Asymmetric Attractor Dynamics in Transformer Generation

2026-04-16 · G. Aytug Akarlar arxiv

We present causal evidence that hallucination in autoregressive language models is an early trajectory commitment governed by asymmetric attractor dynamics. Using same-prompt bifurcation, in which we repeatedly sample id…

Hallucination Detection and Correction in Medical VLMs via Counter-Evidence Verification

2026-06-17 · Nan Zhou, Ke Zou, Meng Liu, Linchao He 외 arxiv

Vision-Language models (VLMs) reliability in medical diagnosis is challenged by trust-undermining hallucinations. Existing hallucination detection approaches mainly focus on identifying factual inconsistencies between ge…

Medical Report GenerationMedical DiagnosisVisual Grounding

FactSelfCheck: Fact-Level Black-Box Hallucination Detection for LLMs

2025-03-21 · Albert Sawczyn, Jakub Binkowski, Denis Janiak, Bogdan Gabrys 외

Large Language Models (LLMs) frequently generate hallucinated content, posing significant challenges for applications where factuality is crucial. While existing hallucination detection methods typically operate at the s…

HallucinationKnowledge GraphsSentence

Medico: Towards Hallucination Detection and Correction with Multi-source Evidence Fusion

2024-10-14 · Xinping Zhao, Jindi Yu, Zhenyu Liu, Jifang Wang 외

As we all know, hallucinations prevail in Large Language Models (LLMs), where the generated content is coherent but factually incorrect, which inflicts a heavy blow on the widespread application of LLMs. Previous studies…

Hallucination

Mitigating Entity-Level Hallucination in Large Language Models

2024-07-12 · Weihang Su, Yichen Tang, Qingyao Ai, Changyue Wang 외

The emergence of Large Language Models (LLMs) has revolutionized how users access information, shifting from traditional search engines to direct question-and-answer interactions with LLMs. However, the widespread adopti…

HallucinationInformation RetrievalRetrieval