Metraipolis › archive › dc_bace52819371
Dynamic Expert Pruning for Multi-Agent Systems — arXiv:2610.02951
- Source
- arXiv
- Original
- https://arxiv.org/abs/2610.02951
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-05T22:48:06.620Z
- SHA-256 of the archived text
d3c05a5d99a715a7669a446311164cb57ae0a01caf2f9e81c747f611246f979a- Stable URL of this page
https://metraipolis.com/documento/dc_bace52819371
How to cite
Dynamic Expert Pruning for Multi-Agent Systems — arXiv:2610.02951 — arXiv, https://arxiv.org/abs/2610.02951 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bace52819371 (Metraipolis archive dc_bace52819371, sha256 d3c05a5d99a715a7669a446311164cb57ae0a01caf2f9e81c747f611246f979a)
BibTeX
@misc{dcbace52819371,
title = {Dynamic Expert Pruning for Multi-Agent Systems — arXiv:2610.02951},
howpublished = {arXiv},
url = {https://arxiv.org/abs/2610.02951},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bace52819371; sha256 d3c05a5d99a715a7669a446311164cb57ae0a01caf2f9e81c747f611246f979a},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.02951. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Mixture-of-Experts (MoE) architectures scale language models efficiently by activating only a few experts per token, but the saving is confined to computation: every expert must stay resident on the accelerator, so memory bounds where these models can be deployed. Expert pruning reduces this footprint, yet existing methods are static --- a single mask, calibrated offline, is applied to the model for every subsequent request. This assumption can fail when the workload is heterogeneous, most prominently in multi-agent systems, where one backbone serves many tasks and roles at once: our analysis shows that different tasks and roles recruit different experts, while static methods assign one fixed subset to all of them. We therefore propose Dynamic Expert Pruning (DEP), which rests on a finding we establish here: an agent's system and task prompts are by themselves sufficient to identify the experts that agent and its task require, since that text already describes what the agent will do. A lightweight predictor, trained once on workflow transcripts, turns those prompts into a specialized per-request mask in a single forward pass, with no per-configuration calibration. Across diverse tasks and roles, model scales, and MoE architectures, DEP achieves better overall accuracy than static pruning and merging baselines, and generalizes to workflows unseen in training without retraining. Its margin over those baselines is largest when few experts are retained, suggesting that the role specialization inherent to multi-agent systems permits sparser serving than static pruning allows. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.MA, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA). Pagina del lavoro: https://arxiv.org/abs/2610.02951 — 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 2026-10-05T22:45:40.653Z dalla pagina https://arxiv.org/list/cs.MA/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).