Metraipolis › archive › dc_906ae618fce9

When Are Nearby Experts Composable? Cross-Task Model Merging in Neural Thickets — arXiv:2611.11177

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.11177
License (as archived)
CC0-1.0
Archived
2026-11-10T18:41:19.717Z
SHA-256 of the archived text
d3d3923ffc4f67b439d4ac423e89fe16343567adda3730b300fd508154167c33
Stable URL of this page
https://metraipolis.com/documento/dc_906ae618fce9

How to cite

When Are Nearby Experts Composable? Cross-Task Model Merging in Neural Thickets — arXiv:2611.11177 — arXiv (en), https://arxiv.org/abs/2611.11177 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_906ae618fce9 (Metraipolis archive dc_906ae618fce9, sha256 d3d3923ffc4f67b439d4ac423e89fe16343567adda3730b300fd508154167c33)
BibTeX
@misc{dc906ae618fce9,
  title = {When Are Nearby Experts Composable? Cross-Task Model Merging in Neural Thickets — arXiv:2611.11177},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.11177},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_906ae618fce9; sha256 d3d3923ffc4f67b439d4ac423e89fe16343567adda3730b300fd508154167c33},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.11177. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Random perturbations around a shared checkpoint can produce experts for distinct tasks, but shared origin does not ensure that their gains survive composition. We formulate expert-pair selection for joint base-relative retention, measured by strict joint success and worst-task gain with unconditional abstention accounting. An exact finite-difference decomposition and an arbitrarily small orthogonal counterexample show why proximity and orthogonality alone are insufficient. We instantiate this formulation with quality-constrained spectral selection, a fixed-budget framework combining cross-task screening, layerwise pair geometry, separate validation, and Base fallback. Across six checkpoints, 21 task pairs, and five independent perturbation pools, the procedure improves strict joint success over Top-1 averaging throughout and leads the four-method model-scale comparison on four of six checkpoints. For the five models with at least 1.5B parameters, matched quality-only, global-cosine, and spectral ranking attain 49.1%, 50.5%, and 53.1% joint success. Under this matched protocol, the spectral selector also yields the highest worst-task gain and Weak Pareto rate and the lowest regression rate. Mean worst-task gain reaches +0.61 percentage points, while 17.1% regression and 13.0% abstention show that composability remains conditional. Project page: this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2611.11177 — 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 2026-11-10T18:37:37.510Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).