Metraipolis › archive › dc_c8fc0fc3f3e3
Error-Subspace Calibration for Multi-Task Model Merging — arXiv:2610.33483
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.33483
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-27T08:00:12.744Z
- SHA-256 of the archived text
440bacaea7699fdd0880a656dc4b5dc5917d25528f0431326c1e8b01c569fdc6- Stable URL of this page
https://metraipolis.com/documento/dc_c8fc0fc3f3e3
How to cite
Error-Subspace Calibration for Multi-Task Model Merging — arXiv:2610.33483 — arXiv (en), https://arxiv.org/abs/2610.33483 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_c8fc0fc3f3e3 (Metraipolis archive dc_c8fc0fc3f3e3, sha256 440bacaea7699fdd0880a656dc4b5dc5917d25528f0431326c1e8b01c569fdc6)
BibTeX
@misc{dcc8fc0fc3f3e3,
title = {Error-Subspace Calibration for Multi-Task Model Merging — arXiv:2610.33483},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.33483},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_c8fc0fc3f3e3; sha256 440bacaea7699fdd0880a656dc4b5dc5917d25528f0431326c1e8b01c569fdc6},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.33483. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Model merging combines task-specific expert models into a single model without joint training on their original datasets, but often incurs a performance gap relative to the individual experts. Post-merging calibration can reduce this gap through task-specific representation alignment, but the alignment objective alone does not specify which feature directions to correct. We propose Error-Subspace Calibration (ESCal), a post-merging calibration framework that identifies low-dimensional error subspaces from weight differences between expert and merged models. At each intermediate block boundary, a bridge module projects the merged representation onto the next block's input-projection error subspace and maps these coordinates to an additive correction in the current block's output-projection error subspace. The subspace bases remain fixed, so calibration learns only these linear maps, yielding a parameter-efficient mechanism for feature correction. We find that ESCal brings the merged model closer to the experts in both representations and predictions, indicating more faithful recovery of task-specific behavior. Experiments on vision and natural language understanding benchmarks show that ESCal improves merged-model performance across diverse architectures and merging methods using a small calibration set for each task. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.33483 — 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 2026-10-27T07:55:30.828Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).