Metraipolis › archive › dc_19cc429d72dc

DeMTS: Denoising Trajectories as Multivariate Time Series for Hallucination Detection in Diffusion Language Models — arXiv:2608.14632

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.14632
License (as archived)
CC0-1.0
Archived
2026-10-18T17:51:17.856Z
SHA-256 of the archived text
9758f8a535936829f546af937aba90560ae185045e795f7756053819b42fa81b
Stable URL of this page
https://metraipolis.com/documento/dc_19cc429d72dc

How to cite

DeMTS: Denoising Trajectories as Multivariate Time Series for Hallucination Detection in Diffusion Language Models — arXiv:2608.14632 — arXiv (en), https://arxiv.org/abs/2608.14632 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_19cc429d72dc (Metraipolis archive dc_19cc429d72dc, sha256 9758f8a535936829f546af937aba90560ae185045e795f7756053819b42fa81b)
BibTeX
@misc{dc19cc429d72dc,
  title = {DeMTS: Denoising Trajectories as Multivariate Time Series for Hallucination Detection in Diffusion Language Models — arXiv:2608.14632},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.14632},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_19cc429d72dc; sha256 9758f8a535936829f546af937aba90560ae185045e795f7756053819b42fa81b},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.14632. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Diffusion large language models (D-LLMs) have emerged as a promising paradigm for text generation. However, similar to autoregressive LLMs, D-LLMs remain vulnerable to hallucinations, where fluent outputs may contain factually incorrect or unsupported content. Although existing hallucination detection methods for D-LLMs attempt to leverage uncertainty trajectories of the denoising process to better identify hallucination signals, they typically compress the trajectories along either the temporal or token dimension, overlooking the useful information encoded in the complete two-dimensional token-step structure. Consequently, they may fail to capture hallucination-relevant patterns, such as inconsistent convergence and cross-token fault propagation, leading to suboptimal detection performance. To bridge this gap, we propose a D-LLM hallucination detection framework that formulates the Denoising trajectories as Multivariate Time Series over learnable latent variables (DeMTS for short). DeMTS employs a trajectory-preserving token-to-variable assignment module to convert token signals into stable latent variables. Based on these variables, we propose dynamic multivariate temporal modeling to progressively integrate inter-variable dependency modeling with temporal encoding for hallucination prediction. Extensive experiments on two D-LLMs backbones and three benchmarks demonstrate that DeMTS outperforms existing hallucination detection methods while maintaining strong robustness, efficiency, and cross-task transferability.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-18 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.14632 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-18T16:29:13.919Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).