Metraipolis › archive › dc_cd62d1f470e7

DMDIntel: Interpreting Large Language Models via Dynamic Mode Decomposition — arXiv:2608.13048

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.13048
License (as archived)
CC0-1.0
Archived
2026-10-19T04:17:28.919Z
SHA-256 of the archived text
4a686dc144b8b3295d8476b31ab923c5d5d7f150c72a8f15b3e302bc61053e05
Stable URL of this page
https://metraipolis.com/documento/dc_cd62d1f470e7

How to cite

DMDIntel: Interpreting Large Language Models via Dynamic Mode Decomposition — arXiv:2608.13048 — arXiv (en), https://arxiv.org/abs/2608.13048 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cd62d1f470e7 (Metraipolis archive dc_cd62d1f470e7, sha256 4a686dc144b8b3295d8476b31ab923c5d5d7f150c72a8f15b3e302bc61053e05)
BibTeX
@misc{dccd62d1f470e7,
  title = {DMDIntel: Interpreting Large Language Models via Dynamic Mode Decomposition — arXiv:2608.13048},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.13048},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cd62d1f470e7; sha256 4a686dc144b8b3295d8476b31ab923c5d5d7f150c72a8f15b3e302bc61053e05},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.13048. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

In this work, we introduce DMDIntel which uses dynamic mode decomposition (DMD) to make the predictions made by LLMs in a classification task interpretable. It develops an input attribution pipeline, that first decomposes the hidden states of an LLM into prominent patterns, also known as modes, and then associates ranks to the input tokens based on the projection values on those modes. Rigorous experiments across three datasets and three model families consistently show that the ranked attribution of input tokens obtained using DMDIntel by far outperforms state-of-the-art techniques such as principal component analysis, integrated gradients and SHAP.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-14 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.13048 — 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 2026-10-19T04:13:06.569Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-14?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).