Metraipolis › archive › dc_b18fc0fba22d

DUD: Decoupled Update Dynamics for Reliable Uncertainty Quantification in Large Language Models — arXiv:2608.03411

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.03411
License (as archived)
CC0-1.0
Archived
2026-10-31T12:31:02.070Z
SHA-256 of the archived text
af14ff9031c562f50cc1f894b3927979d479403da23a030fdcb6aa6a6e62bcf3
Stable URL of this page
https://metraipolis.com/documento/dc_b18fc0fba22d

How to cite

DUD: Decoupled Update Dynamics for Reliable Uncertainty Quantification in Large Language Models — arXiv:2608.03411 — arXiv (en), https://arxiv.org/abs/2608.03411 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_b18fc0fba22d (Metraipolis archive dc_b18fc0fba22d, sha256 af14ff9031c562f50cc1f894b3927979d479403da23a030fdcb6aa6a6e62bcf3)
BibTeX
@misc{dcb18fc0fba22d,
  title = {DUD: Decoupled Update Dynamics for Reliable Uncertainty Quantification in Large Language Models — arXiv:2608.03411},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.03411},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_b18fc0fba22d; sha256 af14ff9031c562f50cc1f894b3927979d479403da23a030fdcb6aa6a6e62bcf3},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.03411. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Accurate Uncertainty Quantification (UQ) is critical for reliable deployment of Large Language Models (LLMs), yet traditional probability-based metrics often fail to capture the model's true epistemic state. While recent mechanistic approaches leverage hidden state dynamics, they typically aggregate residual stream updates, conflating the distinct roles of parametric memory (Feed-Forward Networks) and contextual processing (Attention). We argue that this aggregation obscures fine-grained mechanistic conflicts, such as memory-context misalignment, that are fundamental indicators of uncertainty. To address this, we introduce \textbf{D}ecoupled \textbf{U}pdate \textbf{D}ynamics \textbf{(DUD)}, a framework that explicitly decouples FFN and Attention contributions via noise-induced causal interventions. By quantifying the independent restoration capabilities of each module, we construct a dual-stream dynamic profile that captures the model's internal fragility. Extensive experiments demonstrate that DUD significantly outperforms state-of-the-art baselines in both uncertainty estimation and calibration, while exhibiting superior cross-dataset generalization, validating decoupled dynamics as a robust proxy for model faithfulness.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-05 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.03411 — 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 2026-10-31T12:17:37.756Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-05?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).