ICR Probe: Tracking Hidden State Dynamics for Reliable Hallucination Detection in LLMs
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866911069993172992 |
|---|---|
| author | Zhang, Zhenliang Hu, Xinyu Zhang, Huixuan Zhang, Junzhe Wan, Xiaojun |
| author_facet | Zhang, Zhenliang Hu, Xinyu Zhang, Huixuan Zhang, Junzhe Wan, Xiaojun |
| contents | Large language models (LLMs) excel at various natural language processing tasks, but their tendency to generate hallucinations undermines their reliability. Existing hallucination detection methods leveraging hidden states predominantly focus on static and isolated representations, overlooking their dynamic evolution across layers, which limits efficacy. To address this limitation, we shift the focus to the hidden state update process and introduce a novel metric, the ICR Score (Information Contribution to Residual Stream), which quantifies the contribution of modules to the hidden states' update. We empirically validate that the ICR Score is effective and reliable in distinguishing hallucinations. Building on these insights, we propose a hallucination detection method, the ICR Probe, which captures the cross-layer evolution of hidden states. Experimental results show that the ICR Probe achieves superior performance with significantly fewer parameters. Furthermore, ablation studies and case analyses offer deeper insights into the underlying mechanism of this method, improving its interpretability. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_16488 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | ICR Probe: Tracking Hidden State Dynamics for Reliable Hallucination Detection in LLMs Zhang, Zhenliang Hu, Xinyu Zhang, Huixuan Zhang, Junzhe Wan, Xiaojun Computation and Language Artificial Intelligence Large language models (LLMs) excel at various natural language processing tasks, but their tendency to generate hallucinations undermines their reliability. Existing hallucination detection methods leveraging hidden states predominantly focus on static and isolated representations, overlooking their dynamic evolution across layers, which limits efficacy. To address this limitation, we shift the focus to the hidden state update process and introduce a novel metric, the ICR Score (Information Contribution to Residual Stream), which quantifies the contribution of modules to the hidden states' update. We empirically validate that the ICR Score is effective and reliable in distinguishing hallucinations. Building on these insights, we propose a hallucination detection method, the ICR Probe, which captures the cross-layer evolution of hidden states. Experimental results show that the ICR Probe achieves superior performance with significantly fewer parameters. Furthermore, ablation studies and case analyses offer deeper insights into the underlying mechanism of this method, improving its interpretability. |
| title | ICR Probe: Tracking Hidden State Dynamics for Reliable Hallucination Detection in LLMs |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2507.16488 |